1 systemd System and Service Manager
3 CHANGES WITH 242 in spe:
5 * In .link files, MACAddressPolicy=persistent (the default) is changed
6 to cover more devices. For devices like bridges, tun, tap, bond, and
7 similar interfaces that do not have other identifying information,
8 the interface name is used as the basis for persistent seed for MAC
9 and IPv4LL addresses. The way that devices that were handled
10 previously is not changed, and this change is about covering more
11 devices then previously by the "persistent" policy.
13 MACAddressPolicy=random may be used to force randomized MACs and
14 IPv4LL addresses for a device if desired.
16 Hint: the log output from udev (at debug level) was enhanced to
17 clarify what policy is followed and which attributes are used.
18 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
19 may be used to view this.
21 * The .device units generated by systemd-fstab-generator and other
22 generators do not automatically pull in the corresponding .mount unit
23 as a Wants= dependency. This means that simply plugging in the device
24 will not cause the mount unit to be started automatically. But please
25 note that the mount unit may be started for other reasons, in
26 particular if it is part of local-fs.target, and any unit which
27 (transitively) depends on local-fs.target is started.
29 * networkctl list/status/lldp now accept globbing wildcards for network
30 interface names to match against all existing interfaces.
32 * The $PIDFILE environment variable is set to point the absolute path
33 configured with PIDFile= for processes of that service.
35 * The fallback DNS server list was augmented with Cloudflare public DNS
36 servers. Use `-Ddns-servers=` to set a different fallback.
38 * A new special target usb-gadget.target will be started automatically
39 when a USB Device Controller is detected (which means that the system
42 * A new unit setting CPUQuotaPeriodSec= assigns the time period
43 relatively to which the CPU time quota specified by CPUQuota= is
46 * A new unit setting ProtectHostname= may be used to prevent services
47 from modifying hostname information (even if they otherwise would
48 have privileges to do so).
50 * A new unit setting NetworkNamespacePath= may be used to specify a
51 namespace for service or socket units through a path referring to a
52 Linux network namespace pseudo-file.
54 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
55 have an effect on .socket units: when used the listening socket is
56 created within the configured network namespace instead of the host
59 * ExecStart= command lines in unit files may now be prefixed with ':'
60 in which case environment variable substitution is
61 disabled. (Supported for the other ExecXYZ= settings, too.)
63 * systemd-networkd recognizes a new operation state 'enslaved', used
64 (instead of 'degraded' or 'carrier') for interfaces which form a
65 bridge, bond, or similar, and an new 'degraded-carrier' operational
66 state used for the bond or bridge master interface when one of the
67 enslaved devices is not operational.
69 * .network files learnt the new IgnoreCarrierLoss= option for leaving
70 networks configured even if the carrier is lost.
72 * The RequiredForOnline= setting in .network files may now specify a
73 minimum operational state required for the interface to be considered
74 "online" by systemd-networkd-wait-online. Related to this
75 systemd-networkd-wait-online gained a new option --operational-state=
76 to configure the same, and its --interface= option was updated to
77 optionally also take an operational state specific for an interface.
79 * systemd-networkd now implements L2TP tunnels.
81 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
82 may be used to cause autonomous and onlink prefixes received in IPv6
83 Router Advertisements to be ignored.
85 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
86 file settings may be used to tweak bridge behaviour.
88 * The new TripleSampling= option in .network files may be used to
89 configure CAN triple sampling.
91 * A new .netdev setting PrivateKeyFile= may be used to point to private
92 key for a WireGuard interface.
94 * /etc/crypttab now supports the same-cpu-crypt and
95 submit-from-crypt-cpus options to tweak encryption work scheduling
98 * systemd-tmpfiles will now take a BSD file lock before operating on a
99 contents of directory. This may be used to temporarily exclude
100 directories from aging by taking the same lock (useful for example
101 when extracting a tarball into /tmp or /var/tmp as a privileged user,
102 which might create files with really old timestamps, which
103 nevertheless should not be deleted). For further details, see:
105 https://systemd.io/TEMPORARY_DIRECTORIES
107 * systemd-tmpfiles' h line type gained support for the
108 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
109 controlling project quota inheritance.
111 * sd-boot and bootctl now implement support for an Extended Boot Loader
112 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
113 addition to the ESP partition mounted to /efi or /boot/efi.
114 Configuration file fragments, kernels, initrds and other EFI images
115 to boot will be loaded from both the ESP and XBOOTLDR partitions.
116 The XBOOTLDR partition was previously described by the Boot Loader
117 Specification, but implementation was missing in sd-boot. Support for
118 this concept allows using the sd-boot boot loader in more
119 conservative scenarios where the boot loader itself is placed in the
120 ESP but the kernels to boot (and their metadata) in a separate
123 * A system may now be booted with systemd.volatile=overlay on the
124 kernel command line, which causes the root file system to be set up
125 an overlayfs mount combining the root-only root directory with a
126 writable tmpfs. In this setup, the underlying root device is not
127 modified, and any changes are lost at reboot.
129 * Similar, systemd-nspawn can now boot containers with a volatile
130 overlayfs root with the new --volatile=overlay switch.
132 * systemd-nspawn can now consume OCI runtime bundles using a new
133 --oci-bundle= option. This implementation is fully usable, with most
134 features in the specification implemented, but since this a lot of
135 new code and functionality, this feature should most likely not
136 be used in production yet.
138 * systemd-nspawn now supports various options described by the OCI
139 runtime specification on the command-line and in .nspawn files:
140 --inaccessible=/Inaccessible= may be used to mask parts of the file
141 system tree, --console=/--pipe may be used to configure how standard
142 input, output, and error are set up.
144 * busctl learned the `emit` verb to generate D-Bus signals.
146 * systemd-analyze cat-config may be used to gather and display
147 configuration spread over multiple files, for example system and user
148 presets, tmpfiles.d, sysusers.d, udev rules, etc.
150 * systemd-analyze calendar now takes an optional new parameter
151 --iterations= which may be used to show a maximum number of iterations
152 the specified expression will elapse next.
154 * The sd-bus C API gained support for naming method parameters in the
157 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
158 the reboot() system call expects.
160 * journalctl learnt a new --cursor-file= option that points to a file
161 from which a cursor should be loaded in the beginning and to which
162 the updated cursor should be stored at the end.
164 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
165 detected by systemd-detect-virt (and may also be used in
166 ConditionVirtualization=).
168 * The behaviour of systemd-logind may now be modified with environment
169 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
170 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
171 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
172 skip the relevant operation completely (when set to false), or to
173 create a flag file in /run/systemd (when set to true), instead of
174 actually commencing the real operation when requested. The presence
175 of /run/systemd/reboot-to-firmware-setup,
176 /run/systemd/reboot-to-boot-loader-menu, and
177 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
178 boot loader implementations to replace some steps logind performs
179 during reboot with their own operations.
181 * systemctl can be used to request a reboot into the boot loader menu
182 or a specific boot loader entry with the new --boot-load-menu= and
183 --boot-loader-entry= options to a reboot command. (This requires a
184 boot loader that supports this, for example sd-boot.)
186 * kernel-install will no longer unconditionally create the output
187 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
188 snippets, but will do only if the machine-specific parent directory
189 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
190 to create this parent directory during sd-boot installation.
192 This makes it easier to use kernel-install with plugins which support
193 a different layout of the bootloader partitions (for example grub2).
195 * During package installation (with `ninja install`), we would create
196 symlinks for systemd-networkd.service, systemd-networkd.socket,
197 systemd-resolved.service, remote-cryptsetup.target, remote-fs.target,
198 systemd-networkd-wait-online.service, and systemd-timesyncd.service
199 in /etc, as if `systemctl enable` was called for those units, to make
200 the system usable immediately after installation. Now this is not
201 done anymore, and instead calling `systemctl preset-all` is
202 recommended after the first installation of systemd.
204 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
205 is built on seccomp. When turned on creation of SUID/SGID files is
208 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
209 implied if DynamicUser= is turned on for a service. This hardens
210 these services, so that they neither can benefit from nor create
211 SUID/SGID executables. This is a minor compatibility breakage, given
212 that when DynamicUser= was first introduced SUID/SGID behaviour was
213 unaffected. However, the security benefit of these two options is
214 substantial, and the setting is still relatively new, hence we opted
215 to make it mandatory for services with dynamic users.
217 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
218 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
219 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
220 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
221 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
222 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
223 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
224 Pyo, Jan Engelhardt, Jonathan Lebon, Jonathon Kowalski, Jörg Sommer,
225 Jörg Thalheim, Kai-Heng Feng, Lennart Poettering, Lubomir Rintel,
226 Martin Pitt, Matthias Klumpp, Michael Biebl, Michael Niewöhner, Michael
227 Olbrich, Michal Sekletar, Mike Lothian, Piotr Drąg, Riccardo Schirone,
228 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
229 Gonzalez, Stephane Chazelas, StKob, Susant Sahani, Sylvain Plantefève,
230 Szabolcs Fruhwald, Taro Yamada, Theo Ouzhinski, Thomas Haller, Tobias
231 Jungel, Tom Yan, Tony Asleson, Topi Miettinen, unixsysadmin, Van Laser,
232 Vesa Jääskeläinen, Yu, Li-Yu, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
234 — Somewhere, 2019-0X-YZ
238 * The default locale can now be configured at compile time. Otherwise,
239 a suitable default will be selected automatically (one of C.UTF-8,
242 * The version string shown by systemd and other tools now includes the
243 git commit hash when built from git. An override may be specified
244 during compilation, which is intended to be used by distributions to
245 include the package release information.
247 * systemd-cat can now filter standard input and standard error streams
248 for different syslog priorities using the new --stderr-priority=
251 * systemd-journald and systemd-journal-remote reject entries which
252 contain too many fields (CVE-2018-16865) and set limits on the
253 process' command line length (CVE-2018-16864).
255 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
258 * A new network device NamePolicy "keep" is implemented for link files,
259 and used by default in 99-default.link (the fallback configuration
260 provided by systemd). With this policy, if the network device name
261 was already set by userspace, the device will not be renamed again.
262 This matches the naming scheme that was implemented before
263 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
264 is also enabled by default, even if not specified. Effectively, this
265 means that if naming-scheme >= 240 is specified, network devices will
266 be renamed according to the configuration, even if they have been
267 renamed already, if "keep" is not specified as the naming policy in
268 the .link file. The 99-default.link file provided by systemd includes
269 "keep" for backwards compatibility, but it is recommended for user
270 installed .link files to *not* include it.
272 The "kernel" policy, which keeps kernel names declared to be
273 "persistent", now works again as documented.
275 * kernel-install script now optionally takes the paths to one or more
276 initrd files, and passes them to all plugins.
278 * The mincore() system call has been dropped from the @system-service
279 system call filter group, as it is pretty exotic and may potentially
280 used for side-channel attacks.
282 * -fPIE is dropped from compiler and linker options. Please specify
283 -Db_pie=true option to meson to build position-independent
284 executables. Note that the meson option is supported since meson-0.49.
286 * The fs.protected_regular and fs.protected_fifos sysctls, which were
287 added in Linux 4.19 to make some data spoofing attacks harder, are
288 now enabled by default. While this will hopefully improve the
289 security of most installations, it is technically a backwards
290 incompatible change; to disable these sysctls again, place the
291 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
293 fs.protected_regular = 0
294 fs.protected_fifos = 0
296 Note that the similar hardlink and symlink protection has been
297 enabled since v199, and may be disabled likewise.
299 * The files read from the EnvironmentFile= setting in unit files now
300 parse backslashes inside quotes literally, matching the behaviour of
303 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
304 now automatically become NOPs when run in a chroot() environment.
306 * The tmpfiles.d/ "C" line type will now copy directory trees not only
307 when the destination is so far missing, but also if it already exists
308 as a directory and is empty. This is useful to cater for systems
309 where directory trees are put together from multiple separate mount
310 points but otherwise empty.
312 * A new function sd_bus_close_unref() (and the associated
313 sd_bus_close_unrefp()) has been added to libsystemd, that combines
314 sd_bus_close() and sd_bus_unref() in one.
316 * udevadm control learnt a new option for --ping for testing whether a
317 systemd-udevd instance is running and reacting.
319 * udevadm trigger learnt a new option for --wait-daemon for waiting
320 systemd-udevd daemon to be initialized.
322 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
323 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
324 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
325 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
326 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
327 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
328 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
329 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
330 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
331 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
332 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
333 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
334 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
335 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
336 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
337 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
338 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
344 * NoNewPrivileges=yes has been set for all long-running services
345 implemented by systemd. Previously, this was problematic due to
346 SELinux (as this would also prohibit the transition from PID1's label
347 to the service's label). This restriction has since been lifted, but
348 an SELinux policy update is required.
349 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
351 * DynamicUser=yes is dropped from systemd-networkd.service,
352 systemd-resolved.service and systemd-timesyncd.service, which was
353 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
354 and since v236 for systemd-timesyncd.service. The users and groups
355 systemd-network, systemd-resolve and systemd-timesync are created
356 by systemd-sysusers again. Distributors or system administrators
357 may need to create these users and groups if they not exist (or need
358 to re-enable DynamicUser= for those units) while upgrading systemd.
359 Also, the clock file for systemd-timesyncd may need to move from
360 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
362 * When unit files are loaded from disk, previously systemd would
363 sometimes (depending on the unit loading order) load units from the
364 target path of symlinks in .wants/ or .requires/ directories of other
365 units. This meant that unit could be loaded from different paths
366 depending on whether the unit was requested explicitly or as a
367 dependency of another unit, not honouring the priority of directories
368 in search path. It also meant that it was possible to successfully
369 load and start units which are not found in the unit search path, as
370 long as they were requested as a dependency and linked to from
371 .wants/ or .requires/. The target paths of those symlinks are not
372 used for loading units anymore and the unit file must be found in
375 * A new service type has been added: Type=exec. It's very similar to
376 Type=simple but ensures the service manager will wait for both fork()
377 and execve() of the main service binary to complete before proceeding
378 with follow-up units. This is primarily useful so that the manager
379 propagates any errors in the preparation phase of service execution
380 back to the job that requested the unit to be started. For example,
381 consider a service that has ExecStart= set to a file system binary
382 that doesn't exist. With Type=simple starting the unit would be
383 considered instantly successful, as only fork() has to complete
384 successfully and the manager does not wait for execve(), and hence
385 its failure is seen "too late". With the new Type=exec service type
386 starting the unit will fail, as the manager will wait for the
387 execve() and notice its failure, which is then propagated back to the
390 NOTE: with the next release 241 of systemd we intend to change the
391 systemd-run tool to default to Type=exec for transient services
392 started by it. This should be mostly safe, but in specific corner
393 cases might result in problems, as the systemd-run tool will then
394 block on NSS calls (such as user name look-ups due to User=) done
395 between the fork() and execve(), which under specific circumstances
396 might cause problems. It is recommended to specify "-p Type=simple"
397 explicitly in the few cases where this applies. For regular,
398 non-transient services (i.e. those defined with unit files on disk)
399 we will continue to default to Type=simple.
401 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
402 userspace processes is set to 1024 (soft) and 4096
403 (hard). Previously, systemd passed this on unmodified to all
404 processes it forked off. With this systemd release the hard limit
405 systemd passes on is increased to 512K, overriding the kernel's
406 defaults and substantially increasing the number of simultaneous file
407 descriptors unprivileged userspace processes can allocate. Note that
408 the soft limit remains at 1024 for compatibility reasons: the
409 traditional UNIX select() call cannot deal with file descriptors >=
410 1024 and increasing the soft limit globally might thus result in
411 programs unexpectedly allocating a high file descriptor and thus
412 failing abnormally when attempting to use it with select() (of
413 course, programs shouldn't use select() anymore, and prefer
414 poll()/epoll, but the call unfortunately remains undeservedly popular
415 at this time). This change reflects the fact that file descriptor
416 handling in the Linux kernel has been optimized in more recent
417 kernels and allocating large numbers of them should be much cheaper
418 both in memory and in performance than it used to be. Programs that
419 want to take benefit of the increased limit have to "opt-in" into
420 high file descriptors explicitly by raising their soft limit. Of
421 course, when they do that they must acknowledge that they cannot use
422 select() anymore (and neither can any shared library they use — or
423 any shared library used by any shared library they use and so on).
424 Which default hard limit is most appropriate is of course hard to
425 decide. However, given reports that ~300K file descriptors are used
426 in real-life applications we believe 512K is sufficiently high as new
427 default for now. Note that there are also reports that using very
428 high hard limits (e.g. 1G) is problematic: some software allocates
429 large arrays with one element for each potential file descriptor
430 (Java, …) — a high hard limit thus triggers excessively large memory
431 allocations in these applications. Hopefully, the new default of 512K
432 is a good middle ground: higher than what real-life applications
433 currently need, and low enough for avoid triggering excessively large
434 allocations in problematic software. (And yes, somebody should fix
437 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
438 to the highest possible values, as separate accounting of file
439 descriptors is no longer necessary, as memcg tracks them correctly as
440 part of the memory accounting anyway. Thus, from the four limits on
441 file descriptors currently enforced (fs.file-max, fs.nr_open,
442 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
443 and keep only the latter two. A set of build-time options
444 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
445 has been added to revert this change in behaviour, which might be
446 an option for systems that turn off memcg in the kernel.
448 * When no /etc/locale.conf file exists (and hence no locale settings
449 are in place), systemd will now use the "C.UTF-8" locale by default,
450 and set LANG= to it. This locale is supported by various
451 distributions including Fedora, with clear indications that upstream
452 glibc is going to make it available too. This locale enables UTF-8
453 mode by default, which appears appropriate for 2018.
455 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
456 default. This effectively switches the RFC3704 Reverse Path filtering
457 from Strict mode to Loose mode. This is more appropriate for hosts
458 that have multiple links with routes to the same networks (e.g.
459 a client with a Wi-Fi and Ethernet both connected to the internet).
461 Consult the kernel documentation for details on this sysctl:
462 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
464 * CPUAccounting=yes no longer enables the CPU controller when using
465 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
466 statistics are now provided independently from the CPU controller.
468 * Support for disabling a particular cgroup controller within a sub-tree
469 has been added through the DisableControllers= directive.
471 * cgroup_no_v1=all on the kernel command line now also implies
472 using the unified cgroup hierarchy, unless one explicitly passes
473 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
475 * The new "MemoryMin=" unit file property may now be used to set the
476 memory usage protection limit of processes invoked by the unit. This
477 controls the cgroup v2 memory.min attribute. Similarly, the new
478 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
479 cgroup v2 io.latency cgroup property for configuring per-service I/O
482 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
483 to the cgroup v1 "devices" cgroup controller.
485 * systemd-escape now is able to combine --unescape with --template. It
486 also learnt a new option --instance for extracting and unescaping the
487 instance part of a unit name.
489 * sd-bus now provides the sd_bus_message_readv() which is similar to
490 sd_bus_message_read() but takes a va_list object. The pair
491 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
492 has been added for configuring the default method call timeout to
493 use. sd_bus_error_move() may be used to efficiently move the contents
494 from one sd_bus_error structure to another, invalidating the
495 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
496 be used to control whether a bus connection object is automatically
497 flushed when an sd-event loop is exited.
499 * When processing classic BSD syslog log messages, journald will now
500 save the original time-stamp string supplied in the new
501 SYSLOG_TIMESTAMP= journal field. This permits consumers to
502 reconstruct the original BSD syslog message more correctly.
504 * StandardOutput=/StandardError= in service files gained support for
505 new "append:…" parameters, for connecting STDOUT/STDERR of a service
506 to a file, and appending to it.
508 * The signal to use as last step of killing of unit processes is now
509 configurable. Previously it was hard-coded to SIGKILL, which may now
510 be overridden with the new KillSignal= setting. Note that this is the
511 signal used when regular termination (i.e. SIGTERM) does not suffice.
512 Similarly, the signal used when aborting a program in case of a
513 watchdog timeout may now be configured too (WatchdogSignal=).
515 * The XDG_SESSION_DESKTOP environment variable may now be configured in
516 the pam_systemd argument line, using the new desktop= switch. This is
517 useful to initialize it properly from a display manager without
518 having to touch C code.
520 * Most configuration options that previously accepted percentage values
521 now also accept permille values with the '‰' suffix (instead of '%').
523 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
526 * systemd-resolved's configuration file resolved.conf gained a new
527 option ReadEtcHosts= which may be used to turn off processing and
528 honoring /etc/hosts entries.
530 * The "--wait" switch may now be passed to "systemctl
531 is-system-running", in which case the tool will synchronously wait
532 until the system finished start-up.
534 * hostnamed gained a new bus call to determine the DMI product UUID.
536 * On x86-64 systemd will now prefer using the RDRAND processor
537 instruction over /dev/urandom whenever it requires randomness that
538 neither has to be crypto-grade nor should be reproducible. This
539 should substantially reduce the amount of entropy systemd requests
540 from the kernel during initialization on such systems, though not
541 reduce it to zero. (Why not zero? systemd still needs to allocate
542 UUIDs and such uniquely, which require high-quality randomness.)
544 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
545 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
546 for forcing the "Other Information" bit in IPv6 RA messages. The
547 bonding logic gained four new options AdActorSystemPriority=,
548 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
549 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
550 shuffling of flows. The tunnel logic gained a new
551 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
552 Deployment. The policy rule logic gained four new options IPProtocol=,
553 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
554 support for the MulticastToUnicast= option. networkd also gained
555 support for configuring static IPv4 ARP or IPv6 neighbor entries.
557 * .preset files (as read by 'systemctl preset') may now be used to
558 instantiate services.
560 * /etc/crypttab now understands the sector-size= option to configure
561 the sector size for an encrypted partition.
563 * Key material for encrypted disks may now be placed on a formatted
564 medium, and referenced from /etc/crypttab by the UUID of the file
565 system, followed by "=" suffixed by the path to the key file.
567 * The "collect" udev component has been removed without replacement, as
568 it is neither used nor maintained.
570 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
571 LogsDirectory=, ConfigurationDirectory= settings are used in a
572 service the executed processes will now receive a set of environment
573 variables containing the full paths of these directories.
574 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
575 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
576 are used. Note that these options may be used multiple times per
577 service in which case the resulting paths will be concatenated and
580 * Predictable interface naming has been extended to cover InfiniBand
581 NICs. They will be exposed with an "ib" prefix.
583 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
584 which case the respective line failing is ignored.
586 * .link files may now be used to configure the equivalent to the
587 "ethtool advertise" commands.
589 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
590 alternative to libudev.h. Previously, the latter was just an internal
591 wrapper around the former, but now these two APIs are exposed
594 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
595 which calculates an app-specific boot ID similar to how
596 sd_id128_get_machine_app_specific() generates an app-specific machine
599 * A new tool systemd-id128 has been added that can be used to determine
600 and generate various 128bit IDs.
602 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
605 * systemd-hibernate-resume-generator will now honor the "noresume"
606 kernel command line option, in which case it will bypass resuming
607 from any hibernated image.
609 * The systemd-sleep.conf configuration file gained new options
610 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
611 AllowHybridSleep= for prohibiting specific sleep modes even if the
614 * portablectl is now officially supported and has thus moved to
617 * bootctl learnt the two new commands "set-default" and "set-oneshot"
618 for setting the default boot loader item to boot to (either
619 persistently or only for the next boot). This is currently only
620 compatible with sd-boot, but may be implemented on other boot loaders
621 too, that follow the boot loader interface. The updated interface is
624 https://systemd.io/BOOT_LOADER_INTERFACE
626 * A new kernel command line option systemd.early_core_pattern= is now
627 understood which may be used to influence the core_pattern PID 1
628 installs during early boot.
630 * busctl learnt two new options -j and --json= for outputting method
631 call replies, properties and monitoring output in JSON.
633 * journalctl's JSON output now supports simple ANSI coloring as well as
634 a new "json-seq" mode for generating RFC7464 output.
636 * Unit files now support the %g/%G specifiers that resolve to the UNIX
637 group/GID of the service manager runs as, similar to the existing
638 %u/%U specifiers that resolve to the UNIX user/UID.
640 * systemd-logind learnt a new global configuration option
641 UserStopDelaySec= that may be set in logind.conf. It specifies how
642 long the systemd --user instance shall remain started after a user
643 logs out. This is useful to speed up repetitive re-connections of the
644 same user, as it means the user's service manager doesn't have to be
645 stopped/restarted on each iteration, but can be reused between
646 subsequent options. This setting defaults to 10s. systemd-logind also
647 exports two new properties on its Manager D-Bus objects indicating
648 whether the system's lid is currently closed, and whether the system
651 * systemd gained support for a generic boot counting logic, which
652 generically permits automatic reverting to older boot loader entries
653 if newer updated ones don't work. The boot loader side is implemented
654 in sd-boot, but is kept open for other boot loaders too. For details
657 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
659 * The SuccessAction=/FailureAction= unit file settings now learnt two
660 new parameters: "exit" and "exit-force", which result in immediate
661 exiting of the service manager, and are only useful in systemd --user
662 and container environments.
664 * Unit files gained support for a pair of options
665 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
666 exit status to use as service manager exit status when
667 SuccessAction=/FailureAction= is set to exit or exit-force.
669 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
670 options may now be used to configure the log rate limiting applied by
671 journald per-service.
673 * systemd-analyze gained a new verb "timespan" for parsing and
674 normalizing time span values (i.e. strings like "5min 7s 8us").
676 * systemd-analyze also gained a new verb "security" for analyzing the
677 security and sand-boxing settings of services in order to determine an
678 "exposure level" for them, indicating whether a service would benefit
679 from more sand-boxing options turned on for them.
681 * "systemd-analyze syscall-filter" will now also show system calls
682 supported by the local kernel but not included in any of the defined
685 * .nspawn files now understand the Ephemeral= setting, matching the
686 --ephemeral command line switch.
688 * sd-event gained the new APIs sd_event_source_get_floating() and
689 sd_event_source_set_floating() for controlling whether a specific
690 event source is "floating", i.e. destroyed along with the even loop
693 * Unit objects on D-Bus gained a new "Refs" property that lists all
694 clients that currently have a reference on the unit (to ensure it is
697 * The JoinControllers= option in system.conf is no longer supported, as
698 it didn't work correctly, is hard to support properly, is legacy (as
699 the concept only exists on cgroup v1) and apparently wasn't used.
701 * Journal messages that are generated whenever a unit enters the failed
702 state are now tagged with a unique MESSAGE_ID. Similarly, messages
703 generated whenever a service process exits are now made recognizable,
704 too. A taged message is also emitted whenever a unit enters the
705 "dead" state on success.
707 * systemd-run gained a new switch --working-directory= for configuring
708 the working directory of the service to start. A shortcut -d is
709 equivalent, setting the working directory of the service to the
710 current working directory of the invoking program. The new --shell
711 (or just -S) option has been added for invoking the $SHELL of the
712 caller as a service, and implies --pty --same-dir --wait --collect
713 --service-type=exec. Or in other words, "systemd-run -S" is now the
714 quickest way to quickly get an interactive in a fully clean and
715 well-defined system service context.
717 * machinectl gained a new verb "import-fs" for importing an OS tree
718 from a directory. Moreover, when a directory or tarball is imported
719 and single top-level directory found with the OS itself below the OS
720 tree is automatically mangled and moved one level up.
722 * systemd-importd will no longer set up an implicit btrfs loop-back
723 file system on /var/lib/machines. If one is already set up, it will
726 * A new generator "systemd-run-generator" has been added. It will
727 synthesize a unit from one or more program command lines included in
728 the kernel command line. This is very useful in container managers
731 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
733 This will run "systemd-nspawn" on an image, invoke the specified
734 command line and immediately shut down the container again, returning
735 the command line's exit code.
737 * The block device locking logic is now documented:
739 https://systemd.io/BLOCK_DEVICE_LOCKING
741 * loginctl and machinectl now optionally output the various tables in
742 JSON using the --output= switch. It is our intention to add similar
743 support to systemctl and all other commands.
745 * udevadm's query and trigger verb now optionally take a .device unit
748 * systemd-udevd's network naming logic now understands a new
749 net.naming-scheme= kernel command line switch, which may be used to
750 pick a specific version of the naming scheme. This helps stabilizing
751 interface names even as systemd/udev are updated and the naming logic
754 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
755 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
756 initialize one to all 0xFF.
758 * After loading the SELinux policy systemd will now recursively relabel
759 all files and directories listed in
760 /run/systemd/relabel-extra.d/*.relabel (which should be simple
761 newline separated lists of paths) in addition to the ones it already
762 implicitly relabels in /run, /dev and /sys. After the relabelling is
763 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
764 removed. This is useful to permit initrds (i.e. code running before
765 the SELinux policy is in effect) to generate files in the host
766 filesystem safely and ensure that the correct label is applied during
767 the transition to the host OS.
769 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
770 mknod() handling in user namespaces. Previously mknod() would always
771 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
772 but device nodes generated that way cannot be opened, and attempts to
773 open them result in EPERM. This breaks the "graceful fallback" logic
774 in systemd's PrivateDevices= sand-boxing option. This option is
775 implemented defensively, so that when systemd detects it runs in a
776 restricted environment (such as a user namespace, or an environment
777 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
778 where device nodes cannot be created the effect of PrivateDevices= is
779 bypassed (following the logic that 2nd-level sand-boxing is not
780 essential if the system systemd runs in is itself already sand-boxed
781 as a whole). This logic breaks with 4.18 in container managers where
782 user namespacing is used: suddenly PrivateDevices= succeeds setting
783 up a private /dev/ file system containing devices nodes — but when
784 these are opened they don't work.
786 At this point is is recommended that container managers utilizing
787 user namespaces that intend to run systemd in the payload explicitly
788 block mknod() with seccomp or similar, so that the graceful fallback
791 We are very sorry for the breakage and the requirement to change
792 container configurations for newer kernels. It's purely caused by an
793 incompatible kernel change. The relevant kernel developers have been
794 notified about this userspace breakage quickly, but they chose to
797 * PermissionsStartOnly= setting is deprecated (but is still supported
798 for backwards compatibility). The same functionality is provided by
799 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
802 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
805 * The naming scheme for network devices was changed to always rename
806 devices, even if they were already renamed by userspace. The "kernel"
807 policy was changed to only apply as a fallback, if no other naming
810 * The requirements to build systemd is bumped to meson-0.46 and
813 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
814 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
815 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
816 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
817 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
818 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
819 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
820 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
821 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
822 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
823 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
824 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
825 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
826 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
827 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
828 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
829 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
830 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
831 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
832 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
833 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
834 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
835 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
836 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
837 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
838 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
839 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
840 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
841 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
842 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
843 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
844 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
845 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
846 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
847 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
848 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
849 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
850 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
851 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
852 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
853 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
854 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
855 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
856 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
857 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
863 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
864 builtin will name network interfaces differently than in previous
865 versions for virtual network interfaces created with SR-IOV and NPAR
866 and for devices where the PCI network controller device does not have
867 a slot number associated.
869 SR-IOV virtual devices are now named based on the name of the parent
870 interface, with a suffix of "v<N>", where <N> is the virtual device
871 number. Previously those virtual devices were named as if completely
874 The ninth and later NPAR virtual devices will be named following the
875 scheme used for the first eight NPAR partitions. Previously those
876 devices were not renamed and the kernel default (eth<n>) was used.
878 "net_id" will also generate names for PCI devices where the PCI
879 network controller device does not have an associated slot number
880 itself, but one of its parents does. Previously those devices were
881 not renamed and the kernel default (eth<n>) was used.
883 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
884 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
885 the unit. So, it is expected that the default behavior of
886 systemd-logind is not changed. However, if distribution packagers or
887 administrators disabled or modified IPAddressDeny= setting by a
888 drop-in config file, then it may be necessary to update the file to
889 re-enable AF_INET and AF_INET6 to support network user name services,
892 * When the RestrictNamespaces= unit property is specified multiple
893 times, then the specified types are merged now. Previously, only the
894 last assignment was used. So, if distribution packagers or
895 administrators modified the setting by a drop-in config file, then it
896 may be necessary to update the file.
898 * When OnFailure= is used in combination with Restart= on a service
899 unit, then the specified units will no longer be triggered on
900 failures that result in restarting. Previously, the specified units
901 would be activated each time the unit failed, even when the unit was
902 going to be restarted automatically. This behaviour contradicted the
903 documentation. With this release the code is adjusted to match the
906 * systemd-tmpfiles will now print a notice whenever it encounters
907 tmpfiles.d/ lines referencing the /var/run/ directory. It will
908 recommend reworking them to use the /run/ directory instead (for
909 which /var/run/ is simply a symlinked compatibility alias). This way
910 systemd-tmpfiles can properly detect line conflicts and merge lines
911 referencing the same file by two paths, without having to access
914 * systemctl disable/unmask/preset/preset-all cannot be used with
915 --runtime. Previously this was allowed, but resulted in unintuitive
916 behaviour that wasn't useful. systemctl disable/unmask will now undo
917 both runtime and persistent enablement/masking, i.e. it will remove
918 any relevant symlinks both in /run and /etc.
920 * Note that all long-running system services shipped with systemd will
921 now default to a system call whitelist (rather than a blacklist, as
922 before). In particular, systemd-udevd will now enforce one too. For
923 most cases this should be safe, however downstream distributions
924 which disabled sandboxing of systemd-udevd (specifically the
925 MountFlags= setting), might want to disable this security feature
926 too, as the default whitelisting will prohibit all mount, swap,
927 reboot and clock changing operations from udev rules.
929 * sd-boot acquired new loader configuration settings to optionally turn
930 off Windows and MacOS boot partition discovery as well as
931 reboot-into-firmware menu items. It is also able to pick a better
932 screen resolution for HiDPI systems, and now provides loader
933 configuration settings to change the resolution explicitly.
935 * systemd-resolved now supports DNS-over-TLS. It's still
936 turned off by default, use DNSOverTLS=opportunistic to turn it on in
937 resolved.conf. We intend to make this the default as soon as couple
938 of additional techniques for optimizing the initial latency caused by
939 establishing a TLS/TCP connection are implemented.
941 * systemd-resolved.service and systemd-networkd.service now set
942 DynamicUser=yes. The users systemd-resolve and systemd-network are
943 not created by systemd-sysusers anymore.
945 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
946 that embedd a network facing module into any process using getpwuid()
947 or related call: the dynamic allocation of the user ID for
948 systemd-resolved.service means the service manager has to check NSS
949 if the user name is already taken when forking off the service. Since
950 the user in the common case won't be defined in /etc/passwd the
951 lookup is likely to trigger nss-ldap which in turn might use NSS to
952 ask systemd-resolved for hostname lookups. This will hence result in
953 a deadlock: a user name lookup in order to start
954 systemd-resolved.service will result in a host name lookup for which
955 systemd-resolved.service needs to be started already. There are
956 multiple ways to work around this problem: pre-allocate the
957 "systemd-resolve" user on such systems, so that nss-ldap won't be
958 triggered; or use a different NSS package that doesn't do networking
959 in-process but provides a local asynchronous name cache; or configure
960 the NSS package to avoid lookups for UIDs in the range `pkg-config
961 systemd --variable=dynamicuidmin` … `pkg-config systemd
962 --variable=dynamicuidmax`, so that it does not consider itself
963 authoritative for the same UID range systemd allocates dynamic users
966 * The systemd-resolve tool has been renamed to resolvectl (it also
967 remains available under the old name, for compatibility), and its
968 interface is now verb-based, similar in style to the other <xyz>ctl
969 tools, such as systemctl or loginctl.
971 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
972 compatibility. It may be symlinked under the 'resolvconf' name, in
973 which case it will take arguments and input compatible with the
974 Debian and FreeBSD resolvconf tool.
976 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
977 where the system initially suspends, and after a timeout resumes and
980 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
981 set the client will only send a DUID as client identifier.
983 * The nss-systemd glibc NSS module will now enumerate dynamic users and
984 groups in effect. Previously, it could resolve UIDs/GIDs to user
985 names/groups and vice versa, but did not support enumeration.
987 * journald's Compress= configuration setting now optionally accepts a
988 byte threshold value. All journal objects larger than this threshold
989 will be compressed, smaller ones will not. Previously this threshold
990 was not configurable and set to 512.
992 * A new system.conf setting NoNewPrivileges= is now available which may
993 be used to turn off acquisition of new privileges system-wide
994 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
995 for all its children). Note that turning this option on means setuid
996 binaries and file system capabilities lose their special powers.
997 While turning on this option is a big step towards a more secure
998 system, doing so is likely to break numerous pre-existing UNIX tools,
999 in particular su and sudo.
1001 * A new service systemd-time-sync-wait.service has been added. If
1002 enabled it will delay the time-sync.target unit at boot until time
1003 synchronization has been received from the network. This
1004 functionality is useful on systems lacking a local RTC or where it is
1005 acceptable that the boot process shall be delayed by external network
1008 * When hibernating, systemd will now inform the kernel of the image
1009 write offset, on kernels new enough to support this. This means swap
1010 files should work for hibernation now.
1012 * When loading unit files, systemd will now look for drop-in unit files
1013 extensions in additional places. Previously, for a unit file name
1014 "foo-bar-baz.service" it would look for dropin files in
1015 "foo-bar-baz.service.d/*.conf". Now, it will also look in
1016 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
1017 service name truncated after all inner dashes. This scheme allows
1018 writing drop-ins easily that apply to a whole set of unit files at
1019 once. It's particularly useful for mount and slice units (as their
1020 naming is prefix based), but is also useful for service and other
1021 units, for packages that install multiple unit files at once,
1022 following a strict naming regime of beginning the unit file name with
1023 the package's name. Two new specifiers are now supported in unit
1024 files to match this: %j and %J are replaced by the part of the unit
1025 name following the last dash.
1027 * Unit files and other configuration files that support specifier
1028 expansion now understand another three new specifiers: %T and %V will
1029 resolve to /tmp and /var/tmp respectively, or whatever temporary
1030 directory has been set for the calling user. %E will expand to either
1031 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
1033 * The ExecStart= lines of unit files are no longer required to
1034 reference absolute paths. If non-absolute paths are specified the
1035 specified binary name is searched within the service manager's
1036 built-in $PATH, which may be queried with 'systemd-path
1037 search-binaries-default'. It's generally recommended to continue to
1038 use absolute paths for all binaries specified in unit files.
1040 * Units gained a new load state "bad-setting", which is used when a
1041 unit file was loaded, but contained fatal errors which prevent it
1042 from being started (for example, a service unit has been defined
1043 lacking both ExecStart= and ExecStop= lines).
1045 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
1046 support alternative debuggers, for example lldb. The old name
1047 continues to be available however, for compatibility reasons. Use the
1048 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
1049 to pick an alternative debugger instead of the default gdb.
1051 * systemctl and the other tools will now output escape sequences that
1052 generate proper clickable hyperlinks in various terminal emulators
1053 where useful (for example, in the "systemctl status" output you can
1054 now click on the unit file name to quickly open it in the
1055 editor/viewer of your choice). Note that not all terminal emulators
1056 support this functionality yet, but many do. Unfortunately, the
1057 "less" pager doesn't support this yet, hence this functionality is
1058 currently automatically turned off when a pager is started (which
1059 happens quite often due to auto-paging). We hope to remove this
1060 limitation as soon as "less" learns these escape sequences. This new
1061 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
1062 environment variable. For details on these escape sequences see:
1063 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
1065 * networkd's .network files now support a new IPv6MTUBytes= option for
1066 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
1067 option in the [Route] section to configure the MTU to use for
1068 specific routes. It also gained support for configuration of the DHCP
1069 "UserClass" option through the new UserClass= setting. It gained
1070 three new options in the new [CAN] section for configuring CAN
1071 networks. The MULTICAST and ALLMULTI interface flags may now be
1072 controlled explicitly with the new Multicast= and AllMulticast=
1075 * networkd will now automatically make use of the kernel's route
1076 expiration feature, if it is available.
1078 * udevd's .link files now support setting the number of receive and
1079 transmit channels, using the RxChannels=, TxChannels=,
1080 OtherChannels=, CombinedChannels= settings.
1082 * Support for UDPSegmentationOffload= has been removed, given its
1083 limited support in hardware, and waning software support.
1085 * networkd's .netdev files now support creating "netdevsim" interfaces.
1087 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
1088 to query the unit belonging to a specific kernel control group.
1090 * systemd-analyze gained a new verb "cat-config", which may be used to
1091 dump the contents of any configuration file, with all its matching
1092 drop-in files added in, and honouring the usual search and masking
1093 logic applied to systemd configuration files. For example use
1094 "systemd-analyze cat-config systemd/system.conf" to get the complete
1095 system configuration file of systemd how it would be loaded by PID 1
1096 itself. Similar to this, various tools such as systemd-tmpfiles or
1097 systemd-sysusers, gained a new option "--cat-config", which does the
1098 corresponding operation for their own configuration settings. For
1099 example, "systemd-tmpfiles --cat-config" will now output the full
1100 list of tmpfiles.d/ lines in place.
1102 * timedatectl gained three new verbs: "show" shows bus properties of
1103 systemd-timedated, "timesync-status" shows the current NTP
1104 synchronization state of systemd-timesyncd, and "show-timesync"
1105 shows bus properties of systemd-timesyncd.
1107 * systemd-timesyncd gained a bus interface on which it exposes details
1110 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
1111 understood by systemd-timedated. It takes a colon-separated list of
1112 unit names of NTP client services. The list is used by
1113 "timedatectl set-ntp".
1115 * systemd-nspawn gained a new --rlimit= switch for setting initial
1116 resource limits for the container payload. There's a new switch
1117 --hostname= to explicitly override the container's hostname. A new
1118 --no-new-privileges= switch may be used to control the
1119 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
1120 --oom-score-adjust= switch controls the OOM scoring adjustment value
1121 for the payload. The new --cpu-affinity= switch controls the CPU
1122 affinity of the container payload. The new --resolv-conf= switch
1123 allows more detailed control of /etc/resolv.conf handling of the
1124 container. Similarly, the new --timezone= switch allows more detailed
1125 control of /etc/localtime handling of the container.
1127 * systemd-detect-virt gained a new --list switch, which will print a
1128 list of all currently known VM and container environments.
1130 * Support for "Portable Services" has been added, see
1131 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
1132 experimental, but this is expected to change soon. Reflecting this
1133 experimental state, the "portablectl" binary is not installed into
1134 /usr/bin yet. The binary has to be called with the full path
1135 /usr/lib/systemd/portablectl instead.
1137 * journalctl's and systemctl's -o switch now knows a new log output
1138 mode "with-unit". The output it generates is very similar to the
1139 regular "short" mode, but displays the unit name instead of the
1140 syslog tag for each log line. Also, the date is shown with timezone
1141 information. This mode is probably more useful than the classic
1142 "short" output mode for most purposes, except where pixel-perfect
1143 compatibility with classic /var/log/messages formatting is required.
1145 * A new --dump-bus-properties switch has been added to the systemd
1146 binary, which may be used to dump all supported D-Bus properties.
1147 (Options which are still supported, but are deprecated, are *not*
1150 * sd-bus gained a set of new calls:
1151 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
1152 enable/disable the "floating" state of a bus slot object,
1153 i.e. whether the slot object pins the bus it is allocated for into
1154 memory or if the bus slot object gets disconnected when the bus goes
1155 away. sd_bus_open_with_description(),
1156 sd_bus_open_user_with_description(),
1157 sd_bus_open_system_with_description() may be used to allocate bus
1158 objects and set their description string already during allocation.
1160 * sd-event gained support for watching inotify events from the event
1161 loop, in an efficient way, sharing inotify handles between multiple
1162 users. For this a new function sd_event_add_inotify() has been added.
1164 * sd-event and sd-bus gained support for calling special user-supplied
1165 destructor functions for userdata pointers associated with
1166 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
1167 functions sd_bus_slot_set_destroy_callback,
1168 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
1169 sd_bus_track_get_destroy_callback,
1170 sd_event_source_set_destroy_callback,
1171 sd_event_source_get_destroy_callback have been added.
1173 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
1175 * PID 1 will now automatically reschedule .timer units whenever the
1176 local timezone changes. (They previously got rescheduled
1177 automatically when the system clock changed.)
1179 * New documentation has been added to document cgroups delegation,
1180 portable services and the various code quality tools we have set up:
1182 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
1183 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
1184 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
1186 * The Boot Loader Specification has been added to the source tree.
1188 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
1190 While moving it into our source tree we have updated it and further
1191 changes are now accepted through the usual github PR workflow.
1193 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
1194 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
1195 earlier PAM modules. The data in these fields is used to initialize
1196 the session scope's resource properties. Thus external PAM modules
1197 may now configure per-session limits, for example sourced from
1198 external user databases.
1200 * socket units with Accept=yes will now maintain a "refused" counter in
1201 addition to the existing "accepted" counter, counting connections
1202 refused due to the enforced limits.
1204 * The "systemd-path search-binaries-default" command may now be use to
1205 query the default, built-in $PATH PID 1 will pass to the services it
1208 * A new unit file setting PrivateMounts= has been added. It's a boolean
1209 option. If enabled the unit's processes are invoked in their own file
1210 system namespace. Note that this behaviour is also implied if any
1211 other file system namespacing options (such as PrivateTmp=,
1212 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
1213 primarily useful for services that do not use any of the other file
1214 system namespacing options. One such service is systemd-udevd.service
1215 wher this is now used by default.
1217 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
1218 when the system is booted in UEFI "secure mode".
1220 * A new unit "system-update-pre.target" is added, which defines an
1221 optional synchronization point for offline system updates, as
1222 implemented by the pre-existing "system-update.target" unit. It
1223 allows ordering services before the service that executes the actual
1224 update process in a generic way.
1226 * Systemd now emits warnings whenever .include syntax is used.
1228 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
1229 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
1230 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
1231 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
1232 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
1233 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
1234 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
1235 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
1236 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
1237 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
1238 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
1239 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
1240 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
1241 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
1242 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
1243 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
1244 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
1245 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
1246 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
1247 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
1248 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
1249 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
1250 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
1251 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
1252 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
1253 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
1254 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
1255 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
1256 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1258 — Berlin, 2018-06-22
1262 * The MemoryAccounting= unit property now defaults to on. After
1263 discussions with the upstream control group maintainers we learnt
1264 that the negative impact of cgroup memory accounting on current
1265 kernels is finally relatively minimal, so that it should be safe to
1266 enable this by default without affecting system performance. Besides
1267 memory accounting only task accounting is turned on by default, all
1268 other forms of resource accounting (CPU, IO, IP) remain off for now,
1269 because it's not clear yet that their impact is small enough to move
1270 from opt-in to opt-out. We recommend downstreams to leave memory
1271 accounting on by default if kernel 4.14 or higher is primarily
1272 used. On very resource constrained systems or when support for old
1273 kernels is a necessity, -Dmemory-accounting-default=false can be used
1274 to revert this change.
1276 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
1277 %udev_rules_update) and the journal catalog (%journal_catalog_update)
1278 from the upgrade scriptlets of individual packages now do nothing.
1279 Transfiletriggers have been added which will perform those updates
1280 once at the end of the transaction.
1282 Similar transfiletriggers have been added to execute any sysctl.d
1283 and binfmt.d rules. Thus, it should be unnecessary to provide any
1284 scriptlets to execute this configuration from package installation
1287 * systemd-sysusers gained a mode where the configuration to execute is
1288 specified on the command line, but this configuration is not executed
1289 directly, but instead it is merged with the configuration on disk,
1290 and the result is executed. This is useful for package installation
1291 scripts which want to create the user before installing any files on
1292 disk (in case some of those files are owned by that user), while
1293 still allowing local admin overrides.
1295 This functionality is exposed to rpm scriptlets through a new
1296 %sysusers_create_package macro. Old %sysusers_create and
1297 %sysusers_create_inline macros are deprecated.
1299 A transfiletrigger for sysusers.d configuration is now installed,
1300 which means that it should be unnecessary to call systemd-sysusers from
1301 package installation scripts, unless the package installs any files
1302 owned by those newly-created users, in which case
1303 %sysusers_create_package should be used.
1305 * Analogous change has been done for systemd-tmpfiles: it gained a mode
1306 where the command-line configuration is merged with the configuration
1307 on disk. This is exposed as the new %tmpfiles_create_package macro,
1308 and %tmpfiles_create is deprecated. A transfiletrigger is installed
1309 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
1310 from package installation scripts.
1312 * sysusers.d configuration for a user may now also specify the group
1313 number, in addition to the user number ("u username 123:456"), or
1314 without the user number ("u username -:456").
1316 * Configution items for systemd-sysusers can now be specified as
1317 positional arguments when the new --inline switch is used.
1319 * The login shell of users created through sysusers.d may now be
1320 specified (previously, it was always /bin/sh for root and
1321 /sbin/nologin for other users).
1323 * systemd-analyze gained a new --global switch to look at global user
1324 configuration. It also gained a unit-paths verb to list the unit load
1325 paths that are compiled into systemd (which can be used with
1326 --systemd, --user, or --global).
1328 * udevadm trigger gained a new --settle/-w option to wait for any
1329 triggered events to finish (but just those, and not any other events
1330 which are triggered meanwhile).
1332 * The action that systemd-logind takes when the lid is closed and the
1333 machine is connected to external power can now be configured using
1334 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
1335 was determined by HandleLidSwitch=, and, for backwards compatibility,
1336 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
1338 * journalctl will periodically call sd_journal_process() to make it
1339 resilient against inotify queue overruns when journal files are
1340 rotated very quickly.
1342 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
1343 sd_bus_get_n_queued_write — may be used to check the number of
1344 pending bus messages.
1346 * systemd gained a new
1347 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
1348 which can be used to migrate foreign processes to scope and service
1349 units. The primary user for this new API is systemd itself: the
1350 systemd --user instance uses this call of the systemd --system
1351 instance to migrate processes if it itself gets the request to
1352 migrate processes and the kernel refuses this due to access
1353 restrictions. Thanks to this "systemd-run --scope --user …" works
1354 again in pure cgroup v2 environments when invoked from the user
1357 * A new TemporaryFileSystem= setting can be used to mask out part of
1358 the real file system tree with tmpfs mounts. This may be combined
1359 with BindPaths= and BindReadOnlyPaths= to hide files or directories
1360 not relevant to the unit, while still allowing some paths lower in
1361 the tree to be accessed.
1363 ProtectHome=tmpfs may now be used to hide user home and runtime
1364 directories from units, in a way that is mostly equivalent to
1365 "TemporaryFileSystem=/home /run/user /root".
1367 * Non-service units are now started with KeyringMode=shared by default.
1368 This means that mount and swapon and other mount tools have access
1369 to keys in the main keyring.
1371 * /sys/fs/bpf is now mounted automatically.
1373 * QNX virtualization is now detected by systemd-detect-virt and may
1374 be used in ConditionVirtualization=.
1376 * IPAccounting= may now be enabled also for slice units.
1378 * A new -Dsplit-bin= build configuration switch may be used to specify
1379 whether bin and sbin directories are merged, or if they should be
1380 included separately in $PATH and various listings of executable
1381 directories. The build configuration scripts will try to autodetect
1382 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
1383 system, but distributions are encouraged to configure this
1386 * A new -Dok-color= build configuration switch may be used to change
1387 the colour of "OK" status messages.
1389 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
1390 PrivateNetwork=yes was buggy in previous versions of systemd. This
1391 means that after the upgrade and daemon-reexec, any such units must
1394 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
1395 will not exclude read-only files owned by root from cleanup.
1397 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
1398 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
1399 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
1400 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
1401 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
1402 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
1403 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
1404 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
1405 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
1406 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
1407 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
1408 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
1409 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
1410 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
1411 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
1412 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
1414 — Warsaw, 2018-03-05
1418 * Some keyboards come with a zoom see-saw or rocker which until now got
1419 mapped to the Linux "zoomin/out" keys in hwdb. However, these
1420 keycodes are not recognized by any major desktop. They now produce
1421 Up/Down key events so that they can be used for scrolling.
1423 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
1424 slightly: previously, if an argument was specified for lines of this
1425 type (i.e. the right-most column was set) this string was appended to
1426 existing files each time systemd-tmpfiles was run. This behaviour was
1427 different from what the documentation said, and not particularly
1428 useful, as repeated systemd-tmpfiles invocations would not be
1429 idempotent and grow such files without bounds. With this release
1430 behaviour has been altered to match what the documentation says:
1431 lines of this type only have an effect if the indicated files don't
1432 exist yet, and only then the argument string is written to the file.
1434 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
1435 systemd-tmpfiles behaviour: previously, read-only files owned by root
1436 were always excluded from the file "aging" algorithm (i.e. the
1437 automatic clean-up of directories like /tmp based on
1438 atime/mtime/ctime). We intend to drop this restriction, and age files
1439 by default even when owned by root and read-only. This behaviour was
1440 inherited from older tools, but there have been requests to remove
1441 it, and it's not obvious why this restriction was made in the first
1442 place. Please speak up now, if you are aware of software that reqires
1443 this behaviour, otherwise we'll remove the restriction in v238.
1445 * A new environment variable $SYSTEMD_OFFLINE is now understood by
1446 systemctl. It takes a boolean argument. If on, systemctl assumes it
1447 operates on an "offline" OS tree, and will not attempt to talk to the
1448 service manager. Previously, this mode was implicitly enabled if a
1449 chroot() environment was detected, and this new environment variable
1450 now provides explicit control.
1452 * .path and .socket units may now be created transiently, too.
1453 Previously only service, mount, automount and timer units were
1454 supported as transient units. The systemd-run tool has been updated
1455 to expose this new functionality, you may hence use it now to bind
1456 arbitrary commands to path or socket activation on-the-fly from the
1457 command line. Moreover, almost all properties are now exposed for the
1458 unit types that already supported transient operation.
1460 * The systemd-mount command gained support for a new --owner= parameter
1461 which takes a user name, which is then resolved and included in uid=
1462 and gid= mount options string of the file system to mount.
1464 * A new unit condition ConditionControlGroupController= has been added
1465 that checks whether a specific cgroup controller is available.
1467 * Unit files, udev's .link files, and systemd-networkd's .netdev and
1468 .network files all gained support for a new condition
1469 ConditionKernelVersion= for checking against specific kernel
1472 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
1473 support for configuring device flags in the Flags= setting. In the
1474 same files, the [Tunnel] section gained support for configuring
1475 AllowLocalRemote=. The [Route] section in .network files gained
1476 support for configuring InitialCongestionWindow=,
1477 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
1478 understands RapidCommit=.
1480 * systemd-networkd's DHCPv6 support gained support for Prefix
1483 * sd-bus gained support for a new "watch-bind" feature. When this
1484 feature is enabled, an sd_bus connection may be set up to connect to
1485 an AF_UNIX socket in the file system as soon as it is created. This
1486 functionality is useful for writing early-boot services that
1487 automatically connect to the system bus as soon as it is started,
1488 without ugly time-based polling. systemd-networkd and
1489 systemd-resolved have been updated to make use of this
1490 functionality. busctl exposes this functionality in a new
1491 --watch-bind= command line switch.
1493 * sd-bus will now optionally synthesize a local "Connected" signal as
1494 soon as a D-Bus connection is set up fully. This message mirrors the
1495 already existing "Disconnected" signal which is synthesized when the
1496 connection is terminated. This signal is generally useful but
1497 particularly handy in combination with the "watch-bind" feature
1498 described above. Synthesizing of this message has to be requested
1499 explicitly through the new API call sd_bus_set_connected_signal(). In
1500 addition a new call sd_bus_is_ready() has been added that checks
1501 whether a connection is fully set up (i.e. between the "Connected" and
1502 "Disconnected" signals).
1504 * sd-bus gained two new calls sd_bus_request_name_async() and
1505 sd_bus_release_name_async() for asynchronously registering bus
1506 names. Similar, there is now sd_bus_add_match_async() for installing
1507 a signal match asynchronously. All of systemd's own services have
1508 been updated to make use of these calls. Doing these operations
1509 asynchronously has two benefits: it reduces the risk of deadlocks in
1510 case of cyclic dependencies between bus services, and it speeds up
1511 service initialization since synchronization points for bus
1512 round-trips are removed.
1514 * sd-bus gained two new calls sd_bus_match_signal() and
1515 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
1516 and sd_bus_add_match_async() but instead of taking a D-Bus match
1517 string take match fields as normal function parameters.
1519 * sd-bus gained two new calls sd_bus_set_sender() and
1520 sd_bus_message_set_sender() for setting the sender name of outgoing
1521 messages (either for all outgoing messages or for just one specific
1522 one). These calls are only useful in direct connections as on
1523 brokered connections the broker fills in the sender anyway,
1524 overwriting whatever the client filled in.
1526 * sd-event gained a new pseudo-handle that may be specified on all API
1527 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
1528 used this refers to the default event loop object of the calling
1529 thread. Note however that this does not implicitly allocate one —
1530 which has to be done prior by using sd_event_default(). Similarly
1531 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
1532 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
1533 to the default bus of the specified type of the calling thread. Here
1534 too this does not implicitly allocate bus connection objects, this
1535 has to be done prior with sd_bus_default() and friends.
1537 * sd-event gained a new call pair
1538 sd_event_source_{get|set}_io_fd_own(). This may be used to request
1539 automatic closure of the file descriptor an IO event source watches
1540 when the event source is destroyed.
1542 * systemd-networkd gained support for natively configuring WireGuard
1545 * In previous versions systemd synthesized user records both for the
1546 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
1547 internally. In order to simplify distribution-wide renames of the
1548 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
1549 new transitional flag file has been added: if
1550 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
1551 user and group record within the systemd codebase is disabled.
1553 * systemd-notify gained a new --uid= option for selecting the source
1554 user/UID to use for notification messages sent to the service
1557 * journalctl gained a new --grep= option to list only entries in which
1558 the message matches a certain pattern. By default matching is case
1559 insensitive if the pattern is lowercase, and case sensitive
1560 otherwise. Option --case-sensitive=yes|no can be used to override
1561 this an specify case sensitivity or case insensitivity.
1563 * There's now a "systemd-analyze service-watchdogs" command for printing
1564 the current state of the service runtime watchdog, and optionally
1565 enabling or disabling the per-service watchdogs system-wide if given a
1566 boolean argument (i.e. the concept you configure in WatchdogSec=), for
1567 debugging purposes. There's also a kernel command line option
1568 systemd.service_watchdogs= for controlling the same.
1570 * Two new "log-level" and "log-target" options for systemd-analyze were
1571 added that merge the now deprecated get-log-level, set-log-level and
1572 get-log-target, set-log-target pairs. The deprecated options are still
1573 understood for backwards compatibility. The two new options print the
1574 current value when no arguments are given, and set them when a
1575 level/target is given as an argument.
1577 * sysusers.d's "u" lines now optionally accept both a UID and a GID
1578 specification, separated by a ":" character, in order to create users
1579 where UID and GID do not match.
1581 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
1582 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
1583 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
1584 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
1585 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
1586 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
1587 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
1588 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
1589 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
1590 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
1591 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
1592 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
1593 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
1594 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
1595 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
1596 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
1597 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
1598 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
1599 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
1600 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
1607 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
1608 in v235 has been extended to also set the dummy.ko module option
1609 numdummies=0, preventing the kernel from automatically creating
1610 dummy0. All dummy interfaces must now be explicitly created.
1612 * Unknown '%' specifiers in configuration files are now rejected. This
1613 applies to units and tmpfiles.d configuration. Any percent characters
1614 that are followed by a letter or digit that are not supposed to be
1615 interpreted as the beginning of a specifier should be escaped by
1616 doubling ("%%"). (So "size=5%" is still accepted, as well as
1617 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
1618 valid specifiers today.)
1620 * systemd-resolved now maintains a new dynamic
1621 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
1622 recommended to make /etc/resolv.conf a symlink to it. This file
1623 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
1624 includes dynamically acquired search domains, achieving more correct
1625 DNS resolution by software that bypasses local DNS APIs such as NSS.
1627 * The "uaccess" udev tag has been dropped from /dev/kvm and
1628 /dev/dri/renderD*. These devices now have the 0666 permissions by
1629 default (but this may be changed at build-time). /dev/dri/renderD*
1630 will now be owned by the "render" group along with /dev/kfd.
1632 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
1633 systemd-journal-gatewayd.service and
1634 systemd-journal-upload.service. This means "nss-systemd" must be
1635 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
1636 services are resolved properly.
1638 * In /etc/fstab two new mount options are now understood:
1639 x-systemd.makefs and x-systemd.growfs. The former has the effect that
1640 the configured file system is formatted before it is mounted, the
1641 latter that the file system is resized to the full block device size
1642 after it is mounted (i.e. if the file system is smaller than the
1643 partition it resides on, it's grown). This is similar to the fsck
1644 logic in /etc/fstab, and pulls in systemd-makefs@.service and
1645 systemd-growfs@.service as necessary, similar to
1646 systemd-fsck@.service. Resizing is currently only supported on ext4
1649 * In systemd-networkd, the IPv6 RA logic now optionally may announce
1650 DNS server and domain information.
1652 * Support for the LUKS2 on-disk format for encrypted partitions has
1653 been added. This requires libcryptsetup2 during compilation and
1656 * The systemd --user instance will now signal "readiness" when its
1657 basic.target unit has been reached, instead of when the run queue ran
1658 empty for the first time.
1660 * Tmpfiles.d with user configuration are now also supported.
1661 systemd-tmpfiles gained a new --user switch, and snippets placed in
1662 ~/.config/user-tmpfiles.d/ and corresponding directories will be
1663 executed by systemd-tmpfiles --user running in the new
1664 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
1665 running in the user session.
1667 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
1668 %S resolves to the top-level state directory (/var/lib for the system
1669 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
1670 top-level cache directory (/var/cache for the system instance,
1671 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
1672 logs directory (/var/log for the system instance,
1673 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
1674 existing %t specifier, that resolves to the top-level runtime
1675 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
1678 * journalctl learnt a new parameter --output-fields= for limiting the
1679 set of journal fields to output in verbose and JSON output modes.
1681 * systemd-timesyncd's configuration file gained a new option
1682 RootDistanceMaxSec= for setting the maximum root distance of servers
1683 it'll use, as well as the new options PollIntervalMinSec= and
1684 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
1686 * bootctl gained a new command "list" for listing all available boot
1687 menu items on systems that follow the boot loader specification.
1689 * systemctl gained a new --dry-run switch that shows what would be done
1690 instead of doing it, and is currently supported by the shutdown and
1693 * ConditionSecurity= can now detect the TOMOYO security module.
1695 * Unit file [Install] sections are now also respected in unit drop-in
1696 files. This is intended to be used by drop-ins under /usr/lib/.
1698 * systemd-firstboot may now also set the initial keyboard mapping.
1700 * Udev "changed" events for devices which are exposed as systemd
1701 .device units are now propagated to units specified in
1702 ReloadPropagatedFrom= as reload requests.
1704 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
1705 unit template name (i.e. a name in the form of 'foobar@.service',
1706 without the instance component between the '@' and - the '.'), then
1707 the escaped sysfs path of the device is automatically used as the
1710 * SystemCallFilter= in unit files has been extended so that an "errno"
1711 can be specified individually for each system call. Example:
1712 SystemCallFilter=~uname:EILSEQ.
1714 * The cgroup delegation logic has been substantially updated. Delegate=
1715 now optionally takes a list of controllers (instead of a boolean, as
1716 before), which lists the controllers to delegate at least.
1718 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
1720 * A new LogLevelMax= setting configures the maximum log level any
1721 process of the service may log at (i.e. anything with a lesser
1722 priority than what is specified is automatically dropped). A new
1723 LogExtraFields= setting allows configuration of additional journal
1724 fields to attach to all log records generated by any of the unit's
1727 * New StandardInputData= and StandardInputText= settings along with the
1728 new option StandardInput=data may be used to configure textual or
1729 binary data that shall be passed to the executed service process via
1730 standard input, encoded in-line in the unit file.
1732 * StandardInput=, StandardOutput= and StandardError= may now be used to
1733 connect stdin/stdout/stderr of executed processes directly with a
1734 file or AF_UNIX socket in the file system, using the new "file:" option.
1736 * A new unit file option CollectMode= has been added, that allows
1737 tweaking the garbage collection logic for units. It may be used to
1738 tell systemd to garbage collect units that have failed automatically
1739 (normally it only GCs units that exited successfully). systemd-run
1740 and systemd-mount expose this new functionality with a new -G option.
1742 * "machinectl bind" may now be used to bind mount non-directories
1743 (i.e. regularfiles, devices, fifos, sockets).
1745 * systemd-analyze gained a new verb "calendar" for validating and
1746 testing calendar time specifications to use for OnCalendar= in timer
1747 units. Besides validating the expression it will calculate the next
1748 time the specified expression would elapse.
1750 * In addition to the pre-existing FailureAction= unit file setting
1751 there's now SuccessAction=, for configuring a shutdown action to
1752 execute when a unit completes successfully. This is useful in
1753 particular inside containers that shall terminate after some workload
1754 has been completed. Also, both options are now supported for all unit
1755 types, not just services.
1757 * networkds's IP rule support gained two new options
1758 IncomingInterface= and OutgoingInterface= for configuring the incoming
1759 and outgoing interfaces of configured rules. systemd-networkd also
1760 gained support for "vxcan" network devices.
1762 * networkd gained a new setting RequiredForOnline=, taking a
1763 boolean. If set, systemd-wait-online will take it into consideration
1764 when determining that the system is up, otherwise it will ignore the
1765 interface for this purpose.
1767 * The sd_notify() protocol gained support for a new operation: with
1768 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
1769 store again, ahead of POLLHUP or POLLERR when they are removed
1772 * A new document doc/UIDS-GIDS.md has been added to the source tree,
1773 that documents the UID/GID range and assignment assumptions and
1774 requirements of systemd.
1776 * The watchdog device PID 1 will ping may now be configured through the
1777 WatchdogDevice= configuration file setting, or by setting the
1778 systemd.watchdog_service= kernel commandline option.
1780 * systemd-resolved's gained support for registering DNS-SD services on
1781 the local network using MulticastDNS. Services may either be
1782 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
1783 the same dir below /run, /usr/lib), or through its D-Bus API.
1785 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
1786 extend the effective start, runtime, and stop time. The service must
1787 continue to send EXTEND_TIMEOUT_USEC within the period specified to
1788 prevent the service manager from making the service as timedout.
1790 * systemd-resolved's DNSSEC support gained support for RFC 8080
1791 (Ed25519 keys and signatures).
1793 * The systemd-resolve command line tool gained a new set of options
1794 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
1795 --set-nta= and --revert to configure per-interface DNS configuration
1796 dynamically during runtime. It's useful for pushing DNS information
1797 into systemd-resolved from DNS hook scripts that various interface
1798 managing software supports (such as pppd).
1800 * systemd-nspawn gained a new --network-namespace-path= command line
1801 option, which may be used to make a container join an existing
1802 network namespace, by specifying a path to a "netns" file.
1804 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
1805 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
1806 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
1807 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
1808 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
1809 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
1810 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
1811 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
1812 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
1813 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
1814 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
1815 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
1816 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
1817 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
1818 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
1819 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
1820 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
1821 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
1822 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
1823 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
1824 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
1825 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
1826 Jędrzejewski-Szmek, Zeal Jagannatha
1828 — Berlin, 2017-12-14
1832 * INCOMPATIBILITY: systemd-logind.service and other long-running
1833 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
1834 communication with the outside. This generally improves security of
1835 the system, and is in almost all cases a safe and good choice, as
1836 these services do not and should not provide any network-facing
1837 functionality. However, systemd-logind uses the glibc NSS API to
1838 query the user database. This creates problems on systems where NSS
1839 is set up to directly consult network services for user database
1840 lookups. In particular, this creates incompatibilities with the
1841 "nss-nis" module, which attempts to directly contact the NIS/YP
1842 network servers it is configured for, and will now consistently
1843 fail. In such cases, it is possible to turn off IP sandboxing for
1844 systemd-logind.service (set IPAddressDeny= in its [Service] section
1845 to the empty string, via a .d/ unit file drop-in). Downstream
1846 distributions might want to update their nss-nis packaging to include
1847 such a drop-in snippet, accordingly, to hide this incompatibility
1848 from the user. Another option is to make use of glibc's nscd service
1849 to proxy such network requests through a privilege-separated, minimal
1850 local caching daemon, or to switch to more modern technologies such
1851 sssd, whose NSS hook-ups generally do not involve direct network
1852 access. In general, we think it's definitely time to question the
1853 implementation choices of nss-nis, i.e. whether it's a good idea
1854 today to embed a network-facing loadable module into all local
1855 processes that need to query the user database, including the most
1856 trivial and benign ones, such as "ls". For more details about
1857 IPAddressDeny= see below.
1859 * A new modprobe.d drop-in is now shipped by default that sets the
1860 bonding module option max_bonds=0. This overrides the kernel default,
1861 to avoid conflicts and ambiguity as to whether or not bond0 should be
1862 managed by systemd-networkd or not. This resolves multiple issues
1863 with bond0 properties not being applied, when bond0 is configured
1864 with systemd-networkd. Distributors may choose to not package this,
1865 however in that case users will be prevented from correctly managing
1866 bond0 interface using systemd-networkd.
1868 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
1869 which print the logging level and target of the system manager. They
1870 complement the existing "set-log-level" and "set-log-target" verbs
1871 used to change those values.
1873 * journald.conf gained a new boolean setting ReadKMsg= which defaults
1874 to on. If turned off kernel log messages will not be read by
1875 systemd-journald or included in the logs. It also gained a new
1876 setting LineMax= for configuring the maximum line length in
1877 STDOUT/STDERR log streams. The new default for this value is 48K, up
1878 from the previous hardcoded 2048.
1880 * A new unit setting RuntimeDirectoryPreserve= has been added, which
1881 allows more detailed control of what to do with a runtime directory
1882 configured with RuntimeDirectory= (i.e. a directory below /run or
1883 $XDG_RUNTIME_DIR) after a unit is stopped.
1885 * The RuntimeDirectory= setting for units gained support for creating
1886 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
1887 one top-level directory.
1889 * Units gained new options StateDirectory=, CacheDirectory=,
1890 LogsDirectory= and ConfigurationDirectory= which are closely related
1891 to RuntimeDirectory= but manage per-service directories below
1892 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
1893 possible to write unit files which when activated automatically gain
1894 properly owned service specific directories in these locations, thus
1895 making unit files self-contained and increasing compatibility with
1896 stateless systems and factory reset where /etc or /var are
1897 unpopulated at boot. Matching these new settings there's also
1898 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
1899 ConfigurationDirectoryMode= for configuring the access mode of these
1900 directories. These settings are particularly useful in combination
1901 with DynamicUser=yes as they provide secure, properly-owned,
1902 writable, and stateful locations for storage, excluded from the
1903 sandbox that such services live in otherwise.
1905 * Automake support has been removed from this release. systemd is now
1908 * systemd-journald will now aggressively cache client metadata during
1909 runtime, speeding up log write performance under pressure. This comes
1910 at a small price though: as much of the metadata is read
1911 asynchronously from /proc/ (and isn't implicitly attached to log
1912 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
1913 metadata stored alongside a log entry might be slightly
1914 out-of-date. Previously it could only be slightly newer than the log
1915 message. The time window is small however, and given that the kernel
1916 is unlikely to be improved anytime soon in this regard, this appears
1919 * nss-myhostname/systemd-resolved will now by default synthesize an
1920 A/AAAA resource record for the "_gateway" hostname, pointing to the
1921 current default IP gateway. Previously it did that for the "gateway"
1922 name, hampering adoption, as some distributions wanted to leave that
1923 host name open for local use. The old behaviour may still be
1924 requested at build time.
1926 * systemd-networkd's [Address] section in .network files gained a new
1927 Scope= setting for configuring the IP address scope. The [Network]
1928 section gained a new boolean setting ConfigureWithoutCarrier= that
1929 tells systemd-networkd to ignore link sensing when configuring the
1930 device. The [DHCP] section gained a new Anonymize= boolean option for
1931 turning on a number of options suggested in RFC 7844. A new
1932 [RoutingPolicyRule] section has been added for configuring the IP
1933 routing policy. The [Route] section has gained support for a new
1934 Type= setting which permits configuring
1935 blackhole/unreachable/prohibit routes.
1937 * The [VRF] section in .netdev files gained a new Table= setting for
1938 configuring the routing table to use. The [Tunnel] section gained a
1939 new Independent= boolean field for configuring tunnels independent of
1940 an underlying network interface. The [Bridge] section gained a new
1941 GroupForwardMask= option for configuration of propagation of link
1942 local frames between bridge ports.
1944 * The WakeOnLan= setting in .link files gained support for a number of
1945 new modes. A new TCP6SegmentationOffload= setting has been added for
1946 configuring TCP/IPv6 hardware segmentation offload.
1948 * The IPv6 RA sender implementation may now optionally send out RDNSS
1949 and RDNSSL records to supply DNS configuration to peers.
1951 * systemd-nspawn gained support for a new --system-call-filter= command
1952 line option for adding and removing entries in the default system
1953 call filter it applies. Moreover systemd-nspawn has been changed to
1954 implement a system call whitelist instead of a blacklist.
1956 * systemd-run gained support for a new --pipe command line option. If
1957 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
1958 are directly passed on to the activated transient service
1959 executable. This allows invoking arbitrary processes as systemd
1960 services (for example to take benefit of dependency management,
1961 accounting management, resource management or log management that is
1962 done automatically for services) — while still allowing them to be
1963 integrated in a classic UNIX shell pipeline.
1965 * When a service sends RELOAD=1 via sd_notify() and reload propagation
1966 using ReloadPropagationTo= is configured, a reload is now propagated
1967 to configured units. (Previously this was only done on explicitly
1968 requested reloads, using "systemctl reload" or an equivalent
1971 * For each service unit a restart counter is now kept: it is increased
1972 each time the service is restarted due to Restart=, and may be
1973 queried using "systemctl show -p NRestarts …".
1975 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
1976 @signal and @timer have been added, for usage with SystemCallFilter=
1977 in unit files and the new --system-call-filter= command line option
1978 of systemd-nspawn (see above).
1980 * ExecStart= lines in unit files gained two new modifiers: when a
1981 command line is prefixed with "!" the command will be executed as
1982 configured, except for the credentials applied by
1983 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
1984 "+", but does still apply namespacing options unlike "+". There's
1985 also "!!" now, which is mostly identical, but becomes a NOP on
1986 systems that support ambient capabilities. This is useful to write
1987 unit files that work with ambient capabilities where possible but
1988 automatically fall back to traditional privilege dropping mechanisms
1989 on systems where this is not supported.
1991 * ListenNetlink= settings in socket units now support RDMA netlink
1994 * A new unit file setting LockPersonality= has been added which permits
1995 locking down the chosen execution domain ("personality") of a service
1998 * A new special target "getty-pre.target" has been added, which is
1999 ordered before all text logins, and may be used to order services
2000 before textual logins acquire access to the console.
2002 * systemd will now attempt to load the virtio-rng.ko kernel module very
2003 early on if a VM environment supporting this is detected. This should
2004 improve entropy during early boot in virtualized environments.
2006 * A _netdev option is now supported in /etc/crypttab that operates in a
2007 similar way as the same option in /etc/fstab: it permits configuring
2008 encrypted devices that need to be ordered after the network is up.
2009 Following this logic, two new special targets
2010 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
2011 added that are to cryptsetup.target what remote-fs.target and
2012 remote-fs-pre.target are to local-fs.target.
2014 * Service units gained a new UnsetEnvironment= setting which permits
2015 unsetting specific environment variables for services that are
2016 normally passed to it (for example in order to mask out locale
2017 settings for specific services that can't deal with it).
2019 * Units acquired a new boolean option IPAccounting=. When turned on, IP
2020 traffic accounting (packet count as well as byte count) is done for
2021 the service, and shown as part of "systemctl status" or "systemd-run
2024 * Service units acquired two new options IPAddressAllow= and
2025 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
2026 for configuring a simple IP access control list for all sockets of
2027 the unit. These options are available also on .slice and .socket
2028 units, permitting flexible access list configuration for individual
2029 services as well as groups of services (as defined by a slice unit),
2030 including system-wide. Note that IP ACLs configured this way are
2031 enforced on every single IPv4 and IPv6 socket created by any process
2032 of the service unit, and apply to ingress as well as egress traffic.
2034 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
2035 structured log message is generated each time the unit is stopped,
2036 containing information about the consumed resources of this
2039 * A new setting KeyringMode= has been added to unit files, which may be
2040 used to control how the kernel keyring is set up for executed
2043 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
2044 "systemctl kexec" and "systemctl exit" are now always asynchronous in
2045 behaviour (that is: these commands return immediately after the
2046 operation was enqueued instead of waiting for the operation to
2047 complete). Previously, "systemctl poweroff" and "systemctl reboot"
2048 were asynchronous on systems using systemd-logind (i.e. almost
2049 always, and like they were on sysvinit), and the other three commands
2050 were unconditionally synchronous. With this release this is cleaned
2051 up, and callers will see the same asynchronous behaviour on all
2052 systems for all five operations.
2054 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
2057 * .timer units now accept calendar specifications in other timezones
2058 than UTC or the local timezone.
2060 * The tmpfiles snippet var.conf has been changed to create
2061 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
2062 the "utmp" group already, and it appears to be generally understood
2063 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
2064 databases. Previously this was implemented correctly for all these
2065 databases excepts btmp, which has been opened up like this now
2066 too. Note that while the other databases are world-readable
2067 (i.e. 0644), btmp is not and remains more restrictive.
2069 * The systemd-resolve tool gained a new --reset-server-features
2070 switch. When invoked like this systemd-resolved will forget
2071 everything it learnt about the features supported by the configured
2072 upstream DNS servers, and restarts the feature probing logic on the
2073 next resolver look-up for them at the highest feature level
2076 * The status dump systemd-resolved sends to the logs upon receiving
2077 SIGUSR1 now also includes information about all DNS servers it is
2078 configured to use, and the features levels it probed for them.
2080 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
2081 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
2082 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
2083 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
2084 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
2085 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
2086 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
2087 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
2088 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
2089 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
2090 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
2091 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
2092 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
2093 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
2094 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
2095 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
2096 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
2097 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
2098 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
2099 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2101 — Berlin, 2017-10-06
2105 * Meson is now supported as build system in addition to Automake. It is
2106 our plan to remove Automake in one of our next releases, so that
2107 Meson becomes our exclusive build system. Hence, please start using
2108 the Meson build system in your downstream packaging. There's plenty
2109 of documentation around how to use Meson, the extremely brief
2112 ./autogen.sh && ./configure && make && sudo make install
2116 meson build && ninja -C build && sudo ninja -C build install
2118 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
2119 which permits configuring a timeout on the time a job is
2120 running. This is particularly useful for setting timeouts on jobs for
2123 * Unit files gained two new options ConditionUser= and ConditionGroup=
2124 for conditionalizing units based on the identity of the user/group
2125 running a systemd user instance.
2127 * systemd-networkd now understands a new FlowLabel= setting in the
2128 [VXLAN] section of .network files, as well as a Priority= in
2129 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
2130 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
2131 gained support for configuration of GENEVE links, and IPv6 address
2132 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
2134 * .link files now understand a new Port= setting.
2136 * systemd-networkd's DHCP support gained support for DHCP option 119
2137 (domain search list).
2139 * systemd-networkd gained support for serving IPv6 address ranges using
2140 the Router Advertisement protocol. The new .network configuration
2141 section [IPv6Prefix] may be used to configure the ranges to
2142 serve. This is implemented based on a new, minimal, native server
2143 implementation of RA.
2145 * journalctl's --output= switch gained support for a new parameter
2146 "short-iso-precise" for a mode where timestamps are shown as precise
2149 * systemd-udevd's "net_id" builtin may now generate stable network
2150 interface names from IBM PowerVM VIO devices as well as ACPI platform
2153 * MulticastDNS support in systemd-resolved may now be explicitly
2154 enabled/disabled using the new MulticastDNS= configuration file
2157 * systemd-resolved may now optionally use libidn2 instead of the libidn
2158 for processing internationalized domain names. Support for libidn2
2159 should be considered experimental and should not be enabled by
2162 * "machinectl pull-tar" and related call may now do verification of
2163 downloaded images using SUSE-style .sha256 checksum files in addition
2164 to the already existing support for validating using Ubuntu-style
2167 * sd-bus gained support for a new sd_bus_message_appendv() call which
2168 is va_list equivalent of sd_bus_message_append().
2170 * sd-boot gained support for validating images using SHIM/MOK.
2172 * The SMACK code learnt support for "onlycap".
2174 * systemd-mount --umount is now much smarter in figuring out how to
2175 properly unmount a device given its mount or device path.
2177 * The code to call libnss_dns as a fallback from libnss_resolve when
2178 the communication with systemd-resolved fails was removed. This
2179 fallback was redundant and interfered with the [!UNAVAIL=return]
2180 suffix. See nss-resolve(8) for the recommended configuration.
2182 * systemd-logind may now be restarted without losing state. It stores
2183 the file descriptors for devices it manages in the system manager
2184 using the FDSTORE= mechanism. Please note that further changes in
2185 other components may be required to make use of this (for example
2186 Xorg has code to listen for stops of systemd-logind and terminate
2187 itself when logind is stopped or restarted, in order to avoid using
2188 stale file descriptors for graphical devices, which is now
2189 counterproductive and must be reverted in order for restarts of
2190 systemd-logind to be safe. See
2191 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
2193 * All kernel-install plugins are called with the environment variable
2194 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
2195 /etc/machine-id. If the machine ID could not be determined,
2196 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
2197 anything in the entry directory (passed as the second argument) if
2198 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatiblity, a
2199 temporary directory is passed as the entry directory and removed
2200 after all the plugins exit.
2202 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
2203 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
2204 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
2205 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
2206 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
2207 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
2208 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
2209 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
2210 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
2211 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
2212 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
2213 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
2214 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
2215 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
2216 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
2217 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
2218 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
2219 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
2220 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
2221 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
2222 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
2223 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
2224 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
2225 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
2226 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
2227 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
2228 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
2229 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
2230 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
2233 — Berlin, 2017-07-12
2237 * The "hybrid" control group mode has been modified to improve
2238 compatibility with "legacy" cgroups-v1 setups. Specifically, the
2239 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
2240 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
2241 cgroups-v1 hierarchy), the only externally visible change being that
2242 the cgroups-v2 hierarchy is also mounted, to
2243 /sys/fs/cgroup/unified. This should provide a large degree of
2244 compatibility with "legacy" cgroups-v1, while taking benefit of the
2245 better management capabilities of cgroups-v2.
2247 * The default control group setup mode may be selected both a boot-time
2248 via a set of kernel command line parameters (specifically:
2249 systemd.unified_cgroup_hierarchy= and
2250 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
2251 default selected on the configure command line
2252 (--with-default-hierarchy=). The upstream default is "hybrid"
2253 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
2254 this will change in a future systemd version to be "unified" (pure
2255 cgroups-v2 mode). The third option for the compile time option is
2256 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
2257 distributions to default to "hybrid" mode for release distributions,
2258 starting with v233. We recommend "unified" for development
2259 distributions (specifically: distributions such as Fedora's rawhide)
2260 as that's where things are headed in the long run. Use "legacy" for
2261 greatest stability and compatibility only.
2263 * Note one current limitation of "unified" and "hybrid" control group
2264 setup modes: the kernel currently does not permit the systemd --user
2265 instance (i.e. unprivileged code) to migrate processes between two
2266 disconnected cgroup subtrees, even if both are managed and owned by
2267 the user. This effectively means "systemd-run --user --scope" doesn't
2268 work when invoked from outside of any "systemd --user" service or
2269 scope. Specifically, it is not supported from session scopes. We are
2270 working on fixing this in a future systemd version. (See #3388 for
2271 further details about this.)
2273 * DBus policy files are now installed into /usr rather than /etc. Make
2274 sure your system has dbus >= 1.9.18 running before upgrading to this
2275 version, or override the install path with --with-dbuspolicydir= .
2277 * All python scripts shipped with systemd (specifically: the various
2278 tests written in Python) now require Python 3.
2280 * systemd unit tests can now run standalone (without the source or
2281 build directories), and can be installed into /usr/lib/systemd/tests/
2282 with 'make install-tests'.
2284 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
2285 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
2288 * Support for the %c, %r, %R specifiers in unit files has been
2289 removed. Specifiers are not supposed to be dependent on configuration
2290 in the unit file itself (so that they resolve the same regardless
2291 where used in the unit files), but these specifiers were influenced
2292 by the Slice= option.
2294 * The shell invoked by debug-shell.service now defaults to /bin/sh in
2295 all cases. If distributions want to use a different shell for this
2296 purpose (for example Fedora's /sbin/sushell) they need to specify
2297 this explicitly at configure time using --with-debug-shell=.
2299 * The confirmation spawn prompt has been reworked to offer the
2302 (c)ontinue, proceed without asking anymore
2303 (D)ump, show the state of the unit
2304 (f)ail, don't execute the command and pretend it failed
2306 (i)nfo, show a short summary of the unit
2307 (j)obs, show jobs that are in progress
2308 (s)kip, don't execute the command and pretend it succeeded
2309 (y)es, execute the command
2311 The 'n' choice for the confirmation spawn prompt has been removed,
2312 because its meaning was confusing.
2314 The prompt may now also be redirected to an alternative console by
2315 specifying the console as parameter to systemd.confirm_spawn=.
2317 * Services of Type=notify require a READY=1 notification to be sent
2318 during startup. If no such message is sent, the service now fails,
2319 even if the main process exited with a successful exit code.
2321 * Services that fail to start up correctly now always have their
2322 ExecStopPost= commands executed. Previously, they'd enter "failed"
2323 state directly, without executing these commands.
2325 * The option MulticastDNS= of network configuration files has acquired
2326 an actual implementation. With MulticastDNS=yes a host can resolve
2327 names of remote hosts and reply to mDNS A and AAAA requests.
2329 * When units are about to be started an additional check is now done to
2330 ensure that all dependencies of type BindsTo= (when used in
2331 combination with After=) have been started.
2333 * systemd-analyze gained a new verb "syscall-filter" which shows which
2334 system call groups are defined for the SystemCallFilter= unit file
2335 setting, and which system calls they contain.
2337 * A new system call filter group "@filesystem" has been added,
2338 consisting of various file system related system calls. Group
2339 "@reboot" has been added, covering reboot, kexec and shutdown related
2340 calls. Finally, group "@swap" has been added covering swap
2341 configuration related calls.
2343 * A new unit file option RestrictNamespaces= has been added that may be
2344 used to restrict access to the various process namespace types the
2345 Linux kernel provides. Specifically, it may be used to take away the
2346 right for a service unit to create additional file system, network,
2347 user, and other namespaces. This sandboxing option is particularly
2348 relevant due to the high amount of recently discovered namespacing
2349 related vulnerabilities in the kernel.
2351 * systemd-udev's .link files gained support for a new AutoNegotiation=
2352 setting for configuring Ethernet auto-negotiation.
2354 * systemd-networkd's .network files gained support for a new
2355 ListenPort= setting in the [DHCP] section to explicitly configure the
2356 UDP client port the DHCP client shall listen on.
2358 * .network files gained a new Unmanaged= boolean setting for explicitly
2359 excluding one or more interfaces from management by systemd-networkd.
2361 * The systemd-networkd ProxyARP= option has been renamed to
2362 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
2363 renamed to ReduceARPProxy=. The old names continue to be available
2366 * systemd-networkd gained support for configuring IPv6 Proxy NDP
2367 addresses via the new IPv6ProxyNDPAddress= .network file setting.
2369 * systemd-networkd's bonding device support gained support for two new
2370 configuration options ActiveSlave= and PrimarySlave=.
2372 * The various options in the [Match] section of .network files gained
2373 support for negative matching.
2375 * New systemd-specific mount options are now understood in /etc/fstab:
2377 x-systemd.mount-timeout= may be used to configure the maximum
2378 permitted runtime of the mount command.
2380 x-systemd.device-bound may be set to bind a mount point to its
2381 backing device unit, in order to automatically remove a mount point
2382 if its backing device is unplugged. This option may also be
2383 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
2384 on the block device, which is now automatically set for all CDROM
2385 drives, so that mounted CDs are automatically unmounted when they are
2386 removed from the drive.
2388 x-systemd.after= and x-systemd.before= may be used to explicitly
2389 order a mount after or before another unit or mount point.
2391 * Enqueued start jobs for device units are now automatically garbage
2392 collected if there are no jobs waiting for them anymore.
2394 * systemctl list-jobs gained two new switches: with --after, for every
2395 queued job the jobs it's waiting for are shown; with --before the
2396 jobs which it's blocking are shown.
2398 * systemd-nspawn gained support for ephemeral boots from disk images
2399 (or in other words: --ephemeral and --image= may now be
2400 combined). Moreover, ephemeral boots are now supported for normal
2401 directories, even if the backing file system is not btrfs. Of course,
2402 if the file system does not support file system snapshots or
2403 reflinks, the initial copy operation will be relatively expensive, but
2404 this should still be suitable for many use cases.
2406 * Calendar time specifications in .timer units now support
2407 specifications relative to the end of a month by using "~" instead of
2408 "-" as separator between month and day. For example, "*-02~03" means
2409 "the third last day in February". In addition a new syntax for
2410 repeated events has been added using the "/" character. For example,
2411 "9..17/2:00" means "every two hours from 9am to 5pm".
2413 * systemd-socket-proxyd gained a new parameter --connections-max= for
2414 configuring the maximum number of concurrent connections.
2416 * sd-id128 gained a new API for generating unique IDs for the host in a
2417 way that does not leak the machine ID. Specifically,
2418 sd_id128_get_machine_app_specific() derives an ID based on the
2419 machine ID a in well-defined, non-reversible, stable way. This is
2420 useful whenever an identifier for the host is needed but where the
2421 identifier shall not be useful to identify the system beyond the
2422 scope of the application itself. (Internally this uses HMAC-SHA256 as
2423 keyed hash function using the machine ID as input.)
2425 * NotifyAccess= gained a new supported value "exec". When set
2426 notifications are accepted from all processes systemd itself invoked,
2427 including all control processes.
2429 * .nspawn files gained support for defining overlay mounts using the
2430 Overlay= and OverlayReadOnly= options. Previously this functionality
2431 was only available on the systemd-nspawn command line.
2433 * systemd-nspawn's --bind= and --overlay= options gained support for
2434 bind/overlay mounts whose source lies within the container tree by
2435 prefixing the source path with "+".
2437 * systemd-nspawn's --bind= and --overlay= options gained support for
2438 automatically allocating a temporary source directory in /var/tmp
2439 that is removed when the container dies. Specifically, if the source
2440 directory is specified as empty string this mechanism is selected. An
2441 example usage is --overlay=+/var::/var, which creates an overlay
2442 mount based on the original /var contained in the image, overlayed
2443 with a temporary directory in the host's /var/tmp. This way changes
2444 to /var are automatically flushed when the container shuts down.
2446 * systemd-nspawn --image= option does now permit raw file system block
2447 devices (in addition to images containing partition tables, as
2450 * The disk image dissection logic in systemd-nspawn gained support for
2451 automatically setting up LUKS encrypted as well as Verity protected
2452 partitions. When a container is booted from an encrypted image the
2453 passphrase is queried at start-up time. When a container with Verity
2454 data is started, the root hash is search in a ".roothash" file
2455 accompanying the disk image (alternatively, pass the root hash via
2456 the new --root-hash= command line option).
2458 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
2459 be used to dissect disk images the same way as systemd-nspawn does
2460 it, following the Bootable Partition Specification. It may even be
2461 used to mount disk images with complex partition setups (including
2462 LUKS and Verity partitions) to a local host directory, in order to
2463 inspect them. This tool is not considered public API (yet), and is
2464 thus not installed into /usr/bin. Please do not rely on its
2465 existence, since it might go away or be changed in later systemd
2468 * A new generator "systemd-verity-generator" has been added, similar in
2469 style to "systemd-cryptsetup-generator", permitting automatic setup of
2470 Verity root partitions when systemd boots up. In order to make use of
2471 this your partition setup should follow the Discoverable Partitions
2472 Specification, and the GPT partition ID of the root file system
2473 partition should be identical to the upper 128bit of the Verity root
2474 hash. The GPT partition ID of the Verity partition protecting it
2475 should be the lower 128bit of the Verity root hash. If the partition
2476 image follows this model it is sufficient to specify a single
2477 "roothash=" kernel command line argument to both configure which root
2478 image and verity partition to use as well as the root hash for
2479 it. Note that systemd-nspawn's Verity support follows the same
2480 semantics, meaning that disk images with proper Verity data in place
2481 may be booted in containers with systemd-nspawn as well as on
2482 physical systems via the verity generator. Also note that the "mkosi"
2483 tool available at https://github.com/systemd/mkosi has been updated
2484 to generate Verity protected disk images following this scheme. In
2485 fact, it has been updated to generate disk images that optionally
2486 implement a complete UEFI SecureBoot trust chain, involving a signed
2487 kernel and initrd image that incorporates such a root hash as well as
2488 a Verity-enabled root partition.
2490 * The hardware database (hwdb) udev supports has been updated to carry
2491 accelerometer quirks.
2493 * All system services are now run with a fresh kernel keyring set up
2494 for them. The invocation ID is stored by default in it, thus
2495 providing a safe, non-overridable way to determine the invocation
2498 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
2499 options for bind mounting arbitrary paths in a service-specific
2500 way. When these options are used, arbitrary host or service files and
2501 directories may be mounted to arbitrary locations in the service's
2504 * Documentation has been added that lists all of systemd's low-level
2505 environment variables:
2507 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
2509 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
2510 whether a specific socket file descriptor matches a specified socket
2513 * systemd-firstboot has been updated to check for the
2514 systemd.firstboot= kernel command line option. It accepts a boolean
2515 and when set to false the first boot questions are skipped.
2517 * systemd-fstab-generator has been updated to check for the
2518 systemd.volatile= kernel command line option, which either takes an
2519 optional boolean parameter or the special value "state". If used the
2520 system may be booted in a "volatile" boot mode. Specifically,
2521 "systemd.volatile" is used, the root directory will be mounted as
2522 tmpfs, and only /usr is mounted from the actual root file system. If
2523 "systemd.volatile=state" is used, the root directory will be mounted
2524 as usual, but /var is mounted as tmpfs. This concept provides similar
2525 functionality as systemd-nspawn's --volatile= option, but provides it
2526 on physical boots. Use this option for implementing stateless
2527 systems, or testing systems with all state and/or configuration reset
2528 to the defaults. (Note though that many distributions are not
2529 prepared to boot up without a populated /etc or /var, though.)
2531 * systemd-gpt-auto-generator gained support for LUKS encrypted root
2532 partitions. Previously it only supported LUKS encrypted partitions
2533 for all other uses, except for the root partition itself.
2535 * Socket units gained support for listening on AF_VSOCK sockets for
2536 communication in virtualized QEMU environments.
2538 * The "configure" script gained a new option --with-fallback-hostname=
2539 for specifying the fallback hostname to use if none is configured in
2540 /etc/hostname. For example, by specifying
2541 --with-fallback-hostname=fedora it is possible to default to a
2542 hostname of "fedora" on pristine installations.
2544 * systemd-cgls gained support for a new --unit= switch for listing only
2545 the control groups of a specific unit. Similar --user-unit= has been
2546 added for listing only the control groups of a specific user unit.
2548 * systemd-mount gained a new --umount switch for unmounting a mount or
2549 automount point (and all mount/automount points below it).
2551 * systemd will now refuse full configuration reloads (via systemctl
2552 daemon-reload and related calls) unless at least 16MiB of free space
2553 are available in /run. This is a safety precaution in order to ensure
2554 that generators can safely operate after the reload completed.
2556 * A new unit file option RootImage= has been added, which has a similar
2557 effect as RootDirectory= but mounts the service's root directory from
2558 a disk image instead of plain directory. This logic reuses the same
2559 image dissection and mount logic that systemd-nspawn already uses,
2560 and hence supports any disk images systemd-nspawn supports, including
2561 those following the Discoverable Partition Specification, as well as
2562 Verity enabled images. This option enables systemd to run system
2563 services directly off disk images acting as resource bundles,
2564 possibly even including full integrity data.
2566 * A new MountAPIVFS= unit file option has been added, taking a boolean
2567 argument. If enabled /proc, /sys and /dev (collectively called the
2568 "API VFS") will be mounted for the service. This is only relevant if
2569 RootDirectory= or RootImage= is used for the service, as these mounts
2570 are of course in place in the host mount namespace anyway.
2572 * systemd-nspawn gained support for a new --pivot-root= switch. If
2573 specified the root directory within the container image is pivoted to
2574 the specified mount point, while the original root disk is moved to a
2575 different place. This option enables booting of ostree images
2576 directly with systemd-nspawn.
2578 * The systemd build scripts will no longer complain if the NTP server
2579 addresses are not changed from the defaults. Google now supports
2580 these NTP servers officially. We still recommend downstreams to
2581 properly register an NTP pool with the NTP pool project though.
2583 * coredumpctl gained a new "--reverse" option for printing the list
2584 of coredumps in reverse order.
2586 * coredumpctl will now show additional information about truncated and
2587 inaccessible coredumps, as well as coredumps that are still being
2588 processed. It also gained a new --quiet switch for suppressing
2589 additional informational message in its output.
2591 * coredumpctl gained support for only showing coredumps newer and/or
2592 older than specific timestamps, using the new --since= and --until=
2593 options, reminiscent of journalctl's options by the same name.
2595 * The systemd-coredump logic has been improved so that it may be reused
2596 to collect backtraces in non-compiled languages, for example in
2597 scripting languages such as Python.
2599 * machinectl will now show the UID shift of local containers, if user
2600 namespacing is enabled for them.
2602 * systemd will now optionally run "environment generator" binaries at
2603 configuration load time. They may be used to add environment
2604 variables to the environment block passed to services invoked. One
2605 user environment generator is shipped by default that sets up
2606 environment variables based on files dropped into /etc/environment.d
2607 and ~/.config/environment.d/.
2609 * systemd-resolved now includes the new, recently published 2017 DNSSEC
2612 * hostnamed has been updated to report a new chassis type of
2613 "convertible" to cover "foldable" laptops that can both act as a
2614 tablet and as a laptop, such as various Lenovo Yoga devices.
2616 Contributions from: Adrián López, Alexander Galanin, Alexander
2617 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
2618 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
2619 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
2620 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
2621 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
2622 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
2623 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
2624 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
2625 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
2626 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
2627 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
2628 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
2629 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
2630 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
2631 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
2632 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
2633 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
2634 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
2635 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
2636 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
2637 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
2638 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
2639 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
2640 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
2641 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
2642 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
2643 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
2646 — Berlin, 2017-03-01
2650 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
2651 RestrictAddressFamilies= enabled. These sandboxing options should
2652 generally be compatible with the various external udev call-out
2653 binaries we are aware of, however there may be exceptions, in
2654 particular when exotic languages for these call-outs are used. In
2655 this case, consider turning off these settings locally.
2657 * The new RemoveIPC= option can be used to remove IPC objects owned by
2658 the user or group of a service when that service exits.
2660 * The new ProtectKernelModules= option can be used to disable explicit
2661 load and unload operations of kernel modules by a service. In
2662 addition access to /usr/lib/modules is removed if this option is set.
2664 * ProtectSystem= option gained a new value "strict", which causes the
2665 whole file system tree with the exception of /dev, /proc, and /sys,
2666 to be remounted read-only for a service.
2668 * The new ProtectKernelTunables= option can be used to disable
2669 modification of configuration files in /sys and /proc by a service.
2670 Various directories and files are remounted read-only, so access is
2671 restricted even if the file permissions would allow it.
2673 * The new ProtectControlGroups= option can be used to disable write
2674 access by a service to /sys/fs/cgroup.
2676 * Various systemd services have been hardened with
2677 ProtectKernelTunables=yes, ProtectControlGroups=yes,
2678 RestrictAddressFamilies=.
2680 * Support for dynamically creating users for the lifetime of a service
2681 has been added. If DynamicUser=yes is specified, user and group IDs
2682 will be allocated from the range 61184..65519 for the lifetime of the
2683 service. They can be resolved using the new nss-systemd.so NSS
2684 module. The module must be enabled in /etc/nsswitch.conf. Services
2685 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
2686 any resources allocated by the service will be cleaned up when the
2687 service exits. They also have ProtectHome=read-only and
2688 ProtectSystem=strict enabled, so they are not able to make any
2689 permanent modifications to the system.
2691 * The nss-systemd module also always resolves root and nobody, making
2692 it possible to have no /etc/passwd or /etc/group files in minimal
2693 container or chroot environments.
2695 * Services may be started with their own user namespace using the new
2696 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
2697 under which the service is running are mapped. All other users are
2700 * Support for the cgroup namespace has been added to systemd-nspawn. If
2701 supported by kernel, the container system started by systemd-nspawn
2702 will have its own view of the cgroup hierarchy. This new behaviour
2703 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
2705 * The new MemorySwapMax= option can be used to limit the maximum swap
2706 usage under the unified cgroup hierarchy.
2708 * Support for the CPU controller in the unified cgroup hierarchy has
2709 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
2710 options. This controller requires out-of-tree patches for the kernel
2711 and the support is provisional.
2713 * Mount and automount units may now be created transiently
2714 (i.e. dynamically at runtime via the bus API, instead of requiring
2715 unit files in the file system).
2717 * systemd-mount is a new tool which may mount file systems – much like
2718 mount(8), optionally pulling in additional dependencies through
2719 transient .mount and .automount units. For example, this tool
2720 automatically runs fsck on a backing block device before mounting,
2721 and allows the automount logic to be used dynamically from the
2722 command line for establishing mount points. This tool is particularly
2723 useful when dealing with removable media, as it will ensure fsck is
2724 run – if necessary – before the first access and that the file system
2725 is quickly unmounted after each access by utilizing the automount
2726 logic. This maximizes the chance that the file system on the
2727 removable media stays in a clean state, and if it isn't in a clean
2728 state is fixed automatically.
2730 * LazyUnmount=yes option for mount units has been added to expose the
2731 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
2734 * /efi will be used as the mount point of the EFI boot partition, if
2735 the directory is present, and the mount point was not configured
2736 through other means (e.g. fstab). If /efi directory does not exist,
2737 /boot will be used as before. This makes it easier to automatically
2738 mount the EFI partition on systems where /boot is used for something
2741 * When operating on GPT disk images for containers, systemd-nspawn will
2742 now mount the ESP to /boot or /efi according to the same rules as PID
2743 1 running on a host. This allows tools like "bootctl" to operate
2744 correctly within such containers, in order to make container images
2745 bootable on physical systems.
2747 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
2749 * Two new user session targets have been added to support running
2750 graphical sessions under the systemd --user instance:
2751 graphical-session.target and graphical-session-pre.target. See
2752 systemd.special(7) for a description of how those targets should be
2755 * The vconsole initialization code has been significantly reworked to
2756 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
2757 support unicode keymaps. Font and keymap configuration will now be
2758 copied to all allocated virtual consoles.
2760 * FreeBSD's bhyve virtualization is now detected.
2762 * Information recorded in the journal for core dumps now includes the
2763 contents of /proc/mountinfo and the command line of the process at
2764 the top of the process hierarchy (which is usually the init process
2767 * systemd-journal-gatewayd learned the --directory= option to serve
2768 files from the specified location.
2770 * journalctl --root=… can be used to peruse the journal in the
2771 /var/log/ directories inside of a container tree. This is similar to
2772 the existing --machine= option, but does not require the container to
2775 * The hardware database has been extended to support
2776 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
2779 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
2780 specify the click rate for mice which include a horizontal wheel with
2781 a click rate that is different than the one for the vertical wheel.
2783 * systemd-run gained a new --wait option that makes service execution
2784 synchronous. (Specifically, the command will not return until the
2785 specified service binary exited.)
2787 * systemctl gained a new --wait option that causes the start command to
2788 wait until the units being started have terminated again.
2790 * A new journal output mode "short-full" has been added which displays
2791 timestamps with abbreviated English day names and adds a timezone
2792 suffix. Those timestamps include more information than the default
2793 "short" output mode, and can be passed directly to journalctl's
2794 --since= and --until= options.
2796 * /etc/resolv.conf will be bind-mounted into containers started by
2797 systemd-nspawn, if possible, so any changes to resolv.conf contents
2798 are automatically propagated to the container.
2800 * The number of instances for socket-activated services originating
2801 from a single IP address can be limited with
2802 MaxConnectionsPerSource=, extending the existing setting of
2805 * systemd-networkd gained support for vcan ("Virtual CAN") interface
2808 * .netdev and .network configuration can now be extended through
2811 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
2812 Segmentation Offload, Generic Receive Offload, Large Receive Offload
2813 can be enabled and disabled using the new UDPSegmentationOffload=,
2814 TCPSegmentationOffload=, GenericSegmentationOffload=,
2815 GenericReceiveOffload=, LargeReceiveOffload= options in the
2816 [Link] section of .link files.
2818 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
2819 Port VLAN ID can be configured for bridge devices using the new STP=,
2820 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
2821 section of .netdev files.
2823 * The route table to which routes received over DHCP or RA should be
2824 added can be configured with the new RouteTable= option in the [DHCP]
2825 and [IPv6AcceptRA] sections of .network files.
2827 * The Address Resolution Protocol can be disabled on links managed by
2828 systemd-networkd using the ARP=no setting in the [Link] section of
2831 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
2832 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
2833 encode information about the result and exit codes of the current
2834 service runtime cycle.
2836 * systemd-sysctl will now configure kernel parameters in the order
2837 they occur in the configuration files. This matches what sysctl
2838 has been traditionally doing.
2840 * kernel-install "plugins" that are executed to perform various
2841 tasks after a new kernel is added and before an old one is removed
2842 can now return a special value to terminate the procedure and
2843 prevent any later plugins from running.
2845 * Journald's SplitMode=login setting has been deprecated. It has been
2846 removed from documentation, and its use is discouraged. In a future
2847 release it will be completely removed, and made equivalent to current
2848 default of SplitMode=uid.
2850 * Storage=both option setting in /etc/systemd/coredump.conf has been
2851 removed. With fast LZ4 compression storing the core dump twice is not
2854 * The --share-system systemd-nspawn option has been replaced with an
2855 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
2856 this functionality is discouraged. In addition the variables
2857 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
2858 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
2859 individual namespaces.
2861 * "machinectl list" now shows the IP address of running containers in
2862 the output, as well as OS release information.
2864 * "loginctl list" now shows the TTY of each session in the output.
2866 * sd-bus gained new API calls sd_bus_track_set_recursive(),
2867 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
2868 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
2869 tracking objects in a "recursive" mode, where a single client can be
2870 counted multiple times, if it takes multiple references.
2872 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
2873 sd_bus_get_exit_on_disconnect(). They may be used to make a
2874 process using sd-bus automatically exit if the bus connection is
2877 * Bus clients of the service manager may now "pin" loaded units into
2878 memory, by taking an explicit reference on them. This is useful to
2879 ensure the client can retrieve runtime data about the service even
2880 after the service completed execution. Taking such a reference is
2881 available only for privileged clients and should be helpful to watch
2882 running services in a race-free manner, and in particular collect
2883 information about exit statuses and results.
2885 * The nss-resolve module has been changed to strictly return UNAVAIL
2886 when communication via D-Bus with resolved failed, and NOTFOUND when
2887 a lookup completed but was negative. This means it is now possible to
2888 neatly configure fallbacks using nsswitch.conf result checking
2889 expressions. Taking benefit of this, the new recommended
2890 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
2892 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
2894 * A new setting CtrlAltDelBurstAction= has been added to
2895 /etc/systemd/system.conf which may be used to configure the precise
2896 behaviour if the user on the console presses Ctrl-Alt-Del more often
2897 than 7 times in 2s. Previously this would unconditionally result in
2898 an expedited, immediate reboot. With this new setting the precise
2899 operation may be configured in more detail, and also turned off
2902 * In .netdev files two new settings RemoteChecksumTx= and
2903 RemoteChecksumRx= are now understood that permit configuring the
2904 remote checksumming logic for VXLAN networks.
2906 * The service manager learnt a new "invocation ID" concept for invoked
2907 services. Each runtime cycle of a service will get a new invocation
2908 ID (a 128bit random UUID) assigned that identifies the current
2909 run of the service uniquely and globally. A new invocation ID
2910 is generated each time a service starts up. The journal will store
2911 the invocation ID of a service along with any logged messages, thus
2912 making the invocation ID useful for matching the online runtime of a
2913 service with the offline log data it generated in a safe way without
2914 relying on synchronized timestamps. In many ways this new service
2915 invocation ID concept is similar to the kernel's boot ID concept that
2916 uniquely and globally identifies the runtime of each boot. The
2917 invocation ID of a service is passed to the service itself via an
2918 environment variable ($INVOCATION_ID). A new bus call
2919 GetUnitByInvocationID() has been added that is similar to GetUnit()
2920 but instead of retrieving the bus path for a unit by its name
2921 retrieves it by its invocation ID. The returned path is valid only as
2922 long as the passed invocation ID is current.
2924 * systemd-resolved gained a new "DNSStubListener" setting in
2925 resolved.conf. It either takes a boolean value or the special values
2926 "udp" and "tcp", and configures whether to enable the stub DNS
2927 listener on 127.0.0.53:53.
2929 * IP addresses configured via networkd may now carry additional
2930 configuration settings supported by the kernel. New options include:
2931 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
2932 PrefixRoute=, AutoJoin=.
2934 * The PAM configuration fragment file for "user@.service" shipped with
2935 systemd (i.e. the --user instance of systemd) has been stripped to
2936 the minimum necessary to make the system boot. Previously, it
2937 contained Fedora-specific stanzas that did not apply to other
2938 distributions. It is expected that downstream distributions add
2939 additional configuration lines, matching their needs to this file,
2940 using it only as rough template of what systemd itself needs. Note
2941 that this reduced fragment does not even include an invocation of
2942 pam_limits which most distributions probably want to add, even though
2943 systemd itself does not need it. (There's also the new build time
2944 option --with-pamconfdir=no to disable installation of the PAM
2947 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
2948 capability is now also dropped from its set (in addition to
2949 CAP_SYS_MKNOD as before).
2951 * In service unit files it is now possible to connect a specific named
2952 file descriptor with stdin/stdout/stdout of an executed service. The
2953 name may be specified in matching .socket units using the
2954 FileDescriptorName= setting.
2956 * A number of journal settings may now be configured on the kernel
2957 command line. Specifically, the following options are now understood:
2958 systemd.journald.max_level_console=,
2959 systemd.journald.max_level_store=,
2960 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
2961 systemd.journald.max_level_wall=.
2963 * "systemctl is-enabled --full" will now show by which symlinks a unit
2964 file is enabled in the unit dependency tree.
2966 * Support for VeraCrypt encrypted partitions has been added to the
2967 "cryptsetup" logic and /etc/crypttab.
2969 * systemd-detect-virt gained support for a new --private-users switch
2970 that checks whether the invoking processes are running inside a user
2971 namespace. Similar, a new special value "private-users" for the
2972 existing ConditionVirtualization= setting has been added, permitting
2973 skipping of specific units in user namespace environments.
2975 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
2976 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
2977 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
2978 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
2979 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
2980 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
2981 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
2982 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
2983 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
2984 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
2985 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
2986 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
2987 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
2988 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
2989 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
2990 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
2991 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
2992 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
2993 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
2994 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
2995 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
2996 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
2997 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
2998 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
2999 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
3000 Jędrzejewski-Szmek, Zeal Jagannatha
3002 — Santa Fe, 2016-11-03
3006 * In service units the various ExecXYZ= settings have been extended
3007 with an additional special character as first argument of the
3008 assigned value: if the character '+' is used the specified command
3009 line it will be run with full privileges, regardless of User=,
3010 Group=, CapabilityBoundingSet= and similar options. The effect is
3011 similar to the existing PermissionsStartOnly= option, but allows
3012 configuration of this concept for each executed command line
3015 * Services may now alter the service watchdog timeout at runtime by
3016 sending a WATCHDOG_USEC= message via sd_notify().
3018 * MemoryLimit= and related unit settings now optionally take percentage
3019 specifications. The percentage is taken relative to the amount of
3020 physical memory in the system (or in case of containers, the assigned
3021 amount of memory). This allows scaling service resources neatly with
3022 the amount of RAM available on the system. Similarly, systemd-logind's
3023 RuntimeDirectorySize= option now also optionally takes percentage
3026 * In similar fashion TasksMax= takes percentage values now, too. The
3027 value is taken relative to the configured maximum number of processes
3028 on the system. The per-service task maximum has been changed to 15%
3029 using this functionality. (Effectively this is an increase of 512 →
3030 4915 for service units, given the kernel's default pid_max setting.)
3032 * Calendar time specifications in .timer units now understand a ".."
3033 syntax for time ranges. Example: "4..7:10" may now be used for
3034 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
3037 * The InaccessableDirectories=, ReadOnlyDirectories= and
3038 ReadWriteDirectories= unit file settings have been renamed to
3039 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
3040 applied to all kinds of file nodes, and not just directories, with
3041 the exception of symlinks. Specifically these settings may now be
3042 used on block and character device nodes, UNIX sockets and FIFOS as
3043 well as regular files. The old names of these settings remain
3044 available for compatibility.
3046 * systemd will now log about all service processes it kills forcibly
3047 (using SIGKILL) because they remained after the clean shutdown phase
3048 of the service completed. This should help identifying services that
3049 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
3050 systemd-logind's configuration a similar log message is generated for
3051 processes killed at the end of each session due to this setting.
3053 * systemd will now set the $JOURNAL_STREAM environment variable for all
3054 services whose stdout/stderr are connected to the Journal (which
3055 effectively means by default: all services). The variable contains
3056 the device and inode number of the file descriptor used for
3057 stdout/stderr. This may be used by invoked programs to detect whether
3058 their stdout/stderr is connected to the Journal, in which case they
3059 can switch over to direct Journal communication, thus being able to
3060 pass extended, structured metadata along with their log messages. As
3061 one example, this is now used by glib's logging primitives.
3063 * When using systemd's default tmp.mount unit for /tmp, the mount point
3064 will now be established with the "nosuid" and "nodev" options. This
3065 avoids privilege escalation attacks that put traps and exploits into
3066 /tmp. However, this might cause problems if you e. g. put container
3067 images or overlays into /tmp; if you need this, override tmp.mount's
3068 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
3071 * systemd now supports the "memory" cgroup controller also on
3074 * The systemd-cgtop tool now optionally takes a control group path as
3075 command line argument. If specified, the control group list shown is
3076 limited to subgroups of that group.
3078 * The SystemCallFilter= unit file setting gained support for
3079 pre-defined, named system call filter sets. For example
3080 SystemCallFilter=@clock is now an effective way to make all clock
3081 changing-related system calls unavailable to a service. A number of
3082 similar pre-defined groups are defined. Writing system call filters
3083 for system services is simplified substantially with this new
3084 concept. Accordingly, all of systemd's own, long-running services now
3085 enable system call filtering based on this, by default.
3087 * A new service setting MemoryDenyWriteExecute= has been added, taking
3088 a boolean value. If turned on, a service may no longer create memory
3089 mappings that are writable and executable at the same time. This
3090 enhances security for services where this is enabled as it becomes
3091 harder to dynamically write and then execute memory in exploited
3092 service processes. This option has been enabled for all of systemd's
3093 own long-running services.
3095 * A new RestrictRealtime= service setting has been added, taking a
3096 boolean argument. If set the service's processes may no longer
3097 acquire realtime scheduling. This improves security as realtime
3098 scheduling may otherwise be used to easily freeze the system.
3100 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
3101 value. This may be used for requesting that the system manager inside
3102 of the container reports start-up completion to nspawn which then
3103 propagates this notification further to the service manager
3104 supervising nspawn itself. A related option NotifyReady= in .nspawn
3105 files has been added too. This functionality allows ordering of the
3106 start-up of multiple containers using the usual systemd ordering
3109 * machinectl gained a new command "stop" that is an alias for
3112 * systemd-resolved gained support for contacting DNS servers on
3113 link-local IPv6 addresses.
3115 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
3116 its caches. A method call for requesting the same operation has been
3117 added to the bus API too, and is made available via "systemd-resolve
3120 * systemd-resolve gained a new --status switch. If passed a brief
3121 summary of the used DNS configuration with per-interface information
3124 * resolved.conf gained a new Cache= boolean option, defaulting to
3125 on. If turned off local DNS caching is disabled. This comes with a
3126 performance penalty in particular when DNSSEC is enabled. Note that
3127 resolved disables its internal caching implicitly anyway, when the
3128 configured DNS server is on a host-local IP address such as ::1 or
3129 127.0.0.1, thus automatically avoiding double local caching.
3131 * systemd-resolved now listens on the local IP address 127.0.0.53:53
3132 for DNS requests. This improves compatibility with local programs
3133 that do not use the libc NSS or systemd-resolved's bus APIs for name
3134 resolution. This minimal DNS service is only available to local
3135 programs and does not implement the full DNS protocol, but enough to
3136 cover local DNS clients. A new, static resolv.conf file, listing just
3137 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
3138 now recommended to make /etc/resolv.conf a symlink to this file in
3139 order to route all DNS lookups to systemd-resolved, regardless if
3140 done via NSS, the bus API or raw DNS packets. Note that this local
3141 DNS service is not as fully featured as the libc NSS or
3142 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
3143 used to deliver link-local address information (as this implies
3144 sending a local interface index along), LLMNR/mDNS support via this
3145 interface is severely restricted. It is thus strongly recommended for
3146 all applications to use the libc NSS API or native systemd-resolved
3149 * systemd-networkd's bridge support learned a new setting
3150 VLANFiltering= for controlling VLAN filtering. Moreover a new section
3151 in .network files has been added for configuring VLAN bridging in
3152 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
3154 * systemd-networkd's IPv6 Router Advertisement code now makes use of
3155 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
3156 now be acquired without relying on DHCPv6. Two new options
3157 UseDomains= and UseDNS= have been added to configure this behaviour.
3159 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
3160 renamed IPv6AcceptRA=, without altering its behaviour. The old
3161 setting name remains available for compatibility reasons.
3163 * The systemd-networkd VTI/VTI6 tunneling support gained new options
3164 Key=, InputKey= and OutputKey=.
3166 * systemd-networkd gained support for VRF ("Virtual Routing Function")
3167 interface configuration.
3169 * "systemctl edit" may now be used to create new unit files by
3170 specifying the --force switch.
3172 * sd-event gained a new function sd_event_get_iteration() for
3173 requesting the current iteration counter of the event loop. It starts
3174 at zero and is increased by one with each event loop iteration.
3176 * A new rpm macro %systemd_ordering is provided by the macros.systemd
3177 file. It can be used in lieu of %systemd_requires in packages which
3178 don't use any systemd functionality and are intended to be installed
3179 in minimal containers without systemd present. This macro provides
3180 ordering dependencies to ensure that if the package is installed in
3181 the same rpm transaction as systemd, systemd will be installed before
3182 the scriptlets for the package are executed, allowing unit presets
3185 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
3186 been added to simplify packaging of generators.
3188 * The os-release file gained VERSION_CODENAME field for the
3189 distribution nickname (e.g. VERSION_CODENAME=woody).
3191 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
3192 can be set to disable parsing of metadata and the creation
3193 of persistent symlinks for that device.
3195 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
3196 to make them available to logged-in users has been reverted.
3198 * Much of the common code of the various systemd components is now
3199 built into an internal shared library libsystemd-shared-231.so
3200 (incorporating the systemd version number in the name, to be updated
3201 with future releases) that the components link to. This should
3202 decrease systemd footprint both in memory during runtime and on
3203 disk. Note that the shared library is not for public use, and is
3204 neither API nor ABI stable, but is likely to change with every new
3205 released update. Packagers need to make sure that binaries
3206 linking to libsystemd-shared.so are updated in step with the
3209 * Configuration for "mkosi" is now part of the systemd
3210 repository. mkosi is a tool to easily build legacy-free OS images,
3211 and is available on github: https://github.com/systemd/mkosi. If
3212 "mkosi" is invoked in the build tree a new raw OS image is generated
3213 incorporating the systemd sources currently being worked on and a
3214 clean, fresh distribution installation. The generated OS image may be
3215 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
3216 UEFI PC. This functionality is particularly useful to easily test
3217 local changes made to systemd in a pristine, defined environment. See
3218 doc/HACKING for details.
3220 * configure learned the --with-support-url= option to specify the
3221 distribution's bugtracker.
3223 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
3224 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
3225 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
3226 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
3227 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
3228 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
3229 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
3230 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
3231 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
3232 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
3233 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
3234 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
3235 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
3236 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
3237 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
3238 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
3239 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
3240 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
3241 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
3243 — Berlin, 2016-07-25
3247 * DNSSEC is now turned on by default in systemd-resolved (in
3248 "allow-downgrade" mode), but may be turned off during compile time by
3249 passing "--with-default-dnssec=no" to "configure" (and of course,
3250 during runtime with DNSSEC= in resolved.conf). We recommend
3251 downstreams to leave this on at least during development cycles and
3252 report any issues with the DNSSEC logic upstream. We are very
3253 interested in collecting feedback about the DNSSEC validator and its
3254 limitations in the wild. Note however, that DNSSEC support is
3255 probably nothing downstreams should turn on in stable distros just
3256 yet, as it might create incompatibilities with a few DNS servers and
3257 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
3258 automatically whenever we detect such incompatible setups, but there
3259 might be systems we do not cover yet. Hence: please help us testing
3260 the DNSSEC code, leave this on where you can, report back, but then
3261 again don't consider turning this on in your stable, LTS or
3262 production release just yet. (Note that you have to enable
3263 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
3264 and its DNSSEC mode for host name resolution from local
3267 * systemd-resolve conveniently resolves DANE records with the --tlsa
3268 option and OPENPGPKEY records with the --openpgp option. It also
3269 supports dumping raw DNS record data via the new --raw= switch.
3271 * systemd-logind will now by default terminate user processes that are
3272 part of the user session scope unit (session-XX.scope) when the user
3273 logs out. This behavior is controlled by the KillUserProcesses=
3274 setting in logind.conf, and the previous default of "no" is now
3275 changed to "yes". This means that user sessions will be properly
3276 cleaned up after, but additional steps are necessary to allow
3277 intentionally long-running processes to survive logout.
3279 While the user is logged in at least once, user@.service is running,
3280 and any service that should survive the end of any individual login
3281 session can be started at a user service or scope using systemd-run.
3282 systemd-run(1) man page has been extended with an example which shows
3283 how to run screen in a scope unit underneath user@.service. The same
3284 command works for tmux.
3286 After the user logs out of all sessions, user@.service will be
3287 terminated too, by default, unless the user has "lingering" enabled.
3288 To effectively allow users to run long-term tasks even if they are
3289 logged out, lingering must be enabled for them. See loginctl(1) for
3290 details. The default polkit policy was modified to allow users to
3291 set lingering for themselves without authentication.
3293 Previous defaults can be restored at compile time by the
3294 --without-kill-user-processes option to "configure".
3296 * systemd-logind gained new configuration settings SessionsMax= and
3297 InhibitorsMax=, both with a default of 8192. It will not register new
3298 user sessions or inhibitors above this limit.
3300 * systemd-logind will now reload configuration on SIGHUP.
3302 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
3303 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
3304 enable. Also, support for the "io" cgroup controller in the unified
3305 hierarchy has been added, so that the "memory", "pids" and "io" are
3306 now the controllers that are supported on the unified hierarchy.
3308 WARNING: it is not possible to use previous systemd versions with
3309 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
3310 is necessary to also update systemd in the initramfs if using the
3311 unified hierarchy. An updated SELinux policy is also required.
3313 * LLDP support has been extended, and both passive (receive-only) and
3314 active (sender) modes are supported. Passive mode ("routers-only") is
3315 enabled by default in systemd-networkd. Active LLDP mode is enabled
3316 by default for containers on the internal network. The "networkctl
3317 lldp" command may be used to list information gathered. "networkctl
3318 status" will also show basic LLDP information on connected peers now.
3320 * The IAID and DUID unique identifier sent in DHCP requests may now be
3321 configured for the system and each .network file managed by
3322 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
3324 * systemd-networkd gained support for configuring proxy ARP support for
3325 each interface, via the ProxyArp= setting in .network files. It also
3326 gained support for configuring the multicast querier feature of
3327 bridge devices, via the new MulticastQuerier= setting in .netdev
3328 files. Similarly, snooping on the IGMP traffic can be controlled
3329 via the new setting MulticastSnooping=.
3331 A new setting PreferredLifetime= has been added for addresses
3332 configured in .network file to configure the lifetime intended for an
3335 The systemd-networkd DHCP server gained the option EmitRouter=, which
3336 defaults to yes, to configure whether the DHCP Option 3 (Router)
3339 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
3340 systemd-socket-activate and installed into /usr/bin. It is now fully
3343 * systemd-journald now uses separate threads to flush changes to disk
3344 when closing journal files, thus reducing impact of slow disk I/O on
3345 logging performance.
3347 * The sd-journal API gained two new calls
3348 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
3349 can be used to open journal files using file descriptors instead of
3350 file or directory paths. sd_journal_open_container() has been
3351 deprecated, sd_journal_open_directory_fd() should be used instead
3352 with the flag SD_JOURNAL_OS_ROOT.
3354 * journalctl learned a new output mode "-o short-unix" that outputs log
3355 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
3356 UTC). It also gained support for a new --no-hostname setting to
3357 suppress the hostname column in the family of "short" output modes.
3359 * systemd-ask-password now optionally skips printing of the password to
3360 stdout with --no-output which can be useful in scripts.
3362 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
3363 (devices tagged with ID_MAKER_TOOL) are now tagged with
3364 "uaccess" and are available to logged in users.
3366 * The DeviceAllow= unit setting now supports specifiers (with "%").
3368 * "systemctl show" gained a new --value switch, which allows print a
3369 only the contents of a specific unit property, without also printing
3370 the property's name. Similar support was added to "show*" verbs
3371 of loginctl and machinectl that output "key=value" lists.
3373 * A new unit type "generated" was added for files dynamically generated
3374 by generator tools. Similarly, a new unit type "transient" is used
3375 for unit files created using the runtime API. "systemctl enable" will
3376 refuse to operate on such files.
3378 * A new command "systemctl revert" has been added that may be used to
3379 revert to the vendor version of a unit file, in case local changes
3380 have been made by adding drop-ins or overriding the unit file.
3382 * "machinectl clean" gained a new verb to automatically remove all or
3383 just hidden container images.
3385 * systemd-tmpfiles gained support for a new line type "e" for emptying
3386 directories, if they exist, without creating them if they don't.
3388 * systemd-nspawn gained support for automatically patching the UID/GIDs
3389 of the owners and the ACLs of all files and directories in a
3390 container tree to match the UID/GID user namespacing range selected
3391 for the container invocation. This mode is enabled via the new
3392 --private-users-chown switch. It also gained support for
3393 automatically choosing a free, previously unused UID/GID range when
3394 starting a container, via the new --private-users=pick setting (which
3395 implies --private-users-chown). Together, these options for the first
3396 time make user namespacing for nspawn containers fully automatic and
3397 thus deployable. The systemd-nspawn@.service template unit file has
3398 been changed to use this functionality by default.
3400 * systemd-nspawn gained a new --network-zone= switch, that allows
3401 creating ad-hoc virtual Ethernet links between multiple containers,
3402 that only exist as long as at least one container referencing them is
3403 running. This allows easy connecting of multiple containers with a
3404 common link that implements an Ethernet broadcast domain. Each of
3405 these network "zones" may be named relatively freely by the user, and
3406 may be referenced by any number of containers, but each container may
3407 only reference one of these "zones". On the lower level, this is
3408 implemented by an automatically managed bridge network interface for
3409 each zone, that is created when the first container referencing its
3410 zone is created and removed when the last one referencing its zone
3413 * The default start timeout may now be configured on the kernel command
3414 line via systemd.default_timeout_start_sec=. It was already
3415 configurable via the DefaultTimeoutStartSec= option in
3416 /etc/systemd/system.conf.
3418 * Socket units gained a new TriggerLimitIntervalSec= and
3419 TriggerLimitBurst= setting to configure a limit on the activation
3420 rate of the socket unit.
3422 * The LimitNICE= setting now optionally takes normal UNIX nice values
3423 in addition to the raw integer limit value. If the specified
3424 parameter is prefixed with "+" or "-" and is in the range -20..19 the
3425 value is understood as UNIX nice value. If not prefixed like this it
3426 is understood as raw RLIMIT_NICE limit.
3428 * Note that the effect of the PrivateDevices= unit file setting changed
3429 slightly with this release: the per-device /dev file system will be
3430 mounted read-only from this version on, and will have "noexec"
3431 set. This (minor) change of behavior might cause some (exceptional)
3432 legacy software to break, when PrivateDevices=yes is set for its
3433 service. Please leave PrivateDevices= off if you run into problems
3436 * systemd-bootchart has been split out to a separate repository:
3437 https://github.com/systemd/systemd-bootchart
3439 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
3440 merged into the kernel in its current form.
3442 * The compatibility libraries libsystemd-daemon.so,
3443 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
3444 which have been deprecated since systemd-209 have been removed along
3445 with the corresponding pkg-config files. All symbols provided by
3446 those libraries are provided by libsystemd.so.
3448 * The Capabilities= unit file setting has been removed (it is ignored
3449 for backwards compatibility). AmbientCapabilities= and
3450 CapabilityBoundingSet= should be used instead.
3452 * A new special target has been added, initrd-root-device.target,
3453 which creates a synchronization point for dependencies of the root
3454 device in early userspace. Initramfs builders must ensure that this
3455 target is now included in early userspace.
3457 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
3458 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
3459 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
3460 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
3461 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
3462 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
3463 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
3464 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
3465 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
3466 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
3467 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
3468 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
3469 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
3470 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
3471 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
3472 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
3473 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
3474 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
3475 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
3476 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
3477 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
3478 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
3479 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
3480 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
3481 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
3484 — Fairfax, 2016-05-21
3488 * The systemd-resolved DNS resolver service has gained a substantial
3489 set of new features, most prominently it may now act as a DNSSEC
3490 validating stub resolver. DNSSEC mode is currently turned off by
3491 default, but is expected to be turned on by default in one of the
3492 next releases. For now, we invite everybody to test the DNSSEC logic
3493 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
3494 service also gained a full set of D-Bus interfaces, including calls
3495 to configure DNS and DNSSEC settings per link (for use by external
3496 network management software). systemd-resolved and systemd-networkd
3497 now distinguish between "search" and "routing" domains. The former
3498 are used to qualify single-label names, the latter are used purely
3499 for routing lookups within certain domains to specific links.
3500 resolved now also synthesizes RRs for all entries from /etc/hosts.
3502 * The systemd-resolve tool (which is a client utility for
3503 systemd-resolved) has been improved considerably and is now fully
3504 supported and documented. Hence it has moved from /usr/lib/systemd to
3507 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
3510 * The coredump collection logic has been reworked: when a coredump is
3511 collected it is now written to disk, compressed and processed
3512 (including stacktrace extraction) from a new instantiated service
3513 systemd-coredump@.service, instead of directly from the
3514 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
3515 processing large coredumps can take up a substantial amount of
3516 resources and time, and this previously happened entirely outside of
3517 systemd's service supervision. With the new logic the core_pattern
3518 hook only does minimal metadata collection before passing off control
3519 to the new instantiated service, which is configured with a time
3520 limit, a nice level and other settings to minimize negative impact on
3521 the rest of the system. Also note that the new logic will honour the
3522 RLIMIT_CORE setting of the crashed process, which now allows users
3523 and processes to turn off coredumping for their processes by setting
3526 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
3527 and all forked processes by default. Previously, PID 1 would leave
3528 the setting at "0" for all processes, as set by the kernel. Note that
3529 the resource limit traditionally has no effect on the generated
3530 coredumps on the system if the /proc/sys/kernel/core_pattern hook
3531 logic is used. Since the limit is now honoured (see above) its
3532 default has been changed so that the coredumping logic is enabled by
3533 default for all processes, while allowing specific opt-out.
3535 * When the stacktrace is extracted from processes of system users, this
3536 is now done as "systemd-coredump" user, in order to sandbox this
3537 potentially security sensitive parsing operation. (Note that when
3538 processing coredumps of normal users this is done under the user ID
3539 of process that crashed, as before.) Packagers should take notice
3540 that it is now necessary to create the "systemd-coredump" system user
3541 and group at package installation time.
3543 * The systemd-activate socket activation testing tool gained support
3544 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
3545 and --seqpacket switches. It also has been extended to support both
3546 new-style and inetd-style file descriptor passing. Use the new
3547 --inetd switch to request inetd-style file descriptor passing.
3549 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
3550 variable, which takes a boolean value. If set to false, ANSI color
3551 output is disabled in the tools even when run on a terminal that
3554 * The VXLAN support in networkd now supports two new settings
3555 DestinationPort= and PortRange=.
3557 * A new systemd.machine_id= kernel command line switch has been added,
3558 that may be used to set the machine ID in /etc/machine-id if it is
3559 not initialized yet. This command line option has no effect if the
3560 file is already initialized.
3562 * systemd-nspawn gained a new --as-pid2 switch that invokes any
3563 specified command line as PID 2 rather than PID 1 in the
3564 container. In this mode PID 1 is a minimal stub init process that
3565 implements the special POSIX and Linux semantics of PID 1 regarding
3566 signal and child process management. Note that this stub init process
3567 is implemented in nspawn itself and requires no support from the
3568 container image. This new logic is useful to support running
3569 arbitrary commands in the container, as normal processes are
3570 generally not prepared to run as PID 1.
3572 * systemd-nspawn gained a new --chdir= switch for setting the current
3573 working directory for the process started in the container.
3575 * "journalctl /dev/sda" will now output all kernel log messages for
3576 specified device from the current boot, in addition to all devices
3577 that are parents of it. This should make log output about devices
3578 pretty useful, as long as kernel drivers attach enough metadata to
3579 the log messages. (The usual SATA drivers do.)
3581 * The sd-journal API gained two new calls
3582 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
3583 that report whether log data from /run or /var has been found.
3585 * journalctl gained a new switch "--fields" that prints all journal
3586 record field names currently in use in the journal. This is backed
3587 by two new sd-journal API calls sd_journal_enumerate_fields() and
3588 sd_journal_restart_fields().
3590 * Most configurable timeouts in systemd now expect an argument of
3591 "infinity" to turn them off, instead of "0" as before. The semantics
3592 from now on is that a timeout of "0" means "now", and "infinity"
3593 means "never". To maintain backwards compatibility, "0" continues to
3594 turn off previously existing timeout settings.
3596 * "systemctl reload-or-try-restart" has been renamed to "systemctl
3597 try-reload-or-restart" to clarify what it actually does: the "try"
3598 logic applies to both reloading and restarting, not just restarting.
3599 The old name continues to be accepted for compatibility.
3601 * On boot-up, when PID 1 detects that the system clock is behind the
3602 release date of the systemd version in use, the clock is now set
3603 to the latter. Previously, this was already done in timesyncd, in order
3604 to avoid running with clocks set to the various clock epochs such as
3605 1902, 1938 or 1970. With this change the logic is now done in PID 1
3606 in addition to timesyncd during early boot-up, so that it is enforced
3607 before the first process is spawned by systemd. Note that the logic
3608 in timesyncd remains, as it is more comprehensive and ensures
3609 clock monotonicity by maintaining a persistent timestamp file in
3610 /var. Since /var is generally not available in earliest boot or the
3611 initrd, this part of the logic remains in timesyncd, and is not done
3614 * Support for tweaking details in net_cls.class_id through the
3615 NetClass= configuration directive has been removed, as the kernel
3616 people have decided to deprecate that controller in cgroup v2.
3617 Userspace tools such as nftables are moving over to setting rules
3618 that are specific to the full cgroup path of a task, which obsoletes
3619 these controllers anyway. The NetClass= directive is kept around for
3620 legacy compatibility reasons. For a more in-depth description of the
3621 kernel change, please refer to the respective upstream commit:
3623 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
3625 * A new service setting RuntimeMaxSec= has been added that may be used
3626 to specify a maximum runtime for a service. If the timeout is hit, the
3627 service is terminated and put into a failure state.
3629 * A new service setting AmbientCapabilities= has been added. It allows
3630 configuration of additional Linux process capabilities that are
3631 passed to the activated processes. This is only available on very
3634 * The process resource limit settings in service units may now be used
3635 to configure hard and soft limits individually.
3637 * The various libsystemd APIs such as sd-bus or sd-event now publicly
3638 expose support for gcc's __attribute__((cleanup())) C extension.
3639 Specifically, for many object destructor functions alternative
3640 versions have been added that have names suffixed with "p" and take a
3641 pointer to a pointer to the object to destroy, instead of just a
3642 pointer to the object itself. This is useful because these destructor
3643 functions may be used directly as parameters to the cleanup
3644 construct. Internally, systemd has been a heavy user of this GCC
3645 extension for a long time, and with this change similar support is
3646 now available to consumers of the library outside of systemd. Note
3647 that by using this extension in your sources compatibility with old
3648 and strictly ANSI compatible C compilers is lost. However, all gcc or
3649 LLVM versions of recent years support this extension.
3651 * Timer units gained support for a new setting RandomizedDelaySec= that
3652 allows configuring some additional randomized delay to the configured
3653 time. This is useful to spread out timer events to avoid load peaks in
3654 clusters or larger setups.
3656 * Calendar time specifications now support sub-second accuracy.
3658 * Socket units now support listening on SCTP and UDP-lite protocol
3661 * The sd-event API now comes with a full set of man pages.
3663 * Older versions of systemd contained experimental support for
3664 compressing journal files and coredumps with the LZ4 compressor that
3665 was not compatible with the lz4 binary (due to API limitations of the
3666 lz4 library). This support has been removed; only support for files
3667 compatible with the lz4 binary remains. This LZ4 logic is now
3668 officially supported and no longer considered experimental.
3670 * The dkr image import logic has been removed again from importd. dkr's
3671 micro-services focus doesn't fit into the machine image focus of
3672 importd, and quickly got out of date with the upstream dkr API.
3674 * Creation of the /run/lock/lockdev/ directory was dropped from
3675 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
3676 been available for many years. If you still need this, you need to
3677 create your own tmpfiles.d config file with:
3679 d /run/lock/lockdev 0775 root lock -
3681 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
3682 and RebootArgument= have been moved from the [Service] section of
3683 unit files to [Unit], and they are now supported on all unit types,
3684 not just service units. Of course, systemd will continue to
3685 understand these settings also at the old location, in order to
3686 maintain compatibility.
3688 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
3689 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
3690 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
3691 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
3692 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
3693 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
3694 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
3695 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
3696 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
3697 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
3698 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
3699 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
3700 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
3701 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
3702 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
3703 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
3704 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
3705 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
3706 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
3708 — Berlin, 2016-02-11
3712 * A number of properties previously only settable in unit
3713 files are now also available as properties to set when
3714 creating transient units programmatically via the bus, as it
3715 is exposed with systemd-run's --property=
3716 setting. Specifically, these are: SyslogIdentifier=,
3717 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
3718 EnvironmentFile=, ReadWriteDirectories=,
3719 ReadOnlyDirectories=, InaccessibleDirectories=,
3720 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
3722 * When creating transient services via the bus API it is now
3723 possible to pass in a set of file descriptors to use as
3724 STDIN/STDOUT/STDERR for the invoked process.
3726 * Slice units may now be created transiently via the bus APIs,
3727 similar to the way service and scope units may already be
3728 created transiently.
3730 * Wherever systemd expects a calendar timestamp specification
3731 (like in journalctl's --since= and --until= switches) UTC
3732 timestamps are now supported. Timestamps suffixed with "UTC"
3733 are now considered to be in Universal Time Coordinated
3734 instead of the local timezone. Also, timestamps may now
3735 optionally be specified with sub-second accuracy. Both of
3736 these additions also apply to recurring calendar event
3737 specification, such as OnCalendar= in timer units.
3739 * journalctl gained a new "--sync" switch that asks the
3740 journal daemon to write all so far unwritten log messages to
3741 disk and sync the files, before returning.
3743 * systemd-tmpfiles learned two new line types "q" and "Q" that
3744 operate like "v", but also set up a basic btrfs quota
3745 hierarchy when used on a btrfs file system with quota
3748 * tmpfiles' "v", "q" and "Q" will now create a plain directory
3749 instead of a subvolume (even on a btrfs file system) if the
3750 root directory is a plain directory, and not a
3751 subvolume. This should simplify things with certain chroot()
3752 environments which are not aware of the concept of btrfs
3755 * systemd-detect-virt gained a new --chroot switch to detect
3756 whether execution takes place in a chroot() environment.
3758 * CPUAffinity= now takes CPU index ranges in addition to
3761 * The various memory-related resource limit settings (such as
3762 LimitAS=) now understand the usual K, M, G, ... suffixes to
3763 the base of 1024 (IEC). Similar, the time-related resource
3764 limit settings understand the usual min, h, day, ...
3767 * There's a new system.conf setting DefaultTasksMax= to
3768 control the default TasksMax= setting for services and
3769 scopes running on the system. (TasksMax= is the primary
3770 setting that exposes the "pids" cgroup controller on systemd
3771 and was introduced in the previous systemd release.) The
3772 setting now defaults to 512, which means services that are
3773 not explicitly configured otherwise will only be able to
3774 create 512 processes or threads at maximum, from this
3775 version on. Note that this means that thread- or
3776 process-heavy services might need to be reconfigured to set
3777 TasksMax= to a higher value. It is sufficient to set
3778 TasksMax= in these specific unit files to a higher value, or
3779 even "infinity". Similar, there's now a logind.conf setting
3780 UserTasksMax= that defaults to 4096 and limits the total
3781 number of processes or tasks each user may own
3782 concurrently. nspawn containers also have the TasksMax=
3783 value set by default now, to 8192. Note that all of this
3784 only has an effect if the "pids" cgroup controller is
3785 enabled in the kernel. The general benefit of these changes
3786 should be a more robust and safer system, that provides a
3787 certain amount of per-service fork() bomb protection.
3789 * systemd-nspawn gained the new --network-veth-extra= switch
3790 to define additional and arbitrarily-named virtual Ethernet
3791 links between the host and the container.
3793 * A new service execution setting PassEnvironment= has been
3794 added that allows importing select environment variables
3795 from PID1's environment block into the environment block of
3798 * Timer units gained support for a new RemainAfterElapse=
3799 setting which takes a boolean argument. It defaults to on,
3800 exposing behaviour unchanged to previous releases. If set to
3801 off, timer units are unloaded after they elapsed if they
3802 cannot elapse again. This is particularly useful for
3803 transient timer units, which shall not stay around longer
3804 than until they first elapse.
3806 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
3807 default now (the kernel default is 16). This is beneficial
3808 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
3809 allows substantially larger numbers of queued
3810 datagrams. This should increase the capability of systemd to
3811 parallelize boot-up, as logging and sd_notify() are unlikely
3812 to stall execution anymore. If you need to change the value
3813 from the new defaults, use the usual sysctl.d/ snippets.
3815 * The compression framing format used by the journal or
3816 coredump processing has changed to be in line with what the
3817 official LZ4 tools generate. LZ4 compression support in
3818 systemd was considered unsupported previously, as the format
3819 was not compatible with the normal tools. With this release
3820 this has changed now, and it is hence safe for downstream
3821 distributions to turn it on. While not compressing as well
3822 as the XZ, LZ4 is substantially faster, which makes
3823 it a good default choice for the compression logic in the
3824 journal and in coredump handling.
3826 * Any reference to /etc/mtab has been dropped from
3827 systemd. The file has been obsolete since a while, but
3828 systemd refused to work on systems where it was incorrectly
3829 set up (it should be a symlink or non-existent). Please make
3830 sure to update to util-linux 2.27.1 or newer in conjunction
3831 with this systemd release, which also drops any reference to
3832 /etc/mtab. If you maintain a distribution make sure that no
3833 software you package still references it, as this is a
3834 likely source of bugs. There's also a glibc bug pending,
3835 asking for removal of any reference to this obsolete file:
3837 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
3839 Note that only util-linux versions built with
3840 --enable-libmount-force-mountinfo are supported.
3842 * Support for the ".snapshot" unit type has been removed. This
3843 feature turned out to be little useful and little used, and
3844 has now been removed from the core and from systemctl.
3846 * The dependency types RequiresOverridable= and
3847 RequisiteOverridable= have been removed from systemd. They
3848 have been used only very sparingly to our knowledge and
3849 other options that provide a similar effect (such as
3850 systemctl --mode=ignore-dependencies) are much more useful
3851 and commonly used. Moreover, they were only half-way
3852 implemented as the option to control behaviour regarding
3853 these dependencies was never added to systemctl. By removing
3854 these dependency types the execution engine becomes a bit
3855 simpler. Unit files that use these dependencies should be
3856 changed to use the non-Overridable dependency types
3857 instead. In fact, when parsing unit files with these
3858 options, that's what systemd will automatically convert them
3859 too, but it will also warn, asking users to fix the unit
3860 files accordingly. Removal of these dependency types should
3861 only affect a negligible number of unit files in the wild.
3863 * Behaviour of networkd's IPForward= option changed
3864 (again). It will no longer maintain a per-interface setting,
3865 but propagate one way from interfaces where this is enabled
3866 to the global kernel setting. The global setting will be
3867 enabled when requested by a network that is set up, but
3868 never be disabled again. This change was made to make sure
3869 IPv4 and IPv6 behaviour regarding packet forwarding is
3870 similar (as the Linux IPv6 stack does not support
3871 per-interface control of this setting) and to minimize
3874 * In unit files the behaviour of %u, %U, %h, %s has
3875 changed. These specifiers will now unconditionally resolve
3876 to the various user database fields of the user that the
3877 systemd instance is running as, instead of the user
3878 configured in the specific unit via User=. Note that this
3879 effectively doesn't change much, as resolving of these
3880 specifiers was already turned off in the --system instance
3881 of systemd, as we cannot do NSS lookups from PID 1. In the
3882 --user instance of systemd these specifiers where correctly
3883 resolved, but hardly made any sense, since the user instance
3884 lacks privileges to do user switches anyway, and User= is
3885 hence useless. Moreover, even in the --user instance of
3886 systemd behaviour was awkward as it would only take settings
3887 from User= assignment placed before the specifier into
3888 account. In order to unify and simplify the logic around
3889 this the specifiers will now always resolve to the
3890 credentials of the user invoking the manager (which in case
3891 of PID 1 is the root user).
3893 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
3894 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
3895 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
3896 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
3897 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
3898 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
3899 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
3900 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
3901 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
3902 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
3903 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
3904 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
3905 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
3906 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
3909 — Berlin, 2015-11-18
3913 * systemd now depends on util-linux v2.27. More specifically,
3914 the newly added mount monitor feature in libmount now
3915 replaces systemd's former own implementation.
3917 * libmount mandates /etc/mtab not to be regular file, and
3918 systemd now enforces this condition at early boot.
3919 /etc/mtab has been deprecated and warned about for a very
3920 long time, so systems running systemd should already have
3921 stopped having this file around as anything else than a
3922 symlink to /proc/self/mounts.
3924 * Support for the "pids" cgroup controller has been added. It
3925 allows accounting the number of tasks in a cgroup and
3926 enforcing limits on it. This adds two new setting
3927 TasksAccounting= and TasksMax= to each unit, as well as a
3928 global option DefaultTasksAccounting=.
3930 * Support for the "net_cls" cgroup controller has been added.
3931 It allows assigning a net class ID to each task in the
3932 cgroup, which can then be used in firewall rules and traffic
3933 shaping configurations. Note that the kernel netfilter net
3934 class code does not currently work reliably for ingress
3935 packets on unestablished sockets.
3937 This adds a new config directive called NetClass= to CGroup
3938 enabled units. Allowed values are positive numbers for fixed
3939 assignments and "auto" for picking a free value
3942 * 'systemctl is-system-running' now returns 'offline' if the
3943 system is not booted with systemd. This command can now be
3944 used as a substitute for 'systemd-notify --booted'.
3946 * Watchdog timeouts have been increased to 3 minutes for all
3947 in-tree service files. Apparently, disk IO issues are more
3948 frequent than we hoped, and user reported >1 minute waiting
3951 * 'machine-id-commit' functionality has been merged into
3952 'machine-id-setup --commit'. The separate binary has been
3955 * The WorkingDirectory= directive in unit files may now be set
3956 to the special value '~'. In this case, the working
3957 directory is set to the home directory of the user
3958 configured in User=.
3960 * "machinectl shell" will now open the shell in the home
3961 directory of the selected user by default.
3963 * The CrashChVT= configuration file setting is renamed to
3964 CrashChangeVT=, following our usual logic of not
3965 abbreviating unnecessarily. The old directive is still
3966 supported for compat reasons. Also, this directive now takes
3967 an integer value between 1 and 63, or a boolean value. The
3968 formerly supported '-1' value for disabling stays around for
3971 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
3972 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
3973 RootDirectory= properties can now be set for transient
3976 * The systemd-analyze tool gained a new "set-log-target" verb
3977 to change the logging target the system manager logs to
3978 dynamically during runtime. This is similar to how
3979 "systemd-analyze set-log-level" already changes the log
3982 * In nspawn /sys is now mounted as tmpfs, with only a selected
3983 set of subdirectories mounted in from the real sysfs. This
3984 enhances security slightly, and is useful for ensuring user
3985 namespaces work correctly.
3987 * Support for USB FunctionFS activation has been added. This
3988 allows implementation of USB gadget services that are
3989 activated as soon as they are requested, so that they don't
3990 have to run continuously, similar to classic socket
3993 * The "systemctl exit" command now optionally takes an
3994 additional parameter that sets the exit code to return from
3995 the systemd manager when exiting. This is only relevant when
3996 running the systemd user instance, or when running the
3997 system instance in a container.
3999 * sd-bus gained the new API calls sd_bus_path_encode_many()
4000 and sd_bus_path_decode_many() that allow easy encoding and
4001 decoding of multiple identifier strings inside a D-Bus
4002 object path. Another new call sd_bus_default_flush_close()
4003 has been added to flush and close per-thread default
4006 * systemd-cgtop gained support for a -M/--machine= switch to
4007 show the control groups within a certain container only.
4009 * "systemctl kill" gained support for an optional --fail
4010 switch. If specified the requested operation will fail of no
4011 processes have been killed, because the unit had no
4012 processes attached, or similar.
4014 * A new systemd.crash_reboot=1 kernel command line option has
4015 been added that triggers a reboot after crashing. This can
4016 also be set through CrashReboot= in systemd.conf.
4018 * The RuntimeDirectory= setting now understands unit
4019 specifiers like %i or %f.
4021 * A new (still internal) library API sd-ipv4acd has been added,
4022 that implements address conflict detection for IPv4. It's
4023 based on code from sd-ipv4ll, and will be useful for
4024 detecting DHCP address conflicts.
4026 * File descriptors passed during socket activation may now be
4027 named. A new API sd_listen_fds_with_names() is added to
4028 access the names. The default names may be overridden,
4029 either in the .socket file using the FileDescriptorName=
4030 parameter, or by passing FDNAME= when storing the file
4031 descriptors using sd_notify().
4033 * systemd-networkd gained support for:
4035 - Setting the IPv6 Router Advertisement settings via
4036 IPv6AcceptRouterAdvertisements= in .network files.
4038 - Configuring the HelloTimeSec=, MaxAgeSec= and
4039 ForwardDelaySec= bridge parameters in .netdev files.
4041 - Configuring PreferredSource= for static routes in
4044 * The "ask-password" framework used to query for LUKS harddisk
4045 passwords or SSL passwords during boot gained support for
4046 caching passwords in the kernel keyring, if it is
4047 available. This makes sure that the user only has to type in
4048 a passphrase once if there are multiple objects to unlock
4049 with the same one. Previously, such password caching was
4050 available only when Plymouth was used; this moves the
4051 caching logic into the systemd codebase itself. The
4052 "systemd-ask-password" utility gained a new --keyname=
4053 switch to control which kernel keyring key to use for
4054 caching a password in. This functionality is also useful for
4055 enabling display managers such as gdm to automatically
4056 unlock the user's GNOME keyring if its passphrase, the
4057 user's password and the harddisk password are the same, if
4058 gdm-autologin is used.
4060 * When downloading tar or raw images using "machinectl
4061 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
4062 file is now also downloaded, if it is available and stored
4063 next to the image file.
4065 * Units of type ".socket" gained a new boolean setting
4066 Writable= which is only useful in conjunction with
4067 ListenSpecial=. If true, enables opening the specified
4068 special file in O_RDWR mode rather than O_RDONLY mode.
4070 * systemd-rfkill has been reworked to become a singleton
4071 service that is activated through /dev/rfkill on each rfkill
4072 state change and saves the settings to disk. This way,
4073 systemd-rfkill is now compatible with devices that exist
4074 only intermittendly, and even restores state if the previous
4075 system shutdown was abrupt rather than clean.
4077 * The journal daemon gained support for vacuuming old journal
4078 files controlled by the number of files that shall remain,
4079 in addition to the already existing control by size and by
4080 date. This is useful as journal interleaving performance
4081 degrades with too many separate journal files, and allows
4082 putting an effective limit on them. The new setting defaults
4083 to 100, but this may be changed by setting SystemMaxFiles=
4084 and RuntimeMaxFiles= in journald.conf. Also, the
4085 "journalctl" tool gained the new --vacuum-files= switch to
4086 manually vacuum journal files to leave only the specified
4087 number of files in place.
4089 * udev will now create /dev/disk/by-path links for ATA devices
4090 on kernels where that is supported.
4092 * Galician, Serbian, Turkish and Korean translations were added.
4094 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
4095 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
4096 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
4097 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
4098 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
4099 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
4100 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
4101 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
4102 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
4103 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
4104 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
4105 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
4106 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
4107 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
4108 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
4109 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
4110 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
4111 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
4113 — Berlin, 2015-10-07
4117 * The DHCP implementation of systemd-networkd gained a set of
4120 - The DHCP server now supports emitting DNS and NTP
4121 information. It may be enabled and configured via
4122 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
4123 and NTP information is enabled, but no servers are
4124 configured, the corresponding uplink information (if there
4125 is any) is propagated.
4127 - Server and client now support transmission and reception
4128 of timezone information. It can be configured via the
4129 newly introduced network options UseTimezone=,
4130 EmitTimezone=, and Timezone=. Transmission of timezone
4131 information is enabled between host and containers by
4132 default now: the container will change its local timezone
4133 to what the host has set.
4135 - Lease timeouts can now be configured via
4136 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
4138 - The DHCP server improved on the stability of
4139 leases. Clients are more likely to get the same lease
4140 information back, even if the server loses state.
4142 - The DHCP server supports two new configuration options to
4143 control the lease address pool metrics, PoolOffset= and
4146 * The encapsulation limit of tunnels in systemd-networkd may
4147 now be configured via 'EncapsulationLimit='. It allows
4148 modifying the maximum additional levels of encapsulation
4149 that are permitted to be prepended to a packet.
4151 * systemd now supports the concept of user buses replacing
4152 session buses, if used with dbus-1.10 (and enabled via dbus
4153 --enable-user-session). It previously only supported this on
4154 kdbus-enabled systems, and this release expands this to
4155 'dbus-daemon' systems.
4157 * systemd-networkd now supports predictable interface names
4160 * systemd now optionally supports the new Linux kernel
4161 "unified" control group hierarchy. If enabled via the kernel
4162 command-line option 'systemd.unified_cgroup_hierarchy=1',
4163 systemd will try to mount the unified cgroup hierarchy
4164 directly on /sys/fs/cgroup. If not enabled, or not
4165 available, systemd will fall back to the legacy cgroup
4166 hierarchy setup, as before. Host system and containers can
4167 mix and match legacy and unified hierarchies as they
4168 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
4169 environment variable to individually select the hierarchy to
4170 use for executed containers. By default, nspawn will use the
4171 unified hierarchy for the containers if the host uses the
4172 unified hierarchy, and the legacy hierarchy otherwise.
4173 Please note that at this point the unified hierarchy is an
4174 experimental kernel feature and is likely to change in one
4175 of the next kernel releases. Therefore, it should not be
4176 enabled by default in downstream distributions yet. The
4177 minimum required kernel version for the unified hierarchy to
4178 work is 4.2. Note that when the unified hierarchy is used
4179 for the first time delegated access to controllers is
4180 safe. Because of this systemd-nspawn containers will get
4181 access to controllers now, as will systemd user
4182 sessions. This means containers and user sessions may now
4183 manage their own resources, partitioning up what the system
4186 * A new special scope unit "init.scope" has been introduced
4187 that encapsulates PID 1 of the system. It may be used to
4188 determine resource usage and enforce resource limits on PID
4189 1 itself. PID 1 hence moved out of the root of the control
4192 * The cgtop tool gained support for filtering out kernel
4193 threads when counting tasks in a control group. Also, the
4194 count of processes is now recursively summed up by
4195 default. Two options -k and --recursive= have been added to
4196 revert to old behaviour. The tool has also been updated to
4197 work correctly in containers now.
4199 * systemd-nspawn's --bind= and --bind-ro= options have been
4200 extended to allow creation of non-recursive bind mounts.
4202 * libsystemd gained two new calls sd_pid_get_cgroup() and
4203 sd_peer_get_cgroup() which return the control group path of
4204 a process or peer of a connected AF_UNIX socket. This
4205 function call is particularly useful when implementing
4206 delegated subtrees support in the control group hierarchy.
4208 * The "sd-event" event loop API of libsystemd now supports
4209 correct dequeuing of real-time signals, without losing
4212 * When systemd requests a polkit decision when managing units it
4213 will now add additional fields to the request, including unit
4214 name and desired operation. This enables more powerful polkit
4215 policies, that make decisions depending on these parameters.
4217 * nspawn learnt support for .nspawn settings files, that may
4218 accompany the image files or directories of containers, and
4219 may contain additional settings for the container. This is
4220 an alternative to configuring container parameters via the
4221 nspawn command line.
4223 Contributions from: Cristian Rodríguez, Daniel Mack, David
4224 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
4225 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
4226 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
4227 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
4228 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
4229 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
4230 Andersen, Tom Gundersen, Torstein Husebø
4232 — Berlin, 2015-09-08
4236 * machinectl gained a new verb 'shell' which opens a fresh
4237 shell on the target container or the host. It is similar to
4238 the existing 'login' command of machinectl, but spawns the
4239 shell directly without prompting for username or
4240 password. The pseudo machine '.host' now refers to the local
4241 host and is used by default. Hence, 'machinectl shell' can
4242 be used as replacement for 'su -' which spawns a session as
4243 a fresh systemd unit in a way that is fully isolated from
4244 the originating session.
4246 * systemd-networkd learned to cope with private-zone DHCP
4247 options and allows other programs to query the values.
4249 * SELinux access control when enabling/disabling units is no
4250 longer enforced with this release. The previous implementation
4251 was incorrect, and a new corrected implementation is not yet
4252 available. As unit file operations are still protected via
4253 polkit and D-Bus policy this is not a security problem. Yet,
4254 distributions which care about optimal SELinux support should
4255 probably not stabilize on this release.
4257 * sd-bus gained support for matches of type "arg0has=", that
4258 test for membership of strings in string arrays sent in bus
4261 * systemd-resolved now dumps the contents of its DNS and LLMNR
4262 caches to the logs on reception of the SIGUSR1 signal. This
4263 is useful to debug DNS behaviour.
4265 * The coredumpctl tool gained a new --directory= option to
4266 operate on journal files in a specific directory.
4268 * "systemctl reboot" and related commands gained a new
4269 "--message=" option which may be used to set a free-text
4270 wall message when shutting down or rebooting the
4271 system. This message is also logged, which is useful for
4272 figuring out the reason for a reboot or shutdown a
4275 * The "systemd-resolve-host" tool's -i switch now takes
4276 network interface numbers as alternative to interface names.
4278 * A new unit file setting for services has been introduced:
4279 UtmpMode= allows configuration of how precisely systemd
4280 handles utmp and wtmp entries for the service if this is
4281 enabled. This allows writing services that appear similar to
4282 user sessions in the output of the "w", "who", "last" and
4285 * systemd-resolved will now locally synthesize DNS resource
4286 records for the "localhost" and "gateway" domains as well as
4287 the local hostname. This should ensure that clients querying
4288 RRs via resolved will get similar results as those going via
4289 NSS, if nss-myhostname is enabled.
4291 Contributions from: Alastair Hughes, Alex Crawford, Daniel
4292 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
4293 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
4294 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
4295 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
4296 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
4297 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
4298 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
4299 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
4300 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
4301 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
4302 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4304 — Berlin, 2015-08-27
4308 * The systemd-efi-boot-generator functionality was merged into
4309 systemd-gpt-auto-generator.
4311 * systemd-networkd now supports Group Policy for vxlan
4312 devices. It can be enabled via the new boolean configuration
4313 option called 'GroupPolicyExtension='.
4315 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
4316 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
4317 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
4319 — Berlin, 2015-07-31
4323 * The python-systemd code has been removed from the systemd repository.
4324 A new repository has been created which accommodates the code from
4325 now on, and we kindly ask distributions to create a separate package
4326 for this: https://github.com/systemd/python-systemd
4328 * The systemd daemon will now reload its main configuration
4329 (/etc/systemd/system.conf) on daemon-reload.
4331 * sd-dhcp now exposes vendor specific extensions via
4332 sd_dhcp_lease_get_vendor_specific().
4334 * systemd-networkd gained a number of new configuration options.
4336 - A new boolean configuration option for TAP devices called
4337 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
4338 device, thus allowing to send and receive GSO packets.
4340 - A new tunnel configuration option called 'CopyDSCP='.
4341 If enabled, the DSCP field of ip6 tunnels is copied into the
4342 decapsulated packet.
4344 - A set of boolean bridge configuration options were added.
4345 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
4346 and 'UnicastFlood=' are now parsed by networkd and applied to the
4347 respective bridge link device via the respective IFLA_BRPORT_*
4350 - A new string configuration option to override the hostname sent
4351 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
4352 is true, networkd will use the configured hostname instead of the
4353 system hostname when sending DHCP requests.
4355 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
4356 networkd will configure the IPv6 flow-label of the tunnel device
4357 according to RFC2460.
4359 - The 'macvtap' virtual network devices are now supported, similar to
4360 the already supported 'macvlan' devices.
4362 * systemd-resolved now implements RFC5452 to improve resilience against
4363 cache poisoning. Additionally, source port randomization is enabled
4364 by default to further protect against DNS spoofing attacks.
4366 * nss-mymachines now supports translating UIDs and GIDs of running
4367 containers with user-namespaces enabled. If a container 'foo'
4368 translates a host uid 'UID' to the container uid 'TUID', then
4369 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
4370 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
4371 mapped as 'vg-foo-TGID'.
4373 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
4374 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
4375 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
4376 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
4377 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
4378 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
4379 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
4380 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
4381 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
4382 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
4384 — Berlin, 2015-07-29
4388 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
4389 There are no known issues with current sysfs, and udev does not need
4390 or should be used to work around such bugs.
4392 * udev does no longer enable USB HID power management. Several reports
4393 indicate, that some devices cannot handle that setting.
4395 * The udev accelerometer helper was removed. The functionality
4396 is now fully included in iio-sensor-proxy. But this means,
4397 older iio-sensor-proxy versions will no longer provide
4398 accelerometer/orientation data with this systemd version.
4399 Please upgrade iio-sensor-proxy to version 1.0.
4401 * networkd gained a new configuration option IPv6PrivacyExtensions=
4402 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
4403 for Stateless Address") on selected networks.
4405 * For the sake of fewer build-time dependencies and less code in the
4406 main repository, the python bindings are about to be removed in the
4407 next release. A new repository has been created which accommodates
4408 the code from now on, and we kindly ask distributions to create a
4409 separate package for this. The removal will take place in v223.
4411 https://github.com/systemd/python-systemd
4413 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
4414 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
4415 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
4416 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
4417 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
4418 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
4419 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
4420 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
4421 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
4422 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4424 — Berlin, 2015-07-07
4428 * The sd-bus.h and sd-event.h APIs have now been declared
4429 stable and have been added to the official interface of
4430 libsystemd.so. sd-bus implements an alternative D-Bus client
4431 library, that is relatively easy to use, very efficient and
4432 supports both classic D-Bus as well as kdbus as transport
4433 backend. sd-event is a generic event loop abstraction that
4434 is built around Linux epoll, but adds features such as event
4435 prioritization or efficient timer handling. Both APIs are good
4436 choices for C programs looking for a bus and/or event loop
4437 implementation that is minimal and does not have to be
4438 portable to other kernels.
4440 * kdbus support is no longer compile-time optional. It is now
4441 always built-in. However, it can still be disabled at
4442 runtime using the kdbus=0 kernel command line setting, and
4443 that setting may be changed to default to off, by specifying
4444 --disable-kdbus at build-time. Note though that the kernel
4445 command line setting has no effect if the kdbus.ko kernel
4446 module is not installed, in which case kdbus is (obviously)
4447 also disabled. We encourage all downstream distributions to
4448 begin testing kdbus by adding it to the kernel images in the
4449 development distributions, and leaving kdbus support in
4452 * The minimal required util-linux version has been bumped to
4455 * Support for chkconfig (--enable-chkconfig) was removed in
4456 favor of calling an abstraction tool
4457 /lib/systemd/systemd-sysv-install. This needs to be
4458 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
4459 in README for details.
4461 * If there's a systemd unit and a SysV init script for the
4462 same service name, and the user executes "systemctl enable"
4463 for it (or a related call), then this will now enable both
4464 (or execute the related operation on both), not just the
4467 * The libudev API documentation has been converted from gtkdoc
4470 * gudev has been removed from the systemd tree, it is now an
4473 * The systemd-cgtop tool learnt a new --raw switch to generate
4474 "raw" (machine parsable) output.
4476 * networkd's IPForwarding= .network file setting learnt the
4477 new setting "kernel", which ensures that networkd does not
4478 change the IP forwarding sysctl from the default kernel
4481 * The systemd-logind bus API now exposes a new boolean
4482 property "Docked" that reports whether logind considers the
4483 system "docked", i.e. connected to a docking station or not.
4485 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
4486 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
4487 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
4488 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
4489 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
4490 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
4491 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
4492 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
4493 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
4494 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
4495 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
4496 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
4497 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
4498 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
4499 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
4500 Fink, Zbigniew Jędrzejewski-Szmek
4502 — Berlin, 2015-06-19
4506 * The gudev library has been extracted into a separate repository
4507 available at: https://git.gnome.org/browse/libgudev/
4508 It is now managed as part of the Gnome project. Distributions
4509 are recommended to pass --disable-gudev to systemd and use
4510 gudev from the Gnome project instead. gudev is still included
4511 in systemd, for now. It will be removed soon, though. Please
4512 also see the announcement-thread on systemd-devel:
4513 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
4515 * systemd now exposes a CPUUsageNSec= property for each
4516 service unit on the bus, that contains the overall consumed
4517 CPU time of a service (the sum of what each process of the
4518 service consumed). This value is only available if
4519 CPUAccounting= is turned on for a service, and is then shown
4520 in the "systemctl status" output.
4522 * Support for configuring alternative mappings of the old SysV
4523 runlevels to systemd targets has been removed. They are now
4524 hardcoded in a way that runlevels 2, 3, 4 all map to
4525 multi-user.target and 5 to graphical.target (which
4526 previously was already the default behaviour).
4528 * The auto-mounter logic gained support for mount point
4529 expiry, using a new TimeoutIdleSec= setting in .automount
4530 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
4532 * The EFI System Partition (ESP) as mounted to /boot by
4533 systemd-efi-boot-generator will now be unmounted
4534 automatically after 2 minutes of not being used. This should
4535 minimize the risk of ESP corruptions.
4537 * New /etc/fstab options x-systemd.requires= and
4538 x-systemd.requires-mounts-for= are now supported to express
4539 additional dependencies for mounts. This is useful for
4540 journalling file systems that support external journal
4541 devices or overlay file systems that require underlying file
4542 systems to be mounted.
4544 * systemd does not support direct live-upgrades (via systemctl
4545 daemon-reexec) from versions older than v44 anymore. As no
4546 distribution we are aware of shipped such old versions in a
4547 stable release this should not be problematic.
4549 * When systemd forks off a new per-connection service instance
4550 it will now set the $REMOTE_ADDR environment variable to the
4551 remote IP address, and $REMOTE_PORT environment variable to
4552 the remote IP port. This behaviour is similar to the
4553 corresponding environment variables defined by CGI.
4555 * systemd-networkd gained support for uplink failure
4556 detection. The BindCarrier= option allows binding interface
4557 configuration dynamically to the link sense of other
4558 interfaces. This is useful to achieve behaviour like in
4561 * systemd-networkd gained support for configuring the DHCP
4562 client identifier to use when requesting leases.
4564 * systemd-networkd now has a per-network UseNTP= option to
4565 configure whether NTP server information acquired via DHCP
4566 is passed on to services like systemd-timesyncd.
4568 * systemd-networkd gained support for vti6 tunnels.
4570 * Note that systemd-networkd manages the sysctl variable
4571 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
4572 it is configured for since v219. The variable controls IP
4573 forwarding, and is a per-interface alternative to the global
4574 /proc/sys/net/ipv[46]/ip_forward. This setting is
4575 configurable in the IPForward= option, which defaults to
4576 "no". This means if networkd is used for an interface it is
4577 no longer sufficient to set the global sysctl option to turn
4578 on IP forwarding! Instead, the .network file option
4579 IPForward= needs to be turned on! Note that the
4580 implementation of this behaviour was broken in v219 and has
4583 * Many bonding and vxlan options are now configurable in
4586 * systemd-nspawn gained a new --property= setting to set unit
4587 properties for the container scope. This is useful for
4588 setting resource parameters (e.g. "CPUShares=500") on
4589 containers started from the command line.
4591 * systemd-nspawn gained a new --private-users= switch to make
4592 use of user namespacing available on recent Linux kernels.
4594 * systemd-nspawn may now be called as part of a shell pipeline
4595 in which case the pipes used for stdin and stdout are passed
4596 directly to the process invoked in the container, without
4597 indirection via a pseudo tty.
4599 * systemd-nspawn gained a new switch to control the UNIX
4600 signal to use when killing the init process of the container
4603 * systemd-nspawn gained a new --overlay= switch for mounting
4604 overlay file systems into the container using the new kernel
4607 * When a container image is imported via systemd-importd and
4608 the host file system is not btrfs, a loopback block device
4609 file is created in /var/lib/machines.raw with a btrfs file
4610 system inside. It is then mounted to /var/lib/machines to
4611 enable btrfs features for container management. The loopback
4612 file and btrfs file system is grown as needed when container
4613 images are imported via systemd-importd.
4615 * systemd-machined/systemd-importd gained support for btrfs
4616 quota, to enforce container disk space limits on disk. This
4617 is exposed in "machinectl set-limit".
4619 * systemd-importd now can import containers from local .tar,
4620 .raw and .qcow2 images, and export them to .tar and .raw. It
4621 can also import dkr v2 images now from the network (on top
4624 * systemd-importd gained support for verifying downloaded
4625 images with gpg2 (previously only gpg1 was supported).
4627 * systemd-machined, systemd-logind, systemd: most bus calls are
4628 now accessible to unprivileged processes via polkit. Also,
4629 systemd-logind will now allow users to kill their own sessions
4630 without further privileges or authorization.
4632 * systemd-shutdownd has been removed. This service was
4633 previously responsible for implementing scheduled shutdowns
4634 as exposed in /usr/bin/shutdown's time parameter. This
4635 functionality has now been moved into systemd-logind and is
4636 accessible via a bus interface.
4638 * "systemctl reboot" gained a new switch --firmware-setup that
4639 can be used to reboot into the EFI firmware setup, if that
4640 is available. systemd-logind now exposes an API on the bus
4641 to trigger such reboots, in case graphical desktop UIs want
4642 to cover this functionality.
4644 * "systemctl enable", "systemctl disable" and "systemctl mask"
4645 now support a new "--now" switch. If specified the units
4646 that are enabled will also be started, and the ones
4647 disabled/masked also stopped.
4649 * The Gummiboot EFI boot loader tool has been merged into
4650 systemd, and renamed to "systemd-boot". The bootctl tool has been
4651 updated to support systemd-boot.
4653 * An EFI kernel stub has been added that may be used to create
4654 kernel EFI binaries that contain not only the actual kernel,
4655 but also an initrd, boot splash, command line and OS release
4656 information. This combined binary can then be signed as a
4657 single image, so that the firmware can verify it all in one
4658 step. systemd-boot has special support for EFI binaries created
4659 like this and can extract OS release information from them
4660 and show them in the boot menu. This functionality is useful
4661 to implement cryptographically verified boot schemes.
4663 * Optional support has been added to systemd-fsck to pass
4664 fsck's progress report to an AF_UNIX socket in the file
4667 * udev will no longer create device symlinks for all block
4668 devices by default. A blacklist for excluding special block
4669 devices from this logic has been turned into a whitelist
4670 that requires picking block devices explicitly that require
4673 * A new (currently still internal) API sd-device.h has been
4674 added to libsystemd. This modernized API is supposed to
4675 replace libudev eventually. In fact, already much of libudev
4676 is now just a wrapper around sd-device.h.
4678 * A new hwdb database for storing metadata about pointing
4679 stick devices has been added.
4681 * systemd-tmpfiles gained support for setting file attributes
4682 similar to the "chattr" tool with new 'h' and 'H' lines.
4684 * systemd-journald will no longer unconditionally set the
4685 btrfs NOCOW flag on new journal files. This is instead done
4686 with tmpfiles snippet using the new 'h' line type. This
4687 allows easy disabling of this logic, by masking the
4688 journal-nocow.conf tmpfiles file.
4690 * systemd-journald will now translate audit message types to
4691 human readable identifiers when writing them to the
4692 journal. This should improve readability of audit messages.
4694 * The LUKS logic gained support for the offset= and skip=
4695 options in /etc/crypttab, as previously implemented by
4698 * /usr/lib/os-release gained a new optional field VARIANT= for
4699 distributions that support multiple variants (such as a
4700 desktop edition, a server edition, ...)
4702 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
4703 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
4704 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
4705 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
4706 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
4707 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
4708 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
4709 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
4710 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
4711 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
4712 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
4713 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
4714 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
4715 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
4716 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
4717 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
4718 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
4719 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
4720 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
4721 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
4722 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
4723 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
4724 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
4725 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
4726 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
4727 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
4728 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
4730 — Berlin, 2015-05-22
4734 * Introduce a new API "sd-hwdb.h" for querying the hardware
4735 metadata database. With this minimal interface one can query
4736 and enumerate the udev hwdb, decoupled from the old libudev
4737 library. libudev's interface for this is now only a wrapper
4738 around sd-hwdb. A new tool systemd-hwdb has been added to
4739 interface with and update the database.
4741 * When any of systemd's tools copies files (for example due to
4742 tmpfiles' C lines) a btrfs reflink will attempted first,
4743 before bytewise copying is done.
4745 * systemd-nspawn gained a new --ephemeral switch. When
4746 specified a btrfs snapshot is taken of the container's root
4747 directory, and immediately removed when the container
4748 terminates again. Thus, a container can be started whose
4749 changes never alter the container's root directory, and are
4750 lost on container termination. This switch can also be used
4751 for starting a container off the root file system of the
4752 host without affecting the host OS. This switch is only
4753 available on btrfs file systems.
4755 * systemd-nspawn gained a new --template= switch. It takes the
4756 path to a container tree to use as template for the tree
4757 specified via --directory=, should that directory be
4758 missing. This allows instantiating containers dynamically,
4759 on first run. This switch is only available on btrfs file
4762 * When a .mount unit refers to a mount point on which multiple
4763 mounts are stacked, and the .mount unit is stopped all of
4764 the stacked mount points will now be unmounted until no
4765 mount point remains.
4767 * systemd now has an explicit notion of supported and
4768 unsupported unit types. Jobs enqueued for unsupported unit
4769 types will now fail with an "unsupported" error code. More
4770 specifically .swap, .automount and .device units are not
4771 supported in containers, .busname units are not supported on
4772 non-kdbus systems. .swap and .automount are also not
4773 supported if their respective kernel compile time options
4776 * machinectl gained support for two new "copy-from" and
4777 "copy-to" commands for copying files from a running
4778 container to the host or vice versa.
4780 * machinectl gained support for a new "bind" command to bind
4781 mount host directories into local containers. This is
4782 currently only supported for nspawn containers.
4784 * networkd gained support for configuring bridge forwarding
4785 database entries (fdb) from .network files.
4787 * A new tiny daemon "systemd-importd" has been added that can
4788 download container images in tar, raw, qcow2 or dkr formats,
4789 and make them available locally in /var/lib/machines, so
4790 that they can run as nspawn containers. The daemon can GPG
4791 verify the downloads (not supported for dkr, since it has no
4792 provisions for verifying downloads). It will transparently
4793 decompress bz2, xz, gzip compressed downloads if necessary,
4794 and restore sparse files on disk. The daemon uses privilege
4795 separation to ensure the actual download logic runs with
4796 fewer privileges than the daemon itself. machinectl has
4797 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
4798 make the functionality of importd available to the
4799 user. With this in place the Fedora and Ubuntu "Cloud"
4800 images can be downloaded and booted as containers unmodified
4801 (the Fedora images lack the appropriate GPG signature files
4802 currently, so they cannot be verified, but this will change
4803 soon, hopefully). Note that downloading images is currently
4804 only fully supported on btrfs.
4806 * machinectl is now able to list container images found in
4807 /var/lib/machines, along with some metadata about sizes of
4808 disk and similar. If the directory is located on btrfs and
4809 quota is enabled, this includes quota display. A new command
4810 "image-status" has been added that shows additional
4811 information about images.
4813 * machinectl is now able to clone container images
4814 efficiently, if the underlying file system (btrfs) supports
4815 it, with the new "machinectl clone" command. It also
4816 gained commands for renaming and removing images, as well as
4817 marking them read-only or read-write (supported also on
4818 legacy file systems).
4820 * networkd gained support for collecting LLDP network
4821 announcements, from hardware that supports this. This is
4822 shown in networkctl output.
4824 * systemd-run gained support for a new -t (--pty) switch for
4825 invoking a binary on a pty whose input and output is
4826 connected to the invoking terminal. This allows executing
4827 processes as system services while interactively
4828 communicating with them via the terminal. Most interestingly
4829 this is supported across container boundaries. Invoking
4830 "systemd-run -t /bin/bash" is an alternative to running a
4831 full login session, the difference being that the former
4832 will not register a session, nor go through the PAM session
4835 * tmpfiles gained support for a new "v" line type for creating
4836 btrfs subvolumes. If the underlying file system is a legacy
4837 file system, this automatically degrades to creating a
4838 normal directory. Among others /var/lib/machines is now
4839 created like this at boot, should it be missing.
4841 * The directory /var/lib/containers/ has been deprecated and
4842 been replaced by /var/lib/machines. The term "machines" has
4843 been used in the systemd context as generic term for both
4844 VMs and containers, and hence appears more appropriate for
4845 this, as the directory can also contain raw images bootable
4848 * systemd-nspawn when invoked with -M but without --directory=
4849 or --image= is now capable of searching for the container
4850 root directory, subvolume or disk image automatically, in
4851 /var/lib/machines. systemd-nspawn@.service has been updated
4852 to make use of this, thus allowing it to be used for raw
4855 * A new machines.target unit has been introduced that is
4856 supposed to group all containers/VMs invoked as services on
4857 the system. systemd-nspawn@.service has been updated to
4858 integrate with that.
4860 * machinectl gained a new "start" command, for invoking a
4861 container as a service. "machinectl start foo" is mostly
4862 equivalent to "systemctl start systemd-nspawn@foo.service",
4863 but handles escaping in a nicer way.
4865 * systemd-nspawn will now mount most of the cgroupfs tree
4866 read-only into each container, with the exception of the
4867 container's own subtree in the name=systemd hierarchy.
4869 * journald now sets the special FS_NOCOW file flag for its
4870 journal files. This should improve performance on btrfs, by
4871 avoiding heavy fragmentation when journald's write-pattern
4872 is used on COW file systems. It degrades btrfs' data
4873 integrity guarantees for the files to the same levels as for
4874 ext3/ext4 however. This should be OK though as journald does
4875 its own data integrity checks and all its objects are
4876 checksummed on disk. Also, journald should handle btrfs disk
4877 full events a lot more gracefully now, by processing SIGBUS
4878 errors, and not relying on fallocate() anymore.
4880 * When journald detects that journal files it is writing to
4881 have been deleted it will immediately start new journal
4884 * systemd now provides a way to store file descriptors
4885 per-service in PID 1. This is useful for daemons to ensure
4886 that fds they require are not lost during a daemon
4887 restart. The fds are passed to the daemon on the next
4888 invocation in the same way socket activation fds are
4889 passed. This is now used by journald to ensure that the
4890 various sockets connected to all the system's stdout/stderr
4891 are not lost when journald is restarted. File descriptors
4892 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
4893 an extension to sd_notify(). Note that a limit is enforced
4894 on the number of fds a service can store in PID 1, and it
4895 defaults to 0, so that no fds may be stored, unless this is
4896 explicitly turned on.
4898 * The default TERM variable to use for units connected to a
4899 terminal, when no other value is explicitly is set is now
4900 vt220 rather than vt102. This should be fairly safe still,
4901 but allows PgUp/PgDn work.
4903 * The /etc/crypttab option header= as known from Debian is now
4906 * "loginctl user-status" and "loginctl session-status" will
4907 now show the last 10 lines of log messages of the
4908 user/session following the status output. Similar,
4909 "machinectl status" will show the last 10 log lines
4910 associated with a virtual machine or container
4911 service. (Note that this is usually not the log messages
4912 done in the VM/container itself, but simply what the
4913 container manager logs. For nspawn this includes all console
4916 * "loginctl session-status" without further argument will now
4917 show the status of the session of the caller. Similar,
4918 "lock-session", "unlock-session", "activate",
4919 "enable-linger", "disable-linger" may now be called without
4920 session/user parameter in which case they apply to the
4921 caller's session/user.
4923 * An X11 session scriptlet is now shipped that uploads
4924 $DISPLAY and $XAUTHORITY into the environment of the systemd
4925 --user daemon if a session begins. This should improve
4926 compatibility with X11 enabled applications run as systemd
4929 * Generators are now subject to masking via /etc and /run, the
4930 same way as unit files.
4932 * networkd .network files gained support for configuring
4933 per-link IPv4/IPv6 packet forwarding as well as IPv4
4934 masquerading. This is by default turned on for veth links to
4935 containers, as registered by systemd-nspawn. This means that
4936 nspawn containers run with --network-veth will now get
4937 automatic routed access to the host's networks without any
4938 further configuration or setup, as long as networkd runs on
4941 * systemd-nspawn gained the --port= (-p) switch to expose TCP
4942 or UDP posts of a container on the host. With this in place
4943 it is possible to run containers with private veth links
4944 (--network-veth), and have their functionality exposed on
4945 the host as if their services were running directly on the
4948 * systemd-nspawn's --network-veth switch now gained a short
4949 version "-n", since with the changes above it is now truly
4950 useful out-of-the-box. The systemd-nspawn@.service has been
4951 updated to make use of it too by default.
4953 * systemd-nspawn will now maintain a per-image R/W lock, to
4954 ensure that the same image is not started more than once
4955 writable. (It's OK to run an image multiple times
4956 simultaneously in read-only mode.)
4958 * systemd-nspawn's --image= option is now capable of
4959 dissecting and booting MBR and GPT disk images that contain
4960 only a single active Linux partition. Previously it
4961 supported only GPT disk images with proper GPT type
4962 IDs. This allows running cloud images from major
4963 distributions directly with systemd-nspawn, without
4966 * In addition to collecting mouse dpi data in the udev
4967 hardware database, there's now support for collecting angle
4968 information for mouse scroll wheels. The database is
4969 supposed to guarantee similar scrolling behavior on mice
4970 that it knows about. There's also support for collecting
4971 information about Touchpad types.
4973 * udev's input_id built-in will now also collect touch screen
4974 dimension data and attach it to probed devices.
4976 * /etc/os-release gained support for a Distribution Privacy
4979 * networkd gained support for creating "ipvlan", "gretap",
4980 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
4982 * systemd-tmpfiles gained support for "a" lines for setting
4985 * systemd-nspawn will now mount /tmp in the container to
4986 tmpfs, automatically.
4988 * systemd now exposes the memory.usage_in_bytes cgroup
4989 attribute and shows it for each service in the "systemctl
4990 status" output, if available.
4992 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
4993 immediate reboot is triggered. This useful if shutdown is
4994 hung and is unable to complete, to expedite the
4995 operation. Note that this kind of reboot will still unmount
4996 all file systems, and hence should not result in fsck being
4999 * A .device unit for an optical block device will now be
5000 considered active only when a medium is in the drive. Also,
5001 mount units are now bound to their backing devices thus
5002 triggering automatic unmounting when devices become
5003 unavailable. With this in place systemd will now
5004 automatically unmount left-over mounts when a CD-ROM is
5005 ejected or an USB stick is yanked from the system.
5007 * networkd-wait-online now has support for waiting for
5008 specific interfaces only (with globbing), and for giving up
5009 after a configurable timeout.
5011 * networkd now exits when idle. It will be automatically
5012 restarted as soon as interfaces show up, are removed or
5013 change state. networkd will stay around as long as there is
5014 at least one DHCP state machine or similar around, that keep
5017 * networkd may now configure IPv6 link-local addressing in
5018 addition to IPv4 link-local addressing.
5020 * The IPv6 "token" for use in SLAAC may now be configured for
5021 each .network interface in networkd.
5023 * Routes configured with networkd may now be assigned a scope
5026 * networkd's [Match] sections now support globbing and lists
5027 of multiple space-separated matches per item.
5029 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
5030 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
5031 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
5032 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
5033 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
5034 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
5035 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
5036 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
5037 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
5038 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
5039 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
5040 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
5041 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
5042 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
5043 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
5044 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
5045 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
5046 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
5047 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
5048 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
5049 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
5050 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
5051 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
5052 Hoffmann, Zbigniew Jędrzejewski-Szmek
5054 — Berlin, 2015-02-16
5058 * When querying unit file enablement status (for example via
5059 "systemctl is-enabled"), a new state "indirect" is now known
5060 which indicates that a unit might not be enabled itself, but
5061 another unit listed in its Also= setting might be.
5063 * Similar to the various existing ConditionXYZ= settings for
5064 units, there are now matching AssertXYZ= settings. While
5065 failing conditions cause a unit to be skipped, but its job
5066 to succeed, failing assertions declared like this will cause
5067 a unit start operation and its job to fail.
5069 * hostnamed now knows a new chassis type "embedded".
5071 * systemctl gained a new "edit" command. When used on a unit
5072 file, this allows extending unit files with .d/ drop-in
5073 configuration snippets or editing the full file (after
5074 copying it from /usr/lib to /etc). This will invoke the
5075 user's editor (as configured with $EDITOR), and reload the
5076 modified configuration after editing.
5078 * "systemctl status" now shows the suggested enablement state
5079 for a unit, as declared in the (usually vendor-supplied)
5080 system preset files.
5082 * nss-myhostname will now resolve the single-label host name
5083 "gateway" to the locally configured default IP routing
5084 gateways, ordered by their metrics. This assigns a stable
5085 name to the used gateways, regardless which ones are
5086 currently configured. Note that the name will only be
5087 resolved after all other name sources (if nss-myhostname is
5088 configured properly) and should hence not negatively impact
5089 systems that use the single-label host name "gateway" in
5092 * systemd-inhibit now allows filtering by mode when listing
5095 * Scope and service units gained a new "Delegate" boolean
5096 property, which, when set, allows processes running inside the
5097 unit to further partition resources. This is primarily
5098 useful for systemd user instances as well as container
5101 * journald will now pick up audit messages directly from
5102 the kernel, and log them like any other log message. The
5103 audit fields are split up and fully indexed. This means that
5104 journalctl in many ways is now a (nicer!) alternative to
5105 ausearch, the traditional audit client. Note that this
5106 implements only a minimal audit client. If you want the
5107 special audit modes like reboot-on-log-overflow, please use
5108 the traditional auditd instead, which can be used in
5109 parallel to journald.
5111 * The ConditionSecurity= unit file option now understands the
5112 special string "audit" to check whether auditing is
5115 * journalctl gained two new commands --vacuum-size= and
5116 --vacuum-time= to delete old journal files until the
5117 remaining ones take up no more than the specified size on disk,
5118 or are not older than the specified time.
5120 * A new, native PPPoE library has been added to sd-network,
5121 systemd's library of light-weight networking protocols. This
5122 library will be used in a future version of networkd to
5123 enable PPPoE communication without an external pppd daemon.
5125 * The busctl tool now understands a new "capture" verb that
5126 works similar to "monitor", but writes a packet capture
5127 trace to STDOUT that can be redirected to a file which is
5128 compatible with libcap's capture file format. This can then
5129 be loaded in Wireshark and similar tools to inspect bus
5132 * The busctl tool now understands a new "tree" verb that shows
5133 the object trees of a specific service on the bus, or of all
5136 * The busctl tool now understands a new "introspect" verb that
5137 shows all interfaces and members of objects on the bus,
5138 including their signature and values. This is particularly
5139 useful to get more information about bus objects shown by
5140 the new "busctl tree" command.
5142 * The busctl tool now understands new verbs "call",
5143 "set-property" and "get-property" for invoking bus method
5144 calls, setting and getting bus object properties in a
5147 * busctl gained a new --augment-creds= argument that controls
5148 whether the tool shall augment credential information it
5149 gets from the bus with data from /proc, in a possibly
5152 * nspawn's --link-journal= switch gained two new values
5153 "try-guest" and "try-host" that work like "guest" and
5154 "host", but do not fail if the host has no persistent
5155 journalling enabled. -j is now equivalent to
5156 --link-journal=try-guest.
5158 * macvlan network devices created by nspawn will now have
5159 stable MAC addresses.
5161 * A new SmackProcessLabel= unit setting has been added, which
5162 controls the SMACK security label processes forked off by
5163 the respective unit shall use.
5165 * If compiled with --enable-xkbcommon, systemd-localed will
5166 verify x11 keymap settings by compiling the given keymap. It
5167 will spew out warnings if the compilation fails. This
5168 requires libxkbcommon to be installed.
5170 * When a coredump is collected, a larger number of metadata
5171 fields is now collected and included in the journal records
5172 created for it. More specifically, control group membership,
5173 environment variables, memory maps, working directory,
5174 chroot directory, /proc/$PID/status, and a list of open file
5175 descriptors is now stored in the log entry.
5177 * The udev hwdb now contains DPI information for mice. For
5180 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
5182 * All systemd programs that read standalone configuration
5183 files in /etc now also support a corresponding series of
5184 .conf.d configuration directories in /etc/, /run/,
5185 /usr/local/lib/, /usr/lib/, and (if configured with
5186 --enable-split-usr) /lib/. In particular, the following
5187 configuration files now have corresponding configuration
5188 directories: system.conf user.conf, logind.conf,
5189 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
5190 resolved.conf, timesyncd.conf, journal-remote.conf, and
5191 journal-upload.conf. Note that distributions should use the
5192 configuration directories in /usr/lib/; the directories in
5193 /etc/ are reserved for the system administrator.
5195 * systemd-rfkill will no longer take the rfkill device name
5196 into account when storing rfkill state on disk, as the name
5197 might be dynamically assigned and not stable. Instead, the
5198 ID_PATH udev variable combined with the rfkill type (wlan,
5199 bluetooth, ...) is used.
5201 * A new service systemd-machine-id-commit.service has been
5202 added. When used on systems where /etc is read-only during
5203 boot, and /etc/machine-id is not initialized (but an empty
5204 file), this service will copy the temporary machine ID
5205 created as replacement into /etc after the system is fully
5206 booted up. This is useful for systems that are freshly
5207 installed with a non-initialized machine ID, but should get
5208 a fixed machine ID for subsequent boots.
5210 * networkd's .netdev files now provide a large set of
5211 configuration parameters for VXLAN devices. Similarly, the
5212 bridge port cost parameter is now configurable in .network
5213 files. There's also new support for configuring IP source
5214 routing. networkd .link files gained support for a new
5215 OriginalName= match that is useful to match against the
5216 original interface name the kernel assigned. .network files
5217 may include MTU= and MACAddress= fields for altering the MTU
5218 and MAC address while being connected to a specific network
5221 * The LUKS logic gained supported for configuring
5222 UUID-specific key files. There's also new support for naming
5223 LUKS device from the kernel command line, using the new
5224 luks.name= argument.
5226 * Timer units may now be transiently created via the bus API
5227 (this was previously already available for scope and service
5228 units). In addition it is now possible to create multiple
5229 transient units at the same time with a single bus call. The
5230 "systemd-run" tool has been updated to make use of this for
5231 running commands on a specified time, in at(1)-style.
5233 * tmpfiles gained support for "t" lines, for assigning
5234 extended attributes to files. Among other uses this may be
5235 used to assign SMACK labels to files.
5237 Contributions from: Alin Rauta, Alison Chaiken, Andrej
5238 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
5239 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
5240 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
5241 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
5242 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
5243 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
5244 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
5245 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
5246 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
5247 Michael Biebl, Michael Chapman, Michael Marineau, Michal
5248 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
5249 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
5250 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
5251 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
5252 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5253 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
5254 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
5256 — Berlin, 2014-12-10
5260 * journalctl gained the new options -t/--identifier= to match
5261 on the syslog identifier (aka "tag"), as well as --utc to
5262 show log timestamps in the UTC timezone. journalctl now also
5263 accepts -n/--lines=all to disable line capping in a pager.
5265 * journalctl gained a new switch, --flush, that synchronously
5266 flushes logs from /run/log/journal to /var/log/journal if
5267 persistent storage is enabled. systemd-journal-flush.service
5268 now waits until the operation is complete.
5270 * Services can notify the manager before they start a reload
5271 (by sending RELOADING=1) or shutdown (by sending
5272 STOPPING=1). This allows the manager to track and show the
5273 internal state of daemons and closes a race condition when
5274 the process is still running but has closed its D-Bus
5277 * Services with Type=oneshot do not have to have any ExecStart
5280 * User units are now loaded also from
5281 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
5282 /run/systemd/user directory that was already previously
5283 supported, but is under the control of the user.
5285 * Job timeouts (i.e. timeouts on the time a job that is
5286 queued stays in the run queue) can now optionally result in
5287 immediate reboot or power-off actions (JobTimeoutAction= and
5288 JobTimeoutRebootArgument=). This is useful on ".target"
5289 units, to limit the maximum time a target remains
5290 undispatched in the run queue, and to trigger an emergency
5291 operation in such a case. This is now used by default to
5292 turn off the system if boot-up (as defined by everything in
5293 basic.target) hangs and does not complete for at least
5294 15min. Also, if power-off or reboot hang for at least 30min
5295 an immediate power-off/reboot operation is triggered. This
5296 functionality is particularly useful to increase reliability
5297 on embedded devices, but also on laptops which might
5298 accidentally get powered on when carried in a backpack and
5299 whose boot stays stuck in a hard disk encryption passphrase
5302 * systemd-logind can be configured to also handle lid switch
5303 events even when the machine is docked or multiple displays
5304 are attached (HandleLidSwitchDocked= option).
5306 * A helper binary and a service have been added which can be
5307 used to resume from hibernation in the initramfs. A
5308 generator will parse the resume= option on the kernel
5309 command line to trigger resume.
5311 * A user console daemon systemd-consoled has been
5312 added. Currently, it is a preview, and will so far open a
5313 single terminal on each session of the user marked as
5314 Desktop=systemd-console.
5316 * Route metrics can be specified for DHCP routes added by
5319 * The SELinux context of socket-activated services can be set
5320 from the information provided by the networking stack
5321 (SELinuxContextFromNet= option).
5323 * Userspace firmware loading support has been removed and
5324 the minimum supported kernel version is thus bumped to 3.7.
5326 * Timeout for udev workers has been increased from 1 to 3
5327 minutes, but a warning will be printed after 1 minute to
5328 help diagnose kernel modules that take a long time to load.
5330 * Udev rules can now remove tags on devices with TAG-="foobar".
5332 * systemd's readahead implementation has been removed. In many
5333 circumstances it didn't give expected benefits even for
5334 rotational disk drives and was becoming less relevant in the
5335 age of SSDs. As none of the developers has been using
5336 rotating media anymore, and nobody stepped up to actively
5337 maintain this component of systemd it has now been removed.
5339 * Swap units can use Options= to specify discard options.
5340 Discard options specified for swaps in /etc/fstab are now
5343 * Docker containers are now detected as a separate type of
5346 * The Password Agent protocol gained support for queries where
5347 the user input is shown, useful e.g. for user names.
5348 systemd-ask-password gained a new --echo option to turn that
5351 * The default sysctl.d/ snippets will now set:
5353 net.core.default_qdisc = fq_codel
5355 This selects Fair Queuing Controlled Delay as the default
5356 queuing discipline for network interfaces. fq_codel helps
5357 fight the network bufferbloat problem. It is believed to be
5358 a good default with no tuning required for most workloads.
5359 Downstream distributions may override this choice. On 10Gbit
5360 servers that do not do forwarding, "fq" may perform better.
5361 Systems without a good clocksource should use "pfifo_fast".
5363 * If kdbus is enabled during build a new option BusPolicy= is
5364 available for service units, that allows locking all service
5365 processes into a stricter bus policy, in order to limit
5366 access to various bus services, or even hide most of them
5367 from the service's view entirely.
5369 * networkctl will now show the .network and .link file
5370 networkd has applied to a specific interface.
5372 * sd-login gained a new API call sd_session_get_desktop() to
5373 query which desktop environment has been selected for a
5376 * UNIX utmp support is now compile-time optional to support
5377 legacy-free systems.
5379 * systemctl gained two new commands "add-wants" and
5380 "add-requires" for pulling in units from specific targets
5383 * If the word "rescue" is specified on the kernel command line
5384 the system will now boot into rescue mode (aka
5385 rescue.target), which was previously available only by
5386 specifying "1" or "systemd.unit=rescue.target" on the kernel
5387 command line. This new kernel command line option nicely
5388 mirrors the already existing "emergency" kernel command line
5391 * New kernel command line options mount.usr=, mount.usrflags=,
5392 mount.usrfstype= have been added that match root=, rootflags=,
5393 rootfstype= but allow mounting a specific file system to
5396 * The $NOTIFY_SOCKET is now also passed to control processes of
5397 services, not only the main process.
5399 * This version reenables support for fsck's -l switch. This
5400 means at least version v2.25 of util-linux is required for
5401 operation, otherwise dead-locks on device nodes may
5402 occur. Again: you need to update util-linux to at least
5403 v2.25 when updating systemd to v217.
5405 * The "multi-seat-x" tool has been removed from systemd, as
5406 its functionality has been integrated into X servers 1.16,
5407 and the tool is hence redundant. It is recommended to update
5408 display managers invoking this tool to simply invoke X
5409 directly from now on, again.
5411 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
5412 message flag has been added for all of systemd's polkit
5413 authenticated method calls has been added. In particular this
5414 now allows optional interactive authorization via polkit for
5415 many of PID1's privileged operations such as unit file
5416 enabling and disabling.
5418 * "udevadm hwdb --update" learnt a new switch "--usr" for
5419 placing the rebuilt hardware database in /usr instead of
5420 /etc. When used only hardware database entries stored in
5421 /usr will be used, and any user database entries in /etc are
5422 ignored. This functionality is useful for vendors to ship a
5423 pre-built database on systems where local configuration is
5424 unnecessary or unlikely.
5426 * Calendar time specifications in .timer units now also
5427 understand the strings "semi-annually", "quarterly" and
5428 "minutely" as shortcuts (in addition to the preexisting
5429 "anually", "hourly", ...).
5431 * systemd-tmpfiles will now correctly create files in /dev
5432 at boot which are marked for creation only at boot. It is
5433 recommended to always create static device nodes with 'c!'
5434 and 'b!', so that they are created only at boot and not
5435 overwritten at runtime.
5437 * When the watchdog logic is used for a service (WatchdogSec=)
5438 and the watchdog timeout is hit the service will now be
5439 terminated with SIGABRT (instead of just SIGTERM), in order
5440 to make sure a proper coredump and backtrace is
5441 generated. This ensures that hanging services will result in
5442 similar coredump/backtrace behaviour as services that hit a
5445 Contributions from: Andreas Henriksson, Andrei Borzenkov,
5446 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
5447 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
5448 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
5449 Herrmann, David Sommerseth, David Strauss, Emil Renner
5450 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
5451 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
5452 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
5453 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
5454 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
5455 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
5456 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
5457 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
5458 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
5459 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
5460 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
5461 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
5462 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
5463 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
5464 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
5465 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
5468 — Berlin, 2014-10-28
5472 * timedated no longer reads NTP implementation unit names from
5473 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
5474 implementations should add a
5476 Conflicts=systemd-timesyncd.service
5478 to their unit files to take over and replace systemd's NTP
5479 default functionality.
5481 * systemd-sysusers gained a new line type "r" for configuring
5482 which UID/GID ranges to allocate system users/groups
5483 from. Lines of type "u" may now add an additional column
5484 that specifies the home directory for the system user to be
5485 created. Also, systemd-sysusers may now optionally read user
5486 information from STDIN instead of a file. This is useful for
5487 invoking it from RPM preinst scriptlets that need to create
5488 users before the first RPM file is installed since these
5489 files might need to be owned by them. A new
5490 %sysusers_create_inline RPM macro has been introduced to do
5491 just that. systemd-sysusers now updates the shadow files as
5492 well as the user/group databases, which should enhance
5493 compatibility with certain tools like grpck.
5495 * A number of bus APIs of PID 1 now optionally consult polkit to
5496 permit access for otherwise unprivileged clients under certain
5497 conditions. Note that this currently doesn't support
5498 interactive authentication yet, but this is expected to be
5499 added eventually, too.
5501 * /etc/machine-info now has new fields for configuring the
5502 deployment environment of the machine, as well as the
5503 location of the machine. hostnamectl has been updated with
5504 new command to update these fields.
5506 * systemd-timesyncd has been updated to automatically acquire
5507 NTP server information from systemd-networkd, which might
5508 have been discovered via DHCP.
5510 * systemd-resolved now includes a caching DNS stub resolver
5511 and a complete LLMNR name resolution implementation. A new
5512 NSS module "nss-resolve" has been added which can be used
5513 instead of glibc's own "nss-dns" to resolve hostnames via
5514 systemd-resolved. Hostnames, addresses and arbitrary RRs may
5515 be resolved via systemd-resolved D-Bus APIs. In contrast to
5516 the glibc internal resolver systemd-resolved is aware of
5517 multi-homed system, and keeps DNS server and caches separate
5518 and per-interface. Queries are sent simultaneously on all
5519 interfaces that have DNS servers configured, in order to
5520 properly handle VPNs and local LANs which might resolve
5521 separate sets of domain names. systemd-resolved may acquire
5522 DNS server information from systemd-networkd automatically,
5523 which in turn might have discovered them via DHCP. A tool
5524 "systemd-resolve-host" has been added that may be used to
5525 query the DNS logic in resolved. systemd-resolved implements
5526 IDNA and automatically uses IDNA or UTF-8 encoding depending
5527 on whether classic DNS or LLMNR is used as transport. In the
5528 next releases we intend to add a DNSSEC and mDNS/DNS-SD
5529 implementation to systemd-resolved.
5531 * A new NSS module nss-mymachines has been added, that
5532 automatically resolves the names of all local registered
5533 containers to their respective IP addresses.
5535 * A new client tool "networkctl" for systemd-networkd has been
5536 added. It currently is entirely passive and will query
5537 networking configuration from udev, rtnetlink and networkd,
5538 and present it to the user in a very friendly
5539 way. Eventually, we hope to extend it to become a full
5540 control utility for networkd.
5542 * .socket units gained a new DeferAcceptSec= setting that
5543 controls the kernels' TCP_DEFER_ACCEPT sockopt for
5544 TCP. Similarly, support for controlling TCP keep-alive
5545 settings has been added (KeepAliveTimeSec=,
5546 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
5547 turning off Nagle's algorithm on TCP has been added
5550 * logind learned a new session type "web", for use in projects
5551 like Cockpit which register web clients as PAM sessions.
5553 * timer units with at least one OnCalendar= setting will now
5554 be started only after time-sync.target has been
5555 reached. This way they will not elapse before the system
5556 clock has been corrected by a local NTP client or
5557 similar. This is particular useful on RTC-less embedded
5558 machines, that come up with an invalid system clock.
5560 * systemd-nspawn's --network-veth= switch should now result in
5561 stable MAC addresses for both the outer and the inner side
5564 * systemd-nspawn gained a new --volatile= switch for running
5565 container instances with /etc or /var unpopulated.
5567 * The kdbus client code has been updated to use the new Linux
5568 3.17 memfd subsystem instead of the old kdbus-specific one.
5570 * systemd-networkd's DHCP client and server now support
5571 FORCERENEW. There are also new configuration options to
5572 configure the vendor client identifier and broadcast mode
5575 * systemd will no longer inform the kernel about the current
5576 timezone, as this is necessarily incorrect and racy as the
5577 kernel has no understanding of DST and similar
5578 concepts. This hence means FAT timestamps will be always
5579 considered UTC, similar to what Android is already
5580 doing. Also, when the RTC is configured to the local time
5581 (rather than UTC) systemd will never synchronize back to it,
5582 as this might confuse Windows at a later boot.
5584 * systemd-analyze gained a new command "verify" for offline
5585 validation of unit files.
5587 * systemd-networkd gained support for a couple of additional
5588 settings for bonding networking setups. Also, the metric for
5589 statically configured routes may now be configured. For
5590 network interfaces where this is appropriate the peer IP
5591 address may now be configured.
5593 * systemd-networkd's DHCP client will no longer request
5594 broadcasting by default, as this tripped up some networks.
5595 For hardware where broadcast is required the feature should
5596 be switched back on using RequestBroadcast=yes.
5598 * systemd-networkd will now set up IPv4LL addresses (when
5599 enabled) even if DHCP is configured successfully.
5601 * udev will now default to respect network device names given
5602 by the kernel when the kernel indicates that these are
5603 predictable. This behavior can be tweaked by changing
5604 NamePolicy= in the relevant .link file.
5606 * A new library systemd-terminal has been added that
5607 implements full TTY stream parsing and rendering. This
5608 library is supposed to be used later on for implementing a
5609 full userspace VT subsystem, replacing the current kernel
5612 * A new tool systemd-journal-upload has been added to push
5613 journal data to a remote system running
5614 systemd-journal-remote.
5616 * journald will no longer forward all local data to another
5617 running syslog daemon. This change has been made because
5618 rsyslog (which appears to be the most commonly used syslog
5619 implementation these days) no longer makes use of this, and
5620 instead pulls the data out of the journal on its own. Since
5621 forwarding the messages to a non-existent syslog server is
5622 more expensive than we assumed we have now turned this
5623 off. If you run a syslog server that is not a recent rsyslog
5624 version, you have to turn this option on again
5625 (ForwardToSyslog= in journald.conf).
5627 * journald now optionally supports the LZ4 compressor for
5628 larger journal fields. This compressor should perform much
5629 better than XZ which was the previous default.
5631 * machinectl now shows the IP addresses of local containers,
5632 if it knows them, plus the interface name of the container.
5634 * A new tool "systemd-escape" has been added that makes it
5635 easy to escape strings to build unit names and similar.
5637 * sd_notify() messages may now include a new ERRNO= field
5638 which is parsed and collected by systemd and shown among the
5639 "systemctl status" output for a service.
5641 * A new component "systemd-firstboot" has been added that
5642 queries the most basic systemd information (timezone,
5643 hostname, root password) interactively on first
5644 boot. Alternatively it may also be used to provision these
5645 things offline on OS images installed into directories.
5647 * The default sysctl.d/ snippets will now set
5649 net.ipv4.conf.default.promote_secondaries=1
5651 This has the benefit of no flushing secondary IP addresses
5652 when primary addresses are removed.
5654 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
5655 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
5656 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
5657 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
5658 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
5659 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
5660 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
5661 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
5662 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
5663 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
5664 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
5665 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
5666 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
5667 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
5668 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
5670 — Berlin, 2014-08-19
5674 * A new tool systemd-sysusers has been added. This tool
5675 creates system users and groups in /etc/passwd and
5676 /etc/group, based on static declarative system user/group
5677 definitions in /usr/lib/sysusers.d/. This is useful to
5678 enable factory resets and volatile systems that boot up with
5679 an empty /etc directory, and thus need system users and
5680 groups created during early boot. systemd now also ships
5681 with two default sysusers.d/ files for the most basic
5682 users and groups systemd and the core operating system
5685 * A new tmpfiles snippet has been added that rebuilds the
5686 essential files in /etc on boot, should they be missing.
5688 * A directive for ensuring automatic clean-up of
5689 /var/cache/man/ has been removed from the default
5690 configuration. This line should now be shipped by the man
5691 implementation. The necessary change has been made to the
5692 man-db implementation. Note that you need to update your man
5693 implementation to one that ships this line, otherwise no
5694 automatic clean-up of /var/cache/man will take place.
5696 * A new condition ConditionNeedsUpdate= has been added that
5697 may conditionalize services to only run when /etc or /var
5698 are "older" than the vendor operating system resources in
5699 /usr. This is useful for reconstructing or updating /etc
5700 after an offline update of /usr or a factory reset, on the
5701 next reboot. Services that want to run once after such an
5702 update or reset should use this condition and order
5703 themselves before the new systemd-update-done.service, which
5704 will mark the two directories as fully updated. A number of
5705 service files have been added making use of this, to rebuild
5706 the udev hardware database, the journald message catalog and
5707 dynamic loader cache (ldconfig). The systemd-sysusers tool
5708 described above also makes use of this now. With this in
5709 place it is now possible to start up a minimal operating
5710 system with /etc empty cleanly. For more information on the
5711 concepts involved see this recent blog story:
5713 http://0pointer.de/blog/projects/stateless.html
5715 * A new system group "input" has been introduced, and all
5716 input device nodes get this group assigned. This is useful
5717 for system-level software to get access to input devices. It
5718 complements what is already done for "audio" and "video".
5720 * systemd-networkd learnt minimal DHCPv4 server support in
5721 addition to the existing DHCPv4 client support. It also
5722 learnt DHCPv6 client and IPv6 Router Solicitation client
5723 support. The DHCPv4 client gained support for static routes
5724 passed in from the server. Note that the [DHCPv4] section
5725 known in older systemd-networkd versions has been renamed to
5726 [DHCP] and is now also used by the DHCPv6 client. Existing
5727 .network files using settings of this section should be
5728 updated, though compatibility is maintained. Optionally, the
5729 client hostname may now be sent to the DHCP server.
5731 * networkd gained support for vxlan virtual networks as well
5732 as tun/tap and dummy devices.
5734 * networkd gained support for automatic allocation of address
5735 ranges for interfaces from a system-wide pool of
5736 addresses. This is useful for dynamically managing a large
5737 number of interfaces with a single network configuration
5738 file. In particular this is useful to easily assign
5739 appropriate IP addresses to the veth links of a large number
5740 of nspawn instances.
5742 * RPM macros for processing sysusers, sysctl and binfmt
5743 drop-in snippets at package installation time have been
5746 * The /etc/os-release file should now be placed in
5747 /usr/lib/os-release. The old location is automatically
5748 created as symlink. /usr/lib is the more appropriate
5749 location of this file, since it shall actually describe the
5750 vendor operating system shipped in /usr, and not the
5751 configuration stored in /etc.
5753 * .mount units gained a new boolean SloppyOptions= setting
5754 that maps to mount(8)'s -s option which enables permissive
5755 parsing of unknown mount options.
5757 * tmpfiles learnt a new "L+" directive which creates a symlink
5758 but (unlike "L") deletes a pre-existing file first, should
5759 it already exist and not already be the correct
5760 symlink. Similarly, "b+", "c+" and "p+" directives have been
5761 added as well, which create block and character devices, as
5762 well as fifos in the filesystem, possibly removing any
5763 pre-existing files of different types.
5765 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
5766 'argument' field (which so far specified the source to
5767 symlink/copy the files from) is now optional. If omitted the
5768 same file os copied from /usr/share/factory/ suffixed by the
5769 full destination path. This is useful for populating /etc
5770 with essential files, by copying them from vendor defaults
5771 shipped in /usr/share/factory/etc.
5773 * A new command "systemctl preset-all" has been added that
5774 applies the service preset settings to all installed unit
5775 files. A new switch --preset-mode= has been added that
5776 controls whether only enable or only disable operations
5779 * A new command "systemctl is-system-running" has been added
5780 that allows checking the overall state of the system, for
5781 example whether it is fully up and running.
5783 * When the system boots up with an empty /etc, the equivalent
5784 to "systemctl preset-all" is executed during early boot, to
5785 make sure all default services are enabled after a factory
5788 * systemd now contains a minimal preset file that enables the
5789 most basic services systemd ships by default.
5791 * Unit files' [Install] section gained a new DefaultInstance=
5792 field for defining the default instance to create if a
5793 template unit is enabled with no instance specified.
5795 * A new passive target cryptsetup-pre.target has been added
5796 that may be used by services that need to make they run and
5797 finish before the first LUKS cryptographic device is set up.
5799 * The /dev/loop-control and /dev/btrfs-control device nodes
5800 are now owned by the "disk" group by default, opening up
5801 access to this group.
5803 * systemd-coredump will now automatically generate a
5804 stack trace of all core dumps taking place on the system,
5805 based on elfutils' libdw library. This stack trace is logged
5808 * systemd-coredump may now optionally store coredumps directly
5809 on disk (in /var/lib/systemd/coredump, possibly compressed),
5810 instead of storing them unconditionally in the journal. This
5811 mode is the new default. A new configuration file
5812 /etc/systemd/coredump.conf has been added to configure this
5813 and other parameters of systemd-coredump.
5815 * coredumpctl gained a new "info" verb to show details about a
5816 specific coredump. A new switch "-1" has also been added
5817 that makes sure to only show information about the most
5818 recent entry instead of all entries. Also, as the tool is
5819 generally useful now the "systemd-" prefix of the binary
5820 name has been removed. Distributions that want to maintain
5821 compatibility with the old name should add a symlink from
5822 the old name to the new name.
5824 * journald's SplitMode= now defaults to "uid". This makes sure
5825 that unprivileged users can access their own coredumps with
5826 coredumpctl without restrictions.
5828 * New kernel command line options "systemd.wants=" (for
5829 pulling an additional unit during boot), "systemd.mask="
5830 (for masking a specific unit for the boot), and
5831 "systemd.debug-shell" (for enabling the debug shell on tty9)
5832 have been added. This is implemented in the new generator
5833 "systemd-debug-generator".
5835 * systemd-nspawn will now by default filter a couple of
5836 syscalls for containers, among them those required for
5837 kernel module loading, direct x86 IO port access, swap
5838 management, and kexec. Most importantly though
5839 open_by_handle_at() is now prohibited for containers,
5840 closing a hole similar to a recently discussed vulnerability
5841 in docker regarding access to files on file hierarchies the
5842 container should normally not have access to. Note that, for
5843 nspawn, we generally make no security claims anyway (and
5844 this is explicitly documented in the man page), so this is
5845 just a fix for one of the most obvious problems.
5847 * A new man page file-hierarchy(7) has been added that
5848 contains a minimized, modernized version of the file system
5849 layout systemd expects, similar in style to the FHS
5850 specification or hier(5). A new tool systemd-path(1) has
5851 been added to query many of these paths for the local
5854 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
5855 longer done. Since the directory now has a per-user size
5856 limit, and is cleaned on logout this appears unnecessary,
5857 in particular since this now brings the lifecycle of this
5858 directory closer in line with how IPC objects are handled.
5860 * systemd.pc now exports a number of additional directories,
5861 including $libdir (which is useful to identify the library
5862 path for the primary architecture of the system), and a
5863 couple of drop-in directories.
5865 * udev's predictable network interface names now use the dev_port
5866 sysfs attribute, introduced in linux 3.15 instead of dev_id to
5867 distinguish between ports of the same PCI function. dev_id should
5868 only be used for ports using the same HW address, hence the need
5871 * machined has been updated to export the OS version of a
5872 container (read from /etc/os-release and
5873 /usr/lib/os-release) on the bus. This is now shown in
5874 "machinectl status" for a machine.
5876 * A new service setting RestartForceExitStatus= has been
5877 added. If configured to a set of exit signals or process
5878 return values, the service will be restarted when the main
5879 daemon process exits with any of them, regardless of the
5882 * systemctl's -H switch for connecting to remote systemd
5883 machines has been extended so that it may be used to
5884 directly connect to a specific container on the
5885 host. "systemctl -H root@foobar:waldi" will now connect as
5886 user "root" to host "foobar", and then proceed directly to
5887 the container named "waldi". Note that currently you have to
5888 authenticate as user "root" for this to work, as entering
5889 containers is a privileged operation.
5891 Contributions from: Andreas Henriksson, Benjamin Steinwender,
5892 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
5893 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
5894 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
5895 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
5896 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
5897 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
5898 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
5899 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
5900 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
5901 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
5902 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
5904 — Berlin, 2014-07-03
5908 * As an experimental feature, udev now tries to lock the
5909 disk device node (flock(LOCK_SH|LOCK_NB)) while it
5910 executes events for the disk or any of its partitions.
5911 Applications like partitioning programs can lock the
5912 disk device node (flock(LOCK_EX)) and claim temporary
5913 device ownership that way; udev will entirely skip all event
5914 handling for this disk and its partitions. If the disk
5915 was opened for writing, the close will trigger a partition
5916 table rescan in udev's "watch" facility, and if needed
5917 synthesize "change" events for the disk and all its partitions.
5918 This is now unconditionally enabled, and if it turns out to
5919 cause major problems, we might turn it on only for specific
5920 devices, or might need to disable it entirely. Device Mapper
5921 devices are excluded from this logic.
5923 * We temporarily dropped the "-l" switch for fsck invocations,
5924 since they collide with the flock() logic above. util-linux
5925 upstream has been changed already to avoid this conflict,
5926 and we will readd "-l" as soon as util-linux with this
5927 change has been released.
5929 * The dependency on libattr has been removed. Since a long
5930 time, the extended attribute calls have moved to glibc, and
5931 libattr is thus unnecessary.
5933 * Virtualization detection works without privileges now. This
5934 means the systemd-detect-virt binary no longer requires
5935 CAP_SYS_PTRACE file capabilities, and our daemons can run
5936 with fewer privileges.
5938 * systemd-networkd now runs under its own "systemd-network"
5939 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
5940 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
5941 loses the ability to write to files owned by root this way.
5943 * Similarly, systemd-resolved now runs under its own
5944 "systemd-resolve" user with no capabilities remaining.
5946 * Similarly, systemd-bus-proxyd now runs under its own
5947 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
5949 * systemd-networkd gained support for setting up "veth"
5950 virtual Ethernet devices for container connectivity, as well
5951 as GRE and VTI tunnels.
5953 * systemd-networkd will no longer automatically attempt to
5954 manually load kernel modules necessary for certain tunnel
5955 transports. Instead, it is assumed the kernel loads them
5956 automatically when required. This only works correctly on
5957 very new kernels. On older kernels, please consider adding
5958 the kernel modules to /etc/modules-load.d/ as a work-around.
5960 * The resolv.conf file systemd-resolved generates has been
5961 moved to /run/systemd/resolve/. If you have a symlink from
5962 /etc/resolv.conf, it might be necessary to correct it.
5964 * Two new service settings, ProtectHome= and ProtectSystem=,
5965 have been added. When enabled, they will make the user data
5966 (such as /home) inaccessible or read-only and the system
5967 (such as /usr) read-only, for specific services. This allows
5968 very light-weight per-service sandboxing to avoid
5969 modifications of user data or system files from
5970 services. These two new switches have been enabled for all
5971 of systemd's long-running services, where appropriate.
5973 * Socket units gained new SocketUser= and SocketGroup=
5974 settings to set the owner user and group of AF_UNIX sockets
5975 and FIFOs in the file system.
5977 * Socket units gained a new RemoveOnStop= setting. If enabled,
5978 all FIFOS and sockets in the file system will be removed
5979 when the specific socket unit is stopped.
5981 * Socket units gained a new Symlinks= setting. It takes a list
5982 of symlinks to create to file system sockets or FIFOs
5983 created by the specific Unix sockets. This is useful to
5984 manage symlinks to socket nodes with the same lifecycle as
5987 * The /dev/log socket and /dev/initctl FIFO have been moved to
5988 /run, and have been replaced by symlinks. This allows
5989 connecting to these facilities even if PrivateDevices=yes is
5990 used for a service (which makes /dev/log itself unavailable,
5991 but /run is left). This also has the benefit of ensuring
5992 that /dev only contains device nodes, directories and
5993 symlinks, and nothing else.
5995 * sd-daemon gained two new calls sd_pid_notify() and
5996 sd_pid_notifyf(). They are similar to sd_notify() and
5997 sd_notifyf(), but allow overriding of the source PID of
5998 notification messages if permissions permit this. This is
5999 useful to send notify messages on behalf of a different
6000 process (for example, the parent process). The
6001 systemd-notify tool has been updated to make use of this
6002 when sending messages (so that notification messages now
6003 originate from the shell script invoking systemd-notify and
6004 not the systemd-notify process itself. This should minimize
6005 a race where systemd fails to associate notification
6006 messages to services when the originating process already
6009 * A new "on-abnormal" setting for Restart= has been added. If
6010 set, it will result in automatic restarts on all "abnormal"
6011 reasons for a process to exit, which includes unclean
6012 signals, core dumps, timeouts and watchdog timeouts, but
6013 does not include clean and unclean exit codes or clean
6014 signals. Restart=on-abnormal is an alternative for
6015 Restart=on-failure for services that shall be able to
6016 terminate and avoid restarts on certain errors, by
6017 indicating so with an unclean exit code. Restart=on-failure
6018 or Restart=on-abnormal is now the recommended setting for
6019 all long-running services.
6021 * If the InaccessibleDirectories= service setting points to a
6022 mount point (or if there are any submounts contained within
6023 it), it is now attempted to completely unmount it, to make
6024 the file systems truly unavailable for the respective
6027 * The ReadOnlyDirectories= service setting and
6028 systemd-nspawn's --read-only parameter are now recursively
6029 applied to all submounts, too.
6031 * Mount units may now be created transiently via the bus APIs.
6033 * The support for SysV and LSB init scripts has been removed
6034 from the systemd daemon itself. Instead, it is now
6035 implemented as a generator that creates native systemd units
6036 from these scripts when needed. This enables us to remove a
6037 substantial amount of legacy code from PID 1, following the
6038 fact that many distributions only ship a very small number
6039 of LSB/SysV init scripts nowadays.
6041 * Privileged Xen (dom0) domains are not considered
6042 virtualization anymore by the virtualization detection
6043 logic. After all, they generally have unrestricted access to
6044 the hardware and usually are used to manage the unprivileged
6047 * systemd-tmpfiles gained a new "C" line type, for copying
6048 files or entire directories.
6050 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
6051 lines. So far, they have been non-globbing versions of the
6052 latter, and have thus been redundant. In future, it is
6053 recommended to only use "z". "m" has hence been removed
6054 from the documentation, even though it stays supported.
6056 * A tmpfiles snippet to recreate the most basic structure in
6057 /var has been added. This is enough to create the /var/run →
6058 /run symlink and create a couple of structural
6059 directories. This allows systems to boot up with an empty or
6060 volatile /var. Of course, while with this change, the core OS
6061 now is capable with dealing with a volatile /var, not all
6062 user services are ready for it. However, we hope that sooner
6063 or later, many service daemons will be changed upstream so
6064 that they are able to automatically create their necessary
6065 directories in /var at boot, should they be missing. This is
6066 the first step to allow state-less systems that only require
6067 the vendor image for /usr to boot.
6069 * systemd-nspawn has gained a new --tmpfs= switch to mount an
6070 empty tmpfs instance to a specific directory. This is
6071 particularly useful for making use of the automatic
6072 reconstruction of /var (see above), by passing --tmpfs=/var.
6074 * Access modes specified in tmpfiles snippets may now be
6075 prefixed with "~", which indicates that they shall be masked
6076 by whether the existing file or directory is currently
6077 writable, readable or executable at all. Also, if specified,
6078 the sgid/suid/sticky bits will be masked for all
6081 * A new passive target unit "network-pre.target" has been
6082 added which is useful for services that shall run before any
6083 network is configured, for example firewall scripts.
6085 * The "floppy" group that previously owned the /dev/fd*
6086 devices is no longer used. The "disk" group is now used
6087 instead. Distributions should probably deprecate usage of
6090 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
6091 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
6092 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
6093 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
6094 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
6095 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
6098 — Berlin, 2014-06-11
6102 * A new "systemd-timesyncd" daemon has been added for
6103 synchronizing the system clock across the network. It
6104 implements an SNTP client. In contrast to NTP
6105 implementations such as chrony or the NTP reference server,
6106 this only implements a client side, and does not bother with
6107 the full NTP complexity, focusing only on querying time from
6108 one remote server and synchronizing the local clock to
6109 it. Unless you intend to serve NTP to networked clients or
6110 want to connect to local hardware clocks, this simple NTP
6111 client should be more than appropriate for most
6112 installations. The daemon runs with minimal privileges, and
6113 has been hooked up with networkd to only operate when
6114 network connectivity is available. The daemon saves the
6115 current clock to disk every time a new NTP sync has been
6116 acquired, and uses this to possibly correct the system clock
6117 early at bootup, in order to accommodate for systems that
6118 lack an RTC such as the Raspberry Pi and embedded devices,
6119 and to make sure that time monotonically progresses on these
6120 systems, even if it is not always correct. To make use of
6121 this daemon, a new system user and group "systemd-timesync"
6122 needs to be created on installation of systemd.
6124 * The queue "seqnum" interface of libudev has been disabled, as
6125 it was generally incompatible with device namespacing as
6126 sequence numbers of devices go "missing" if the devices are
6127 part of a different namespace.
6129 * "systemctl list-timers" and "systemctl list-sockets" gained
6130 a --recursive switch for showing units of these types also
6131 for all local containers, similar in style to the already
6132 supported --recursive switch for "systemctl list-units".
6134 * A new RebootArgument= setting has been added for service
6135 units, which may be used to specify a kernel reboot argument
6136 to use when triggering reboots with StartLimitAction=.
6138 * A new FailureAction= setting has been added for service
6139 units which may be used to specify an operation to trigger
6140 when a service fails. This works similarly to
6141 StartLimitAction=, but unlike it, controls what is done
6142 immediately rather than only after several attempts to
6143 restart the service in question.
6145 * hostnamed got updated to also expose the kernel name,
6146 release, and version on the bus. This is useful for
6147 executing commands like hostnamectl with the -H switch.
6148 systemd-analyze makes use of this to properly display
6149 details when running non-locally.
6151 * The bootchart tool can now show cgroup information in the
6152 graphs it generates.
6154 * The CFS CPU quota cgroup attribute is now exposed for
6155 services. The new CPUQuota= switch has been added for this
6156 which takes a percentage value. Setting this will have the
6157 result that a service may never get more CPU time than the
6158 specified percentage, even if the machine is otherwise idle.
6160 * systemd-networkd learned IPIP and SIT tunnel support.
6162 * LSB init scripts exposing a dependency on $network will now
6163 get a dependency on network-online.target rather than simply
6164 network.target. This should bring LSB handling closer to
6165 what it was on SysV systems.
6167 * A new fsck.repair= kernel option has been added to control
6168 how fsck shall deal with unclean file systems at boot.
6170 * The (.ini) configuration file parser will now silently
6171 ignore sections whose name begins with "X-". This may be
6172 used to maintain application-specific extension sections in unit
6175 * machined gained a new API to query the IP addresses of
6176 registered containers. "machinectl status" has been updated
6177 to show these addresses in its output.
6179 * A new call sd_uid_get_display() has been added to the
6180 sd-login APIs for querying the "primary" session of a
6181 user. The "primary" session of the user is elected from the
6182 user's sessions and generally a graphical session is
6183 preferred over a text one.
6185 * A minimal systemd-resolved daemon has been added. It
6186 currently simply acts as a companion to systemd-networkd and
6187 manages resolv.conf based on per-interface DNS
6188 configuration, possibly supplied via DHCP. In the long run
6189 we hope to extend this into a local DNSSEC enabled DNS and
6192 * The systemd-networkd-wait-online tool is now enabled by
6193 default. It will delay network-online.target until a network
6194 connection has been configured. The tool primarily integrates
6195 with networkd, but will also make a best effort to make sense
6196 of network configuration performed in some other way.
6198 * Two new service options StartupCPUShares= and
6199 StartupBlockIOWeight= have been added that work similarly to
6200 CPUShares= and BlockIOWeight= however only apply during
6201 system startup. This is useful to prioritize certain services
6202 differently during bootup than during normal runtime.
6204 * hostnamed has been changed to prefer the statically
6205 configured hostname in /etc/hostname (unless set to
6206 'localhost' or empty) over any dynamic one supplied by
6207 dhcp. With this change, the rules for picking the hostname
6208 match more closely the rules of other configuration settings
6209 where the local administrator's configuration in /etc always
6210 overrides any other settings.
6212 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
6213 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
6214 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
6215 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
6216 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
6217 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
6218 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
6219 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
6220 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
6221 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
6222 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
6223 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
6224 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
6225 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
6226 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
6227 Lindskog, WaLyong Cho, Will Woods, Zbigniew
6230 — Beijing, 2014-05-28
6234 * When restoring the screen brightness at boot, stay away from
6235 the darkest setting or from the lowest 5% of the available
6236 range, depending on which is the larger value of both. This
6237 should effectively protect the user from rebooting into a
6238 black screen, should the brightness have been set to minimum
6241 * sd-login gained a new sd_machine_get_class() call to
6242 determine the class ("vm" or "container") of a machine
6243 registered with machined.
6245 * sd-login gained new calls
6246 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
6247 to query the identity of the peer of a local AF_UNIX
6248 connection. They operate similarly to their sd_pid_get_xyz()
6251 * PID 1 will now maintain a system-wide system state engine
6252 with the states "starting", "running", "degraded",
6253 "maintenance", "stopping". These states are bound to system
6254 startup, normal runtime, runtime with at least one failed
6255 service, rescue/emergency mode and system shutdown. This
6256 state is shown in the "systemctl status" output when no unit
6257 name is passed. It is useful to determine system state, in
6258 particularly when doing so for many systems or containers at
6261 * A new command "list-machines" has been added to "systemctl"
6262 that lists all local OS containers and shows their system
6263 state (see above), if systemd runs inside of them.
6265 * systemctl gained a new "-r" switch to recursively enumerate
6266 units on all local containers, when used with the
6267 "list-unit" command (which is the default one that is
6268 executed when no parameters are specified).
6270 * The GPT automatic partition discovery logic will now honour
6271 two GPT partition flags: one may be set on a partition to
6272 cause it to be mounted read-only, and the other may be set
6273 on a partition to ignore it during automatic discovery.
6275 * Two new GPT type UUIDs have been added for automatic root
6276 partition discovery, for 32-bit and 64-bit ARM. This is not
6277 particularly useful for discovering the root directory on
6278 these architectures during bare-metal boots (since UEFI is
6279 not common there), but still very useful to allow booting of
6280 ARM disk images in nspawn with the -i option.
6282 * MAC addresses of interfaces created with nspawn's
6283 --network-interface= switch will now be generated from the
6284 machine name, and thus be stable between multiple invocations
6287 * logind will now automatically remove all IPC objects owned
6288 by a user if she or he fully logs out. This makes sure that
6289 users who are logged out cannot continue to consume IPC
6290 resources. This covers SysV memory, semaphores and message
6291 queues as well as POSIX shared memory and message
6292 queues. Traditionally, SysV and POSIX IPC had no lifecycle
6293 limits. With this functionality, that is corrected. This may
6294 be turned off by using the RemoveIPC= switch of logind.conf.
6296 * The systemd-machine-id-setup and tmpfiles tools gained a
6297 --root= switch to operate on a specific root directory,
6300 * journald can now forward logged messages to the TTYs of all
6301 logged in users ("wall"). This is the default for all
6302 emergency messages now.
6304 * A new tool systemd-journal-remote has been added to stream
6305 journal log messages across the network.
6307 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
6308 controller trees are mounted into it. Note that the
6309 directories mounted beneath it are not read-only. This is a
6310 security measure and is particularly useful because glibc
6311 actually includes a search logic to pick any tmpfs it can
6312 find to implement shm_open() if /dev/shm is not available
6313 (which it might very well be in namespaced setups).
6315 * machinectl gained a new "poweroff" command to cleanly power
6316 down a local OS container.
6318 * The PrivateDevices= unit file setting will now also drop the
6319 CAP_MKNOD capability from the capability bound set, and
6320 imply DevicePolicy=closed.
6322 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
6323 comprehensively on all long-running systemd services where
6324 this is appropriate.
6326 * systemd-udevd will now run in a disassociated mount
6327 namespace. To mount directories from udev rules, make sure to
6328 pull in mount units via SYSTEMD_WANTS properties.
6330 * The kdbus support gained support for uploading policy into
6331 the kernel. sd-bus gained support for creating "monitoring"
6332 connections that can eavesdrop into all bus communication
6333 for debugging purposes.
6335 * Timestamps may now be specified in seconds since the UNIX
6336 epoch Jan 1st, 1970 by specifying "@" followed by the value
6339 * Native tcpwrap support in systemd has been removed. tcpwrap
6340 is old code, not really maintained anymore and has serious
6341 shortcomings, and better options such as firewalls
6342 exist. For setups that require tcpwrap usage, please
6343 consider invoking your socket-activated service via tcpd,
6344 like on traditional inetd.
6346 * A new system.conf configuration option
6347 DefaultTimerAccuracySec= has been added that controls the
6348 default AccuracySec= setting of .timer units.
6350 * Timer units gained a new WakeSystem= switch. If enabled,
6351 timers configured this way will cause the system to resume
6352 from system suspend (if the system supports that, which most
6355 * Timer units gained a new Persistent= switch. If enabled,
6356 timers configured this way will save to disk when they have
6357 been last triggered. This information is then used on next
6358 reboot to possible execute overdue timer events, that
6359 could not take place because the system was powered off.
6360 This enables simple anacron-like behaviour for timer units.
6362 * systemctl's "list-timers" will now also list the time a
6363 timer unit was last triggered in addition to the next time
6364 it will be triggered.
6366 * systemd-networkd will now assign predictable IPv4LL
6367 addresses to its local interfaces.
6369 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
6370 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
6371 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
6372 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
6373 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
6374 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
6375 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
6376 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
6379 — Berlin, 2014-03-25
6383 * A new unit file setting RestrictAddressFamilies= has been
6384 added to restrict which socket address families unit
6385 processes gain access to. This takes address family names
6386 like "AF_INET" or "AF_UNIX", and is useful to minimize the
6387 attack surface of services via exotic protocol stacks. This
6388 is built on seccomp system call filters.
6390 * Two new unit file settings RuntimeDirectory= and
6391 RuntimeDirectoryMode= have been added that may be used to
6392 manage a per-daemon runtime directories below /run. This is
6393 an alternative for setting up directory permissions with
6394 tmpfiles snippets, and has the advantage that the runtime
6395 directory's lifetime is bound to the daemon runtime and that
6396 the daemon starts up with an empty directory each time. This
6397 is particularly useful when writing services that drop
6398 privileges using the User= or Group= setting.
6400 * The DeviceAllow= unit setting now supports globbing for
6401 matching against device group names.
6403 * The systemd configuration file system.conf gained new
6404 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
6405 DefaultMemoryAccounting= to globally turn on/off accounting
6406 for specific resources (cgroups) for all units. These
6407 settings may still be overridden individually in each unit
6410 * systemd-gpt-auto-generator is now able to discover /srv and
6411 root partitions in addition to /home and swap partitions. It
6412 also supports LUKS-encrypted partitions now. With this in
6413 place, automatic discovery of partitions to mount following
6414 the Discoverable Partitions Specification
6415 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
6416 is now a lot more complete. This allows booting without
6417 /etc/fstab and without root= on the kernel command line on
6418 systems prepared appropriately.
6420 * systemd-nspawn gained a new --image= switch which allows
6421 booting up disk images and Linux installations on any block
6422 device that follow the Discoverable Partitions Specification
6423 (see above). This means that installations made with
6424 appropriately updated installers may now be started and
6425 deployed using container managers, completely
6426 unmodified. (We hope that libvirt-lxc will add support for
6427 this feature soon, too.)
6429 * systemd-nspawn gained a new --network-macvlan= setting to
6430 set up a private macvlan interface for the
6431 container. Similarly, systemd-networkd gained a new
6432 Kind=macvlan setting in .netdev files.
6434 * systemd-networkd now supports configuring local addresses
6437 * A new tool systemd-network-wait-online has been added to
6438 synchronously wait for network connectivity using
6441 * The sd-bus.h bus API gained a new sd_bus_track object for
6442 tracking the lifecycle of bus peers. Note that sd-bus.h is
6443 still not a public API though (unless you specify
6444 --enable-kdbus on the configure command line, which however
6445 voids your warranty and you get no API stability guarantee).
6447 * The $XDG_RUNTIME_DIR runtime directories for each user are
6448 now individual tmpfs instances, which has the benefit of
6449 introducing separate pools for each user, with individual
6450 size limits, and thus making sure that unprivileged clients
6451 can no longer negatively impact the system or other users by
6452 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
6453 RuntimeDirectorySize= has been introduced that allows
6454 controlling the default size limit for all users. It
6455 defaults to 10% of the available physical memory. This is no
6456 replacement for quotas on tmpfs though (which the kernel
6457 still does not support), as /dev/shm and /tmp are still
6458 shared resources used by both the system and unprivileged
6461 * logind will now automatically turn off automatic suspending
6462 on laptop lid close when more than one display is
6463 connected. This was previously expected to be implemented
6464 individually in desktop environments (such as GNOME),
6465 however has been added to logind now, in order to fix a
6466 boot-time race where a desktop environment might not have
6467 been started yet and thus not been able to take an inhibitor
6468 lock at the time where logind already suspends the system
6469 due to a closed lid.
6471 * logind will now wait at least 30s after each system
6472 suspend/resume cycle, and 3min after system boot before
6473 suspending the system due to a closed laptop lid. This
6474 should give USB docking stations and similar enough time to
6475 be probed and configured after system resume and boot in
6476 order to then act as suspend blocker.
6478 * systemd-run gained a new --property= setting which allows
6479 initialization of resource control properties (and others)
6480 for the created scope or service unit. Example: "systemd-run
6481 --property=BlockIOWeight=10 updatedb" may be used to run
6482 updatedb at a low block IO scheduling weight.
6484 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
6485 now also work in --scope mode.
6487 * When systemd is compiled with kdbus support, basic support
6488 for enforced policies is now in place. (Note that enabling
6489 kdbus still voids your warranty and no API compatibility
6492 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
6493 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
6494 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
6495 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
6496 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
6497 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
6498 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
6499 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
6500 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
6501 Zbigniew Jędrzejewski-Szmek
6503 — Berlin, 2014-03-12
6507 * systemd will now relabel /dev after loading the SMACK policy
6508 according to SMACK rules.
6510 * A new unit file option AppArmorProfile= has been added to
6511 set the AppArmor profile for the processes of a unit.
6513 * A new condition check ConditionArchitecture= has been added
6514 to conditionalize units based on the system architecture, as
6515 reported by uname()'s "machine" field.
6517 * systemd-networkd now supports matching on the system
6518 virtualization, architecture, kernel command line, host name
6521 * logind is now a lot more aggressive when suspending the
6522 machine due to a closed laptop lid. Instead of acting only
6523 on the lid close action, it will continuously watch the lid
6524 status and act on it. This is useful for laptops where the
6525 power button is on the outside of the chassis so that it can
6526 be reached without opening the lid (such as the Lenovo
6527 Yoga). On those machines, logind will now immediately
6528 re-suspend the machine if the power button has been
6529 accidentally pressed while the laptop was suspended and in a
6530 backpack or similar.
6532 * logind will now watch SW_DOCK switches and inhibit reaction
6533 to the lid switch if it is pressed. This means that logind
6534 will not suspend the machine anymore if the lid is closed
6535 and the system is docked, if the laptop supports SW_DOCK
6536 notifications via the input layer. Note that ACPI docking
6537 stations do not generate this currently. Also note that this
6538 logic is usually not fully sufficient and Desktop
6539 Environments should take a lid switch inhibitor lock when an
6540 external display is connected, as systemd will not watch
6543 * nspawn will now make use of the devices cgroup controller by
6544 default, and only permit creation of and access to the usual
6545 API device nodes like /dev/null or /dev/random, as well as
6546 access to (but not creation of) the pty devices.
6548 * We will now ship a default .network file for
6549 systemd-networkd that automatically configures DHCP for
6550 network interfaces created by nspawn's --network-veth or
6551 --network-bridge= switches.
6553 * systemd will now understand the usual M, K, G, T suffixes
6554 according to SI conventions (i.e. to the base 1000) when
6555 referring to throughput and hardware metrics. It will stay
6556 with IEC conventions (i.e. to the base 1024) for software
6557 metrics, according to what is customary according to
6558 Wikipedia. We explicitly document which base applies for
6559 each configuration option.
6561 * The DeviceAllow= setting in unit files now supports a syntax
6562 to whitelist an entire group of devices node majors at once,
6563 based on the /proc/devices listing. For example, with the
6564 string "char-pts", it is now possible to whitelist all
6565 current and future pseudo-TTYs at once.
6567 * sd-event learned a new "post" event source. Event sources of
6568 this type are triggered by the dispatching of any event
6569 source of a type that is not "post". This is useful for
6570 implementing clean-up and check event sources that are
6571 triggered by other work being done in the program.
6573 * systemd-networkd is no longer statically enabled, but uses
6574 the usual [Install] sections so that it can be
6575 enabled/disabled using systemctl. It still is enabled by
6578 * When creating a veth interface pair with systemd-nspawn, the
6579 host side will now be prefixed with "vb-" if
6580 --network-bridge= is used, and with "ve-" if --network-veth
6581 is used. This way, it is easy to distinguish these cases on
6582 the host, for example to apply different configuration to
6583 them with systemd-networkd.
6585 * The compatibility libraries for libsystemd-journal.so,
6586 libsystem-id128.so, libsystemd-login.so and
6587 libsystemd-daemon.so do not make use of IFUNC
6588 anymore. Instead, we now build libsystemd.so multiple times
6589 under these alternative names. This means that the footprint
6590 is drastically increased, but given that these are
6591 transitional compatibility libraries, this should not matter
6592 much. This change has been made necessary to support the ARM
6593 platform for these compatibility libraries, as the ARM
6594 toolchain is not really at the same level as the toolchain
6595 for other architectures like x86 and does not support
6596 IFUNC. Please make sure to use --enable-compat-libs only
6597 during a transitional period!
6599 * The .include syntax has been deprecated and is not documented
6600 anymore. Drop-in files in .d directories should be used instead.
6602 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
6603 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
6604 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
6605 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
6606 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
6607 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
6608 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
6609 Zbigniew Jędrzejewski-Szmek
6611 — Berlin, 2014-02-24
6615 * A new component "systemd-networkd" has been added that can
6616 be used to configure local network interfaces statically or
6617 via DHCP. It is capable of bringing up bridges, VLANs, and
6618 bonding. Currently, no hook-ups for interactive network
6619 configuration are provided. Use this for your initrd,
6620 container, embedded, or server setup if you need a simple,
6621 yet powerful, network configuration solution. This
6622 configuration subsystem is quite nifty, as it allows wildcard
6623 hotplug matching in interfaces. For example, with a single
6624 configuration snippet, you can configure that all Ethernet
6625 interfaces showing up are automatically added to a bridge,
6626 or similar. It supports link-sensing and more.
6628 * A new tool "systemd-socket-proxyd" has been added which can
6629 act as a bidirectional proxy for TCP sockets. This is
6630 useful for adding socket activation support to services that
6631 do not actually support socket activation, including virtual
6632 machines and the like.
6634 * Add a new tool to save/restore rfkill state on
6637 * Save/restore state of keyboard backlights in addition to
6638 display backlights on shutdown/boot.
6640 * udev learned a new SECLABEL{} construct to label device
6641 nodes with a specific security label when they appear. For
6642 now, only SECLABEL{selinux} is supported, but the syntax is
6643 prepared for additional security frameworks.
6645 * udev gained a new scheme to configure link-level attributes
6646 from files in /etc/systemd/network/*.link. These files can
6647 match against MAC address, device path, driver name and type,
6648 and will apply attributes like the naming policy, link speed,
6649 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
6650 address assignment policy (randomized, ...).
6652 * The configuration of network interface naming rules for
6653 "permanent interface names" has changed: a new NamePolicy=
6654 setting in the [Link] section of .link files determines the
6655 priority of possible naming schemes (onboard, slot, MAC,
6656 path). The default value of this setting is determined by
6657 /usr/lib/net/links/99-default.link. Old
6658 80-net-name-slot.rules udev configuration file has been
6659 removed, so local configuration overriding this file should
6660 be adapted to override 99-default.link instead.
6662 * When the User= switch is used in a unit file, also
6663 initialize $SHELL= based on the user database entry.
6665 * systemd no longer depends on libdbus. All communication is
6666 now done with sd-bus, systemd's low-level bus library
6669 * kdbus support has been added to PID 1 itself. When kdbus is
6670 enabled, this causes PID 1 to set up the system bus and
6671 enable support for a new ".busname" unit type that
6672 encapsulates bus name activation on kdbus. It works a little
6673 bit like ".socket" units, except for bus names. A new
6674 generator has been added that converts classic dbus1 service
6675 activation files automatically into native systemd .busname
6678 * sd-bus: add a light-weight vtable implementation that allows
6679 defining objects on the bus with a simple static const
6680 vtable array of its methods, signals and properties.
6682 * systemd will not generate or install static dbus
6683 introspection data anymore to /usr/share/dbus-1/interfaces,
6684 as the precise format of these files is unclear, and
6685 nothing makes use of it.
6687 * A proxy daemon is now provided to proxy clients connecting
6688 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
6689 compatibility with classic D-Bus.
6691 * A bus driver implementation has been added that supports the
6692 classic D-Bus bus driver calls on kdbus, also for
6693 compatibility purposes.
6695 * A new API "sd-event.h" has been added that implements a
6696 minimal event loop API built around epoll. It provides a
6697 couple of features that direct epoll usage is lacking:
6698 prioritization of events, scales to large numbers of timer
6699 events, per-event timer slack (accuracy), system-wide
6700 coalescing of timer events, exit handlers, watchdog
6701 supervision support using systemd's sd_notify() API, child
6704 * A new API "sd-rntl.h" has been added that provides an API
6705 around the route netlink interface of the kernel, similar in
6706 style to "sd-bus.h".
6708 * A new API "sd-dhcp-client.h" has been added that provides a
6709 small DHCPv4 client-side implementation. This is used by
6712 * There is a new kernel command line option
6713 "systemd.restore_state=0|1". When set to "0", none of the
6714 systemd tools will restore saved runtime state to hardware
6715 devices. More specifically, the rfkill and backlight states
6718 * The FsckPassNo= compatibility option in mount/service units
6719 has been removed. The fstab generator will now add the
6720 necessary dependencies automatically, and does not require
6721 PID1's support for that anymore.
6723 * journalctl gained a new switch, --list-boots, that lists
6724 recent boots with their times and boot IDs.
6726 * The various tools like systemctl, loginctl, timedatectl,
6727 busctl, systemd-run, ... have gained a new switch "-M" to
6728 connect to a specific, local OS container (as direct
6729 connection, without requiring SSH). This works on any
6730 container that is registered with machined, such as those
6731 created by libvirt-lxc or nspawn.
6733 * systemd-run and systemd-analyze also gained support for "-H"
6734 to connect to remote hosts via SSH. This is particularly
6735 useful for systemd-run because it enables queuing of jobs
6736 onto remote systems.
6738 * machinectl gained a new command "login" to open a getty
6739 login in any local container. This works with any container
6740 that is registered with machined (such as those created by
6741 libvirt-lxc or nspawn), and which runs systemd inside.
6743 * machinectl gained a new "reboot" command that may be used to
6744 trigger a reboot on a specific container that is registered
6745 with machined. This works on any container that runs an init
6746 system of some kind.
6748 * systemctl gained a new "list-timers" command to print a nice
6749 listing of installed timer units with the times they elapse
6752 * Alternative reboot() parameters may now be specified on the
6753 "systemctl reboot" command line and are passed to the
6754 reboot() system call.
6756 * systemctl gained a new --job-mode= switch to configure the
6757 mode to queue a job with. This is a more generic version of
6758 --fail, --irreversible, and --ignore-dependencies, which are
6759 still available but not advertised anymore.
6761 * /etc/systemd/system.conf gained new settings to configure
6762 various default timeouts of units, as well as the default
6763 start limit interval and burst. These may still be overridden
6766 * PID1 will now export on the bus profile data of the security
6767 policy upload process (such as the SELinux policy upload to
6770 * journald: when forwarding logs to the console, include
6771 timestamps (following the setting in
6772 /sys/module/printk/parameters/time).
6774 * OnCalendar= in timer units now understands the special
6775 strings "yearly" and "annually". (Both are equivalent)
6777 * The accuracy of timer units is now configurable with the new
6778 AccuracySec= setting. It defaults to 1min.
6780 * A new dependency type JoinsNamespaceOf= has been added that
6781 allows running two services within the same /tmp and network
6782 namespace, if PrivateNetwork= or PrivateTmp= are used.
6784 * A new command "cat" has been added to systemctl. It outputs
6785 the original unit file of a unit, and concatenates the
6786 contents of additional "drop-in" unit file snippets, so that
6787 the full configuration is shown.
6789 * systemctl now supports globbing on the various "list-xyz"
6790 commands, like "list-units" or "list-sockets", as well as on
6791 those commands which take multiple unit names.
6793 * journalctl's --unit= switch gained support for globbing.
6795 * All systemd daemons now make use of the watchdog logic so
6796 that systemd automatically notices when they hang.
6798 * If the $container_ttys environment variable is set,
6799 getty-generator will automatically spawn a getty for each
6800 listed tty. This is useful for container managers to request
6801 login gettys to be spawned on as many ttys as needed.
6803 * %h, %s, %U specifier support is not available anymore when
6804 used in unit files for PID 1. This is because NSS calls are
6805 not safe from PID 1. They stay available for --user
6806 instances of systemd, and as special case for the root user.
6808 * loginctl gained a new "--no-legend" switch to turn off output
6811 * The "sd-login.h" API gained three new calls:
6812 sd_session_is_remote(), sd_session_get_remote_user(),
6813 sd_session_get_remote_host() to query information about
6816 * The udev hardware database now also carries vendor/product
6817 information of SDIO devices.
6819 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
6820 determine whether watchdog notifications are requested by
6823 * Socket-activated per-connection services now include a
6824 short description of the connection parameters in the
6827 * tmpfiles gained a new "--boot" option. When this is not used,
6828 only lines where the command character is not suffixed with
6829 "!" are executed. When this option is specified, those
6830 options are executed too. This partitions tmpfiles
6831 directives into those that can be safely executed at any
6832 time, and those which should be run only at boot (for
6833 example, a line that creates /run/nologin).
6835 * A new API "sd-resolve.h" has been added which provides a simple
6836 asynchronous wrapper around glibc NSS host name resolution
6837 calls, such as getaddrinfo(). In contrast to glibc's
6838 getaddrinfo_a(), it does not use signals. In contrast to most
6839 other asynchronous name resolution libraries, this one does
6840 not reimplement DNS, but reuses NSS, so that alternate
6841 host name resolution systems continue to work, such as mDNS,
6842 LDAP, etc. This API is based on libasyncns, but it has been
6843 cleaned up for inclusion in systemd.
6845 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
6846 "sd-daemon.h" are no longer found in individual libraries
6847 libsystemd-journal.so, libsystemd-login.so,
6848 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
6849 merged them into a single library, libsystemd.so, which
6850 provides all symbols. The reason for this is cyclic
6851 dependencies, as these libraries tend to use each other's
6852 symbols. So far, we have managed to workaround that by linking
6853 a copy of a good part of our code into each of these
6854 libraries again and again, which, however, makes certain
6855 things hard to do, like sharing static variables. Also, it
6856 substantially increases footprint. With this change, there
6857 is only one library for the basic APIs systemd
6858 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
6859 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
6860 library as well, however are subject to the --enable-kdbus
6861 switch (see below). Note that "sd-dhcp-client.h" is not part
6862 of this library (this is because it only consumes, never
6863 provides, services of/to other APIs). To make the transition
6864 easy from the separate libraries to the unified one, we
6865 provide the --enable-compat-libs compile-time switch which
6866 will generate stub libraries that are compatible with the
6867 old ones but redirect all calls to the new one.
6869 * All of the kdbus logic and the new APIs "sd-bus.h",
6870 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
6871 and "sd-utf8.h" are compile-time optional via the
6872 "--enable-kdbus" switch, and they are not compiled in by
6873 default. To make use of kdbus, you have to explicitly enable
6874 the switch. Note however, that neither the kernel nor the
6875 userspace API for all of this is considered stable yet. We
6876 want to maintain the freedom to still change the APIs for
6877 now. By specifying this build-time switch, you acknowledge
6878 that you are aware of the instability of the current
6881 * Also, note that while kdbus is pretty much complete,
6882 it lacks one thing: proper policy support. This means you
6883 can build a fully working system with all features; however,
6884 it will be highly insecure. Policy support will be added in
6885 one of the next releases, at the same time that we will
6886 declare the APIs stable.
6888 * When the kernel command line argument "kdbus" is specified,
6889 systemd will automatically load the kdbus.ko kernel module. At
6890 this stage of development, it is only useful for testing kdbus
6891 and should not be used in production. Note: if "--enable-kdbus"
6892 is specified, and the kdbus.ko kernel module is available, and
6893 "kdbus" is added to the kernel command line, the entire system
6894 runs with kdbus instead of dbus-daemon, with the above mentioned
6895 problem of missing the system policy enforcement. Also a future
6896 version of kdbus.ko or a newer systemd will not be compatible with
6897 each other, and will unlikely be able to boot the machine if only
6898 one of them is updated.
6900 * systemctl gained a new "import-environment" command which
6901 uploads the caller's environment (or parts thereof) into the
6902 service manager so that it is inherited by services started
6903 by the manager. This is useful to upload variables like
6904 $DISPLAY into the user service manager.
6906 * A new PrivateDevices= switch has been added to service units
6907 which allows running a service with a namespaced /dev
6908 directory that does not contain any device nodes for
6909 physical devices. More specifically, it only includes devices
6910 such as /dev/null, /dev/urandom, and /dev/zero which are API
6913 * logind has been extended to support behaviour like VT
6914 switching on seats that do not support a VT. This makes
6915 multi-session available on seats that are not the first seat
6916 (seat0), and on systems where kernel support for VTs has
6917 been disabled at compile-time.
6919 * If a process holds a delay lock for system sleep or shutdown
6920 and fails to release it in time, we will now log its
6921 identity. This makes it easier to identify processes that
6922 cause slow suspends or power-offs.
6924 * When parsing /etc/crypttab, support for a new key-slot=
6925 option as supported by Debian is added. It allows indicating
6926 which LUKS slot to use on disk, speeding up key loading.
6928 * The sd_journal_sendv() API call has been checked and
6929 officially declared to be async-signal-safe so that it may
6930 be invoked from signal handlers for logging purposes.
6932 * Boot-time status output is now enabled automatically after a
6933 short timeout if boot does not progress, in order to give
6934 the user an indication what she or he is waiting for.
6936 * The boot-time output has been improved to show how much time
6937 remains until jobs expire.
6939 * The KillMode= switch in service units gained a new possible
6940 value "mixed". If set, and the unit is shut down, then the
6941 initial SIGTERM signal is sent only to the main daemon
6942 process, while the following SIGKILL signal is sent to
6943 all remaining processes of the service.
6945 * When a scope unit is registered, a new property "Controller"
6946 may be set. If set to a valid bus name, systemd will send a
6947 RequestStop() signal to this name when it would like to shut
6948 down the scope. This may be used to hook manager logic into
6949 the shutdown logic of scope units. Also, scope units may now
6950 be put in a special "abandoned" state, in which case the
6951 manager process which created them takes no further
6952 responsibilities for it.
6954 * When reading unit files, systemd will now verify
6955 the access mode of these files, and warn about certain
6956 suspicious combinations. This has been added to make it
6957 easier to track down packaging bugs where unit files are
6958 marked executable or world-writable.
6960 * systemd-nspawn gained a new "--setenv=" switch to set
6961 container-wide environment variables. The similar option in
6962 systemd-activate was renamed from "--environment=" to
6963 "--setenv=" for consistency.
6965 * systemd-nspawn has been updated to create a new kdbus domain
6966 for each container that is invoked, thus allowing each
6967 container to have its own set of system and user buses,
6968 independent of the host.
6970 * systemd-nspawn gained a new --drop-capability= switch to run
6971 the container with less capabilities than the default. Both
6972 --drop-capability= and --capability= now take the special
6973 string "all" for dropping or keeping all capabilities.
6975 * systemd-nspawn gained new switches for executing containers
6976 with specific SELinux labels set.
6978 * systemd-nspawn gained a new --quiet switch to not generate
6979 any additional output but the container's own console
6982 * systemd-nspawn gained a new --share-system switch to run a
6983 container without PID namespacing enabled.
6985 * systemd-nspawn gained a new --register= switch to control
6986 whether the container is registered with systemd-machined or
6987 not. This is useful for containers that do not run full
6988 OS images, but only specific apps.
6990 * systemd-nspawn gained a new --keep-unit which may be used
6991 when invoked as the only program from a service unit, and
6992 results in registration of the unit service itself in
6993 systemd-machined, instead of a newly opened scope unit.
6995 * systemd-nspawn gained a new --network-interface= switch for
6996 moving arbitrary interfaces to the container. The new
6997 --network-veth switch creates a virtual Ethernet connection
6998 between host and container. The new --network-bridge=
6999 switch then allows assigning the host side of this virtual
7000 Ethernet connection to a bridge device.
7002 * systemd-nspawn gained a new --personality= switch for
7003 setting the kernel personality for the container. This is
7004 useful when running a 32-bit container on a 64-bit host. A
7005 similar option Personality= is now also available for service
7008 * logind will now also track a "Desktop" identifier for each
7009 session which encodes the desktop environment of it. This is
7010 useful for desktop environments that want to identify
7011 multiple running sessions of itself easily.
7013 * A new SELinuxContext= setting for service units has been
7014 added that allows setting a specific SELinux execution
7015 context for a service.
7017 * Most systemd client tools will now honour $SYSTEMD_LESS for
7018 settings of the "less" pager. By default, these tools will
7019 override $LESS to allow certain operations to work, such as
7020 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
7021 influence this logic.
7023 * systemd's "seccomp" hook-up has been changed to make use of
7024 the libseccomp library instead of using its own
7025 implementation. This has benefits for portability among
7028 * For usage together with SystemCallFilter=, a new
7029 SystemCallErrorNumber= setting has been introduced that
7030 allows configuration of a system error number to be returned
7031 on filtered system calls, instead of immediately killing the
7032 process. Also, SystemCallArchitectures= has been added to
7033 limit access to system calls of a particular architecture
7034 (in order to turn off support for unused secondary
7035 architectures). There is also a global
7036 SystemCallArchitectures= setting in system.conf now to turn
7037 off support for non-native system calls system-wide.
7039 * systemd requires a kernel with a working name_to_handle_at(),
7040 please see the kernel config requirements in the README file.
7042 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
7043 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
7044 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
7045 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
7046 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
7047 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
7048 Elia Pinto, Florian Weimer, George McCollister, Goffredo
7049 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
7050 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
7051 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
7052 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
7053 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
7054 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
7055 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
7056 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
7057 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
7058 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
7059 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
7060 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
7061 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
7062 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
7063 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
7064 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
7065 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
7067 — Berlin, 2014-02-20
7071 * logind has gained support for facilitating privileged input
7072 and drm device access for unprivileged clients. This work is
7073 useful to allow Wayland display servers (and similar
7074 programs, such as kmscon) to run under the user's ID and
7075 access input and drm devices which are normally
7076 protected. When this is used (and the kernel is new enough)
7077 logind will "mute" IO on the file descriptors passed to
7078 Wayland as long as it is in the background and "unmute" it
7079 if it returns into the foreground. This allows secure
7080 session switching without allowing background sessions to
7081 eavesdrop on input and display data. This also introduces
7082 session switching support if VT support is turned off in the
7083 kernel, and on seats that are not seat0.
7085 * A new kernel command line option luks.options= is understood
7086 now which allows specifying LUKS options for usage for LUKS
7087 encrypted partitions specified with luks.uuid=.
7089 * tmpfiles.d(5) snippets may now use specifier expansion in
7090 path names. More specifically %m, %b, %H, %v, are now
7091 replaced by the local machine id, boot id, hostname, and
7092 kernel version number.
7094 * A new tmpfiles.d(5) command "m" has been introduced which
7095 may be used to change the owner/group/access mode of a file
7096 or directory if it exists, but do nothing if it does not.
7098 * This release removes high-level support for the
7099 MemorySoftLimit= cgroup setting. The underlying kernel
7100 cgroup attribute memory.soft_limit= is currently badly
7101 designed and likely to be removed from the kernel API in its
7102 current form, hence we should not expose it for now.
7104 * The memory.use_hierarchy cgroup attribute is now enabled for
7105 all cgroups systemd creates in the memory cgroup
7106 hierarchy. This option is likely to be come the built-in
7107 default in the kernel anyway, and the non-hierarchical mode
7108 never made much sense in the intrinsically hierarchical
7111 * A new field _SYSTEMD_SLICE= is logged along with all journal
7112 messages containing the slice a message was generated
7113 from. This is useful to allow easy per-customer filtering of
7114 logs among other things.
7116 * systemd-journald will no longer adjust the group of journal
7117 files it creates to the "systemd-journal" group. Instead we
7118 rely on the journal directory to be owned by the
7119 "systemd-journal" group, and its setgid bit set, so that the
7120 kernel file system layer will automatically enforce that
7121 journal files inherit this group assignment. The reason for
7122 this change is that we cannot allow NSS look-ups from
7123 journald which would be necessary to resolve
7124 "systemd-journal" to a numeric GID, because this might
7125 create deadlocks if NSS involves synchronous queries to
7126 other daemons (such as nscd, or sssd) which in turn are
7127 logging clients of journald and might block on it, which
7128 would then dead lock. A tmpfiles.d(5) snippet included in
7129 systemd will make sure the setgid bit and group are
7130 properly set on the journal directory if it exists on every
7131 boot. However, we recommend adjusting it manually after
7132 upgrades too (or from RPM scriptlets), so that the change is
7133 not delayed until next reboot.
7135 * Backlight and random seed files in /var/lib/ have moved into
7136 the /var/lib/systemd/ directory, in order to centralize all
7137 systemd generated files in one directory.
7139 * Boot time performance measurements (as displayed by
7140 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
7141 performance information if that's available to determine how
7142 much time BIOS and boot loader initialization required. With
7143 a sufficiently new BIOS you hence no longer need to boot
7144 with Gummiboot to get access to such information.
7146 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
7147 Cristian Rodríguez, Dave Reisner, David Herrmann, David
7148 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
7149 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
7150 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
7151 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
7152 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7154 — Berlin, 2013-10-02
7158 * The Restart= option for services now understands a new
7159 on-watchdog setting, which will restart the service
7160 automatically if the service stops sending out watchdog keep
7161 alive messages (as configured with WatchdogSec=).
7163 * The getty generator (which is responsible for bringing up a
7164 getty on configured serial consoles) will no longer only
7165 start a getty on the primary kernel console but on all
7166 others, too. This makes the order in which console= is
7167 specified on the kernel command line less important.
7169 * libsystemd-logind gained a new sd_session_get_vt() call to
7170 retrieve the VT number of a session.
7172 * If the option "tries=0" is set for an entry of /etc/crypttab
7173 its passphrase is queried indefinitely instead of any
7174 maximum number of tries.
7176 * If a service with a configure PID file terminates its PID
7177 file will now be removed automatically if it still exists
7178 afterwards. This should put an end to stale PID files.
7180 * systemd-run will now also take relative binary path names
7181 for execution and no longer insists on absolute paths.
7183 * InaccessibleDirectories= and ReadOnlyDirectories= now take
7184 paths that are optionally prefixed with "-" to indicate that
7185 it should not be considered a failure if they do not exist.
7187 * journalctl -o (and similar commands) now understands a new
7188 output mode "short-precise", it is similar to "short" but
7189 shows timestamps with usec accuracy.
7191 * The option "discard" (as known from Debian) is now
7192 synonymous to "allow-discards" in /etc/crypttab. In fact,
7193 "discard" is preferred now (since it is easier to remember
7196 * Some licensing clean-ups were made, so that more code is now
7197 LGPL-2.1 licensed than before.
7199 * A minimal tool to save/restore the display backlight
7200 brightness across reboots has been added. It will store the
7201 backlight setting as late as possible at shutdown, and
7202 restore it as early as possible during reboot.
7204 * A logic to automatically discover and enable home and swap
7205 partitions on GPT disks has been added. With this in place
7206 /etc/fstab becomes optional for many setups as systemd can
7207 discover certain partitions located on the root disk
7208 automatically. Home partitions are recognized under their
7209 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
7210 partitions are recognized under their GPT type ID
7211 0657fd6da4ab43c484e50933c84b4f4f.
7213 * systemd will no longer pass any environment from the kernel
7214 or initrd to system services. If you want to set an
7215 environment for all services, do so via the kernel command
7216 line systemd.setenv= assignment.
7218 * The systemd-sysctl tool no longer natively reads the file
7219 /etc/sysctl.conf. If desired, the file should be symlinked
7220 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
7221 legacy support by a symlink rather than built-in code, it
7222 also makes the otherwise hidden order of application of the
7223 different files visible. (Note that this partly reverts to a
7224 pre-198 application order of sysctl knobs!)
7226 * The "systemctl set-log-level" and "systemctl dump" commands
7227 have been moved to systemd-analyze.
7229 * systemd-run learned the new --remain-after-exit switch,
7230 which causes the scope unit not to be cleaned up
7231 automatically after the process terminated.
7233 * tmpfiles learned a new --exclude-prefix= switch to exclude
7234 certain paths from operation.
7236 * journald will now automatically flush all messages to disk
7237 as soon as a message at the log level CRIT, ALERT or EMERG
7240 Contributions from: Andrew Cook, Brandon Philips, Christian
7241 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
7242 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
7243 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
7244 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
7245 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
7246 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
7247 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
7248 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
7249 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
7250 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
7251 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
7252 William Giokas, Zbigniew Jędrzejewski-Szmek
7254 — Berlin, 2013-09-13
7258 * The documentation has been updated to cover the various new
7259 concepts introduced with 205.
7261 * Unit files now understand the new %v specifier which
7262 resolves to the kernel version string as returned by "uname
7265 * systemctl now supports filtering the unit list output by
7266 load state, active state and sub state, using the new
7269 * "systemctl status" will now show the results of the
7270 condition checks (like ConditionPathExists= and similar) of
7271 the last start attempts of the unit. They are also logged to
7274 * "journalctl -b" may now be used to look for boot output of a
7275 specific boot. Try "journalctl -b -1" for the previous boot,
7276 but the syntax is substantially more powerful.
7278 * "journalctl --show-cursor" has been added which prints the
7279 cursor string the last shown log line. This may then be used
7280 with the new "journalctl --after-cursor=" switch to continue
7281 browsing logs from that point on.
7283 * "journalctl --force" may now be used to force regeneration
7286 * Creation of "dead" device nodes has been moved from udev
7287 into kmod and tmpfiles. Previously, udev would read the kmod
7288 databases to pre-generate dead device nodes based on meta
7289 information contained in kernel modules, so that these would
7290 be auto-loaded on access rather then at boot. As this
7291 does not really have much to do with the exposing actual
7292 kernel devices to userspace this has always been slightly
7293 alien in the udev codebase. Following the new scheme kmod
7294 will now generate a runtime snippet for tmpfiles from the
7295 module meta information and it now is tmpfiles' job to the
7296 create the nodes. This also allows overriding access and
7297 other parameters for the nodes using the usual tmpfiles
7298 facilities. As side effect this allows us to remove the
7299 CAP_SYS_MKNOD capability bit from udevd entirely.
7301 * logind's device ACLs may now be applied to these "dead"
7302 devices nodes too, thus finally allowing managed access to
7303 devices such as /dev/snd/sequencer without loading the
7304 backing module right-away.
7306 * A new RPM macro has been added that may be used to apply
7307 tmpfiles configuration during package installation.
7309 * systemd-detect-virt and ConditionVirtualization= now can
7310 detect User-Mode-Linux machines (UML).
7312 * journald will now implicitly log the effective capabilities
7313 set of processes in the message metadata.
7315 * systemd-cryptsetup has gained support for TrueCrypt volumes.
7317 * The initrd interface has been simplified (more specifically,
7318 support for passing performance data via environment
7319 variables and fsck results via files in /run has been
7320 removed). These features were non-essential, and are
7321 nowadays available in a much nicer way by having systemd in
7322 the initrd serialize its state and have the hosts systemd
7323 deserialize it again.
7325 * The udev "keymap" data files and tools to apply keyboard
7326 specific mappings of scan to key codes, and force-release
7327 scan code lists have been entirely replaced by a udev
7328 "keyboard" builtin and a hwdb data file.
7330 * systemd will now honour the kernel's "quiet" command line
7331 argument also during late shutdown, resulting in a
7332 completely silent shutdown when used.
7334 * There's now an option to control the SO_REUSEPORT socket
7335 option in .socket units.
7337 * Instance units will now automatically get a per-template
7338 subslice of system.slice unless something else is explicitly
7339 configured. For example, instances of sshd@.service will now
7340 implicitly be placed in system-sshd.slice rather than
7341 system.slice as before.
7343 * Test coverage support may now be enabled at build time.
7345 Contributions from: Dave Reisner, Frederic Crozat, Harald
7346 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
7347 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
7348 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
7349 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
7350 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
7351 Giokas, Zbigniew Jędrzejewski-Szmek
7353 — Berlin, 2013-07-23
7357 * Two new unit types have been introduced:
7359 Scope units are very similar to service units, however, are
7360 created out of pre-existing processes — instead of PID 1
7361 forking off the processes. By using scope units it is
7362 possible for system services and applications to group their
7363 own child processes (worker processes) in a powerful way
7364 which then maybe used to organize them, or kill them
7365 together, or apply resource limits on them.
7367 Slice units may be used to partition system resources in an
7368 hierarchical fashion and then assign other units to them. By
7369 default there are now three slices: system.slice (for all
7370 system services), user.slice (for all user sessions),
7371 machine.slice (for VMs and containers).
7373 Slices and scopes have been introduced primarily in
7374 context of the work to move cgroup handling to a
7375 single-writer scheme, where only PID 1
7376 creates/removes/manages cgroups.
7378 * There's a new concept of "transient" units. In contrast to
7379 normal units these units are created via an API at runtime,
7380 not from configuration from disk. More specifically this
7381 means it is now possible to run arbitrary programs as
7382 independent services, with all execution parameters passed
7383 in via bus APIs rather than read from disk. Transient units
7384 make systemd substantially more dynamic then it ever was,
7385 and useful as a general batch manager.
7387 * logind has been updated to make use of scope and slice units
7388 for managing user sessions. As a user logs in he will get
7389 his own private slice unit, to which all sessions are added
7390 as scope units. We also added support for automatically
7391 adding an instance of user@.service for the user into the
7392 slice. Effectively logind will no longer create cgroup
7393 hierarchies on its own now, it will defer entirely to PID 1
7394 for this by means of scope, service and slice units. Since
7395 user sessions this way become entities managed by PID 1
7396 the output of "systemctl" is now a lot more comprehensive.
7398 * A new mini-daemon "systemd-machined" has been added which
7399 may be used by virtualization managers to register local
7400 VMs/containers. nspawn has been updated accordingly, and
7401 libvirt will be updated shortly. machined will collect a bit
7402 of meta information about the VMs/containers, and assign
7403 them their own scope unit (see above). The collected
7404 meta-data is then made available via the "machinectl" tool,
7405 and exposed in "ps" and similar tools. machined/machinectl
7406 is compile-time optional.
7408 * As discussed earlier, the low-level cgroup configuration
7409 options ControlGroup=, ControlGroupModify=,
7410 ControlGroupPersistent=, ControlGroupAttribute= have been
7411 removed. Please use high-level attribute settings instead as
7412 well as slice units.
7414 * A new bus call SetUnitProperties() has been added to alter
7415 various runtime parameters of a unit. This is primarily
7416 useful to alter cgroup parameters dynamically in a nice way,
7417 but will be extended later on to make more properties
7418 modifiable at runtime. systemctl gained a new set-properties
7419 command that wraps this call.
7421 * A new tool "systemd-run" has been added which can be used to
7422 run arbitrary command lines as transient services or scopes,
7423 while configuring a number of settings via the command
7424 line. This tool is currently very basic, however already
7425 very useful. We plan to extend this tool to even allow
7426 queuing of execution jobs with time triggers from the
7427 command line, similar in fashion to "at".
7429 * nspawn will now inform the user explicitly that kernels with
7430 audit enabled break containers, and suggest the user to turn
7433 * Support for detecting the IMA and AppArmor security
7434 frameworks with ConditionSecurity= has been added.
7436 * journalctl gained a new "-k" switch for showing only kernel
7437 messages, mimicking dmesg output; in addition to "--user"
7438 and "--system" switches for showing only user's own logs
7441 * systemd-delta can now show information about drop-in
7442 snippets extending unit files.
7444 * libsystemd-bus has been substantially updated but is still
7445 not available as public API.
7447 * systemd will now look for the "debug" argument on the kernel
7448 command line and enable debug logging, similar to what
7449 "systemd.log_level=debug" already did before.
7451 * "systemctl set-default", "systemctl get-default" has been
7452 added to configure the default.target symlink, which
7453 controls what to boot into by default.
7455 * "systemctl set-log-level" has been added as a convenient
7456 way to raise and lower systemd logging threshold.
7458 * "systemd-analyze plot" will now show the time the various
7459 generators needed for execution, as well as information
7460 about the unit file loading.
7462 * libsystemd-journal gained a new sd_journal_open_files() call
7463 for opening specific journal files. journactl also gained a
7464 new switch to expose this new functionality. Previously we
7465 only supported opening all files from a directory, or all
7466 files from the system, as opening individual files only is
7467 racy due to journal file rotation.
7469 * systemd gained the new DefaultEnvironment= setting in
7470 /etc/systemd/system.conf to set environment variables for
7473 * If a privileged process logs a journal message with the
7474 OBJECT_PID= field set, then journald will automatically
7475 augment this with additional OBJECT_UID=, OBJECT_GID=,
7476 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
7477 system services want to log events about specific client
7478 processes. journactl/systemctl has been updated to make use
7479 of this information if all log messages regarding a specific
7482 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
7483 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
7484 Reisner, David Coppa, David King, David Strauss, Eelco
7485 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
7486 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
7487 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
7488 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
7489 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
7490 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
7491 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
7492 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
7493 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
7494 Łukasz Stelmach, 장동준
7498 * The Python bindings gained some minimal support for the APIs
7499 exposed by libsystemd-logind.
7501 * ConditionSecurity= gained support for detecting SMACK. Since
7502 this condition already supports SELinux and AppArmor we only
7503 miss IMA for this. Patches welcome!
7505 Contributions from: Karol Lewandowski, Lennart Poettering,
7506 Zbigniew Jędrzejewski-Szmek
7510 * systemd-nspawn will now create /etc/resolv.conf if
7511 necessary, before bind-mounting the host's file onto it.
7513 * systemd-nspawn will now store meta information about a
7514 container on the container's cgroup as extended attribute
7515 fields, including the root directory.
7517 * The cgroup hierarchy has been reworked in many ways. All
7518 objects any of the components systemd creates in the cgroup
7519 tree are now suffixed. More specifically, user sessions are
7520 now placed in cgroups suffixed with ".session", users in
7521 cgroups suffixed with ".user", and nspawn containers in
7522 cgroups suffixed with ".nspawn". Furthermore, all cgroup
7523 names are now escaped in a simple scheme to avoid collision
7524 of userspace object names with kernel filenames. This work
7525 is preparation for making these objects relocatable in the
7526 cgroup tree, in order to allow easy resource partitioning of
7527 these objects without causing naming conflicts.
7529 * systemctl list-dependencies gained the new switches
7530 --plain, --reverse, --after and --before.
7532 * systemd-inhibit now shows the process name of processes that
7533 have taken an inhibitor lock.
7535 * nss-myhostname will now also resolve "localhost"
7536 implicitly. This makes /etc/hosts an optional file and
7537 nicely handles that on IPv6 ::1 maps to both "localhost" and
7540 * libsystemd-logind.so gained a new call
7541 sd_get_machine_names() to enumerate running containers and
7542 VMs (currently only supported by very new libvirt and
7543 nspawn). sd_login_monitor can now be used to watch
7544 VMs/containers coming and going.
7546 * .include is not allowed recursively anymore, and only in
7547 unit files. Usually it is better to use drop-in snippets in
7548 .d/*.conf anyway, as introduced with systemd 198.
7550 * systemd-analyze gained a new "critical-chain" command that
7551 determines the slowest chain of units run during system
7552 boot-up. It is very useful for tracking down where
7553 optimizing boot time is the most beneficial.
7555 * systemd will no longer allow manipulating service paths in
7556 the name=systemd:/system cgroup tree using ControlGroup= in
7557 units. (But is still fine with it in all other dirs.)
7559 * There's a new systemd-nspawn@.service service file that may
7560 be used to easily run nspawn containers as system
7561 services. With the container's root directory in
7562 /var/lib/container/foobar it is now sufficient to run
7563 "systemctl start systemd-nspawn@foobar.service" to boot it.
7565 * systemd-cgls gained a new parameter "--machine" to list only
7566 the processes within a certain container.
7568 * ConditionSecurity= now can check for "apparmor". We still
7569 are lacking checks for SMACK and IMA for this condition
7570 check though. Patches welcome!
7572 * A new configuration file /etc/systemd/sleep.conf has been
7573 added that may be used to configure which kernel operation
7574 systemd is supposed to execute when "suspend", "hibernate"
7575 or "hybrid-sleep" is requested. This makes the new kernel
7576 "freeze" state accessible to the user.
7578 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
7579 the passed argument if applicable.
7581 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
7582 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
7583 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
7584 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
7585 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
7586 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
7587 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
7592 * The output of 'systemctl list-jobs' got some polishing. The
7593 '--type=' argument may now be passed more than once. A new
7594 command 'systemctl list-sockets' has been added which shows
7595 a list of kernel sockets systemd is listening on with the
7596 socket units they belong to, plus the units these socket
7599 * The experimental libsystemd-bus library got substantial
7600 updates to work in conjunction with the (also experimental)
7601 kdbus kernel project. It works well enough to exchange
7602 messages with some sophistication. Note that kdbus is not
7603 ready yet, and the library is mostly an elaborate test case
7604 for now, and not installable.
7606 * systemd gained a new unit 'systemd-static-nodes.service'
7607 that generates static device nodes earlier during boot, and
7608 can run in conjunction with udev.
7610 * libsystemd-login gained a new call sd_pid_get_user_unit()
7611 to retrieve the user systemd unit a process is running
7612 in. This is useful for systems where systemd is used as
7615 * systemd-nspawn now places all containers in the new /machine
7616 top-level cgroup directory in the name=systemd
7617 hierarchy. libvirt will soon do the same, so that we get a
7618 uniform separation of /system, /user and /machine for system
7619 services, user processes and containers/virtual
7620 machines. This new cgroup hierarchy is also useful to stick
7621 stable names to specific container instances, which can be
7622 recognized later this way (this name may be controlled
7623 via systemd-nspawn's new -M switch). libsystemd-login also
7624 gained a new call sd_pid_get_machine_name() to retrieve the
7625 name of the container/VM a specific process belongs to.
7627 * bootchart can now store its data in the journal.
7629 * libsystemd-journal gained a new call
7630 sd_journal_add_conjunction() for AND expressions to the
7631 matching logic. This can be used to express more complex
7632 logical expressions.
7634 * journactl can now take multiple --unit= and --user-unit=
7637 * The cryptsetup logic now understands the "luks.key=" kernel
7638 command line switch for specifying a file to read the
7639 decryption key from. Also, if a configured key file is not
7640 found the tool will now automatically fall back to prompting
7643 * Python systemd.journal module was updated to wrap recently
7644 added functions from libsystemd-journal. The interface was
7645 changed to bring the low level interface in s.j._Reader
7646 closer to the C API, and the high level interface in
7647 s.j.Reader was updated to wrap and convert all data about
7650 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
7651 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
7652 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
7653 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
7654 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
7655 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7659 * journalctl --update-catalog now understands a new --root=
7660 option to operate on catalogs found in a different root
7663 * During shutdown after systemd has terminated all running
7664 services a final killing loop kills all remaining left-over
7665 processes. We will now print the name of these processes
7666 when we send SIGKILL to them, since this usually indicates a
7669 * If /etc/crypttab refers to password files stored on
7670 configured mount points automatic dependencies will now be
7671 generated to ensure the specific mount is established first
7672 before the key file is attempted to be read.
7674 * 'systemctl status' will now show information about the
7675 network sockets a socket unit is listening on.
7677 * 'systemctl status' will also shown information about any
7678 drop-in configuration file for units. (Drop-In configuration
7679 files in this context are files such as
7680 /etc/systemd/systemd/foobar.service.d/*.conf)
7682 * systemd-cgtop now optionally shows summed up CPU times of
7683 cgroups. Press '%' while running cgtop to switch between
7684 percentage and absolute mode. This is useful to determine
7685 which cgroups use up the most CPU time over the entire
7686 runtime of the system. systemd-cgtop has also been updated
7687 to be 'pipeable' for processing with further shell tools.
7689 * 'hostnamectl set-hostname' will now allow setting of FQDN
7692 * The formatting and parsing of time span values has been
7693 changed. The parser now understands fractional expressions
7694 such as "5.5h". The formatter will now output fractional
7695 expressions for all time spans under 1min, i.e. "5.123456s"
7696 rather than "5s 123ms 456us". For time spans under 1s
7697 millisecond values are shown, for those under 1ms
7698 microsecond values are shown. This should greatly improve
7699 all time-related output of systemd.
7701 * libsystemd-login and libsystemd-journal gained new
7702 functions for querying the poll() events mask and poll()
7703 timeout value for integration into arbitrary event
7706 * localectl gained the ability to list available X11 keymaps
7707 (models, layouts, variants, options).
7709 * 'systemd-analyze dot' gained the ability to filter for
7710 specific units via shell-style globs, to create smaller,
7711 more useful graphs. I.e. it is now possible to create simple
7712 graphs of all the dependencies between only target units, or
7713 of all units that Avahi has dependencies with.
7715 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
7716 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
7717 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
7718 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
7719 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
7720 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
7721 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
7725 * The boot-time readahead implementation for rotating media
7726 will now read the read-ahead data in multiple passes which
7727 consist of all read requests made in equidistant time
7728 intervals. This means instead of strictly reading read-ahead
7729 data in its physical order on disk we now try to find a
7730 middle ground between physical and access time order.
7732 * /etc/os-release files gained a new BUILD_ID= field for usage
7733 on operating systems that provide continuous builds of OS
7736 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
7737 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
7738 William Douglas, Zbigniew Jędrzejewski-Szmek
7742 * systemd-python gained an API exposing libsystemd-daemon.
7744 * The SMACK setup logic gained support for uploading CIPSO
7747 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
7748 ReadOnlyDirectories= and InaccessibleDirectories= has
7749 changed. The private /tmp and /var/tmp directories are now
7750 shared by all processes of a service (which means
7751 ExecStartPre= may now leave data in /tmp that ExecStart= of
7752 the same service can still access). When a service is
7753 stopped its temporary directories are immediately deleted
7754 (normal clean-up with tmpfiles is still done in addition to
7757 * By default, systemd will now set a couple of sysctl
7758 variables in the kernel: the safe sysrq options are turned
7759 on, IP route verification is turned on, and source routing
7760 disabled. The recently added hardlink and softlink
7761 protection of the kernel is turned on. These settings should
7762 be reasonably safe, and good defaults for all new systems.
7764 * The predictable network naming logic may now be turned off
7765 with a new kernel command line switch: net.ifnames=0.
7767 * A new libsystemd-bus module has been added that implements a
7768 pretty complete D-Bus client library. For details see:
7770 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
7772 * journald will now explicitly flush the journal files to disk
7773 at the latest 5min after each write. The file will then also
7774 be marked offline until the next write. This should increase
7775 reliability in case of a crash. The synchronization delay
7776 can be configured via SyncIntervalSec= in journald.conf.
7778 * There's a new remote-fs-setup.target unit that can be used
7779 to pull in specific services when at least one remote file
7780 system is to be mounted.
7782 * There are new targets timers.target and paths.target as
7783 canonical targets to pull user timer and path units in
7784 from. This complements sockets.target with a similar
7785 purpose for socket units.
7787 * libudev gained a new call udev_device_set_attribute_value()
7788 to set sysfs attributes of a device.
7790 * The udev daemon now sets the default number of worker
7791 processes executed in parallel based on the number of available
7792 CPUs instead of the amount of available RAM. This is supposed
7793 to provide a more reliable default and limit a too aggressive
7794 parallelism for setups with 1000s of devices connected.
7796 Contributions from: Auke Kok, Colin Walters, Cristian
7797 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
7798 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
7799 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
7800 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
7801 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
7802 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
7803 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
7804 Zbigniew Jędrzejewski-Szmek
7808 * Configuration of unit files may now be extended via drop-in
7809 files without having to edit/override the unit files
7810 themselves. More specifically, if the administrator wants to
7811 change one value for a service file foobar.service he can
7812 now do so by dropping in a configuration snippet into
7813 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
7814 will load all these snippets and apply them on top of the
7815 main unit configuration file, possibly extending or
7816 overriding its settings. Using these drop-in snippets is
7817 generally nicer than the two earlier options for changing
7818 unit files locally: copying the files from
7819 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
7820 them there; or creating a new file in /etc/systemd/system/
7821 that incorporates the original one via ".include". Drop-in
7822 snippets into these .d/ directories can be placed in any
7823 directory systemd looks for units in, and the usual
7824 overriding semantics between /usr/lib, /etc and /run apply
7827 * Most unit file settings which take lists of items can now be
7828 reset by assigning the empty string to them. For example,
7829 normally, settings such as Environment=FOO=BAR append a new
7830 environment variable assignment to the environment block,
7831 each time they are used. By assigning Environment= the empty
7832 string the environment block can be reset to empty. This is
7833 particularly useful with the .d/*.conf drop-in snippets
7834 mentioned above, since this adds the ability to reset list
7835 settings from vendor unit files via these drop-ins.
7837 * systemctl gained a new "list-dependencies" command for
7838 listing the dependencies of a unit recursively.
7840 * Inhibitors are now honored and listed by "systemctl
7841 suspend", "systemctl poweroff" (and similar) too, not only
7842 GNOME. These commands will also list active sessions by
7845 * Resource limits (as exposed by the various control group
7846 controllers) can now be controlled dynamically at runtime
7847 for all units. More specifically, you can now use a command
7848 like "systemctl set-cgroup-attr foobar.service cpu.shares
7849 2000" to alter the CPU shares a specific service gets. These
7850 settings are stored persistently on disk, and thus allow the
7851 administrator to easily adjust the resource usage of
7852 services with a few simple commands. This dynamic resource
7853 management logic is also available to other programs via the
7854 bus. Almost any kernel cgroup attribute and controller is
7857 * systemd-vconsole-setup will now copy all font settings to
7858 all allocated VTs, where it previously applied them only to
7861 * libsystemd-login gained the new sd_session_get_tty() API
7864 * This release drops support for a few legacy or
7865 distribution-specific LSB facility names when parsing init
7866 scripts: $x-display-manager, $mail-transfer-agent,
7867 $mail-transport-agent, $mail-transfer-agent, $smtp,
7868 $null. Also, the mail-transfer-agent.target unit backing
7869 this has been removed. Distributions which want to retain
7870 compatibility with this should carry the burden for
7871 supporting this themselves and patch support for these back
7872 in, if they really need to. Also, the facilities $syslog and
7873 $local_fs are now ignored, since systemd does not support
7874 early-boot LSB init scripts anymore, and these facilities
7875 are implied anyway for normal services. syslog.target has
7878 * There are new bus calls on PID1's Manager object for
7879 cancelling jobs, and removing snapshot units. Previously,
7880 both calls were only available on the Job and Snapshot
7883 * systemd-journal-gatewayd gained SSL support.
7885 * The various "environment" files, such as /etc/locale.conf
7886 now support continuation lines with a backslash ("\") as
7887 last character in the line, similarly in style (but different)
7888 to how this is supported in shells.
7890 * For normal user processes the _SYSTEMD_USER_UNIT= field is
7891 now implicitly appended to every log entry logged. systemctl
7892 has been updated to filter by this field when operating on a
7893 user systemd instance.
7895 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
7896 CAP_AUDIT_CONTROL capabilities to the capabilities set for
7897 the container. This makes it easier to boot unmodified
7898 Fedora systems in a container, which however still requires
7899 audit=0 to be passed on the kernel command line. Auditing in
7900 kernel and userspace is unfortunately still too broken in
7901 context of containers, hence we recommend compiling it out
7902 of the kernel or using audit=0. Hopefully this will be fixed
7903 one day for good in the kernel.
7905 * nspawn gained the new --bind= and --bind-ro= parameters to
7906 bind mount specific directories from the host into the
7909 * nspawn will now mount its own devpts file system instance
7910 into the container, in order not to leak pty devices from
7911 the host into the container.
7913 * systemd will now read the firmware boot time performance
7914 information from the EFI variables, if the used boot loader
7915 supports this, and takes it into account for boot performance
7916 analysis via "systemd-analyze". This is currently supported
7917 only in conjunction with Gummiboot, but could be supported
7918 by other boot loaders too. For details see:
7920 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
7922 * A new generator has been added that automatically mounts the
7923 EFI System Partition (ESP) to /boot, if that directory
7924 exists, is empty, and no other file system has been
7925 configured to be mounted there.
7927 * logind will now send out PrepareForSleep(false) out
7928 unconditionally, after coming back from suspend. This may be
7929 used by applications as asynchronous notification for
7930 system resume events.
7932 * "systemctl unlock-sessions" has been added, that allows
7933 unlocking the screens of all user sessions at once, similar
7934 to how "systemctl lock-sessions" already locked all users
7935 sessions. This is backed by a new D-Bus call UnlockSessions().
7937 * "loginctl seat-status" will now show the master device of a
7938 seat. (i.e. the device of a seat that needs to be around for
7939 the seat to be considered available, usually the graphics
7942 * tmpfiles gained a new "X" line type, that allows
7943 configuration of files and directories (with wildcards) that
7944 shall be excluded from automatic cleanup ("aging").
7946 * udev default rules set the device node permissions now only
7947 at "add" events, and do not change them any longer with a
7948 later "change" event.
7950 * The log messages for lid events and power/sleep keypresses
7951 now carry a message ID.
7953 * We now have a substantially larger unit test suite, but this
7954 continues to be work in progress.
7956 * udevadm hwdb gained a new --root= parameter to change the
7957 root directory to operate relative to.
7959 * logind will now issue a background sync() request to the kernel
7960 early at shutdown, so that dirty buffers are flushed to disk early
7961 instead of at the last moment, in order to optimize shutdown
7964 * A new bootctl tool has been added that is an interface for
7965 certain boot loader operations. This is currently a preview
7966 and is likely to be extended into a small mechanism daemon
7967 like timedated, localed, hostnamed, and can be used by
7968 graphical UIs to enumerate available boot options, and
7969 request boot into firmware operations.
7971 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
7972 the rest of the package. It also has been updated to work
7973 correctly in initrds.
7975 * polkit previously has been runtime optional, and is now also
7976 compile time optional via a configure switch.
7978 * systemd-analyze has been reimplemented in C. Also "systemctl
7979 dot" has moved into systemd-analyze.
7981 * "systemctl status" with no further parameters will now print
7982 the status of all active or failed units.
7984 * Operations such as "systemctl start" can now be executed
7985 with a new mode "--irreversible" which may be used to queue
7986 operations that cannot accidentally be reversed by a later
7987 job queuing. This is by default used to make shutdown
7988 requests more robust.
7990 * The Python API of systemd now gained a new module for
7991 reading journal files.
7993 * A new tool kernel-install has been added that can install
7994 kernel images according to the Boot Loader Specification:
7996 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
7998 * Boot time console output has been improved to provide
7999 animated boot time output for hanging jobs.
8001 * A new tool systemd-activate has been added which can be used
8002 to test socket activation with, directly from the command
8003 line. This should make it much easier to test and debug
8004 socket activation in daemons.
8006 * journalctl gained a new "--reverse" (or -r) option to show
8007 journal output in reverse order (i.e. newest line first).
8009 * journalctl gained a new "--pager-end" (or -e) option to jump
8010 to immediately jump to the end of the journal in the
8011 pager. This is only supported in conjunction with "less".
8013 * journalctl gained a new "--user-unit=" option, that works
8014 similarly to "--unit=" but filters for user units rather than
8017 * A number of unit files to ease adoption of systemd in
8018 initrds has been added. This moves some minimal logic from
8019 the various initrd implementations into systemd proper.
8021 * The journal files are now owned by a new group
8022 "systemd-journal", which exists specifically to allow access
8023 to the journal, and nothing else. Previously, we used the
8024 "adm" group for that, which however possibly covers more
8025 than just journal/log file access. This new group is now
8026 already used by systemd-journal-gatewayd to ensure this
8027 daemon gets access to the journal files and as little else
8028 as possible. Note that "make install" will also set FS ACLs
8029 up for /var/log/journal to give "adm" and "wheel" read
8030 access to it, in addition to "systemd-journal" which owns
8031 the journal files. We recommend that packaging scripts also
8032 add read access to "adm" + "wheel" to /var/log/journal, and
8033 all existing/future journal files. To normal users and
8034 administrators little changes, however packagers need to
8035 ensure to create the "systemd-journal" system group at
8036 package installation time.
8038 * The systemd-journal-gatewayd now runs as unprivileged user
8039 systemd-journal-gateway:systemd-journal-gateway. Packaging
8040 scripts need to create these system user/group at
8043 * timedated now exposes a new boolean property CanNTP that
8044 indicates whether a local NTP service is available or not.
8046 * systemd-detect-virt will now also detect xen PVs
8048 * The pstore file system is now mounted by default, if it is
8051 * In addition to the SELinux and IMA policies we will now also
8052 load SMACK policies at early boot.
8054 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
8055 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
8056 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
8057 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
8058 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
8059 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
8060 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
8061 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
8062 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
8063 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
8064 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
8065 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
8066 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
8067 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
8071 * Timer units now support calendar time events in addition to
8072 monotonic time events. That means you can now trigger a unit
8073 based on a calendar time specification such as "Thu,Fri
8074 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
8075 or fifth day of any month of the year 2013, given that it is
8076 a thursday or friday. This brings timer event support
8077 considerably closer to cron's capabilities. For details on
8078 the supported calendar time specification language see
8081 * udev now supports a number of different naming policies for
8082 network interfaces for predictable names, and a combination
8083 of these policies is now the default. Please see this wiki
8084 document for details:
8086 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
8088 * Auke Kok's bootchart implementation has been added to the
8089 systemd tree. It is an optional component that can graph the
8090 boot in quite some detail. It is one of the best bootchart
8091 implementations around and minimal in its code and
8094 * nss-myhostname has been integrated into the systemd source
8095 tree. nss-myhostname guarantees that the local hostname
8096 always stays resolvable via NSS. It has been a weak
8097 requirement of systemd-hostnamed since a long time, and
8098 since its code is actually trivial we decided to just
8099 include it in systemd's source tree. It can be turned off
8100 with a configure switch.
8102 * The read-ahead logic is now capable of properly detecting
8103 whether a btrfs file system is on SSD or rotating media, in
8104 order to optimize the read-ahead scheme. Previously, it was
8105 only capable of detecting this on traditional file systems
8108 * In udev, additional device properties are now read from the
8109 IAB in addition to the OUI database. Also, Bluetooth company
8110 identities are attached to the devices as well.
8112 * In service files %U may be used as specifier that is
8113 replaced by the configured user name of the service.
8115 * nspawn may now be invoked without a controlling TTY. This
8116 makes it suitable for invocation as its own service. This
8117 may be used to set up a simple containerized server system
8118 using only core OS tools.
8120 * systemd and nspawn can now accept socket file descriptors
8121 when they are started for socket activation. This enables
8122 implementation of socket activated nspawn
8123 containers. i.e. think about autospawning an entire OS image
8124 when the first SSH or HTTP connection is received. We expect
8125 that similar functionality will also be added to libvirt-lxc
8128 * journalctl will now suppress ANSI color codes when
8129 presenting log data.
8131 * systemctl will no longer show control group information for
8132 a unit if the control group is empty anyway.
8134 * logind can now automatically suspend/hibernate/shutdown the
8137 * /etc/machine-info and hostnamed now also expose the chassis
8138 type of the system. This can be used to determine whether
8139 the local system is a laptop, desktop, handset or
8140 tablet. This information may either be configured by the
8141 user/vendor or is automatically determined from ACPI and DMI
8142 information if possible.
8144 * A number of polkit actions are now bound together with "imply"
8145 rules. This should simplify creating UIs because many actions
8146 will now authenticate similar ones as well.
8148 * Unit files learnt a new condition ConditionACPower= which
8149 may be used to conditionalize a unit depending on whether an
8150 AC power source is connected or not, of whether the system
8151 is running on battery power.
8153 * systemctl gained a new "is-failed" verb that may be used in
8154 shell scripts and suchlike to check whether a specific unit
8155 is in the "failed" state.
8157 * The EnvironmentFile= setting in unit files now supports file
8158 globbing, and can hence be used to easily read a number of
8159 environment files at once.
8161 * systemd will no longer detect and recognize specific
8162 distributions. All distribution-specific #ifdeffery has been
8163 removed, systemd is now fully generic and
8164 distribution-agnostic. Effectively, not too much is lost as
8165 a lot of the code is still accessible via explicit configure
8166 switches. However, support for some distribution specific
8167 legacy configuration file formats has been dropped. We
8168 recommend distributions to simply adopt the configuration
8169 files everybody else uses now and convert the old
8170 configuration from packaging scripts. Most distributions
8171 already did that. If that's not possible or desirable,
8172 distributions are welcome to forward port the specific
8173 pieces of code locally from the git history.
8175 * When logging a message about a unit systemd will now always
8176 log the unit name in the message meta data.
8178 * localectl will now also discover system locale data that is
8179 not stored in locale archives, but directly unpacked.
8181 * logind will no longer unconditionally use framebuffer
8182 devices as seat masters, i.e. as devices that are required
8183 to be existing before a seat is considered preset. Instead,
8184 it will now look for all devices that are tagged as
8185 "seat-master" in udev. By default, framebuffer devices will
8186 be marked as such, but depending on local systems, other
8187 devices might be marked as well. This may be used to
8188 integrate graphics cards using closed source drivers (such
8189 as NVidia ones) more nicely into logind. Note however, that
8190 we recommend using the open source NVidia drivers instead,
8191 and no udev rules for the closed-source drivers will be
8192 shipped from us upstream.
8194 Contributions from: Adam Williamson, Alessandro Crismani, Auke
8195 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
8196 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
8197 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
8198 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
8199 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
8200 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
8201 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
8202 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
8203 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
8204 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
8205 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
8210 * udev gained support for loading additional device properties
8211 from an indexed database that is keyed by vendor/product IDs
8212 and similar device identifiers. For the beginning this
8213 "hwdb" is populated with data from the well-known PCI and
8214 USB database, but also includes PNP, ACPI and OID data. In
8215 the longer run this indexed database shall grow into
8216 becoming the one central database for non-essential
8217 userspace device metadata. Previously, data from the PCI/USB
8218 database was only attached to select devices, since the
8219 lookup was a relatively expensive operation due to O(n) time
8220 complexity (with n being the number of entries in the
8221 database). Since this is now O(1), we decided to add in this
8222 data for all devices where this is available, by
8223 default. Note that the indexed database needs to be rebuilt
8224 when new data files are installed. To achieve this you need
8225 to update your packaging scripts to invoke "udevadm hwdb
8226 --update" after installation of hwdb data files. For
8227 RPM-based distributions we introduced the new
8228 %udev_hwdb_update macro for this purpose.
8230 * The Journal gained support for the "Message Catalog", an
8231 indexed database to link up additional information with
8232 journal entries. For further details please check:
8234 https://www.freedesktop.org/wiki/Software/systemd/catalog
8236 The indexed message catalog database also needs to be
8237 rebuilt after installation of message catalog files. Use
8238 "journalctl --update-catalog" for this. For RPM-based
8239 distributions we introduced the %journal_catalog_update
8240 macro for this purpose.
8242 * The Python Journal bindings gained support for the standard
8243 Python logging framework.
8245 * The Journal API gained new functions for checking whether
8246 the underlying file system of a journal file is capable of
8247 properly reporting file change notifications, or whether
8248 applications that want to reflect journal changes "live"
8249 need to recheck journal files continuously in appropriate
8252 * It is now possible to set the "age" field for tmpfiles
8253 entries to 0, indicating that files matching this entry
8254 shall always be removed when the directories are cleaned up.
8256 * coredumpctl gained a new "gdb" verb which invokes gdb
8257 right-away on the selected coredump.
8259 * There's now support for "hybrid sleep" on kernels that
8260 support this, in addition to "suspend" and "hibernate". Use
8261 "systemctl hybrid-sleep" to make use of this.
8263 * logind's HandleSuspendKey= setting (and related settings)
8264 now gained support for a new "lock" setting to simply
8265 request the screen lock on all local sessions, instead of
8266 actually executing a suspend or hibernation.
8268 * systemd will now mount the EFI variables file system by
8271 * Socket units now gained support for configuration of the
8272 SMACK security label.
8274 * timedatectl will now output the time of the last and next
8275 daylight saving change.
8277 * We dropped support for various legacy and distro-specific
8278 concepts, such as insserv, early-boot SysV services
8279 (i.e. those for non-standard runlevels such as 'b' or 'S')
8280 or ArchLinux /etc/rc.conf support. We recommend the
8281 distributions who still need support this to either continue
8282 to maintain the necessary patches downstream, or find a
8283 different solution. (Talk to us if you have questions!)
8285 * Various systemd components will now bypass polkit checks for
8286 root and otherwise handle properly if polkit is not found to
8287 be around. This should fix most issues for polkit-less
8288 systems. Quite frankly this should have been this way since
8289 day one. It is absolutely our intention to make systemd work
8290 fine on polkit-less systems, and we consider it a bug if
8291 something does not work as it should if polkit is not around.
8293 * For embedded systems it is now possible to build udev and
8294 systemd without blkid and/or kmod support.
8296 * "systemctl switch-root" is now capable of switching root
8297 more than once. I.e. in addition to transitions from the
8298 initrd to the host OS it is now possible to transition to
8299 further OS images from the host. This is useful to implement
8300 offline updating tools.
8302 * Various other additions have been made to the RPM macros
8303 shipped with systemd. Use %udev_rules_update() after
8304 installing new udev rules files. %_udevhwdbdir,
8305 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
8306 %_sysctldir are now available which resolve to the right
8307 directories for packages to place various data files in.
8309 * journalctl gained the new --full switch (in addition to
8310 --all, to disable ellipsation for long messages.
8312 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
8313 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
8314 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
8315 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
8316 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
8317 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
8318 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
8319 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
8320 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
8324 * journalctl gained new --since= and --until= switches to
8325 filter by time. It also now supports nice filtering for
8326 units via --unit=/-u.
8328 * Type=oneshot services may use ExecReload= and do the
8331 * The journal daemon now supports time-based rotation and
8332 vacuuming, in addition to the usual disk-space based
8335 * The journal will now index the available field values for
8336 each field name. This enables clients to show pretty drop
8337 downs of available match values when filtering. The bash
8338 completion of journalctl has been updated
8339 accordingly. journalctl gained a new switch -F to list all
8340 values a certain field takes in the journal database.
8342 * More service events are now written as structured messages
8343 to the journal, and made recognizable via message IDs.
8345 * The timedated, localed and hostnamed mini-services which
8346 previously only provided support for changing time, locale
8347 and hostname settings from graphical DEs such as GNOME now
8348 also have a minimal (but very useful) text-based client
8349 utility each. This is probably the nicest way to changing
8350 these settings from the command line now, especially since
8351 it lists available options and is fully integrated with bash
8354 * There's now a new tool "systemd-coredumpctl" to list and
8355 extract coredumps from the journal.
8357 * We now install a README each in /var/log/ and
8358 /etc/rc.d/init.d explaining where the system logs and init
8359 scripts went. This hopefully should help folks who go to
8360 that dirs and look into the otherwise now empty void and
8361 scratch their heads.
8363 * When user-services are invoked (by systemd --user) the
8364 $MANAGERPID env var is set to the PID of systemd.
8366 * SIGRTMIN+24 when sent to a --user instance will now result
8367 in immediate termination of systemd.
8369 * gatewayd received numerous feature additions such as a
8370 "follow" mode, for live syncing and filtering.
8372 * browse.html now allows filtering and showing detailed
8373 information on specific entries. Keyboard navigation and
8374 mouse screen support has been added.
8376 * gatewayd/journalctl now supports HTML5/JSON
8377 Server-Sent-Events as output.
8379 * The SysV init script compatibility logic will now
8380 heuristically determine whether a script supports the
8381 "reload" verb, and only then make this available as
8384 * "systemctl status --follow" has been removed, use "journalctl
8387 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
8388 have been removed since they are hardly useful to be
8391 * And I'd like to take the opportunity to specifically mention
8392 Zbigniew for his great contributions. Zbigniew, you rock!
8394 Contributions from: Andrew Eikum, Christian Hesse, Colin
8395 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
8396 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
8397 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
8398 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
8399 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
8400 Jędrzejewski-Szmek, Сковорода Никита Андреевич
8404 * If /etc/vconsole.conf is non-existent or empty we will no
8405 longer load any console font or key map at boot by
8406 default. Instead the kernel defaults will be left
8407 intact. This is definitely the right thing to do, as no
8408 configuration should mean no configuration, and hard-coding
8409 font names that are different on all archs is probably a bad
8410 idea. Also, the kernel default key map and font should be
8411 good enough for most cases anyway, and mostly identical to
8412 the userspace fonts/key maps we previously overloaded them
8413 with. If distributions want to continue to default to a
8414 non-kernel font or key map they should ship a default
8415 /etc/vconsole.conf with the appropriate contents.
8417 Contributions from: Colin Walters, Daniel J Walsh, Dave
8418 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
8419 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8423 * journalctl gained a new --cursor= switch to show entries
8424 starting from the specified location in the journal.
8426 * We now enforce a size limit on journal entry fields exported
8427 with "-o json" in journalctl. Fields larger than 4K will be
8428 assigned null. This can be turned off with --all.
8430 * An (optional) journal gateway daemon is now available as
8431 "systemd-journal-gatewayd.service". This service provides
8432 access to the journal via HTTP and JSON. This functionality
8433 will be used to implement live log synchronization in both
8434 pull and push modes, but has various other users too, such
8435 as easy log access for debugging of embedded devices. Right
8436 now it is already useful to retrieve the journal via HTTP:
8438 # systemctl start systemd-journal-gatewayd.service
8439 # wget http://localhost:19531/entries
8441 This will download the journal contents in a
8442 /var/log/messages compatible format. The same as JSON:
8444 # curl -H"Accept: application/json" http://localhost:19531/entries
8446 This service is also accessible via a web browser where a
8447 single static HTML5 app is served that uses the JSON logic
8448 to enable the user to do some basic browsing of the
8449 journal. This will be extended later on. Here's an example
8450 screenshot of this app in its current state:
8452 http://0pointer.de/public/journal-gatewayd
8454 Contributions from: Kay Sievers, Lennart Poettering, Robert
8455 Milasan, Tom Gundersen
8459 * The bash completion logic is now available for journalctl
8462 * We do not mount the "cpuset" controller anymore together with
8463 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
8464 started if no parameters are assigned to it. "cpuset" hence
8465 broke code that assumed it could create "cpu" groups and
8468 * journalctl -f will now subscribe to terminal size changes,
8469 and line break accordingly.
8471 Contributions from: Dave Reisner, Kay Sievers, Lennart
8472 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
8476 * nspawn will now create a symlink /etc/localtime in the
8477 container environment, copying the host's timezone
8478 setting. Previously this has been done via a bind mount, but
8479 since symlinks cannot be bind mounted this has now been
8480 changed to create/update the appropriate symlink.
8482 * journalctl -n's line number argument is now optional, and
8483 will default to 10 if omitted.
8485 * journald will now log the maximum size the journal files may
8486 take up on disk. This is particularly useful if the default
8487 built-in logic of determining this parameter from the file
8488 system size is used. Use "systemctl status
8489 systemd-journald.service" to see this information.
8491 * The multi-seat X wrapper tool has been stripped down. As X
8492 is now capable of enumerating graphics devices via udev in a
8493 seat-aware way the wrapper is not strictly necessary
8494 anymore. A stripped down temporary stop-gap is still shipped
8495 until the upstream display managers have been updated to
8496 fully support the new X logic. Expect this wrapper to be
8497 removed entirely in one of the next releases.
8499 * HandleSleepKey= in logind.conf has been split up into
8500 HandleSuspendKey= and HandleHibernateKey=. The old setting
8501 is not available anymore. X11 and the kernel are
8502 distinguishing between these keys and we should too. This
8503 also means the inhibition lock for these keys has been split
8506 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
8507 Poettering, Lukas Nykryn, Václav Pavlín
8511 * Whenever a unit changes state we will now log this to the
8512 journal and show along the unit's own log output in
8515 * ConditionPathIsMountPoint= can now properly detect bind
8516 mount points too. (Previously, a bind mount of one file
8517 system to another place in the same file system could not be
8518 detected as mount, since they shared struct stat's st_dev
8521 * We will now mount the cgroup controllers cpu, cpuacct,
8522 cpuset and the controllers net_cls, net_prio together by
8525 * nspawn containers will now have a virtualized boot
8526 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
8527 over with a randomized ID at container initialization). This
8528 has the effect of making "journalctl -b" do the right thing
8531 * The JSON output journal serialization has been updated not
8532 to generate "endless" list objects anymore, but rather one
8533 JSON object per line. This is more in line how most JSON
8534 parsers expect JSON objects. The new output mode
8535 "json-pretty" has been added to provide similar output, but
8536 neatly aligned for readability by humans.
8538 * We dropped all explicit sync() invocations in the shutdown
8539 code. The kernel does this implicitly anyway in the kernel
8540 reboot() syscall. halt(8)'s -n option is now a compatibility
8543 * We now support virtualized reboot() in containers, as
8544 supported by newer kernels. We will fall back to exit() if
8545 CAP_SYS_REBOOT is not available to the container. Also,
8546 nspawn makes use of this now and will actually reboot the
8547 container if the containerized OS asks for that.
8549 * journalctl will only show local log output by default
8550 now. Use --merge (-m) to show remote log output, too.
8552 * libsystemd-journal gained the new sd_journal_get_usage()
8553 call to determine the current disk usage of all journal
8554 files. This is exposed in the new "journalctl --disk-usage"
8557 * journald gained a new configuration setting SplitMode= in
8558 journald.conf which may be used to control how user journals
8559 are split off. See journald.conf(5) for details.
8561 * A new condition type ConditionFileNotEmpty= has been added.
8563 * tmpfiles' "w" lines now support file globbing, to write
8564 multiple files at once.
8566 * We added Python bindings for the journal submission
8567 APIs. More Python APIs for a number of selected APIs will
8568 likely follow. Note that we intend to add native bindings
8569 only for the Python language, as we consider it common
8570 enough to deserve bindings shipped within systemd. There are
8571 various projects outside of systemd that provide bindings
8572 for languages such as PHP or Lua.
8574 * Many conditions will now resolve specifiers such as %i. In
8575 addition, PathChanged= and related directives of .path units
8576 now support specifiers as well.
8578 * There's now a new RPM macro definition for the system preset
8581 * journald will now warn if it ca not forward a message to the
8582 syslog daemon because its socket is full.
8584 * timedated will no longer write or process /etc/timezone,
8585 except on Debian. As we do not support late mounted /usr
8586 anymore /etc/localtime always being a symlink is now safe,
8587 and hence the information in /etc/timezone is not necessary
8590 * logind will now always reserve one VT for a text getty (VT6
8591 by default). Previously if more than 6 X sessions where
8592 started they took up all the VTs with auto-spawned gettys,
8593 so that no text gettys were available anymore.
8595 * udev will now automatically inform the btrfs kernel logic
8596 about btrfs RAID components showing up. This should make
8597 simple hotplug based btrfs RAID assembly work.
8599 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
8600 (but not for its children which will stay at the kernel
8601 default). This should allow setups with a lot more listening
8604 * systemd will now always pass the configured timezone to the
8605 kernel at boot. timedated will do the same when the timezone
8608 * logind's inhibition logic has been updated. By default,
8609 logind will now handle the lid switch, the power and sleep
8610 keys all the time, even in graphical sessions. If DEs want
8611 to handle these events on their own they should take the new
8612 handle-power-key, handle-sleep-key and handle-lid-switch
8613 inhibitors during their runtime. A simple way to achieve
8614 that is to invoke the DE wrapped in an invocation of:
8616 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
8618 * Access to unit operations is now checked via SELinux taking
8619 the unit file label and client process label into account.
8621 * systemd will now notify the administrator in the journal
8622 when he over-mounts a non-empty directory.
8624 * There are new specifiers that are resolved in unit files,
8625 for the host name (%H), the machine ID (%m) and the boot ID
8628 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
8629 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
8630 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
8631 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
8632 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
8633 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
8634 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
8638 * Support for reading structured kernel messages from
8639 /dev/kmsg has now been added and is enabled by default.
8641 * Support for reading kernel messages from /proc/kmsg has now
8642 been removed. If you want kernel messages in the journal
8643 make sure to run a recent kernel (>= 3.5) that supports
8644 reading structured messages from /dev/kmsg (see
8645 above). /proc/kmsg is now exclusive property of classic
8646 syslog daemons again.
8648 * The libudev API gained the new
8649 udev_device_new_from_device_id() call.
8651 * The logic for file system namespace (ReadOnlyDirectory=,
8652 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
8653 require pivot_root() anymore. This means fewer temporary
8654 directories are created below /tmp for this feature.
8656 * nspawn containers will now see and receive all submounts
8657 made on the host OS below the root file system of the
8660 * Forward Secure Sealing is now supported for Journal files,
8661 which provide cryptographical sealing of journal files so
8662 that attackers cannot alter log history anymore without this
8663 being detectable. Lennart will soon post a blog story about
8664 this explaining it in more detail.
8666 * There are two new service settings RestartPreventExitStatus=
8667 and SuccessExitStatus= which allow configuration of exit
8668 status (exit code or signal) which will be excepted from the
8669 restart logic, resp. consider successful.
8671 * journalctl gained the new --verify switch that can be used
8672 to check the integrity of the structure of journal files and
8673 (if Forward Secure Sealing is enabled) the contents of
8676 * nspawn containers will now be run with /dev/stdin, /dev/fd/
8677 and similar symlinks pre-created. This makes running shells
8678 as container init process a lot more fun.
8680 * The fstab support can now handle PARTUUID= and PARTLABEL=
8683 * A new ConditionHost= condition has been added to match
8684 against the hostname (with globs) and machine ID. This is
8685 useful for clusters where a single OS image is used to
8686 provision a large number of hosts which shall run slightly
8687 different sets of services.
8689 * Services which hit the restart limit will now be placed in a
8692 Contributions from: Bertram Poettering, Dave Reisner, Huang
8693 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
8694 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
8698 * When running in --user mode systemd will now become a
8699 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
8700 tree a lot more organized.
8702 * A new PartOf= unit dependency type has been introduced that
8703 may be used to group services in a natural way.
8705 * "systemctl enable" may now be used to enable instances of
8708 * journalctl now prints error log levels in red, and
8709 warning/notice log levels in bright white. It also supports
8710 filtering by log level now.
8712 * cgtop gained a new -n switch (similar to top), to configure
8713 the maximum number of iterations to run for. It also gained
8714 -b, to run in batch mode (accepting no input).
8716 * The suffix ".service" may now be omitted on most systemctl
8717 command lines involving service unit names.
8719 * There's a new bus call in logind to lock all sessions, as
8720 well as a loginctl verb for it "lock-sessions".
8722 * libsystemd-logind.so gained a new call sd_journal_perror()
8723 that works similar to libc perror() but logs to the journal
8724 and encodes structured information about the error number.
8726 * /etc/crypttab entries now understand the new keyfile-size=
8729 * shutdown(8) now can send a (configurable) wall message when
8730 a shutdown is cancelled.
8732 * The mount propagation mode for the root file system will now
8733 default to "shared", which is useful to make containers work
8734 nicely out-of-the-box so that they receive new mounts from
8735 the host. This can be undone locally by running "mount
8736 --make-rprivate /" if needed.
8738 * The prefdm.service file has been removed. Distributions
8739 should maintain this unit downstream if they intend to keep
8740 it around. However, we recommend writing normal unit files
8741 for display managers instead.
8743 * Since systemd is a crucial part of the OS we will now
8744 default to a number of compiler switches that improve
8745 security (hardening) such as read-only relocations, stack
8746 protection, and suchlike.
8748 * The TimeoutSec= setting for services is now split into
8749 TimeoutStartSec= and TimeoutStopSec= to allow configuration
8750 of individual time outs for the start and the stop phase of
8753 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
8754 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
8755 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
8756 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
8757 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
8758 Gundersen, Zbigniew Jędrzejewski-Szmek
8762 * The journal and id128 C APIs are now fully documented as man
8765 * Extra safety checks have been added when transitioning from
8766 the initial RAM disk to the main system to avoid accidental
8769 * /etc/crypttab entries now understand the new keyfile-offset=
8772 * systemctl -t can now be used to filter by unit load state.
8774 * The journal C API gained the new sd_journal_wait() call to
8775 make writing synchronous journal clients easier.
8777 * journalctl gained the new -D switch to show journals from a
8780 * journalctl now displays a special marker between log
8781 messages of two different boots.
8783 * The journal is now explicitly flushed to /var via a service
8784 systemd-journal-flush.service, rather than implicitly simply
8785 by seeing /var/log/journal to be writable.
8787 * journalctl (and the journal C APIs) can now match for much
8788 more complex expressions, with alternatives and
8791 * When transitioning from the initial RAM disk to the main
8792 system we will now kill all processes in a killing spree to
8793 ensure no processes stay around by accident.
8795 * Three new specifiers may be used in unit files: %u, %h, %s
8796 resolve to the user name, user home directory resp. user
8797 shell. This is useful for running systemd user instances.
8799 * We now automatically rotate journal files if their data
8800 object hash table gets a fill level > 75%. We also size the
8801 hash table based on the configured maximum file size. This
8802 together should lower hash collisions drastically and thus
8803 speed things up a bit.
8805 * journalctl gained the new "--header" switch to introspect
8806 header data of journal files.
8808 * A new setting SystemCallFilters= has been added to services
8809 which may be used to apply blacklists or whitelists to
8810 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
8812 * nspawn gained a new --link-journal= switch (and quicker: -j)
8813 to link the container journal with the host. This makes it
8814 very easy to centralize log viewing on the host for all
8815 guests while still keeping the journal files separated.
8817 * Many bugfixes and optimizations
8819 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
8820 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
8821 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
8826 * Several tools now understand kernel command line arguments,
8827 which are only read when run in an initial RAM disk. They
8828 usually follow closely their normal counterparts, but are
8831 * There's a new tool to analyze the readahead files that are
8832 automatically generated at boot. Use:
8834 /usr/lib/systemd/systemd-readahead analyze /.readahead
8836 * We now provide an early debug shell on tty9 if this enabled. Use:
8838 systemctl enable debug-shell.service
8840 * All plymouth related units have been moved into the Plymouth
8841 package. Please make sure to upgrade your Plymouth version
8844 * systemd-tmpfiles now supports getting passed the basename of
8845 a configuration file only, in which case it will look for it
8846 in all appropriate directories automatically.
8848 * udevadm info now takes a /dev or /sys path as argument, and
8849 does the right thing. Example:
8851 udevadm info /dev/sda
8852 udevadm info /sys/class/block/sda
8854 * systemctl now prints a warning if a unit is stopped but a
8855 unit that might trigger it continues to run. Example: a
8856 service is stopped but the socket that activates it is left
8859 * "systemctl status" will now mention if the log output was
8860 shortened due to rotation since a service has been started.
8862 * The journal API now exposes functions to determine the
8863 "cutoff" times due to rotation.
8865 * journald now understands SIGUSR1 and SIGUSR2 for triggering
8866 immediately flushing of runtime logs to /var if possible,
8867 resp. for triggering immediate rotation of the journal
8870 * It is now considered an error if a service is attempted to
8871 be stopped that is not loaded.
8873 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
8875 * systemd-analyze now supports Python 3
8877 * tmpfiles now supports cleaning up directories via aging
8878 where the first level dirs are always kept around but
8879 directories beneath it automatically aged. This is enabled
8880 by prefixing the age field with '~'.
8882 * Seat objects now expose CanGraphical, CanTTY properties
8883 which is required to deal with very fast bootups where the
8884 display manager might be running before the graphics drivers
8885 completed initialization.
8887 * Seat objects now expose a State property.
8889 * We now include RPM macros for service enabling/disabling
8890 based on the preset logic. We recommend RPM based
8891 distributions to make use of these macros if possible. This
8892 makes it simpler to reuse RPM spec files across
8895 * We now make sure that the collected systemd unit name is
8896 always valid when services log to the journal via
8899 * There's a new man page kernel-command-line(7) detailing all
8900 command line options we understand.
8902 * The fstab generator may now be disabled at boot by passing
8903 fstab=0 on the kernel command line.
8905 * A new kernel command line option modules-load= is now understood
8906 to load a specific kernel module statically, early at boot.
8908 * Unit names specified on the systemctl command line are now
8909 automatically escaped as needed. Also, if file system or
8910 device paths are specified they are automatically turned
8911 into the appropriate mount or device unit names. Example:
8913 systemctl status /home
8914 systemctl status /dev/sda
8916 * The SysVConsole= configuration option has been removed from
8917 system.conf parsing.
8919 * The SysV search path is no longer exported on the D-Bus
8922 * The Names= option has been removed from unit file parsing.
8924 * There's a new man page bootup(7) detailing the boot process.
8926 * Every unit and every generator we ship with systemd now
8927 comes with full documentation. The self-explanatory boot is
8930 * A couple of services gained "systemd-" prefixes in their
8931 name if they wrap systemd code, rather than only external
8932 code. Among them fsck@.service which is now
8933 systemd-fsck@.service.
8935 * The HaveWatchdog property has been removed from the D-Bus
8938 * systemd.confirm_spawn= on the kernel command line should now
8941 * There's a new man page crypttab(5) which details all options
8942 we actually understand.
8944 * systemd-nspawn gained a new --capability= switch to pass
8945 additional capabilities to the container.
8947 * timedated will now read known NTP implementation unit names
8948 from /usr/lib/systemd/ntp-units.d/*.list,
8949 systemd-timedated-ntp.target has been removed.
8951 * journalctl gained a new switch "-b" that lists log data of
8952 the current boot only.
8954 * The notify socket is in the abstract namespace again, in
8955 order to support daemons which chroot() at start-up.
8957 * There is a new Storage= configuration option for journald
8958 which allows configuration of where log data should go. This
8959 also provides a way to disable journal logging entirely, so
8960 that data collected is only forwarded to the console, the
8961 kernel log buffer or another syslog implementation.
8963 * Many bugfixes and optimizations
8965 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
8966 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
8967 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
8968 Shawn Landden, Tom Gundersen
8972 * "systemctl help <unit>" now shows the man page if one is
8975 * Several new man pages have been added.
8977 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
8978 MaxLevelConsole= can now be specified in
8979 journald.conf. These options allow reducing the amount of
8980 data stored on disk or forwarded by the log level.
8982 * TimerSlackNSec= can now be specified in system.conf for
8983 PID1. This allows system-wide power savings.
8985 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
8986 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
8991 * logind is now capable of (optionally) handling power and
8992 sleep keys as well as the lid switch.
8994 * journalctl now understands the syntax "journalctl
8995 /usr/bin/avahi-daemon" to get all log output of a specific
8998 * CapabilityBoundingSet= in system.conf now also influences
8999 the capability bound set of usermode helpers of the kernel.
9001 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
9002 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
9003 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
9004 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
9008 * Note that we skipped 139 releases here in order to set the
9009 new version to something that is greater than both udev's
9010 and systemd's most recent version number.
9012 * udev: all udev sources are merged into the systemd source tree now.
9013 All future udev development will happen in the systemd tree. It
9014 is still fully supported to use the udev daemon and tools without
9015 systemd running, like in initramfs or other init systems. Building
9016 udev though, will require the *build* of the systemd tree, but
9017 udev can be properly *run* without systemd.
9019 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
9020 should be used to create dead device nodes as workarounds for broken
9023 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
9024 no longer supported. udev_monitor_new_from_netlink() needs to be
9025 used to subscribe to events.
9027 * udev: when udevd is started by systemd, processes which are left
9028 behind by forking them off of udev rules, are unconditionally cleaned
9029 up and killed now after the event handling has finished. Services or
9030 daemons must be started as systemd services. Services can be
9031 pulled-in by udev to get started, but they can no longer be directly
9032 forked by udev rules.
9034 * udev: the daemon binary is called systemd-udevd now and installed
9035 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
9036 to adapt to that, create symlink, or rename the binary after building
9039 * libudev no longer provides these symbols:
9040 udev_monitor_from_socket()
9041 udev_queue_get_failed_list_entry()
9042 udev_get_{dev,sys,run}_path()
9043 The versions number was bumped and symbol versioning introduced.
9045 * systemd-loginctl and systemd-journalctl have been renamed
9046 to loginctl and journalctl to match systemctl.
9048 * The config files: /etc/systemd/systemd-logind.conf and
9049 /etc/systemd/systemd-journald.conf have been renamed to
9050 logind.conf and journald.conf. Package updates should rename
9051 the files to the new names on upgrade.
9053 * For almost all files the license is now LGPL2.1+, changed
9054 from the previous GPL2.0+. Exceptions are some minor stuff
9055 of udev (which will be changed to LGPL2.1 eventually, too),
9056 and the MIT licensed sd-daemon.[ch] library that is suitable
9057 to be used as drop-in files.
9059 * systemd and logind now handle system sleep states, in
9060 particular suspending and hibernating.
9062 * logind now implements a sleep/shutdown/idle inhibiting logic
9063 suitable for a variety of uses. Soonishly Lennart will blog
9064 about this in more detail.
9066 * var-run.mount and var-lock.mount are no longer provided
9067 (which previously bind mounted these directories to their new
9068 places). Distributions which have not converted these
9069 directories to symlinks should consider stealing these files
9070 from git history and add them downstream.
9072 * We introduced the Documentation= field for units and added
9073 this to all our shipped units. This is useful to make it
9074 easier to explore the boot and the purpose of the various
9077 * All smaller setup units (such as
9078 systemd-vconsole-setup.service) now detect properly if they
9079 are run in a container and are skipped when
9080 appropriate. This guarantees an entirely noise-free boot in
9081 Linux container environments such as systemd-nspawn.
9083 * A framework for implementing offline system updates is now
9084 integrated, for details see:
9085 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
9087 * A new service type Type=idle is available now which helps us
9088 avoiding ugly interleaving of getty output and boot status
9091 * There's now a system-wide CapabilityBoundingSet= option to
9092 globally reduce the set of capabilities for the
9093 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
9094 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
9095 even CAP_NET_ADMIN system-wide for secure systems.
9097 * There are now system-wide DefaultLimitXXX= options to
9098 globally change the defaults of the various resource limits
9099 for all units started by PID 1.
9101 * Harald Hoyer's systemd test suite has been integrated into
9102 systemd which allows easy testing of systemd builds in qemu
9103 and nspawn. (This is really awesome! Ask us for details!)
9105 * The fstab parser is now implemented as generator, not inside
9108 * systemctl will now warn you if .mount units generated from
9109 /etc/fstab are out of date due to changes in fstab that
9110 have not been read by systemd yet.
9112 * systemd is now suitable for usage in initrds. Dracut has
9113 already been updated to make use of this. With this in place
9114 initrds get a slight bit faster but primarily are much
9115 easier to introspect and debug since "systemctl status" in
9116 the host system can be used to introspect initrd services,
9117 and the journal from the initrd is kept around too.
9119 * systemd-delta has been added, a tool to explore differences
9120 between user/admin configuration and vendor defaults.
9122 * PrivateTmp= now affects both /tmp and /var/tmp.
9124 * Boot time status messages are now much prettier and feature
9125 proper english language. Booting up systemd has never been
9128 * Read-ahead pack files now include the inode number of all
9129 files to pre-cache. When the inode changes the pre-caching
9130 is not attempted. This should be nicer to deal with updated
9131 packages which might result in changes of read-ahead
9134 * We now temporaritly lower the kernel's read_ahead_kb variable
9135 when collecting read-ahead data to ensure the kernel's
9136 built-in read-ahead does not add noise to our measurements
9137 of necessary blocks to pre-cache.
9139 * There's now RequiresMountsFor= to add automatic dependencies
9140 for all mounts necessary for a specific file system path.
9142 * MountAuto= and SwapAuto= have been removed from
9143 system.conf. Mounting file systems at boot has to take place
9146 * nspawn now learned a new switch --uuid= to set the machine
9147 ID on the command line.
9149 * nspawn now learned the -b switch to automatically search
9152 * vt102 is now the default TERM for serial TTYs, upgraded from
9155 * systemd-logind now works on VT-less systems.
9157 * The build tree has been reorganized. The individual
9158 components now have directories of their own.
9160 * A new condition type ConditionPathIsReadWrite= is now available.
9162 * nspawn learned the new -C switch to create cgroups for the
9163 container in other hierarchies.
9165 * We now have support for hardware watchdogs, configurable in
9168 * The scheduled shutdown logic now has a public API.
9170 * We now mount /tmp as tmpfs by default, but this can be
9171 masked and /etc/fstab can override it.
9173 * Since udisks does not make use of /media anymore we are not
9174 mounting a tmpfs on it anymore.
9176 * journalctl gained a new --local switch to only interleave
9177 locally generated journal files.
9179 * We can now load the IMA policy at boot automatically.
9181 * The GTK tools have been split off into a systemd-ui.
9183 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
9184 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
9185 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
9186 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
9187 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
9188 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
9189 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
9190 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
9191 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
9196 * This is mostly a bugfix release
9198 * Support optional initialization of the machine ID from the
9199 KVM or container configured UUID.
9201 * Support immediate reboots with "systemctl reboot -ff"
9203 * Show /etc/os-release data in systemd-analyze output
9205 * Many bugfixes for the journal, including endianness fixes and
9206 ensuring that disk space enforcement works
9208 * sd-login.h is C++ compatible again
9210 * Extend the /etc/os-release format on request of the Debian
9213 * We now refuse non-UTF8 strings used in various configuration
9214 and unit files. This is done to ensure we do not pass invalid
9215 data over D-Bus or expose it elsewhere.
9217 * Register Mimo USB Screens as suitable for automatic seat
9220 * Read SELinux client context from journal clients in a race
9223 * Reorder configuration file lookup order. /etc now always
9224 overrides /run in order to allow the administrator to always
9225 and unconditionally override vendor-supplied or
9226 automatically generated data.
9228 * The various user visible bits of the journal now have man
9229 pages. We still lack man pages for the journal API calls
9232 * We now ship all man pages in HTML format again in the
9235 Contributions from: Dave Reisner, Dirk Eibach, Frederic
9236 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
9237 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
9242 * This is mostly a bugfix release
9244 * systems lacking /etc/os-release are no longer supported.
9246 * Various functionality updates to libsystemd-login.so
9248 * Track class of PAM logins to distinguish greeters from
9251 Contributions from: Kay Sievers, Lennart Poettering, Michael
9256 * This is an important bugfix release for v41.
9258 * Building man pages is now optional which should be useful
9259 for those building systemd from git but unwilling to install
9262 * Watchdog support for supervising services is now usable. In
9263 a future release support for hardware watchdogs
9264 (i.e. /dev/watchdog) will be added building on this.
9266 * Service start rate limiting is now configurable and can be
9267 turned off per service. When a start rate limit is hit a
9268 reboot can automatically be triggered.
9270 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
9272 Contributions from: Benjamin Franzke, Bill Nottingham,
9273 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
9274 Schmidt, Michał Górny, Piotr Drąg
9278 * The systemd binary is installed /usr/lib/systemd/systemd now;
9279 An existing /sbin/init symlink needs to be adapted with the
9282 * The code that loads kernel modules has been ported to invoke
9283 libkmod directly, instead of modprobe. This means we do not
9284 support systems with module-init-tools anymore.
9286 * Watchdog support is now already useful, but still not
9289 * A new kernel command line option systemd.setenv= is
9290 understood to set system wide environment variables
9291 dynamically at boot.
9293 * We now limit the set of capabilities of systemd-journald.
9295 * We now set SIGPIPE to ignore by default, since it only is
9296 useful in shell pipelines, and has little use in general
9297 code. This can be disabled with IgnoreSIPIPE=no in unit
9300 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
9301 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
9306 * This is mostly a bugfix release
9308 * We now expose the reason why a service failed in the
9309 "Result" D-Bus property.
9311 * Rudimentary service watchdog support (will be completed over
9312 the next few releases.)
9314 * When systemd forks off in order execute some service we will
9315 now immediately changes its argv[0] to reflect which process
9316 it will execute. This is useful to minimize the time window
9317 with a generic argv[0], which makes bootcharts more useful
9319 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
9320 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
9321 Mike Kazantsev, Ray Strode
9325 * This is mostly a test release, but incorporates many
9328 * New systemd-cgtop tool to show control groups by their
9331 * Linking against libacl for ACLs is optional again. If
9332 disabled, support tracking device access for active logins
9333 goes becomes unavailable, and so does access to the user
9334 journals by the respective users.
9336 * If a group "adm" exists, journal files are automatically
9337 owned by them, thus allow members of this group full access
9338 to the system journal as well as all user journals.
9340 * The journal now stores the SELinux context of the logging
9341 client for all entries.
9343 * Add C++ inclusion guards to all public headers
9345 * New output mode "cat" in the journal to print only text
9346 messages, without any meta data like date or time.
9348 * Include tiny X server wrapper as a temporary stop-gap to
9349 teach XOrg udev display enumeration. This is used by display
9350 managers such as gdm, and will go away as soon as XOrg
9351 learned native udev hotplugging for display devices.
9353 * Add new systemd-cat tool for executing arbitrary programs
9354 with STDERR/STDOUT connected to the journal. Can also act as
9355 BSD logger replacement, and does so by default.
9357 * Optionally store all locally generated coredumps in the
9358 journal along with meta data.
9360 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
9361 writing short strings to files (for usage for /sys), and for
9362 creating symlinks, character and block device nodes.
9364 * New unit file option ControlGroupPersistent= to make cgroups
9365 persistent, following the mechanisms outlined in
9366 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
9368 * Support multiple local RTCs in a sane way
9370 * No longer monopolize IO when replaying readahead data on
9371 rotating disks, since we might starve non-file-system IO to
9372 death, since fanotify() will not see accesses done by blkid,
9375 * Do not show kernel threads in systemd-cgls anymore, unless
9376 requested with new -k switch.
9378 Contributions from: Dan Horák, Kay Sievers, Lennart
9379 Poettering, Michal Schmidt
9383 * This is mostly a test release, but incorporates many
9386 * The git repository moved to:
9387 git://anongit.freedesktop.org/systemd/systemd
9388 ssh://git.freedesktop.org/git/systemd/systemd
9390 * First release with the journal
9391 http://0pointer.de/blog/projects/the-journal.html
9393 * The journal replaces both systemd-kmsg-syslogd and
9394 systemd-stdout-bridge.
9396 * New sd_pid_get_unit() API call in libsystemd-logind
9398 * Many systemadm clean-ups
9400 * Introduce remote-fs-pre.target which is ordered before all
9401 remote mounts and may be used to start services before all
9404 * Added Mageia support
9406 * Add bash completion for systemd-loginctl
9408 * Actively monitor PID file creation for daemons which exit in
9409 the parent process before having finished writing the PID
9410 file in the daemon process. Daemons which do this need to be
9411 fixed (i.e. PID file creation must have finished before the
9412 parent exits), but we now react a bit more gracefully to them.
9414 * Add colourful boot output, mimicking the well-known output
9415 of existing distributions.
9417 * New option PassCredentials= for socket units, for
9418 compatibility with a recent kernel ABI breakage.
9420 * /etc/rc.local is now hooked in via a generator binary, and
9421 thus will no longer act as synchronization point during
9424 * systemctl list-unit-files now supports --root=.
9426 * systemd-tmpfiles now understands two new commands: z, Z for
9427 relabelling files according to the SELinux database. This is
9428 useful to apply SELinux labels to specific files in /sys,
9431 * Output of SysV services is now forwarded to both the console
9432 and the journal by default, not only just the console.
9434 * New man pages for all APIs from libsystemd-login.
9436 * The build tree got reorganized and the build system is a
9437 lot more modular allowing embedded setups to specifically
9438 select the components of systemd they are interested in.
9440 * Support for Linux systems lacking the kernel VT subsystem is
9443 * configure's --with-rootdir= got renamed to
9444 --with-rootprefix= to follow the naming used by udev and
9447 * Unless specified otherwise we will now install to /usr instead
9448 of /usr/local by default.
9450 * Processes with '@' in argv[0][0] are now excluded from the
9451 final shut-down killing spree, following the logic explained
9453 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
9455 * All processes remaining in a service cgroup when we enter
9456 the START or START_PRE states are now killed with
9457 SIGKILL. That means it is no longer possible to spawn
9458 background processes from ExecStart= lines (which was never
9459 supported anyway, and bad style).
9461 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
9462 reloading of units together.
9464 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
9465 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
9466 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
9467 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
9468 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek