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. They
16 complement the existing "set-log-level" and "set-log-target" verbs
17 used to change those values.
19 * journald.conf gained a new boolean setting ReadKMsg= which defaults
20 to on. If turned off kernel log messages will not be read by
21 systemd-journald or included in the logs. It also gained a new
22 setting LineMax= for configuring the maximum line length in
23 STDOUT/STDERR log streams. The new default for this value is 48K, up
24 from the previous hardcoded 2048.
26 * A new unit setting RuntimeDirectoryPreserve= has been added, which
27 allows more detailed control of what to do with a runtime directory
28 configured with RuntimeDirectory= (i.e. a directory below /run or
29 $XDG_RUNTIME_DIR) after a unit is stopped.
31 * The RuntimeDirectory= setting for units gained support for creating
32 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
33 one top-level directory.
35 * Units gained new options StateDirectory=, CacheDirectory=,
36 LogsDirectory= and ConfigurationDirectory= which are closely related
37 to RuntimeDirectory= but manage per-service directories below
38 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
39 possible to write unit files which when activated automatically gain
40 properly owned service specific directories in these locations, thus
41 making unit files self-contained and increasing compatibility with
42 stateless systems and factory reset where /etc or /var are
43 unpopulated at boot. Matching these new settings there's also
44 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
45 ConfigurationDirectoryMode= for configuring the access mode of these
48 * Automake support has been removed from this release. systemd is now
51 * systemd-journald will now aggressively cache client metadata during
52 runtime, speeding up log write performance under pressure. This comes
53 at a small price though: as much of the metadata is read
54 asynchronously from /proc/ (and isn't implicitly attached to log
55 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
56 metadata stored alongside a log entry might be slightly
57 out-of-date. Previously it could only be slightly newer than the log
58 message. The time window is small however, and given that the kernel
59 is unlikely to be improved anytime soon in this regard, this appears
62 * nss-myhostname/systemd-resolved will now by default synthesize an
63 A/AAAA resource record for the "_gateway" hostname, pointing to the
64 current default IP gateway. Previously it did that for the "gateway"
65 name, hampering adoption, as some distributions wanted to leave that
66 host name open for local use. The old behaviour may still be
67 requested at build time.
69 * systemd-networkd's [Address] section in .network files gained a new
70 Scope= setting for configuring the IP address scope. The [Network]
71 section gained a new boolean setting ConfigureWithoutCarrier= that
72 tells systemd-networkd to ignore link sensing when configuring the
73 device. The [DHCP] section gained a new Anonymize= boolean option for
74 turning on a number of options suggested in RFC 7844. A new
75 [RoutingPolicyRule] section has been added for configuring the IP
76 routing policy. The [Route] section has gained support for a new
77 Type= setting which permits configuring
78 blackhole/unreachable/prohibit routes.
80 * The [VRF] section in .netdev files gained a new Table= setting for
81 configuring the routing table to use. The [Tunnel] section gained a
82 new Independent= boolean field for configuring tunnels independent of
83 an underlying network interface. The [Bridge] section gained a new
84 GroupForwardMask= option for configuration of propagation of link
85 local frames between bridge ports.
87 * The WakeOnLan= setting in .link files gained support for a number of
88 new modes. A new TCP6SegmentationOffload= setting has been added for
89 configuring TCP/IPv6 hardware segmentation offload.
91 * The IPv6 RA sender implementation may now optionally send out RDNSS
92 and RDNSSL records to supply DNS configuration to peers.
94 * systemd-nspawn gained support for a new --system-call-filter= command
95 line option for adding and removing entries in the default system
96 call filter it applies. Moreover systemd-nspawn has been changed to
97 implement a system call whitelist instead of a blacklist.
99 * systemd-run gained support for a new --pipe command line option. If
100 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
101 are directly passed on to the activated transient service
102 executable. This allows invoking arbitrary processes as systemd
103 services (for example to take benefit of dependency management,
104 accounting management, resource management or log management that is
105 done automatically for services) — while still allowing them to be
106 integrated in a classic UNIX shell pipeline.
108 * When a service sends RELOAD=1 via sd_notify() and reload propagation
109 using ReloadPropagationTo= is configured, a reload is now propagated
110 to configured units. (Previously this was only done on explicitly
111 requested reloads, using "systemctl reload" or an equivalent
114 * For each service unit a restart counter is now kept: it is increased
115 each time the service is restarted due to Restart=, and may be
116 queried using "systemctl show -p NRestarts …".
118 * New system call filter groups @setuid, @memlock, @signal and
119 @timer have been added, for usage with SystemCallFilter=
120 in unit files and the new --system-call-filter= command line option
121 of systemd-nspawn (see above).
123 * ExecStart= lines in unit files gained two new modifiers: when a
124 command line is prefixed with "!" the command will be executed as
125 configured, except for the credentials applied by
126 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
127 "+", but does still apply namespacing options unlike "+". There's
128 also "!!" now, which is mostly identical, but becomes a NOP on
129 systems that support ambient capabilities. This is useful to write
130 unit files that work with ambient capabilities where possible but
131 automatically fall back to traditional privilege dropping mechanisms
132 on systems where this is not supported.
134 * ListenNetlink= settings in socket units now support RDMA netlink
137 * A new unit file setting LockPersonality= has been added which permits
138 locking down the chosen execution domain ("personality") of a service
141 * A new special target "getty-pre.target" has been added, which is
142 ordered before all text logins, and may be used to order services
143 before textual logins acquire access to the console.
145 * systemd will now attempt to load the virtio-rng.ko kernel module very
146 early on if a VM environment supporting this is detected. This should
147 improve entropy during early boot in virtualized environments.
149 * A _netdev option is now supported in /etc/crypttab that operates in a
150 similar way as the same option in /etc/fstab: it permits configuring
151 encrypted devices that need to be ordered after the network is up.
152 Following this logic, two new special targets
153 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
154 added that are to cryptsetup.target what remote-fs.target and
155 remote-fs-pre.target are to local-fs.target.
157 * Service units gained a new UnsetEnvironment= setting which permits
158 unsetting specific environment variables for services that are
159 normally passed to it (for example in order to mask out locale
160 settings for specific services that can't deal with it).
162 * Units acquired a new boolean option IPAccounting=. When turned on, IP
163 traffic accounting (packet count as well as byte count) is done for
164 the service, and shown as part of "systemctl status" or "systemd-run
167 * Service units acquired two new options IPAddressAllow= and
168 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
169 for configuring a simple IP access control list for all sockets of
170 the unit. These options are available also on .slice and .socket
171 units, permitting flexible access list configuration for individual
172 services as well as groups of services (as defined by a slice unit),
173 including system-wide. Note that IP ACLs configured this way are
174 enforced on every single IPv4 and IPv6 socket created by any process
175 of the service unit, and apply to ingress as well as egress traffic.
177 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
178 structured log message is generated each time the unit is stopped,
179 containing information about the consumed resources of this
182 * A new setting KeyringMode= has been added to unit files, which may be
183 used to control how the kernel keyring is set up for executed
186 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
187 "systemctl kexec" and "systemctl exit" are now always asynchronous in
188 behaviour (that is: these commands return immediately after the
189 operation was enqueued instead of waiting for the operation to
190 complete). Previously, "systemctl poweroff" and "systemctl reboot"
191 were asynchronous on systems using systemd-logind (i.e. almost
192 always, and like they were on sysvinit), and the other three commands
193 were unconditionally synchronous. With this release this is cleaned
194 up, and callers will see the same asynchronous behaviour on all
195 systems for all five operations.
197 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
200 * .timer units now accept calendar specifications in other timezones
201 than UTC or the local timezone.
203 * The tmpfiles snippet var.conf has been changed to create
204 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
205 the "utmp" group already, and it appears to be generally understood
206 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
207 databases. Previously this was implemented correctly for all these
208 databases excepts btmp, which has been opened up like this now
209 too. Note that while the other databases are world-readable
210 (i.e. 0644), btmp is not and remains more restrictive.
212 * The systemd-resolve tool gained a new --reset-server-features
213 switch. When invoked like this systemd-resolved will forget
214 everything it learnt about the features supported by the configured
215 upstream DNS servers, and restarts the feature probing logic on the
216 next resolver look-up for them at the highest feature level again.
218 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
219 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
220 Burchardt, b1tninja, bengal, Benjamin Berg, Benjamin Robin, Charles
221 Huber, Christian Hesse, Daniel Berrange, Daniel Mack, Daniel Rusek,
222 dasj19, Davide Cavalca, Dimitri John Ledkov, Diogo Pereira, Djalal
223 Harouni, dkg, dmig, Dmitry Torokhov, ettavolt, Evgeny Vereshchagin,
224 Fabio Kung, Felipe Sateler, Franck Bui, g0tar, Hans de Goede, Harald
225 Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov, Jakub Wilk, Jan
226 Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen, John Lin,
227 jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg Thalheim,
228 Jouke Witteveen, juga0, Justin Michaud, Kai-Heng Feng, Lennart
229 Poettering, Lion Yang, Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn,
230 Marcel Hollerbach, Marcus Lundblad, Martin Pitt, Michael Biebl, Michael
231 Grzeschik, Michal Sekletar, Mike Gilbert, Neil Brown, Nicolas Iooss,
232 Patrik Flykt, pEJipE, Piotr Drąg, Russell Stuart, S. Fan, Shengyao Xue,
233 Stefan Pietsch, Susant Sahani, Tejun Heo, Thomas Miller, Thomas Sailer,
234 Tobias Hunger, Tom Gundersen, Tommi Rantala, Topi Miettinen, Torstein
235 Husebø, userwithuid, Vasilis Liaskovitis, Vito Caputo, WaLyong Cho,
236 William Douglas, Xiang Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
242 * Meson is now supported as build system in addition to Automake. It is
243 our plan to remove Automake in one of our next releases, so that
244 Meson becomes our exclusive build system. Hence, please start using
245 the Meson build system in your downstream packaging. There's plenty
246 of documentation around how to use Meson, the extremely brief
249 ./autogen.sh && ./configure && make && sudo make install
253 meson build && ninja -C build && sudo ninja -C build install
255 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
256 which permits configuring a timeout on the time a job is
257 running. This is particularly useful for setting timeouts on jobs for
260 * Unit files gained two new options ConditionUser= and ConditionGroup=
261 for conditionalizing units based on the identity of the user/group
262 running a systemd user instance.
264 * systemd-networkd now understands a new FlowLabel= setting in the
265 [VXLAN] section of .network files, as well as a Priority= in
266 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
267 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
268 gained support for configuration of GENEVE links, and IPv6 address
269 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
271 * .link files now understand a new Port= setting.
273 * systemd-networkd's DHCP support gained support for DHCP option 119
274 (domain search list).
276 * systemd-networkd gained support for serving IPv6 address ranges using
277 the Router Advertisment protocol. The new .network configuration
278 section [IPv6Prefix] may be used to configure the ranges to
279 serve. This is implemented based on a new, minimal, native server
280 implementation of RA.
282 * journalctl's --output= switch gained support for a new parameter
283 "short-iso-precise" for a mode where timestamps are shown as precise
286 * systemd-udevd's "net_id" builtin may now generate stable network
287 interface names from IBM PowerVM VIO devices as well as ACPI platform
290 * MulticastDNS support in systemd-resolved may now be explicitly
291 enabled/disabled using the new MulticastDNS= configuration file
294 * systemd-resolved may now optionally use libidn2 instead of the libidn
295 for processing internationalized domain names. Support for libidn2
296 should be considered experimental and should not be enabled by
299 * "machinectl pull-tar" and related call may now do verification of
300 downloaded images using SUSE-style .sha256 checksum files in addition
301 to the already existing support for validating using Ubuntu-style
304 * sd-bus gained support for a new sd_bus_message_appendv() call which
305 is va_list equivalent of sd_bus_message_append().
307 * sd-boot gained support for validating images using SHIM/MOK.
309 * The SMACK code learnt support for "onlycap".
311 * systemd-mount --umount is now much smarter in figuring out how to
312 properly unmount a device given its mount or device path.
314 * The code to call libnss_dns as a fallback from libnss_resolve when
315 the communication with systemd-resolved fails was removed. This
316 fallback was redundant and interfered with the [!UNAVAIL=return]
317 suffix. See nss-resolve(8) for the recommended configuration.
319 * systemd-logind may now be restarted without losing state. It stores
320 the file descriptors for devices it manages in the system manager
321 using the FDSTORE= mechanism. Please note that further changes in
322 other components may be required to make use of this (for example
323 Xorg has code to listen for stops of systemd-logind and terminate
324 itself when logind is stopped or restarted, in order to avoid using
325 stale file descriptors for graphical devices, which is now
326 counterproductive and must be reverted in order for restarts of
327 systemd-logind to be safe. See
328 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
330 * All kernel install plugins are called with the environment variable
331 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
332 /etc/machine-id. If the file is missing or empty, the variable is
333 empty and BOOT_DIR_ABS is the path of a temporary directory which is
334 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
335 is empty, all plugins should not put anything in BOOT_DIR_ABS.
337 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
338 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
339 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
340 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
341 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
342 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
343 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
344 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
345 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
346 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
347 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
348 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
349 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
350 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
351 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
352 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
353 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
354 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
355 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
356 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
357 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
358 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
359 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
360 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
361 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
362 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
363 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
364 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
365 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
372 * The "hybrid" control group mode has been modified to improve
373 compatibility with "legacy" cgroups-v1 setups. Specifically, the
374 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
375 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
376 cgroups-v1 hierarchy), the only externally visible change being that
377 the cgroups-v2 hierarchy is also mounted, to
378 /sys/fs/cgroup/unified. This should provide a large degree of
379 compatibility with "legacy" cgroups-v1, while taking benefit of the
380 better management capabilities of cgroups-v2.
382 * The default control group setup mode may be selected both a boot-time
383 via a set of kernel command line parameters (specifically:
384 systemd.unified_cgroup_hierarchy= and
385 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
386 default selected on the configure command line
387 (--with-default-hierarchy=). The upstream default is "hybrid"
388 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
389 this will change in a future systemd version to be "unified" (pure
390 cgroups-v2 mode). The third option for the compile time option is
391 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
392 distributions to default to "hybrid" mode for release distributions,
393 starting with v233. We recommend "unified" for development
394 distributions (specifically: distributions such as Fedora's rawhide)
395 as that's where things are headed in the long run. Use "legacy" for
396 greatest stability and compatibility only.
398 * Note one current limitation of "unified" and "hybrid" control group
399 setup modes: the kernel currently does not permit the systemd --user
400 instance (i.e. unprivileged code) to migrate processes between two
401 disconnected cgroup subtrees, even if both are managed and owned by
402 the user. This effectively means "systemd-run --user --scope" doesn't
403 work when invoked from outside of any "systemd --user" service or
404 scope. Specifically, it is not supported from session scopes. We are
405 working on fixing this in a future systemd version. (See #3388 for
406 further details about this.)
408 * DBus policy files are now installed into /usr rather than /etc. Make
409 sure your system has dbus >= 1.9.18 running before upgrading to this
410 version, or override the install path with --with-dbuspolicydir= .
412 * All python scripts shipped with systemd (specifically: the various
413 tests written in Python) now require Python 3.
415 * systemd unit tests can now run standalone (without the source or
416 build directories), and can be installed into /usr/lib/systemd/tests/
417 with 'make install-tests'.
419 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
420 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
423 * Support for the %c, %r, %R specifiers in unit files has been
424 removed. Specifiers are not supposed to be dependent on configuration
425 in the unit file itself (so that they resolve the same regardless
426 where used in the unit files), but these specifiers were influenced
427 by the Slice= option.
429 * The shell invoked by debug-shell.service now defaults to /bin/sh in
430 all cases. If distributions want to use a different shell for this
431 purpose (for example Fedora's /sbin/sushell) they need to specify
432 this explicitly at configure time using --with-debug-shell=.
434 * The confirmation spawn prompt has been reworked to offer the
437 (c)ontinue, proceed without asking anymore
438 (D)ump, show the state of the unit
439 (f)ail, don't execute the command and pretend it failed
441 (i)nfo, show a short summary of the unit
442 (j)obs, show jobs that are in progress
443 (s)kip, don't execute the command and pretend it succeeded
444 (y)es, execute the command
446 The 'n' choice for the confirmation spawn prompt has been removed,
447 because its meaning was confusing.
449 The prompt may now also be redirected to an alternative console by
450 specifying the console as parameter to systemd.confirm_spawn=.
452 * Services of Type=notify require a READY=1 notification to be sent
453 during startup. If no such message is sent, the service now fails,
454 even if the main process exited with a successful exit code.
456 * Services that fail to start up correctly now always have their
457 ExecStopPost= commands executed. Previously, they'd enter "failed"
458 state directly, without executing these commands.
460 * The option MulticastDNS= of network configuration files has acquired
461 an actual implementation. With MulticastDNS=yes a host can resolve
462 names of remote hosts and reply to mDNS A and AAAA requests.
464 * When units are about to be started an additional check is now done to
465 ensure that all dependencies of type BindsTo= (when used in
466 combination with After=) have been started.
468 * systemd-analyze gained a new verb "syscall-filter" which shows which
469 system call groups are defined for the SystemCallFilter= unit file
470 setting, and which system calls they contain.
472 * A new system call filter group "@filesystem" has been added,
473 consisting of various file system related system calls. Group
474 "@reboot" has been added, covering reboot, kexec and shutdown related
475 calls. Finally, group "@swap" has been added covering swap
476 configuration related calls.
478 * A new unit file option RestrictNamespaces= has been added that may be
479 used to restrict access to the various process namespace types the
480 Linux kernel provides. Specifically, it may be used to take away the
481 right for a service unit to create additional file system, network,
482 user, and other namespaces. This sandboxing option is particularly
483 relevant due to the high amount of recently discovered namespacing
484 related vulnerabilities in the kernel.
486 * systemd-udev's .link files gained support for a new AutoNegotiation=
487 setting for configuring Ethernet auto-negotiation.
489 * systemd-networkd's .network files gained support for a new
490 ListenPort= setting in the [DHCP] section to explicitly configure the
491 UDP client port the DHCP client shall listen on.
493 * .network files gained a new Unmanaged= boolean setting for explicitly
494 excluding one or more interfaces from management by systemd-networkd.
496 * The systemd-networkd ProxyARP= option has been renamed to
497 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
498 renamed to ReduceARPProxy=. The old names continue to be available
501 * systemd-networkd gained support for configuring IPv6 Proxy NDP
502 addresses via the new IPv6ProxyNDPAddress= .network file setting.
504 * systemd-networkd's bonding device support gained support for two new
505 configuration options ActiveSlave= and PrimarySlave=.
507 * The various options in the [Match] section of .network files gained
508 support for negative matching.
510 * New systemd-specific mount options are now understood in /etc/fstab:
512 x-systemd.mount-timeout= may be used to configure the maximum
513 permitted runtime of the mount command.
515 x-systemd.device-bound may be set to bind a mount point to its
516 backing device unit, in order to automatically remove a mount point
517 if its backing device is unplugged. This option may also be
518 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
519 on the block device, which is now automatically set for all CDROM
520 drives, so that mounted CDs are automatically unmounted when they are
521 removed from the drive.
523 x-systemd.after= and x-systemd.before= may be used to explicitly
524 order a mount after or before another unit or mount point.
526 * Enqueued start jobs for device units are now automatically garbage
527 collected if there are no jobs waiting for them anymore.
529 * systemctl list-jobs gained two new switches: with --after, for every
530 queued job the jobs it's waiting for are shown; with --before the
531 jobs which it's blocking are shown.
533 * systemd-nspawn gained support for ephemeral boots from disk images
534 (or in other words: --ephemeral and --image= may now be
535 combined). Moreover, ephemeral boots are now supported for normal
536 directories, even if the backing file system is not btrfs. Of course,
537 if the file system does not support file system snapshots or
538 reflinks, the initial copy operation will be relatively expensive, but
539 this should still be suitable for many use cases.
541 * Calendar time specifications in .timer units now support
542 specifications relative to the end of a month by using "~" instead of
543 "-" as separator between month and day. For example, "*-02~03" means
544 "the third last day in February". In addition a new syntax for
545 repeated events has been added using the "/" character. For example,
546 "9..17/2:00" means "every two hours from 9am to 5pm".
548 * systemd-socket-proxyd gained a new parameter --connections-max= for
549 configuring the maximum number of concurrent connections.
551 * sd-id128 gained a new API for generating unique IDs for the host in a
552 way that does not leak the machine ID. Specifically,
553 sd_id128_get_machine_app_specific() derives an ID based on the
554 machine ID a in well-defined, non-reversible, stable way. This is
555 useful whenever an identifier for the host is needed but where the
556 identifier shall not be useful to identify the system beyond the
557 scope of the application itself. (Internally this uses HMAC-SHA256 as
558 keyed hash function using the machine ID as input.)
560 * NotifyAccess= gained a new supported value "exec". When set
561 notifications are accepted from all processes systemd itself invoked,
562 including all control processes.
564 * .nspawn files gained support for defining overlay mounts using the
565 Overlay= and OverlayReadOnly= options. Previously this functionality
566 was only available on the systemd-nspawn command line.
568 * systemd-nspawn's --bind= and --overlay= options gained support for
569 bind/overlay mounts whose source lies within the container tree by
570 prefixing the source path with "+".
572 * systemd-nspawn's --bind= and --overlay= options gained support for
573 automatically allocating a temporary source directory in /var/tmp
574 that is removed when the container dies. Specifically, if the source
575 directory is specified as empty string this mechanism is selected. An
576 example usage is --overlay=+/var::/var, which creates an overlay
577 mount based on the original /var contained in the image, overlayed
578 with a temporary directory in the host's /var/tmp. This way changes
579 to /var are automatically flushed when the container shuts down.
581 * systemd-nspawn --image= option does now permit raw file system block
582 devices (in addition to images containing partition tables, as
585 * The disk image dissection logic in systemd-nspawn gained support for
586 automatically setting up LUKS encrypted as well as Verity protected
587 partitions. When a container is booted from an encrypted image the
588 passphrase is queried at start-up time. When a container with Verity
589 data is started, the root hash is search in a ".roothash" file
590 accompanying the disk image (alternatively, pass the root hash via
591 the new --root-hash= command line option).
593 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
594 be used to dissect disk images the same way as systemd-nspawn does
595 it, following the Bootable Partition Specification. It may even be
596 used to mount disk images with complex partition setups (including
597 LUKS and Verity partitions) to a local host directory, in order to
598 inspect them. This tool is not considered public API (yet), and is
599 thus not installed into /usr/bin. Please do not rely on its
600 existence, since it might go away or be changed in later systemd
603 * A new generator "systemd-verity-generator" has been added, similar in
604 style to "systemd-cryptsetup-generator", permitting automatic setup of
605 Verity root partitions when systemd boots up. In order to make use of
606 this your partition setup should follow the Discoverable Partitions
607 Specification, and the GPT partition ID of the root file system
608 partition should be identical to the upper 128bit of the Verity root
609 hash. The GPT partition ID of the Verity partition protecting it
610 should be the lower 128bit of the Verity root hash. If the partition
611 image follows this model it is sufficient to specify a single
612 "roothash=" kernel command line argument to both configure which root
613 image and verity partition to use as well as the root hash for
614 it. Note that systemd-nspawn's Verity support follows the same
615 semantics, meaning that disk images with proper Verity data in place
616 may be booted in containers with systemd-nspawn as well as on
617 physical systems via the verity generator. Also note that the "mkosi"
618 tool available at https://github.com/systemd/mkosi has been updated
619 to generate Verity protected disk images following this scheme. In
620 fact, it has been updated to generate disk images that optionally
621 implement a complete UEFI SecureBoot trust chain, involving a signed
622 kernel and initrd image that incorporates such a root hash as well as
623 a Verity-enabled root partition.
625 * The hardware database (hwdb) udev supports has been updated to carry
626 accelerometer quirks.
628 * All system services are now run with a fresh kernel keyring set up
629 for them. The invocation ID is stored by default in it, thus
630 providing a safe, non-overridable way to determine the invocation
633 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
634 options for bind mounting arbitrary paths in a service-specific
635 way. When these options are used, arbitrary host or service files and
636 directories may be mounted to arbitrary locations in the service's
639 * Documentation has been added that lists all of systemd's low-level
640 environment variables:
642 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
644 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
645 whether a specific socket file descriptor matches a specified socket
648 * systemd-firstboot has been updated to check for the
649 systemd.firstboot= kernel command line option. It accepts a boolean
650 and when set to false the first boot questions are skipped.
652 * systemd-fstab-generator has been updated to check for the
653 systemd.volatile= kernel command line option, which either takes an
654 optional boolean parameter or the special value "state". If used the
655 system may be booted in a "volatile" boot mode. Specifically,
656 "systemd.volatile" is used, the root directory will be mounted as
657 tmpfs, and only /usr is mounted from the actual root file system. If
658 "systemd.volatile=state" is used, the root directory will be mounted
659 as usual, but /var is mounted as tmpfs. This concept provides similar
660 functionality as systemd-nspawn's --volatile= option, but provides it
661 on physical boots. Use this option for implementing stateless
662 systems, or testing systems with all state and/or configuration reset
663 to the defaults. (Note though that many distributions are not
664 prepared to boot up without a populated /etc or /var, though.)
666 * systemd-gpt-auto-generator gained support for LUKS encrypted root
667 partitions. Previously it only supported LUKS encrypted partitions
668 for all other uses, except for the root partition itself.
670 * Socket units gained support for listening on AF_VSOCK sockets for
671 communication in virtualized QEMU environments.
673 * The "configure" script gained a new option --with-fallback-hostname=
674 for specifying the fallback hostname to use if none is configured in
675 /etc/hostname. For example, by specifying
676 --with-fallback-hostname=fedora it is possible to default to a
677 hostname of "fedora" on pristine installations.
679 * systemd-cgls gained support for a new --unit= switch for listing only
680 the control groups of a specific unit. Similar --user-unit= has been
681 added for listing only the control groups of a specific user unit.
683 * systemd-mount gained a new --umount switch for unmounting a mount or
684 automount point (and all mount/automount points below it).
686 * systemd will now refuse full configuration reloads (via systemctl
687 daemon-reload and related calls) unless at least 16MiB of free space
688 are available in /run. This is a safety precaution in order to ensure
689 that generators can safely operate after the reload completed.
691 * A new unit file option RootImage= has been added, which has a similar
692 effect as RootDirectory= but mounts the service's root directory from
693 a disk image instead of plain directory. This logic reuses the same
694 image dissection and mount logic that systemd-nspawn already uses,
695 and hence supports any disk images systemd-nspawn supports, including
696 those following the Discoverable Partition Specification, as well as
697 Verity enabled images. This option enables systemd to run system
698 services directly off disk images acting as resource bundles,
699 possibly even including full integrity data.
701 * A new MountAPIVFS= unit file option has been added, taking a boolean
702 argument. If enabled /proc, /sys and /dev (collectively called the
703 "API VFS") will be mounted for the service. This is only relevant if
704 RootDirectory= or RootImage= is used for the service, as these mounts
705 are of course in place in the host mount namespace anyway.
707 * systemd-nspawn gained support for a new --pivot-root= switch. If
708 specified the root directory within the container image is pivoted to
709 the specified mount point, while the original root disk is moved to a
710 different place. This option enables booting of ostree images
711 directly with systemd-nspawn.
713 * The systemd build scripts will no longer complain if the NTP server
714 addresses are not changed from the defaults. Google now supports
715 these NTP servers officially. We still recommend downstreams to
716 properly register an NTP pool with the NTP pool project though.
718 * coredumpctl gained a new "--reverse" option for printing the list
719 of coredumps in reverse order.
721 * coredumpctl will now show additional information about truncated and
722 inaccessible coredumps, as well as coredumps that are still being
723 processed. It also gained a new --quiet switch for suppressing
724 additional informational message in its output.
726 * coredumpctl gained support for only showing coredumps newer and/or
727 older than specific timestamps, using the new --since= and --until=
728 options, reminiscent of journalctl's options by the same name.
730 * The systemd-coredump logic has been improved so that it may be reused
731 to collect backtraces in non-compiled languages, for example in
732 scripting languages such as Python.
734 * machinectl will now show the UID shift of local containers, if user
735 namespacing is enabled for them.
737 * systemd will now optionally run "environment generator" binaries at
738 configuration load time. They may be used to add environment
739 variables to the environment block passed to services invoked. One
740 user environment generator is shipped by default that sets up
741 environment variables based on files dropped into /etc/environment.d
742 and ~/.config/environment.d/.
744 * systemd-resolved now includes the new, recently published 2017 DNSSEC
747 * hostnamed has been updated to report a new chassis type of
748 "convertible" to cover "foldable" laptops that can both act as a
749 tablet and as a laptop, such as various Lenovo Yoga devices.
751 Contributions from: Adrián López, Alexander Galanin, Alexander
752 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
753 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
754 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
755 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
756 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
757 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
758 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
759 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
760 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
761 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
762 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
763 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
764 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
765 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
766 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
767 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
768 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
769 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
770 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
771 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
772 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
773 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
774 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
775 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
776 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
777 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
778 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
785 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
786 RestrictAddressFamilies= enabled. These sandboxing options should
787 generally be compatible with the various external udev call-out
788 binaries we are aware of, however there may be exceptions, in
789 particular when exotic languages for these call-outs are used. In
790 this case, consider turning off these settings locally.
792 * The new RemoveIPC= option can be used to remove IPC objects owned by
793 the user or group of a service when that service exits.
795 * The new ProtectKernelModules= option can be used to disable explicit
796 load and unload operations of kernel modules by a service. In
797 addition access to /usr/lib/modules is removed if this option is set.
799 * ProtectSystem= option gained a new value "strict", which causes the
800 whole file system tree with the exception of /dev, /proc, and /sys,
801 to be remounted read-only for a service.
803 * The new ProtectKernelTunables= option can be used to disable
804 modification of configuration files in /sys and /proc by a service.
805 Various directories and files are remounted read-only, so access is
806 restricted even if the file permissions would allow it.
808 * The new ProtectControlGroups= option can be used to disable write
809 access by a service to /sys/fs/cgroup.
811 * Various systemd services have been hardened with
812 ProtectKernelTunables=yes, ProtectControlGroups=yes,
813 RestrictAddressFamilies=.
815 * Support for dynamically creating users for the lifetime of a service
816 has been added. If DynamicUser=yes is specified, user and group IDs
817 will be allocated from the range 61184..65519 for the lifetime of the
818 service. They can be resolved using the new nss-systemd.so NSS
819 module. The module must be enabled in /etc/nsswitch.conf. Services
820 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
821 any resources allocated by the service will be cleaned up when the
822 service exits. They also have ProtectHome=read-only and
823 ProtectSystem=strict enabled, so they are not able to make any
824 permanent modifications to the system.
826 * The nss-systemd module also always resolves root and nobody, making
827 it possible to have no /etc/passwd or /etc/group files in minimal
828 container or chroot environments.
830 * Services may be started with their own user namespace using the new
831 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
832 under which the service is running are mapped. All other users are
835 * Support for the cgroup namespace has been added to systemd-nspawn. If
836 supported by kernel, the container system started by systemd-nspawn
837 will have its own view of the cgroup hierarchy. This new behaviour
838 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
840 * The new MemorySwapMax= option can be used to limit the maximum swap
841 usage under the unified cgroup hierarchy.
843 * Support for the CPU controller in the unified cgroup hierarchy has
844 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
845 options. This controller requires out-of-tree patches for the kernel
846 and the support is provisional.
848 * Mount and automount units may now be created transiently
849 (i.e. dynamically at runtime via the bus API, instead of requiring
850 unit files in the file system).
852 * systemd-mount is a new tool which may mount file systems – much like
853 mount(8), optionally pulling in additional dependencies through
854 transient .mount and .automount units. For example, this tool
855 automatically runs fsck on a backing block device before mounting,
856 and allows the automount logic to be used dynamically from the
857 command line for establishing mount points. This tool is particularly
858 useful when dealing with removable media, as it will ensure fsck is
859 run – if necessary – before the first access and that the file system
860 is quickly unmounted after each access by utilizing the automount
861 logic. This maximizes the chance that the file system on the
862 removable media stays in a clean state, and if it isn't in a clean
863 state is fixed automatically.
865 * LazyUnmount=yes option for mount units has been added to expose the
866 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
869 * /efi will be used as the mount point of the EFI boot partition, if
870 the directory is present, and the mount point was not configured
871 through other means (e.g. fstab). If /efi directory does not exist,
872 /boot will be used as before. This makes it easier to automatically
873 mount the EFI partition on systems where /boot is used for something
876 * When operating on GPT disk images for containers, systemd-nspawn will
877 now mount the ESP to /boot or /efi according to the same rules as PID
878 1 running on a host. This allows tools like "bootctl" to operate
879 correctly within such containers, in order to make container images
880 bootable on physical systems.
882 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
884 * Two new user session targets have been added to support running
885 graphical sessions under the systemd --user instance:
886 graphical-session.target and graphical-session-pre.target. See
887 systemd.special(7) for a description of how those targets should be
890 * The vconsole initialization code has been significantly reworked to
891 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
892 support unicode keymaps. Font and keymap configuration will now be
893 copied to all allocated virtual consoles.
895 * FreeBSD's bhyve virtualization is now detected.
897 * Information recorded in the journal for core dumps now includes the
898 contents of /proc/mountinfo and the command line of the process at
899 the top of the process hierarchy (which is usually the init process
902 * systemd-journal-gatewayd learned the --directory= option to serve
903 files from the specified location.
905 * journalctl --root=… can be used to peruse the journal in the
906 /var/log/ directories inside of a container tree. This is similar to
907 the existing --machine= option, but does not require the container to
910 * The hardware database has been extended to support
911 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
914 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
915 specify the click rate for mice which include a horizontal wheel with
916 a click rate that is different than the one for the vertical wheel.
918 * systemd-run gained a new --wait option that makes service execution
919 synchronous. (Specifically, the command will not return until the
920 specified service binary exited.)
922 * systemctl gained a new --wait option that causes the start command to
923 wait until the units being started have terminated again.
925 * A new journal output mode "short-full" has been added which displays
926 timestamps with abbreviated English day names and adds a timezone
927 suffix. Those timestamps include more information than the default
928 "short" output mode, and can be passed directly to journalctl's
929 --since= and --until= options.
931 * /etc/resolv.conf will be bind-mounted into containers started by
932 systemd-nspawn, if possible, so any changes to resolv.conf contents
933 are automatically propagated to the container.
935 * The number of instances for socket-activated services originating
936 from a single IP address can be limited with
937 MaxConnectionsPerSource=, extending the existing setting of
940 * systemd-networkd gained support for vcan ("Virtual CAN") interface
943 * .netdev and .network configuration can now be extended through
946 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
947 Segmentation Offload, Generic Receive Offload, Large Receive Offload
948 can be enabled and disabled using the new UDPSegmentationOffload=,
949 TCPSegmentationOffload=, GenericSegmentationOffload=,
950 GenericReceiveOffload=, LargeReceiveOffload= options in the
951 [Link] section of .link files.
953 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
954 Port VLAN ID can be configured for bridge devices using the new STP=,
955 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
956 section of .netdev files.
958 * The route table to which routes received over DHCP or RA should be
959 added can be configured with the new RouteTable= option in the [DHCP]
960 and [IPv6AcceptRA] sections of .network files.
962 * The Address Resolution Protocol can be disabled on links managed by
963 systemd-networkd using the ARP=no setting in the [Link] section of
966 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
967 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
968 encode information about the result and exit codes of the current
969 service runtime cycle.
971 * systemd-sysctl will now configure kernel parameters in the order
972 they occur in the configuration files. This matches what sysctl
973 has been traditionally doing.
975 * kernel-install "plugins" that are executed to perform various
976 tasks after a new kernel is added and before an old one is removed
977 can now return a special value to terminate the procedure and
978 prevent any later plugins from running.
980 * Journald's SplitMode=login setting has been deprecated. It has been
981 removed from documentation, and its use is discouraged. In a future
982 release it will be completely removed, and made equivalent to current
983 default of SplitMode=uid.
985 * Storage=both option setting in /etc/systemd/coredump.conf has been
986 removed. With fast LZ4 compression storing the core dump twice is not
989 * The --share-system systemd-nspawn option has been replaced with an
990 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
991 this functionality is discouraged. In addition the variables
992 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
993 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
994 individual namespaces.
996 * "machinectl list" now shows the IP address of running containers in
997 the output, as well as OS release information.
999 * "loginctl list" now shows the TTY of each session in the output.
1001 * sd-bus gained new API calls sd_bus_track_set_recursive(),
1002 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
1003 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1004 tracking objects in a "recursive" mode, where a single client can be
1005 counted multiple times, if it takes multiple references.
1007 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1008 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1009 process using sd-bus automatically exit if the bus connection is
1012 * Bus clients of the service manager may now "pin" loaded units into
1013 memory, by taking an explicit reference on them. This is useful to
1014 ensure the client can retrieve runtime data about the service even
1015 after the service completed execution. Taking such a reference is
1016 available only for privileged clients and should be helpful to watch
1017 running services in a race-free manner, and in particular collect
1018 information about exit statuses and results.
1020 * The nss-resolve module has been changed to strictly return UNAVAIL
1021 when communication via D-Bus with resolved failed, and NOTFOUND when
1022 a lookup completed but was negative. This means it is now possible to
1023 neatly configure fallbacks using nsswitch.conf result checking
1024 expressions. Taking benefit of this, the new recommended
1025 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1027 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1029 * A new setting CtrlAltDelBurstAction= has been added to
1030 /etc/systemd/system.conf which may be used to configure the precise
1031 behaviour if the user on the console presses Ctrl-Alt-Del more often
1032 than 7 times in 2s. Previously this would unconditionally result in
1033 an expedited, immediate reboot. With this new setting the precise
1034 operation may be configured in more detail, and also turned off
1037 * In .netdev files two new settings RemoteChecksumTx= and
1038 RemoteChecksumRx= are now understood that permit configuring the
1039 remote checksumming logic for VXLAN networks.
1041 * The service manager learnt a new "invocation ID" concept for invoked
1042 services. Each runtime cycle of a service will get a new invocation
1043 ID (a 128bit random UUID) assigned that identifies the current
1044 run of the service uniquely and globally. A new invocation ID
1045 is generated each time a service starts up. The journal will store
1046 the invocation ID of a service along with any logged messages, thus
1047 making the invocation ID useful for matching the online runtime of a
1048 service with the offline log data it generated in a safe way without
1049 relying on synchronized timestamps. In many ways this new service
1050 invocation ID concept is similar to the kernel's boot ID concept that
1051 uniquely and globally identifies the runtime of each boot. The
1052 invocation ID of a service is passed to the service itself via an
1053 environment variable ($INVOCATION_ID). A new bus call
1054 GetUnitByInvocationID() has been added that is similar to GetUnit()
1055 but instead of retrieving the bus path for a unit by its name
1056 retrieves it by its invocation ID. The returned path is valid only as
1057 long as the passed invocation ID is current.
1059 * systemd-resolved gained a new "DNSStubListener" setting in
1060 resolved.conf. It either takes a boolean value or the special values
1061 "udp" and "tcp", and configures whether to enable the stub DNS
1062 listener on 127.0.0.53:53.
1064 * IP addresses configured via networkd may now carry additional
1065 configuration settings supported by the kernel. New options include:
1066 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1067 PrefixRoute=, AutoJoin=.
1069 * The PAM configuration fragment file for "user@.service" shipped with
1070 systemd (i.e. the --user instance of systemd) has been stripped to
1071 the minimum necessary to make the system boot. Previously, it
1072 contained Fedora-specific stanzas that did not apply to other
1073 distributions. It is expected that downstream distributions add
1074 additional configuration lines, matching their needs to this file,
1075 using it only as rough template of what systemd itself needs. Note
1076 that this reduced fragment does not even include an invocation of
1077 pam_limits which most distributions probably want to add, even though
1078 systemd itself does not need it. (There's also the new build time
1079 option --with-pamconfdir=no to disable installation of the PAM
1082 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1083 capability is now also dropped from its set (in addition to
1084 CAP_SYS_MKNOD as before).
1086 * In service unit files it is now possible to connect a specific named
1087 file descriptor with stdin/stdout/stdout of an executed service. The
1088 name may be specified in matching .socket units using the
1089 FileDescriptorName= setting.
1091 * A number of journal settings may now be configured on the kernel
1092 command line. Specifically, the following options are now understood:
1093 systemd.journald.max_level_console=,
1094 systemd.journald.max_level_store=,
1095 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1096 systemd.journald.max_level_wall=.
1098 * "systemctl is-enabled --full" will now show by which symlinks a unit
1099 file is enabled in the unit dependency tree.
1101 * Support for VeraCrypt encrypted partitions has been added to the
1102 "cryptsetup" logic and /etc/crypttab.
1104 * systemd-detect-virt gained support for a new --private-users switch
1105 that checks whether the invoking processes are running inside a user
1106 namespace. Similar, a new special value "private-users" for the
1107 existing ConditionVirtualization= setting has been added, permitting
1108 skipping of specific units in user namespace environments.
1110 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1111 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1112 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1113 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1114 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1115 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1116 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1117 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1118 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1119 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1120 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1121 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1122 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1123 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1124 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1125 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1126 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1127 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1128 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1129 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1130 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1131 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1132 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1133 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1134 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1135 Jędrzejewski-Szmek, Zeal Jagannatha
1137 — Santa Fe, 2016-11-03
1141 * In service units the various ExecXYZ= settings have been extended
1142 with an additional special character as first argument of the
1143 assigned value: if the character '+' is used the specified command
1144 line it will be run with full privileges, regardless of User=,
1145 Group=, CapabilityBoundingSet= and similar options. The effect is
1146 similar to the existing PermissionsStartOnly= option, but allows
1147 configuration of this concept for each executed command line
1150 * Services may now alter the service watchdog timeout at runtime by
1151 sending a WATCHDOG_USEC= message via sd_notify().
1153 * MemoryLimit= and related unit settings now optionally take percentage
1154 specifications. The percentage is taken relative to the amount of
1155 physical memory in the system (or in case of containers, the assigned
1156 amount of memory). This allows scaling service resources neatly with
1157 the amount of RAM available on the system. Similarly, systemd-logind's
1158 RuntimeDirectorySize= option now also optionally takes percentage
1161 * In similar fashion TasksMax= takes percentage values now, too. The
1162 value is taken relative to the configured maximum number of processes
1163 on the system. The per-service task maximum has been changed to 15%
1164 using this functionality. (Effectively this is an increase of 512 →
1165 4915 for service units, given the kernel's default pid_max setting.)
1167 * Calendar time specifications in .timer units now understand a ".."
1168 syntax for time ranges. Example: "4..7:10" may now be used for
1169 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1172 * The InaccessableDirectories=, ReadOnlyDirectories= and
1173 ReadWriteDirectories= unit file settings have been renamed to
1174 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1175 applied to all kinds of file nodes, and not just directories, with
1176 the exception of symlinks. Specifically these settings may now be
1177 used on block and character device nodes, UNIX sockets and FIFOS as
1178 well as regular files. The old names of these settings remain
1179 available for compatibility.
1181 * systemd will now log about all service processes it kills forcibly
1182 (using SIGKILL) because they remained after the clean shutdown phase
1183 of the service completed. This should help identifying services that
1184 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1185 systemd-logind's configuration a similar log message is generated for
1186 processes killed at the end of each session due to this setting.
1188 * systemd will now set the $JOURNAL_STREAM environment variable for all
1189 services whose stdout/stderr are connected to the Journal (which
1190 effectively means by default: all services). The variable contains
1191 the device and inode number of the file descriptor used for
1192 stdout/stderr. This may be used by invoked programs to detect whether
1193 their stdout/stderr is connected to the Journal, in which case they
1194 can switch over to direct Journal communication, thus being able to
1195 pass extended, structured metadata along with their log messages. As
1196 one example, this is now used by glib's logging primitives.
1198 * When using systemd's default tmp.mount unit for /tmp, the mount point
1199 will now be established with the "nosuid" and "nodev" options. This
1200 avoids privilege escalation attacks that put traps and exploits into
1201 /tmp. However, this might cause problems if you e. g. put container
1202 images or overlays into /tmp; if you need this, override tmp.mount's
1203 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1206 * systemd now supports the "memory" cgroup controller also on
1209 * The systemd-cgtop tool now optionally takes a control group path as
1210 command line argument. If specified, the control group list shown is
1211 limited to subgroups of that group.
1213 * The SystemCallFilter= unit file setting gained support for
1214 pre-defined, named system call filter sets. For example
1215 SystemCallFilter=@clock is now an effective way to make all clock
1216 changing-related system calls unavailable to a service. A number of
1217 similar pre-defined groups are defined. Writing system call filters
1218 for system services is simplified substantially with this new
1219 concept. Accordingly, all of systemd's own, long-running services now
1220 enable system call filtering based on this, by default.
1222 * A new service setting MemoryDenyWriteExecute= has been added, taking
1223 a boolean value. If turned on, a service may no longer create memory
1224 mappings that are writable and executable at the same time. This
1225 enhances security for services where this is enabled as it becomes
1226 harder to dynamically write and then execute memory in exploited
1227 service processes. This option has been enabled for all of systemd's
1228 own long-running services.
1230 * A new RestrictRealtime= service setting has been added, taking a
1231 boolean argument. If set the service's processes may no longer
1232 acquire realtime scheduling. This improves security as realtime
1233 scheduling may otherwise be used to easily freeze the system.
1235 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1236 value. This may be used for requesting that the system manager inside
1237 of the container reports start-up completion to nspawn which then
1238 propagates this notification further to the service manager
1239 supervising nspawn itself. A related option NotifyReady= in .nspawn
1240 files has been added too. This functionality allows ordering of the
1241 start-up of multiple containers using the usual systemd ordering
1244 * machinectl gained a new command "stop" that is an alias for
1247 * systemd-resolved gained support for contacting DNS servers on
1248 link-local IPv6 addresses.
1250 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1251 its caches. A method call for requesting the same operation has been
1252 added to the bus API too, and is made available via "systemd-resolve
1255 * systemd-resolve gained a new --status switch. If passed a brief
1256 summary of the used DNS configuration with per-interface information
1259 * resolved.conf gained a new Cache= boolean option, defaulting to
1260 on. If turned off local DNS caching is disabled. This comes with a
1261 performance penalty in particular when DNSSEC is enabled. Note that
1262 resolved disables its internal caching implicitly anyway, when the
1263 configured DNS server is on a host-local IP address such as ::1 or
1264 127.0.0.1, thus automatically avoiding double local caching.
1266 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1267 for DNS requests. This improves compatibility with local programs
1268 that do not use the libc NSS or systemd-resolved's bus APIs for name
1269 resolution. This minimal DNS service is only available to local
1270 programs and does not implement the full DNS protocol, but enough to
1271 cover local DNS clients. A new, static resolv.conf file, listing just
1272 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1273 now recommended to make /etc/resolv.conf a symlink to this file in
1274 order to route all DNS lookups to systemd-resolved, regardless if
1275 done via NSS, the bus API or raw DNS packets. Note that this local
1276 DNS service is not as fully featured as the libc NSS or
1277 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1278 used to deliver link-local address information (as this implies
1279 sending a local interface index along), LLMNR/mDNS support via this
1280 interface is severely restricted. It is thus strongly recommended for
1281 all applications to use the libc NSS API or native systemd-resolved
1284 * systemd-networkd's bridge support learned a new setting
1285 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1286 in .network files has been added for configuring VLAN bridging in
1287 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1289 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1290 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1291 now be acquired without relying on DHCPv6. Two new options
1292 UseDomains= and UseDNS= have been added to configure this behaviour.
1294 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1295 renamed IPv6AcceptRA=, without altering its behaviour. The old
1296 setting name remains available for compatibility reasons.
1298 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1299 Key=, InputKey= and OutputKey=.
1301 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1302 interface configuration.
1304 * "systemctl edit" may now be used to create new unit files by
1305 specifying the --force switch.
1307 * sd-event gained a new function sd_event_get_iteration() for
1308 requesting the current iteration counter of the event loop. It starts
1309 at zero and is increased by one with each event loop iteration.
1311 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1312 file. It can be used in lieu of %systemd_requires in packages which
1313 don't use any systemd functionality and are intended to be installed
1314 in minimal containers without systemd present. This macro provides
1315 ordering dependencies to ensure that if the package is installed in
1316 the same rpm transaction as systemd, systemd will be installed before
1317 the scriptlets for the package are executed, allowing unit presets
1320 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1321 been added to simplify packaging of generators.
1323 * The os-release file gained VERSION_CODENAME field for the
1324 distribution nickname (e.g. VERSION_CODENAME=woody).
1326 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1327 can be set to disable parsing of metadata and the creation
1328 of persistent symlinks for that device.
1330 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1331 to make them available to logged-in users has been reverted.
1333 * Much of the common code of the various systemd components is now
1334 built into an internal shared library libsystemd-shared-231.so
1335 (incorporating the systemd version number in the name, to be updated
1336 with future releases) that the components link to. This should
1337 decrease systemd footprint both in memory during runtime and on
1338 disk. Note that the shared library is not for public use, and is
1339 neither API not ABI stable, but is likely to change with every new
1340 released update. Packagers need to make sure that binaries
1341 linking to libsystemd-shared.so are updated in step with the
1344 * Configuration for "mkosi" is now part of the systemd
1345 repository. mkosi is a tool to easily build legacy-free OS images,
1346 and is available on github: https://github.com/systemd/mkosi. If
1347 "mkosi" is invoked in the build tree a new raw OS image is generated
1348 incorporating the systemd sources currently being worked on and a
1349 clean, fresh distribution installation. The generated OS image may be
1350 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1351 UEFI PC. This functionality is particularly useful to easily test
1352 local changes made to systemd in a pristine, defined environment. See
1353 HACKING for details.
1355 * configure learned the --with-support-url= option to specify the
1356 distribution's bugtracker.
1358 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1359 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1360 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1361 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1362 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1363 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1364 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1365 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1366 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1367 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1368 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1369 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1370 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1371 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1372 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1373 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1374 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1375 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1376 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1378 — Berlin, 2016-07-25
1382 * DNSSEC is now turned on by default in systemd-resolved (in
1383 "allow-downgrade" mode), but may be turned off during compile time by
1384 passing "--with-default-dnssec=no" to "configure" (and of course,
1385 during runtime with DNSSEC= in resolved.conf). We recommend
1386 downstreams to leave this on at least during development cycles and
1387 report any issues with the DNSSEC logic upstream. We are very
1388 interested in collecting feedback about the DNSSEC validator and its
1389 limitations in the wild. Note however, that DNSSEC support is
1390 probably nothing downstreams should turn on in stable distros just
1391 yet, as it might create incompatibilities with a few DNS servers and
1392 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1393 automatically whenever we detect such incompatible setups, but there
1394 might be systems we do not cover yet. Hence: please help us testing
1395 the DNSSEC code, leave this on where you can, report back, but then
1396 again don't consider turning this on in your stable, LTS or
1397 production release just yet. (Note that you have to enable
1398 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1399 and its DNSSEC mode for host name resolution from local
1402 * systemd-resolve conveniently resolves DANE records with the --tlsa
1403 option and OPENPGPKEY records with the --openpgp option. It also
1404 supports dumping raw DNS record data via the new --raw= switch.
1406 * systemd-logind will now by default terminate user processes that are
1407 part of the user session scope unit (session-XX.scope) when the user
1408 logs out. This behavior is controlled by the KillUserProcesses=
1409 setting in logind.conf, and the previous default of "no" is now
1410 changed to "yes". This means that user sessions will be properly
1411 cleaned up after, but additional steps are necessary to allow
1412 intentionally long-running processes to survive logout.
1414 While the user is logged in at least once, user@.service is running,
1415 and any service that should survive the end of any individual login
1416 session can be started at a user service or scope using systemd-run.
1417 systemd-run(1) man page has been extended with an example which shows
1418 how to run screen in a scope unit underneath user@.service. The same
1419 command works for tmux.
1421 After the user logs out of all sessions, user@.service will be
1422 terminated too, by default, unless the user has "lingering" enabled.
1423 To effectively allow users to run long-term tasks even if they are
1424 logged out, lingering must be enabled for them. See loginctl(1) for
1425 details. The default polkit policy was modified to allow users to
1426 set lingering for themselves without authentication.
1428 Previous defaults can be restored at compile time by the
1429 --without-kill-user-processes option to "configure".
1431 * systemd-logind gained new configuration settings SessionsMax= and
1432 InhibitorsMax=, both with a default of 8192. It will not register new
1433 user sessions or inhibitors above this limit.
1435 * systemd-logind will now reload configuration on SIGHUP.
1437 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1438 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1439 enable. Also, support for the "io" cgroup controller in the unified
1440 hierarchy has been added, so that the "memory", "pids" and "io" are
1441 now the controllers that are supported on the unified hierarchy.
1443 WARNING: it is not possible to use previous systemd versions with
1444 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1445 is necessary to also update systemd in the initramfs if using the
1446 unified hierarchy. An updated SELinux policy is also required.
1448 * LLDP support has been extended, and both passive (receive-only) and
1449 active (sender) modes are supported. Passive mode ("routers-only") is
1450 enabled by default in systemd-networkd. Active LLDP mode is enabled
1451 by default for containers on the internal network. The "networkctl
1452 lldp" command may be used to list information gathered. "networkctl
1453 status" will also show basic LLDP information on connected peers now.
1455 * The IAID and DUID unique identifier sent in DHCP requests may now be
1456 configured for the system and each .network file managed by
1457 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1459 * systemd-networkd gained support for configuring proxy ARP support for
1460 each interface, via the ProxyArp= setting in .network files. It also
1461 gained support for configuring the multicast querier feature of
1462 bridge devices, via the new MulticastQuerier= setting in .netdev
1463 files. Similarly, snooping on the IGMP traffic can be controlled
1464 via the new setting MulticastSnooping=.
1466 A new setting PreferredLifetime= has been added for addresses
1467 configured in .network file to configure the lifetime intended for an
1470 The systemd-networkd DHCP server gained the option EmitRouter=, which
1471 defaults to yes, to configure whether the DHCP Option 3 (Router)
1474 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1475 systemd-socket-activate and installed into /usr/bin. It is now fully
1478 * systemd-journald now uses separate threads to flush changes to disk
1479 when closing journal files, thus reducing impact of slow disk I/O on
1480 logging performance.
1482 * The sd-journal API gained two new calls
1483 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1484 can be used to open journal files using file descriptors instead of
1485 file or directory paths. sd_journal_open_container() has been
1486 deprecated, sd_journal_open_directory_fd() should be used instead
1487 with the flag SD_JOURNAL_OS_ROOT.
1489 * journalctl learned a new output mode "-o short-unix" that outputs log
1490 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1491 UTC). It also gained support for a new --no-hostname setting to
1492 suppress the hostname column in the family of "short" output modes.
1494 * systemd-ask-password now optionally skips printing of the password to
1495 stdout with --no-output which can be useful in scripts.
1497 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1498 (devices tagged with ID_MAKER_TOOL) are now tagged with
1499 "uaccess" and are available to logged in users.
1501 * The DeviceAllow= unit setting now supports specifiers (with "%").
1503 * "systemctl show" gained a new --value switch, which allows print a
1504 only the contents of a specific unit property, without also printing
1505 the property's name. Similar support was added to "show*" verbs
1506 of loginctl and machinectl that output "key=value" lists.
1508 * A new unit type "generated" was added for files dynamically generated
1509 by generator tools. Similarly, a new unit type "transient" is used
1510 for unit files created using the runtime API. "systemctl enable" will
1511 refuse to operate on such files.
1513 * A new command "systemctl revert" has been added that may be used to
1514 revert to the vendor version of a unit file, in case local changes
1515 have been made by adding drop-ins or overriding the unit file.
1517 * "machinectl clean" gained a new verb to automatically remove all or
1518 just hidden container images.
1520 * systemd-tmpfiles gained support for a new line type "e" for emptying
1521 directories, if they exist, without creating them if they don't.
1523 * systemd-nspawn gained support for automatically patching the UID/GIDs
1524 of the owners and the ACLs of all files and directories in a
1525 container tree to match the UID/GID user namespacing range selected
1526 for the container invocation. This mode is enabled via the new
1527 --private-users-chown switch. It also gained support for
1528 automatically choosing a free, previously unused UID/GID range when
1529 starting a container, via the new --private-users=pick setting (which
1530 implies --private-users-chown). Together, these options for the first
1531 time make user namespacing for nspawn containers fully automatic and
1532 thus deployable. The systemd-nspawn@.service template unit file has
1533 been changed to use this functionality by default.
1535 * systemd-nspawn gained a new --network-zone= switch, that allows
1536 creating ad-hoc virtual Ethernet links between multiple containers,
1537 that only exist as long as at least one container referencing them is
1538 running. This allows easy connecting of multiple containers with a
1539 common link that implements an Ethernet broadcast domain. Each of
1540 these network "zones" may be named relatively freely by the user, and
1541 may be referenced by any number of containers, but each container may
1542 only reference one of these "zones". On the lower level, this is
1543 implemented by an automatically managed bridge network interface for
1544 each zone, that is created when the first container referencing its
1545 zone is created and removed when the last one referencing its zone
1548 * The default start timeout may now be configured on the kernel command
1549 line via systemd.default_timeout_start_sec=. It was already
1550 configurable via the DefaultTimeoutStartSec= option in
1551 /etc/systemd/system.conf.
1553 * Socket units gained a new TriggerLimitIntervalSec= and
1554 TriggerLimitBurst= setting to configure a limit on the activation
1555 rate of the socket unit.
1557 * The LimitNICE= setting now optionally takes normal UNIX nice values
1558 in addition to the raw integer limit value. If the specified
1559 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1560 value is understood as UNIX nice value. If not prefixed like this it
1561 is understood as raw RLIMIT_NICE limit.
1563 * Note that the effect of the PrivateDevices= unit file setting changed
1564 slightly with this release: the per-device /dev file system will be
1565 mounted read-only from this version on, and will have "noexec"
1566 set. This (minor) change of behavior might cause some (exceptional)
1567 legacy software to break, when PrivateDevices=yes is set for its
1568 service. Please leave PrivateDevices= off if you run into problems
1571 * systemd-bootchart has been split out to a separate repository:
1572 https://github.com/systemd/systemd-bootchart
1574 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1575 merged into the kernel in its current form.
1577 * The compatibility libraries libsystemd-daemon.so,
1578 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1579 which have been deprecated since systemd-209 have been removed along
1580 with the corresponding pkg-config files. All symbols provided by
1581 those libraries are provided by libsystemd.so.
1583 * The Capabilities= unit file setting has been removed (it is ignored
1584 for backwards compatibility). AmbientCapabilities= and
1585 CapabilityBoundingSet= should be used instead.
1587 * A new special target has been added, initrd-root-device.target,
1588 which creates a synchronization point for dependencies of the root
1589 device in early userspace. Initramfs builders must ensure that this
1590 target is now included in early userspace.
1592 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1593 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1594 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1595 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1596 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1597 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1598 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1599 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1600 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1601 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1602 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1603 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1604 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1605 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1606 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1607 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1608 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1609 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1610 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1611 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1612 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1613 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1614 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1615 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1616 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1619 — Fairfax, 2016-05-21
1623 * The systemd-resolved DNS resolver service has gained a substantial
1624 set of new features, most prominently it may now act as a DNSSEC
1625 validating stub resolver. DNSSEC mode is currently turned off by
1626 default, but is expected to be turned on by default in one of the
1627 next releases. For now, we invite everybody to test the DNSSEC logic
1628 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1629 service also gained a full set of D-Bus interfaces, including calls
1630 to configure DNS and DNSSEC settings per link (for use by external
1631 network management software). systemd-resolved and systemd-networkd
1632 now distinguish between "search" and "routing" domains. The former
1633 are used to qualify single-label names, the latter are used purely
1634 for routing lookups within certain domains to specific links.
1635 resolved now also synthesizes RRs for all entries from /etc/hosts.
1637 * The systemd-resolve tool (which is a client utility for
1638 systemd-resolved) has been improved considerably and is now fully
1639 supported and documented. Hence it has moved from /usr/lib/systemd to
1642 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1645 * The coredump collection logic has been reworked: when a coredump is
1646 collected it is now written to disk, compressed and processed
1647 (including stacktrace extraction) from a new instantiated service
1648 systemd-coredump@.service, instead of directly from the
1649 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1650 processing large coredumps can take up a substantial amount of
1651 resources and time, and this previously happened entirely outside of
1652 systemd's service supervision. With the new logic the core_pattern
1653 hook only does minimal metadata collection before passing off control
1654 to the new instantiated service, which is configured with a time
1655 limit, a nice level and other settings to minimize negative impact on
1656 the rest of the system. Also note that the new logic will honour the
1657 RLIMIT_CORE setting of the crashed process, which now allows users
1658 and processes to turn off coredumping for their processes by setting
1661 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1662 and all forked processes by default. Previously, PID 1 would leave
1663 the setting at "0" for all processes, as set by the kernel. Note that
1664 the resource limit traditionally has no effect on the generated
1665 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1666 logic is used. Since the limit is now honoured (see above) its
1667 default has been changed so that the coredumping logic is enabled by
1668 default for all processes, while allowing specific opt-out.
1670 * When the stacktrace is extracted from processes of system users, this
1671 is now done as "systemd-coredump" user, in order to sandbox this
1672 potentially security sensitive parsing operation. (Note that when
1673 processing coredumps of normal users this is done under the user ID
1674 of process that crashed, as before.) Packagers should take notice
1675 that it is now necessary to create the "systemd-coredump" system user
1676 and group at package installation time.
1678 * The systemd-activate socket activation testing tool gained support
1679 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1680 and --seqpacket switches. It also has been extended to support both
1681 new-style and inetd-style file descriptor passing. Use the new
1682 --inetd switch to request inetd-style file descriptor passing.
1684 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1685 variable, which takes a boolean value. If set to false, ANSI color
1686 output is disabled in the tools even when run on a terminal that
1689 * The VXLAN support in networkd now supports two new settings
1690 DestinationPort= and PortRange=.
1692 * A new systemd.machine_id= kernel command line switch has been added,
1693 that may be used to set the machine ID in /etc/machine-id if it is
1694 not initialized yet. This command line option has no effect if the
1695 file is already initialized.
1697 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1698 specified command line as PID 2 rather than PID 1 in the
1699 container. In this mode PID 1 is a minimal stub init process that
1700 implements the special POSIX and Linux semantics of PID 1 regarding
1701 signal and child process management. Note that this stub init process
1702 is implemented in nspawn itself and requires no support from the
1703 container image. This new logic is useful to support running
1704 arbitrary commands in the container, as normal processes are
1705 generally not prepared to run as PID 1.
1707 * systemd-nspawn gained a new --chdir= switch for setting the current
1708 working directory for the process started in the container.
1710 * "journalctl /dev/sda" will now output all kernel log messages for
1711 specified device from the current boot, in addition to all devices
1712 that are parents of it. This should make log output about devices
1713 pretty useful, as long as kernel drivers attach enough metadata to
1714 the log messages. (The usual SATA drivers do.)
1716 * The sd-journal API gained two new calls
1717 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1718 that report whether log data from /run or /var has been found.
1720 * journalctl gained a new switch "--fields" that prints all journal
1721 record field names currently in use in the journal. This is backed
1722 by two new sd-journal API calls sd_journal_enumerate_fields() and
1723 sd_journal_restart_fields().
1725 * Most configurable timeouts in systemd now expect an argument of
1726 "infinity" to turn them off, instead of "0" as before. The semantics
1727 from now on is that a timeout of "0" means "now", and "infinity"
1728 means "never". To maintain backwards compatibility, "0" continues to
1729 turn off previously existing timeout settings.
1731 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1732 try-reload-or-restart" to clarify what it actually does: the "try"
1733 logic applies to both reloading and restarting, not just restarting.
1734 The old name continues to be accepted for compatibility.
1736 * On boot-up, when PID 1 detects that the system clock is behind the
1737 release date of the systemd version in use, the clock is now set
1738 to the latter. Previously, this was already done in timesyncd, in order
1739 to avoid running with clocks set to the various clock epochs such as
1740 1902, 1938 or 1970. With this change the logic is now done in PID 1
1741 in addition to timesyncd during early boot-up, so that it is enforced
1742 before the first process is spawned by systemd. Note that the logic
1743 in timesyncd remains, as it is more comprehensive and ensures
1744 clock monotonicity by maintaining a persistent timestamp file in
1745 /var. Since /var is generally not available in earliest boot or the
1746 initrd, this part of the logic remains in timesyncd, and is not done
1749 * Support for tweaking details in net_cls.class_id through the
1750 NetClass= configuration directive has been removed, as the kernel
1751 people have decided to deprecate that controller in cgroup v2.
1752 Userspace tools such as nftables are moving over to setting rules
1753 that are specific to the full cgroup path of a task, which obsoletes
1754 these controllers anyway. The NetClass= directive is kept around for
1755 legacy compatibility reasons. For a more in-depth description of the
1756 kernel change, please refer to the respective upstream commit:
1758 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1760 * A new service setting RuntimeMaxSec= has been added that may be used
1761 to specify a maximum runtime for a service. If the timeout is hit, the
1762 service is terminated and put into a failure state.
1764 * A new service setting AmbientCapabilities= has been added. It allows
1765 configuration of additional Linux process capabilities that are
1766 passed to the activated processes. This is only available on very
1769 * The process resource limit settings in service units may now be used
1770 to configure hard and soft limits individually.
1772 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1773 expose support for gcc's __attribute__((cleanup())) C extension.
1774 Specifically, for many object destructor functions alternative
1775 versions have been added that have names suffixed with "p" and take a
1776 pointer to a pointer to the object to destroy, instead of just a
1777 pointer to the object itself. This is useful because these destructor
1778 functions may be used directly as parameters to the cleanup
1779 construct. Internally, systemd has been a heavy user of this GCC
1780 extension for a long time, and with this change similar support is
1781 now available to consumers of the library outside of systemd. Note
1782 that by using this extension in your sources compatibility with old
1783 and strictly ANSI compatible C compilers is lost. However, all gcc or
1784 LLVM versions of recent years support this extension.
1786 * Timer units gained support for a new setting RandomizedDelaySec= that
1787 allows configuring some additional randomized delay to the configured
1788 time. This is useful to spread out timer events to avoid load peaks in
1789 clusters or larger setups.
1791 * Calendar time specifications now support sub-second accuracy.
1793 * Socket units now support listening on SCTP and UDP-lite protocol
1796 * The sd-event API now comes with a full set of man pages.
1798 * Older versions of systemd contained experimental support for
1799 compressing journal files and coredumps with the LZ4 compressor that
1800 was not compatible with the lz4 binary (due to API limitations of the
1801 lz4 library). This support has been removed; only support for files
1802 compatible with the lz4 binary remains. This LZ4 logic is now
1803 officially supported and no longer considered experimental.
1805 * The dkr image import logic has been removed again from importd. dkr's
1806 micro-services focus doesn't fit into the machine image focus of
1807 importd, and quickly got out of date with the upstream dkr API.
1809 * Creation of the /run/lock/lockdev/ directory was dropped from
1810 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1811 been available for many years. If you still need this, you need to
1812 create your own tmpfiles.d config file with:
1814 d /run/lock/lockdev 0775 root lock -
1816 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1817 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1818 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1819 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1820 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1821 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1822 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1823 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1824 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1825 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1826 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1827 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1828 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1829 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1830 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1831 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1832 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1833 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1834 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1836 — Berlin, 2016-02-11
1840 * A number of properties previously only settable in unit
1841 files are now also available as properties to set when
1842 creating transient units programmatically via the bus, as it
1843 is exposed with systemd-run's --property=
1844 setting. Specifically, these are: SyslogIdentifier=,
1845 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1846 EnvironmentFile=, ReadWriteDirectories=,
1847 ReadOnlyDirectories=, InaccessibleDirectories=,
1848 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1850 * When creating transient services via the bus API it is now
1851 possible to pass in a set of file descriptors to use as
1852 STDIN/STDOUT/STDERR for the invoked process.
1854 * Slice units may now be created transiently via the bus APIs,
1855 similar to the way service and scope units may already be
1856 created transiently.
1858 * Wherever systemd expects a calendar timestamp specification
1859 (like in journalctl's --since= and --until= switches) UTC
1860 timestamps are now supported. Timestamps suffixed with "UTC"
1861 are now considered to be in Universal Time Coordinated
1862 instead of the local timezone. Also, timestamps may now
1863 optionally be specified with sub-second accuracy. Both of
1864 these additions also apply to recurring calendar event
1865 specification, such as OnCalendar= in timer units.
1867 * journalctl gained a new "--sync" switch that asks the
1868 journal daemon to write all so far unwritten log messages to
1869 disk and sync the files, before returning.
1871 * systemd-tmpfiles learned two new line types "q" and "Q" that
1872 operate like "v", but also set up a basic btrfs quota
1873 hierarchy when used on a btrfs file system with quota
1876 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1877 instead of a subvolume (even on a btrfs file system) if the
1878 root directory is a plain directory, and not a
1879 subvolume. This should simplify things with certain chroot()
1880 environments which are not aware of the concept of btrfs
1883 * systemd-detect-virt gained a new --chroot switch to detect
1884 whether execution takes place in a chroot() environment.
1886 * CPUAffinity= now takes CPU index ranges in addition to
1889 * The various memory-related resource limit settings (such as
1890 LimitAS=) now understand the usual K, M, G, ... suffixes to
1891 the base of 1024 (IEC). Similar, the time-related resource
1892 limit settings understand the usual min, h, day, ...
1895 * There's a new system.conf setting DefaultTasksMax= to
1896 control the default TasksMax= setting for services and
1897 scopes running on the system. (TasksMax= is the primary
1898 setting that exposes the "pids" cgroup controller on systemd
1899 and was introduced in the previous systemd release.) The
1900 setting now defaults to 512, which means services that are
1901 not explicitly configured otherwise will only be able to
1902 create 512 processes or threads at maximum, from this
1903 version on. Note that this means that thread- or
1904 process-heavy services might need to be reconfigured to set
1905 TasksMax= to a higher value. It is sufficient to set
1906 TasksMax= in these specific unit files to a higher value, or
1907 even "infinity". Similar, there's now a logind.conf setting
1908 UserTasksMax= that defaults to 4096 and limits the total
1909 number of processes or tasks each user may own
1910 concurrently. nspawn containers also have the TasksMax=
1911 value set by default now, to 8192. Note that all of this
1912 only has an effect if the "pids" cgroup controller is
1913 enabled in the kernel. The general benefit of these changes
1914 should be a more robust and safer system, that provides a
1915 certain amount of per-service fork() bomb protection.
1917 * systemd-nspawn gained the new --network-veth-extra= switch
1918 to define additional and arbitrarily-named virtual Ethernet
1919 links between the host and the container.
1921 * A new service execution setting PassEnvironment= has been
1922 added that allows importing select environment variables
1923 from PID1's environment block into the environment block of
1926 * Timer units gained support for a new RemainAfterElapse=
1927 setting which takes a boolean argument. It defaults to on,
1928 exposing behaviour unchanged to previous releases. If set to
1929 off, timer units are unloaded after they elapsed if they
1930 cannot elapse again. This is particularly useful for
1931 transient timer units, which shall not stay around longer
1932 than until they first elapse.
1934 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1935 default now (the kernel default is 16). This is beneficial
1936 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1937 allows substantially larger numbers of queued
1938 datagrams. This should increase the capability of systemd to
1939 parallelize boot-up, as logging and sd_notify() are unlikely
1940 to stall execution anymore. If you need to change the value
1941 from the new defaults, use the usual sysctl.d/ snippets.
1943 * The compression framing format used by the journal or
1944 coredump processing has changed to be in line with what the
1945 official LZ4 tools generate. LZ4 compression support in
1946 systemd was considered unsupported previously, as the format
1947 was not compatible with the normal tools. With this release
1948 this has changed now, and it is hence safe for downstream
1949 distributions to turn it on. While not compressing as well
1950 as the XZ, LZ4 is substantially faster, which makes
1951 it a good default choice for the compression logic in the
1952 journal and in coredump handling.
1954 * Any reference to /etc/mtab has been dropped from
1955 systemd. The file has been obsolete since a while, but
1956 systemd refused to work on systems where it was incorrectly
1957 set up (it should be a symlink or non-existent). Please make
1958 sure to update to util-linux 2.27.1 or newer in conjunction
1959 with this systemd release, which also drops any reference to
1960 /etc/mtab. If you maintain a distribution make sure that no
1961 software you package still references it, as this is a
1962 likely source of bugs. There's also a glibc bug pending,
1963 asking for removal of any reference to this obsolete file:
1965 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1967 Note that only util-linux versions built with
1968 --enable-libmount-force-mountinfo are supported.
1970 * Support for the ".snapshot" unit type has been removed. This
1971 feature turned out to be little useful and little used, and
1972 has now been removed from the core and from systemctl.
1974 * The dependency types RequiresOverridable= and
1975 RequisiteOverridable= have been removed from systemd. They
1976 have been used only very sparingly to our knowledge and
1977 other options that provide a similar effect (such as
1978 systemctl --mode=ignore-dependencies) are much more useful
1979 and commonly used. Moreover, they were only half-way
1980 implemented as the option to control behaviour regarding
1981 these dependencies was never added to systemctl. By removing
1982 these dependency types the execution engine becomes a bit
1983 simpler. Unit files that use these dependencies should be
1984 changed to use the non-Overridable dependency types
1985 instead. In fact, when parsing unit files with these
1986 options, that's what systemd will automatically convert them
1987 too, but it will also warn, asking users to fix the unit
1988 files accordingly. Removal of these dependency types should
1989 only affect a negligible number of unit files in the wild.
1991 * Behaviour of networkd's IPForward= option changed
1992 (again). It will no longer maintain a per-interface setting,
1993 but propagate one way from interfaces where this is enabled
1994 to the global kernel setting. The global setting will be
1995 enabled when requested by a network that is set up, but
1996 never be disabled again. This change was made to make sure
1997 IPv4 and IPv6 behaviour regarding packet forwarding is
1998 similar (as the Linux IPv6 stack does not support
1999 per-interface control of this setting) and to minimize
2002 * In unit files the behaviour of %u, %U, %h, %s has
2003 changed. These specifiers will now unconditionally resolve
2004 to the various user database fields of the user that the
2005 systemd instance is running as, instead of the user
2006 configured in the specific unit via User=. Note that this
2007 effectively doesn't change much, as resolving of these
2008 specifiers was already turned off in the --system instance
2009 of systemd, as we cannot do NSS lookups from PID 1. In the
2010 --user instance of systemd these specifiers where correctly
2011 resolved, but hardly made any sense, since the user instance
2012 lacks privileges to do user switches anyway, and User= is
2013 hence useless. Moreover, even in the --user instance of
2014 systemd behaviour was awkward as it would only take settings
2015 from User= assignment placed before the specifier into
2016 account. In order to unify and simplify the logic around
2017 this the specifiers will now always resolve to the
2018 credentials of the user invoking the manager (which in case
2019 of PID 1 is the root user).
2021 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2022 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2023 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2024 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2025 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2026 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2027 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2028 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2029 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2030 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2031 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2032 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2033 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2034 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2037 — Berlin, 2015-11-18
2041 * systemd now depends on util-linux v2.27. More specifically,
2042 the newly added mount monitor feature in libmount now
2043 replaces systemd's former own implementation.
2045 * libmount mandates /etc/mtab not to be regular file, and
2046 systemd now enforces this condition at early boot.
2047 /etc/mtab has been deprecated and warned about for a very
2048 long time, so systems running systemd should already have
2049 stopped having this file around as anything else than a
2050 symlink to /proc/self/mounts.
2052 * Support for the "pids" cgroup controller has been added. It
2053 allows accounting the number of tasks in a cgroup and
2054 enforcing limits on it. This adds two new setting
2055 TasksAccounting= and TasksMax= to each unit, as well as a
2056 global option DefaultTasksAccounting=.
2058 * Support for the "net_cls" cgroup controller has been added.
2059 It allows assigning a net class ID to each task in the
2060 cgroup, which can then be used in firewall rules and traffic
2061 shaping configurations. Note that the kernel netfilter net
2062 class code does not currently work reliably for ingress
2063 packets on unestablished sockets.
2065 This adds a new config directive called NetClass= to CGroup
2066 enabled units. Allowed values are positive numbers for fixed
2067 assignments and "auto" for picking a free value
2070 * 'systemctl is-system-running' now returns 'offline' if the
2071 system is not booted with systemd. This command can now be
2072 used as a substitute for 'systemd-notify --booted'.
2074 * Watchdog timeouts have been increased to 3 minutes for all
2075 in-tree service files. Apparently, disk IO issues are more
2076 frequent than we hoped, and user reported >1 minute waiting
2079 * 'machine-id-commit' functionality has been merged into
2080 'machine-id-setup --commit'. The separate binary has been
2083 * The WorkingDirectory= directive in unit files may now be set
2084 to the special value '~'. In this case, the working
2085 directory is set to the home directory of the user
2086 configured in User=.
2088 * "machinectl shell" will now open the shell in the home
2089 directory of the selected user by default.
2091 * The CrashChVT= configuration file setting is renamed to
2092 CrashChangeVT=, following our usual logic of not
2093 abbreviating unnecessarily. The old directive is still
2094 supported for compat reasons. Also, this directive now takes
2095 an integer value between 1 and 63, or a boolean value. The
2096 formerly supported '-1' value for disabling stays around for
2099 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2100 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2101 RootDirectory= properties can now be set for transient
2104 * The systemd-analyze tool gained a new "set-log-target" verb
2105 to change the logging target the system manager logs to
2106 dynamically during runtime. This is similar to how
2107 "systemd-analyze set-log-level" already changes the log
2110 * In nspawn /sys is now mounted as tmpfs, with only a selected
2111 set of subdirectories mounted in from the real sysfs. This
2112 enhances security slightly, and is useful for ensuring user
2113 namespaces work correctly.
2115 * Support for USB FunctionFS activation has been added. This
2116 allows implementation of USB gadget services that are
2117 activated as soon as they are requested, so that they don't
2118 have to run continuously, similar to classic socket
2121 * The "systemctl exit" command now optionally takes an
2122 additional parameter that sets the exit code to return from
2123 the systemd manager when exiting. This is only relevant when
2124 running the systemd user instance, or when running the
2125 system instance in a container.
2127 * sd-bus gained the new API calls sd_bus_path_encode_many()
2128 and sd_bus_path_decode_many() that allow easy encoding and
2129 decoding of multiple identifier strings inside a D-Bus
2130 object path. Another new call sd_bus_default_flush_close()
2131 has been added to flush and close per-thread default
2134 * systemd-cgtop gained support for a -M/--machine= switch to
2135 show the control groups within a certain container only.
2137 * "systemctl kill" gained support for an optional --fail
2138 switch. If specified the requested operation will fail of no
2139 processes have been killed, because the unit had no
2140 processes attached, or similar.
2142 * A new systemd.crash_reboot=1 kernel command line option has
2143 been added that triggers a reboot after crashing. This can
2144 also be set through CrashReboot= in systemd.conf.
2146 * The RuntimeDirectory= setting now understands unit
2147 specifiers like %i or %f.
2149 * A new (still internal) library API sd-ipv4acd has been added,
2150 that implements address conflict detection for IPv4. It's
2151 based on code from sd-ipv4ll, and will be useful for
2152 detecting DHCP address conflicts.
2154 * File descriptors passed during socket activation may now be
2155 named. A new API sd_listen_fds_with_names() is added to
2156 access the names. The default names may be overridden,
2157 either in the .socket file using the FileDescriptorName=
2158 parameter, or by passing FDNAME= when storing the file
2159 descriptors using sd_notify().
2161 * systemd-networkd gained support for:
2163 - Setting the IPv6 Router Advertisement settings via
2164 IPv6AcceptRouterAdvertisements= in .network files.
2166 - Configuring the HelloTimeSec=, MaxAgeSec= and
2167 ForwardDelaySec= bridge parameters in .netdev files.
2169 - Configuring PreferredSource= for static routes in
2172 * The "ask-password" framework used to query for LUKS harddisk
2173 passwords or SSL passwords during boot gained support for
2174 caching passwords in the kernel keyring, if it is
2175 available. This makes sure that the user only has to type in
2176 a passphrase once if there are multiple objects to unlock
2177 with the same one. Previously, such password caching was
2178 available only when Plymouth was used; this moves the
2179 caching logic into the systemd codebase itself. The
2180 "systemd-ask-password" utility gained a new --keyname=
2181 switch to control which kernel keyring key to use for
2182 caching a password in. This functionality is also useful for
2183 enabling display managers such as gdm to automatically
2184 unlock the user's GNOME keyring if its passphrase, the
2185 user's password and the harddisk password are the same, if
2186 gdm-autologin is used.
2188 * When downloading tar or raw images using "machinectl
2189 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2190 file is now also downloaded, if it is available and stored
2191 next to the image file.
2193 * Units of type ".socket" gained a new boolean setting
2194 Writable= which is only useful in conjunction with
2195 ListenSpecial=. If true, enables opening the specified
2196 special file in O_RDWR mode rather than O_RDONLY mode.
2198 * systemd-rfkill has been reworked to become a singleton
2199 service that is activated through /dev/rfkill on each rfkill
2200 state change and saves the settings to disk. This way,
2201 systemd-rfkill is now compatible with devices that exist
2202 only intermittendly, and even restores state if the previous
2203 system shutdown was abrupt rather than clean.
2205 * The journal daemon gained support for vacuuming old journal
2206 files controlled by the number of files that shall remain,
2207 in addition to the already existing control by size and by
2208 date. This is useful as journal interleaving performance
2209 degrades with too many separate journal files, and allows
2210 putting an effective limit on them. The new setting defaults
2211 to 100, but this may be changed by setting SystemMaxFiles=
2212 and RuntimeMaxFiles= in journald.conf. Also, the
2213 "journalctl" tool gained the new --vacuum-files= switch to
2214 manually vacuum journal files to leave only the specified
2215 number of files in place.
2217 * udev will now create /dev/disk/by-path links for ATA devices
2218 on kernels where that is supported.
2220 * Galician, Serbian, Turkish and Korean translations were added.
2222 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2223 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2224 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2225 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2226 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2227 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2228 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2229 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2230 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2231 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2232 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2233 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2234 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2235 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2236 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2237 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2238 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2239 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2241 — Berlin, 2015-10-07
2245 * The DHCP implementation of systemd-networkd gained a set of
2248 - The DHCP server now supports emitting DNS and NTP
2249 information. It may be enabled and configured via
2250 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2251 and NTP information is enabled, but no servers are
2252 configured, the corresponding uplink information (if there
2253 is any) is propagated.
2255 - Server and client now support transmission and reception
2256 of timezone information. It can be configured via the
2257 newly introduced network options UseTimezone=,
2258 EmitTimezone=, and Timezone=. Transmission of timezone
2259 information is enabled between host and containers by
2260 default now: the container will change its local timezone
2261 to what the host has set.
2263 - Lease timeouts can now be configured via
2264 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2266 - The DHCP server improved on the stability of
2267 leases. Clients are more likely to get the same lease
2268 information back, even if the server loses state.
2270 - The DHCP server supports two new configuration options to
2271 control the lease address pool metrics, PoolOffset= and
2274 * The encapsulation limit of tunnels in systemd-networkd may
2275 now be configured via 'EncapsulationLimit='. It allows
2276 modifying the maximum additional levels of encapsulation
2277 that are permitted to be prepended to a packet.
2279 * systemd now supports the concept of user buses replacing
2280 session buses, if used with dbus-1.10 (and enabled via dbus
2281 --enable-user-session). It previously only supported this on
2282 kdbus-enabled systems, and this release expands this to
2283 'dbus-daemon' systems.
2285 * systemd-networkd now supports predictable interface names
2288 * systemd now optionally supports the new Linux kernel
2289 "unified" control group hierarchy. If enabled via the kernel
2290 command-line option 'systemd.unified_cgroup_hierarchy=1',
2291 systemd will try to mount the unified cgroup hierarchy
2292 directly on /sys/fs/cgroup. If not enabled, or not
2293 available, systemd will fall back to the legacy cgroup
2294 hierarchy setup, as before. Host system and containers can
2295 mix and match legacy and unified hierarchies as they
2296 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2297 environment variable to individually select the hierarchy to
2298 use for executed containers. By default, nspawn will use the
2299 unified hierarchy for the containers if the host uses the
2300 unified hierarchy, and the legacy hierarchy otherwise.
2301 Please note that at this point the unified hierarchy is an
2302 experimental kernel feature and is likely to change in one
2303 of the next kernel releases. Therefore, it should not be
2304 enabled by default in downstream distributions yet. The
2305 minimum required kernel version for the unified hierarchy to
2306 work is 4.2. Note that when the unified hierarchy is used
2307 for the first time delegated access to controllers is
2308 safe. Because of this systemd-nspawn containers will get
2309 access to controllers now, as will systemd user
2310 sessions. This means containers and user sessions may now
2311 manage their own resources, partitioning up what the system
2314 * A new special scope unit "init.scope" has been introduced
2315 that encapsulates PID 1 of the system. It may be used to
2316 determine resource usage and enforce resource limits on PID
2317 1 itself. PID 1 hence moved out of the root of the control
2320 * The cgtop tool gained support for filtering out kernel
2321 threads when counting tasks in a control group. Also, the
2322 count of processes is now recursively summed up by
2323 default. Two options -k and --recursive= have been added to
2324 revert to old behaviour. The tool has also been updated to
2325 work correctly in containers now.
2327 * systemd-nspawn's --bind= and --bind-ro= options have been
2328 extended to allow creation of non-recursive bind mounts.
2330 * libsystemd gained two new calls sd_pid_get_cgroup() and
2331 sd_peer_get_cgroup() which return the control group path of
2332 a process or peer of a connected AF_UNIX socket. This
2333 function call is particularly useful when implementing
2334 delegated subtrees support in the control group hierarchy.
2336 * The "sd-event" event loop API of libsystemd now supports
2337 correct dequeuing of real-time signals, without losing
2340 * When systemd requests a PolicyKit decision when managing
2341 units it will now add additional fields to the request,
2342 including unit name and desired operation. This enables more
2343 powerful PolicyKit policies, that make decisions depending
2344 on these parameters.
2346 * nspawn learnt support for .nspawn settings files, that may
2347 accompany the image files or directories of containers, and
2348 may contain additional settings for the container. This is
2349 an alternative to configuring container parameters via the
2350 nspawn command line.
2352 Contributions from: Cristian Rodríguez, Daniel Mack, David
2353 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2354 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2355 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2356 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2357 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2358 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2359 Andersen, Tom Gundersen, Torstein Husebø
2361 — Berlin, 2015-09-08
2365 * machinectl gained a new verb 'shell' which opens a fresh
2366 shell on the target container or the host. It is similar to
2367 the existing 'login' command of machinectl, but spawns the
2368 shell directly without prompting for username or
2369 password. The pseudo machine '.host' now refers to the local
2370 host and is used by default. Hence, 'machinectl shell' can
2371 be used as replacement for 'su -' which spawns a session as
2372 a fresh systemd unit in a way that is fully isolated from
2373 the originating session.
2375 * systemd-networkd learned to cope with private-zone DHCP
2376 options and allows other programs to query the values.
2378 * SELinux access control when enabling/disabling units is no
2379 longer enforced with this release. The previous
2380 implementation was incorrect, and a new corrected
2381 implementation is not yet available. As unit file operations
2382 are still protected via PolicyKit and D-Bus policy this is
2383 not a security problem. Yet, distributions which care about
2384 optimal SELinux support should probably not stabilize on
2387 * sd-bus gained support for matches of type "arg0has=", that
2388 test for membership of strings in string arrays sent in bus
2391 * systemd-resolved now dumps the contents of its DNS and LLMNR
2392 caches to the logs on reception of the SIGUSR1 signal. This
2393 is useful to debug DNS behaviour.
2395 * The coredumpctl tool gained a new --directory= option to
2396 operate on journal files in a specific directory.
2398 * "systemctl reboot" and related commands gained a new
2399 "--message=" option which may be used to set a free-text
2400 wall message when shutting down or rebooting the
2401 system. This message is also logged, which is useful for
2402 figuring out the reason for a reboot or shutdown a
2405 * The "systemd-resolve-host" tool's -i switch now takes
2406 network interface numbers as alternative to interface names.
2408 * A new unit file setting for services has been introduced:
2409 UtmpMode= allows configuration of how precisely systemd
2410 handles utmp and wtmp entries for the service if this is
2411 enabled. This allows writing services that appear similar to
2412 user sessions in the output of the "w", "who", "last" and
2415 * systemd-resolved will now locally synthesize DNS resource
2416 records for the "localhost" and "gateway" domains as well as
2417 the local hostname. This should ensure that clients querying
2418 RRs via resolved will get similar results as those going via
2419 NSS, if nss-myhostname is enabled.
2421 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2422 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2423 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2424 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2425 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2426 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2427 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2428 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2429 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2430 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2431 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2432 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2434 — Berlin, 2015-08-27
2438 * The systemd-efi-boot-generator functionality was merged into
2439 systemd-gpt-auto-generator.
2441 * systemd-networkd now supports Group Policy for vxlan
2442 devices. It can be enabled via the new boolean configuration
2443 option called 'GroupPolicyExtension='.
2445 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2446 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2447 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2449 — Berlin, 2015-07-31
2453 * The python-systemd code has been removed from the systemd repository.
2454 A new repository has been created which accommodates the code from
2455 now on, and we kindly ask distributions to create a separate package
2456 for this: https://github.com/systemd/python-systemd
2458 * The systemd daemon will now reload its main configuration
2459 (/etc/systemd/system.conf) on daemon-reload.
2461 * sd-dhcp now exposes vendor specific extensions via
2462 sd_dhcp_lease_get_vendor_specific().
2464 * systemd-networkd gained a number of new configuration options.
2466 - A new boolean configuration option for TAP devices called
2467 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2468 device, thus allowing to send and receive GSO packets.
2470 - A new tunnel configuration option called 'CopyDSCP='.
2471 If enabled, the DSCP field of ip6 tunnels is copied into the
2472 decapsulated packet.
2474 - A set of boolean bridge configuration options were added.
2475 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2476 and 'UnicastFlood=' are now parsed by networkd and applied to the
2477 respective bridge link device via the respective IFLA_BRPORT_*
2480 - A new string configuration option to override the hostname sent
2481 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2482 is true, networkd will use the configured hostname instead of the
2483 system hostname when sending DHCP requests.
2485 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2486 networkd will configure the IPv6 flow-label of the tunnel device
2487 according to RFC2460.
2489 - The 'macvtap' virtual network devices are now supported, similar to
2490 the already supported 'macvlan' devices.
2492 * systemd-resolved now implements RFC5452 to improve resilience against
2493 cache poisoning. Additionally, source port randomization is enabled
2494 by default to further protect against DNS spoofing attacks.
2496 * nss-mymachines now supports translating UIDs and GIDs of running
2497 containers with user-namespaces enabled. If a container 'foo'
2498 translates a host uid 'UID' to the container uid 'TUID', then
2499 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2500 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2501 mapped as 'vg-foo-TGID'.
2503 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2504 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2505 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2506 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2507 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2508 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2509 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2510 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2511 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2512 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2514 — Berlin, 2015-07-29
2518 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2519 There are no known issues with current sysfs, and udev does not need
2520 or should be used to work around such bugs.
2522 * udev does no longer enable USB HID power management. Several reports
2523 indicate, that some devices cannot handle that setting.
2525 * The udev accelerometer helper was removed. The functionality
2526 is now fully included in iio-sensor-proxy. But this means,
2527 older iio-sensor-proxy versions will no longer provide
2528 accelerometer/orientation data with this systemd version.
2529 Please upgrade iio-sensor-proxy to version 1.0.
2531 * networkd gained a new configuration option IPv6PrivacyExtensions=
2532 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2533 for Stateless Address") on selected networks.
2535 * For the sake of fewer build-time dependencies and less code in the
2536 main repository, the python bindings are about to be removed in the
2537 next release. A new repository has been created which accommodates
2538 the code from now on, and we kindly ask distributions to create a
2539 separate package for this. The removal will take place in v223.
2541 https://github.com/systemd/python-systemd
2543 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2544 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2545 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2546 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2547 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2548 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2549 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2550 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2551 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2552 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2554 — Berlin, 2015-07-07
2558 * The sd-bus.h and sd-event.h APIs have now been declared
2559 stable and have been added to the official interface of
2560 libsystemd.so. sd-bus implements an alternative D-Bus client
2561 library, that is relatively easy to use, very efficient and
2562 supports both classic D-Bus as well as kdbus as transport
2563 backend. sd-event is a generic event loop abstraction that
2564 is built around Linux epoll, but adds features such as event
2565 prioritization or efficient timer handling. Both APIs are good
2566 choices for C programs looking for a bus and/or event loop
2567 implementation that is minimal and does not have to be
2568 portable to other kernels.
2570 * kdbus support is no longer compile-time optional. It is now
2571 always built-in. However, it can still be disabled at
2572 runtime using the kdbus=0 kernel command line setting, and
2573 that setting may be changed to default to off, by specifying
2574 --disable-kdbus at build-time. Note though that the kernel
2575 command line setting has no effect if the kdbus.ko kernel
2576 module is not installed, in which case kdbus is (obviously)
2577 also disabled. We encourage all downstream distributions to
2578 begin testing kdbus by adding it to the kernel images in the
2579 development distributions, and leaving kdbus support in
2582 * The minimal required util-linux version has been bumped to
2585 * Support for chkconfig (--enable-chkconfig) was removed in
2586 favor of calling an abstraction tool
2587 /lib/systemd/systemd-sysv-install. This needs to be
2588 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2589 in README for details.
2591 * If there's a systemd unit and a SysV init script for the
2592 same service name, and the user executes "systemctl enable"
2593 for it (or a related call), then this will now enable both
2594 (or execute the related operation on both), not just the
2597 * The libudev API documentation has been converted from gtkdoc
2600 * gudev has been removed from the systemd tree, it is now an
2603 * The systemd-cgtop tool learnt a new --raw switch to generate
2604 "raw" (machine parsable) output.
2606 * networkd's IPForwarding= .network file setting learnt the
2607 new setting "kernel", which ensures that networkd does not
2608 change the IP forwarding sysctl from the default kernel
2611 * The systemd-logind bus API now exposes a new boolean
2612 property "Docked" that reports whether logind considers the
2613 system "docked", i.e. connected to a docking station or not.
2615 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2616 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2617 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2618 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2619 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2620 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2621 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2622 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2623 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2624 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2625 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2626 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2627 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2628 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2629 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2630 Fink, Zbigniew Jędrzejewski-Szmek
2632 — Berlin, 2015-06-19
2636 * The gudev library has been extracted into a separate repository
2637 available at: https://git.gnome.org/browse/libgudev/
2638 It is now managed as part of the Gnome project. Distributions
2639 are recommended to pass --disable-gudev to systemd and use
2640 gudev from the Gnome project instead. gudev is still included
2641 in systemd, for now. It will be removed soon, though. Please
2642 also see the announcement-thread on systemd-devel:
2643 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2645 * systemd now exposes a CPUUsageNSec= property for each
2646 service unit on the bus, that contains the overall consumed
2647 CPU time of a service (the sum of what each process of the
2648 service consumed). This value is only available if
2649 CPUAccounting= is turned on for a service, and is then shown
2650 in the "systemctl status" output.
2652 * Support for configuring alternative mappings of the old SysV
2653 runlevels to systemd targets has been removed. They are now
2654 hardcoded in a way that runlevels 2, 3, 4 all map to
2655 multi-user.target and 5 to graphical.target (which
2656 previously was already the default behaviour).
2658 * The auto-mounter logic gained support for mount point
2659 expiry, using a new TimeoutIdleSec= setting in .automount
2660 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2662 * The EFI System Partition (ESP) as mounted to /boot by
2663 systemd-efi-boot-generator will now be unmounted
2664 automatically after 2 minutes of not being used. This should
2665 minimize the risk of ESP corruptions.
2667 * New /etc/fstab options x-systemd.requires= and
2668 x-systemd.requires-mounts-for= are now supported to express
2669 additional dependencies for mounts. This is useful for
2670 journalling file systems that support external journal
2671 devices or overlay file systems that require underlying file
2672 systems to be mounted.
2674 * systemd does not support direct live-upgrades (via systemctl
2675 daemon-reexec) from versions older than v44 anymore. As no
2676 distribution we are aware of shipped such old versions in a
2677 stable release this should not be problematic.
2679 * When systemd forks off a new per-connection service instance
2680 it will now set the $REMOTE_ADDR environment variable to the
2681 remote IP address, and $REMOTE_PORT environment variable to
2682 the remote IP port. This behaviour is similar to the
2683 corresponding environment variables defined by CGI.
2685 * systemd-networkd gained support for uplink failure
2686 detection. The BindCarrier= option allows binding interface
2687 configuration dynamically to the link sense of other
2688 interfaces. This is useful to achieve behaviour like in
2691 * systemd-networkd gained support for configuring the DHCP
2692 client identifier to use when requesting leases.
2694 * systemd-networkd now has a per-network UseNTP= option to
2695 configure whether NTP server information acquired via DHCP
2696 is passed on to services like systemd-timesyncd.
2698 * systemd-networkd gained support for vti6 tunnels.
2700 * Note that systemd-networkd manages the sysctl variable
2701 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2702 it is configured for since v219. The variable controls IP
2703 forwarding, and is a per-interface alternative to the global
2704 /proc/sys/net/ipv[46]/ip_forward. This setting is
2705 configurable in the IPForward= option, which defaults to
2706 "no". This means if networkd is used for an interface it is
2707 no longer sufficient to set the global sysctl option to turn
2708 on IP forwarding! Instead, the .network file option
2709 IPForward= needs to be turned on! Note that the
2710 implementation of this behaviour was broken in v219 and has
2713 * Many bonding and vxlan options are now configurable in
2716 * systemd-nspawn gained a new --property= setting to set unit
2717 properties for the container scope. This is useful for
2718 setting resource parameters (e.g. "CPUShares=500") on
2719 containers started from the command line.
2721 * systemd-nspawn gained a new --private-users= switch to make
2722 use of user namespacing available on recent Linux kernels.
2724 * systemd-nspawn may now be called as part of a shell pipeline
2725 in which case the pipes used for stdin and stdout are passed
2726 directly to the process invoked in the container, without
2727 indirection via a pseudo tty.
2729 * systemd-nspawn gained a new switch to control the UNIX
2730 signal to use when killing the init process of the container
2733 * systemd-nspawn gained a new --overlay= switch for mounting
2734 overlay file systems into the container using the new kernel
2737 * When a container image is imported via systemd-importd and
2738 the host file system is not btrfs, a loopback block device
2739 file is created in /var/lib/machines.raw with a btrfs file
2740 system inside. It is then mounted to /var/lib/machines to
2741 enable btrfs features for container management. The loopback
2742 file and btrfs file system is grown as needed when container
2743 images are imported via systemd-importd.
2745 * systemd-machined/systemd-importd gained support for btrfs
2746 quota, to enforce container disk space limits on disk. This
2747 is exposed in "machinectl set-limit".
2749 * systemd-importd now can import containers from local .tar,
2750 .raw and .qcow2 images, and export them to .tar and .raw. It
2751 can also import dkr v2 images now from the network (on top
2754 * systemd-importd gained support for verifying downloaded
2755 images with gpg2 (previously only gpg1 was supported).
2757 * systemd-machined, systemd-logind, systemd: most bus calls
2758 are now accessible to unprivileged processes via
2759 PolicyKit. Also, systemd-logind will now allow users to kill
2760 their own sessions without further privileges or
2763 * systemd-shutdownd has been removed. This service was
2764 previously responsible for implementing scheduled shutdowns
2765 as exposed in /usr/bin/shutdown's time parameter. This
2766 functionality has now been moved into systemd-logind and is
2767 accessible via a bus interface.
2769 * "systemctl reboot" gained a new switch --firmware-setup that
2770 can be used to reboot into the EFI firmware setup, if that
2771 is available. systemd-logind now exposes an API on the bus
2772 to trigger such reboots, in case graphical desktop UIs want
2773 to cover this functionality.
2775 * "systemctl enable", "systemctl disable" and "systemctl mask"
2776 now support a new "--now" switch. If specified the units
2777 that are enabled will also be started, and the ones
2778 disabled/masked also stopped.
2780 * The Gummiboot EFI boot loader tool has been merged into
2781 systemd, and renamed to "systemd-boot". The bootctl tool has been
2782 updated to support systemd-boot.
2784 * An EFI kernel stub has been added that may be used to create
2785 kernel EFI binaries that contain not only the actual kernel,
2786 but also an initrd, boot splash, command line and OS release
2787 information. This combined binary can then be signed as a
2788 single image, so that the firmware can verify it all in one
2789 step. systemd-boot has special support for EFI binaries created
2790 like this and can extract OS release information from them
2791 and show them in the boot menu. This functionality is useful
2792 to implement cryptographically verified boot schemes.
2794 * Optional support has been added to systemd-fsck to pass
2795 fsck's progress report to an AF_UNIX socket in the file
2798 * udev will no longer create device symlinks for all block
2799 devices by default. A blacklist for excluding special block
2800 devices from this logic has been turned into a whitelist
2801 that requires picking block devices explicitly that require
2804 * A new (currently still internal) API sd-device.h has been
2805 added to libsystemd. This modernized API is supposed to
2806 replace libudev eventually. In fact, already much of libudev
2807 is now just a wrapper around sd-device.h.
2809 * A new hwdb database for storing metadata about pointing
2810 stick devices has been added.
2812 * systemd-tmpfiles gained support for setting file attributes
2813 similar to the "chattr" tool with new 'h' and 'H' lines.
2815 * systemd-journald will no longer unconditionally set the
2816 btrfs NOCOW flag on new journal files. This is instead done
2817 with tmpfiles snippet using the new 'h' line type. This
2818 allows easy disabling of this logic, by masking the
2819 journal-nocow.conf tmpfiles file.
2821 * systemd-journald will now translate audit message types to
2822 human readable identifiers when writing them to the
2823 journal. This should improve readability of audit messages.
2825 * The LUKS logic gained support for the offset= and skip=
2826 options in /etc/crypttab, as previously implemented by
2829 * /usr/lib/os-release gained a new optional field VARIANT= for
2830 distributions that support multiple variants (such as a
2831 desktop edition, a server edition, ...)
2833 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2834 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2835 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2836 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2837 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2838 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2839 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2840 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2841 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2842 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2843 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2844 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2845 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2846 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2847 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2848 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2849 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2850 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2851 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2852 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2853 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2854 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2855 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2856 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2857 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2858 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2859 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2861 — Berlin, 2015-05-22
2865 * Introduce a new API "sd-hwdb.h" for querying the hardware
2866 metadata database. With this minimal interface one can query
2867 and enumerate the udev hwdb, decoupled from the old libudev
2868 library. libudev's interface for this is now only a wrapper
2869 around sd-hwdb. A new tool systemd-hwdb has been added to
2870 interface with and update the database.
2872 * When any of systemd's tools copies files (for example due to
2873 tmpfiles' C lines) a btrfs reflink will attempted first,
2874 before bytewise copying is done.
2876 * systemd-nspawn gained a new --ephemeral switch. When
2877 specified a btrfs snapshot is taken of the container's root
2878 directory, and immediately removed when the container
2879 terminates again. Thus, a container can be started whose
2880 changes never alter the container's root directory, and are
2881 lost on container termination. This switch can also be used
2882 for starting a container off the root file system of the
2883 host without affecting the host OS. This switch is only
2884 available on btrfs file systems.
2886 * systemd-nspawn gained a new --template= switch. It takes the
2887 path to a container tree to use as template for the tree
2888 specified via --directory=, should that directory be
2889 missing. This allows instantiating containers dynamically,
2890 on first run. This switch is only available on btrfs file
2893 * When a .mount unit refers to a mount point on which multiple
2894 mounts are stacked, and the .mount unit is stopped all of
2895 the stacked mount points will now be unmounted until no
2896 mount point remains.
2898 * systemd now has an explicit notion of supported and
2899 unsupported unit types. Jobs enqueued for unsupported unit
2900 types will now fail with an "unsupported" error code. More
2901 specifically .swap, .automount and .device units are not
2902 supported in containers, .busname units are not supported on
2903 non-kdbus systems. .swap and .automount are also not
2904 supported if their respective kernel compile time options
2907 * machinectl gained support for two new "copy-from" and
2908 "copy-to" commands for copying files from a running
2909 container to the host or vice versa.
2911 * machinectl gained support for a new "bind" command to bind
2912 mount host directories into local containers. This is
2913 currently only supported for nspawn containers.
2915 * networkd gained support for configuring bridge forwarding
2916 database entries (fdb) from .network files.
2918 * A new tiny daemon "systemd-importd" has been added that can
2919 download container images in tar, raw, qcow2 or dkr formats,
2920 and make them available locally in /var/lib/machines, so
2921 that they can run as nspawn containers. The daemon can GPG
2922 verify the downloads (not supported for dkr, since it has no
2923 provisions for verifying downloads). It will transparently
2924 decompress bz2, xz, gzip compressed downloads if necessary,
2925 and restore sparse files on disk. The daemon uses privilege
2926 separation to ensure the actual download logic runs with
2927 fewer privileges than the daemon itself. machinectl has
2928 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2929 make the functionality of importd available to the
2930 user. With this in place the Fedora and Ubuntu "Cloud"
2931 images can be downloaded and booted as containers unmodified
2932 (the Fedora images lack the appropriate GPG signature files
2933 currently, so they cannot be verified, but this will change
2934 soon, hopefully). Note that downloading images is currently
2935 only fully supported on btrfs.
2937 * machinectl is now able to list container images found in
2938 /var/lib/machines, along with some metadata about sizes of
2939 disk and similar. If the directory is located on btrfs and
2940 quota is enabled, this includes quota display. A new command
2941 "image-status" has been added that shows additional
2942 information about images.
2944 * machinectl is now able to clone container images
2945 efficiently, if the underlying file system (btrfs) supports
2946 it, with the new "machinectl clone" command. It also
2947 gained commands for renaming and removing images, as well as
2948 marking them read-only or read-write (supported also on
2949 legacy file systems).
2951 * networkd gained support for collecting LLDP network
2952 announcements, from hardware that supports this. This is
2953 shown in networkctl output.
2955 * systemd-run gained support for a new -t (--pty) switch for
2956 invoking a binary on a pty whose input and output is
2957 connected to the invoking terminal. This allows executing
2958 processes as system services while interactively
2959 communicating with them via the terminal. Most interestingly
2960 this is supported across container boundaries. Invoking
2961 "systemd-run -t /bin/bash" is an alternative to running a
2962 full login session, the difference being that the former
2963 will not register a session, nor go through the PAM session
2966 * tmpfiles gained support for a new "v" line type for creating
2967 btrfs subvolumes. If the underlying file system is a legacy
2968 file system, this automatically degrades to creating a
2969 normal directory. Among others /var/lib/machines is now
2970 created like this at boot, should it be missing.
2972 * The directory /var/lib/containers/ has been deprecated and
2973 been replaced by /var/lib/machines. The term "machines" has
2974 been used in the systemd context as generic term for both
2975 VMs and containers, and hence appears more appropriate for
2976 this, as the directory can also contain raw images bootable
2979 * systemd-nspawn when invoked with -M but without --directory=
2980 or --image= is now capable of searching for the container
2981 root directory, subvolume or disk image automatically, in
2982 /var/lib/machines. systemd-nspawn@.service has been updated
2983 to make use of this, thus allowing it to be used for raw
2986 * A new machines.target unit has been introduced that is
2987 supposed to group all containers/VMs invoked as services on
2988 the system. systemd-nspawn@.service has been updated to
2989 integrate with that.
2991 * machinectl gained a new "start" command, for invoking a
2992 container as a service. "machinectl start foo" is mostly
2993 equivalent to "systemctl start systemd-nspawn@foo.service",
2994 but handles escaping in a nicer way.
2996 * systemd-nspawn will now mount most of the cgroupfs tree
2997 read-only into each container, with the exception of the
2998 container's own subtree in the name=systemd hierarchy.
3000 * journald now sets the special FS_NOCOW file flag for its
3001 journal files. This should improve performance on btrfs, by
3002 avoiding heavy fragmentation when journald's write-pattern
3003 is used on COW file systems. It degrades btrfs' data
3004 integrity guarantees for the files to the same levels as for
3005 ext3/ext4 however. This should be OK though as journald does
3006 its own data integrity checks and all its objects are
3007 checksummed on disk. Also, journald should handle btrfs disk
3008 full events a lot more gracefully now, by processing SIGBUS
3009 errors, and not relying on fallocate() anymore.
3011 * When journald detects that journal files it is writing to
3012 have been deleted it will immediately start new journal
3015 * systemd now provides a way to store file descriptors
3016 per-service in PID 1. This is useful for daemons to ensure
3017 that fds they require are not lost during a daemon
3018 restart. The fds are passed to the daemon on the next
3019 invocation in the same way socket activation fds are
3020 passed. This is now used by journald to ensure that the
3021 various sockets connected to all the system's stdout/stderr
3022 are not lost when journald is restarted. File descriptors
3023 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3024 an extension to sd_notify(). Note that a limit is enforced
3025 on the number of fds a service can store in PID 1, and it
3026 defaults to 0, so that no fds may be stored, unless this is
3027 explicitly turned on.
3029 * The default TERM variable to use for units connected to a
3030 terminal, when no other value is explicitly is set is now
3031 vt220 rather than vt102. This should be fairly safe still,
3032 but allows PgUp/PgDn work.
3034 * The /etc/crypttab option header= as known from Debian is now
3037 * "loginctl user-status" and "loginctl session-status" will
3038 now show the last 10 lines of log messages of the
3039 user/session following the status output. Similar,
3040 "machinectl status" will show the last 10 log lines
3041 associated with a virtual machine or container
3042 service. (Note that this is usually not the log messages
3043 done in the VM/container itself, but simply what the
3044 container manager logs. For nspawn this includes all console
3047 * "loginctl session-status" without further argument will now
3048 show the status of the session of the caller. Similar,
3049 "lock-session", "unlock-session", "activate",
3050 "enable-linger", "disable-linger" may now be called without
3051 session/user parameter in which case they apply to the
3052 caller's session/user.
3054 * An X11 session scriptlet is now shipped that uploads
3055 $DISPLAY and $XAUTHORITY into the environment of the systemd
3056 --user daemon if a session begins. This should improve
3057 compatibility with X11 enabled applications run as systemd
3060 * Generators are now subject to masking via /etc and /run, the
3061 same way as unit files.
3063 * networkd .network files gained support for configuring
3064 per-link IPv4/IPv6 packet forwarding as well as IPv4
3065 masquerading. This is by default turned on for veth links to
3066 containers, as registered by systemd-nspawn. This means that
3067 nspawn containers run with --network-veth will now get
3068 automatic routed access to the host's networks without any
3069 further configuration or setup, as long as networkd runs on
3072 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3073 or UDP posts of a container on the host. With this in place
3074 it is possible to run containers with private veth links
3075 (--network-veth), and have their functionality exposed on
3076 the host as if their services were running directly on the
3079 * systemd-nspawn's --network-veth switch now gained a short
3080 version "-n", since with the changes above it is now truly
3081 useful out-of-the-box. The systemd-nspawn@.service has been
3082 updated to make use of it too by default.
3084 * systemd-nspawn will now maintain a per-image R/W lock, to
3085 ensure that the same image is not started more than once
3086 writable. (It's OK to run an image multiple times
3087 simultaneously in read-only mode.)
3089 * systemd-nspawn's --image= option is now capable of
3090 dissecting and booting MBR and GPT disk images that contain
3091 only a single active Linux partition. Previously it
3092 supported only GPT disk images with proper GPT type
3093 IDs. This allows running cloud images from major
3094 distributions directly with systemd-nspawn, without
3097 * In addition to collecting mouse dpi data in the udev
3098 hardware database, there's now support for collecting angle
3099 information for mouse scroll wheels. The database is
3100 supposed to guarantee similar scrolling behavior on mice
3101 that it knows about. There's also support for collecting
3102 information about Touchpad types.
3104 * udev's input_id built-in will now also collect touch screen
3105 dimension data and attach it to probed devices.
3107 * /etc/os-release gained support for a Distribution Privacy
3110 * networkd gained support for creating "ipvlan", "gretap",
3111 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3113 * systemd-tmpfiles gained support for "a" lines for setting
3116 * systemd-nspawn will now mount /tmp in the container to
3117 tmpfs, automatically.
3119 * systemd now exposes the memory.usage_in_bytes cgroup
3120 attribute and shows it for each service in the "systemctl
3121 status" output, if available.
3123 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3124 immediate reboot is triggered. This useful if shutdown is
3125 hung and is unable to complete, to expedite the
3126 operation. Note that this kind of reboot will still unmount
3127 all file systems, and hence should not result in fsck being
3130 * A .device unit for an optical block device will now be
3131 considered active only when a medium is in the drive. Also,
3132 mount units are now bound to their backing devices thus
3133 triggering automatic unmounting when devices become
3134 unavailable. With this in place systemd will now
3135 automatically unmount left-over mounts when a CD-ROM is
3136 ejected or an USB stick is yanked from the system.
3138 * networkd-wait-online now has support for waiting for
3139 specific interfaces only (with globbing), and for giving up
3140 after a configurable timeout.
3142 * networkd now exits when idle. It will be automatically
3143 restarted as soon as interfaces show up, are removed or
3144 change state. networkd will stay around as long as there is
3145 at least one DHCP state machine or similar around, that keep
3148 * networkd may now configure IPv6 link-local addressing in
3149 addition to IPv4 link-local addressing.
3151 * The IPv6 "token" for use in SLAAC may now be configured for
3152 each .network interface in networkd.
3154 * Routes configured with networkd may now be assigned a scope
3157 * networkd's [Match] sections now support globbing and lists
3158 of multiple space-separated matches per item.
3160 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3161 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3162 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3163 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3164 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3165 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3166 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3167 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3168 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3169 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3170 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3171 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3172 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3173 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3174 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3175 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3176 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3177 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3178 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3179 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3180 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3181 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3182 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3183 Hoffmann, Zbigniew Jędrzejewski-Szmek
3185 — Berlin, 2015-02-16
3189 * When querying unit file enablement status (for example via
3190 "systemctl is-enabled"), a new state "indirect" is now known
3191 which indicates that a unit might not be enabled itself, but
3192 another unit listed in its Also= setting might be.
3194 * Similar to the various existing ConditionXYZ= settings for
3195 units, there are now matching AssertXYZ= settings. While
3196 failing conditions cause a unit to be skipped, but its job
3197 to succeed, failing assertions declared like this will cause
3198 a unit start operation and its job to fail.
3200 * hostnamed now knows a new chassis type "embedded".
3202 * systemctl gained a new "edit" command. When used on a unit
3203 file, this allows extending unit files with .d/ drop-in
3204 configuration snippets or editing the full file (after
3205 copying it from /usr/lib to /etc). This will invoke the
3206 user's editor (as configured with $EDITOR), and reload the
3207 modified configuration after editing.
3209 * "systemctl status" now shows the suggested enablement state
3210 for a unit, as declared in the (usually vendor-supplied)
3211 system preset files.
3213 * nss-myhostname will now resolve the single-label host name
3214 "gateway" to the locally configured default IP routing
3215 gateways, ordered by their metrics. This assigns a stable
3216 name to the used gateways, regardless which ones are
3217 currently configured. Note that the name will only be
3218 resolved after all other name sources (if nss-myhostname is
3219 configured properly) and should hence not negatively impact
3220 systems that use the single-label host name "gateway" in
3223 * systemd-inhibit now allows filtering by mode when listing
3226 * Scope and service units gained a new "Delegate" boolean
3227 property, which, when set, allows processes running inside the
3228 unit to further partition resources. This is primarily
3229 useful for systemd user instances as well as container
3232 * journald will now pick up audit messages directly from
3233 the kernel, and log them like any other log message. The
3234 audit fields are split up and fully indexed. This means that
3235 journalctl in many ways is now a (nicer!) alternative to
3236 ausearch, the traditional audit client. Note that this
3237 implements only a minimal audit client. If you want the
3238 special audit modes like reboot-on-log-overflow, please use
3239 the traditional auditd instead, which can be used in
3240 parallel to journald.
3242 * The ConditionSecurity= unit file option now understands the
3243 special string "audit" to check whether auditing is
3246 * journalctl gained two new commands --vacuum-size= and
3247 --vacuum-time= to delete old journal files until the
3248 remaining ones take up no more than the specified size on disk,
3249 or are not older than the specified time.
3251 * A new, native PPPoE library has been added to sd-network,
3252 systemd's library of light-weight networking protocols. This
3253 library will be used in a future version of networkd to
3254 enable PPPoE communication without an external pppd daemon.
3256 * The busctl tool now understands a new "capture" verb that
3257 works similar to "monitor", but writes a packet capture
3258 trace to STDOUT that can be redirected to a file which is
3259 compatible with libcap's capture file format. This can then
3260 be loaded in Wireshark and similar tools to inspect bus
3263 * The busctl tool now understands a new "tree" verb that shows
3264 the object trees of a specific service on the bus, or of all
3267 * The busctl tool now understands a new "introspect" verb that
3268 shows all interfaces and members of objects on the bus,
3269 including their signature and values. This is particularly
3270 useful to get more information about bus objects shown by
3271 the new "busctl tree" command.
3273 * The busctl tool now understands new verbs "call",
3274 "set-property" and "get-property" for invoking bus method
3275 calls, setting and getting bus object properties in a
3278 * busctl gained a new --augment-creds= argument that controls
3279 whether the tool shall augment credential information it
3280 gets from the bus with data from /proc, in a possibly
3283 * nspawn's --link-journal= switch gained two new values
3284 "try-guest" and "try-host" that work like "guest" and
3285 "host", but do not fail if the host has no persistent
3286 journalling enabled. -j is now equivalent to
3287 --link-journal=try-guest.
3289 * macvlan network devices created by nspawn will now have
3290 stable MAC addresses.
3292 * A new SmackProcessLabel= unit setting has been added, which
3293 controls the SMACK security label processes forked off by
3294 the respective unit shall use.
3296 * If compiled with --enable-xkbcommon, systemd-localed will
3297 verify x11 keymap settings by compiling the given keymap. It
3298 will spew out warnings if the compilation fails. This
3299 requires libxkbcommon to be installed.
3301 * When a coredump is collected, a larger number of metadata
3302 fields is now collected and included in the journal records
3303 created for it. More specifically, control group membership,
3304 environment variables, memory maps, working directory,
3305 chroot directory, /proc/$PID/status, and a list of open file
3306 descriptors is now stored in the log entry.
3308 * The udev hwdb now contains DPI information for mice. For
3311 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3313 * All systemd programs that read standalone configuration
3314 files in /etc now also support a corresponding series of
3315 .conf.d configuration directories in /etc/, /run/,
3316 /usr/local/lib/, /usr/lib/, and (if configured with
3317 --enable-split-usr) /lib/. In particular, the following
3318 configuration files now have corresponding configuration
3319 directories: system.conf user.conf, logind.conf,
3320 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3321 resolved.conf, timesyncd.conf, journal-remote.conf, and
3322 journal-upload.conf. Note that distributions should use the
3323 configuration directories in /usr/lib/; the directories in
3324 /etc/ are reserved for the system administrator.
3326 * systemd-rfkill will no longer take the rfkill device name
3327 into account when storing rfkill state on disk, as the name
3328 might be dynamically assigned and not stable. Instead, the
3329 ID_PATH udev variable combined with the rfkill type (wlan,
3330 bluetooth, ...) is used.
3332 * A new service systemd-machine-id-commit.service has been
3333 added. When used on systems where /etc is read-only during
3334 boot, and /etc/machine-id is not initialized (but an empty
3335 file), this service will copy the temporary machine ID
3336 created as replacement into /etc after the system is fully
3337 booted up. This is useful for systems that are freshly
3338 installed with a non-initialized machine ID, but should get
3339 a fixed machine ID for subsequent boots.
3341 * networkd's .netdev files now provide a large set of
3342 configuration parameters for VXLAN devices. Similarly, the
3343 bridge port cost parameter is now configurable in .network
3344 files. There's also new support for configuring IP source
3345 routing. networkd .link files gained support for a new
3346 OriginalName= match that is useful to match against the
3347 original interface name the kernel assigned. .network files
3348 may include MTU= and MACAddress= fields for altering the MTU
3349 and MAC address while being connected to a specific network
3352 * The LUKS logic gained supported for configuring
3353 UUID-specific key files. There's also new support for naming
3354 LUKS device from the kernel command line, using the new
3355 luks.name= argument.
3357 * Timer units may now be transiently created via the bus API
3358 (this was previously already available for scope and service
3359 units). In addition it is now possible to create multiple
3360 transient units at the same time with a single bus call. The
3361 "systemd-run" tool has been updated to make use of this for
3362 running commands on a specified time, in at(1)-style.
3364 * tmpfiles gained support for "t" lines, for assigning
3365 extended attributes to files. Among other uses this may be
3366 used to assign SMACK labels to files.
3368 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3369 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3370 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3371 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3372 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3373 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3374 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3375 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3376 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3377 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3378 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3379 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3380 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3381 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3382 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3383 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3384 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3385 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3387 — Berlin, 2014-12-10
3391 * journalctl gained the new options -t/--identifier= to match
3392 on the syslog identifier (aka "tag"), as well as --utc to
3393 show log timestamps in the UTC timezone. journalctl now also
3394 accepts -n/--lines=all to disable line capping in a pager.
3396 * journalctl gained a new switch, --flush, that synchronously
3397 flushes logs from /run/log/journal to /var/log/journal if
3398 persistent storage is enabled. systemd-journal-flush.service
3399 now waits until the operation is complete.
3401 * Services can notify the manager before they start a reload
3402 (by sending RELOADING=1) or shutdown (by sending
3403 STOPPING=1). This allows the manager to track and show the
3404 internal state of daemons and closes a race condition when
3405 the process is still running but has closed its D-Bus
3408 * Services with Type=oneshot do not have to have any ExecStart
3411 * User units are now loaded also from
3412 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3413 /run/systemd/user directory that was already previously
3414 supported, but is under the control of the user.
3416 * Job timeouts (i.e. time-outs on the time a job that is
3417 queued stays in the run queue) can now optionally result in
3418 immediate reboot or power-off actions (JobTimeoutAction= and
3419 JobTimeoutRebootArgument=). This is useful on ".target"
3420 units, to limit the maximum time a target remains
3421 undispatched in the run queue, and to trigger an emergency
3422 operation in such a case. This is now used by default to
3423 turn off the system if boot-up (as defined by everything in
3424 basic.target) hangs and does not complete for at least
3425 15min. Also, if power-off or reboot hang for at least 30min
3426 an immediate power-off/reboot operation is triggered. This
3427 functionality is particularly useful to increase reliability
3428 on embedded devices, but also on laptops which might
3429 accidentally get powered on when carried in a backpack and
3430 whose boot stays stuck in a hard disk encryption passphrase
3433 * systemd-logind can be configured to also handle lid switch
3434 events even when the machine is docked or multiple displays
3435 are attached (HandleLidSwitchDocked= option).
3437 * A helper binary and a service have been added which can be
3438 used to resume from hibernation in the initramfs. A
3439 generator will parse the resume= option on the kernel
3440 command line to trigger resume.
3442 * A user console daemon systemd-consoled has been
3443 added. Currently, it is a preview, and will so far open a
3444 single terminal on each session of the user marked as
3445 Desktop=systemd-console.
3447 * Route metrics can be specified for DHCP routes added by
3450 * The SELinux context of socket-activated services can be set
3451 from the information provided by the networking stack
3452 (SELinuxContextFromNet= option).
3454 * Userspace firmware loading support has been removed and
3455 the minimum supported kernel version is thus bumped to 3.7.
3457 * Timeout for udev workers has been increased from 1 to 3
3458 minutes, but a warning will be printed after 1 minute to
3459 help diagnose kernel modules that take a long time to load.
3461 * Udev rules can now remove tags on devices with TAG-="foobar".
3463 * systemd's readahead implementation has been removed. In many
3464 circumstances it didn't give expected benefits even for
3465 rotational disk drives and was becoming less relevant in the
3466 age of SSDs. As none of the developers has been using
3467 rotating media anymore, and nobody stepped up to actively
3468 maintain this component of systemd it has now been removed.
3470 * Swap units can use Options= to specify discard options.
3471 Discard options specified for swaps in /etc/fstab are now
3474 * Docker containers are now detected as a separate type of
3477 * The Password Agent protocol gained support for queries where
3478 the user input is shown, useful e.g. for user names.
3479 systemd-ask-password gained a new --echo option to turn that
3482 * The default sysctl.d/ snippets will now set:
3484 net.core.default_qdisc = fq_codel
3486 This selects Fair Queuing Controlled Delay as the default
3487 queuing discipline for network interfaces. fq_codel helps
3488 fight the network bufferbloat problem. It is believed to be
3489 a good default with no tuning required for most workloads.
3490 Downstream distributions may override this choice. On 10Gbit
3491 servers that do not do forwarding, "fq" may perform better.
3492 Systems without a good clocksource should use "pfifo_fast".
3494 * If kdbus is enabled during build a new option BusPolicy= is
3495 available for service units, that allows locking all service
3496 processes into a stricter bus policy, in order to limit
3497 access to various bus services, or even hide most of them
3498 from the service's view entirely.
3500 * networkctl will now show the .network and .link file
3501 networkd has applied to a specific interface.
3503 * sd-login gained a new API call sd_session_get_desktop() to
3504 query which desktop environment has been selected for a
3507 * UNIX utmp support is now compile-time optional to support
3508 legacy-free systems.
3510 * systemctl gained two new commands "add-wants" and
3511 "add-requires" for pulling in units from specific targets
3514 * If the word "rescue" is specified on the kernel command line
3515 the system will now boot into rescue mode (aka
3516 rescue.target), which was previously available only by
3517 specifying "1" or "systemd.unit=rescue.target" on the kernel
3518 command line. This new kernel command line option nicely
3519 mirrors the already existing "emergency" kernel command line
3522 * New kernel command line options mount.usr=, mount.usrflags=,
3523 mount.usrfstype= have been added that match root=, rootflags=,
3524 rootfstype= but allow mounting a specific file system to
3527 * The $NOTIFY_SOCKET is now also passed to control processes of
3528 services, not only the main process.
3530 * This version reenables support for fsck's -l switch. This
3531 means at least version v2.25 of util-linux is required for
3532 operation, otherwise dead-locks on device nodes may
3533 occur. Again: you need to update util-linux to at least
3534 v2.25 when updating systemd to v217.
3536 * The "multi-seat-x" tool has been removed from systemd, as
3537 its functionality has been integrated into X servers 1.16,
3538 and the tool is hence redundant. It is recommended to update
3539 display managers invoking this tool to simply invoke X
3540 directly from now on, again.
3542 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3543 message flag has been added for all of systemd's PolicyKit
3544 authenticated method calls has been added. In particular
3545 this now allows optional interactive authorization via
3546 PolicyKit for many of PID1's privileged operations such as
3547 unit file enabling and disabling.
3549 * "udevadm hwdb --update" learnt a new switch "--usr" for
3550 placing the rebuilt hardware database in /usr instead of
3551 /etc. When used only hardware database entries stored in
3552 /usr will be used, and any user database entries in /etc are
3553 ignored. This functionality is useful for vendors to ship a
3554 pre-built database on systems where local configuration is
3555 unnecessary or unlikely.
3557 * Calendar time specifications in .timer units now also
3558 understand the strings "semi-annually", "quarterly" and
3559 "minutely" as shortcuts (in addition to the preexisting
3560 "anually", "hourly", ...).
3562 * systemd-tmpfiles will now correctly create files in /dev
3563 at boot which are marked for creation only at boot. It is
3564 recommended to always create static device nodes with 'c!'
3565 and 'b!', so that they are created only at boot and not
3566 overwritten at runtime.
3568 * When the watchdog logic is used for a service (WatchdogSec=)
3569 and the watchdog timeout is hit the service will now be
3570 terminated with SIGABRT (instead of just SIGTERM), in order
3571 to make sure a proper coredump and backtrace is
3572 generated. This ensures that hanging services will result in
3573 similar coredump/backtrace behaviour as services that hit a
3576 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3577 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3578 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3579 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3580 Herrmann, David Sommerseth, David Strauss, Emil Renner
3581 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3582 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3583 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3584 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3585 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3586 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3587 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3588 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3589 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3590 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3591 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3592 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3593 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3594 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3595 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3596 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3599 — Berlin, 2014-10-28
3603 * timedated no longer reads NTP implementation unit names from
3604 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3605 implementations should add a
3607 Conflicts=systemd-timesyncd.service
3609 to their unit files to take over and replace systemd's NTP
3610 default functionality.
3612 * systemd-sysusers gained a new line type "r" for configuring
3613 which UID/GID ranges to allocate system users/groups
3614 from. Lines of type "u" may now add an additional column
3615 that specifies the home directory for the system user to be
3616 created. Also, systemd-sysusers may now optionally read user
3617 information from STDIN instead of a file. This is useful for
3618 invoking it from RPM preinst scriptlets that need to create
3619 users before the first RPM file is installed since these
3620 files might need to be owned by them. A new
3621 %sysusers_create_inline RPM macro has been introduced to do
3622 just that. systemd-sysusers now updates the shadow files as
3623 well as the user/group databases, which should enhance
3624 compatibility with certain tools like grpck.
3626 * A number of bus APIs of PID 1 now optionally consult
3627 PolicyKit to permit access for otherwise unprivileged
3628 clients under certain conditions. Note that this currently
3629 doesn't support interactive authentication yet, but this is
3630 expected to be added eventually, too.
3632 * /etc/machine-info now has new fields for configuring the
3633 deployment environment of the machine, as well as the
3634 location of the machine. hostnamectl has been updated with
3635 new command to update these fields.
3637 * systemd-timesyncd has been updated to automatically acquire
3638 NTP server information from systemd-networkd, which might
3639 have been discovered via DHCP.
3641 * systemd-resolved now includes a caching DNS stub resolver
3642 and a complete LLMNR name resolution implementation. A new
3643 NSS module "nss-resolve" has been added which can be used
3644 instead of glibc's own "nss-dns" to resolve hostnames via
3645 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3646 be resolved via systemd-resolved D-Bus APIs. In contrast to
3647 the glibc internal resolver systemd-resolved is aware of
3648 multi-homed system, and keeps DNS server and caches separate
3649 and per-interface. Queries are sent simultaneously on all
3650 interfaces that have DNS servers configured, in order to
3651 properly handle VPNs and local LANs which might resolve
3652 separate sets of domain names. systemd-resolved may acquire
3653 DNS server information from systemd-networkd automatically,
3654 which in turn might have discovered them via DHCP. A tool
3655 "systemd-resolve-host" has been added that may be used to
3656 query the DNS logic in resolved. systemd-resolved implements
3657 IDNA and automatically uses IDNA or UTF-8 encoding depending
3658 on whether classic DNS or LLMNR is used as transport. In the
3659 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3660 implementation to systemd-resolved.
3662 * A new NSS module nss-mymachines has been added, that
3663 automatically resolves the names of all local registered
3664 containers to their respective IP addresses.
3666 * A new client tool "networkctl" for systemd-networkd has been
3667 added. It currently is entirely passive and will query
3668 networking configuration from udev, rtnetlink and networkd,
3669 and present it to the user in a very friendly
3670 way. Eventually, we hope to extend it to become a full
3671 control utility for networkd.
3673 * .socket units gained a new DeferAcceptSec= setting that
3674 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3675 TCP. Similarly, support for controlling TCP keep-alive
3676 settings has been added (KeepAliveTimeSec=,
3677 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3678 turning off Nagle's algorithm on TCP has been added
3681 * logind learned a new session type "web", for use in projects
3682 like Cockpit which register web clients as PAM sessions.
3684 * timer units with at least one OnCalendar= setting will now
3685 be started only after time-sync.target has been
3686 reached. This way they will not elapse before the system
3687 clock has been corrected by a local NTP client or
3688 similar. This is particular useful on RTC-less embedded
3689 machines, that come up with an invalid system clock.
3691 * systemd-nspawn's --network-veth= switch should now result in
3692 stable MAC addresses for both the outer and the inner side
3695 * systemd-nspawn gained a new --volatile= switch for running
3696 container instances with /etc or /var unpopulated.
3698 * The kdbus client code has been updated to use the new Linux
3699 3.17 memfd subsystem instead of the old kdbus-specific one.
3701 * systemd-networkd's DHCP client and server now support
3702 FORCERENEW. There are also new configuration options to
3703 configure the vendor client identifier and broadcast mode
3706 * systemd will no longer inform the kernel about the current
3707 timezone, as this is necessarily incorrect and racy as the
3708 kernel has no understanding of DST and similar
3709 concepts. This hence means FAT timestamps will be always
3710 considered UTC, similar to what Android is already
3711 doing. Also, when the RTC is configured to the local time
3712 (rather than UTC) systemd will never synchronize back to it,
3713 as this might confuse Windows at a later boot.
3715 * systemd-analyze gained a new command "verify" for offline
3716 validation of unit files.
3718 * systemd-networkd gained support for a couple of additional
3719 settings for bonding networking setups. Also, the metric for
3720 statically configured routes may now be configured. For
3721 network interfaces where this is appropriate the peer IP
3722 address may now be configured.
3724 * systemd-networkd's DHCP client will no longer request
3725 broadcasting by default, as this tripped up some networks.
3726 For hardware where broadcast is required the feature should
3727 be switched back on using RequestBroadcast=yes.
3729 * systemd-networkd will now set up IPv4LL addresses (when
3730 enabled) even if DHCP is configured successfully.
3732 * udev will now default to respect network device names given
3733 by the kernel when the kernel indicates that these are
3734 predictable. This behavior can be tweaked by changing
3735 NamePolicy= in the relevant .link file.
3737 * A new library systemd-terminal has been added that
3738 implements full TTY stream parsing and rendering. This
3739 library is supposed to be used later on for implementing a
3740 full userspace VT subsystem, replacing the current kernel
3743 * A new tool systemd-journal-upload has been added to push
3744 journal data to a remote system running
3745 systemd-journal-remote.
3747 * journald will no longer forward all local data to another
3748 running syslog daemon. This change has been made because
3749 rsyslog (which appears to be the most commonly used syslog
3750 implementation these days) no longer makes use of this, and
3751 instead pulls the data out of the journal on its own. Since
3752 forwarding the messages to a non-existent syslog server is
3753 more expensive than we assumed we have now turned this
3754 off. If you run a syslog server that is not a recent rsyslog
3755 version, you have to turn this option on again
3756 (ForwardToSyslog= in journald.conf).
3758 * journald now optionally supports the LZ4 compressor for
3759 larger journal fields. This compressor should perform much
3760 better than XZ which was the previous default.
3762 * machinectl now shows the IP addresses of local containers,
3763 if it knows them, plus the interface name of the container.
3765 * A new tool "systemd-escape" has been added that makes it
3766 easy to escape strings to build unit names and similar.
3768 * sd_notify() messages may now include a new ERRNO= field
3769 which is parsed and collected by systemd and shown among the
3770 "systemctl status" output for a service.
3772 * A new component "systemd-firstboot" has been added that
3773 queries the most basic systemd information (timezone,
3774 hostname, root password) interactively on first
3775 boot. Alternatively it may also be used to provision these
3776 things offline on OS images installed into directories.
3778 * The default sysctl.d/ snippets will now set
3780 net.ipv4.conf.default.promote_secondaries=1
3782 This has the benefit of no flushing secondary IP addresses
3783 when primary addresses are removed.
3785 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3786 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3787 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3788 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3789 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3790 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3791 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3792 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3793 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3794 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3795 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3796 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3797 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3798 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3799 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3801 — Berlin, 2014-08-19
3805 * A new tool systemd-sysusers has been added. This tool
3806 creates system users and groups in /etc/passwd and
3807 /etc/group, based on static declarative system user/group
3808 definitions in /usr/lib/sysusers.d/. This is useful to
3809 enable factory resets and volatile systems that boot up with
3810 an empty /etc directory, and thus need system users and
3811 groups created during early boot. systemd now also ships
3812 with two default sysusers.d/ files for the most basic
3813 users and groups systemd and the core operating system
3816 * A new tmpfiles snippet has been added that rebuilds the
3817 essential files in /etc on boot, should they be missing.
3819 * A directive for ensuring automatic clean-up of
3820 /var/cache/man/ has been removed from the default
3821 configuration. This line should now be shipped by the man
3822 implementation. The necessary change has been made to the
3823 man-db implementation. Note that you need to update your man
3824 implementation to one that ships this line, otherwise no
3825 automatic clean-up of /var/cache/man will take place.
3827 * A new condition ConditionNeedsUpdate= has been added that
3828 may conditionalize services to only run when /etc or /var
3829 are "older" than the vendor operating system resources in
3830 /usr. This is useful for reconstructing or updating /etc
3831 after an offline update of /usr or a factory reset, on the
3832 next reboot. Services that want to run once after such an
3833 update or reset should use this condition and order
3834 themselves before the new systemd-update-done.service, which
3835 will mark the two directories as fully updated. A number of
3836 service files have been added making use of this, to rebuild
3837 the udev hardware database, the journald message catalog and
3838 dynamic loader cache (ldconfig). The systemd-sysusers tool
3839 described above also makes use of this now. With this in
3840 place it is now possible to start up a minimal operating
3841 system with /etc empty cleanly. For more information on the
3842 concepts involved see this recent blog story:
3844 http://0pointer.de/blog/projects/stateless.html
3846 * A new system group "input" has been introduced, and all
3847 input device nodes get this group assigned. This is useful
3848 for system-level software to get access to input devices. It
3849 complements what is already done for "audio" and "video".
3851 * systemd-networkd learnt minimal DHCPv4 server support in
3852 addition to the existing DHCPv4 client support. It also
3853 learnt DHCPv6 client and IPv6 Router Solicitation client
3854 support. The DHCPv4 client gained support for static routes
3855 passed in from the server. Note that the [DHCPv4] section
3856 known in older systemd-networkd versions has been renamed to
3857 [DHCP] and is now also used by the DHCPv6 client. Existing
3858 .network files using settings of this section should be
3859 updated, though compatibility is maintained. Optionally, the
3860 client hostname may now be sent to the DHCP server.
3862 * networkd gained support for vxlan virtual networks as well
3863 as tun/tap and dummy devices.
3865 * networkd gained support for automatic allocation of address
3866 ranges for interfaces from a system-wide pool of
3867 addresses. This is useful for dynamically managing a large
3868 number of interfaces with a single network configuration
3869 file. In particular this is useful to easily assign
3870 appropriate IP addresses to the veth links of a large number
3871 of nspawn instances.
3873 * RPM macros for processing sysusers, sysctl and binfmt
3874 drop-in snippets at package installation time have been
3877 * The /etc/os-release file should now be placed in
3878 /usr/lib/os-release. The old location is automatically
3879 created as symlink. /usr/lib is the more appropriate
3880 location of this file, since it shall actually describe the
3881 vendor operating system shipped in /usr, and not the
3882 configuration stored in /etc.
3884 * .mount units gained a new boolean SloppyOptions= setting
3885 that maps to mount(8)'s -s option which enables permissive
3886 parsing of unknown mount options.
3888 * tmpfiles learnt a new "L+" directive which creates a symlink
3889 but (unlike "L") deletes a pre-existing file first, should
3890 it already exist and not already be the correct
3891 symlink. Similarly, "b+", "c+" and "p+" directives have been
3892 added as well, which create block and character devices, as
3893 well as fifos in the filesystem, possibly removing any
3894 pre-existing files of different types.
3896 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3897 'argument' field (which so far specified the source to
3898 symlink/copy the files from) is now optional. If omitted the
3899 same file os copied from /usr/share/factory/ suffixed by the
3900 full destination path. This is useful for populating /etc
3901 with essential files, by copying them from vendor defaults
3902 shipped in /usr/share/factory/etc.
3904 * A new command "systemctl preset-all" has been added that
3905 applies the service preset settings to all installed unit
3906 files. A new switch --preset-mode= has been added that
3907 controls whether only enable or only disable operations
3910 * A new command "systemctl is-system-running" has been added
3911 that allows checking the overall state of the system, for
3912 example whether it is fully up and running.
3914 * When the system boots up with an empty /etc, the equivalent
3915 to "systemctl preset-all" is executed during early boot, to
3916 make sure all default services are enabled after a factory
3919 * systemd now contains a minimal preset file that enables the
3920 most basic services systemd ships by default.
3922 * Unit files' [Install] section gained a new DefaultInstance=
3923 field for defining the default instance to create if a
3924 template unit is enabled with no instance specified.
3926 * A new passive target cryptsetup-pre.target has been added
3927 that may be used by services that need to make they run and
3928 finish before the first LUKS cryptographic device is set up.
3930 * The /dev/loop-control and /dev/btrfs-control device nodes
3931 are now owned by the "disk" group by default, opening up
3932 access to this group.
3934 * systemd-coredump will now automatically generate a
3935 stack trace of all core dumps taking place on the system,
3936 based on elfutils' libdw library. This stack trace is logged
3939 * systemd-coredump may now optionally store coredumps directly
3940 on disk (in /var/lib/systemd/coredump, possibly compressed),
3941 instead of storing them unconditionally in the journal. This
3942 mode is the new default. A new configuration file
3943 /etc/systemd/coredump.conf has been added to configure this
3944 and other parameters of systemd-coredump.
3946 * coredumpctl gained a new "info" verb to show details about a
3947 specific coredump. A new switch "-1" has also been added
3948 that makes sure to only show information about the most
3949 recent entry instead of all entries. Also, as the tool is
3950 generally useful now the "systemd-" prefix of the binary
3951 name has been removed. Distributions that want to maintain
3952 compatibility with the old name should add a symlink from
3953 the old name to the new name.
3955 * journald's SplitMode= now defaults to "uid". This makes sure
3956 that unprivileged users can access their own coredumps with
3957 coredumpctl without restrictions.
3959 * New kernel command line options "systemd.wants=" (for
3960 pulling an additional unit during boot), "systemd.mask="
3961 (for masking a specific unit for the boot), and
3962 "systemd.debug-shell" (for enabling the debug shell on tty9)
3963 have been added. This is implemented in the new generator
3964 "systemd-debug-generator".
3966 * systemd-nspawn will now by default filter a couple of
3967 syscalls for containers, among them those required for
3968 kernel module loading, direct x86 IO port access, swap
3969 management, and kexec. Most importantly though
3970 open_by_handle_at() is now prohibited for containers,
3971 closing a hole similar to a recently discussed vulnerability
3972 in docker regarding access to files on file hierarchies the
3973 container should normally not have access to. Note that, for
3974 nspawn, we generally make no security claims anyway (and
3975 this is explicitly documented in the man page), so this is
3976 just a fix for one of the most obvious problems.
3978 * A new man page file-hierarchy(7) has been added that
3979 contains a minimized, modernized version of the file system
3980 layout systemd expects, similar in style to the FHS
3981 specification or hier(5). A new tool systemd-path(1) has
3982 been added to query many of these paths for the local
3985 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3986 longer done. Since the directory now has a per-user size
3987 limit, and is cleaned on logout this appears unnecessary,
3988 in particular since this now brings the lifecycle of this
3989 directory closer in line with how IPC objects are handled.
3991 * systemd.pc now exports a number of additional directories,
3992 including $libdir (which is useful to identify the library
3993 path for the primary architecture of the system), and a
3994 couple of drop-in directories.
3996 * udev's predictable network interface names now use the dev_port
3997 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3998 distinguish between ports of the same PCI function. dev_id should
3999 only be used for ports using the same HW address, hence the need
4002 * machined has been updated to export the OS version of a
4003 container (read from /etc/os-release and
4004 /usr/lib/os-release) on the bus. This is now shown in
4005 "machinectl status" for a machine.
4007 * A new service setting RestartForceExitStatus= has been
4008 added. If configured to a set of exit signals or process
4009 return values, the service will be restarted when the main
4010 daemon process exits with any of them, regardless of the
4013 * systemctl's -H switch for connecting to remote systemd
4014 machines has been extended so that it may be used to
4015 directly connect to a specific container on the
4016 host. "systemctl -H root@foobar:waldi" will now connect as
4017 user "root" to host "foobar", and then proceed directly to
4018 the container named "waldi". Note that currently you have to
4019 authenticate as user "root" for this to work, as entering
4020 containers is a privileged operation.
4022 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4023 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4024 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4025 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4026 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4027 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4028 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4029 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4030 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4031 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4032 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4033 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4035 — Berlin, 2014-07-03
4039 * As an experimental feature, udev now tries to lock the
4040 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4041 executes events for the disk or any of its partitions.
4042 Applications like partitioning programs can lock the
4043 disk device node (flock(LOCK_EX)) and claim temporary
4044 device ownership that way; udev will entirely skip all event
4045 handling for this disk and its partitions. If the disk
4046 was opened for writing, the close will trigger a partition
4047 table rescan in udev's "watch" facility, and if needed
4048 synthesize "change" events for the disk and all its partitions.
4049 This is now unconditionally enabled, and if it turns out to
4050 cause major problems, we might turn it on only for specific
4051 devices, or might need to disable it entirely. Device Mapper
4052 devices are excluded from this logic.
4054 * We temporarily dropped the "-l" switch for fsck invocations,
4055 since they collide with the flock() logic above. util-linux
4056 upstream has been changed already to avoid this conflict,
4057 and we will readd "-l" as soon as util-linux with this
4058 change has been released.
4060 * The dependency on libattr has been removed. Since a long
4061 time, the extended attribute calls have moved to glibc, and
4062 libattr is thus unnecessary.
4064 * Virtualization detection works without privileges now. This
4065 means the systemd-detect-virt binary no longer requires
4066 CAP_SYS_PTRACE file capabilities, and our daemons can run
4067 with fewer privileges.
4069 * systemd-networkd now runs under its own "systemd-network"
4070 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4071 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4072 loses the ability to write to files owned by root this way.
4074 * Similarly, systemd-resolved now runs under its own
4075 "systemd-resolve" user with no capabilities remaining.
4077 * Similarly, systemd-bus-proxyd now runs under its own
4078 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4080 * systemd-networkd gained support for setting up "veth"
4081 virtual Ethernet devices for container connectivity, as well
4082 as GRE and VTI tunnels.
4084 * systemd-networkd will no longer automatically attempt to
4085 manually load kernel modules necessary for certain tunnel
4086 transports. Instead, it is assumed the kernel loads them
4087 automatically when required. This only works correctly on
4088 very new kernels. On older kernels, please consider adding
4089 the kernel modules to /etc/modules-load.d/ as a work-around.
4091 * The resolv.conf file systemd-resolved generates has been
4092 moved to /run/systemd/resolve/. If you have a symlink from
4093 /etc/resolv.conf, it might be necessary to correct it.
4095 * Two new service settings, ProtectHome= and ProtectSystem=,
4096 have been added. When enabled, they will make the user data
4097 (such as /home) inaccessible or read-only and the system
4098 (such as /usr) read-only, for specific services. This allows
4099 very light-weight per-service sandboxing to avoid
4100 modifications of user data or system files from
4101 services. These two new switches have been enabled for all
4102 of systemd's long-running services, where appropriate.
4104 * Socket units gained new SocketUser= and SocketGroup=
4105 settings to set the owner user and group of AF_UNIX sockets
4106 and FIFOs in the file system.
4108 * Socket units gained a new RemoveOnStop= setting. If enabled,
4109 all FIFOS and sockets in the file system will be removed
4110 when the specific socket unit is stopped.
4112 * Socket units gained a new Symlinks= setting. It takes a list
4113 of symlinks to create to file system sockets or FIFOs
4114 created by the specific Unix sockets. This is useful to
4115 manage symlinks to socket nodes with the same life-cycle as
4118 * The /dev/log socket and /dev/initctl FIFO have been moved to
4119 /run, and have been replaced by symlinks. This allows
4120 connecting to these facilities even if PrivateDevices=yes is
4121 used for a service (which makes /dev/log itself unavailable,
4122 but /run is left). This also has the benefit of ensuring
4123 that /dev only contains device nodes, directories and
4124 symlinks, and nothing else.
4126 * sd-daemon gained two new calls sd_pid_notify() and
4127 sd_pid_notifyf(). They are similar to sd_notify() and
4128 sd_notifyf(), but allow overriding of the source PID of
4129 notification messages if permissions permit this. This is
4130 useful to send notify messages on behalf of a different
4131 process (for example, the parent process). The
4132 systemd-notify tool has been updated to make use of this
4133 when sending messages (so that notification messages now
4134 originate from the shell script invoking systemd-notify and
4135 not the systemd-notify process itself. This should minimize
4136 a race where systemd fails to associate notification
4137 messages to services when the originating process already
4140 * A new "on-abnormal" setting for Restart= has been added. If
4141 set, it will result in automatic restarts on all "abnormal"
4142 reasons for a process to exit, which includes unclean
4143 signals, core dumps, timeouts and watchdog timeouts, but
4144 does not include clean and unclean exit codes or clean
4145 signals. Restart=on-abnormal is an alternative for
4146 Restart=on-failure for services that shall be able to
4147 terminate and avoid restarts on certain errors, by
4148 indicating so with an unclean exit code. Restart=on-failure
4149 or Restart=on-abnormal is now the recommended setting for
4150 all long-running services.
4152 * If the InaccessibleDirectories= service setting points to a
4153 mount point (or if there are any submounts contained within
4154 it), it is now attempted to completely unmount it, to make
4155 the file systems truly unavailable for the respective
4158 * The ReadOnlyDirectories= service setting and
4159 systemd-nspawn's --read-only parameter are now recursively
4160 applied to all submounts, too.
4162 * Mount units may now be created transiently via the bus APIs.
4164 * The support for SysV and LSB init scripts has been removed
4165 from the systemd daemon itself. Instead, it is now
4166 implemented as a generator that creates native systemd units
4167 from these scripts when needed. This enables us to remove a
4168 substantial amount of legacy code from PID 1, following the
4169 fact that many distributions only ship a very small number
4170 of LSB/SysV init scripts nowadays.
4172 * Privileged Xen (dom0) domains are not considered
4173 virtualization anymore by the virtualization detection
4174 logic. After all, they generally have unrestricted access to
4175 the hardware and usually are used to manage the unprivileged
4178 * systemd-tmpfiles gained a new "C" line type, for copying
4179 files or entire directories.
4181 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4182 lines. So far, they have been non-globbing versions of the
4183 latter, and have thus been redundant. In future, it is
4184 recommended to only use "z". "m" has hence been removed
4185 from the documentation, even though it stays supported.
4187 * A tmpfiles snippet to recreate the most basic structure in
4188 /var has been added. This is enough to create the /var/run →
4189 /run symlink and create a couple of structural
4190 directories. This allows systems to boot up with an empty or
4191 volatile /var. Of course, while with this change, the core OS
4192 now is capable with dealing with a volatile /var, not all
4193 user services are ready for it. However, we hope that sooner
4194 or later, many service daemons will be changed upstream so
4195 that they are able to automatically create their necessary
4196 directories in /var at boot, should they be missing. This is
4197 the first step to allow state-less systems that only require
4198 the vendor image for /usr to boot.
4200 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4201 empty tmpfs instance to a specific directory. This is
4202 particularly useful for making use of the automatic
4203 reconstruction of /var (see above), by passing --tmpfs=/var.
4205 * Access modes specified in tmpfiles snippets may now be
4206 prefixed with "~", which indicates that they shall be masked
4207 by whether the existing file or directory is currently
4208 writable, readable or executable at all. Also, if specified,
4209 the sgid/suid/sticky bits will be masked for all
4212 * A new passive target unit "network-pre.target" has been
4213 added which is useful for services that shall run before any
4214 network is configured, for example firewall scripts.
4216 * The "floppy" group that previously owned the /dev/fd*
4217 devices is no longer used. The "disk" group is now used
4218 instead. Distributions should probably deprecate usage of
4221 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4222 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4223 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4224 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4225 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4226 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4229 — Berlin, 2014-06-11
4233 * A new "systemd-timesyncd" daemon has been added for
4234 synchronizing the system clock across the network. It
4235 implements an SNTP client. In contrast to NTP
4236 implementations such as chrony or the NTP reference server,
4237 this only implements a client side, and does not bother with
4238 the full NTP complexity, focusing only on querying time from
4239 one remote server and synchronizing the local clock to
4240 it. Unless you intend to serve NTP to networked clients or
4241 want to connect to local hardware clocks, this simple NTP
4242 client should be more than appropriate for most
4243 installations. The daemon runs with minimal privileges, and
4244 has been hooked up with networkd to only operate when
4245 network connectivity is available. The daemon saves the
4246 current clock to disk every time a new NTP sync has been
4247 acquired, and uses this to possibly correct the system clock
4248 early at bootup, in order to accommodate for systems that
4249 lack an RTC such as the Raspberry Pi and embedded devices,
4250 and to make sure that time monotonically progresses on these
4251 systems, even if it is not always correct. To make use of
4252 this daemon, a new system user and group "systemd-timesync"
4253 needs to be created on installation of systemd.
4255 * The queue "seqnum" interface of libudev has been disabled, as
4256 it was generally incompatible with device namespacing as
4257 sequence numbers of devices go "missing" if the devices are
4258 part of a different namespace.
4260 * "systemctl list-timers" and "systemctl list-sockets" gained
4261 a --recursive switch for showing units of these types also
4262 for all local containers, similar in style to the already
4263 supported --recursive switch for "systemctl list-units".
4265 * A new RebootArgument= setting has been added for service
4266 units, which may be used to specify a kernel reboot argument
4267 to use when triggering reboots with StartLimitAction=.
4269 * A new FailureAction= setting has been added for service
4270 units which may be used to specify an operation to trigger
4271 when a service fails. This works similarly to
4272 StartLimitAction=, but unlike it, controls what is done
4273 immediately rather than only after several attempts to
4274 restart the service in question.
4276 * hostnamed got updated to also expose the kernel name,
4277 release, and version on the bus. This is useful for
4278 executing commands like hostnamectl with the -H switch.
4279 systemd-analyze makes use of this to properly display
4280 details when running non-locally.
4282 * The bootchart tool can now show cgroup information in the
4283 graphs it generates.
4285 * The CFS CPU quota cgroup attribute is now exposed for
4286 services. The new CPUQuota= switch has been added for this
4287 which takes a percentage value. Setting this will have the
4288 result that a service may never get more CPU time than the
4289 specified percentage, even if the machine is otherwise idle.
4291 * systemd-networkd learned IPIP and SIT tunnel support.
4293 * LSB init scripts exposing a dependency on $network will now
4294 get a dependency on network-online.target rather than simply
4295 network.target. This should bring LSB handling closer to
4296 what it was on SysV systems.
4298 * A new fsck.repair= kernel option has been added to control
4299 how fsck shall deal with unclean file systems at boot.
4301 * The (.ini) configuration file parser will now silently
4302 ignore sections whose name begins with "X-". This may be
4303 used to maintain application-specific extension sections in unit
4306 * machined gained a new API to query the IP addresses of
4307 registered containers. "machinectl status" has been updated
4308 to show these addresses in its output.
4310 * A new call sd_uid_get_display() has been added to the
4311 sd-login APIs for querying the "primary" session of a
4312 user. The "primary" session of the user is elected from the
4313 user's sessions and generally a graphical session is
4314 preferred over a text one.
4316 * A minimal systemd-resolved daemon has been added. It
4317 currently simply acts as a companion to systemd-networkd and
4318 manages resolv.conf based on per-interface DNS
4319 configuration, possibly supplied via DHCP. In the long run
4320 we hope to extend this into a local DNSSEC enabled DNS and
4323 * The systemd-networkd-wait-online tool is now enabled by
4324 default. It will delay network-online.target until a network
4325 connection has been configured. The tool primarily integrates
4326 with networkd, but will also make a best effort to make sense
4327 of network configuration performed in some other way.
4329 * Two new service options StartupCPUShares= and
4330 StartupBlockIOWeight= have been added that work similarly to
4331 CPUShares= and BlockIOWeight= however only apply during
4332 system startup. This is useful to prioritize certain services
4333 differently during bootup than during normal runtime.
4335 * hostnamed has been changed to prefer the statically
4336 configured hostname in /etc/hostname (unless set to
4337 'localhost' or empty) over any dynamic one supplied by
4338 dhcp. With this change, the rules for picking the hostname
4339 match more closely the rules of other configuration settings
4340 where the local administrator's configuration in /etc always
4341 overrides any other settings.
4343 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4344 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4345 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4346 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4347 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4348 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4349 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4350 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4351 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4352 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4353 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4354 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4355 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4356 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4357 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4358 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4361 — Beijing, 2014-05-28
4365 * When restoring the screen brightness at boot, stay away from
4366 the darkest setting or from the lowest 5% of the available
4367 range, depending on which is the larger value of both. This
4368 should effectively protect the user from rebooting into a
4369 black screen, should the brightness have been set to minimum
4372 * sd-login gained a new sd_machine_get_class() call to
4373 determine the class ("vm" or "container") of a machine
4374 registered with machined.
4376 * sd-login gained new calls
4377 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4378 to query the identity of the peer of a local AF_UNIX
4379 connection. They operate similarly to their sd_pid_get_xyz()
4382 * PID 1 will now maintain a system-wide system state engine
4383 with the states "starting", "running", "degraded",
4384 "maintenance", "stopping". These states are bound to system
4385 startup, normal runtime, runtime with at least one failed
4386 service, rescue/emergency mode and system shutdown. This
4387 state is shown in the "systemctl status" output when no unit
4388 name is passed. It is useful to determine system state, in
4389 particularly when doing so for many systems or containers at
4392 * A new command "list-machines" has been added to "systemctl"
4393 that lists all local OS containers and shows their system
4394 state (see above), if systemd runs inside of them.
4396 * systemctl gained a new "-r" switch to recursively enumerate
4397 units on all local containers, when used with the
4398 "list-unit" command (which is the default one that is
4399 executed when no parameters are specified).
4401 * The GPT automatic partition discovery logic will now honour
4402 two GPT partition flags: one may be set on a partition to
4403 cause it to be mounted read-only, and the other may be set
4404 on a partition to ignore it during automatic discovery.
4406 * Two new GPT type UUIDs have been added for automatic root
4407 partition discovery, for 32-bit and 64-bit ARM. This is not
4408 particularly useful for discovering the root directory on
4409 these architectures during bare-metal boots (since UEFI is
4410 not common there), but still very useful to allow booting of
4411 ARM disk images in nspawn with the -i option.
4413 * MAC addresses of interfaces created with nspawn's
4414 --network-interface= switch will now be generated from the
4415 machine name, and thus be stable between multiple invocations
4418 * logind will now automatically remove all IPC objects owned
4419 by a user if she or he fully logs out. This makes sure that
4420 users who are logged out cannot continue to consume IPC
4421 resources. This covers SysV memory, semaphores and message
4422 queues as well as POSIX shared memory and message
4423 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4424 limits. With this functionality, that is corrected. This may
4425 be turned off by using the RemoveIPC= switch of logind.conf.
4427 * The systemd-machine-id-setup and tmpfiles tools gained a
4428 --root= switch to operate on a specific root directory,
4431 * journald can now forward logged messages to the TTYs of all
4432 logged in users ("wall"). This is the default for all
4433 emergency messages now.
4435 * A new tool systemd-journal-remote has been added to stream
4436 journal log messages across the network.
4438 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4439 controller trees are mounted into it. Note that the
4440 directories mounted beneath it are not read-only. This is a
4441 security measure and is particularly useful because glibc
4442 actually includes a search logic to pick any tmpfs it can
4443 find to implement shm_open() if /dev/shm is not available
4444 (which it might very well be in namespaced setups).
4446 * machinectl gained a new "poweroff" command to cleanly power
4447 down a local OS container.
4449 * The PrivateDevices= unit file setting will now also drop the
4450 CAP_MKNOD capability from the capability bound set, and
4451 imply DevicePolicy=closed.
4453 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4454 comprehensively on all long-running systemd services where
4455 this is appropriate.
4457 * systemd-udevd will now run in a disassociated mount
4458 namespace. To mount directories from udev rules, make sure to
4459 pull in mount units via SYSTEMD_WANTS properties.
4461 * The kdbus support gained support for uploading policy into
4462 the kernel. sd-bus gained support for creating "monitoring"
4463 connections that can eavesdrop into all bus communication
4464 for debugging purposes.
4466 * Timestamps may now be specified in seconds since the UNIX
4467 epoch Jan 1st, 1970 by specifying "@" followed by the value
4470 * Native tcpwrap support in systemd has been removed. tcpwrap
4471 is old code, not really maintained anymore and has serious
4472 shortcomings, and better options such as firewalls
4473 exist. For setups that require tcpwrap usage, please
4474 consider invoking your socket-activated service via tcpd,
4475 like on traditional inetd.
4477 * A new system.conf configuration option
4478 DefaultTimerAccuracySec= has been added that controls the
4479 default AccuracySec= setting of .timer units.
4481 * Timer units gained a new WakeSystem= switch. If enabled,
4482 timers configured this way will cause the system to resume
4483 from system suspend (if the system supports that, which most
4486 * Timer units gained a new Persistent= switch. If enabled,
4487 timers configured this way will save to disk when they have
4488 been last triggered. This information is then used on next
4489 reboot to possible execute overdue timer events, that
4490 could not take place because the system was powered off.
4491 This enables simple anacron-like behaviour for timer units.
4493 * systemctl's "list-timers" will now also list the time a
4494 timer unit was last triggered in addition to the next time
4495 it will be triggered.
4497 * systemd-networkd will now assign predictable IPv4LL
4498 addresses to its local interfaces.
4500 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4501 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4502 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4503 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4504 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4505 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4506 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4507 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4510 — Berlin, 2014-03-25
4514 * A new unit file setting RestrictAddressFamilies= has been
4515 added to restrict which socket address families unit
4516 processes gain access to. This takes address family names
4517 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4518 attack surface of services via exotic protocol stacks. This
4519 is built on seccomp system call filters.
4521 * Two new unit file settings RuntimeDirectory= and
4522 RuntimeDirectoryMode= have been added that may be used to
4523 manage a per-daemon runtime directories below /run. This is
4524 an alternative for setting up directory permissions with
4525 tmpfiles snippets, and has the advantage that the runtime
4526 directory's lifetime is bound to the daemon runtime and that
4527 the daemon starts up with an empty directory each time. This
4528 is particularly useful when writing services that drop
4529 privileges using the User= or Group= setting.
4531 * The DeviceAllow= unit setting now supports globbing for
4532 matching against device group names.
4534 * The systemd configuration file system.conf gained new
4535 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4536 DefaultMemoryAccounting= to globally turn on/off accounting
4537 for specific resources (cgroups) for all units. These
4538 settings may still be overridden individually in each unit
4541 * systemd-gpt-auto-generator is now able to discover /srv and
4542 root partitions in addition to /home and swap partitions. It
4543 also supports LUKS-encrypted partitions now. With this in
4544 place, automatic discovery of partitions to mount following
4545 the Discoverable Partitions Specification
4546 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4547 is now a lot more complete. This allows booting without
4548 /etc/fstab and without root= on the kernel command line on
4549 systems prepared appropriately.
4551 * systemd-nspawn gained a new --image= switch which allows
4552 booting up disk images and Linux installations on any block
4553 device that follow the Discoverable Partitions Specification
4554 (see above). This means that installations made with
4555 appropriately updated installers may now be started and
4556 deployed using container managers, completely
4557 unmodified. (We hope that libvirt-lxc will add support for
4558 this feature soon, too.)
4560 * systemd-nspawn gained a new --network-macvlan= setting to
4561 set up a private macvlan interface for the
4562 container. Similarly, systemd-networkd gained a new
4563 Kind=macvlan setting in .netdev files.
4565 * systemd-networkd now supports configuring local addresses
4568 * A new tool systemd-network-wait-online has been added to
4569 synchronously wait for network connectivity using
4572 * The sd-bus.h bus API gained a new sd_bus_track object for
4573 tracking the life-cycle of bus peers. Note that sd-bus.h is
4574 still not a public API though (unless you specify
4575 --enable-kdbus on the configure command line, which however
4576 voids your warranty and you get no API stability guarantee).
4578 * The $XDG_RUNTIME_DIR runtime directories for each user are
4579 now individual tmpfs instances, which has the benefit of
4580 introducing separate pools for each user, with individual
4581 size limits, and thus making sure that unprivileged clients
4582 can no longer negatively impact the system or other users by
4583 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4584 RuntimeDirectorySize= has been introduced that allows
4585 controlling the default size limit for all users. It
4586 defaults to 10% of the available physical memory. This is no
4587 replacement for quotas on tmpfs though (which the kernel
4588 still does not support), as /dev/shm and /tmp are still
4589 shared resources used by both the system and unprivileged
4592 * logind will now automatically turn off automatic suspending
4593 on laptop lid close when more than one display is
4594 connected. This was previously expected to be implemented
4595 individually in desktop environments (such as GNOME),
4596 however has been added to logind now, in order to fix a
4597 boot-time race where a desktop environment might not have
4598 been started yet and thus not been able to take an inhibitor
4599 lock at the time where logind already suspends the system
4600 due to a closed lid.
4602 * logind will now wait at least 30s after each system
4603 suspend/resume cycle, and 3min after system boot before
4604 suspending the system due to a closed laptop lid. This
4605 should give USB docking stations and similar enough time to
4606 be probed and configured after system resume and boot in
4607 order to then act as suspend blocker.
4609 * systemd-run gained a new --property= setting which allows
4610 initialization of resource control properties (and others)
4611 for the created scope or service unit. Example: "systemd-run
4612 --property=BlockIOWeight=10 updatedb" may be used to run
4613 updatedb at a low block IO scheduling weight.
4615 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4616 now also work in --scope mode.
4618 * When systemd is compiled with kdbus support, basic support
4619 for enforced policies is now in place. (Note that enabling
4620 kdbus still voids your warranty and no API compatibility
4623 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4624 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4625 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4626 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4627 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4628 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4629 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4630 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4631 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4632 Zbigniew Jędrzejewski-Szmek
4634 — Berlin, 2014-03-12
4638 * systemd will now relabel /dev after loading the SMACK policy
4639 according to SMACK rules.
4641 * A new unit file option AppArmorProfile= has been added to
4642 set the AppArmor profile for the processes of a unit.
4644 * A new condition check ConditionArchitecture= has been added
4645 to conditionalize units based on the system architecture, as
4646 reported by uname()'s "machine" field.
4648 * systemd-networkd now supports matching on the system
4649 virtualization, architecture, kernel command line, host name
4652 * logind is now a lot more aggressive when suspending the
4653 machine due to a closed laptop lid. Instead of acting only
4654 on the lid close action, it will continuously watch the lid
4655 status and act on it. This is useful for laptops where the
4656 power button is on the outside of the chassis so that it can
4657 be reached without opening the lid (such as the Lenovo
4658 Yoga). On those machines, logind will now immediately
4659 re-suspend the machine if the power button has been
4660 accidentally pressed while the laptop was suspended and in a
4661 backpack or similar.
4663 * logind will now watch SW_DOCK switches and inhibit reaction
4664 to the lid switch if it is pressed. This means that logind
4665 will not suspend the machine anymore if the lid is closed
4666 and the system is docked, if the laptop supports SW_DOCK
4667 notifications via the input layer. Note that ACPI docking
4668 stations do not generate this currently. Also note that this
4669 logic is usually not fully sufficient and Desktop
4670 Environments should take a lid switch inhibitor lock when an
4671 external display is connected, as systemd will not watch
4674 * nspawn will now make use of the devices cgroup controller by
4675 default, and only permit creation of and access to the usual
4676 API device nodes like /dev/null or /dev/random, as well as
4677 access to (but not creation of) the pty devices.
4679 * We will now ship a default .network file for
4680 systemd-networkd that automatically configures DHCP for
4681 network interfaces created by nspawn's --network-veth or
4682 --network-bridge= switches.
4684 * systemd will now understand the usual M, K, G, T suffixes
4685 according to SI conventions (i.e. to the base 1000) when
4686 referring to throughput and hardware metrics. It will stay
4687 with IEC conventions (i.e. to the base 1024) for software
4688 metrics, according to what is customary according to
4689 Wikipedia. We explicitly document which base applies for
4690 each configuration option.
4692 * The DeviceAllow= setting in unit files now supports a syntax
4693 to whitelist an entire group of devices node majors at once,
4694 based on the /proc/devices listing. For example, with the
4695 string "char-pts", it is now possible to whitelist all
4696 current and future pseudo-TTYs at once.
4698 * sd-event learned a new "post" event source. Event sources of
4699 this type are triggered by the dispatching of any event
4700 source of a type that is not "post". This is useful for
4701 implementing clean-up and check event sources that are
4702 triggered by other work being done in the program.
4704 * systemd-networkd is no longer statically enabled, but uses
4705 the usual [Install] sections so that it can be
4706 enabled/disabled using systemctl. It still is enabled by
4709 * When creating a veth interface pair with systemd-nspawn, the
4710 host side will now be prefixed with "vb-" if
4711 --network-bridge= is used, and with "ve-" if --network-veth
4712 is used. This way, it is easy to distinguish these cases on
4713 the host, for example to apply different configuration to
4714 them with systemd-networkd.
4716 * The compatibility libraries for libsystemd-journal.so,
4717 libsystem-id128.so, libsystemd-login.so and
4718 libsystemd-daemon.so do not make use of IFUNC
4719 anymore. Instead, we now build libsystemd.so multiple times
4720 under these alternative names. This means that the footprint
4721 is drastically increased, but given that these are
4722 transitional compatibility libraries, this should not matter
4723 much. This change has been made necessary to support the ARM
4724 platform for these compatibility libraries, as the ARM
4725 toolchain is not really at the same level as the toolchain
4726 for other architectures like x86 and does not support
4727 IFUNC. Please make sure to use --enable-compat-libs only
4728 during a transitional period!
4730 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4731 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4732 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4733 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4734 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4735 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4736 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4737 Zbigniew Jędrzejewski-Szmek
4739 — Berlin, 2014-02-24
4743 * A new component "systemd-networkd" has been added that can
4744 be used to configure local network interfaces statically or
4745 via DHCP. It is capable of bringing up bridges, VLANs, and
4746 bonding. Currently, no hook-ups for interactive network
4747 configuration are provided. Use this for your initrd,
4748 container, embedded, or server setup if you need a simple,
4749 yet powerful, network configuration solution. This
4750 configuration subsystem is quite nifty, as it allows wildcard
4751 hotplug matching in interfaces. For example, with a single
4752 configuration snippet, you can configure that all Ethernet
4753 interfaces showing up are automatically added to a bridge,
4754 or similar. It supports link-sensing and more.
4756 * A new tool "systemd-socket-proxyd" has been added which can
4757 act as a bidirectional proxy for TCP sockets. This is
4758 useful for adding socket activation support to services that
4759 do not actually support socket activation, including virtual
4760 machines and the like.
4762 * Add a new tool to save/restore rfkill state on
4765 * Save/restore state of keyboard backlights in addition to
4766 display backlights on shutdown/boot.
4768 * udev learned a new SECLABEL{} construct to label device
4769 nodes with a specific security label when they appear. For
4770 now, only SECLABEL{selinux} is supported, but the syntax is
4771 prepared for additional security frameworks.
4773 * udev gained a new scheme to configure link-level attributes
4774 from files in /etc/systemd/network/*.link. These files can
4775 match against MAC address, device path, driver name and type,
4776 and will apply attributes like the naming policy, link speed,
4777 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4778 address assignment policy (randomized, ...).
4780 * The configuration of network interface naming rules for
4781 "permanent interface names" has changed: a new NamePolicy=
4782 setting in the [Link] section of .link files determines the
4783 priority of possible naming schemes (onboard, slot, MAC,
4784 path). The default value of this setting is determined by
4785 /usr/lib/net/links/99-default.link. Old
4786 80-net-name-slot.rules udev configuration file has been
4787 removed, so local configuration overriding this file should
4788 be adapted to override 99-default.link instead.
4790 * When the User= switch is used in a unit file, also
4791 initialize $SHELL= based on the user database entry.
4793 * systemd no longer depends on libdbus. All communication is
4794 now done with sd-bus, systemd's low-level bus library
4797 * kdbus support has been added to PID 1 itself. When kdbus is
4798 enabled, this causes PID 1 to set up the system bus and
4799 enable support for a new ".busname" unit type that
4800 encapsulates bus name activation on kdbus. It works a little
4801 bit like ".socket" units, except for bus names. A new
4802 generator has been added that converts classic dbus1 service
4803 activation files automatically into native systemd .busname
4806 * sd-bus: add a light-weight vtable implementation that allows
4807 defining objects on the bus with a simple static const
4808 vtable array of its methods, signals and properties.
4810 * systemd will not generate or install static dbus
4811 introspection data anymore to /usr/share/dbus-1/interfaces,
4812 as the precise format of these files is unclear, and
4813 nothing makes use of it.
4815 * A proxy daemon is now provided to proxy clients connecting
4816 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4817 compatibility with classic D-Bus.
4819 * A bus driver implementation has been added that supports the
4820 classic D-Bus bus driver calls on kdbus, also for
4821 compatibility purposes.
4823 * A new API "sd-event.h" has been added that implements a
4824 minimal event loop API built around epoll. It provides a
4825 couple of features that direct epoll usage is lacking:
4826 prioritization of events, scales to large numbers of timer
4827 events, per-event timer slack (accuracy), system-wide
4828 coalescing of timer events, exit handlers, watchdog
4829 supervision support using systemd's sd_notify() API, child
4832 * A new API "sd-rntl.h" has been added that provides an API
4833 around the route netlink interface of the kernel, similar in
4834 style to "sd-bus.h".
4836 * A new API "sd-dhcp-client.h" has been added that provides a
4837 small DHCPv4 client-side implementation. This is used by
4840 * There is a new kernel command line option
4841 "systemd.restore_state=0|1". When set to "0", none of the
4842 systemd tools will restore saved runtime state to hardware
4843 devices. More specifically, the rfkill and backlight states
4846 * The FsckPassNo= compatibility option in mount/service units
4847 has been removed. The fstab generator will now add the
4848 necessary dependencies automatically, and does not require
4849 PID1's support for that anymore.
4851 * journalctl gained a new switch, --list-boots, that lists
4852 recent boots with their times and boot IDs.
4854 * The various tools like systemctl, loginctl, timedatectl,
4855 busctl, systemd-run, ... have gained a new switch "-M" to
4856 connect to a specific, local OS container (as direct
4857 connection, without requiring SSH). This works on any
4858 container that is registered with machined, such as those
4859 created by libvirt-lxc or nspawn.
4861 * systemd-run and systemd-analyze also gained support for "-H"
4862 to connect to remote hosts via SSH. This is particularly
4863 useful for systemd-run because it enables queuing of jobs
4864 onto remote systems.
4866 * machinectl gained a new command "login" to open a getty
4867 login in any local container. This works with any container
4868 that is registered with machined (such as those created by
4869 libvirt-lxc or nspawn), and which runs systemd inside.
4871 * machinectl gained a new "reboot" command that may be used to
4872 trigger a reboot on a specific container that is registered
4873 with machined. This works on any container that runs an init
4874 system of some kind.
4876 * systemctl gained a new "list-timers" command to print a nice
4877 listing of installed timer units with the times they elapse
4880 * Alternative reboot() parameters may now be specified on the
4881 "systemctl reboot" command line and are passed to the
4882 reboot() system call.
4884 * systemctl gained a new --job-mode= switch to configure the
4885 mode to queue a job with. This is a more generic version of
4886 --fail, --irreversible, and --ignore-dependencies, which are
4887 still available but not advertised anymore.
4889 * /etc/systemd/system.conf gained new settings to configure
4890 various default timeouts of units, as well as the default
4891 start limit interval and burst. These may still be overridden
4894 * PID1 will now export on the bus profile data of the security
4895 policy upload process (such as the SELinux policy upload to
4898 * journald: when forwarding logs to the console, include
4899 timestamps (following the setting in
4900 /sys/module/printk/parameters/time).
4902 * OnCalendar= in timer units now understands the special
4903 strings "yearly" and "annually". (Both are equivalent)
4905 * The accuracy of timer units is now configurable with the new
4906 AccuracySec= setting. It defaults to 1min.
4908 * A new dependency type JoinsNamespaceOf= has been added that
4909 allows running two services within the same /tmp and network
4910 namespace, if PrivateNetwork= or PrivateTmp= are used.
4912 * A new command "cat" has been added to systemctl. It outputs
4913 the original unit file of a unit, and concatenates the
4914 contents of additional "drop-in" unit file snippets, so that
4915 the full configuration is shown.
4917 * systemctl now supports globbing on the various "list-xyz"
4918 commands, like "list-units" or "list-sockets", as well as on
4919 those commands which take multiple unit names.
4921 * journalctl's --unit= switch gained support for globbing.
4923 * All systemd daemons now make use of the watchdog logic so
4924 that systemd automatically notices when they hang.
4926 * If the $container_ttys environment variable is set,
4927 getty-generator will automatically spawn a getty for each
4928 listed tty. This is useful for container managers to request
4929 login gettys to be spawned on as many ttys as needed.
4931 * %h, %s, %U specifier support is not available anymore when
4932 used in unit files for PID 1. This is because NSS calls are
4933 not safe from PID 1. They stay available for --user
4934 instances of systemd, and as special case for the root user.
4936 * loginctl gained a new "--no-legend" switch to turn off output
4939 * The "sd-login.h" API gained three new calls:
4940 sd_session_is_remote(), sd_session_get_remote_user(),
4941 sd_session_get_remote_host() to query information about
4944 * The udev hardware database now also carries vendor/product
4945 information of SDIO devices.
4947 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4948 determine whether watchdog notifications are requested by
4951 * Socket-activated per-connection services now include a
4952 short description of the connection parameters in the
4955 * tmpfiles gained a new "--boot" option. When this is not used,
4956 only lines where the command character is not suffixed with
4957 "!" are executed. When this option is specified, those
4958 options are executed too. This partitions tmpfiles
4959 directives into those that can be safely executed at any
4960 time, and those which should be run only at boot (for
4961 example, a line that creates /run/nologin).
4963 * A new API "sd-resolve.h" has been added which provides a simple
4964 asynchronous wrapper around glibc NSS host name resolution
4965 calls, such as getaddrinfo(). In contrast to glibc's
4966 getaddrinfo_a(), it does not use signals. In contrast to most
4967 other asynchronous name resolution libraries, this one does
4968 not reimplement DNS, but reuses NSS, so that alternate
4969 host name resolution systems continue to work, such as mDNS,
4970 LDAP, etc. This API is based on libasyncns, but it has been
4971 cleaned up for inclusion in systemd.
4973 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4974 "sd-daemon.h" are no longer found in individual libraries
4975 libsystemd-journal.so, libsystemd-login.so,
4976 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4977 merged them into a single library, libsystemd.so, which
4978 provides all symbols. The reason for this is cyclic
4979 dependencies, as these libraries tend to use each other's
4980 symbols. So far, we have managed to workaround that by linking
4981 a copy of a good part of our code into each of these
4982 libraries again and again, which, however, makes certain
4983 things hard to do, like sharing static variables. Also, it
4984 substantially increases footprint. With this change, there
4985 is only one library for the basic APIs systemd
4986 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4987 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4988 library as well, however are subject to the --enable-kdbus
4989 switch (see below). Note that "sd-dhcp-client.h" is not part
4990 of this library (this is because it only consumes, never
4991 provides, services of/to other APIs). To make the transition
4992 easy from the separate libraries to the unified one, we
4993 provide the --enable-compat-libs compile-time switch which
4994 will generate stub libraries that are compatible with the
4995 old ones but redirect all calls to the new one.
4997 * All of the kdbus logic and the new APIs "sd-bus.h",
4998 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4999 and "sd-utf8.h" are compile-time optional via the
5000 "--enable-kdbus" switch, and they are not compiled in by
5001 default. To make use of kdbus, you have to explicitly enable
5002 the switch. Note however, that neither the kernel nor the
5003 userspace API for all of this is considered stable yet. We
5004 want to maintain the freedom to still change the APIs for
5005 now. By specifying this build-time switch, you acknowledge
5006 that you are aware of the instability of the current
5009 * Also, note that while kdbus is pretty much complete,
5010 it lacks one thing: proper policy support. This means you
5011 can build a fully working system with all features; however,
5012 it will be highly insecure. Policy support will be added in
5013 one of the next releases, at the same time that we will
5014 declare the APIs stable.
5016 * When the kernel command line argument "kdbus" is specified,
5017 systemd will automatically load the kdbus.ko kernel module. At
5018 this stage of development, it is only useful for testing kdbus
5019 and should not be used in production. Note: if "--enable-kdbus"
5020 is specified, and the kdbus.ko kernel module is available, and
5021 "kdbus" is added to the kernel command line, the entire system
5022 runs with kdbus instead of dbus-daemon, with the above mentioned
5023 problem of missing the system policy enforcement. Also a future
5024 version of kdbus.ko or a newer systemd will not be compatible with
5025 each other, and will unlikely be able to boot the machine if only
5026 one of them is updated.
5028 * systemctl gained a new "import-environment" command which
5029 uploads the caller's environment (or parts thereof) into the
5030 service manager so that it is inherited by services started
5031 by the manager. This is useful to upload variables like
5032 $DISPLAY into the user service manager.
5034 * A new PrivateDevices= switch has been added to service units
5035 which allows running a service with a namespaced /dev
5036 directory that does not contain any device nodes for
5037 physical devices. More specifically, it only includes devices
5038 such as /dev/null, /dev/urandom, and /dev/zero which are API
5041 * logind has been extended to support behaviour like VT
5042 switching on seats that do not support a VT. This makes
5043 multi-session available on seats that are not the first seat
5044 (seat0), and on systems where kernel support for VTs has
5045 been disabled at compile-time.
5047 * If a process holds a delay lock for system sleep or shutdown
5048 and fails to release it in time, we will now log its
5049 identity. This makes it easier to identify processes that
5050 cause slow suspends or power-offs.
5052 * When parsing /etc/crypttab, support for a new key-slot=
5053 option as supported by Debian is added. It allows indicating
5054 which LUKS slot to use on disk, speeding up key loading.
5056 * The sd_journald_sendv() API call has been checked and
5057 officially declared to be async-signal-safe so that it may
5058 be invoked from signal handlers for logging purposes.
5060 * Boot-time status output is now enabled automatically after a
5061 short timeout if boot does not progress, in order to give
5062 the user an indication what she or he is waiting for.
5064 * The boot-time output has been improved to show how much time
5065 remains until jobs expire.
5067 * The KillMode= switch in service units gained a new possible
5068 value "mixed". If set, and the unit is shut down, then the
5069 initial SIGTERM signal is sent only to the main daemon
5070 process, while the following SIGKILL signal is sent to
5071 all remaining processes of the service.
5073 * When a scope unit is registered, a new property "Controller"
5074 may be set. If set to a valid bus name, systemd will send a
5075 RequestStop() signal to this name when it would like to shut
5076 down the scope. This may be used to hook manager logic into
5077 the shutdown logic of scope units. Also, scope units may now
5078 be put in a special "abandoned" state, in which case the
5079 manager process which created them takes no further
5080 responsibilities for it.
5082 * When reading unit files, systemd will now verify
5083 the access mode of these files, and warn about certain
5084 suspicious combinations. This has been added to make it
5085 easier to track down packaging bugs where unit files are
5086 marked executable or world-writable.
5088 * systemd-nspawn gained a new "--setenv=" switch to set
5089 container-wide environment variables. The similar option in
5090 systemd-activate was renamed from "--environment=" to
5091 "--setenv=" for consistency.
5093 * systemd-nspawn has been updated to create a new kdbus domain
5094 for each container that is invoked, thus allowing each
5095 container to have its own set of system and user buses,
5096 independent of the host.
5098 * systemd-nspawn gained a new --drop-capability= switch to run
5099 the container with less capabilities than the default. Both
5100 --drop-capability= and --capability= now take the special
5101 string "all" for dropping or keeping all capabilities.
5103 * systemd-nspawn gained new switches for executing containers
5104 with specific SELinux labels set.
5106 * systemd-nspawn gained a new --quiet switch to not generate
5107 any additional output but the container's own console
5110 * systemd-nspawn gained a new --share-system switch to run a
5111 container without PID namespacing enabled.
5113 * systemd-nspawn gained a new --register= switch to control
5114 whether the container is registered with systemd-machined or
5115 not. This is useful for containers that do not run full
5116 OS images, but only specific apps.
5118 * systemd-nspawn gained a new --keep-unit which may be used
5119 when invoked as the only program from a service unit, and
5120 results in registration of the unit service itself in
5121 systemd-machined, instead of a newly opened scope unit.
5123 * systemd-nspawn gained a new --network-interface= switch for
5124 moving arbitrary interfaces to the container. The new
5125 --network-veth switch creates a virtual Ethernet connection
5126 between host and container. The new --network-bridge=
5127 switch then allows assigning the host side of this virtual
5128 Ethernet connection to a bridge device.
5130 * systemd-nspawn gained a new --personality= switch for
5131 setting the kernel personality for the container. This is
5132 useful when running a 32-bit container on a 64-bit host. A
5133 similar option Personality= is now also available for service
5136 * logind will now also track a "Desktop" identifier for each
5137 session which encodes the desktop environment of it. This is
5138 useful for desktop environments that want to identify
5139 multiple running sessions of itself easily.
5141 * A new SELinuxContext= setting for service units has been
5142 added that allows setting a specific SELinux execution
5143 context for a service.
5145 * Most systemd client tools will now honour $SYSTEMD_LESS for
5146 settings of the "less" pager. By default, these tools will
5147 override $LESS to allow certain operations to work, such as
5148 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5149 influence this logic.
5151 * systemd's "seccomp" hook-up has been changed to make use of
5152 the libseccomp library instead of using its own
5153 implementation. This has benefits for portability among
5156 * For usage together with SystemCallFilter=, a new
5157 SystemCallErrorNumber= setting has been introduced that
5158 allows configuration of a system error number to be returned
5159 on filtered system calls, instead of immediately killing the
5160 process. Also, SystemCallArchitectures= has been added to
5161 limit access to system calls of a particular architecture
5162 (in order to turn off support for unused secondary
5163 architectures). There is also a global
5164 SystemCallArchitectures= setting in system.conf now to turn
5165 off support for non-native system calls system-wide.
5167 * systemd requires a kernel with a working name_to_handle_at(),
5168 please see the kernel config requirements in the README file.
5170 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5171 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5172 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5173 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5174 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5175 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5176 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5177 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5178 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5179 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5180 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5181 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5182 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5183 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5184 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5185 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5186 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5187 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5188 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5189 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5190 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5191 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5192 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5193 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5195 — Berlin, 2014-02-20
5199 * logind has gained support for facilitating privileged input
5200 and drm device access for unprivileged clients. This work is
5201 useful to allow Wayland display servers (and similar
5202 programs, such as kmscon) to run under the user's ID and
5203 access input and drm devices which are normally
5204 protected. When this is used (and the kernel is new enough)
5205 logind will "mute" IO on the file descriptors passed to
5206 Wayland as long as it is in the background and "unmute" it
5207 if it returns into the foreground. This allows secure
5208 session switching without allowing background sessions to
5209 eavesdrop on input and display data. This also introduces
5210 session switching support if VT support is turned off in the
5211 kernel, and on seats that are not seat0.
5213 * A new kernel command line option luks.options= is understood
5214 now which allows specifying LUKS options for usage for LUKS
5215 encrypted partitions specified with luks.uuid=.
5217 * tmpfiles.d(5) snippets may now use specifier expansion in
5218 path names. More specifically %m, %b, %H, %v, are now
5219 replaced by the local machine id, boot id, hostname, and
5220 kernel version number.
5222 * A new tmpfiles.d(5) command "m" has been introduced which
5223 may be used to change the owner/group/access mode of a file
5224 or directory if it exists, but do nothing if it does not.
5226 * This release removes high-level support for the
5227 MemorySoftLimit= cgroup setting. The underlying kernel
5228 cgroup attribute memory.soft_limit= is currently badly
5229 designed and likely to be removed from the kernel API in its
5230 current form, hence we should not expose it for now.
5232 * The memory.use_hierarchy cgroup attribute is now enabled for
5233 all cgroups systemd creates in the memory cgroup
5234 hierarchy. This option is likely to be come the built-in
5235 default in the kernel anyway, and the non-hierarchical mode
5236 never made much sense in the intrinsically hierarchical
5239 * A new field _SYSTEMD_SLICE= is logged along with all journal
5240 messages containing the slice a message was generated
5241 from. This is useful to allow easy per-customer filtering of
5242 logs among other things.
5244 * systemd-journald will no longer adjust the group of journal
5245 files it creates to the "systemd-journal" group. Instead we
5246 rely on the journal directory to be owned by the
5247 "systemd-journal" group, and its setgid bit set, so that the
5248 kernel file system layer will automatically enforce that
5249 journal files inherit this group assignment. The reason for
5250 this change is that we cannot allow NSS look-ups from
5251 journald which would be necessary to resolve
5252 "systemd-journal" to a numeric GID, because this might
5253 create deadlocks if NSS involves synchronous queries to
5254 other daemons (such as nscd, or sssd) which in turn are
5255 logging clients of journald and might block on it, which
5256 would then dead lock. A tmpfiles.d(5) snippet included in
5257 systemd will make sure the setgid bit and group are
5258 properly set on the journal directory if it exists on every
5259 boot. However, we recommend adjusting it manually after
5260 upgrades too (or from RPM scriptlets), so that the change is
5261 not delayed until next reboot.
5263 * Backlight and random seed files in /var/lib/ have moved into
5264 the /var/lib/systemd/ directory, in order to centralize all
5265 systemd generated files in one directory.
5267 * Boot time performance measurements (as displayed by
5268 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5269 performance information if that's available to determine how
5270 much time BIOS and boot loader initialization required. With
5271 a sufficiently new BIOS you hence no longer need to boot
5272 with Gummiboot to get access to such information.
5274 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5275 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5276 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5277 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5278 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5279 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5280 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5282 — Berlin, 2013-10-02
5286 * The Restart= option for services now understands a new
5287 on-watchdog setting, which will restart the service
5288 automatically if the service stops sending out watchdog keep
5289 alive messages (as configured with WatchdogSec=).
5291 * The getty generator (which is responsible for bringing up a
5292 getty on configured serial consoles) will no longer only
5293 start a getty on the primary kernel console but on all
5294 others, too. This makes the order in which console= is
5295 specified on the kernel command line less important.
5297 * libsystemd-logind gained a new sd_session_get_vt() call to
5298 retrieve the VT number of a session.
5300 * If the option "tries=0" is set for an entry of /etc/crypttab
5301 its passphrase is queried indefinitely instead of any
5302 maximum number of tries.
5304 * If a service with a configure PID file terminates its PID
5305 file will now be removed automatically if it still exists
5306 afterwards. This should put an end to stale PID files.
5308 * systemd-run will now also take relative binary path names
5309 for execution and no longer insists on absolute paths.
5311 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5312 paths that are optionally prefixed with "-" to indicate that
5313 it should not be considered a failure if they do not exist.
5315 * journalctl -o (and similar commands) now understands a new
5316 output mode "short-precise", it is similar to "short" but
5317 shows timestamps with usec accuracy.
5319 * The option "discard" (as known from Debian) is now
5320 synonymous to "allow-discards" in /etc/crypttab. In fact,
5321 "discard" is preferred now (since it is easier to remember
5324 * Some licensing clean-ups were made, so that more code is now
5325 LGPL-2.1 licensed than before.
5327 * A minimal tool to save/restore the display backlight
5328 brightness across reboots has been added. It will store the
5329 backlight setting as late as possible at shutdown, and
5330 restore it as early as possible during reboot.
5332 * A logic to automatically discover and enable home and swap
5333 partitions on GPT disks has been added. With this in place
5334 /etc/fstab becomes optional for many setups as systemd can
5335 discover certain partitions located on the root disk
5336 automatically. Home partitions are recognized under their
5337 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5338 partitions are recognized under their GPT type ID
5339 0657fd6da4ab43c484e50933c84b4f4f.
5341 * systemd will no longer pass any environment from the kernel
5342 or initrd to system services. If you want to set an
5343 environment for all services, do so via the kernel command
5344 line systemd.setenv= assignment.
5346 * The systemd-sysctl tool no longer natively reads the file
5347 /etc/sysctl.conf. If desired, the file should be symlinked
5348 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5349 legacy support by a symlink rather than built-in code, it
5350 also makes the otherwise hidden order of application of the
5351 different files visible. (Note that this partly reverts to a
5352 pre-198 application order of sysctl knobs!)
5354 * The "systemctl set-log-level" and "systemctl dump" commands
5355 have been moved to systemd-analyze.
5357 * systemd-run learned the new --remain-after-exit switch,
5358 which causes the scope unit not to be cleaned up
5359 automatically after the process terminated.
5361 * tmpfiles learned a new --exclude-prefix= switch to exclude
5362 certain paths from operation.
5364 * journald will now automatically flush all messages to disk
5365 as soon as a message at the log level CRIT, ALERT or EMERG
5368 Contributions from: Andrew Cook, Brandon Philips, Christian
5369 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5370 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5371 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5372 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5373 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5374 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5375 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5376 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5377 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5378 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5379 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5380 William Giokas, Zbigniew Jędrzejewski-Szmek
5382 — Berlin, 2013-09-13
5386 * The documentation has been updated to cover the various new
5387 concepts introduced with 205.
5389 * Unit files now understand the new %v specifier which
5390 resolves to the kernel version string as returned by "uname
5393 * systemctl now supports filtering the unit list output by
5394 load state, active state and sub state, using the new
5397 * "systemctl status" will now show the results of the
5398 condition checks (like ConditionPathExists= and similar) of
5399 the last start attempts of the unit. They are also logged to
5402 * "journalctl -b" may now be used to look for boot output of a
5403 specific boot. Try "journalctl -b -1" for the previous boot,
5404 but the syntax is substantially more powerful.
5406 * "journalctl --show-cursor" has been added which prints the
5407 cursor string the last shown log line. This may then be used
5408 with the new "journalctl --after-cursor=" switch to continue
5409 browsing logs from that point on.
5411 * "journalctl --force" may now be used to force regeneration
5414 * Creation of "dead" device nodes has been moved from udev
5415 into kmod and tmpfiles. Previously, udev would read the kmod
5416 databases to pre-generate dead device nodes based on meta
5417 information contained in kernel modules, so that these would
5418 be auto-loaded on access rather then at boot. As this
5419 does not really have much to do with the exposing actual
5420 kernel devices to userspace this has always been slightly
5421 alien in the udev codebase. Following the new scheme kmod
5422 will now generate a runtime snippet for tmpfiles from the
5423 module meta information and it now is tmpfiles' job to the
5424 create the nodes. This also allows overriding access and
5425 other parameters for the nodes using the usual tmpfiles
5426 facilities. As side effect this allows us to remove the
5427 CAP_SYS_MKNOD capability bit from udevd entirely.
5429 * logind's device ACLs may now be applied to these "dead"
5430 devices nodes too, thus finally allowing managed access to
5431 devices such as /dev/snd/sequencer without loading the
5432 backing module right-away.
5434 * A new RPM macro has been added that may be used to apply
5435 tmpfiles configuration during package installation.
5437 * systemd-detect-virt and ConditionVirtualization= now can
5438 detect User-Mode-Linux machines (UML).
5440 * journald will now implicitly log the effective capabilities
5441 set of processes in the message metadata.
5443 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5445 * The initrd interface has been simplified (more specifically,
5446 support for passing performance data via environment
5447 variables and fsck results via files in /run has been
5448 removed). These features were non-essential, and are
5449 nowadays available in a much nicer way by having systemd in
5450 the initrd serialize its state and have the hosts systemd
5451 deserialize it again.
5453 * The udev "keymap" data files and tools to apply keyboard
5454 specific mappings of scan to key codes, and force-release
5455 scan code lists have been entirely replaced by a udev
5456 "keyboard" builtin and a hwdb data file.
5458 * systemd will now honour the kernel's "quiet" command line
5459 argument also during late shutdown, resulting in a
5460 completely silent shutdown when used.
5462 * There's now an option to control the SO_REUSEPORT socket
5463 option in .socket units.
5465 * Instance units will now automatically get a per-template
5466 subslice of system.slice unless something else is explicitly
5467 configured. For example, instances of sshd@.service will now
5468 implicitly be placed in system-sshd.slice rather than
5469 system.slice as before.
5471 * Test coverage support may now be enabled at build time.
5473 Contributions from: Dave Reisner, Frederic Crozat, Harald
5474 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5475 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5476 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5477 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5478 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5479 Giokas, Zbigniew Jędrzejewski-Szmek
5481 — Berlin, 2013-07-23
5485 * Two new unit types have been introduced:
5487 Scope units are very similar to service units, however, are
5488 created out of pre-existing processes — instead of PID 1
5489 forking off the processes. By using scope units it is
5490 possible for system services and applications to group their
5491 own child processes (worker processes) in a powerful way
5492 which then maybe used to organize them, or kill them
5493 together, or apply resource limits on them.
5495 Slice units may be used to partition system resources in an
5496 hierarchical fashion and then assign other units to them. By
5497 default there are now three slices: system.slice (for all
5498 system services), user.slice (for all user sessions),
5499 machine.slice (for VMs and containers).
5501 Slices and scopes have been introduced primarily in
5502 context of the work to move cgroup handling to a
5503 single-writer scheme, where only PID 1
5504 creates/removes/manages cgroups.
5506 * There's a new concept of "transient" units. In contrast to
5507 normal units these units are created via an API at runtime,
5508 not from configuration from disk. More specifically this
5509 means it is now possible to run arbitrary programs as
5510 independent services, with all execution parameters passed
5511 in via bus APIs rather than read from disk. Transient units
5512 make systemd substantially more dynamic then it ever was,
5513 and useful as a general batch manager.
5515 * logind has been updated to make use of scope and slice units
5516 for managing user sessions. As a user logs in he will get
5517 his own private slice unit, to which all sessions are added
5518 as scope units. We also added support for automatically
5519 adding an instance of user@.service for the user into the
5520 slice. Effectively logind will no longer create cgroup
5521 hierarchies on its own now, it will defer entirely to PID 1
5522 for this by means of scope, service and slice units. Since
5523 user sessions this way become entities managed by PID 1
5524 the output of "systemctl" is now a lot more comprehensive.
5526 * A new mini-daemon "systemd-machined" has been added which
5527 may be used by virtualization managers to register local
5528 VMs/containers. nspawn has been updated accordingly, and
5529 libvirt will be updated shortly. machined will collect a bit
5530 of meta information about the VMs/containers, and assign
5531 them their own scope unit (see above). The collected
5532 meta-data is then made available via the "machinectl" tool,
5533 and exposed in "ps" and similar tools. machined/machinectl
5534 is compile-time optional.
5536 * As discussed earlier, the low-level cgroup configuration
5537 options ControlGroup=, ControlGroupModify=,
5538 ControlGroupPersistent=, ControlGroupAttribute= have been
5539 removed. Please use high-level attribute settings instead as
5540 well as slice units.
5542 * A new bus call SetUnitProperties() has been added to alter
5543 various runtime parameters of a unit. This is primarily
5544 useful to alter cgroup parameters dynamically in a nice way,
5545 but will be extended later on to make more properties
5546 modifiable at runtime. systemctl gained a new set-properties
5547 command that wraps this call.
5549 * A new tool "systemd-run" has been added which can be used to
5550 run arbitrary command lines as transient services or scopes,
5551 while configuring a number of settings via the command
5552 line. This tool is currently very basic, however already
5553 very useful. We plan to extend this tool to even allow
5554 queuing of execution jobs with time triggers from the
5555 command line, similar in fashion to "at".
5557 * nspawn will now inform the user explicitly that kernels with
5558 audit enabled break containers, and suggest the user to turn
5561 * Support for detecting the IMA and AppArmor security
5562 frameworks with ConditionSecurity= has been added.
5564 * journalctl gained a new "-k" switch for showing only kernel
5565 messages, mimicking dmesg output; in addition to "--user"
5566 and "--system" switches for showing only user's own logs
5569 * systemd-delta can now show information about drop-in
5570 snippets extending unit files.
5572 * libsystemd-bus has been substantially updated but is still
5573 not available as public API.
5575 * systemd will now look for the "debug" argument on the kernel
5576 command line and enable debug logging, similar to what
5577 "systemd.log_level=debug" already did before.
5579 * "systemctl set-default", "systemctl get-default" has been
5580 added to configure the default.target symlink, which
5581 controls what to boot into by default.
5583 * "systemctl set-log-level" has been added as a convenient
5584 way to raise and lower systemd logging threshold.
5586 * "systemd-analyze plot" will now show the time the various
5587 generators needed for execution, as well as information
5588 about the unit file loading.
5590 * libsystemd-journal gained a new sd_journal_open_files() call
5591 for opening specific journal files. journactl also gained a
5592 new switch to expose this new functionality. Previously we
5593 only supported opening all files from a directory, or all
5594 files from the system, as opening individual files only is
5595 racy due to journal file rotation.
5597 * systemd gained the new DefaultEnvironment= setting in
5598 /etc/systemd/system.conf to set environment variables for
5601 * If a privileged process logs a journal message with the
5602 OBJECT_PID= field set, then journald will automatically
5603 augment this with additional OBJECT_UID=, OBJECT_GID=,
5604 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5605 system services want to log events about specific client
5606 processes. journactl/systemctl has been updated to make use
5607 of this information if all log messages regarding a specific
5610 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5611 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5612 Reisner, David Coppa, David King, David Strauss, Eelco
5613 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5614 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5615 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5616 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5617 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5618 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5619 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5620 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5621 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5622 Łukasz Stelmach, 장동준
5626 * The Python bindings gained some minimal support for the APIs
5627 exposed by libsystemd-logind.
5629 * ConditionSecurity= gained support for detecting SMACK. Since
5630 this condition already supports SELinux and AppArmor we only
5631 miss IMA for this. Patches welcome!
5633 Contributions from: Karol Lewandowski, Lennart Poettering,
5634 Zbigniew Jędrzejewski-Szmek
5638 * systemd-nspawn will now create /etc/resolv.conf if
5639 necessary, before bind-mounting the host's file onto it.
5641 * systemd-nspawn will now store meta information about a
5642 container on the container's cgroup as extended attribute
5643 fields, including the root directory.
5645 * The cgroup hierarchy has been reworked in many ways. All
5646 objects any of the components systemd creates in the cgroup
5647 tree are now suffixed. More specifically, user sessions are
5648 now placed in cgroups suffixed with ".session", users in
5649 cgroups suffixed with ".user", and nspawn containers in
5650 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5651 names are now escaped in a simple scheme to avoid collision
5652 of userspace object names with kernel filenames. This work
5653 is preparation for making these objects relocatable in the
5654 cgroup tree, in order to allow easy resource partitioning of
5655 these objects without causing naming conflicts.
5657 * systemctl list-dependencies gained the new switches
5658 --plain, --reverse, --after and --before.
5660 * systemd-inhibit now shows the process name of processes that
5661 have taken an inhibitor lock.
5663 * nss-myhostname will now also resolve "localhost"
5664 implicitly. This makes /etc/hosts an optional file and
5665 nicely handles that on IPv6 ::1 maps to both "localhost" and
5668 * libsystemd-logind.so gained a new call
5669 sd_get_machine_names() to enumerate running containers and
5670 VMs (currently only supported by very new libvirt and
5671 nspawn). sd_login_monitor can now be used to watch
5672 VMs/containers coming and going.
5674 * .include is not allowed recursively anymore, and only in
5675 unit files. Usually it is better to use drop-in snippets in
5676 .d/*.conf anyway, as introduced with systemd 198.
5678 * systemd-analyze gained a new "critical-chain" command that
5679 determines the slowest chain of units run during system
5680 boot-up. It is very useful for tracking down where
5681 optimizing boot time is the most beneficial.
5683 * systemd will no longer allow manipulating service paths in
5684 the name=systemd:/system cgroup tree using ControlGroup= in
5685 units. (But is still fine with it in all other dirs.)
5687 * There's a new systemd-nspawn@.service service file that may
5688 be used to easily run nspawn containers as system
5689 services. With the container's root directory in
5690 /var/lib/container/foobar it is now sufficient to run
5691 "systemctl start systemd-nspawn@foobar.service" to boot it.
5693 * systemd-cgls gained a new parameter "--machine" to list only
5694 the processes within a certain container.
5696 * ConditionSecurity= now can check for "apparmor". We still
5697 are lacking checks for SMACK and IMA for this condition
5698 check though. Patches welcome!
5700 * A new configuration file /etc/systemd/sleep.conf has been
5701 added that may be used to configure which kernel operation
5702 systemd is supposed to execute when "suspend", "hibernate"
5703 or "hybrid-sleep" is requested. This makes the new kernel
5704 "freeze" state accessible to the user.
5706 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5707 the passed argument if applicable.
5709 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5710 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5711 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5712 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5713 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5714 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5715 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5720 * The output of 'systemctl list-jobs' got some polishing. The
5721 '--type=' argument may now be passed more than once. A new
5722 command 'systemctl list-sockets' has been added which shows
5723 a list of kernel sockets systemd is listening on with the
5724 socket units they belong to, plus the units these socket
5727 * The experimental libsystemd-bus library got substantial
5728 updates to work in conjunction with the (also experimental)
5729 kdbus kernel project. It works well enough to exchange
5730 messages with some sophistication. Note that kdbus is not
5731 ready yet, and the library is mostly an elaborate test case
5732 for now, and not installable.
5734 * systemd gained a new unit 'systemd-static-nodes.service'
5735 that generates static device nodes earlier during boot, and
5736 can run in conjunction with udev.
5738 * libsystemd-login gained a new call sd_pid_get_user_unit()
5739 to retrieve the user systemd unit a process is running
5740 in. This is useful for systems where systemd is used as
5743 * systemd-nspawn now places all containers in the new /machine
5744 top-level cgroup directory in the name=systemd
5745 hierarchy. libvirt will soon do the same, so that we get a
5746 uniform separation of /system, /user and /machine for system
5747 services, user processes and containers/virtual
5748 machines. This new cgroup hierarchy is also useful to stick
5749 stable names to specific container instances, which can be
5750 recognized later this way (this name may be controlled
5751 via systemd-nspawn's new -M switch). libsystemd-login also
5752 gained a new call sd_pid_get_machine_name() to retrieve the
5753 name of the container/VM a specific process belongs to.
5755 * bootchart can now store its data in the journal.
5757 * libsystemd-journal gained a new call
5758 sd_journal_add_conjunction() for AND expressions to the
5759 matching logic. This can be used to express more complex
5760 logical expressions.
5762 * journactl can now take multiple --unit= and --user-unit=
5765 * The cryptsetup logic now understands the "luks.key=" kernel
5766 command line switch for specifying a file to read the
5767 decryption key from. Also, if a configured key file is not
5768 found the tool will now automatically fall back to prompting
5771 * Python systemd.journal module was updated to wrap recently
5772 added functions from libsystemd-journal. The interface was
5773 changed to bring the low level interface in s.j._Reader
5774 closer to the C API, and the high level interface in
5775 s.j.Reader was updated to wrap and convert all data about
5778 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5779 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5780 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5781 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5782 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5783 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5787 * journalctl --update-catalog now understands a new --root=
5788 option to operate on catalogs found in a different root
5791 * During shutdown after systemd has terminated all running
5792 services a final killing loop kills all remaining left-over
5793 processes. We will now print the name of these processes
5794 when we send SIGKILL to them, since this usually indicates a
5797 * If /etc/crypttab refers to password files stored on
5798 configured mount points automatic dependencies will now be
5799 generated to ensure the specific mount is established first
5800 before the key file is attempted to be read.
5802 * 'systemctl status' will now show information about the
5803 network sockets a socket unit is listening on.
5805 * 'systemctl status' will also shown information about any
5806 drop-in configuration file for units. (Drop-In configuration
5807 files in this context are files such as
5808 /etc/systemd/systemd/foobar.service.d/*.conf)
5810 * systemd-cgtop now optionally shows summed up CPU times of
5811 cgroups. Press '%' while running cgtop to switch between
5812 percentage and absolute mode. This is useful to determine
5813 which cgroups use up the most CPU time over the entire
5814 runtime of the system. systemd-cgtop has also been updated
5815 to be 'pipeable' for processing with further shell tools.
5817 * 'hostnamectl set-hostname' will now allow setting of FQDN
5820 * The formatting and parsing of time span values has been
5821 changed. The parser now understands fractional expressions
5822 such as "5.5h". The formatter will now output fractional
5823 expressions for all time spans under 1min, i.e. "5.123456s"
5824 rather than "5s 123ms 456us". For time spans under 1s
5825 millisecond values are shown, for those under 1ms
5826 microsecond values are shown. This should greatly improve
5827 all time-related output of systemd.
5829 * libsystemd-login and libsystemd-journal gained new
5830 functions for querying the poll() events mask and poll()
5831 timeout value for integration into arbitrary event
5834 * localectl gained the ability to list available X11 keymaps
5835 (models, layouts, variants, options).
5837 * 'systemd-analyze dot' gained the ability to filter for
5838 specific units via shell-style globs, to create smaller,
5839 more useful graphs. I.e. it is now possible to create simple
5840 graphs of all the dependencies between only target units, or
5841 of all units that Avahi has dependencies with.
5843 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5844 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5845 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5846 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5847 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5848 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5849 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5853 * The boot-time readahead implementation for rotating media
5854 will now read the read-ahead data in multiple passes which
5855 consist of all read requests made in equidistant time
5856 intervals. This means instead of strictly reading read-ahead
5857 data in its physical order on disk we now try to find a
5858 middle ground between physical and access time order.
5860 * /etc/os-release files gained a new BUILD_ID= field for usage
5861 on operating systems that provide continuous builds of OS
5864 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5865 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5866 William Douglas, Zbigniew Jędrzejewski-Szmek
5870 * systemd-python gained an API exposing libsystemd-daemon.
5872 * The SMACK setup logic gained support for uploading CIPSO
5875 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5876 ReadOnlyDirectories= and InaccessibleDirectories= has
5877 changed. The private /tmp and /var/tmp directories are now
5878 shared by all processes of a service (which means
5879 ExecStartPre= may now leave data in /tmp that ExecStart= of
5880 the same service can still access). When a service is
5881 stopped its temporary directories are immediately deleted
5882 (normal clean-up with tmpfiles is still done in addition to
5885 * By default, systemd will now set a couple of sysctl
5886 variables in the kernel: the safe sysrq options are turned
5887 on, IP route verification is turned on, and source routing
5888 disabled. The recently added hardlink and softlink
5889 protection of the kernel is turned on. These settings should
5890 be reasonably safe, and good defaults for all new systems.
5892 * The predictable network naming logic may now be turned off
5893 with a new kernel command line switch: net.ifnames=0.
5895 * A new libsystemd-bus module has been added that implements a
5896 pretty complete D-Bus client library. For details see:
5898 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5900 * journald will now explicitly flush the journal files to disk
5901 at the latest 5min after each write. The file will then also
5902 be marked offline until the next write. This should increase
5903 reliability in case of a crash. The synchronization delay
5904 can be configured via SyncIntervalSec= in journald.conf.
5906 * There's a new remote-fs-setup.target unit that can be used
5907 to pull in specific services when at least one remote file
5908 system is to be mounted.
5910 * There are new targets timers.target and paths.target as
5911 canonical targets to pull user timer and path units in
5912 from. This complements sockets.target with a similar
5913 purpose for socket units.
5915 * libudev gained a new call udev_device_set_attribute_value()
5916 to set sysfs attributes of a device.
5918 * The udev daemon now sets the default number of worker
5919 processes executed in parallel based on the number of available
5920 CPUs instead of the amount of available RAM. This is supposed
5921 to provide a more reliable default and limit a too aggressive
5922 parallelism for setups with 1000s of devices connected.
5924 Contributions from: Auke Kok, Colin Walters, Cristian
5925 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5926 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5927 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5928 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5929 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5930 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5931 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5932 Zbigniew Jędrzejewski-Szmek
5936 * Configuration of unit files may now be extended via drop-in
5937 files without having to edit/override the unit files
5938 themselves. More specifically, if the administrator wants to
5939 change one value for a service file foobar.service he can
5940 now do so by dropping in a configuration snippet into
5941 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5942 will load all these snippets and apply them on top of the
5943 main unit configuration file, possibly extending or
5944 overriding its settings. Using these drop-in snippets is
5945 generally nicer than the two earlier options for changing
5946 unit files locally: copying the files from
5947 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5948 them there; or creating a new file in /etc/systemd/system/
5949 that incorporates the original one via ".include". Drop-in
5950 snippets into these .d/ directories can be placed in any
5951 directory systemd looks for units in, and the usual
5952 overriding semantics between /usr/lib, /etc and /run apply
5955 * Most unit file settings which take lists of items can now be
5956 reset by assigning the empty string to them. For example,
5957 normally, settings such as Environment=FOO=BAR append a new
5958 environment variable assignment to the environment block,
5959 each time they are used. By assigning Environment= the empty
5960 string the environment block can be reset to empty. This is
5961 particularly useful with the .d/*.conf drop-in snippets
5962 mentioned above, since this adds the ability to reset list
5963 settings from vendor unit files via these drop-ins.
5965 * systemctl gained a new "list-dependencies" command for
5966 listing the dependencies of a unit recursively.
5968 * Inhibitors are now honored and listed by "systemctl
5969 suspend", "systemctl poweroff" (and similar) too, not only
5970 GNOME. These commands will also list active sessions by
5973 * Resource limits (as exposed by the various control group
5974 controllers) can now be controlled dynamically at runtime
5975 for all units. More specifically, you can now use a command
5976 like "systemctl set-cgroup-attr foobar.service cpu.shares
5977 2000" to alter the CPU shares a specific service gets. These
5978 settings are stored persistently on disk, and thus allow the
5979 administrator to easily adjust the resource usage of
5980 services with a few simple commands. This dynamic resource
5981 management logic is also available to other programs via the
5982 bus. Almost any kernel cgroup attribute and controller is
5985 * systemd-vconsole-setup will now copy all font settings to
5986 all allocated VTs, where it previously applied them only to
5989 * libsystemd-login gained the new sd_session_get_tty() API
5992 * This release drops support for a few legacy or
5993 distribution-specific LSB facility names when parsing init
5994 scripts: $x-display-manager, $mail-transfer-agent,
5995 $mail-transport-agent, $mail-transfer-agent, $smtp,
5996 $null. Also, the mail-transfer-agent.target unit backing
5997 this has been removed. Distributions which want to retain
5998 compatibility with this should carry the burden for
5999 supporting this themselves and patch support for these back
6000 in, if they really need to. Also, the facilities $syslog and
6001 $local_fs are now ignored, since systemd does not support
6002 early-boot LSB init scripts anymore, and these facilities
6003 are implied anyway for normal services. syslog.target has
6006 * There are new bus calls on PID1's Manager object for
6007 cancelling jobs, and removing snapshot units. Previously,
6008 both calls were only available on the Job and Snapshot
6011 * systemd-journal-gatewayd gained SSL support.
6013 * The various "environment" files, such as /etc/locale.conf
6014 now support continuation lines with a backslash ("\") as
6015 last character in the line, similarly in style (but different)
6016 to how this is supported in shells.
6018 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6019 now implicitly appended to every log entry logged. systemctl
6020 has been updated to filter by this field when operating on a
6021 user systemd instance.
6023 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6024 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6025 the container. This makes it easier to boot unmodified
6026 Fedora systems in a container, which however still requires
6027 audit=0 to be passed on the kernel command line. Auditing in
6028 kernel and userspace is unfortunately still too broken in
6029 context of containers, hence we recommend compiling it out
6030 of the kernel or using audit=0. Hopefully this will be fixed
6031 one day for good in the kernel.
6033 * nspawn gained the new --bind= and --bind-ro= parameters to
6034 bind mount specific directories from the host into the
6037 * nspawn will now mount its own devpts file system instance
6038 into the container, in order not to leak pty devices from
6039 the host into the container.
6041 * systemd will now read the firmware boot time performance
6042 information from the EFI variables, if the used boot loader
6043 supports this, and takes it into account for boot performance
6044 analysis via "systemd-analyze". This is currently supported
6045 only in conjunction with Gummiboot, but could be supported
6046 by other boot loaders too. For details see:
6048 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6050 * A new generator has been added that automatically mounts the
6051 EFI System Partition (ESP) to /boot, if that directory
6052 exists, is empty, and no other file system has been
6053 configured to be mounted there.
6055 * logind will now send out PrepareForSleep(false) out
6056 unconditionally, after coming back from suspend. This may be
6057 used by applications as asynchronous notification for
6058 system resume events.
6060 * "systemctl unlock-sessions" has been added, that allows
6061 unlocking the screens of all user sessions at once, similar
6062 to how "systemctl lock-sessions" already locked all users
6063 sessions. This is backed by a new D-Bus call UnlockSessions().
6065 * "loginctl seat-status" will now show the master device of a
6066 seat. (i.e. the device of a seat that needs to be around for
6067 the seat to be considered available, usually the graphics
6070 * tmpfiles gained a new "X" line type, that allows
6071 configuration of files and directories (with wildcards) that
6072 shall be excluded from automatic cleanup ("aging").
6074 * udev default rules set the device node permissions now only
6075 at "add" events, and do not change them any longer with a
6076 later "change" event.
6078 * The log messages for lid events and power/sleep keypresses
6079 now carry a message ID.
6081 * We now have a substantially larger unit test suite, but this
6082 continues to be work in progress.
6084 * udevadm hwdb gained a new --root= parameter to change the
6085 root directory to operate relative to.
6087 * logind will now issue a background sync() request to the kernel
6088 early at shutdown, so that dirty buffers are flushed to disk early
6089 instead of at the last moment, in order to optimize shutdown
6092 * A new bootctl tool has been added that is an interface for
6093 certain boot loader operations. This is currently a preview
6094 and is likely to be extended into a small mechanism daemon
6095 like timedated, localed, hostnamed, and can be used by
6096 graphical UIs to enumerate available boot options, and
6097 request boot into firmware operations.
6099 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6100 the rest of the package. It also has been updated to work
6101 correctly in initrds.
6103 * Policykit previously has been runtime optional, and is now
6104 also compile time optional via a configure switch.
6106 * systemd-analyze has been reimplemented in C. Also "systemctl
6107 dot" has moved into systemd-analyze.
6109 * "systemctl status" with no further parameters will now print
6110 the status of all active or failed units.
6112 * Operations such as "systemctl start" can now be executed
6113 with a new mode "--irreversible" which may be used to queue
6114 operations that cannot accidentally be reversed by a later
6115 job queuing. This is by default used to make shutdown
6116 requests more robust.
6118 * The Python API of systemd now gained a new module for
6119 reading journal files.
6121 * A new tool kernel-install has been added that can install
6122 kernel images according to the Boot Loader Specification:
6124 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6126 * Boot time console output has been improved to provide
6127 animated boot time output for hanging jobs.
6129 * A new tool systemd-activate has been added which can be used
6130 to test socket activation with, directly from the command
6131 line. This should make it much easier to test and debug
6132 socket activation in daemons.
6134 * journalctl gained a new "--reverse" (or -r) option to show
6135 journal output in reverse order (i.e. newest line first).
6137 * journalctl gained a new "--pager-end" (or -e) option to jump
6138 to immediately jump to the end of the journal in the
6139 pager. This is only supported in conjunction with "less".
6141 * journalctl gained a new "--user-unit=" option, that works
6142 similarly to "--unit=" but filters for user units rather than
6145 * A number of unit files to ease adoption of systemd in
6146 initrds has been added. This moves some minimal logic from
6147 the various initrd implementations into systemd proper.
6149 * The journal files are now owned by a new group
6150 "systemd-journal", which exists specifically to allow access
6151 to the journal, and nothing else. Previously, we used the
6152 "adm" group for that, which however possibly covers more
6153 than just journal/log file access. This new group is now
6154 already used by systemd-journal-gatewayd to ensure this
6155 daemon gets access to the journal files and as little else
6156 as possible. Note that "make install" will also set FS ACLs
6157 up for /var/log/journal to give "adm" and "wheel" read
6158 access to it, in addition to "systemd-journal" which owns
6159 the journal files. We recommend that packaging scripts also
6160 add read access to "adm" + "wheel" to /var/log/journal, and
6161 all existing/future journal files. To normal users and
6162 administrators little changes, however packagers need to
6163 ensure to create the "systemd-journal" system group at
6164 package installation time.
6166 * The systemd-journal-gatewayd now runs as unprivileged user
6167 systemd-journal-gateway:systemd-journal-gateway. Packaging
6168 scripts need to create these system user/group at
6171 * timedated now exposes a new boolean property CanNTP that
6172 indicates whether a local NTP service is available or not.
6174 * systemd-detect-virt will now also detect xen PVs
6176 * The pstore file system is now mounted by default, if it is
6179 * In addition to the SELinux and IMA policies we will now also
6180 load SMACK policies at early boot.
6182 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6183 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6184 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6185 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6186 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6187 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6188 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6189 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6190 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6191 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6192 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6193 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6194 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6195 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6199 * Timer units now support calendar time events in addition to
6200 monotonic time events. That means you can now trigger a unit
6201 based on a calendar time specification such as "Thu,Fri
6202 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6203 or fifth day of any month of the year 2013, given that it is
6204 a thursday or friday. This brings timer event support
6205 considerably closer to cron's capabilities. For details on
6206 the supported calendar time specification language see
6209 * udev now supports a number of different naming policies for
6210 network interfaces for predictable names, and a combination
6211 of these policies is now the default. Please see this wiki
6212 document for details:
6214 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6216 * Auke Kok's bootchart implementation has been added to the
6217 systemd tree. It is an optional component that can graph the
6218 boot in quite some detail. It is one of the best bootchart
6219 implementations around and minimal in its code and
6222 * nss-myhostname has been integrated into the systemd source
6223 tree. nss-myhostname guarantees that the local hostname
6224 always stays resolvable via NSS. It has been a weak
6225 requirement of systemd-hostnamed since a long time, and
6226 since its code is actually trivial we decided to just
6227 include it in systemd's source tree. It can be turned off
6228 with a configure switch.
6230 * The read-ahead logic is now capable of properly detecting
6231 whether a btrfs file system is on SSD or rotating media, in
6232 order to optimize the read-ahead scheme. Previously, it was
6233 only capable of detecting this on traditional file systems
6236 * In udev, additional device properties are now read from the
6237 IAB in addition to the OUI database. Also, Bluetooth company
6238 identities are attached to the devices as well.
6240 * In service files %U may be used as specifier that is
6241 replaced by the configured user name of the service.
6243 * nspawn may now be invoked without a controlling TTY. This
6244 makes it suitable for invocation as its own service. This
6245 may be used to set up a simple containerized server system
6246 using only core OS tools.
6248 * systemd and nspawn can now accept socket file descriptors
6249 when they are started for socket activation. This enables
6250 implementation of socket activated nspawn
6251 containers. i.e. think about autospawning an entire OS image
6252 when the first SSH or HTTP connection is received. We expect
6253 that similar functionality will also be added to libvirt-lxc
6256 * journalctl will now suppress ANSI color codes when
6257 presenting log data.
6259 * systemctl will no longer show control group information for
6260 a unit if the control group is empty anyway.
6262 * logind can now automatically suspend/hibernate/shutdown the
6265 * /etc/machine-info and hostnamed now also expose the chassis
6266 type of the system. This can be used to determine whether
6267 the local system is a laptop, desktop, handset or
6268 tablet. This information may either be configured by the
6269 user/vendor or is automatically determined from ACPI and DMI
6270 information if possible.
6272 * A number of PolicyKit actions are now bound together with
6273 "imply" rules. This should simplify creating UIs because
6274 many actions will now authenticate similar ones as well.
6276 * Unit files learnt a new condition ConditionACPower= which
6277 may be used to conditionalize a unit depending on whether an
6278 AC power source is connected or not, of whether the system
6279 is running on battery power.
6281 * systemctl gained a new "is-failed" verb that may be used in
6282 shell scripts and suchlike to check whether a specific unit
6283 is in the "failed" state.
6285 * The EnvironmentFile= setting in unit files now supports file
6286 globbing, and can hence be used to easily read a number of
6287 environment files at once.
6289 * systemd will no longer detect and recognize specific
6290 distributions. All distribution-specific #ifdeffery has been
6291 removed, systemd is now fully generic and
6292 distribution-agnostic. Effectively, not too much is lost as
6293 a lot of the code is still accessible via explicit configure
6294 switches. However, support for some distribution specific
6295 legacy configuration file formats has been dropped. We
6296 recommend distributions to simply adopt the configuration
6297 files everybody else uses now and convert the old
6298 configuration from packaging scripts. Most distributions
6299 already did that. If that's not possible or desirable,
6300 distributions are welcome to forward port the specific
6301 pieces of code locally from the git history.
6303 * When logging a message about a unit systemd will now always
6304 log the unit name in the message meta data.
6306 * localectl will now also discover system locale data that is
6307 not stored in locale archives, but directly unpacked.
6309 * logind will no longer unconditionally use framebuffer
6310 devices as seat masters, i.e. as devices that are required
6311 to be existing before a seat is considered preset. Instead,
6312 it will now look for all devices that are tagged as
6313 "seat-master" in udev. By default, framebuffer devices will
6314 be marked as such, but depending on local systems, other
6315 devices might be marked as well. This may be used to
6316 integrate graphics cards using closed source drivers (such
6317 as NVidia ones) more nicely into logind. Note however, that
6318 we recommend using the open source NVidia drivers instead,
6319 and no udev rules for the closed-source drivers will be
6320 shipped from us upstream.
6322 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6323 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6324 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6325 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6326 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6327 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6328 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6329 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6330 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6331 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6332 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6333 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6338 * udev gained support for loading additional device properties
6339 from an indexed database that is keyed by vendor/product IDs
6340 and similar device identifiers. For the beginning this
6341 "hwdb" is populated with data from the well-known PCI and
6342 USB database, but also includes PNP, ACPI and OID data. In
6343 the longer run this indexed database shall grow into
6344 becoming the one central database for non-essential
6345 userspace device metadata. Previously, data from the PCI/USB
6346 database was only attached to select devices, since the
6347 lookup was a relatively expensive operation due to O(n) time
6348 complexity (with n being the number of entries in the
6349 database). Since this is now O(1), we decided to add in this
6350 data for all devices where this is available, by
6351 default. Note that the indexed database needs to be rebuilt
6352 when new data files are installed. To achieve this you need
6353 to update your packaging scripts to invoke "udevadm hwdb
6354 --update" after installation of hwdb data files. For
6355 RPM-based distributions we introduced the new
6356 %udev_hwdb_update macro for this purpose.
6358 * The Journal gained support for the "Message Catalog", an
6359 indexed database to link up additional information with
6360 journal entries. For further details please check:
6362 https://www.freedesktop.org/wiki/Software/systemd/catalog
6364 The indexed message catalog database also needs to be
6365 rebuilt after installation of message catalog files. Use
6366 "journalctl --update-catalog" for this. For RPM-based
6367 distributions we introduced the %journal_catalog_update
6368 macro for this purpose.
6370 * The Python Journal bindings gained support for the standard
6371 Python logging framework.
6373 * The Journal API gained new functions for checking whether
6374 the underlying file system of a journal file is capable of
6375 properly reporting file change notifications, or whether
6376 applications that want to reflect journal changes "live"
6377 need to recheck journal files continuously in appropriate
6380 * It is now possible to set the "age" field for tmpfiles
6381 entries to 0, indicating that files matching this entry
6382 shall always be removed when the directories are cleaned up.
6384 * coredumpctl gained a new "gdb" verb which invokes gdb
6385 right-away on the selected coredump.
6387 * There's now support for "hybrid sleep" on kernels that
6388 support this, in addition to "suspend" and "hibernate". Use
6389 "systemctl hybrid-sleep" to make use of this.
6391 * logind's HandleSuspendKey= setting (and related settings)
6392 now gained support for a new "lock" setting to simply
6393 request the screen lock on all local sessions, instead of
6394 actually executing a suspend or hibernation.
6396 * systemd will now mount the EFI variables file system by
6399 * Socket units now gained support for configuration of the
6400 SMACK security label.
6402 * timedatectl will now output the time of the last and next
6403 daylight saving change.
6405 * We dropped support for various legacy and distro-specific
6406 concepts, such as insserv, early-boot SysV services
6407 (i.e. those for non-standard runlevels such as 'b' or 'S')
6408 or ArchLinux /etc/rc.conf support. We recommend the
6409 distributions who still need support this to either continue
6410 to maintain the necessary patches downstream, or find a
6411 different solution. (Talk to us if you have questions!)
6413 * Various systemd components will now bypass PolicyKit checks
6414 for root and otherwise handle properly if PolicyKit is not
6415 found to be around. This should fix most issues for
6416 PolicyKit-less systems. Quite frankly this should have been
6417 this way since day one. It is absolutely our intention to
6418 make systemd work fine on PolicyKit-less systems, and we
6419 consider it a bug if something does not work as it should if
6420 PolicyKit is not around.
6422 * For embedded systems it is now possible to build udev and
6423 systemd without blkid and/or kmod support.
6425 * "systemctl switch-root" is now capable of switching root
6426 more than once. I.e. in addition to transitions from the
6427 initrd to the host OS it is now possible to transition to
6428 further OS images from the host. This is useful to implement
6429 offline updating tools.
6431 * Various other additions have been made to the RPM macros
6432 shipped with systemd. Use %udev_rules_update() after
6433 installing new udev rules files. %_udevhwdbdir,
6434 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6435 %_sysctldir are now available which resolve to the right
6436 directories for packages to place various data files in.
6438 * journalctl gained the new --full switch (in addition to
6439 --all, to disable ellipsation for long messages.
6441 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6442 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6443 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6444 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6445 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6446 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6447 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6448 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6449 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6453 * journalctl gained new --since= and --until= switches to
6454 filter by time. It also now supports nice filtering for
6455 units via --unit=/-u.
6457 * Type=oneshot services may use ExecReload= and do the
6460 * The journal daemon now supports time-based rotation and
6461 vacuuming, in addition to the usual disk-space based
6464 * The journal will now index the available field values for
6465 each field name. This enables clients to show pretty drop
6466 downs of available match values when filtering. The bash
6467 completion of journalctl has been updated
6468 accordingly. journalctl gained a new switch -F to list all
6469 values a certain field takes in the journal database.
6471 * More service events are now written as structured messages
6472 to the journal, and made recognizable via message IDs.
6474 * The timedated, localed and hostnamed mini-services which
6475 previously only provided support for changing time, locale
6476 and hostname settings from graphical DEs such as GNOME now
6477 also have a minimal (but very useful) text-based client
6478 utility each. This is probably the nicest way to changing
6479 these settings from the command line now, especially since
6480 it lists available options and is fully integrated with bash
6483 * There's now a new tool "systemd-coredumpctl" to list and
6484 extract coredumps from the journal.
6486 * We now install a README each in /var/log/ and
6487 /etc/rc.d/init.d explaining where the system logs and init
6488 scripts went. This hopefully should help folks who go to
6489 that dirs and look into the otherwise now empty void and
6490 scratch their heads.
6492 * When user-services are invoked (by systemd --user) the
6493 $MANAGERPID env var is set to the PID of systemd.
6495 * SIGRTMIN+24 when sent to a --user instance will now result
6496 in immediate termination of systemd.
6498 * gatewayd received numerous feature additions such as a
6499 "follow" mode, for live syncing and filtering.
6501 * browse.html now allows filtering and showing detailed
6502 information on specific entries. Keyboard navigation and
6503 mouse screen support has been added.
6505 * gatewayd/journalctl now supports HTML5/JSON
6506 Server-Sent-Events as output.
6508 * The SysV init script compatibility logic will now
6509 heuristically determine whether a script supports the
6510 "reload" verb, and only then make this available as
6513 * "systemctl status --follow" has been removed, use "journalctl
6516 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6517 have been removed since they are hardly useful to be
6520 * And I'd like to take the opportunity to specifically mention
6521 Zbigniew for his great contributions. Zbigniew, you rock!
6523 Contributions from: Andrew Eikum, Christian Hesse, Colin
6524 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6525 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6526 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6527 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6528 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6529 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6533 * If /etc/vconsole.conf is non-existent or empty we will no
6534 longer load any console font or key map at boot by
6535 default. Instead the kernel defaults will be left
6536 intact. This is definitely the right thing to do, as no
6537 configuration should mean no configuration, and hard-coding
6538 font names that are different on all archs is probably a bad
6539 idea. Also, the kernel default key map and font should be
6540 good enough for most cases anyway, and mostly identical to
6541 the userspace fonts/key maps we previously overloaded them
6542 with. If distributions want to continue to default to a
6543 non-kernel font or key map they should ship a default
6544 /etc/vconsole.conf with the appropriate contents.
6546 Contributions from: Colin Walters, Daniel J Walsh, Dave
6547 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6548 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6552 * journalctl gained a new --cursor= switch to show entries
6553 starting from the specified location in the journal.
6555 * We now enforce a size limit on journal entry fields exported
6556 with "-o json" in journalctl. Fields larger than 4K will be
6557 assigned null. This can be turned off with --all.
6559 * An (optional) journal gateway daemon is now available as
6560 "systemd-journal-gatewayd.service". This service provides
6561 access to the journal via HTTP and JSON. This functionality
6562 will be used to implement live log synchronization in both
6563 pull and push modes, but has various other users too, such
6564 as easy log access for debugging of embedded devices. Right
6565 now it is already useful to retrieve the journal via HTTP:
6567 # systemctl start systemd-journal-gatewayd.service
6568 # wget http://localhost:19531/entries
6570 This will download the journal contents in a
6571 /var/log/messages compatible format. The same as JSON:
6573 # curl -H"Accept: application/json" http://localhost:19531/entries
6575 This service is also accessible via a web browser where a
6576 single static HTML5 app is served that uses the JSON logic
6577 to enable the user to do some basic browsing of the
6578 journal. This will be extended later on. Here's an example
6579 screenshot of this app in its current state:
6581 http://0pointer.de/public/journal-gatewayd
6583 Contributions from: Kay Sievers, Lennart Poettering, Robert
6584 Milasan, Tom Gundersen
6588 * The bash completion logic is now available for journalctl
6591 * We do not mount the "cpuset" controller anymore together with
6592 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6593 started if no parameters are assigned to it. "cpuset" hence
6594 broke code that assumed it could create "cpu" groups and
6597 * journalctl -f will now subscribe to terminal size changes,
6598 and line break accordingly.
6600 Contributions from: Dave Reisner, Kay Sievers, Lennart
6601 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6605 * nspawn will now create a symlink /etc/localtime in the
6606 container environment, copying the host's timezone
6607 setting. Previously this has been done via a bind mount, but
6608 since symlinks cannot be bind mounted this has now been
6609 changed to create/update the appropriate symlink.
6611 * journalctl -n's line number argument is now optional, and
6612 will default to 10 if omitted.
6614 * journald will now log the maximum size the journal files may
6615 take up on disk. This is particularly useful if the default
6616 built-in logic of determining this parameter from the file
6617 system size is used. Use "systemctl status
6618 systemd-journald.service" to see this information.
6620 * The multi-seat X wrapper tool has been stripped down. As X
6621 is now capable of enumerating graphics devices via udev in a
6622 seat-aware way the wrapper is not strictly necessary
6623 anymore. A stripped down temporary stop-gap is still shipped
6624 until the upstream display managers have been updated to
6625 fully support the new X logic. Expect this wrapper to be
6626 removed entirely in one of the next releases.
6628 * HandleSleepKey= in logind.conf has been split up into
6629 HandleSuspendKey= and HandleHibernateKey=. The old setting
6630 is not available anymore. X11 and the kernel are
6631 distinguishing between these keys and we should too. This
6632 also means the inhibition lock for these keys has been split
6635 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6636 Poettering, Lukas Nykryn, Václav Pavlín
6640 * Whenever a unit changes state we will now log this to the
6641 journal and show along the unit's own log output in
6644 * ConditionPathIsMountPoint= can now properly detect bind
6645 mount points too. (Previously, a bind mount of one file
6646 system to another place in the same file system could not be
6647 detected as mount, since they shared struct stat's st_dev
6650 * We will now mount the cgroup controllers cpu, cpuacct,
6651 cpuset and the controllers net_cls, net_prio together by
6654 * nspawn containers will now have a virtualized boot
6655 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6656 over with a randomized ID at container initialization). This
6657 has the effect of making "journalctl -b" do the right thing
6660 * The JSON output journal serialization has been updated not
6661 to generate "endless" list objects anymore, but rather one
6662 JSON object per line. This is more in line how most JSON
6663 parsers expect JSON objects. The new output mode
6664 "json-pretty" has been added to provide similar output, but
6665 neatly aligned for readability by humans.
6667 * We dropped all explicit sync() invocations in the shutdown
6668 code. The kernel does this implicitly anyway in the kernel
6669 reboot() syscall. halt(8)'s -n option is now a compatibility
6672 * We now support virtualized reboot() in containers, as
6673 supported by newer kernels. We will fall back to exit() if
6674 CAP_SYS_REBOOT is not available to the container. Also,
6675 nspawn makes use of this now and will actually reboot the
6676 container if the containerized OS asks for that.
6678 * journalctl will only show local log output by default
6679 now. Use --merge (-m) to show remote log output, too.
6681 * libsystemd-journal gained the new sd_journal_get_usage()
6682 call to determine the current disk usage of all journal
6683 files. This is exposed in the new "journalctl --disk-usage"
6686 * journald gained a new configuration setting SplitMode= in
6687 journald.conf which may be used to control how user journals
6688 are split off. See journald.conf(5) for details.
6690 * A new condition type ConditionFileNotEmpty= has been added.
6692 * tmpfiles' "w" lines now support file globbing, to write
6693 multiple files at once.
6695 * We added Python bindings for the journal submission
6696 APIs. More Python APIs for a number of selected APIs will
6697 likely follow. Note that we intend to add native bindings
6698 only for the Python language, as we consider it common
6699 enough to deserve bindings shipped within systemd. There are
6700 various projects outside of systemd that provide bindings
6701 for languages such as PHP or Lua.
6703 * Many conditions will now resolve specifiers such as %i. In
6704 addition, PathChanged= and related directives of .path units
6705 now support specifiers as well.
6707 * There's now a new RPM macro definition for the system preset
6710 * journald will now warn if it ca not forward a message to the
6711 syslog daemon because its socket is full.
6713 * timedated will no longer write or process /etc/timezone,
6714 except on Debian. As we do not support late mounted /usr
6715 anymore /etc/localtime always being a symlink is now safe,
6716 and hence the information in /etc/timezone is not necessary
6719 * logind will now always reserve one VT for a text getty (VT6
6720 by default). Previously if more than 6 X sessions where
6721 started they took up all the VTs with auto-spawned gettys,
6722 so that no text gettys were available anymore.
6724 * udev will now automatically inform the btrfs kernel logic
6725 about btrfs RAID components showing up. This should make
6726 simple hotplug based btrfs RAID assembly work.
6728 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6729 (but not for its children which will stay at the kernel
6730 default). This should allow setups with a lot more listening
6733 * systemd will now always pass the configured timezone to the
6734 kernel at boot. timedated will do the same when the timezone
6737 * logind's inhibition logic has been updated. By default,
6738 logind will now handle the lid switch, the power and sleep
6739 keys all the time, even in graphical sessions. If DEs want
6740 to handle these events on their own they should take the new
6741 handle-power-key, handle-sleep-key and handle-lid-switch
6742 inhibitors during their runtime. A simple way to achieve
6743 that is to invoke the DE wrapped in an invocation of:
6745 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6747 * Access to unit operations is now checked via SELinux taking
6748 the unit file label and client process label into account.
6750 * systemd will now notify the administrator in the journal
6751 when he over-mounts a non-empty directory.
6753 * There are new specifiers that are resolved in unit files,
6754 for the host name (%H), the machine ID (%m) and the boot ID
6757 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6758 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6759 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6760 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6761 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6762 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6763 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6767 * Support for reading structured kernel messages from
6768 /dev/kmsg has now been added and is enabled by default.
6770 * Support for reading kernel messages from /proc/kmsg has now
6771 been removed. If you want kernel messages in the journal
6772 make sure to run a recent kernel (>= 3.5) that supports
6773 reading structured messages from /dev/kmsg (see
6774 above). /proc/kmsg is now exclusive property of classic
6775 syslog daemons again.
6777 * The libudev API gained the new
6778 udev_device_new_from_device_id() call.
6780 * The logic for file system namespace (ReadOnlyDirectory=,
6781 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6782 require pivot_root() anymore. This means fewer temporary
6783 directories are created below /tmp for this feature.
6785 * nspawn containers will now see and receive all submounts
6786 made on the host OS below the root file system of the
6789 * Forward Secure Sealing is now supported for Journal files,
6790 which provide cryptographical sealing of journal files so
6791 that attackers cannot alter log history anymore without this
6792 being detectable. Lennart will soon post a blog story about
6793 this explaining it in more detail.
6795 * There are two new service settings RestartPreventExitStatus=
6796 and SuccessExitStatus= which allow configuration of exit
6797 status (exit code or signal) which will be excepted from the
6798 restart logic, resp. consider successful.
6800 * journalctl gained the new --verify switch that can be used
6801 to check the integrity of the structure of journal files and
6802 (if Forward Secure Sealing is enabled) the contents of
6805 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6806 and similar symlinks pre-created. This makes running shells
6807 as container init process a lot more fun.
6809 * The fstab support can now handle PARTUUID= and PARTLABEL=
6812 * A new ConditionHost= condition has been added to match
6813 against the hostname (with globs) and machine ID. This is
6814 useful for clusters where a single OS image is used to
6815 provision a large number of hosts which shall run slightly
6816 different sets of services.
6818 * Services which hit the restart limit will now be placed in a
6821 Contributions from: Bertram Poettering, Dave Reisner, Huang
6822 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6823 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6827 * When running in --user mode systemd will now become a
6828 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6829 tree a lot more organized.
6831 * A new PartOf= unit dependency type has been introduced that
6832 may be used to group services in a natural way.
6834 * "systemctl enable" may now be used to enable instances of
6837 * journalctl now prints error log levels in red, and
6838 warning/notice log levels in bright white. It also supports
6839 filtering by log level now.
6841 * cgtop gained a new -n switch (similar to top), to configure
6842 the maximum number of iterations to run for. It also gained
6843 -b, to run in batch mode (accepting no input).
6845 * The suffix ".service" may now be omitted on most systemctl
6846 command lines involving service unit names.
6848 * There's a new bus call in logind to lock all sessions, as
6849 well as a loginctl verb for it "lock-sessions".
6851 * libsystemd-logind.so gained a new call sd_journal_perror()
6852 that works similar to libc perror() but logs to the journal
6853 and encodes structured information about the error number.
6855 * /etc/crypttab entries now understand the new keyfile-size=
6858 * shutdown(8) now can send a (configurable) wall message when
6859 a shutdown is cancelled.
6861 * The mount propagation mode for the root file system will now
6862 default to "shared", which is useful to make containers work
6863 nicely out-of-the-box so that they receive new mounts from
6864 the host. This can be undone locally by running "mount
6865 --make-rprivate /" if needed.
6867 * The prefdm.service file has been removed. Distributions
6868 should maintain this unit downstream if they intend to keep
6869 it around. However, we recommend writing normal unit files
6870 for display managers instead.
6872 * Since systemd is a crucial part of the OS we will now
6873 default to a number of compiler switches that improve
6874 security (hardening) such as read-only relocations, stack
6875 protection, and suchlike.
6877 * The TimeoutSec= setting for services is now split into
6878 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6879 of individual time outs for the start and the stop phase of
6882 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6883 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6884 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6885 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6886 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6887 Gundersen, Zbigniew Jędrzejewski-Szmek
6891 * The journal and id128 C APIs are now fully documented as man
6894 * Extra safety checks have been added when transitioning from
6895 the initial RAM disk to the main system to avoid accidental
6898 * /etc/crypttab entries now understand the new keyfile-offset=
6901 * systemctl -t can now be used to filter by unit load state.
6903 * The journal C API gained the new sd_journal_wait() call to
6904 make writing synchronous journal clients easier.
6906 * journalctl gained the new -D switch to show journals from a
6909 * journalctl now displays a special marker between log
6910 messages of two different boots.
6912 * The journal is now explicitly flushed to /var via a service
6913 systemd-journal-flush.service, rather than implicitly simply
6914 by seeing /var/log/journal to be writable.
6916 * journalctl (and the journal C APIs) can now match for much
6917 more complex expressions, with alternatives and
6920 * When transitioning from the initial RAM disk to the main
6921 system we will now kill all processes in a killing spree to
6922 ensure no processes stay around by accident.
6924 * Three new specifiers may be used in unit files: %u, %h, %s
6925 resolve to the user name, user home directory resp. user
6926 shell. This is useful for running systemd user instances.
6928 * We now automatically rotate journal files if their data
6929 object hash table gets a fill level > 75%. We also size the
6930 hash table based on the configured maximum file size. This
6931 together should lower hash collisions drastically and thus
6932 speed things up a bit.
6934 * journalctl gained the new "--header" switch to introspect
6935 header data of journal files.
6937 * A new setting SystemCallFilters= has been added to services
6938 which may be used to apply blacklists or whitelists to
6939 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6941 * nspawn gained a new --link-journal= switch (and quicker: -j)
6942 to link the container journal with the host. This makes it
6943 very easy to centralize log viewing on the host for all
6944 guests while still keeping the journal files separated.
6946 * Many bugfixes and optimizations
6948 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6949 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6950 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6955 * Several tools now understand kernel command line arguments,
6956 which are only read when run in an initial RAM disk. They
6957 usually follow closely their normal counterparts, but are
6960 * There's a new tool to analyze the readahead files that are
6961 automatically generated at boot. Use:
6963 /usr/lib/systemd/systemd-readahead analyze /.readahead
6965 * We now provide an early debug shell on tty9 if this enabled. Use:
6967 systemctl enable debug-shell.service
6969 * All plymouth related units have been moved into the Plymouth
6970 package. Please make sure to upgrade your Plymouth version
6973 * systemd-tmpfiles now supports getting passed the basename of
6974 a configuration file only, in which case it will look for it
6975 in all appropriate directories automatically.
6977 * udevadm info now takes a /dev or /sys path as argument, and
6978 does the right thing. Example:
6980 udevadm info /dev/sda
6981 udevadm info /sys/class/block/sda
6983 * systemctl now prints a warning if a unit is stopped but a
6984 unit that might trigger it continues to run. Example: a
6985 service is stopped but the socket that activates it is left
6988 * "systemctl status" will now mention if the log output was
6989 shortened due to rotation since a service has been started.
6991 * The journal API now exposes functions to determine the
6992 "cutoff" times due to rotation.
6994 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6995 immediately flushing of runtime logs to /var if possible,
6996 resp. for triggering immediate rotation of the journal
6999 * It is now considered an error if a service is attempted to
7000 be stopped that is not loaded.
7002 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7004 * systemd-analyze now supports Python 3
7006 * tmpfiles now supports cleaning up directories via aging
7007 where the first level dirs are always kept around but
7008 directories beneath it automatically aged. This is enabled
7009 by prefixing the age field with '~'.
7011 * Seat objects now expose CanGraphical, CanTTY properties
7012 which is required to deal with very fast bootups where the
7013 display manager might be running before the graphics drivers
7014 completed initialization.
7016 * Seat objects now expose a State property.
7018 * We now include RPM macros for service enabling/disabling
7019 based on the preset logic. We recommend RPM based
7020 distributions to make use of these macros if possible. This
7021 makes it simpler to reuse RPM spec files across
7024 * We now make sure that the collected systemd unit name is
7025 always valid when services log to the journal via
7028 * There's a new man page kernel-command-line(7) detailing all
7029 command line options we understand.
7031 * The fstab generator may now be disabled at boot by passing
7032 fstab=0 on the kernel command line.
7034 * A new kernel command line option modules-load= is now understood
7035 to load a specific kernel module statically, early at boot.
7037 * Unit names specified on the systemctl command line are now
7038 automatically escaped as needed. Also, if file system or
7039 device paths are specified they are automatically turned
7040 into the appropriate mount or device unit names. Example:
7042 systemctl status /home
7043 systemctl status /dev/sda
7045 * The SysVConsole= configuration option has been removed from
7046 system.conf parsing.
7048 * The SysV search path is no longer exported on the D-Bus
7051 * The Names= option has been removed from unit file parsing.
7053 * There's a new man page bootup(7) detailing the boot process.
7055 * Every unit and every generator we ship with systemd now
7056 comes with full documentation. The self-explanatory boot is
7059 * A couple of services gained "systemd-" prefixes in their
7060 name if they wrap systemd code, rather than only external
7061 code. Among them fsck@.service which is now
7062 systemd-fsck@.service.
7064 * The HaveWatchdog property has been removed from the D-Bus
7067 * systemd.confirm_spawn= on the kernel command line should now
7070 * There's a new man page crypttab(5) which details all options
7071 we actually understand.
7073 * systemd-nspawn gained a new --capability= switch to pass
7074 additional capabilities to the container.
7076 * timedated will now read known NTP implementation unit names
7077 from /usr/lib/systemd/ntp-units.d/*.list,
7078 systemd-timedated-ntp.target has been removed.
7080 * journalctl gained a new switch "-b" that lists log data of
7081 the current boot only.
7083 * The notify socket is in the abstract namespace again, in
7084 order to support daemons which chroot() at start-up.
7086 * There is a new Storage= configuration option for journald
7087 which allows configuration of where log data should go. This
7088 also provides a way to disable journal logging entirely, so
7089 that data collected is only forwarded to the console, the
7090 kernel log buffer or another syslog implementation.
7092 * Many bugfixes and optimizations
7094 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7095 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7096 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7097 Shawn Landden, Tom Gundersen
7101 * "systemctl help <unit>" now shows the man page if one is
7104 * Several new man pages have been added.
7106 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7107 MaxLevelConsole= can now be specified in
7108 journald.conf. These options allow reducing the amount of
7109 data stored on disk or forwarded by the log level.
7111 * TimerSlackNSec= can now be specified in system.conf for
7112 PID1. This allows system-wide power savings.
7114 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7115 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7120 * logind is now capable of (optionally) handling power and
7121 sleep keys as well as the lid switch.
7123 * journalctl now understands the syntax "journalctl
7124 /usr/bin/avahi-daemon" to get all log output of a specific
7127 * CapabilityBoundingSet= in system.conf now also influences
7128 the capability bound set of usermode helpers of the kernel.
7130 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7131 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7132 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7133 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7137 * Note that we skipped 139 releases here in order to set the
7138 new version to something that is greater than both udev's
7139 and systemd's most recent version number.
7141 * udev: all udev sources are merged into the systemd source tree now.
7142 All future udev development will happen in the systemd tree. It
7143 is still fully supported to use the udev daemon and tools without
7144 systemd running, like in initramfs or other init systems. Building
7145 udev though, will require the *build* of the systemd tree, but
7146 udev can be properly *run* without systemd.
7148 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7149 should be used to create dead device nodes as workarounds for broken
7152 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7153 no longer supported. udev_monitor_new_from_netlink() needs to be
7154 used to subscribe to events.
7156 * udev: when udevd is started by systemd, processes which are left
7157 behind by forking them off of udev rules, are unconditionally cleaned
7158 up and killed now after the event handling has finished. Services or
7159 daemons must be started as systemd services. Services can be
7160 pulled-in by udev to get started, but they can no longer be directly
7161 forked by udev rules.
7163 * udev: the daemon binary is called systemd-udevd now and installed
7164 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7165 to adapt to that, create symlink, or rename the binary after building
7168 * libudev no longer provides these symbols:
7169 udev_monitor_from_socket()
7170 udev_queue_get_failed_list_entry()
7171 udev_get_{dev,sys,run}_path()
7172 The versions number was bumped and symbol versioning introduced.
7174 * systemd-loginctl and systemd-journalctl have been renamed
7175 to loginctl and journalctl to match systemctl.
7177 * The config files: /etc/systemd/systemd-logind.conf and
7178 /etc/systemd/systemd-journald.conf have been renamed to
7179 logind.conf and journald.conf. Package updates should rename
7180 the files to the new names on upgrade.
7182 * For almost all files the license is now LGPL2.1+, changed
7183 from the previous GPL2.0+. Exceptions are some minor stuff
7184 of udev (which will be changed to LGPL2.1 eventually, too),
7185 and the MIT licensed sd-daemon.[ch] library that is suitable
7186 to be used as drop-in files.
7188 * systemd and logind now handle system sleep states, in
7189 particular suspending and hibernating.
7191 * logind now implements a sleep/shutdown/idle inhibiting logic
7192 suitable for a variety of uses. Soonishly Lennart will blog
7193 about this in more detail.
7195 * var-run.mount and var-lock.mount are no longer provided
7196 (which previously bind mounted these directories to their new
7197 places). Distributions which have not converted these
7198 directories to symlinks should consider stealing these files
7199 from git history and add them downstream.
7201 * We introduced the Documentation= field for units and added
7202 this to all our shipped units. This is useful to make it
7203 easier to explore the boot and the purpose of the various
7206 * All smaller setup units (such as
7207 systemd-vconsole-setup.service) now detect properly if they
7208 are run in a container and are skipped when
7209 appropriate. This guarantees an entirely noise-free boot in
7210 Linux container environments such as systemd-nspawn.
7212 * A framework for implementing offline system updates is now
7213 integrated, for details see:
7214 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7216 * A new service type Type=idle is available now which helps us
7217 avoiding ugly interleaving of getty output and boot status
7220 * There's now a system-wide CapabilityBoundingSet= option to
7221 globally reduce the set of capabilities for the
7222 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7223 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7224 even CAP_NET_ADMIN system-wide for secure systems.
7226 * There are now system-wide DefaultLimitXXX= options to
7227 globally change the defaults of the various resource limits
7228 for all units started by PID 1.
7230 * Harald Hoyer's systemd test suite has been integrated into
7231 systemd which allows easy testing of systemd builds in qemu
7232 and nspawn. (This is really awesome! Ask us for details!)
7234 * The fstab parser is now implemented as generator, not inside
7237 * systemctl will now warn you if .mount units generated from
7238 /etc/fstab are out of date due to changes in fstab that
7239 have not been read by systemd yet.
7241 * systemd is now suitable for usage in initrds. Dracut has
7242 already been updated to make use of this. With this in place
7243 initrds get a slight bit faster but primarily are much
7244 easier to introspect and debug since "systemctl status" in
7245 the host system can be used to introspect initrd services,
7246 and the journal from the initrd is kept around too.
7248 * systemd-delta has been added, a tool to explore differences
7249 between user/admin configuration and vendor defaults.
7251 * PrivateTmp= now affects both /tmp and /var/tmp.
7253 * Boot time status messages are now much prettier and feature
7254 proper english language. Booting up systemd has never been
7257 * Read-ahead pack files now include the inode number of all
7258 files to pre-cache. When the inode changes the pre-caching
7259 is not attempted. This should be nicer to deal with updated
7260 packages which might result in changes of read-ahead
7263 * We now temporaritly lower the kernel's read_ahead_kb variable
7264 when collecting read-ahead data to ensure the kernel's
7265 built-in read-ahead does not add noise to our measurements
7266 of necessary blocks to pre-cache.
7268 * There's now RequiresMountsFor= to add automatic dependencies
7269 for all mounts necessary for a specific file system path.
7271 * MountAuto= and SwapAuto= have been removed from
7272 system.conf. Mounting file systems at boot has to take place
7275 * nspawn now learned a new switch --uuid= to set the machine
7276 ID on the command line.
7278 * nspawn now learned the -b switch to automatically search
7281 * vt102 is now the default TERM for serial TTYs, upgraded from
7284 * systemd-logind now works on VT-less systems.
7286 * The build tree has been reorganized. The individual
7287 components now have directories of their own.
7289 * A new condition type ConditionPathIsReadWrite= is now available.
7291 * nspawn learned the new -C switch to create cgroups for the
7292 container in other hierarchies.
7294 * We now have support for hardware watchdogs, configurable in
7297 * The scheduled shutdown logic now has a public API.
7299 * We now mount /tmp as tmpfs by default, but this can be
7300 masked and /etc/fstab can override it.
7302 * Since udisks does not make use of /media anymore we are not
7303 mounting a tmpfs on it anymore.
7305 * journalctl gained a new --local switch to only interleave
7306 locally generated journal files.
7308 * We can now load the IMA policy at boot automatically.
7310 * The GTK tools have been split off into a systemd-ui.
7312 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7313 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7314 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7315 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7316 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7317 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7318 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7319 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7320 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7325 * This is mostly a bugfix release
7327 * Support optional initialization of the machine ID from the
7328 KVM or container configured UUID.
7330 * Support immediate reboots with "systemctl reboot -ff"
7332 * Show /etc/os-release data in systemd-analyze output
7334 * Many bugfixes for the journal, including endianness fixes and
7335 ensuring that disk space enforcement works
7337 * sd-login.h is C++ compatible again
7339 * Extend the /etc/os-release format on request of the Debian
7342 * We now refuse non-UTF8 strings used in various configuration
7343 and unit files. This is done to ensure we do not pass invalid
7344 data over D-Bus or expose it elsewhere.
7346 * Register Mimo USB Screens as suitable for automatic seat
7349 * Read SELinux client context from journal clients in a race
7352 * Reorder configuration file lookup order. /etc now always
7353 overrides /run in order to allow the administrator to always
7354 and unconditionally override vendor-supplied or
7355 automatically generated data.
7357 * The various user visible bits of the journal now have man
7358 pages. We still lack man pages for the journal API calls
7361 * We now ship all man pages in HTML format again in the
7364 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7365 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7366 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7371 * This is mostly a bugfix release
7373 * systems lacking /etc/os-release are no longer supported.
7375 * Various functionality updates to libsystemd-login.so
7377 * Track class of PAM logins to distinguish greeters from
7380 Contributions from: Kay Sievers, Lennart Poettering, Michael
7385 * This is an important bugfix release for v41.
7387 * Building man pages is now optional which should be useful
7388 for those building systemd from git but unwilling to install
7391 * Watchdog support for supervising services is now usable. In
7392 a future release support for hardware watchdogs
7393 (i.e. /dev/watchdog) will be added building on this.
7395 * Service start rate limiting is now configurable and can be
7396 turned off per service. When a start rate limit is hit a
7397 reboot can automatically be triggered.
7399 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7401 Contributions from: Benjamin Franzke, Bill Nottingham,
7402 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7403 Schmidt, Michał Górny, Piotr Drąg
7407 * The systemd binary is installed /usr/lib/systemd/systemd now;
7408 An existing /sbin/init symlink needs to be adapted with the
7411 * The code that loads kernel modules has been ported to invoke
7412 libkmod directly, instead of modprobe. This means we do not
7413 support systems with module-init-tools anymore.
7415 * Watchdog support is now already useful, but still not
7418 * A new kernel command line option systemd.setenv= is
7419 understood to set system wide environment variables
7420 dynamically at boot.
7422 * We now limit the set of capabilities of systemd-journald.
7424 * We now set SIGPIPE to ignore by default, since it only is
7425 useful in shell pipelines, and has little use in general
7426 code. This can be disabled with IgnoreSIPIPE=no in unit
7429 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7430 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7435 * This is mostly a bugfix release
7437 * We now expose the reason why a service failed in the
7438 "Result" D-Bus property.
7440 * Rudimentary service watchdog support (will be completed over
7441 the next few releases.)
7443 * When systemd forks off in order execute some service we will
7444 now immediately changes its argv[0] to reflect which process
7445 it will execute. This is useful to minimize the time window
7446 with a generic argv[0], which makes bootcharts more useful
7448 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7449 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7450 Mike Kazantsev, Ray Strode
7454 * This is mostly a test release, but incorporates many
7457 * New systemd-cgtop tool to show control groups by their
7460 * Linking against libacl for ACLs is optional again. If
7461 disabled, support tracking device access for active logins
7462 goes becomes unavailable, and so does access to the user
7463 journals by the respective users.
7465 * If a group "adm" exists, journal files are automatically
7466 owned by them, thus allow members of this group full access
7467 to the system journal as well as all user journals.
7469 * The journal now stores the SELinux context of the logging
7470 client for all entries.
7472 * Add C++ inclusion guards to all public headers
7474 * New output mode "cat" in the journal to print only text
7475 messages, without any meta data like date or time.
7477 * Include tiny X server wrapper as a temporary stop-gap to
7478 teach XOrg udev display enumeration. This is used by display
7479 managers such as gdm, and will go away as soon as XOrg
7480 learned native udev hotplugging for display devices.
7482 * Add new systemd-cat tool for executing arbitrary programs
7483 with STDERR/STDOUT connected to the journal. Can also act as
7484 BSD logger replacement, and does so by default.
7486 * Optionally store all locally generated coredumps in the
7487 journal along with meta data.
7489 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7490 writing short strings to files (for usage for /sys), and for
7491 creating symlinks, character and block device nodes.
7493 * New unit file option ControlGroupPersistent= to make cgroups
7494 persistent, following the mechanisms outlined in
7495 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7497 * Support multiple local RTCs in a sane way
7499 * No longer monopolize IO when replaying readahead data on
7500 rotating disks, since we might starve non-file-system IO to
7501 death, since fanotify() will not see accesses done by blkid,
7504 * Do not show kernel threads in systemd-cgls anymore, unless
7505 requested with new -k switch.
7507 Contributions from: Dan Horák, Kay Sievers, Lennart
7508 Poettering, Michal Schmidt
7512 * This is mostly a test release, but incorporates many
7515 * The git repository moved to:
7516 git://anongit.freedesktop.org/systemd/systemd
7517 ssh://git.freedesktop.org/git/systemd/systemd
7519 * First release with the journal
7520 http://0pointer.de/blog/projects/the-journal.html
7522 * The journal replaces both systemd-kmsg-syslogd and
7523 systemd-stdout-bridge.
7525 * New sd_pid_get_unit() API call in libsystemd-logind
7527 * Many systemadm clean-ups
7529 * Introduce remote-fs-pre.target which is ordered before all
7530 remote mounts and may be used to start services before all
7533 * Added Mageia support
7535 * Add bash completion for systemd-loginctl
7537 * Actively monitor PID file creation for daemons which exit in
7538 the parent process before having finished writing the PID
7539 file in the daemon process. Daemons which do this need to be
7540 fixed (i.e. PID file creation must have finished before the
7541 parent exits), but we now react a bit more gracefully to them.
7543 * Add colourful boot output, mimicking the well-known output
7544 of existing distributions.
7546 * New option PassCredentials= for socket units, for
7547 compatibility with a recent kernel ABI breakage.
7549 * /etc/rc.local is now hooked in via a generator binary, and
7550 thus will no longer act as synchronization point during
7553 * systemctl list-unit-files now supports --root=.
7555 * systemd-tmpfiles now understands two new commands: z, Z for
7556 relabelling files according to the SELinux database. This is
7557 useful to apply SELinux labels to specific files in /sys,
7560 * Output of SysV services is now forwarded to both the console
7561 and the journal by default, not only just the console.
7563 * New man pages for all APIs from libsystemd-login.
7565 * The build tree got reorganized and the build system is a
7566 lot more modular allowing embedded setups to specifically
7567 select the components of systemd they are interested in.
7569 * Support for Linux systems lacking the kernel VT subsystem is
7572 * configure's --with-rootdir= got renamed to
7573 --with-rootprefix= to follow the naming used by udev and
7576 * Unless specified otherwise we will now install to /usr instead
7577 of /usr/local by default.
7579 * Processes with '@' in argv[0][0] are now excluded from the
7580 final shut-down killing spree, following the logic explained
7582 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7584 * All processes remaining in a service cgroup when we enter
7585 the START or START_PRE states are now killed with
7586 SIGKILL. That means it is no longer possible to spawn
7587 background processes from ExecStart= lines (which was never
7588 supported anyway, and bad style).
7590 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7591 reloading of units together.
7593 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7594 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7595 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7596 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7597 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek