1 systemd System and Service Manager
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 * .timer units gained two new boolean settings OnClockChange= and
64 OnTimezoneChange= which may be used to also trigger a unit when the
65 system clock is changed or the local timezone is
66 modified. systemd-run has been updated to make these options easily
67 accessible from the command line for transient timers.
69 * Two new conditions for units have been added: ConditionMemory= may be
70 used to conditionalize a unit based on installed system
71 RAM. ConditionCPUs= may be used to conditionalize a unit based on
74 * The @default system call filter group understood by SystemCallFilter=
75 has been updated to include the new rseq() system call introduced in
78 * "systemctl start" (and related commands) learnt a new
79 --show-transaction option. If specified brief information about all
80 jobs queued because of the requested operation is shown.
82 * systemd-networkd recognizes a new operation state 'enslaved', used
83 (instead of 'degraded' or 'carrier') for interfaces which form a
84 bridge, bond, or similar, and an new 'degraded-carrier' operational
85 state used for the bond or bridge master interface when one of the
86 enslaved devices is not operational.
88 * .network files learnt the new IgnoreCarrierLoss= option for leaving
89 networks configured even if the carrier is lost.
91 * The RequiredForOnline= setting in .network files may now specify a
92 minimum operational state required for the interface to be considered
93 "online" by systemd-networkd-wait-online. Related to this
94 systemd-networkd-wait-online gained a new option --operational-state=
95 to configure the same, and its --interface= option was updated to
96 optionally also take an operational state specific for an interface.
98 * systemd-networkd-wait-online gained a new setting --any for waiting
99 for only one of the requested interfaces instead of all of them.
101 * systemd-networkd now implements L2TP tunnels.
103 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
104 may be used to cause autonomous and onlink prefixes received in IPv6
105 Router Advertisements to be ignored.
107 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
108 file settings may be used to tweak bridge behaviour.
110 * The new TripleSampling= option in .network files may be used to
111 configure CAN triple sampling.
113 * A new .netdev setting PrivateKeyFile= may be used to point to private
114 key for a WireGuard interface.
116 * /etc/crypttab now supports the same-cpu-crypt and
117 submit-from-crypt-cpus options to tweak encryption work scheduling
120 * systemd-tmpfiles will now take a BSD file lock before operating on a
121 contents of directory. This may be used to temporarily exclude
122 directories from aging by taking the same lock (useful for example
123 when extracting a tarball into /tmp or /var/tmp as a privileged user,
124 which might create files with really old timestamps, which
125 nevertheless should not be deleted). For further details, see:
127 https://systemd.io/TEMPORARY_DIRECTORIES
129 * systemd-tmpfiles' h line type gained support for the
130 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
131 controlling project quota inheritance.
133 * sd-boot and bootctl now implement support for an Extended Boot Loader
134 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
135 addition to the ESP partition mounted to /efi or /boot/efi.
136 Configuration file fragments, kernels, initrds and other EFI images
137 to boot will be loaded from both the ESP and XBOOTLDR partitions.
138 The XBOOTLDR partition was previously described by the Boot Loader
139 Specification, but implementation was missing in sd-boot. Support for
140 this concept allows using the sd-boot boot loader in more
141 conservative scenarios where the boot loader itself is placed in the
142 ESP but the kernels to boot (and their metadata) in a separate
145 * A system may now be booted with systemd.volatile=overlay on the
146 kernel command line, which causes the root file system to be set up
147 an overlayfs mount combining the root-only root directory with a
148 writable tmpfs. In this setup, the underlying root device is not
149 modified, and any changes are lost at reboot.
151 * Similar, systemd-nspawn can now boot containers with a volatile
152 overlayfs root with the new --volatile=overlay switch.
154 * systemd-nspawn can now consume OCI runtime bundles using a new
155 --oci-bundle= option. This implementation is fully usable, with most
156 features in the specification implemented, but since this a lot of
157 new code and functionality, this feature should most likely not
158 be used in production yet.
160 * systemd-nspawn now supports various options described by the OCI
161 runtime specification on the command-line and in .nspawn files:
162 --inaccessible=/Inaccessible= may be used to mask parts of the file
163 system tree, --console=/--pipe may be used to configure how standard
164 input, output, and error are set up.
166 * busctl learned the `emit` verb to generate D-Bus signals.
168 * systemd-analyze cat-config may be used to gather and display
169 configuration spread over multiple files, for example system and user
170 presets, tmpfiles.d, sysusers.d, udev rules, etc.
172 * systemd-analyze calendar now takes an optional new parameter
173 --iterations= which may be used to show a maximum number of iterations
174 the specified expression will elapse next.
176 * The sd-bus C API gained support for naming method parameters in the
179 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
180 the reboot() system call expects.
182 * journalctl learnt a new --cursor-file= option that points to a file
183 from which a cursor should be loaded in the beginning and to which
184 the updated cursor should be stored at the end.
186 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
187 detected by systemd-detect-virt (and may also be used in
188 ConditionVirtualization=).
190 * The behaviour of systemd-logind may now be modified with environment
191 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
192 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
193 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
194 skip the relevant operation completely (when set to false), or to
195 create a flag file in /run/systemd (when set to true), instead of
196 actually commencing the real operation when requested. The presence
197 of /run/systemd/reboot-to-firmware-setup,
198 /run/systemd/reboot-to-boot-loader-menu, and
199 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
200 boot loader implementations to replace some steps logind performs
201 during reboot with their own operations.
203 * systemctl can be used to request a reboot into the boot loader menu
204 or a specific boot loader entry with the new --boot-load-menu= and
205 --boot-loader-entry= options to a reboot command. (This requires a
206 boot loader that supports this, for example sd-boot.)
208 * kernel-install will no longer unconditionally create the output
209 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
210 snippets, but will do only if the machine-specific parent directory
211 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
212 to create this parent directory during sd-boot installation.
214 This makes it easier to use kernel-install with plugins which support
215 a different layout of the bootloader partitions (for example grub2).
217 * During package installation (with `ninja install`), we would create
218 symlinks for systemd-networkd.service, systemd-networkd.socket,
219 systemd-resolved.service, remote-cryptsetup.target, remote-fs.target,
220 systemd-networkd-wait-online.service, and systemd-timesyncd.service
221 in /etc, as if `systemctl enable` was called for those units, to make
222 the system usable immediately after installation. Now this is not
223 done anymore, and instead calling `systemctl preset-all` is
224 recommended after the first installation of systemd.
226 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
227 is built on seccomp. When turned on creation of SUID/SGID files is
230 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
231 implied if DynamicUser= is turned on for a service. This hardens
232 these services, so that they neither can benefit from nor create
233 SUID/SGID executables. This is a minor compatibility breakage, given
234 that when DynamicUser= was first introduced SUID/SGID behaviour was
235 unaffected. However, the security benefit of these two options is
236 substantial, and the setting is still relatively new, hence we opted
237 to make it mandatory for services with dynamic users.
239 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
240 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
241 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
242 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
243 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
244 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
245 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
246 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
247 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
248 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
249 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
250 Sekletar, Mike Lothian, Piotr Drąg, Riccardo Schirone, Robin Elvedi,
251 Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan Gonzalez, Stephane
252 Chazelas, StKob, Susant Sahani, Sylvain Plantefève, Szabolcs Fruhwald,
253 Taro Yamada, Theo Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan,
254 Tony Asleson, Topi Miettinen, unixsysadmin, Van Laser, Vesa
255 Jääskeläinen, Yu, Li-Yu, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
261 * The default locale can now be configured at compile time. Otherwise,
262 a suitable default will be selected automatically (one of C.UTF-8,
265 * The version string shown by systemd and other tools now includes the
266 git commit hash when built from git. An override may be specified
267 during compilation, which is intended to be used by distributions to
268 include the package release information.
270 * systemd-cat can now filter standard input and standard error streams
271 for different syslog priorities using the new --stderr-priority=
274 * systemd-journald and systemd-journal-remote reject entries which
275 contain too many fields (CVE-2018-16865) and set limits on the
276 process' command line length (CVE-2018-16864).
278 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
281 * A new network device NamePolicy "keep" is implemented for link files,
282 and used by default in 99-default.link (the fallback configuration
283 provided by systemd). With this policy, if the network device name
284 was already set by userspace, the device will not be renamed again.
285 This matches the naming scheme that was implemented before
286 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
287 is also enabled by default, even if not specified. Effectively, this
288 means that if naming-scheme >= 240 is specified, network devices will
289 be renamed according to the configuration, even if they have been
290 renamed already, if "keep" is not specified as the naming policy in
291 the .link file. The 99-default.link file provided by systemd includes
292 "keep" for backwards compatibility, but it is recommended for user
293 installed .link files to *not* include it.
295 The "kernel" policy, which keeps kernel names declared to be
296 "persistent", now works again as documented.
298 * kernel-install script now optionally takes the paths to one or more
299 initrd files, and passes them to all plugins.
301 * The mincore() system call has been dropped from the @system-service
302 system call filter group, as it is pretty exotic and may potentially
303 used for side-channel attacks.
305 * -fPIE is dropped from compiler and linker options. Please specify
306 -Db_pie=true option to meson to build position-independent
307 executables. Note that the meson option is supported since meson-0.49.
309 * The fs.protected_regular and fs.protected_fifos sysctls, which were
310 added in Linux 4.19 to make some data spoofing attacks harder, are
311 now enabled by default. While this will hopefully improve the
312 security of most installations, it is technically a backwards
313 incompatible change; to disable these sysctls again, place the
314 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
316 fs.protected_regular = 0
317 fs.protected_fifos = 0
319 Note that the similar hardlink and symlink protection has been
320 enabled since v199, and may be disabled likewise.
322 * The files read from the EnvironmentFile= setting in unit files now
323 parse backslashes inside quotes literally, matching the behaviour of
326 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
327 now automatically become NOPs when run in a chroot() environment.
329 * The tmpfiles.d/ "C" line type will now copy directory trees not only
330 when the destination is so far missing, but also if it already exists
331 as a directory and is empty. This is useful to cater for systems
332 where directory trees are put together from multiple separate mount
333 points but otherwise empty.
335 * A new function sd_bus_close_unref() (and the associated
336 sd_bus_close_unrefp()) has been added to libsystemd, that combines
337 sd_bus_close() and sd_bus_unref() in one.
339 * udevadm control learnt a new option for --ping for testing whether a
340 systemd-udevd instance is running and reacting.
342 * udevadm trigger learnt a new option for --wait-daemon for waiting
343 systemd-udevd daemon to be initialized.
345 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
346 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
347 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
348 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
349 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
350 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
351 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
352 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
353 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
354 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
355 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
356 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
357 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
358 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
359 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
360 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
361 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
367 * NoNewPrivileges=yes has been set for all long-running services
368 implemented by systemd. Previously, this was problematic due to
369 SELinux (as this would also prohibit the transition from PID1's label
370 to the service's label). This restriction has since been lifted, but
371 an SELinux policy update is required.
372 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
374 * DynamicUser=yes is dropped from systemd-networkd.service,
375 systemd-resolved.service and systemd-timesyncd.service, which was
376 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
377 and since v236 for systemd-timesyncd.service. The users and groups
378 systemd-network, systemd-resolve and systemd-timesync are created
379 by systemd-sysusers again. Distributors or system administrators
380 may need to create these users and groups if they not exist (or need
381 to re-enable DynamicUser= for those units) while upgrading systemd.
382 Also, the clock file for systemd-timesyncd may need to move from
383 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
385 * When unit files are loaded from disk, previously systemd would
386 sometimes (depending on the unit loading order) load units from the
387 target path of symlinks in .wants/ or .requires/ directories of other
388 units. This meant that unit could be loaded from different paths
389 depending on whether the unit was requested explicitly or as a
390 dependency of another unit, not honouring the priority of directories
391 in search path. It also meant that it was possible to successfully
392 load and start units which are not found in the unit search path, as
393 long as they were requested as a dependency and linked to from
394 .wants/ or .requires/. The target paths of those symlinks are not
395 used for loading units anymore and the unit file must be found in
398 * A new service type has been added: Type=exec. It's very similar to
399 Type=simple but ensures the service manager will wait for both fork()
400 and execve() of the main service binary to complete before proceeding
401 with follow-up units. This is primarily useful so that the manager
402 propagates any errors in the preparation phase of service execution
403 back to the job that requested the unit to be started. For example,
404 consider a service that has ExecStart= set to a file system binary
405 that doesn't exist. With Type=simple starting the unit would be
406 considered instantly successful, as only fork() has to complete
407 successfully and the manager does not wait for execve(), and hence
408 its failure is seen "too late". With the new Type=exec service type
409 starting the unit will fail, as the manager will wait for the
410 execve() and notice its failure, which is then propagated back to the
413 NOTE: with the next release 241 of systemd we intend to change the
414 systemd-run tool to default to Type=exec for transient services
415 started by it. This should be mostly safe, but in specific corner
416 cases might result in problems, as the systemd-run tool will then
417 block on NSS calls (such as user name look-ups due to User=) done
418 between the fork() and execve(), which under specific circumstances
419 might cause problems. It is recommended to specify "-p Type=simple"
420 explicitly in the few cases where this applies. For regular,
421 non-transient services (i.e. those defined with unit files on disk)
422 we will continue to default to Type=simple.
424 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
425 userspace processes is set to 1024 (soft) and 4096
426 (hard). Previously, systemd passed this on unmodified to all
427 processes it forked off. With this systemd release the hard limit
428 systemd passes on is increased to 512K, overriding the kernel's
429 defaults and substantially increasing the number of simultaneous file
430 descriptors unprivileged userspace processes can allocate. Note that
431 the soft limit remains at 1024 for compatibility reasons: the
432 traditional UNIX select() call cannot deal with file descriptors >=
433 1024 and increasing the soft limit globally might thus result in
434 programs unexpectedly allocating a high file descriptor and thus
435 failing abnormally when attempting to use it with select() (of
436 course, programs shouldn't use select() anymore, and prefer
437 poll()/epoll, but the call unfortunately remains undeservedly popular
438 at this time). This change reflects the fact that file descriptor
439 handling in the Linux kernel has been optimized in more recent
440 kernels and allocating large numbers of them should be much cheaper
441 both in memory and in performance than it used to be. Programs that
442 want to take benefit of the increased limit have to "opt-in" into
443 high file descriptors explicitly by raising their soft limit. Of
444 course, when they do that they must acknowledge that they cannot use
445 select() anymore (and neither can any shared library they use — or
446 any shared library used by any shared library they use and so on).
447 Which default hard limit is most appropriate is of course hard to
448 decide. However, given reports that ~300K file descriptors are used
449 in real-life applications we believe 512K is sufficiently high as new
450 default for now. Note that there are also reports that using very
451 high hard limits (e.g. 1G) is problematic: some software allocates
452 large arrays with one element for each potential file descriptor
453 (Java, …) — a high hard limit thus triggers excessively large memory
454 allocations in these applications. Hopefully, the new default of 512K
455 is a good middle ground: higher than what real-life applications
456 currently need, and low enough for avoid triggering excessively large
457 allocations in problematic software. (And yes, somebody should fix
460 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
461 to the highest possible values, as separate accounting of file
462 descriptors is no longer necessary, as memcg tracks them correctly as
463 part of the memory accounting anyway. Thus, from the four limits on
464 file descriptors currently enforced (fs.file-max, fs.nr_open,
465 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
466 and keep only the latter two. A set of build-time options
467 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
468 has been added to revert this change in behaviour, which might be
469 an option for systems that turn off memcg in the kernel.
471 * When no /etc/locale.conf file exists (and hence no locale settings
472 are in place), systemd will now use the "C.UTF-8" locale by default,
473 and set LANG= to it. This locale is supported by various
474 distributions including Fedora, with clear indications that upstream
475 glibc is going to make it available too. This locale enables UTF-8
476 mode by default, which appears appropriate for 2018.
478 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
479 default. This effectively switches the RFC3704 Reverse Path filtering
480 from Strict mode to Loose mode. This is more appropriate for hosts
481 that have multiple links with routes to the same networks (e.g.
482 a client with a Wi-Fi and Ethernet both connected to the internet).
484 Consult the kernel documentation for details on this sysctl:
485 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
487 * CPUAccounting=yes no longer enables the CPU controller when using
488 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
489 statistics are now provided independently from the CPU controller.
491 * Support for disabling a particular cgroup controller within a sub-tree
492 has been added through the DisableControllers= directive.
494 * cgroup_no_v1=all on the kernel command line now also implies
495 using the unified cgroup hierarchy, unless one explicitly passes
496 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
498 * The new "MemoryMin=" unit file property may now be used to set the
499 memory usage protection limit of processes invoked by the unit. This
500 controls the cgroup v2 memory.min attribute. Similarly, the new
501 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
502 cgroup v2 io.latency cgroup property for configuring per-service I/O
505 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
506 to the cgroup v1 "devices" cgroup controller.
508 * systemd-escape now is able to combine --unescape with --template. It
509 also learnt a new option --instance for extracting and unescaping the
510 instance part of a unit name.
512 * sd-bus now provides the sd_bus_message_readv() which is similar to
513 sd_bus_message_read() but takes a va_list object. The pair
514 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
515 has been added for configuring the default method call timeout to
516 use. sd_bus_error_move() may be used to efficiently move the contents
517 from one sd_bus_error structure to another, invalidating the
518 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
519 be used to control whether a bus connection object is automatically
520 flushed when an sd-event loop is exited.
522 * When processing classic BSD syslog log messages, journald will now
523 save the original time-stamp string supplied in the new
524 SYSLOG_TIMESTAMP= journal field. This permits consumers to
525 reconstruct the original BSD syslog message more correctly.
527 * StandardOutput=/StandardError= in service files gained support for
528 new "append:…" parameters, for connecting STDOUT/STDERR of a service
529 to a file, and appending to it.
531 * The signal to use as last step of killing of unit processes is now
532 configurable. Previously it was hard-coded to SIGKILL, which may now
533 be overridden with the new KillSignal= setting. Note that this is the
534 signal used when regular termination (i.e. SIGTERM) does not suffice.
535 Similarly, the signal used when aborting a program in case of a
536 watchdog timeout may now be configured too (WatchdogSignal=).
538 * The XDG_SESSION_DESKTOP environment variable may now be configured in
539 the pam_systemd argument line, using the new desktop= switch. This is
540 useful to initialize it properly from a display manager without
541 having to touch C code.
543 * Most configuration options that previously accepted percentage values
544 now also accept permille values with the '‰' suffix (instead of '%').
546 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
549 * systemd-resolved's configuration file resolved.conf gained a new
550 option ReadEtcHosts= which may be used to turn off processing and
551 honoring /etc/hosts entries.
553 * The "--wait" switch may now be passed to "systemctl
554 is-system-running", in which case the tool will synchronously wait
555 until the system finished start-up.
557 * hostnamed gained a new bus call to determine the DMI product UUID.
559 * On x86-64 systemd will now prefer using the RDRAND processor
560 instruction over /dev/urandom whenever it requires randomness that
561 neither has to be crypto-grade nor should be reproducible. This
562 should substantially reduce the amount of entropy systemd requests
563 from the kernel during initialization on such systems, though not
564 reduce it to zero. (Why not zero? systemd still needs to allocate
565 UUIDs and such uniquely, which require high-quality randomness.)
567 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
568 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
569 for forcing the "Other Information" bit in IPv6 RA messages. The
570 bonding logic gained four new options AdActorSystemPriority=,
571 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
572 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
573 shuffling of flows. The tunnel logic gained a new
574 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
575 Deployment. The policy rule logic gained four new options IPProtocol=,
576 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
577 support for the MulticastToUnicast= option. networkd also gained
578 support for configuring static IPv4 ARP or IPv6 neighbor entries.
580 * .preset files (as read by 'systemctl preset') may now be used to
581 instantiate services.
583 * /etc/crypttab now understands the sector-size= option to configure
584 the sector size for an encrypted partition.
586 * Key material for encrypted disks may now be placed on a formatted
587 medium, and referenced from /etc/crypttab by the UUID of the file
588 system, followed by "=" suffixed by the path to the key file.
590 * The "collect" udev component has been removed without replacement, as
591 it is neither used nor maintained.
593 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
594 LogsDirectory=, ConfigurationDirectory= settings are used in a
595 service the executed processes will now receive a set of environment
596 variables containing the full paths of these directories.
597 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
598 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
599 are used. Note that these options may be used multiple times per
600 service in which case the resulting paths will be concatenated and
603 * Predictable interface naming has been extended to cover InfiniBand
604 NICs. They will be exposed with an "ib" prefix.
606 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
607 which case the respective line failing is ignored.
609 * .link files may now be used to configure the equivalent to the
610 "ethtool advertise" commands.
612 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
613 alternative to libudev.h. Previously, the latter was just an internal
614 wrapper around the former, but now these two APIs are exposed
617 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
618 which calculates an app-specific boot ID similar to how
619 sd_id128_get_machine_app_specific() generates an app-specific machine
622 * A new tool systemd-id128 has been added that can be used to determine
623 and generate various 128bit IDs.
625 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
628 * systemd-hibernate-resume-generator will now honor the "noresume"
629 kernel command line option, in which case it will bypass resuming
630 from any hibernated image.
632 * The systemd-sleep.conf configuration file gained new options
633 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
634 AllowHybridSleep= for prohibiting specific sleep modes even if the
637 * portablectl is now officially supported and has thus moved to
640 * bootctl learnt the two new commands "set-default" and "set-oneshot"
641 for setting the default boot loader item to boot to (either
642 persistently or only for the next boot). This is currently only
643 compatible with sd-boot, but may be implemented on other boot loaders
644 too, that follow the boot loader interface. The updated interface is
647 https://systemd.io/BOOT_LOADER_INTERFACE
649 * A new kernel command line option systemd.early_core_pattern= is now
650 understood which may be used to influence the core_pattern PID 1
651 installs during early boot.
653 * busctl learnt two new options -j and --json= for outputting method
654 call replies, properties and monitoring output in JSON.
656 * journalctl's JSON output now supports simple ANSI coloring as well as
657 a new "json-seq" mode for generating RFC7464 output.
659 * Unit files now support the %g/%G specifiers that resolve to the UNIX
660 group/GID of the service manager runs as, similar to the existing
661 %u/%U specifiers that resolve to the UNIX user/UID.
663 * systemd-logind learnt a new global configuration option
664 UserStopDelaySec= that may be set in logind.conf. It specifies how
665 long the systemd --user instance shall remain started after a user
666 logs out. This is useful to speed up repetitive re-connections of the
667 same user, as it means the user's service manager doesn't have to be
668 stopped/restarted on each iteration, but can be reused between
669 subsequent options. This setting defaults to 10s. systemd-logind also
670 exports two new properties on its Manager D-Bus objects indicating
671 whether the system's lid is currently closed, and whether the system
674 * systemd gained support for a generic boot counting logic, which
675 generically permits automatic reverting to older boot loader entries
676 if newer updated ones don't work. The boot loader side is implemented
677 in sd-boot, but is kept open for other boot loaders too. For details
680 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
682 * The SuccessAction=/FailureAction= unit file settings now learnt two
683 new parameters: "exit" and "exit-force", which result in immediate
684 exiting of the service manager, and are only useful in systemd --user
685 and container environments.
687 * Unit files gained support for a pair of options
688 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
689 exit status to use as service manager exit status when
690 SuccessAction=/FailureAction= is set to exit or exit-force.
692 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
693 options may now be used to configure the log rate limiting applied by
694 journald per-service.
696 * systemd-analyze gained a new verb "timespan" for parsing and
697 normalizing time span values (i.e. strings like "5min 7s 8us").
699 * systemd-analyze also gained a new verb "security" for analyzing the
700 security and sand-boxing settings of services in order to determine an
701 "exposure level" for them, indicating whether a service would benefit
702 from more sand-boxing options turned on for them.
704 * "systemd-analyze syscall-filter" will now also show system calls
705 supported by the local kernel but not included in any of the defined
708 * .nspawn files now understand the Ephemeral= setting, matching the
709 --ephemeral command line switch.
711 * sd-event gained the new APIs sd_event_source_get_floating() and
712 sd_event_source_set_floating() for controlling whether a specific
713 event source is "floating", i.e. destroyed along with the even loop
716 * Unit objects on D-Bus gained a new "Refs" property that lists all
717 clients that currently have a reference on the unit (to ensure it is
720 * The JoinControllers= option in system.conf is no longer supported, as
721 it didn't work correctly, is hard to support properly, is legacy (as
722 the concept only exists on cgroup v1) and apparently wasn't used.
724 * Journal messages that are generated whenever a unit enters the failed
725 state are now tagged with a unique MESSAGE_ID. Similarly, messages
726 generated whenever a service process exits are now made recognizable,
727 too. A taged message is also emitted whenever a unit enters the
728 "dead" state on success.
730 * systemd-run gained a new switch --working-directory= for configuring
731 the working directory of the service to start. A shortcut -d is
732 equivalent, setting the working directory of the service to the
733 current working directory of the invoking program. The new --shell
734 (or just -S) option has been added for invoking the $SHELL of the
735 caller as a service, and implies --pty --same-dir --wait --collect
736 --service-type=exec. Or in other words, "systemd-run -S" is now the
737 quickest way to quickly get an interactive in a fully clean and
738 well-defined system service context.
740 * machinectl gained a new verb "import-fs" for importing an OS tree
741 from a directory. Moreover, when a directory or tarball is imported
742 and single top-level directory found with the OS itself below the OS
743 tree is automatically mangled and moved one level up.
745 * systemd-importd will no longer set up an implicit btrfs loop-back
746 file system on /var/lib/machines. If one is already set up, it will
749 * A new generator "systemd-run-generator" has been added. It will
750 synthesize a unit from one or more program command lines included in
751 the kernel command line. This is very useful in container managers
754 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
756 This will run "systemd-nspawn" on an image, invoke the specified
757 command line and immediately shut down the container again, returning
758 the command line's exit code.
760 * The block device locking logic is now documented:
762 https://systemd.io/BLOCK_DEVICE_LOCKING
764 * loginctl and machinectl now optionally output the various tables in
765 JSON using the --output= switch. It is our intention to add similar
766 support to systemctl and all other commands.
768 * udevadm's query and trigger verb now optionally take a .device unit
771 * systemd-udevd's network naming logic now understands a new
772 net.naming-scheme= kernel command line switch, which may be used to
773 pick a specific version of the naming scheme. This helps stabilizing
774 interface names even as systemd/udev are updated and the naming logic
777 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
778 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
779 initialize one to all 0xFF.
781 * After loading the SELinux policy systemd will now recursively relabel
782 all files and directories listed in
783 /run/systemd/relabel-extra.d/*.relabel (which should be simple
784 newline separated lists of paths) in addition to the ones it already
785 implicitly relabels in /run, /dev and /sys. After the relabelling is
786 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
787 removed. This is useful to permit initrds (i.e. code running before
788 the SELinux policy is in effect) to generate files in the host
789 filesystem safely and ensure that the correct label is applied during
790 the transition to the host OS.
792 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
793 mknod() handling in user namespaces. Previously mknod() would always
794 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
795 but device nodes generated that way cannot be opened, and attempts to
796 open them result in EPERM. This breaks the "graceful fallback" logic
797 in systemd's PrivateDevices= sand-boxing option. This option is
798 implemented defensively, so that when systemd detects it runs in a
799 restricted environment (such as a user namespace, or an environment
800 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
801 where device nodes cannot be created the effect of PrivateDevices= is
802 bypassed (following the logic that 2nd-level sand-boxing is not
803 essential if the system systemd runs in is itself already sand-boxed
804 as a whole). This logic breaks with 4.18 in container managers where
805 user namespacing is used: suddenly PrivateDevices= succeeds setting
806 up a private /dev/ file system containing devices nodes — but when
807 these are opened they don't work.
809 At this point is is recommended that container managers utilizing
810 user namespaces that intend to run systemd in the payload explicitly
811 block mknod() with seccomp or similar, so that the graceful fallback
814 We are very sorry for the breakage and the requirement to change
815 container configurations for newer kernels. It's purely caused by an
816 incompatible kernel change. The relevant kernel developers have been
817 notified about this userspace breakage quickly, but they chose to
820 * PermissionsStartOnly= setting is deprecated (but is still supported
821 for backwards compatibility). The same functionality is provided by
822 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
825 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
828 * The naming scheme for network devices was changed to always rename
829 devices, even if they were already renamed by userspace. The "kernel"
830 policy was changed to only apply as a fallback, if no other naming
833 * The requirements to build systemd is bumped to meson-0.46 and
836 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
837 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
838 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
839 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
840 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
841 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
842 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
843 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
844 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
845 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
846 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
847 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
848 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
849 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
850 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
851 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
852 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
853 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
854 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
855 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
856 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
857 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
858 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
859 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
860 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
861 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
862 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
863 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
864 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
865 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
866 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
867 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
868 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
869 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
870 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
871 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
872 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
873 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
874 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
875 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
876 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
877 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
878 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
879 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
880 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
886 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
887 builtin will name network interfaces differently than in previous
888 versions for virtual network interfaces created with SR-IOV and NPAR
889 and for devices where the PCI network controller device does not have
890 a slot number associated.
892 SR-IOV virtual devices are now named based on the name of the parent
893 interface, with a suffix of "v<N>", where <N> is the virtual device
894 number. Previously those virtual devices were named as if completely
897 The ninth and later NPAR virtual devices will be named following the
898 scheme used for the first eight NPAR partitions. Previously those
899 devices were not renamed and the kernel default (eth<n>) was used.
901 "net_id" will also generate names for PCI devices where the PCI
902 network controller device does not have an associated slot number
903 itself, but one of its parents does. Previously those devices were
904 not renamed and the kernel default (eth<n>) was used.
906 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
907 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
908 the unit. So, it is expected that the default behavior of
909 systemd-logind is not changed. However, if distribution packagers or
910 administrators disabled or modified IPAddressDeny= setting by a
911 drop-in config file, then it may be necessary to update the file to
912 re-enable AF_INET and AF_INET6 to support network user name services,
915 * When the RestrictNamespaces= unit property is specified multiple
916 times, then the specified types are merged now. Previously, only the
917 last assignment was used. So, if distribution packagers or
918 administrators modified the setting by a drop-in config file, then it
919 may be necessary to update the file.
921 * When OnFailure= is used in combination with Restart= on a service
922 unit, then the specified units will no longer be triggered on
923 failures that result in restarting. Previously, the specified units
924 would be activated each time the unit failed, even when the unit was
925 going to be restarted automatically. This behaviour contradicted the
926 documentation. With this release the code is adjusted to match the
929 * systemd-tmpfiles will now print a notice whenever it encounters
930 tmpfiles.d/ lines referencing the /var/run/ directory. It will
931 recommend reworking them to use the /run/ directory instead (for
932 which /var/run/ is simply a symlinked compatibility alias). This way
933 systemd-tmpfiles can properly detect line conflicts and merge lines
934 referencing the same file by two paths, without having to access
937 * systemctl disable/unmask/preset/preset-all cannot be used with
938 --runtime. Previously this was allowed, but resulted in unintuitive
939 behaviour that wasn't useful. systemctl disable/unmask will now undo
940 both runtime and persistent enablement/masking, i.e. it will remove
941 any relevant symlinks both in /run and /etc.
943 * Note that all long-running system services shipped with systemd will
944 now default to a system call whitelist (rather than a blacklist, as
945 before). In particular, systemd-udevd will now enforce one too. For
946 most cases this should be safe, however downstream distributions
947 which disabled sandboxing of systemd-udevd (specifically the
948 MountFlags= setting), might want to disable this security feature
949 too, as the default whitelisting will prohibit all mount, swap,
950 reboot and clock changing operations from udev rules.
952 * sd-boot acquired new loader configuration settings to optionally turn
953 off Windows and MacOS boot partition discovery as well as
954 reboot-into-firmware menu items. It is also able to pick a better
955 screen resolution for HiDPI systems, and now provides loader
956 configuration settings to change the resolution explicitly.
958 * systemd-resolved now supports DNS-over-TLS. It's still
959 turned off by default, use DNSOverTLS=opportunistic to turn it on in
960 resolved.conf. We intend to make this the default as soon as couple
961 of additional techniques for optimizing the initial latency caused by
962 establishing a TLS/TCP connection are implemented.
964 * systemd-resolved.service and systemd-networkd.service now set
965 DynamicUser=yes. The users systemd-resolve and systemd-network are
966 not created by systemd-sysusers anymore.
968 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
969 that embedd a network facing module into any process using getpwuid()
970 or related call: the dynamic allocation of the user ID for
971 systemd-resolved.service means the service manager has to check NSS
972 if the user name is already taken when forking off the service. Since
973 the user in the common case won't be defined in /etc/passwd the
974 lookup is likely to trigger nss-ldap which in turn might use NSS to
975 ask systemd-resolved for hostname lookups. This will hence result in
976 a deadlock: a user name lookup in order to start
977 systemd-resolved.service will result in a host name lookup for which
978 systemd-resolved.service needs to be started already. There are
979 multiple ways to work around this problem: pre-allocate the
980 "systemd-resolve" user on such systems, so that nss-ldap won't be
981 triggered; or use a different NSS package that doesn't do networking
982 in-process but provides a local asynchronous name cache; or configure
983 the NSS package to avoid lookups for UIDs in the range `pkg-config
984 systemd --variable=dynamicuidmin` … `pkg-config systemd
985 --variable=dynamicuidmax`, so that it does not consider itself
986 authoritative for the same UID range systemd allocates dynamic users
989 * The systemd-resolve tool has been renamed to resolvectl (it also
990 remains available under the old name, for compatibility), and its
991 interface is now verb-based, similar in style to the other <xyz>ctl
992 tools, such as systemctl or loginctl.
994 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
995 compatibility. It may be symlinked under the 'resolvconf' name, in
996 which case it will take arguments and input compatible with the
997 Debian and FreeBSD resolvconf tool.
999 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
1000 where the system initially suspends, and after a timeout resumes and
1003 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
1004 set the client will only send a DUID as client identifier.
1006 * The nss-systemd glibc NSS module will now enumerate dynamic users and
1007 groups in effect. Previously, it could resolve UIDs/GIDs to user
1008 names/groups and vice versa, but did not support enumeration.
1010 * journald's Compress= configuration setting now optionally accepts a
1011 byte threshold value. All journal objects larger than this threshold
1012 will be compressed, smaller ones will not. Previously this threshold
1013 was not configurable and set to 512.
1015 * A new system.conf setting NoNewPrivileges= is now available which may
1016 be used to turn off acquisition of new privileges system-wide
1017 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
1018 for all its children). Note that turning this option on means setuid
1019 binaries and file system capabilities lose their special powers.
1020 While turning on this option is a big step towards a more secure
1021 system, doing so is likely to break numerous pre-existing UNIX tools,
1022 in particular su and sudo.
1024 * A new service systemd-time-sync-wait.service has been added. If
1025 enabled it will delay the time-sync.target unit at boot until time
1026 synchronization has been received from the network. This
1027 functionality is useful on systems lacking a local RTC or where it is
1028 acceptable that the boot process shall be delayed by external network
1031 * When hibernating, systemd will now inform the kernel of the image
1032 write offset, on kernels new enough to support this. This means swap
1033 files should work for hibernation now.
1035 * When loading unit files, systemd will now look for drop-in unit files
1036 extensions in additional places. Previously, for a unit file name
1037 "foo-bar-baz.service" it would look for dropin files in
1038 "foo-bar-baz.service.d/*.conf". Now, it will also look in
1039 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
1040 service name truncated after all inner dashes. This scheme allows
1041 writing drop-ins easily that apply to a whole set of unit files at
1042 once. It's particularly useful for mount and slice units (as their
1043 naming is prefix based), but is also useful for service and other
1044 units, for packages that install multiple unit files at once,
1045 following a strict naming regime of beginning the unit file name with
1046 the package's name. Two new specifiers are now supported in unit
1047 files to match this: %j and %J are replaced by the part of the unit
1048 name following the last dash.
1050 * Unit files and other configuration files that support specifier
1051 expansion now understand another three new specifiers: %T and %V will
1052 resolve to /tmp and /var/tmp respectively, or whatever temporary
1053 directory has been set for the calling user. %E will expand to either
1054 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
1056 * The ExecStart= lines of unit files are no longer required to
1057 reference absolute paths. If non-absolute paths are specified the
1058 specified binary name is searched within the service manager's
1059 built-in $PATH, which may be queried with 'systemd-path
1060 search-binaries-default'. It's generally recommended to continue to
1061 use absolute paths for all binaries specified in unit files.
1063 * Units gained a new load state "bad-setting", which is used when a
1064 unit file was loaded, but contained fatal errors which prevent it
1065 from being started (for example, a service unit has been defined
1066 lacking both ExecStart= and ExecStop= lines).
1068 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
1069 support alternative debuggers, for example lldb. The old name
1070 continues to be available however, for compatibility reasons. Use the
1071 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
1072 to pick an alternative debugger instead of the default gdb.
1074 * systemctl and the other tools will now output escape sequences that
1075 generate proper clickable hyperlinks in various terminal emulators
1076 where useful (for example, in the "systemctl status" output you can
1077 now click on the unit file name to quickly open it in the
1078 editor/viewer of your choice). Note that not all terminal emulators
1079 support this functionality yet, but many do. Unfortunately, the
1080 "less" pager doesn't support this yet, hence this functionality is
1081 currently automatically turned off when a pager is started (which
1082 happens quite often due to auto-paging). We hope to remove this
1083 limitation as soon as "less" learns these escape sequences. This new
1084 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
1085 environment variable. For details on these escape sequences see:
1086 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
1088 * networkd's .network files now support a new IPv6MTUBytes= option for
1089 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
1090 option in the [Route] section to configure the MTU to use for
1091 specific routes. It also gained support for configuration of the DHCP
1092 "UserClass" option through the new UserClass= setting. It gained
1093 three new options in the new [CAN] section for configuring CAN
1094 networks. The MULTICAST and ALLMULTI interface flags may now be
1095 controlled explicitly with the new Multicast= and AllMulticast=
1098 * networkd will now automatically make use of the kernel's route
1099 expiration feature, if it is available.
1101 * udevd's .link files now support setting the number of receive and
1102 transmit channels, using the RxChannels=, TxChannels=,
1103 OtherChannels=, CombinedChannels= settings.
1105 * Support for UDPSegmentationOffload= has been removed, given its
1106 limited support in hardware, and waning software support.
1108 * networkd's .netdev files now support creating "netdevsim" interfaces.
1110 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
1111 to query the unit belonging to a specific kernel control group.
1113 * systemd-analyze gained a new verb "cat-config", which may be used to
1114 dump the contents of any configuration file, with all its matching
1115 drop-in files added in, and honouring the usual search and masking
1116 logic applied to systemd configuration files. For example use
1117 "systemd-analyze cat-config systemd/system.conf" to get the complete
1118 system configuration file of systemd how it would be loaded by PID 1
1119 itself. Similar to this, various tools such as systemd-tmpfiles or
1120 systemd-sysusers, gained a new option "--cat-config", which does the
1121 corresponding operation for their own configuration settings. For
1122 example, "systemd-tmpfiles --cat-config" will now output the full
1123 list of tmpfiles.d/ lines in place.
1125 * timedatectl gained three new verbs: "show" shows bus properties of
1126 systemd-timedated, "timesync-status" shows the current NTP
1127 synchronization state of systemd-timesyncd, and "show-timesync"
1128 shows bus properties of systemd-timesyncd.
1130 * systemd-timesyncd gained a bus interface on which it exposes details
1133 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
1134 understood by systemd-timedated. It takes a colon-separated list of
1135 unit names of NTP client services. The list is used by
1136 "timedatectl set-ntp".
1138 * systemd-nspawn gained a new --rlimit= switch for setting initial
1139 resource limits for the container payload. There's a new switch
1140 --hostname= to explicitly override the container's hostname. A new
1141 --no-new-privileges= switch may be used to control the
1142 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
1143 --oom-score-adjust= switch controls the OOM scoring adjustment value
1144 for the payload. The new --cpu-affinity= switch controls the CPU
1145 affinity of the container payload. The new --resolv-conf= switch
1146 allows more detailed control of /etc/resolv.conf handling of the
1147 container. Similarly, the new --timezone= switch allows more detailed
1148 control of /etc/localtime handling of the container.
1150 * systemd-detect-virt gained a new --list switch, which will print a
1151 list of all currently known VM and container environments.
1153 * Support for "Portable Services" has been added, see
1154 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
1155 experimental, but this is expected to change soon. Reflecting this
1156 experimental state, the "portablectl" binary is not installed into
1157 /usr/bin yet. The binary has to be called with the full path
1158 /usr/lib/systemd/portablectl instead.
1160 * journalctl's and systemctl's -o switch now knows a new log output
1161 mode "with-unit". The output it generates is very similar to the
1162 regular "short" mode, but displays the unit name instead of the
1163 syslog tag for each log line. Also, the date is shown with timezone
1164 information. This mode is probably more useful than the classic
1165 "short" output mode for most purposes, except where pixel-perfect
1166 compatibility with classic /var/log/messages formatting is required.
1168 * A new --dump-bus-properties switch has been added to the systemd
1169 binary, which may be used to dump all supported D-Bus properties.
1170 (Options which are still supported, but are deprecated, are *not*
1173 * sd-bus gained a set of new calls:
1174 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
1175 enable/disable the "floating" state of a bus slot object,
1176 i.e. whether the slot object pins the bus it is allocated for into
1177 memory or if the bus slot object gets disconnected when the bus goes
1178 away. sd_bus_open_with_description(),
1179 sd_bus_open_user_with_description(),
1180 sd_bus_open_system_with_description() may be used to allocate bus
1181 objects and set their description string already during allocation.
1183 * sd-event gained support for watching inotify events from the event
1184 loop, in an efficient way, sharing inotify handles between multiple
1185 users. For this a new function sd_event_add_inotify() has been added.
1187 * sd-event and sd-bus gained support for calling special user-supplied
1188 destructor functions for userdata pointers associated with
1189 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
1190 functions sd_bus_slot_set_destroy_callback,
1191 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
1192 sd_bus_track_get_destroy_callback,
1193 sd_event_source_set_destroy_callback,
1194 sd_event_source_get_destroy_callback have been added.
1196 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
1198 * PID 1 will now automatically reschedule .timer units whenever the
1199 local timezone changes. (They previously got rescheduled
1200 automatically when the system clock changed.)
1202 * New documentation has been added to document cgroups delegation,
1203 portable services and the various code quality tools we have set up:
1205 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
1206 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
1207 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
1209 * The Boot Loader Specification has been added to the source tree.
1211 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
1213 While moving it into our source tree we have updated it and further
1214 changes are now accepted through the usual github PR workflow.
1216 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
1217 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
1218 earlier PAM modules. The data in these fields is used to initialize
1219 the session scope's resource properties. Thus external PAM modules
1220 may now configure per-session limits, for example sourced from
1221 external user databases.
1223 * socket units with Accept=yes will now maintain a "refused" counter in
1224 addition to the existing "accepted" counter, counting connections
1225 refused due to the enforced limits.
1227 * The "systemd-path search-binaries-default" command may now be use to
1228 query the default, built-in $PATH PID 1 will pass to the services it
1231 * A new unit file setting PrivateMounts= has been added. It's a boolean
1232 option. If enabled the unit's processes are invoked in their own file
1233 system namespace. Note that this behaviour is also implied if any
1234 other file system namespacing options (such as PrivateTmp=,
1235 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
1236 primarily useful for services that do not use any of the other file
1237 system namespacing options. One such service is systemd-udevd.service
1238 wher this is now used by default.
1240 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
1241 when the system is booted in UEFI "secure mode".
1243 * A new unit "system-update-pre.target" is added, which defines an
1244 optional synchronization point for offline system updates, as
1245 implemented by the pre-existing "system-update.target" unit. It
1246 allows ordering services before the service that executes the actual
1247 update process in a generic way.
1249 * Systemd now emits warnings whenever .include syntax is used.
1251 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
1252 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
1253 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
1254 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
1255 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
1256 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
1257 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
1258 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
1259 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
1260 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
1261 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
1262 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
1263 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
1264 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
1265 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
1266 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
1267 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
1268 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
1269 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
1270 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
1271 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
1272 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
1273 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
1274 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
1275 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
1276 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
1277 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
1278 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
1279 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1281 — Berlin, 2018-06-22
1285 * The MemoryAccounting= unit property now defaults to on. After
1286 discussions with the upstream control group maintainers we learnt
1287 that the negative impact of cgroup memory accounting on current
1288 kernels is finally relatively minimal, so that it should be safe to
1289 enable this by default without affecting system performance. Besides
1290 memory accounting only task accounting is turned on by default, all
1291 other forms of resource accounting (CPU, IO, IP) remain off for now,
1292 because it's not clear yet that their impact is small enough to move
1293 from opt-in to opt-out. We recommend downstreams to leave memory
1294 accounting on by default if kernel 4.14 or higher is primarily
1295 used. On very resource constrained systems or when support for old
1296 kernels is a necessity, -Dmemory-accounting-default=false can be used
1297 to revert this change.
1299 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
1300 %udev_rules_update) and the journal catalog (%journal_catalog_update)
1301 from the upgrade scriptlets of individual packages now do nothing.
1302 Transfiletriggers have been added which will perform those updates
1303 once at the end of the transaction.
1305 Similar transfiletriggers have been added to execute any sysctl.d
1306 and binfmt.d rules. Thus, it should be unnecessary to provide any
1307 scriptlets to execute this configuration from package installation
1310 * systemd-sysusers gained a mode where the configuration to execute is
1311 specified on the command line, but this configuration is not executed
1312 directly, but instead it is merged with the configuration on disk,
1313 and the result is executed. This is useful for package installation
1314 scripts which want to create the user before installing any files on
1315 disk (in case some of those files are owned by that user), while
1316 still allowing local admin overrides.
1318 This functionality is exposed to rpm scriptlets through a new
1319 %sysusers_create_package macro. Old %sysusers_create and
1320 %sysusers_create_inline macros are deprecated.
1322 A transfiletrigger for sysusers.d configuration is now installed,
1323 which means that it should be unnecessary to call systemd-sysusers from
1324 package installation scripts, unless the package installs any files
1325 owned by those newly-created users, in which case
1326 %sysusers_create_package should be used.
1328 * Analogous change has been done for systemd-tmpfiles: it gained a mode
1329 where the command-line configuration is merged with the configuration
1330 on disk. This is exposed as the new %tmpfiles_create_package macro,
1331 and %tmpfiles_create is deprecated. A transfiletrigger is installed
1332 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
1333 from package installation scripts.
1335 * sysusers.d configuration for a user may now also specify the group
1336 number, in addition to the user number ("u username 123:456"), or
1337 without the user number ("u username -:456").
1339 * Configution items for systemd-sysusers can now be specified as
1340 positional arguments when the new --inline switch is used.
1342 * The login shell of users created through sysusers.d may now be
1343 specified (previously, it was always /bin/sh for root and
1344 /sbin/nologin for other users).
1346 * systemd-analyze gained a new --global switch to look at global user
1347 configuration. It also gained a unit-paths verb to list the unit load
1348 paths that are compiled into systemd (which can be used with
1349 --systemd, --user, or --global).
1351 * udevadm trigger gained a new --settle/-w option to wait for any
1352 triggered events to finish (but just those, and not any other events
1353 which are triggered meanwhile).
1355 * The action that systemd-logind takes when the lid is closed and the
1356 machine is connected to external power can now be configured using
1357 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
1358 was determined by HandleLidSwitch=, and, for backwards compatibility,
1359 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
1361 * journalctl will periodically call sd_journal_process() to make it
1362 resilient against inotify queue overruns when journal files are
1363 rotated very quickly.
1365 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
1366 sd_bus_get_n_queued_write — may be used to check the number of
1367 pending bus messages.
1369 * systemd gained a new
1370 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
1371 which can be used to migrate foreign processes to scope and service
1372 units. The primary user for this new API is systemd itself: the
1373 systemd --user instance uses this call of the systemd --system
1374 instance to migrate processes if it itself gets the request to
1375 migrate processes and the kernel refuses this due to access
1376 restrictions. Thanks to this "systemd-run --scope --user …" works
1377 again in pure cgroup v2 environments when invoked from the user
1380 * A new TemporaryFileSystem= setting can be used to mask out part of
1381 the real file system tree with tmpfs mounts. This may be combined
1382 with BindPaths= and BindReadOnlyPaths= to hide files or directories
1383 not relevant to the unit, while still allowing some paths lower in
1384 the tree to be accessed.
1386 ProtectHome=tmpfs may now be used to hide user home and runtime
1387 directories from units, in a way that is mostly equivalent to
1388 "TemporaryFileSystem=/home /run/user /root".
1390 * Non-service units are now started with KeyringMode=shared by default.
1391 This means that mount and swapon and other mount tools have access
1392 to keys in the main keyring.
1394 * /sys/fs/bpf is now mounted automatically.
1396 * QNX virtualization is now detected by systemd-detect-virt and may
1397 be used in ConditionVirtualization=.
1399 * IPAccounting= may now be enabled also for slice units.
1401 * A new -Dsplit-bin= build configuration switch may be used to specify
1402 whether bin and sbin directories are merged, or if they should be
1403 included separately in $PATH and various listings of executable
1404 directories. The build configuration scripts will try to autodetect
1405 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
1406 system, but distributions are encouraged to configure this
1409 * A new -Dok-color= build configuration switch may be used to change
1410 the colour of "OK" status messages.
1412 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
1413 PrivateNetwork=yes was buggy in previous versions of systemd. This
1414 means that after the upgrade and daemon-reexec, any such units must
1417 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
1418 will not exclude read-only files owned by root from cleanup.
1420 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
1421 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
1422 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
1423 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
1424 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
1425 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
1426 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
1427 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
1428 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
1429 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
1430 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
1431 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
1432 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
1433 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
1434 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
1435 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
1437 — Warsaw, 2018-03-05
1441 * Some keyboards come with a zoom see-saw or rocker which until now got
1442 mapped to the Linux "zoomin/out" keys in hwdb. However, these
1443 keycodes are not recognized by any major desktop. They now produce
1444 Up/Down key events so that they can be used for scrolling.
1446 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
1447 slightly: previously, if an argument was specified for lines of this
1448 type (i.e. the right-most column was set) this string was appended to
1449 existing files each time systemd-tmpfiles was run. This behaviour was
1450 different from what the documentation said, and not particularly
1451 useful, as repeated systemd-tmpfiles invocations would not be
1452 idempotent and grow such files without bounds. With this release
1453 behaviour has been altered to match what the documentation says:
1454 lines of this type only have an effect if the indicated files don't
1455 exist yet, and only then the argument string is written to the file.
1457 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
1458 systemd-tmpfiles behaviour: previously, read-only files owned by root
1459 were always excluded from the file "aging" algorithm (i.e. the
1460 automatic clean-up of directories like /tmp based on
1461 atime/mtime/ctime). We intend to drop this restriction, and age files
1462 by default even when owned by root and read-only. This behaviour was
1463 inherited from older tools, but there have been requests to remove
1464 it, and it's not obvious why this restriction was made in the first
1465 place. Please speak up now, if you are aware of software that reqires
1466 this behaviour, otherwise we'll remove the restriction in v238.
1468 * A new environment variable $SYSTEMD_OFFLINE is now understood by
1469 systemctl. It takes a boolean argument. If on, systemctl assumes it
1470 operates on an "offline" OS tree, and will not attempt to talk to the
1471 service manager. Previously, this mode was implicitly enabled if a
1472 chroot() environment was detected, and this new environment variable
1473 now provides explicit control.
1475 * .path and .socket units may now be created transiently, too.
1476 Previously only service, mount, automount and timer units were
1477 supported as transient units. The systemd-run tool has been updated
1478 to expose this new functionality, you may hence use it now to bind
1479 arbitrary commands to path or socket activation on-the-fly from the
1480 command line. Moreover, almost all properties are now exposed for the
1481 unit types that already supported transient operation.
1483 * The systemd-mount command gained support for a new --owner= parameter
1484 which takes a user name, which is then resolved and included in uid=
1485 and gid= mount options string of the file system to mount.
1487 * A new unit condition ConditionControlGroupController= has been added
1488 that checks whether a specific cgroup controller is available.
1490 * Unit files, udev's .link files, and systemd-networkd's .netdev and
1491 .network files all gained support for a new condition
1492 ConditionKernelVersion= for checking against specific kernel
1495 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
1496 support for configuring device flags in the Flags= setting. In the
1497 same files, the [Tunnel] section gained support for configuring
1498 AllowLocalRemote=. The [Route] section in .network files gained
1499 support for configuring InitialCongestionWindow=,
1500 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
1501 understands RapidCommit=.
1503 * systemd-networkd's DHCPv6 support gained support for Prefix
1506 * sd-bus gained support for a new "watch-bind" feature. When this
1507 feature is enabled, an sd_bus connection may be set up to connect to
1508 an AF_UNIX socket in the file system as soon as it is created. This
1509 functionality is useful for writing early-boot services that
1510 automatically connect to the system bus as soon as it is started,
1511 without ugly time-based polling. systemd-networkd and
1512 systemd-resolved have been updated to make use of this
1513 functionality. busctl exposes this functionality in a new
1514 --watch-bind= command line switch.
1516 * sd-bus will now optionally synthesize a local "Connected" signal as
1517 soon as a D-Bus connection is set up fully. This message mirrors the
1518 already existing "Disconnected" signal which is synthesized when the
1519 connection is terminated. This signal is generally useful but
1520 particularly handy in combination with the "watch-bind" feature
1521 described above. Synthesizing of this message has to be requested
1522 explicitly through the new API call sd_bus_set_connected_signal(). In
1523 addition a new call sd_bus_is_ready() has been added that checks
1524 whether a connection is fully set up (i.e. between the "Connected" and
1525 "Disconnected" signals).
1527 * sd-bus gained two new calls sd_bus_request_name_async() and
1528 sd_bus_release_name_async() for asynchronously registering bus
1529 names. Similar, there is now sd_bus_add_match_async() for installing
1530 a signal match asynchronously. All of systemd's own services have
1531 been updated to make use of these calls. Doing these operations
1532 asynchronously has two benefits: it reduces the risk of deadlocks in
1533 case of cyclic dependencies between bus services, and it speeds up
1534 service initialization since synchronization points for bus
1535 round-trips are removed.
1537 * sd-bus gained two new calls sd_bus_match_signal() and
1538 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
1539 and sd_bus_add_match_async() but instead of taking a D-Bus match
1540 string take match fields as normal function parameters.
1542 * sd-bus gained two new calls sd_bus_set_sender() and
1543 sd_bus_message_set_sender() for setting the sender name of outgoing
1544 messages (either for all outgoing messages or for just one specific
1545 one). These calls are only useful in direct connections as on
1546 brokered connections the broker fills in the sender anyway,
1547 overwriting whatever the client filled in.
1549 * sd-event gained a new pseudo-handle that may be specified on all API
1550 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
1551 used this refers to the default event loop object of the calling
1552 thread. Note however that this does not implicitly allocate one —
1553 which has to be done prior by using sd_event_default(). Similarly
1554 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
1555 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
1556 to the default bus of the specified type of the calling thread. Here
1557 too this does not implicitly allocate bus connection objects, this
1558 has to be done prior with sd_bus_default() and friends.
1560 * sd-event gained a new call pair
1561 sd_event_source_{get|set}_io_fd_own(). This may be used to request
1562 automatic closure of the file descriptor an IO event source watches
1563 when the event source is destroyed.
1565 * systemd-networkd gained support for natively configuring WireGuard
1568 * In previous versions systemd synthesized user records both for the
1569 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
1570 internally. In order to simplify distribution-wide renames of the
1571 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
1572 new transitional flag file has been added: if
1573 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
1574 user and group record within the systemd codebase is disabled.
1576 * systemd-notify gained a new --uid= option for selecting the source
1577 user/UID to use for notification messages sent to the service
1580 * journalctl gained a new --grep= option to list only entries in which
1581 the message matches a certain pattern. By default matching is case
1582 insensitive if the pattern is lowercase, and case sensitive
1583 otherwise. Option --case-sensitive=yes|no can be used to override
1584 this an specify case sensitivity or case insensitivity.
1586 * There's now a "systemd-analyze service-watchdogs" command for printing
1587 the current state of the service runtime watchdog, and optionally
1588 enabling or disabling the per-service watchdogs system-wide if given a
1589 boolean argument (i.e. the concept you configure in WatchdogSec=), for
1590 debugging purposes. There's also a kernel command line option
1591 systemd.service_watchdogs= for controlling the same.
1593 * Two new "log-level" and "log-target" options for systemd-analyze were
1594 added that merge the now deprecated get-log-level, set-log-level and
1595 get-log-target, set-log-target pairs. The deprecated options are still
1596 understood for backwards compatibility. The two new options print the
1597 current value when no arguments are given, and set them when a
1598 level/target is given as an argument.
1600 * sysusers.d's "u" lines now optionally accept both a UID and a GID
1601 specification, separated by a ":" character, in order to create users
1602 where UID and GID do not match.
1604 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
1605 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
1606 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
1607 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
1608 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
1609 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
1610 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
1611 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
1612 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
1613 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
1614 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
1615 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
1616 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
1617 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
1618 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
1619 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
1620 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
1621 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
1622 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
1623 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
1630 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
1631 in v235 has been extended to also set the dummy.ko module option
1632 numdummies=0, preventing the kernel from automatically creating
1633 dummy0. All dummy interfaces must now be explicitly created.
1635 * Unknown '%' specifiers in configuration files are now rejected. This
1636 applies to units and tmpfiles.d configuration. Any percent characters
1637 that are followed by a letter or digit that are not supposed to be
1638 interpreted as the beginning of a specifier should be escaped by
1639 doubling ("%%"). (So "size=5%" is still accepted, as well as
1640 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
1641 valid specifiers today.)
1643 * systemd-resolved now maintains a new dynamic
1644 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
1645 recommended to make /etc/resolv.conf a symlink to it. This file
1646 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
1647 includes dynamically acquired search domains, achieving more correct
1648 DNS resolution by software that bypasses local DNS APIs such as NSS.
1650 * The "uaccess" udev tag has been dropped from /dev/kvm and
1651 /dev/dri/renderD*. These devices now have the 0666 permissions by
1652 default (but this may be changed at build-time). /dev/dri/renderD*
1653 will now be owned by the "render" group along with /dev/kfd.
1655 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
1656 systemd-journal-gatewayd.service and
1657 systemd-journal-upload.service. This means "nss-systemd" must be
1658 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
1659 services are resolved properly.
1661 * In /etc/fstab two new mount options are now understood:
1662 x-systemd.makefs and x-systemd.growfs. The former has the effect that
1663 the configured file system is formatted before it is mounted, the
1664 latter that the file system is resized to the full block device size
1665 after it is mounted (i.e. if the file system is smaller than the
1666 partition it resides on, it's grown). This is similar to the fsck
1667 logic in /etc/fstab, and pulls in systemd-makefs@.service and
1668 systemd-growfs@.service as necessary, similar to
1669 systemd-fsck@.service. Resizing is currently only supported on ext4
1672 * In systemd-networkd, the IPv6 RA logic now optionally may announce
1673 DNS server and domain information.
1675 * Support for the LUKS2 on-disk format for encrypted partitions has
1676 been added. This requires libcryptsetup2 during compilation and
1679 * The systemd --user instance will now signal "readiness" when its
1680 basic.target unit has been reached, instead of when the run queue ran
1681 empty for the first time.
1683 * Tmpfiles.d with user configuration are now also supported.
1684 systemd-tmpfiles gained a new --user switch, and snippets placed in
1685 ~/.config/user-tmpfiles.d/ and corresponding directories will be
1686 executed by systemd-tmpfiles --user running in the new
1687 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
1688 running in the user session.
1690 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
1691 %S resolves to the top-level state directory (/var/lib for the system
1692 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
1693 top-level cache directory (/var/cache for the system instance,
1694 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
1695 logs directory (/var/log for the system instance,
1696 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
1697 existing %t specifier, that resolves to the top-level runtime
1698 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
1701 * journalctl learnt a new parameter --output-fields= for limiting the
1702 set of journal fields to output in verbose and JSON output modes.
1704 * systemd-timesyncd's configuration file gained a new option
1705 RootDistanceMaxSec= for setting the maximum root distance of servers
1706 it'll use, as well as the new options PollIntervalMinSec= and
1707 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
1709 * bootctl gained a new command "list" for listing all available boot
1710 menu items on systems that follow the boot loader specification.
1712 * systemctl gained a new --dry-run switch that shows what would be done
1713 instead of doing it, and is currently supported by the shutdown and
1716 * ConditionSecurity= can now detect the TOMOYO security module.
1718 * Unit file [Install] sections are now also respected in unit drop-in
1719 files. This is intended to be used by drop-ins under /usr/lib/.
1721 * systemd-firstboot may now also set the initial keyboard mapping.
1723 * Udev "changed" events for devices which are exposed as systemd
1724 .device units are now propagated to units specified in
1725 ReloadPropagatedFrom= as reload requests.
1727 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
1728 unit template name (i.e. a name in the form of 'foobar@.service',
1729 without the instance component between the '@' and - the '.'), then
1730 the escaped sysfs path of the device is automatically used as the
1733 * SystemCallFilter= in unit files has been extended so that an "errno"
1734 can be specified individually for each system call. Example:
1735 SystemCallFilter=~uname:EILSEQ.
1737 * The cgroup delegation logic has been substantially updated. Delegate=
1738 now optionally takes a list of controllers (instead of a boolean, as
1739 before), which lists the controllers to delegate at least.
1741 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
1743 * A new LogLevelMax= setting configures the maximum log level any
1744 process of the service may log at (i.e. anything with a lesser
1745 priority than what is specified is automatically dropped). A new
1746 LogExtraFields= setting allows configuration of additional journal
1747 fields to attach to all log records generated by any of the unit's
1750 * New StandardInputData= and StandardInputText= settings along with the
1751 new option StandardInput=data may be used to configure textual or
1752 binary data that shall be passed to the executed service process via
1753 standard input, encoded in-line in the unit file.
1755 * StandardInput=, StandardOutput= and StandardError= may now be used to
1756 connect stdin/stdout/stderr of executed processes directly with a
1757 file or AF_UNIX socket in the file system, using the new "file:" option.
1759 * A new unit file option CollectMode= has been added, that allows
1760 tweaking the garbage collection logic for units. It may be used to
1761 tell systemd to garbage collect units that have failed automatically
1762 (normally it only GCs units that exited successfully). systemd-run
1763 and systemd-mount expose this new functionality with a new -G option.
1765 * "machinectl bind" may now be used to bind mount non-directories
1766 (i.e. regularfiles, devices, fifos, sockets).
1768 * systemd-analyze gained a new verb "calendar" for validating and
1769 testing calendar time specifications to use for OnCalendar= in timer
1770 units. Besides validating the expression it will calculate the next
1771 time the specified expression would elapse.
1773 * In addition to the pre-existing FailureAction= unit file setting
1774 there's now SuccessAction=, for configuring a shutdown action to
1775 execute when a unit completes successfully. This is useful in
1776 particular inside containers that shall terminate after some workload
1777 has been completed. Also, both options are now supported for all unit
1778 types, not just services.
1780 * networkds's IP rule support gained two new options
1781 IncomingInterface= and OutgoingInterface= for configuring the incoming
1782 and outgoing interfaces of configured rules. systemd-networkd also
1783 gained support for "vxcan" network devices.
1785 * networkd gained a new setting RequiredForOnline=, taking a
1786 boolean. If set, systemd-wait-online will take it into consideration
1787 when determining that the system is up, otherwise it will ignore the
1788 interface for this purpose.
1790 * The sd_notify() protocol gained support for a new operation: with
1791 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
1792 store again, ahead of POLLHUP or POLLERR when they are removed
1795 * A new document doc/UIDS-GIDS.md has been added to the source tree,
1796 that documents the UID/GID range and assignment assumptions and
1797 requirements of systemd.
1799 * The watchdog device PID 1 will ping may now be configured through the
1800 WatchdogDevice= configuration file setting, or by setting the
1801 systemd.watchdog_service= kernel commandline option.
1803 * systemd-resolved's gained support for registering DNS-SD services on
1804 the local network using MulticastDNS. Services may either be
1805 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
1806 the same dir below /run, /usr/lib), or through its D-Bus API.
1808 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
1809 extend the effective start, runtime, and stop time. The service must
1810 continue to send EXTEND_TIMEOUT_USEC within the period specified to
1811 prevent the service manager from making the service as timedout.
1813 * systemd-resolved's DNSSEC support gained support for RFC 8080
1814 (Ed25519 keys and signatures).
1816 * The systemd-resolve command line tool gained a new set of options
1817 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
1818 --set-nta= and --revert to configure per-interface DNS configuration
1819 dynamically during runtime. It's useful for pushing DNS information
1820 into systemd-resolved from DNS hook scripts that various interface
1821 managing software supports (such as pppd).
1823 * systemd-nspawn gained a new --network-namespace-path= command line
1824 option, which may be used to make a container join an existing
1825 network namespace, by specifying a path to a "netns" file.
1827 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
1828 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
1829 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
1830 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
1831 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
1832 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
1833 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
1834 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
1835 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
1836 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
1837 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
1838 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
1839 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
1840 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
1841 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
1842 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
1843 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
1844 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
1845 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
1846 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
1847 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
1848 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
1849 Jędrzejewski-Szmek, Zeal Jagannatha
1851 — Berlin, 2017-12-14
1855 * INCOMPATIBILITY: systemd-logind.service and other long-running
1856 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
1857 communication with the outside. This generally improves security of
1858 the system, and is in almost all cases a safe and good choice, as
1859 these services do not and should not provide any network-facing
1860 functionality. However, systemd-logind uses the glibc NSS API to
1861 query the user database. This creates problems on systems where NSS
1862 is set up to directly consult network services for user database
1863 lookups. In particular, this creates incompatibilities with the
1864 "nss-nis" module, which attempts to directly contact the NIS/YP
1865 network servers it is configured for, and will now consistently
1866 fail. In such cases, it is possible to turn off IP sandboxing for
1867 systemd-logind.service (set IPAddressDeny= in its [Service] section
1868 to the empty string, via a .d/ unit file drop-in). Downstream
1869 distributions might want to update their nss-nis packaging to include
1870 such a drop-in snippet, accordingly, to hide this incompatibility
1871 from the user. Another option is to make use of glibc's nscd service
1872 to proxy such network requests through a privilege-separated, minimal
1873 local caching daemon, or to switch to more modern technologies such
1874 sssd, whose NSS hook-ups generally do not involve direct network
1875 access. In general, we think it's definitely time to question the
1876 implementation choices of nss-nis, i.e. whether it's a good idea
1877 today to embed a network-facing loadable module into all local
1878 processes that need to query the user database, including the most
1879 trivial and benign ones, such as "ls". For more details about
1880 IPAddressDeny= see below.
1882 * A new modprobe.d drop-in is now shipped by default that sets the
1883 bonding module option max_bonds=0. This overrides the kernel default,
1884 to avoid conflicts and ambiguity as to whether or not bond0 should be
1885 managed by systemd-networkd or not. This resolves multiple issues
1886 with bond0 properties not being applied, when bond0 is configured
1887 with systemd-networkd. Distributors may choose to not package this,
1888 however in that case users will be prevented from correctly managing
1889 bond0 interface using systemd-networkd.
1891 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
1892 which print the logging level and target of the system manager. They
1893 complement the existing "set-log-level" and "set-log-target" verbs
1894 used to change those values.
1896 * journald.conf gained a new boolean setting ReadKMsg= which defaults
1897 to on. If turned off kernel log messages will not be read by
1898 systemd-journald or included in the logs. It also gained a new
1899 setting LineMax= for configuring the maximum line length in
1900 STDOUT/STDERR log streams. The new default for this value is 48K, up
1901 from the previous hardcoded 2048.
1903 * A new unit setting RuntimeDirectoryPreserve= has been added, which
1904 allows more detailed control of what to do with a runtime directory
1905 configured with RuntimeDirectory= (i.e. a directory below /run or
1906 $XDG_RUNTIME_DIR) after a unit is stopped.
1908 * The RuntimeDirectory= setting for units gained support for creating
1909 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
1910 one top-level directory.
1912 * Units gained new options StateDirectory=, CacheDirectory=,
1913 LogsDirectory= and ConfigurationDirectory= which are closely related
1914 to RuntimeDirectory= but manage per-service directories below
1915 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
1916 possible to write unit files which when activated automatically gain
1917 properly owned service specific directories in these locations, thus
1918 making unit files self-contained and increasing compatibility with
1919 stateless systems and factory reset where /etc or /var are
1920 unpopulated at boot. Matching these new settings there's also
1921 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
1922 ConfigurationDirectoryMode= for configuring the access mode of these
1923 directories. These settings are particularly useful in combination
1924 with DynamicUser=yes as they provide secure, properly-owned,
1925 writable, and stateful locations for storage, excluded from the
1926 sandbox that such services live in otherwise.
1928 * Automake support has been removed from this release. systemd is now
1931 * systemd-journald will now aggressively cache client metadata during
1932 runtime, speeding up log write performance under pressure. This comes
1933 at a small price though: as much of the metadata is read
1934 asynchronously from /proc/ (and isn't implicitly attached to log
1935 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
1936 metadata stored alongside a log entry might be slightly
1937 out-of-date. Previously it could only be slightly newer than the log
1938 message. The time window is small however, and given that the kernel
1939 is unlikely to be improved anytime soon in this regard, this appears
1942 * nss-myhostname/systemd-resolved will now by default synthesize an
1943 A/AAAA resource record for the "_gateway" hostname, pointing to the
1944 current default IP gateway. Previously it did that for the "gateway"
1945 name, hampering adoption, as some distributions wanted to leave that
1946 host name open for local use. The old behaviour may still be
1947 requested at build time.
1949 * systemd-networkd's [Address] section in .network files gained a new
1950 Scope= setting for configuring the IP address scope. The [Network]
1951 section gained a new boolean setting ConfigureWithoutCarrier= that
1952 tells systemd-networkd to ignore link sensing when configuring the
1953 device. The [DHCP] section gained a new Anonymize= boolean option for
1954 turning on a number of options suggested in RFC 7844. A new
1955 [RoutingPolicyRule] section has been added for configuring the IP
1956 routing policy. The [Route] section has gained support for a new
1957 Type= setting which permits configuring
1958 blackhole/unreachable/prohibit routes.
1960 * The [VRF] section in .netdev files gained a new Table= setting for
1961 configuring the routing table to use. The [Tunnel] section gained a
1962 new Independent= boolean field for configuring tunnels independent of
1963 an underlying network interface. The [Bridge] section gained a new
1964 GroupForwardMask= option for configuration of propagation of link
1965 local frames between bridge ports.
1967 * The WakeOnLan= setting in .link files gained support for a number of
1968 new modes. A new TCP6SegmentationOffload= setting has been added for
1969 configuring TCP/IPv6 hardware segmentation offload.
1971 * The IPv6 RA sender implementation may now optionally send out RDNSS
1972 and RDNSSL records to supply DNS configuration to peers.
1974 * systemd-nspawn gained support for a new --system-call-filter= command
1975 line option for adding and removing entries in the default system
1976 call filter it applies. Moreover systemd-nspawn has been changed to
1977 implement a system call whitelist instead of a blacklist.
1979 * systemd-run gained support for a new --pipe command line option. If
1980 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
1981 are directly passed on to the activated transient service
1982 executable. This allows invoking arbitrary processes as systemd
1983 services (for example to take benefit of dependency management,
1984 accounting management, resource management or log management that is
1985 done automatically for services) — while still allowing them to be
1986 integrated in a classic UNIX shell pipeline.
1988 * When a service sends RELOAD=1 via sd_notify() and reload propagation
1989 using ReloadPropagationTo= is configured, a reload is now propagated
1990 to configured units. (Previously this was only done on explicitly
1991 requested reloads, using "systemctl reload" or an equivalent
1994 * For each service unit a restart counter is now kept: it is increased
1995 each time the service is restarted due to Restart=, and may be
1996 queried using "systemctl show -p NRestarts …".
1998 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
1999 @signal and @timer have been added, for usage with SystemCallFilter=
2000 in unit files and the new --system-call-filter= command line option
2001 of systemd-nspawn (see above).
2003 * ExecStart= lines in unit files gained two new modifiers: when a
2004 command line is prefixed with "!" the command will be executed as
2005 configured, except for the credentials applied by
2006 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
2007 "+", but does still apply namespacing options unlike "+". There's
2008 also "!!" now, which is mostly identical, but becomes a NOP on
2009 systems that support ambient capabilities. This is useful to write
2010 unit files that work with ambient capabilities where possible but
2011 automatically fall back to traditional privilege dropping mechanisms
2012 on systems where this is not supported.
2014 * ListenNetlink= settings in socket units now support RDMA netlink
2017 * A new unit file setting LockPersonality= has been added which permits
2018 locking down the chosen execution domain ("personality") of a service
2021 * A new special target "getty-pre.target" has been added, which is
2022 ordered before all text logins, and may be used to order services
2023 before textual logins acquire access to the console.
2025 * systemd will now attempt to load the virtio-rng.ko kernel module very
2026 early on if a VM environment supporting this is detected. This should
2027 improve entropy during early boot in virtualized environments.
2029 * A _netdev option is now supported in /etc/crypttab that operates in a
2030 similar way as the same option in /etc/fstab: it permits configuring
2031 encrypted devices that need to be ordered after the network is up.
2032 Following this logic, two new special targets
2033 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
2034 added that are to cryptsetup.target what remote-fs.target and
2035 remote-fs-pre.target are to local-fs.target.
2037 * Service units gained a new UnsetEnvironment= setting which permits
2038 unsetting specific environment variables for services that are
2039 normally passed to it (for example in order to mask out locale
2040 settings for specific services that can't deal with it).
2042 * Units acquired a new boolean option IPAccounting=. When turned on, IP
2043 traffic accounting (packet count as well as byte count) is done for
2044 the service, and shown as part of "systemctl status" or "systemd-run
2047 * Service units acquired two new options IPAddressAllow= and
2048 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
2049 for configuring a simple IP access control list for all sockets of
2050 the unit. These options are available also on .slice and .socket
2051 units, permitting flexible access list configuration for individual
2052 services as well as groups of services (as defined by a slice unit),
2053 including system-wide. Note that IP ACLs configured this way are
2054 enforced on every single IPv4 and IPv6 socket created by any process
2055 of the service unit, and apply to ingress as well as egress traffic.
2057 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
2058 structured log message is generated each time the unit is stopped,
2059 containing information about the consumed resources of this
2062 * A new setting KeyringMode= has been added to unit files, which may be
2063 used to control how the kernel keyring is set up for executed
2066 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
2067 "systemctl kexec" and "systemctl exit" are now always asynchronous in
2068 behaviour (that is: these commands return immediately after the
2069 operation was enqueued instead of waiting for the operation to
2070 complete). Previously, "systemctl poweroff" and "systemctl reboot"
2071 were asynchronous on systems using systemd-logind (i.e. almost
2072 always, and like they were on sysvinit), and the other three commands
2073 were unconditionally synchronous. With this release this is cleaned
2074 up, and callers will see the same asynchronous behaviour on all
2075 systems for all five operations.
2077 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
2080 * .timer units now accept calendar specifications in other timezones
2081 than UTC or the local timezone.
2083 * The tmpfiles snippet var.conf has been changed to create
2084 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
2085 the "utmp" group already, and it appears to be generally understood
2086 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
2087 databases. Previously this was implemented correctly for all these
2088 databases excepts btmp, which has been opened up like this now
2089 too. Note that while the other databases are world-readable
2090 (i.e. 0644), btmp is not and remains more restrictive.
2092 * The systemd-resolve tool gained a new --reset-server-features
2093 switch. When invoked like this systemd-resolved will forget
2094 everything it learnt about the features supported by the configured
2095 upstream DNS servers, and restarts the feature probing logic on the
2096 next resolver look-up for them at the highest feature level
2099 * The status dump systemd-resolved sends to the logs upon receiving
2100 SIGUSR1 now also includes information about all DNS servers it is
2101 configured to use, and the features levels it probed for them.
2103 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
2104 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
2105 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
2106 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
2107 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
2108 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
2109 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
2110 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
2111 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
2112 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
2113 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
2114 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
2115 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
2116 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
2117 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
2118 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
2119 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
2120 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
2121 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
2122 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2124 — Berlin, 2017-10-06
2128 * Meson is now supported as build system in addition to Automake. It is
2129 our plan to remove Automake in one of our next releases, so that
2130 Meson becomes our exclusive build system. Hence, please start using
2131 the Meson build system in your downstream packaging. There's plenty
2132 of documentation around how to use Meson, the extremely brief
2135 ./autogen.sh && ./configure && make && sudo make install
2139 meson build && ninja -C build && sudo ninja -C build install
2141 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
2142 which permits configuring a timeout on the time a job is
2143 running. This is particularly useful for setting timeouts on jobs for
2146 * Unit files gained two new options ConditionUser= and ConditionGroup=
2147 for conditionalizing units based on the identity of the user/group
2148 running a systemd user instance.
2150 * systemd-networkd now understands a new FlowLabel= setting in the
2151 [VXLAN] section of .network files, as well as a Priority= in
2152 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
2153 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
2154 gained support for configuration of GENEVE links, and IPv6 address
2155 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
2157 * .link files now understand a new Port= setting.
2159 * systemd-networkd's DHCP support gained support for DHCP option 119
2160 (domain search list).
2162 * systemd-networkd gained support for serving IPv6 address ranges using
2163 the Router Advertisement protocol. The new .network configuration
2164 section [IPv6Prefix] may be used to configure the ranges to
2165 serve. This is implemented based on a new, minimal, native server
2166 implementation of RA.
2168 * journalctl's --output= switch gained support for a new parameter
2169 "short-iso-precise" for a mode where timestamps are shown as precise
2172 * systemd-udevd's "net_id" builtin may now generate stable network
2173 interface names from IBM PowerVM VIO devices as well as ACPI platform
2176 * MulticastDNS support in systemd-resolved may now be explicitly
2177 enabled/disabled using the new MulticastDNS= configuration file
2180 * systemd-resolved may now optionally use libidn2 instead of the libidn
2181 for processing internationalized domain names. Support for libidn2
2182 should be considered experimental and should not be enabled by
2185 * "machinectl pull-tar" and related call may now do verification of
2186 downloaded images using SUSE-style .sha256 checksum files in addition
2187 to the already existing support for validating using Ubuntu-style
2190 * sd-bus gained support for a new sd_bus_message_appendv() call which
2191 is va_list equivalent of sd_bus_message_append().
2193 * sd-boot gained support for validating images using SHIM/MOK.
2195 * The SMACK code learnt support for "onlycap".
2197 * systemd-mount --umount is now much smarter in figuring out how to
2198 properly unmount a device given its mount or device path.
2200 * The code to call libnss_dns as a fallback from libnss_resolve when
2201 the communication with systemd-resolved fails was removed. This
2202 fallback was redundant and interfered with the [!UNAVAIL=return]
2203 suffix. See nss-resolve(8) for the recommended configuration.
2205 * systemd-logind may now be restarted without losing state. It stores
2206 the file descriptors for devices it manages in the system manager
2207 using the FDSTORE= mechanism. Please note that further changes in
2208 other components may be required to make use of this (for example
2209 Xorg has code to listen for stops of systemd-logind and terminate
2210 itself when logind is stopped or restarted, in order to avoid using
2211 stale file descriptors for graphical devices, which is now
2212 counterproductive and must be reverted in order for restarts of
2213 systemd-logind to be safe. See
2214 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
2216 * All kernel-install plugins are called with the environment variable
2217 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
2218 /etc/machine-id. If the machine ID could not be determined,
2219 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
2220 anything in the entry directory (passed as the second argument) if
2221 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatiblity, a
2222 temporary directory is passed as the entry directory and removed
2223 after all the plugins exit.
2225 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
2226 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
2227 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
2228 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
2229 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
2230 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
2231 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
2232 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
2233 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
2234 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
2235 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
2236 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
2237 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
2238 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
2239 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
2240 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
2241 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
2242 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
2243 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
2244 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
2245 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
2246 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
2247 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
2248 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
2249 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
2250 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
2251 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
2252 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
2253 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
2256 — Berlin, 2017-07-12
2260 * The "hybrid" control group mode has been modified to improve
2261 compatibility with "legacy" cgroups-v1 setups. Specifically, the
2262 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
2263 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
2264 cgroups-v1 hierarchy), the only externally visible change being that
2265 the cgroups-v2 hierarchy is also mounted, to
2266 /sys/fs/cgroup/unified. This should provide a large degree of
2267 compatibility with "legacy" cgroups-v1, while taking benefit of the
2268 better management capabilities of cgroups-v2.
2270 * The default control group setup mode may be selected both a boot-time
2271 via a set of kernel command line parameters (specifically:
2272 systemd.unified_cgroup_hierarchy= and
2273 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
2274 default selected on the configure command line
2275 (--with-default-hierarchy=). The upstream default is "hybrid"
2276 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
2277 this will change in a future systemd version to be "unified" (pure
2278 cgroups-v2 mode). The third option for the compile time option is
2279 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
2280 distributions to default to "hybrid" mode for release distributions,
2281 starting with v233. We recommend "unified" for development
2282 distributions (specifically: distributions such as Fedora's rawhide)
2283 as that's where things are headed in the long run. Use "legacy" for
2284 greatest stability and compatibility only.
2286 * Note one current limitation of "unified" and "hybrid" control group
2287 setup modes: the kernel currently does not permit the systemd --user
2288 instance (i.e. unprivileged code) to migrate processes between two
2289 disconnected cgroup subtrees, even if both are managed and owned by
2290 the user. This effectively means "systemd-run --user --scope" doesn't
2291 work when invoked from outside of any "systemd --user" service or
2292 scope. Specifically, it is not supported from session scopes. We are
2293 working on fixing this in a future systemd version. (See #3388 for
2294 further details about this.)
2296 * DBus policy files are now installed into /usr rather than /etc. Make
2297 sure your system has dbus >= 1.9.18 running before upgrading to this
2298 version, or override the install path with --with-dbuspolicydir= .
2300 * All python scripts shipped with systemd (specifically: the various
2301 tests written in Python) now require Python 3.
2303 * systemd unit tests can now run standalone (without the source or
2304 build directories), and can be installed into /usr/lib/systemd/tests/
2305 with 'make install-tests'.
2307 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
2308 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
2311 * Support for the %c, %r, %R specifiers in unit files has been
2312 removed. Specifiers are not supposed to be dependent on configuration
2313 in the unit file itself (so that they resolve the same regardless
2314 where used in the unit files), but these specifiers were influenced
2315 by the Slice= option.
2317 * The shell invoked by debug-shell.service now defaults to /bin/sh in
2318 all cases. If distributions want to use a different shell for this
2319 purpose (for example Fedora's /sbin/sushell) they need to specify
2320 this explicitly at configure time using --with-debug-shell=.
2322 * The confirmation spawn prompt has been reworked to offer the
2325 (c)ontinue, proceed without asking anymore
2326 (D)ump, show the state of the unit
2327 (f)ail, don't execute the command and pretend it failed
2329 (i)nfo, show a short summary of the unit
2330 (j)obs, show jobs that are in progress
2331 (s)kip, don't execute the command and pretend it succeeded
2332 (y)es, execute the command
2334 The 'n' choice for the confirmation spawn prompt has been removed,
2335 because its meaning was confusing.
2337 The prompt may now also be redirected to an alternative console by
2338 specifying the console as parameter to systemd.confirm_spawn=.
2340 * Services of Type=notify require a READY=1 notification to be sent
2341 during startup. If no such message is sent, the service now fails,
2342 even if the main process exited with a successful exit code.
2344 * Services that fail to start up correctly now always have their
2345 ExecStopPost= commands executed. Previously, they'd enter "failed"
2346 state directly, without executing these commands.
2348 * The option MulticastDNS= of network configuration files has acquired
2349 an actual implementation. With MulticastDNS=yes a host can resolve
2350 names of remote hosts and reply to mDNS A and AAAA requests.
2352 * When units are about to be started an additional check is now done to
2353 ensure that all dependencies of type BindsTo= (when used in
2354 combination with After=) have been started.
2356 * systemd-analyze gained a new verb "syscall-filter" which shows which
2357 system call groups are defined for the SystemCallFilter= unit file
2358 setting, and which system calls they contain.
2360 * A new system call filter group "@filesystem" has been added,
2361 consisting of various file system related system calls. Group
2362 "@reboot" has been added, covering reboot, kexec and shutdown related
2363 calls. Finally, group "@swap" has been added covering swap
2364 configuration related calls.
2366 * A new unit file option RestrictNamespaces= has been added that may be
2367 used to restrict access to the various process namespace types the
2368 Linux kernel provides. Specifically, it may be used to take away the
2369 right for a service unit to create additional file system, network,
2370 user, and other namespaces. This sandboxing option is particularly
2371 relevant due to the high amount of recently discovered namespacing
2372 related vulnerabilities in the kernel.
2374 * systemd-udev's .link files gained support for a new AutoNegotiation=
2375 setting for configuring Ethernet auto-negotiation.
2377 * systemd-networkd's .network files gained support for a new
2378 ListenPort= setting in the [DHCP] section to explicitly configure the
2379 UDP client port the DHCP client shall listen on.
2381 * .network files gained a new Unmanaged= boolean setting for explicitly
2382 excluding one or more interfaces from management by systemd-networkd.
2384 * The systemd-networkd ProxyARP= option has been renamed to
2385 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
2386 renamed to ReduceARPProxy=. The old names continue to be available
2389 * systemd-networkd gained support for configuring IPv6 Proxy NDP
2390 addresses via the new IPv6ProxyNDPAddress= .network file setting.
2392 * systemd-networkd's bonding device support gained support for two new
2393 configuration options ActiveSlave= and PrimarySlave=.
2395 * The various options in the [Match] section of .network files gained
2396 support for negative matching.
2398 * New systemd-specific mount options are now understood in /etc/fstab:
2400 x-systemd.mount-timeout= may be used to configure the maximum
2401 permitted runtime of the mount command.
2403 x-systemd.device-bound may be set to bind a mount point to its
2404 backing device unit, in order to automatically remove a mount point
2405 if its backing device is unplugged. This option may also be
2406 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
2407 on the block device, which is now automatically set for all CDROM
2408 drives, so that mounted CDs are automatically unmounted when they are
2409 removed from the drive.
2411 x-systemd.after= and x-systemd.before= may be used to explicitly
2412 order a mount after or before another unit or mount point.
2414 * Enqueued start jobs for device units are now automatically garbage
2415 collected if there are no jobs waiting for them anymore.
2417 * systemctl list-jobs gained two new switches: with --after, for every
2418 queued job the jobs it's waiting for are shown; with --before the
2419 jobs which it's blocking are shown.
2421 * systemd-nspawn gained support for ephemeral boots from disk images
2422 (or in other words: --ephemeral and --image= may now be
2423 combined). Moreover, ephemeral boots are now supported for normal
2424 directories, even if the backing file system is not btrfs. Of course,
2425 if the file system does not support file system snapshots or
2426 reflinks, the initial copy operation will be relatively expensive, but
2427 this should still be suitable for many use cases.
2429 * Calendar time specifications in .timer units now support
2430 specifications relative to the end of a month by using "~" instead of
2431 "-" as separator between month and day. For example, "*-02~03" means
2432 "the third last day in February". In addition a new syntax for
2433 repeated events has been added using the "/" character. For example,
2434 "9..17/2:00" means "every two hours from 9am to 5pm".
2436 * systemd-socket-proxyd gained a new parameter --connections-max= for
2437 configuring the maximum number of concurrent connections.
2439 * sd-id128 gained a new API for generating unique IDs for the host in a
2440 way that does not leak the machine ID. Specifically,
2441 sd_id128_get_machine_app_specific() derives an ID based on the
2442 machine ID a in well-defined, non-reversible, stable way. This is
2443 useful whenever an identifier for the host is needed but where the
2444 identifier shall not be useful to identify the system beyond the
2445 scope of the application itself. (Internally this uses HMAC-SHA256 as
2446 keyed hash function using the machine ID as input.)
2448 * NotifyAccess= gained a new supported value "exec". When set
2449 notifications are accepted from all processes systemd itself invoked,
2450 including all control processes.
2452 * .nspawn files gained support for defining overlay mounts using the
2453 Overlay= and OverlayReadOnly= options. Previously this functionality
2454 was only available on the systemd-nspawn command line.
2456 * systemd-nspawn's --bind= and --overlay= options gained support for
2457 bind/overlay mounts whose source lies within the container tree by
2458 prefixing the source path with "+".
2460 * systemd-nspawn's --bind= and --overlay= options gained support for
2461 automatically allocating a temporary source directory in /var/tmp
2462 that is removed when the container dies. Specifically, if the source
2463 directory is specified as empty string this mechanism is selected. An
2464 example usage is --overlay=+/var::/var, which creates an overlay
2465 mount based on the original /var contained in the image, overlayed
2466 with a temporary directory in the host's /var/tmp. This way changes
2467 to /var are automatically flushed when the container shuts down.
2469 * systemd-nspawn --image= option does now permit raw file system block
2470 devices (in addition to images containing partition tables, as
2473 * The disk image dissection logic in systemd-nspawn gained support for
2474 automatically setting up LUKS encrypted as well as Verity protected
2475 partitions. When a container is booted from an encrypted image the
2476 passphrase is queried at start-up time. When a container with Verity
2477 data is started, the root hash is search in a ".roothash" file
2478 accompanying the disk image (alternatively, pass the root hash via
2479 the new --root-hash= command line option).
2481 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
2482 be used to dissect disk images the same way as systemd-nspawn does
2483 it, following the Bootable Partition Specification. It may even be
2484 used to mount disk images with complex partition setups (including
2485 LUKS and Verity partitions) to a local host directory, in order to
2486 inspect them. This tool is not considered public API (yet), and is
2487 thus not installed into /usr/bin. Please do not rely on its
2488 existence, since it might go away or be changed in later systemd
2491 * A new generator "systemd-verity-generator" has been added, similar in
2492 style to "systemd-cryptsetup-generator", permitting automatic setup of
2493 Verity root partitions when systemd boots up. In order to make use of
2494 this your partition setup should follow the Discoverable Partitions
2495 Specification, and the GPT partition ID of the root file system
2496 partition should be identical to the upper 128bit of the Verity root
2497 hash. The GPT partition ID of the Verity partition protecting it
2498 should be the lower 128bit of the Verity root hash. If the partition
2499 image follows this model it is sufficient to specify a single
2500 "roothash=" kernel command line argument to both configure which root
2501 image and verity partition to use as well as the root hash for
2502 it. Note that systemd-nspawn's Verity support follows the same
2503 semantics, meaning that disk images with proper Verity data in place
2504 may be booted in containers with systemd-nspawn as well as on
2505 physical systems via the verity generator. Also note that the "mkosi"
2506 tool available at https://github.com/systemd/mkosi has been updated
2507 to generate Verity protected disk images following this scheme. In
2508 fact, it has been updated to generate disk images that optionally
2509 implement a complete UEFI SecureBoot trust chain, involving a signed
2510 kernel and initrd image that incorporates such a root hash as well as
2511 a Verity-enabled root partition.
2513 * The hardware database (hwdb) udev supports has been updated to carry
2514 accelerometer quirks.
2516 * All system services are now run with a fresh kernel keyring set up
2517 for them. The invocation ID is stored by default in it, thus
2518 providing a safe, non-overridable way to determine the invocation
2521 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
2522 options for bind mounting arbitrary paths in a service-specific
2523 way. When these options are used, arbitrary host or service files and
2524 directories may be mounted to arbitrary locations in the service's
2527 * Documentation has been added that lists all of systemd's low-level
2528 environment variables:
2530 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
2532 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
2533 whether a specific socket file descriptor matches a specified socket
2536 * systemd-firstboot has been updated to check for the
2537 systemd.firstboot= kernel command line option. It accepts a boolean
2538 and when set to false the first boot questions are skipped.
2540 * systemd-fstab-generator has been updated to check for the
2541 systemd.volatile= kernel command line option, which either takes an
2542 optional boolean parameter or the special value "state". If used the
2543 system may be booted in a "volatile" boot mode. Specifically,
2544 "systemd.volatile" is used, the root directory will be mounted as
2545 tmpfs, and only /usr is mounted from the actual root file system. If
2546 "systemd.volatile=state" is used, the root directory will be mounted
2547 as usual, but /var is mounted as tmpfs. This concept provides similar
2548 functionality as systemd-nspawn's --volatile= option, but provides it
2549 on physical boots. Use this option for implementing stateless
2550 systems, or testing systems with all state and/or configuration reset
2551 to the defaults. (Note though that many distributions are not
2552 prepared to boot up without a populated /etc or /var, though.)
2554 * systemd-gpt-auto-generator gained support for LUKS encrypted root
2555 partitions. Previously it only supported LUKS encrypted partitions
2556 for all other uses, except for the root partition itself.
2558 * Socket units gained support for listening on AF_VSOCK sockets for
2559 communication in virtualized QEMU environments.
2561 * The "configure" script gained a new option --with-fallback-hostname=
2562 for specifying the fallback hostname to use if none is configured in
2563 /etc/hostname. For example, by specifying
2564 --with-fallback-hostname=fedora it is possible to default to a
2565 hostname of "fedora" on pristine installations.
2567 * systemd-cgls gained support for a new --unit= switch for listing only
2568 the control groups of a specific unit. Similar --user-unit= has been
2569 added for listing only the control groups of a specific user unit.
2571 * systemd-mount gained a new --umount switch for unmounting a mount or
2572 automount point (and all mount/automount points below it).
2574 * systemd will now refuse full configuration reloads (via systemctl
2575 daemon-reload and related calls) unless at least 16MiB of free space
2576 are available in /run. This is a safety precaution in order to ensure
2577 that generators can safely operate after the reload completed.
2579 * A new unit file option RootImage= has been added, which has a similar
2580 effect as RootDirectory= but mounts the service's root directory from
2581 a disk image instead of plain directory. This logic reuses the same
2582 image dissection and mount logic that systemd-nspawn already uses,
2583 and hence supports any disk images systemd-nspawn supports, including
2584 those following the Discoverable Partition Specification, as well as
2585 Verity enabled images. This option enables systemd to run system
2586 services directly off disk images acting as resource bundles,
2587 possibly even including full integrity data.
2589 * A new MountAPIVFS= unit file option has been added, taking a boolean
2590 argument. If enabled /proc, /sys and /dev (collectively called the
2591 "API VFS") will be mounted for the service. This is only relevant if
2592 RootDirectory= or RootImage= is used for the service, as these mounts
2593 are of course in place in the host mount namespace anyway.
2595 * systemd-nspawn gained support for a new --pivot-root= switch. If
2596 specified the root directory within the container image is pivoted to
2597 the specified mount point, while the original root disk is moved to a
2598 different place. This option enables booting of ostree images
2599 directly with systemd-nspawn.
2601 * The systemd build scripts will no longer complain if the NTP server
2602 addresses are not changed from the defaults. Google now supports
2603 these NTP servers officially. We still recommend downstreams to
2604 properly register an NTP pool with the NTP pool project though.
2606 * coredumpctl gained a new "--reverse" option for printing the list
2607 of coredumps in reverse order.
2609 * coredumpctl will now show additional information about truncated and
2610 inaccessible coredumps, as well as coredumps that are still being
2611 processed. It also gained a new --quiet switch for suppressing
2612 additional informational message in its output.
2614 * coredumpctl gained support for only showing coredumps newer and/or
2615 older than specific timestamps, using the new --since= and --until=
2616 options, reminiscent of journalctl's options by the same name.
2618 * The systemd-coredump logic has been improved so that it may be reused
2619 to collect backtraces in non-compiled languages, for example in
2620 scripting languages such as Python.
2622 * machinectl will now show the UID shift of local containers, if user
2623 namespacing is enabled for them.
2625 * systemd will now optionally run "environment generator" binaries at
2626 configuration load time. They may be used to add environment
2627 variables to the environment block passed to services invoked. One
2628 user environment generator is shipped by default that sets up
2629 environment variables based on files dropped into /etc/environment.d
2630 and ~/.config/environment.d/.
2632 * systemd-resolved now includes the new, recently published 2017 DNSSEC
2635 * hostnamed has been updated to report a new chassis type of
2636 "convertible" to cover "foldable" laptops that can both act as a
2637 tablet and as a laptop, such as various Lenovo Yoga devices.
2639 Contributions from: Adrián López, Alexander Galanin, Alexander
2640 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
2641 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
2642 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
2643 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
2644 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
2645 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
2646 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
2647 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
2648 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
2649 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
2650 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
2651 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
2652 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
2653 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
2654 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
2655 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
2656 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
2657 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
2658 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
2659 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
2660 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
2661 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
2662 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
2663 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
2664 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
2665 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
2666 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
2669 — Berlin, 2017-03-01
2673 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
2674 RestrictAddressFamilies= enabled. These sandboxing options should
2675 generally be compatible with the various external udev call-out
2676 binaries we are aware of, however there may be exceptions, in
2677 particular when exotic languages for these call-outs are used. In
2678 this case, consider turning off these settings locally.
2680 * The new RemoveIPC= option can be used to remove IPC objects owned by
2681 the user or group of a service when that service exits.
2683 * The new ProtectKernelModules= option can be used to disable explicit
2684 load and unload operations of kernel modules by a service. In
2685 addition access to /usr/lib/modules is removed if this option is set.
2687 * ProtectSystem= option gained a new value "strict", which causes the
2688 whole file system tree with the exception of /dev, /proc, and /sys,
2689 to be remounted read-only for a service.
2691 * The new ProtectKernelTunables= option can be used to disable
2692 modification of configuration files in /sys and /proc by a service.
2693 Various directories and files are remounted read-only, so access is
2694 restricted even if the file permissions would allow it.
2696 * The new ProtectControlGroups= option can be used to disable write
2697 access by a service to /sys/fs/cgroup.
2699 * Various systemd services have been hardened with
2700 ProtectKernelTunables=yes, ProtectControlGroups=yes,
2701 RestrictAddressFamilies=.
2703 * Support for dynamically creating users for the lifetime of a service
2704 has been added. If DynamicUser=yes is specified, user and group IDs
2705 will be allocated from the range 61184..65519 for the lifetime of the
2706 service. They can be resolved using the new nss-systemd.so NSS
2707 module. The module must be enabled in /etc/nsswitch.conf. Services
2708 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
2709 any resources allocated by the service will be cleaned up when the
2710 service exits. They also have ProtectHome=read-only and
2711 ProtectSystem=strict enabled, so they are not able to make any
2712 permanent modifications to the system.
2714 * The nss-systemd module also always resolves root and nobody, making
2715 it possible to have no /etc/passwd or /etc/group files in minimal
2716 container or chroot environments.
2718 * Services may be started with their own user namespace using the new
2719 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
2720 under which the service is running are mapped. All other users are
2723 * Support for the cgroup namespace has been added to systemd-nspawn. If
2724 supported by kernel, the container system started by systemd-nspawn
2725 will have its own view of the cgroup hierarchy. This new behaviour
2726 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
2728 * The new MemorySwapMax= option can be used to limit the maximum swap
2729 usage under the unified cgroup hierarchy.
2731 * Support for the CPU controller in the unified cgroup hierarchy has
2732 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
2733 options. This controller requires out-of-tree patches for the kernel
2734 and the support is provisional.
2736 * Mount and automount units may now be created transiently
2737 (i.e. dynamically at runtime via the bus API, instead of requiring
2738 unit files in the file system).
2740 * systemd-mount is a new tool which may mount file systems – much like
2741 mount(8), optionally pulling in additional dependencies through
2742 transient .mount and .automount units. For example, this tool
2743 automatically runs fsck on a backing block device before mounting,
2744 and allows the automount logic to be used dynamically from the
2745 command line for establishing mount points. This tool is particularly
2746 useful when dealing with removable media, as it will ensure fsck is
2747 run – if necessary – before the first access and that the file system
2748 is quickly unmounted after each access by utilizing the automount
2749 logic. This maximizes the chance that the file system on the
2750 removable media stays in a clean state, and if it isn't in a clean
2751 state is fixed automatically.
2753 * LazyUnmount=yes option for mount units has been added to expose the
2754 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
2757 * /efi will be used as the mount point of the EFI boot partition, if
2758 the directory is present, and the mount point was not configured
2759 through other means (e.g. fstab). If /efi directory does not exist,
2760 /boot will be used as before. This makes it easier to automatically
2761 mount the EFI partition on systems where /boot is used for something
2764 * When operating on GPT disk images for containers, systemd-nspawn will
2765 now mount the ESP to /boot or /efi according to the same rules as PID
2766 1 running on a host. This allows tools like "bootctl" to operate
2767 correctly within such containers, in order to make container images
2768 bootable on physical systems.
2770 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
2772 * Two new user session targets have been added to support running
2773 graphical sessions under the systemd --user instance:
2774 graphical-session.target and graphical-session-pre.target. See
2775 systemd.special(7) for a description of how those targets should be
2778 * The vconsole initialization code has been significantly reworked to
2779 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
2780 support unicode keymaps. Font and keymap configuration will now be
2781 copied to all allocated virtual consoles.
2783 * FreeBSD's bhyve virtualization is now detected.
2785 * Information recorded in the journal for core dumps now includes the
2786 contents of /proc/mountinfo and the command line of the process at
2787 the top of the process hierarchy (which is usually the init process
2790 * systemd-journal-gatewayd learned the --directory= option to serve
2791 files from the specified location.
2793 * journalctl --root=… can be used to peruse the journal in the
2794 /var/log/ directories inside of a container tree. This is similar to
2795 the existing --machine= option, but does not require the container to
2798 * The hardware database has been extended to support
2799 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
2802 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
2803 specify the click rate for mice which include a horizontal wheel with
2804 a click rate that is different than the one for the vertical wheel.
2806 * systemd-run gained a new --wait option that makes service execution
2807 synchronous. (Specifically, the command will not return until the
2808 specified service binary exited.)
2810 * systemctl gained a new --wait option that causes the start command to
2811 wait until the units being started have terminated again.
2813 * A new journal output mode "short-full" has been added which displays
2814 timestamps with abbreviated English day names and adds a timezone
2815 suffix. Those timestamps include more information than the default
2816 "short" output mode, and can be passed directly to journalctl's
2817 --since= and --until= options.
2819 * /etc/resolv.conf will be bind-mounted into containers started by
2820 systemd-nspawn, if possible, so any changes to resolv.conf contents
2821 are automatically propagated to the container.
2823 * The number of instances for socket-activated services originating
2824 from a single IP address can be limited with
2825 MaxConnectionsPerSource=, extending the existing setting of
2828 * systemd-networkd gained support for vcan ("Virtual CAN") interface
2831 * .netdev and .network configuration can now be extended through
2834 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
2835 Segmentation Offload, Generic Receive Offload, Large Receive Offload
2836 can be enabled and disabled using the new UDPSegmentationOffload=,
2837 TCPSegmentationOffload=, GenericSegmentationOffload=,
2838 GenericReceiveOffload=, LargeReceiveOffload= options in the
2839 [Link] section of .link files.
2841 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
2842 Port VLAN ID can be configured for bridge devices using the new STP=,
2843 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
2844 section of .netdev files.
2846 * The route table to which routes received over DHCP or RA should be
2847 added can be configured with the new RouteTable= option in the [DHCP]
2848 and [IPv6AcceptRA] sections of .network files.
2850 * The Address Resolution Protocol can be disabled on links managed by
2851 systemd-networkd using the ARP=no setting in the [Link] section of
2854 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
2855 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
2856 encode information about the result and exit codes of the current
2857 service runtime cycle.
2859 * systemd-sysctl will now configure kernel parameters in the order
2860 they occur in the configuration files. This matches what sysctl
2861 has been traditionally doing.
2863 * kernel-install "plugins" that are executed to perform various
2864 tasks after a new kernel is added and before an old one is removed
2865 can now return a special value to terminate the procedure and
2866 prevent any later plugins from running.
2868 * Journald's SplitMode=login setting has been deprecated. It has been
2869 removed from documentation, and its use is discouraged. In a future
2870 release it will be completely removed, and made equivalent to current
2871 default of SplitMode=uid.
2873 * Storage=both option setting in /etc/systemd/coredump.conf has been
2874 removed. With fast LZ4 compression storing the core dump twice is not
2877 * The --share-system systemd-nspawn option has been replaced with an
2878 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
2879 this functionality is discouraged. In addition the variables
2880 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
2881 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
2882 individual namespaces.
2884 * "machinectl list" now shows the IP address of running containers in
2885 the output, as well as OS release information.
2887 * "loginctl list" now shows the TTY of each session in the output.
2889 * sd-bus gained new API calls sd_bus_track_set_recursive(),
2890 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
2891 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
2892 tracking objects in a "recursive" mode, where a single client can be
2893 counted multiple times, if it takes multiple references.
2895 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
2896 sd_bus_get_exit_on_disconnect(). They may be used to make a
2897 process using sd-bus automatically exit if the bus connection is
2900 * Bus clients of the service manager may now "pin" loaded units into
2901 memory, by taking an explicit reference on them. This is useful to
2902 ensure the client can retrieve runtime data about the service even
2903 after the service completed execution. Taking such a reference is
2904 available only for privileged clients and should be helpful to watch
2905 running services in a race-free manner, and in particular collect
2906 information about exit statuses and results.
2908 * The nss-resolve module has been changed to strictly return UNAVAIL
2909 when communication via D-Bus with resolved failed, and NOTFOUND when
2910 a lookup completed but was negative. This means it is now possible to
2911 neatly configure fallbacks using nsswitch.conf result checking
2912 expressions. Taking benefit of this, the new recommended
2913 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
2915 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
2917 * A new setting CtrlAltDelBurstAction= has been added to
2918 /etc/systemd/system.conf which may be used to configure the precise
2919 behaviour if the user on the console presses Ctrl-Alt-Del more often
2920 than 7 times in 2s. Previously this would unconditionally result in
2921 an expedited, immediate reboot. With this new setting the precise
2922 operation may be configured in more detail, and also turned off
2925 * In .netdev files two new settings RemoteChecksumTx= and
2926 RemoteChecksumRx= are now understood that permit configuring the
2927 remote checksumming logic for VXLAN networks.
2929 * The service manager learnt a new "invocation ID" concept for invoked
2930 services. Each runtime cycle of a service will get a new invocation
2931 ID (a 128bit random UUID) assigned that identifies the current
2932 run of the service uniquely and globally. A new invocation ID
2933 is generated each time a service starts up. The journal will store
2934 the invocation ID of a service along with any logged messages, thus
2935 making the invocation ID useful for matching the online runtime of a
2936 service with the offline log data it generated in a safe way without
2937 relying on synchronized timestamps. In many ways this new service
2938 invocation ID concept is similar to the kernel's boot ID concept that
2939 uniquely and globally identifies the runtime of each boot. The
2940 invocation ID of a service is passed to the service itself via an
2941 environment variable ($INVOCATION_ID). A new bus call
2942 GetUnitByInvocationID() has been added that is similar to GetUnit()
2943 but instead of retrieving the bus path for a unit by its name
2944 retrieves it by its invocation ID. The returned path is valid only as
2945 long as the passed invocation ID is current.
2947 * systemd-resolved gained a new "DNSStubListener" setting in
2948 resolved.conf. It either takes a boolean value or the special values
2949 "udp" and "tcp", and configures whether to enable the stub DNS
2950 listener on 127.0.0.53:53.
2952 * IP addresses configured via networkd may now carry additional
2953 configuration settings supported by the kernel. New options include:
2954 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
2955 PrefixRoute=, AutoJoin=.
2957 * The PAM configuration fragment file for "user@.service" shipped with
2958 systemd (i.e. the --user instance of systemd) has been stripped to
2959 the minimum necessary to make the system boot. Previously, it
2960 contained Fedora-specific stanzas that did not apply to other
2961 distributions. It is expected that downstream distributions add
2962 additional configuration lines, matching their needs to this file,
2963 using it only as rough template of what systemd itself needs. Note
2964 that this reduced fragment does not even include an invocation of
2965 pam_limits which most distributions probably want to add, even though
2966 systemd itself does not need it. (There's also the new build time
2967 option --with-pamconfdir=no to disable installation of the PAM
2970 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
2971 capability is now also dropped from its set (in addition to
2972 CAP_SYS_MKNOD as before).
2974 * In service unit files it is now possible to connect a specific named
2975 file descriptor with stdin/stdout/stdout of an executed service. The
2976 name may be specified in matching .socket units using the
2977 FileDescriptorName= setting.
2979 * A number of journal settings may now be configured on the kernel
2980 command line. Specifically, the following options are now understood:
2981 systemd.journald.max_level_console=,
2982 systemd.journald.max_level_store=,
2983 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
2984 systemd.journald.max_level_wall=.
2986 * "systemctl is-enabled --full" will now show by which symlinks a unit
2987 file is enabled in the unit dependency tree.
2989 * Support for VeraCrypt encrypted partitions has been added to the
2990 "cryptsetup" logic and /etc/crypttab.
2992 * systemd-detect-virt gained support for a new --private-users switch
2993 that checks whether the invoking processes are running inside a user
2994 namespace. Similar, a new special value "private-users" for the
2995 existing ConditionVirtualization= setting has been added, permitting
2996 skipping of specific units in user namespace environments.
2998 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
2999 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
3000 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
3001 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
3002 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
3003 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
3004 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
3005 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
3006 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
3007 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
3008 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
3009 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
3010 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
3011 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
3012 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
3013 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
3014 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
3015 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
3016 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
3017 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
3018 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
3019 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
3020 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
3021 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
3022 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
3023 Jędrzejewski-Szmek, Zeal Jagannatha
3025 — Santa Fe, 2016-11-03
3029 * In service units the various ExecXYZ= settings have been extended
3030 with an additional special character as first argument of the
3031 assigned value: if the character '+' is used the specified command
3032 line it will be run with full privileges, regardless of User=,
3033 Group=, CapabilityBoundingSet= and similar options. The effect is
3034 similar to the existing PermissionsStartOnly= option, but allows
3035 configuration of this concept for each executed command line
3038 * Services may now alter the service watchdog timeout at runtime by
3039 sending a WATCHDOG_USEC= message via sd_notify().
3041 * MemoryLimit= and related unit settings now optionally take percentage
3042 specifications. The percentage is taken relative to the amount of
3043 physical memory in the system (or in case of containers, the assigned
3044 amount of memory). This allows scaling service resources neatly with
3045 the amount of RAM available on the system. Similarly, systemd-logind's
3046 RuntimeDirectorySize= option now also optionally takes percentage
3049 * In similar fashion TasksMax= takes percentage values now, too. The
3050 value is taken relative to the configured maximum number of processes
3051 on the system. The per-service task maximum has been changed to 15%
3052 using this functionality. (Effectively this is an increase of 512 →
3053 4915 for service units, given the kernel's default pid_max setting.)
3055 * Calendar time specifications in .timer units now understand a ".."
3056 syntax for time ranges. Example: "4..7:10" may now be used for
3057 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
3060 * The InaccessableDirectories=, ReadOnlyDirectories= and
3061 ReadWriteDirectories= unit file settings have been renamed to
3062 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
3063 applied to all kinds of file nodes, and not just directories, with
3064 the exception of symlinks. Specifically these settings may now be
3065 used on block and character device nodes, UNIX sockets and FIFOS as
3066 well as regular files. The old names of these settings remain
3067 available for compatibility.
3069 * systemd will now log about all service processes it kills forcibly
3070 (using SIGKILL) because they remained after the clean shutdown phase
3071 of the service completed. This should help identifying services that
3072 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
3073 systemd-logind's configuration a similar log message is generated for
3074 processes killed at the end of each session due to this setting.
3076 * systemd will now set the $JOURNAL_STREAM environment variable for all
3077 services whose stdout/stderr are connected to the Journal (which
3078 effectively means by default: all services). The variable contains
3079 the device and inode number of the file descriptor used for
3080 stdout/stderr. This may be used by invoked programs to detect whether
3081 their stdout/stderr is connected to the Journal, in which case they
3082 can switch over to direct Journal communication, thus being able to
3083 pass extended, structured metadata along with their log messages. As
3084 one example, this is now used by glib's logging primitives.
3086 * When using systemd's default tmp.mount unit for /tmp, the mount point
3087 will now be established with the "nosuid" and "nodev" options. This
3088 avoids privilege escalation attacks that put traps and exploits into
3089 /tmp. However, this might cause problems if you e. g. put container
3090 images or overlays into /tmp; if you need this, override tmp.mount's
3091 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
3094 * systemd now supports the "memory" cgroup controller also on
3097 * The systemd-cgtop tool now optionally takes a control group path as
3098 command line argument. If specified, the control group list shown is
3099 limited to subgroups of that group.
3101 * The SystemCallFilter= unit file setting gained support for
3102 pre-defined, named system call filter sets. For example
3103 SystemCallFilter=@clock is now an effective way to make all clock
3104 changing-related system calls unavailable to a service. A number of
3105 similar pre-defined groups are defined. Writing system call filters
3106 for system services is simplified substantially with this new
3107 concept. Accordingly, all of systemd's own, long-running services now
3108 enable system call filtering based on this, by default.
3110 * A new service setting MemoryDenyWriteExecute= has been added, taking
3111 a boolean value. If turned on, a service may no longer create memory
3112 mappings that are writable and executable at the same time. This
3113 enhances security for services where this is enabled as it becomes
3114 harder to dynamically write and then execute memory in exploited
3115 service processes. This option has been enabled for all of systemd's
3116 own long-running services.
3118 * A new RestrictRealtime= service setting has been added, taking a
3119 boolean argument. If set the service's processes may no longer
3120 acquire realtime scheduling. This improves security as realtime
3121 scheduling may otherwise be used to easily freeze the system.
3123 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
3124 value. This may be used for requesting that the system manager inside
3125 of the container reports start-up completion to nspawn which then
3126 propagates this notification further to the service manager
3127 supervising nspawn itself. A related option NotifyReady= in .nspawn
3128 files has been added too. This functionality allows ordering of the
3129 start-up of multiple containers using the usual systemd ordering
3132 * machinectl gained a new command "stop" that is an alias for
3135 * systemd-resolved gained support for contacting DNS servers on
3136 link-local IPv6 addresses.
3138 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
3139 its caches. A method call for requesting the same operation has been
3140 added to the bus API too, and is made available via "systemd-resolve
3143 * systemd-resolve gained a new --status switch. If passed a brief
3144 summary of the used DNS configuration with per-interface information
3147 * resolved.conf gained a new Cache= boolean option, defaulting to
3148 on. If turned off local DNS caching is disabled. This comes with a
3149 performance penalty in particular when DNSSEC is enabled. Note that
3150 resolved disables its internal caching implicitly anyway, when the
3151 configured DNS server is on a host-local IP address such as ::1 or
3152 127.0.0.1, thus automatically avoiding double local caching.
3154 * systemd-resolved now listens on the local IP address 127.0.0.53:53
3155 for DNS requests. This improves compatibility with local programs
3156 that do not use the libc NSS or systemd-resolved's bus APIs for name
3157 resolution. This minimal DNS service is only available to local
3158 programs and does not implement the full DNS protocol, but enough to
3159 cover local DNS clients. A new, static resolv.conf file, listing just
3160 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
3161 now recommended to make /etc/resolv.conf a symlink to this file in
3162 order to route all DNS lookups to systemd-resolved, regardless if
3163 done via NSS, the bus API or raw DNS packets. Note that this local
3164 DNS service is not as fully featured as the libc NSS or
3165 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
3166 used to deliver link-local address information (as this implies
3167 sending a local interface index along), LLMNR/mDNS support via this
3168 interface is severely restricted. It is thus strongly recommended for
3169 all applications to use the libc NSS API or native systemd-resolved
3172 * systemd-networkd's bridge support learned a new setting
3173 VLANFiltering= for controlling VLAN filtering. Moreover a new section
3174 in .network files has been added for configuring VLAN bridging in
3175 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
3177 * systemd-networkd's IPv6 Router Advertisement code now makes use of
3178 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
3179 now be acquired without relying on DHCPv6. Two new options
3180 UseDomains= and UseDNS= have been added to configure this behaviour.
3182 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
3183 renamed IPv6AcceptRA=, without altering its behaviour. The old
3184 setting name remains available for compatibility reasons.
3186 * The systemd-networkd VTI/VTI6 tunneling support gained new options
3187 Key=, InputKey= and OutputKey=.
3189 * systemd-networkd gained support for VRF ("Virtual Routing Function")
3190 interface configuration.
3192 * "systemctl edit" may now be used to create new unit files by
3193 specifying the --force switch.
3195 * sd-event gained a new function sd_event_get_iteration() for
3196 requesting the current iteration counter of the event loop. It starts
3197 at zero and is increased by one with each event loop iteration.
3199 * A new rpm macro %systemd_ordering is provided by the macros.systemd
3200 file. It can be used in lieu of %systemd_requires in packages which
3201 don't use any systemd functionality and are intended to be installed
3202 in minimal containers without systemd present. This macro provides
3203 ordering dependencies to ensure that if the package is installed in
3204 the same rpm transaction as systemd, systemd will be installed before
3205 the scriptlets for the package are executed, allowing unit presets
3208 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
3209 been added to simplify packaging of generators.
3211 * The os-release file gained VERSION_CODENAME field for the
3212 distribution nickname (e.g. VERSION_CODENAME=woody).
3214 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
3215 can be set to disable parsing of metadata and the creation
3216 of persistent symlinks for that device.
3218 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
3219 to make them available to logged-in users has been reverted.
3221 * Much of the common code of the various systemd components is now
3222 built into an internal shared library libsystemd-shared-231.so
3223 (incorporating the systemd version number in the name, to be updated
3224 with future releases) that the components link to. This should
3225 decrease systemd footprint both in memory during runtime and on
3226 disk. Note that the shared library is not for public use, and is
3227 neither API nor ABI stable, but is likely to change with every new
3228 released update. Packagers need to make sure that binaries
3229 linking to libsystemd-shared.so are updated in step with the
3232 * Configuration for "mkosi" is now part of the systemd
3233 repository. mkosi is a tool to easily build legacy-free OS images,
3234 and is available on github: https://github.com/systemd/mkosi. If
3235 "mkosi" is invoked in the build tree a new raw OS image is generated
3236 incorporating the systemd sources currently being worked on and a
3237 clean, fresh distribution installation. The generated OS image may be
3238 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
3239 UEFI PC. This functionality is particularly useful to easily test
3240 local changes made to systemd in a pristine, defined environment. See
3241 doc/HACKING for details.
3243 * configure learned the --with-support-url= option to specify the
3244 distribution's bugtracker.
3246 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
3247 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
3248 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
3249 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
3250 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
3251 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
3252 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
3253 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
3254 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
3255 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
3256 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
3257 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
3258 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
3259 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
3260 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
3261 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
3262 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
3263 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
3264 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
3266 — Berlin, 2016-07-25
3270 * DNSSEC is now turned on by default in systemd-resolved (in
3271 "allow-downgrade" mode), but may be turned off during compile time by
3272 passing "--with-default-dnssec=no" to "configure" (and of course,
3273 during runtime with DNSSEC= in resolved.conf). We recommend
3274 downstreams to leave this on at least during development cycles and
3275 report any issues with the DNSSEC logic upstream. We are very
3276 interested in collecting feedback about the DNSSEC validator and its
3277 limitations in the wild. Note however, that DNSSEC support is
3278 probably nothing downstreams should turn on in stable distros just
3279 yet, as it might create incompatibilities with a few DNS servers and
3280 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
3281 automatically whenever we detect such incompatible setups, but there
3282 might be systems we do not cover yet. Hence: please help us testing
3283 the DNSSEC code, leave this on where you can, report back, but then
3284 again don't consider turning this on in your stable, LTS or
3285 production release just yet. (Note that you have to enable
3286 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
3287 and its DNSSEC mode for host name resolution from local
3290 * systemd-resolve conveniently resolves DANE records with the --tlsa
3291 option and OPENPGPKEY records with the --openpgp option. It also
3292 supports dumping raw DNS record data via the new --raw= switch.
3294 * systemd-logind will now by default terminate user processes that are
3295 part of the user session scope unit (session-XX.scope) when the user
3296 logs out. This behavior is controlled by the KillUserProcesses=
3297 setting in logind.conf, and the previous default of "no" is now
3298 changed to "yes". This means that user sessions will be properly
3299 cleaned up after, but additional steps are necessary to allow
3300 intentionally long-running processes to survive logout.
3302 While the user is logged in at least once, user@.service is running,
3303 and any service that should survive the end of any individual login
3304 session can be started at a user service or scope using systemd-run.
3305 systemd-run(1) man page has been extended with an example which shows
3306 how to run screen in a scope unit underneath user@.service. The same
3307 command works for tmux.
3309 After the user logs out of all sessions, user@.service will be
3310 terminated too, by default, unless the user has "lingering" enabled.
3311 To effectively allow users to run long-term tasks even if they are
3312 logged out, lingering must be enabled for them. See loginctl(1) for
3313 details. The default polkit policy was modified to allow users to
3314 set lingering for themselves without authentication.
3316 Previous defaults can be restored at compile time by the
3317 --without-kill-user-processes option to "configure".
3319 * systemd-logind gained new configuration settings SessionsMax= and
3320 InhibitorsMax=, both with a default of 8192. It will not register new
3321 user sessions or inhibitors above this limit.
3323 * systemd-logind will now reload configuration on SIGHUP.
3325 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
3326 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
3327 enable. Also, support for the "io" cgroup controller in the unified
3328 hierarchy has been added, so that the "memory", "pids" and "io" are
3329 now the controllers that are supported on the unified hierarchy.
3331 WARNING: it is not possible to use previous systemd versions with
3332 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
3333 is necessary to also update systemd in the initramfs if using the
3334 unified hierarchy. An updated SELinux policy is also required.
3336 * LLDP support has been extended, and both passive (receive-only) and
3337 active (sender) modes are supported. Passive mode ("routers-only") is
3338 enabled by default in systemd-networkd. Active LLDP mode is enabled
3339 by default for containers on the internal network. The "networkctl
3340 lldp" command may be used to list information gathered. "networkctl
3341 status" will also show basic LLDP information on connected peers now.
3343 * The IAID and DUID unique identifier sent in DHCP requests may now be
3344 configured for the system and each .network file managed by
3345 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
3347 * systemd-networkd gained support for configuring proxy ARP support for
3348 each interface, via the ProxyArp= setting in .network files. It also
3349 gained support for configuring the multicast querier feature of
3350 bridge devices, via the new MulticastQuerier= setting in .netdev
3351 files. Similarly, snooping on the IGMP traffic can be controlled
3352 via the new setting MulticastSnooping=.
3354 A new setting PreferredLifetime= has been added for addresses
3355 configured in .network file to configure the lifetime intended for an
3358 The systemd-networkd DHCP server gained the option EmitRouter=, which
3359 defaults to yes, to configure whether the DHCP Option 3 (Router)
3362 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
3363 systemd-socket-activate and installed into /usr/bin. It is now fully
3366 * systemd-journald now uses separate threads to flush changes to disk
3367 when closing journal files, thus reducing impact of slow disk I/O on
3368 logging performance.
3370 * The sd-journal API gained two new calls
3371 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
3372 can be used to open journal files using file descriptors instead of
3373 file or directory paths. sd_journal_open_container() has been
3374 deprecated, sd_journal_open_directory_fd() should be used instead
3375 with the flag SD_JOURNAL_OS_ROOT.
3377 * journalctl learned a new output mode "-o short-unix" that outputs log
3378 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
3379 UTC). It also gained support for a new --no-hostname setting to
3380 suppress the hostname column in the family of "short" output modes.
3382 * systemd-ask-password now optionally skips printing of the password to
3383 stdout with --no-output which can be useful in scripts.
3385 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
3386 (devices tagged with ID_MAKER_TOOL) are now tagged with
3387 "uaccess" and are available to logged in users.
3389 * The DeviceAllow= unit setting now supports specifiers (with "%").
3391 * "systemctl show" gained a new --value switch, which allows print a
3392 only the contents of a specific unit property, without also printing
3393 the property's name. Similar support was added to "show*" verbs
3394 of loginctl and machinectl that output "key=value" lists.
3396 * A new unit type "generated" was added for files dynamically generated
3397 by generator tools. Similarly, a new unit type "transient" is used
3398 for unit files created using the runtime API. "systemctl enable" will
3399 refuse to operate on such files.
3401 * A new command "systemctl revert" has been added that may be used to
3402 revert to the vendor version of a unit file, in case local changes
3403 have been made by adding drop-ins or overriding the unit file.
3405 * "machinectl clean" gained a new verb to automatically remove all or
3406 just hidden container images.
3408 * systemd-tmpfiles gained support for a new line type "e" for emptying
3409 directories, if they exist, without creating them if they don't.
3411 * systemd-nspawn gained support for automatically patching the UID/GIDs
3412 of the owners and the ACLs of all files and directories in a
3413 container tree to match the UID/GID user namespacing range selected
3414 for the container invocation. This mode is enabled via the new
3415 --private-users-chown switch. It also gained support for
3416 automatically choosing a free, previously unused UID/GID range when
3417 starting a container, via the new --private-users=pick setting (which
3418 implies --private-users-chown). Together, these options for the first
3419 time make user namespacing for nspawn containers fully automatic and
3420 thus deployable. The systemd-nspawn@.service template unit file has
3421 been changed to use this functionality by default.
3423 * systemd-nspawn gained a new --network-zone= switch, that allows
3424 creating ad-hoc virtual Ethernet links between multiple containers,
3425 that only exist as long as at least one container referencing them is
3426 running. This allows easy connecting of multiple containers with a
3427 common link that implements an Ethernet broadcast domain. Each of
3428 these network "zones" may be named relatively freely by the user, and
3429 may be referenced by any number of containers, but each container may
3430 only reference one of these "zones". On the lower level, this is
3431 implemented by an automatically managed bridge network interface for
3432 each zone, that is created when the first container referencing its
3433 zone is created and removed when the last one referencing its zone
3436 * The default start timeout may now be configured on the kernel command
3437 line via systemd.default_timeout_start_sec=. It was already
3438 configurable via the DefaultTimeoutStartSec= option in
3439 /etc/systemd/system.conf.
3441 * Socket units gained a new TriggerLimitIntervalSec= and
3442 TriggerLimitBurst= setting to configure a limit on the activation
3443 rate of the socket unit.
3445 * The LimitNICE= setting now optionally takes normal UNIX nice values
3446 in addition to the raw integer limit value. If the specified
3447 parameter is prefixed with "+" or "-" and is in the range -20..19 the
3448 value is understood as UNIX nice value. If not prefixed like this it
3449 is understood as raw RLIMIT_NICE limit.
3451 * Note that the effect of the PrivateDevices= unit file setting changed
3452 slightly with this release: the per-device /dev file system will be
3453 mounted read-only from this version on, and will have "noexec"
3454 set. This (minor) change of behavior might cause some (exceptional)
3455 legacy software to break, when PrivateDevices=yes is set for its
3456 service. Please leave PrivateDevices= off if you run into problems
3459 * systemd-bootchart has been split out to a separate repository:
3460 https://github.com/systemd/systemd-bootchart
3462 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
3463 merged into the kernel in its current form.
3465 * The compatibility libraries libsystemd-daemon.so,
3466 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
3467 which have been deprecated since systemd-209 have been removed along
3468 with the corresponding pkg-config files. All symbols provided by
3469 those libraries are provided by libsystemd.so.
3471 * The Capabilities= unit file setting has been removed (it is ignored
3472 for backwards compatibility). AmbientCapabilities= and
3473 CapabilityBoundingSet= should be used instead.
3475 * A new special target has been added, initrd-root-device.target,
3476 which creates a synchronization point for dependencies of the root
3477 device in early userspace. Initramfs builders must ensure that this
3478 target is now included in early userspace.
3480 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
3481 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
3482 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
3483 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
3484 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
3485 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
3486 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
3487 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
3488 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
3489 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
3490 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
3491 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
3492 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
3493 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
3494 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
3495 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
3496 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
3497 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
3498 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
3499 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
3500 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
3501 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
3502 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
3503 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
3504 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
3507 — Fairfax, 2016-05-21
3511 * The systemd-resolved DNS resolver service has gained a substantial
3512 set of new features, most prominently it may now act as a DNSSEC
3513 validating stub resolver. DNSSEC mode is currently turned off by
3514 default, but is expected to be turned on by default in one of the
3515 next releases. For now, we invite everybody to test the DNSSEC logic
3516 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
3517 service also gained a full set of D-Bus interfaces, including calls
3518 to configure DNS and DNSSEC settings per link (for use by external
3519 network management software). systemd-resolved and systemd-networkd
3520 now distinguish between "search" and "routing" domains. The former
3521 are used to qualify single-label names, the latter are used purely
3522 for routing lookups within certain domains to specific links.
3523 resolved now also synthesizes RRs for all entries from /etc/hosts.
3525 * The systemd-resolve tool (which is a client utility for
3526 systemd-resolved) has been improved considerably and is now fully
3527 supported and documented. Hence it has moved from /usr/lib/systemd to
3530 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
3533 * The coredump collection logic has been reworked: when a coredump is
3534 collected it is now written to disk, compressed and processed
3535 (including stacktrace extraction) from a new instantiated service
3536 systemd-coredump@.service, instead of directly from the
3537 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
3538 processing large coredumps can take up a substantial amount of
3539 resources and time, and this previously happened entirely outside of
3540 systemd's service supervision. With the new logic the core_pattern
3541 hook only does minimal metadata collection before passing off control
3542 to the new instantiated service, which is configured with a time
3543 limit, a nice level and other settings to minimize negative impact on
3544 the rest of the system. Also note that the new logic will honour the
3545 RLIMIT_CORE setting of the crashed process, which now allows users
3546 and processes to turn off coredumping for their processes by setting
3549 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
3550 and all forked processes by default. Previously, PID 1 would leave
3551 the setting at "0" for all processes, as set by the kernel. Note that
3552 the resource limit traditionally has no effect on the generated
3553 coredumps on the system if the /proc/sys/kernel/core_pattern hook
3554 logic is used. Since the limit is now honoured (see above) its
3555 default has been changed so that the coredumping logic is enabled by
3556 default for all processes, while allowing specific opt-out.
3558 * When the stacktrace is extracted from processes of system users, this
3559 is now done as "systemd-coredump" user, in order to sandbox this
3560 potentially security sensitive parsing operation. (Note that when
3561 processing coredumps of normal users this is done under the user ID
3562 of process that crashed, as before.) Packagers should take notice
3563 that it is now necessary to create the "systemd-coredump" system user
3564 and group at package installation time.
3566 * The systemd-activate socket activation testing tool gained support
3567 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
3568 and --seqpacket switches. It also has been extended to support both
3569 new-style and inetd-style file descriptor passing. Use the new
3570 --inetd switch to request inetd-style file descriptor passing.
3572 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
3573 variable, which takes a boolean value. If set to false, ANSI color
3574 output is disabled in the tools even when run on a terminal that
3577 * The VXLAN support in networkd now supports two new settings
3578 DestinationPort= and PortRange=.
3580 * A new systemd.machine_id= kernel command line switch has been added,
3581 that may be used to set the machine ID in /etc/machine-id if it is
3582 not initialized yet. This command line option has no effect if the
3583 file is already initialized.
3585 * systemd-nspawn gained a new --as-pid2 switch that invokes any
3586 specified command line as PID 2 rather than PID 1 in the
3587 container. In this mode PID 1 is a minimal stub init process that
3588 implements the special POSIX and Linux semantics of PID 1 regarding
3589 signal and child process management. Note that this stub init process
3590 is implemented in nspawn itself and requires no support from the
3591 container image. This new logic is useful to support running
3592 arbitrary commands in the container, as normal processes are
3593 generally not prepared to run as PID 1.
3595 * systemd-nspawn gained a new --chdir= switch for setting the current
3596 working directory for the process started in the container.
3598 * "journalctl /dev/sda" will now output all kernel log messages for
3599 specified device from the current boot, in addition to all devices
3600 that are parents of it. This should make log output about devices
3601 pretty useful, as long as kernel drivers attach enough metadata to
3602 the log messages. (The usual SATA drivers do.)
3604 * The sd-journal API gained two new calls
3605 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
3606 that report whether log data from /run or /var has been found.
3608 * journalctl gained a new switch "--fields" that prints all journal
3609 record field names currently in use in the journal. This is backed
3610 by two new sd-journal API calls sd_journal_enumerate_fields() and
3611 sd_journal_restart_fields().
3613 * Most configurable timeouts in systemd now expect an argument of
3614 "infinity" to turn them off, instead of "0" as before. The semantics
3615 from now on is that a timeout of "0" means "now", and "infinity"
3616 means "never". To maintain backwards compatibility, "0" continues to
3617 turn off previously existing timeout settings.
3619 * "systemctl reload-or-try-restart" has been renamed to "systemctl
3620 try-reload-or-restart" to clarify what it actually does: the "try"
3621 logic applies to both reloading and restarting, not just restarting.
3622 The old name continues to be accepted for compatibility.
3624 * On boot-up, when PID 1 detects that the system clock is behind the
3625 release date of the systemd version in use, the clock is now set
3626 to the latter. Previously, this was already done in timesyncd, in order
3627 to avoid running with clocks set to the various clock epochs such as
3628 1902, 1938 or 1970. With this change the logic is now done in PID 1
3629 in addition to timesyncd during early boot-up, so that it is enforced
3630 before the first process is spawned by systemd. Note that the logic
3631 in timesyncd remains, as it is more comprehensive and ensures
3632 clock monotonicity by maintaining a persistent timestamp file in
3633 /var. Since /var is generally not available in earliest boot or the
3634 initrd, this part of the logic remains in timesyncd, and is not done
3637 * Support for tweaking details in net_cls.class_id through the
3638 NetClass= configuration directive has been removed, as the kernel
3639 people have decided to deprecate that controller in cgroup v2.
3640 Userspace tools such as nftables are moving over to setting rules
3641 that are specific to the full cgroup path of a task, which obsoletes
3642 these controllers anyway. The NetClass= directive is kept around for
3643 legacy compatibility reasons. For a more in-depth description of the
3644 kernel change, please refer to the respective upstream commit:
3646 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
3648 * A new service setting RuntimeMaxSec= has been added that may be used
3649 to specify a maximum runtime for a service. If the timeout is hit, the
3650 service is terminated and put into a failure state.
3652 * A new service setting AmbientCapabilities= has been added. It allows
3653 configuration of additional Linux process capabilities that are
3654 passed to the activated processes. This is only available on very
3657 * The process resource limit settings in service units may now be used
3658 to configure hard and soft limits individually.
3660 * The various libsystemd APIs such as sd-bus or sd-event now publicly
3661 expose support for gcc's __attribute__((cleanup())) C extension.
3662 Specifically, for many object destructor functions alternative
3663 versions have been added that have names suffixed with "p" and take a
3664 pointer to a pointer to the object to destroy, instead of just a
3665 pointer to the object itself. This is useful because these destructor
3666 functions may be used directly as parameters to the cleanup
3667 construct. Internally, systemd has been a heavy user of this GCC
3668 extension for a long time, and with this change similar support is
3669 now available to consumers of the library outside of systemd. Note
3670 that by using this extension in your sources compatibility with old
3671 and strictly ANSI compatible C compilers is lost. However, all gcc or
3672 LLVM versions of recent years support this extension.
3674 * Timer units gained support for a new setting RandomizedDelaySec= that
3675 allows configuring some additional randomized delay to the configured
3676 time. This is useful to spread out timer events to avoid load peaks in
3677 clusters or larger setups.
3679 * Calendar time specifications now support sub-second accuracy.
3681 * Socket units now support listening on SCTP and UDP-lite protocol
3684 * The sd-event API now comes with a full set of man pages.
3686 * Older versions of systemd contained experimental support for
3687 compressing journal files and coredumps with the LZ4 compressor that
3688 was not compatible with the lz4 binary (due to API limitations of the
3689 lz4 library). This support has been removed; only support for files
3690 compatible with the lz4 binary remains. This LZ4 logic is now
3691 officially supported and no longer considered experimental.
3693 * The dkr image import logic has been removed again from importd. dkr's
3694 micro-services focus doesn't fit into the machine image focus of
3695 importd, and quickly got out of date with the upstream dkr API.
3697 * Creation of the /run/lock/lockdev/ directory was dropped from
3698 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
3699 been available for many years. If you still need this, you need to
3700 create your own tmpfiles.d config file with:
3702 d /run/lock/lockdev 0775 root lock -
3704 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
3705 and RebootArgument= have been moved from the [Service] section of
3706 unit files to [Unit], and they are now supported on all unit types,
3707 not just service units. Of course, systemd will continue to
3708 understand these settings also at the old location, in order to
3709 maintain compatibility.
3711 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
3712 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
3713 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
3714 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
3715 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
3716 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
3717 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
3718 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
3719 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
3720 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
3721 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
3722 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
3723 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
3724 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
3725 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
3726 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
3727 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
3728 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
3729 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
3731 — Berlin, 2016-02-11
3735 * A number of properties previously only settable in unit
3736 files are now also available as properties to set when
3737 creating transient units programmatically via the bus, as it
3738 is exposed with systemd-run's --property=
3739 setting. Specifically, these are: SyslogIdentifier=,
3740 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
3741 EnvironmentFile=, ReadWriteDirectories=,
3742 ReadOnlyDirectories=, InaccessibleDirectories=,
3743 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
3745 * When creating transient services via the bus API it is now
3746 possible to pass in a set of file descriptors to use as
3747 STDIN/STDOUT/STDERR for the invoked process.
3749 * Slice units may now be created transiently via the bus APIs,
3750 similar to the way service and scope units may already be
3751 created transiently.
3753 * Wherever systemd expects a calendar timestamp specification
3754 (like in journalctl's --since= and --until= switches) UTC
3755 timestamps are now supported. Timestamps suffixed with "UTC"
3756 are now considered to be in Universal Time Coordinated
3757 instead of the local timezone. Also, timestamps may now
3758 optionally be specified with sub-second accuracy. Both of
3759 these additions also apply to recurring calendar event
3760 specification, such as OnCalendar= in timer units.
3762 * journalctl gained a new "--sync" switch that asks the
3763 journal daemon to write all so far unwritten log messages to
3764 disk and sync the files, before returning.
3766 * systemd-tmpfiles learned two new line types "q" and "Q" that
3767 operate like "v", but also set up a basic btrfs quota
3768 hierarchy when used on a btrfs file system with quota
3771 * tmpfiles' "v", "q" and "Q" will now create a plain directory
3772 instead of a subvolume (even on a btrfs file system) if the
3773 root directory is a plain directory, and not a
3774 subvolume. This should simplify things with certain chroot()
3775 environments which are not aware of the concept of btrfs
3778 * systemd-detect-virt gained a new --chroot switch to detect
3779 whether execution takes place in a chroot() environment.
3781 * CPUAffinity= now takes CPU index ranges in addition to
3784 * The various memory-related resource limit settings (such as
3785 LimitAS=) now understand the usual K, M, G, ... suffixes to
3786 the base of 1024 (IEC). Similar, the time-related resource
3787 limit settings understand the usual min, h, day, ...
3790 * There's a new system.conf setting DefaultTasksMax= to
3791 control the default TasksMax= setting for services and
3792 scopes running on the system. (TasksMax= is the primary
3793 setting that exposes the "pids" cgroup controller on systemd
3794 and was introduced in the previous systemd release.) The
3795 setting now defaults to 512, which means services that are
3796 not explicitly configured otherwise will only be able to
3797 create 512 processes or threads at maximum, from this
3798 version on. Note that this means that thread- or
3799 process-heavy services might need to be reconfigured to set
3800 TasksMax= to a higher value. It is sufficient to set
3801 TasksMax= in these specific unit files to a higher value, or
3802 even "infinity". Similar, there's now a logind.conf setting
3803 UserTasksMax= that defaults to 4096 and limits the total
3804 number of processes or tasks each user may own
3805 concurrently. nspawn containers also have the TasksMax=
3806 value set by default now, to 8192. Note that all of this
3807 only has an effect if the "pids" cgroup controller is
3808 enabled in the kernel. The general benefit of these changes
3809 should be a more robust and safer system, that provides a
3810 certain amount of per-service fork() bomb protection.
3812 * systemd-nspawn gained the new --network-veth-extra= switch
3813 to define additional and arbitrarily-named virtual Ethernet
3814 links between the host and the container.
3816 * A new service execution setting PassEnvironment= has been
3817 added that allows importing select environment variables
3818 from PID1's environment block into the environment block of
3821 * Timer units gained support for a new RemainAfterElapse=
3822 setting which takes a boolean argument. It defaults to on,
3823 exposing behaviour unchanged to previous releases. If set to
3824 off, timer units are unloaded after they elapsed if they
3825 cannot elapse again. This is particularly useful for
3826 transient timer units, which shall not stay around longer
3827 than until they first elapse.
3829 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
3830 default now (the kernel default is 16). This is beneficial
3831 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
3832 allows substantially larger numbers of queued
3833 datagrams. This should increase the capability of systemd to
3834 parallelize boot-up, as logging and sd_notify() are unlikely
3835 to stall execution anymore. If you need to change the value
3836 from the new defaults, use the usual sysctl.d/ snippets.
3838 * The compression framing format used by the journal or
3839 coredump processing has changed to be in line with what the
3840 official LZ4 tools generate. LZ4 compression support in
3841 systemd was considered unsupported previously, as the format
3842 was not compatible with the normal tools. With this release
3843 this has changed now, and it is hence safe for downstream
3844 distributions to turn it on. While not compressing as well
3845 as the XZ, LZ4 is substantially faster, which makes
3846 it a good default choice for the compression logic in the
3847 journal and in coredump handling.
3849 * Any reference to /etc/mtab has been dropped from
3850 systemd. The file has been obsolete since a while, but
3851 systemd refused to work on systems where it was incorrectly
3852 set up (it should be a symlink or non-existent). Please make
3853 sure to update to util-linux 2.27.1 or newer in conjunction
3854 with this systemd release, which also drops any reference to
3855 /etc/mtab. If you maintain a distribution make sure that no
3856 software you package still references it, as this is a
3857 likely source of bugs. There's also a glibc bug pending,
3858 asking for removal of any reference to this obsolete file:
3860 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
3862 Note that only util-linux versions built with
3863 --enable-libmount-force-mountinfo are supported.
3865 * Support for the ".snapshot" unit type has been removed. This
3866 feature turned out to be little useful and little used, and
3867 has now been removed from the core and from systemctl.
3869 * The dependency types RequiresOverridable= and
3870 RequisiteOverridable= have been removed from systemd. They
3871 have been used only very sparingly to our knowledge and
3872 other options that provide a similar effect (such as
3873 systemctl --mode=ignore-dependencies) are much more useful
3874 and commonly used. Moreover, they were only half-way
3875 implemented as the option to control behaviour regarding
3876 these dependencies was never added to systemctl. By removing
3877 these dependency types the execution engine becomes a bit
3878 simpler. Unit files that use these dependencies should be
3879 changed to use the non-Overridable dependency types
3880 instead. In fact, when parsing unit files with these
3881 options, that's what systemd will automatically convert them
3882 too, but it will also warn, asking users to fix the unit
3883 files accordingly. Removal of these dependency types should
3884 only affect a negligible number of unit files in the wild.
3886 * Behaviour of networkd's IPForward= option changed
3887 (again). It will no longer maintain a per-interface setting,
3888 but propagate one way from interfaces where this is enabled
3889 to the global kernel setting. The global setting will be
3890 enabled when requested by a network that is set up, but
3891 never be disabled again. This change was made to make sure
3892 IPv4 and IPv6 behaviour regarding packet forwarding is
3893 similar (as the Linux IPv6 stack does not support
3894 per-interface control of this setting) and to minimize
3897 * In unit files the behaviour of %u, %U, %h, %s has
3898 changed. These specifiers will now unconditionally resolve
3899 to the various user database fields of the user that the
3900 systemd instance is running as, instead of the user
3901 configured in the specific unit via User=. Note that this
3902 effectively doesn't change much, as resolving of these
3903 specifiers was already turned off in the --system instance
3904 of systemd, as we cannot do NSS lookups from PID 1. In the
3905 --user instance of systemd these specifiers where correctly
3906 resolved, but hardly made any sense, since the user instance
3907 lacks privileges to do user switches anyway, and User= is
3908 hence useless. Moreover, even in the --user instance of
3909 systemd behaviour was awkward as it would only take settings
3910 from User= assignment placed before the specifier into
3911 account. In order to unify and simplify the logic around
3912 this the specifiers will now always resolve to the
3913 credentials of the user invoking the manager (which in case
3914 of PID 1 is the root user).
3916 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
3917 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
3918 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
3919 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
3920 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
3921 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
3922 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
3923 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
3924 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
3925 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
3926 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
3927 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
3928 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
3929 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
3932 — Berlin, 2015-11-18
3936 * systemd now depends on util-linux v2.27. More specifically,
3937 the newly added mount monitor feature in libmount now
3938 replaces systemd's former own implementation.
3940 * libmount mandates /etc/mtab not to be regular file, and
3941 systemd now enforces this condition at early boot.
3942 /etc/mtab has been deprecated and warned about for a very
3943 long time, so systems running systemd should already have
3944 stopped having this file around as anything else than a
3945 symlink to /proc/self/mounts.
3947 * Support for the "pids" cgroup controller has been added. It
3948 allows accounting the number of tasks in a cgroup and
3949 enforcing limits on it. This adds two new setting
3950 TasksAccounting= and TasksMax= to each unit, as well as a
3951 global option DefaultTasksAccounting=.
3953 * Support for the "net_cls" cgroup controller has been added.
3954 It allows assigning a net class ID to each task in the
3955 cgroup, which can then be used in firewall rules and traffic
3956 shaping configurations. Note that the kernel netfilter net
3957 class code does not currently work reliably for ingress
3958 packets on unestablished sockets.
3960 This adds a new config directive called NetClass= to CGroup
3961 enabled units. Allowed values are positive numbers for fixed
3962 assignments and "auto" for picking a free value
3965 * 'systemctl is-system-running' now returns 'offline' if the
3966 system is not booted with systemd. This command can now be
3967 used as a substitute for 'systemd-notify --booted'.
3969 * Watchdog timeouts have been increased to 3 minutes for all
3970 in-tree service files. Apparently, disk IO issues are more
3971 frequent than we hoped, and user reported >1 minute waiting
3974 * 'machine-id-commit' functionality has been merged into
3975 'machine-id-setup --commit'. The separate binary has been
3978 * The WorkingDirectory= directive in unit files may now be set
3979 to the special value '~'. In this case, the working
3980 directory is set to the home directory of the user
3981 configured in User=.
3983 * "machinectl shell" will now open the shell in the home
3984 directory of the selected user by default.
3986 * The CrashChVT= configuration file setting is renamed to
3987 CrashChangeVT=, following our usual logic of not
3988 abbreviating unnecessarily. The old directive is still
3989 supported for compat reasons. Also, this directive now takes
3990 an integer value between 1 and 63, or a boolean value. The
3991 formerly supported '-1' value for disabling stays around for
3994 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
3995 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
3996 RootDirectory= properties can now be set for transient
3999 * The systemd-analyze tool gained a new "set-log-target" verb
4000 to change the logging target the system manager logs to
4001 dynamically during runtime. This is similar to how
4002 "systemd-analyze set-log-level" already changes the log
4005 * In nspawn /sys is now mounted as tmpfs, with only a selected
4006 set of subdirectories mounted in from the real sysfs. This
4007 enhances security slightly, and is useful for ensuring user
4008 namespaces work correctly.
4010 * Support for USB FunctionFS activation has been added. This
4011 allows implementation of USB gadget services that are
4012 activated as soon as they are requested, so that they don't
4013 have to run continuously, similar to classic socket
4016 * The "systemctl exit" command now optionally takes an
4017 additional parameter that sets the exit code to return from
4018 the systemd manager when exiting. This is only relevant when
4019 running the systemd user instance, or when running the
4020 system instance in a container.
4022 * sd-bus gained the new API calls sd_bus_path_encode_many()
4023 and sd_bus_path_decode_many() that allow easy encoding and
4024 decoding of multiple identifier strings inside a D-Bus
4025 object path. Another new call sd_bus_default_flush_close()
4026 has been added to flush and close per-thread default
4029 * systemd-cgtop gained support for a -M/--machine= switch to
4030 show the control groups within a certain container only.
4032 * "systemctl kill" gained support for an optional --fail
4033 switch. If specified the requested operation will fail of no
4034 processes have been killed, because the unit had no
4035 processes attached, or similar.
4037 * A new systemd.crash_reboot=1 kernel command line option has
4038 been added that triggers a reboot after crashing. This can
4039 also be set through CrashReboot= in systemd.conf.
4041 * The RuntimeDirectory= setting now understands unit
4042 specifiers like %i or %f.
4044 * A new (still internal) library API sd-ipv4acd has been added,
4045 that implements address conflict detection for IPv4. It's
4046 based on code from sd-ipv4ll, and will be useful for
4047 detecting DHCP address conflicts.
4049 * File descriptors passed during socket activation may now be
4050 named. A new API sd_listen_fds_with_names() is added to
4051 access the names. The default names may be overridden,
4052 either in the .socket file using the FileDescriptorName=
4053 parameter, or by passing FDNAME= when storing the file
4054 descriptors using sd_notify().
4056 * systemd-networkd gained support for:
4058 - Setting the IPv6 Router Advertisement settings via
4059 IPv6AcceptRouterAdvertisements= in .network files.
4061 - Configuring the HelloTimeSec=, MaxAgeSec= and
4062 ForwardDelaySec= bridge parameters in .netdev files.
4064 - Configuring PreferredSource= for static routes in
4067 * The "ask-password" framework used to query for LUKS harddisk
4068 passwords or SSL passwords during boot gained support for
4069 caching passwords in the kernel keyring, if it is
4070 available. This makes sure that the user only has to type in
4071 a passphrase once if there are multiple objects to unlock
4072 with the same one. Previously, such password caching was
4073 available only when Plymouth was used; this moves the
4074 caching logic into the systemd codebase itself. The
4075 "systemd-ask-password" utility gained a new --keyname=
4076 switch to control which kernel keyring key to use for
4077 caching a password in. This functionality is also useful for
4078 enabling display managers such as gdm to automatically
4079 unlock the user's GNOME keyring if its passphrase, the
4080 user's password and the harddisk password are the same, if
4081 gdm-autologin is used.
4083 * When downloading tar or raw images using "machinectl
4084 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
4085 file is now also downloaded, if it is available and stored
4086 next to the image file.
4088 * Units of type ".socket" gained a new boolean setting
4089 Writable= which is only useful in conjunction with
4090 ListenSpecial=. If true, enables opening the specified
4091 special file in O_RDWR mode rather than O_RDONLY mode.
4093 * systemd-rfkill has been reworked to become a singleton
4094 service that is activated through /dev/rfkill on each rfkill
4095 state change and saves the settings to disk. This way,
4096 systemd-rfkill is now compatible with devices that exist
4097 only intermittendly, and even restores state if the previous
4098 system shutdown was abrupt rather than clean.
4100 * The journal daemon gained support for vacuuming old journal
4101 files controlled by the number of files that shall remain,
4102 in addition to the already existing control by size and by
4103 date. This is useful as journal interleaving performance
4104 degrades with too many separate journal files, and allows
4105 putting an effective limit on them. The new setting defaults
4106 to 100, but this may be changed by setting SystemMaxFiles=
4107 and RuntimeMaxFiles= in journald.conf. Also, the
4108 "journalctl" tool gained the new --vacuum-files= switch to
4109 manually vacuum journal files to leave only the specified
4110 number of files in place.
4112 * udev will now create /dev/disk/by-path links for ATA devices
4113 on kernels where that is supported.
4115 * Galician, Serbian, Turkish and Korean translations were added.
4117 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
4118 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
4119 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
4120 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
4121 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
4122 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
4123 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
4124 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
4125 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
4126 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
4127 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
4128 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
4129 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
4130 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
4131 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
4132 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
4133 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
4134 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
4136 — Berlin, 2015-10-07
4140 * The DHCP implementation of systemd-networkd gained a set of
4143 - The DHCP server now supports emitting DNS and NTP
4144 information. It may be enabled and configured via
4145 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
4146 and NTP information is enabled, but no servers are
4147 configured, the corresponding uplink information (if there
4148 is any) is propagated.
4150 - Server and client now support transmission and reception
4151 of timezone information. It can be configured via the
4152 newly introduced network options UseTimezone=,
4153 EmitTimezone=, and Timezone=. Transmission of timezone
4154 information is enabled between host and containers by
4155 default now: the container will change its local timezone
4156 to what the host has set.
4158 - Lease timeouts can now be configured via
4159 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
4161 - The DHCP server improved on the stability of
4162 leases. Clients are more likely to get the same lease
4163 information back, even if the server loses state.
4165 - The DHCP server supports two new configuration options to
4166 control the lease address pool metrics, PoolOffset= and
4169 * The encapsulation limit of tunnels in systemd-networkd may
4170 now be configured via 'EncapsulationLimit='. It allows
4171 modifying the maximum additional levels of encapsulation
4172 that are permitted to be prepended to a packet.
4174 * systemd now supports the concept of user buses replacing
4175 session buses, if used with dbus-1.10 (and enabled via dbus
4176 --enable-user-session). It previously only supported this on
4177 kdbus-enabled systems, and this release expands this to
4178 'dbus-daemon' systems.
4180 * systemd-networkd now supports predictable interface names
4183 * systemd now optionally supports the new Linux kernel
4184 "unified" control group hierarchy. If enabled via the kernel
4185 command-line option 'systemd.unified_cgroup_hierarchy=1',
4186 systemd will try to mount the unified cgroup hierarchy
4187 directly on /sys/fs/cgroup. If not enabled, or not
4188 available, systemd will fall back to the legacy cgroup
4189 hierarchy setup, as before. Host system and containers can
4190 mix and match legacy and unified hierarchies as they
4191 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
4192 environment variable to individually select the hierarchy to
4193 use for executed containers. By default, nspawn will use the
4194 unified hierarchy for the containers if the host uses the
4195 unified hierarchy, and the legacy hierarchy otherwise.
4196 Please note that at this point the unified hierarchy is an
4197 experimental kernel feature and is likely to change in one
4198 of the next kernel releases. Therefore, it should not be
4199 enabled by default in downstream distributions yet. The
4200 minimum required kernel version for the unified hierarchy to
4201 work is 4.2. Note that when the unified hierarchy is used
4202 for the first time delegated access to controllers is
4203 safe. Because of this systemd-nspawn containers will get
4204 access to controllers now, as will systemd user
4205 sessions. This means containers and user sessions may now
4206 manage their own resources, partitioning up what the system
4209 * A new special scope unit "init.scope" has been introduced
4210 that encapsulates PID 1 of the system. It may be used to
4211 determine resource usage and enforce resource limits on PID
4212 1 itself. PID 1 hence moved out of the root of the control
4215 * The cgtop tool gained support for filtering out kernel
4216 threads when counting tasks in a control group. Also, the
4217 count of processes is now recursively summed up by
4218 default. Two options -k and --recursive= have been added to
4219 revert to old behaviour. The tool has also been updated to
4220 work correctly in containers now.
4222 * systemd-nspawn's --bind= and --bind-ro= options have been
4223 extended to allow creation of non-recursive bind mounts.
4225 * libsystemd gained two new calls sd_pid_get_cgroup() and
4226 sd_peer_get_cgroup() which return the control group path of
4227 a process or peer of a connected AF_UNIX socket. This
4228 function call is particularly useful when implementing
4229 delegated subtrees support in the control group hierarchy.
4231 * The "sd-event" event loop API of libsystemd now supports
4232 correct dequeuing of real-time signals, without losing
4235 * When systemd requests a polkit decision when managing units it
4236 will now add additional fields to the request, including unit
4237 name and desired operation. This enables more powerful polkit
4238 policies, that make decisions depending on these parameters.
4240 * nspawn learnt support for .nspawn settings files, that may
4241 accompany the image files or directories of containers, and
4242 may contain additional settings for the container. This is
4243 an alternative to configuring container parameters via the
4244 nspawn command line.
4246 Contributions from: Cristian Rodríguez, Daniel Mack, David
4247 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
4248 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
4249 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
4250 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
4251 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
4252 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
4253 Andersen, Tom Gundersen, Torstein Husebø
4255 — Berlin, 2015-09-08
4259 * machinectl gained a new verb 'shell' which opens a fresh
4260 shell on the target container or the host. It is similar to
4261 the existing 'login' command of machinectl, but spawns the
4262 shell directly without prompting for username or
4263 password. The pseudo machine '.host' now refers to the local
4264 host and is used by default. Hence, 'machinectl shell' can
4265 be used as replacement for 'su -' which spawns a session as
4266 a fresh systemd unit in a way that is fully isolated from
4267 the originating session.
4269 * systemd-networkd learned to cope with private-zone DHCP
4270 options and allows other programs to query the values.
4272 * SELinux access control when enabling/disabling units is no
4273 longer enforced with this release. The previous implementation
4274 was incorrect, and a new corrected implementation is not yet
4275 available. As unit file operations are still protected via
4276 polkit and D-Bus policy this is not a security problem. Yet,
4277 distributions which care about optimal SELinux support should
4278 probably not stabilize on this release.
4280 * sd-bus gained support for matches of type "arg0has=", that
4281 test for membership of strings in string arrays sent in bus
4284 * systemd-resolved now dumps the contents of its DNS and LLMNR
4285 caches to the logs on reception of the SIGUSR1 signal. This
4286 is useful to debug DNS behaviour.
4288 * The coredumpctl tool gained a new --directory= option to
4289 operate on journal files in a specific directory.
4291 * "systemctl reboot" and related commands gained a new
4292 "--message=" option which may be used to set a free-text
4293 wall message when shutting down or rebooting the
4294 system. This message is also logged, which is useful for
4295 figuring out the reason for a reboot or shutdown a
4298 * The "systemd-resolve-host" tool's -i switch now takes
4299 network interface numbers as alternative to interface names.
4301 * A new unit file setting for services has been introduced:
4302 UtmpMode= allows configuration of how precisely systemd
4303 handles utmp and wtmp entries for the service if this is
4304 enabled. This allows writing services that appear similar to
4305 user sessions in the output of the "w", "who", "last" and
4308 * systemd-resolved will now locally synthesize DNS resource
4309 records for the "localhost" and "gateway" domains as well as
4310 the local hostname. This should ensure that clients querying
4311 RRs via resolved will get similar results as those going via
4312 NSS, if nss-myhostname is enabled.
4314 Contributions from: Alastair Hughes, Alex Crawford, Daniel
4315 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
4316 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
4317 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
4318 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
4319 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
4320 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
4321 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
4322 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
4323 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
4324 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
4325 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4327 — Berlin, 2015-08-27
4331 * The systemd-efi-boot-generator functionality was merged into
4332 systemd-gpt-auto-generator.
4334 * systemd-networkd now supports Group Policy for vxlan
4335 devices. It can be enabled via the new boolean configuration
4336 option called 'GroupPolicyExtension='.
4338 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
4339 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
4340 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
4342 — Berlin, 2015-07-31
4346 * The python-systemd code has been removed from the systemd repository.
4347 A new repository has been created which accommodates the code from
4348 now on, and we kindly ask distributions to create a separate package
4349 for this: https://github.com/systemd/python-systemd
4351 * The systemd daemon will now reload its main configuration
4352 (/etc/systemd/system.conf) on daemon-reload.
4354 * sd-dhcp now exposes vendor specific extensions via
4355 sd_dhcp_lease_get_vendor_specific().
4357 * systemd-networkd gained a number of new configuration options.
4359 - A new boolean configuration option for TAP devices called
4360 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
4361 device, thus allowing to send and receive GSO packets.
4363 - A new tunnel configuration option called 'CopyDSCP='.
4364 If enabled, the DSCP field of ip6 tunnels is copied into the
4365 decapsulated packet.
4367 - A set of boolean bridge configuration options were added.
4368 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
4369 and 'UnicastFlood=' are now parsed by networkd and applied to the
4370 respective bridge link device via the respective IFLA_BRPORT_*
4373 - A new string configuration option to override the hostname sent
4374 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
4375 is true, networkd will use the configured hostname instead of the
4376 system hostname when sending DHCP requests.
4378 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
4379 networkd will configure the IPv6 flow-label of the tunnel device
4380 according to RFC2460.
4382 - The 'macvtap' virtual network devices are now supported, similar to
4383 the already supported 'macvlan' devices.
4385 * systemd-resolved now implements RFC5452 to improve resilience against
4386 cache poisoning. Additionally, source port randomization is enabled
4387 by default to further protect against DNS spoofing attacks.
4389 * nss-mymachines now supports translating UIDs and GIDs of running
4390 containers with user-namespaces enabled. If a container 'foo'
4391 translates a host uid 'UID' to the container uid 'TUID', then
4392 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
4393 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
4394 mapped as 'vg-foo-TGID'.
4396 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
4397 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
4398 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
4399 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
4400 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
4401 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
4402 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
4403 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
4404 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
4405 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
4407 — Berlin, 2015-07-29
4411 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
4412 There are no known issues with current sysfs, and udev does not need
4413 or should be used to work around such bugs.
4415 * udev does no longer enable USB HID power management. Several reports
4416 indicate, that some devices cannot handle that setting.
4418 * The udev accelerometer helper was removed. The functionality
4419 is now fully included in iio-sensor-proxy. But this means,
4420 older iio-sensor-proxy versions will no longer provide
4421 accelerometer/orientation data with this systemd version.
4422 Please upgrade iio-sensor-proxy to version 1.0.
4424 * networkd gained a new configuration option IPv6PrivacyExtensions=
4425 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
4426 for Stateless Address") on selected networks.
4428 * For the sake of fewer build-time dependencies and less code in the
4429 main repository, the python bindings are about to be removed in the
4430 next release. A new repository has been created which accommodates
4431 the code from now on, and we kindly ask distributions to create a
4432 separate package for this. The removal will take place in v223.
4434 https://github.com/systemd/python-systemd
4436 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
4437 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
4438 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
4439 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
4440 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
4441 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
4442 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
4443 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
4444 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
4445 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4447 — Berlin, 2015-07-07
4451 * The sd-bus.h and sd-event.h APIs have now been declared
4452 stable and have been added to the official interface of
4453 libsystemd.so. sd-bus implements an alternative D-Bus client
4454 library, that is relatively easy to use, very efficient and
4455 supports both classic D-Bus as well as kdbus as transport
4456 backend. sd-event is a generic event loop abstraction that
4457 is built around Linux epoll, but adds features such as event
4458 prioritization or efficient timer handling. Both APIs are good
4459 choices for C programs looking for a bus and/or event loop
4460 implementation that is minimal and does not have to be
4461 portable to other kernels.
4463 * kdbus support is no longer compile-time optional. It is now
4464 always built-in. However, it can still be disabled at
4465 runtime using the kdbus=0 kernel command line setting, and
4466 that setting may be changed to default to off, by specifying
4467 --disable-kdbus at build-time. Note though that the kernel
4468 command line setting has no effect if the kdbus.ko kernel
4469 module is not installed, in which case kdbus is (obviously)
4470 also disabled. We encourage all downstream distributions to
4471 begin testing kdbus by adding it to the kernel images in the
4472 development distributions, and leaving kdbus support in
4475 * The minimal required util-linux version has been bumped to
4478 * Support for chkconfig (--enable-chkconfig) was removed in
4479 favor of calling an abstraction tool
4480 /lib/systemd/systemd-sysv-install. This needs to be
4481 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
4482 in README for details.
4484 * If there's a systemd unit and a SysV init script for the
4485 same service name, and the user executes "systemctl enable"
4486 for it (or a related call), then this will now enable both
4487 (or execute the related operation on both), not just the
4490 * The libudev API documentation has been converted from gtkdoc
4493 * gudev has been removed from the systemd tree, it is now an
4496 * The systemd-cgtop tool learnt a new --raw switch to generate
4497 "raw" (machine parsable) output.
4499 * networkd's IPForwarding= .network file setting learnt the
4500 new setting "kernel", which ensures that networkd does not
4501 change the IP forwarding sysctl from the default kernel
4504 * The systemd-logind bus API now exposes a new boolean
4505 property "Docked" that reports whether logind considers the
4506 system "docked", i.e. connected to a docking station or not.
4508 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
4509 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
4510 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
4511 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
4512 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
4513 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
4514 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
4515 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
4516 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
4517 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
4518 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
4519 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
4520 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
4521 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
4522 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
4523 Fink, Zbigniew Jędrzejewski-Szmek
4525 — Berlin, 2015-06-19
4529 * The gudev library has been extracted into a separate repository
4530 available at: https://git.gnome.org/browse/libgudev/
4531 It is now managed as part of the Gnome project. Distributions
4532 are recommended to pass --disable-gudev to systemd and use
4533 gudev from the Gnome project instead. gudev is still included
4534 in systemd, for now. It will be removed soon, though. Please
4535 also see the announcement-thread on systemd-devel:
4536 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
4538 * systemd now exposes a CPUUsageNSec= property for each
4539 service unit on the bus, that contains the overall consumed
4540 CPU time of a service (the sum of what each process of the
4541 service consumed). This value is only available if
4542 CPUAccounting= is turned on for a service, and is then shown
4543 in the "systemctl status" output.
4545 * Support for configuring alternative mappings of the old SysV
4546 runlevels to systemd targets has been removed. They are now
4547 hardcoded in a way that runlevels 2, 3, 4 all map to
4548 multi-user.target and 5 to graphical.target (which
4549 previously was already the default behaviour).
4551 * The auto-mounter logic gained support for mount point
4552 expiry, using a new TimeoutIdleSec= setting in .automount
4553 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
4555 * The EFI System Partition (ESP) as mounted to /boot by
4556 systemd-efi-boot-generator will now be unmounted
4557 automatically after 2 minutes of not being used. This should
4558 minimize the risk of ESP corruptions.
4560 * New /etc/fstab options x-systemd.requires= and
4561 x-systemd.requires-mounts-for= are now supported to express
4562 additional dependencies for mounts. This is useful for
4563 journalling file systems that support external journal
4564 devices or overlay file systems that require underlying file
4565 systems to be mounted.
4567 * systemd does not support direct live-upgrades (via systemctl
4568 daemon-reexec) from versions older than v44 anymore. As no
4569 distribution we are aware of shipped such old versions in a
4570 stable release this should not be problematic.
4572 * When systemd forks off a new per-connection service instance
4573 it will now set the $REMOTE_ADDR environment variable to the
4574 remote IP address, and $REMOTE_PORT environment variable to
4575 the remote IP port. This behaviour is similar to the
4576 corresponding environment variables defined by CGI.
4578 * systemd-networkd gained support for uplink failure
4579 detection. The BindCarrier= option allows binding interface
4580 configuration dynamically to the link sense of other
4581 interfaces. This is useful to achieve behaviour like in
4584 * systemd-networkd gained support for configuring the DHCP
4585 client identifier to use when requesting leases.
4587 * systemd-networkd now has a per-network UseNTP= option to
4588 configure whether NTP server information acquired via DHCP
4589 is passed on to services like systemd-timesyncd.
4591 * systemd-networkd gained support for vti6 tunnels.
4593 * Note that systemd-networkd manages the sysctl variable
4594 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
4595 it is configured for since v219. The variable controls IP
4596 forwarding, and is a per-interface alternative to the global
4597 /proc/sys/net/ipv[46]/ip_forward. This setting is
4598 configurable in the IPForward= option, which defaults to
4599 "no". This means if networkd is used for an interface it is
4600 no longer sufficient to set the global sysctl option to turn
4601 on IP forwarding! Instead, the .network file option
4602 IPForward= needs to be turned on! Note that the
4603 implementation of this behaviour was broken in v219 and has
4606 * Many bonding and vxlan options are now configurable in
4609 * systemd-nspawn gained a new --property= setting to set unit
4610 properties for the container scope. This is useful for
4611 setting resource parameters (e.g. "CPUShares=500") on
4612 containers started from the command line.
4614 * systemd-nspawn gained a new --private-users= switch to make
4615 use of user namespacing available on recent Linux kernels.
4617 * systemd-nspawn may now be called as part of a shell pipeline
4618 in which case the pipes used for stdin and stdout are passed
4619 directly to the process invoked in the container, without
4620 indirection via a pseudo tty.
4622 * systemd-nspawn gained a new switch to control the UNIX
4623 signal to use when killing the init process of the container
4626 * systemd-nspawn gained a new --overlay= switch for mounting
4627 overlay file systems into the container using the new kernel
4630 * When a container image is imported via systemd-importd and
4631 the host file system is not btrfs, a loopback block device
4632 file is created in /var/lib/machines.raw with a btrfs file
4633 system inside. It is then mounted to /var/lib/machines to
4634 enable btrfs features for container management. The loopback
4635 file and btrfs file system is grown as needed when container
4636 images are imported via systemd-importd.
4638 * systemd-machined/systemd-importd gained support for btrfs
4639 quota, to enforce container disk space limits on disk. This
4640 is exposed in "machinectl set-limit".
4642 * systemd-importd now can import containers from local .tar,
4643 .raw and .qcow2 images, and export them to .tar and .raw. It
4644 can also import dkr v2 images now from the network (on top
4647 * systemd-importd gained support for verifying downloaded
4648 images with gpg2 (previously only gpg1 was supported).
4650 * systemd-machined, systemd-logind, systemd: most bus calls are
4651 now accessible to unprivileged processes via polkit. Also,
4652 systemd-logind will now allow users to kill their own sessions
4653 without further privileges or authorization.
4655 * systemd-shutdownd has been removed. This service was
4656 previously responsible for implementing scheduled shutdowns
4657 as exposed in /usr/bin/shutdown's time parameter. This
4658 functionality has now been moved into systemd-logind and is
4659 accessible via a bus interface.
4661 * "systemctl reboot" gained a new switch --firmware-setup that
4662 can be used to reboot into the EFI firmware setup, if that
4663 is available. systemd-logind now exposes an API on the bus
4664 to trigger such reboots, in case graphical desktop UIs want
4665 to cover this functionality.
4667 * "systemctl enable", "systemctl disable" and "systemctl mask"
4668 now support a new "--now" switch. If specified the units
4669 that are enabled will also be started, and the ones
4670 disabled/masked also stopped.
4672 * The Gummiboot EFI boot loader tool has been merged into
4673 systemd, and renamed to "systemd-boot". The bootctl tool has been
4674 updated to support systemd-boot.
4676 * An EFI kernel stub has been added that may be used to create
4677 kernel EFI binaries that contain not only the actual kernel,
4678 but also an initrd, boot splash, command line and OS release
4679 information. This combined binary can then be signed as a
4680 single image, so that the firmware can verify it all in one
4681 step. systemd-boot has special support for EFI binaries created
4682 like this and can extract OS release information from them
4683 and show them in the boot menu. This functionality is useful
4684 to implement cryptographically verified boot schemes.
4686 * Optional support has been added to systemd-fsck to pass
4687 fsck's progress report to an AF_UNIX socket in the file
4690 * udev will no longer create device symlinks for all block
4691 devices by default. A blacklist for excluding special block
4692 devices from this logic has been turned into a whitelist
4693 that requires picking block devices explicitly that require
4696 * A new (currently still internal) API sd-device.h has been
4697 added to libsystemd. This modernized API is supposed to
4698 replace libudev eventually. In fact, already much of libudev
4699 is now just a wrapper around sd-device.h.
4701 * A new hwdb database for storing metadata about pointing
4702 stick devices has been added.
4704 * systemd-tmpfiles gained support for setting file attributes
4705 similar to the "chattr" tool with new 'h' and 'H' lines.
4707 * systemd-journald will no longer unconditionally set the
4708 btrfs NOCOW flag on new journal files. This is instead done
4709 with tmpfiles snippet using the new 'h' line type. This
4710 allows easy disabling of this logic, by masking the
4711 journal-nocow.conf tmpfiles file.
4713 * systemd-journald will now translate audit message types to
4714 human readable identifiers when writing them to the
4715 journal. This should improve readability of audit messages.
4717 * The LUKS logic gained support for the offset= and skip=
4718 options in /etc/crypttab, as previously implemented by
4721 * /usr/lib/os-release gained a new optional field VARIANT= for
4722 distributions that support multiple variants (such as a
4723 desktop edition, a server edition, ...)
4725 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
4726 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
4727 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
4728 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
4729 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
4730 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
4731 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
4732 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
4733 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
4734 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
4735 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
4736 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
4737 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
4738 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
4739 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
4740 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
4741 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
4742 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
4743 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
4744 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
4745 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
4746 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
4747 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
4748 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
4749 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
4750 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
4751 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
4753 — Berlin, 2015-05-22
4757 * Introduce a new API "sd-hwdb.h" for querying the hardware
4758 metadata database. With this minimal interface one can query
4759 and enumerate the udev hwdb, decoupled from the old libudev
4760 library. libudev's interface for this is now only a wrapper
4761 around sd-hwdb. A new tool systemd-hwdb has been added to
4762 interface with and update the database.
4764 * When any of systemd's tools copies files (for example due to
4765 tmpfiles' C lines) a btrfs reflink will attempted first,
4766 before bytewise copying is done.
4768 * systemd-nspawn gained a new --ephemeral switch. When
4769 specified a btrfs snapshot is taken of the container's root
4770 directory, and immediately removed when the container
4771 terminates again. Thus, a container can be started whose
4772 changes never alter the container's root directory, and are
4773 lost on container termination. This switch can also be used
4774 for starting a container off the root file system of the
4775 host without affecting the host OS. This switch is only
4776 available on btrfs file systems.
4778 * systemd-nspawn gained a new --template= switch. It takes the
4779 path to a container tree to use as template for the tree
4780 specified via --directory=, should that directory be
4781 missing. This allows instantiating containers dynamically,
4782 on first run. This switch is only available on btrfs file
4785 * When a .mount unit refers to a mount point on which multiple
4786 mounts are stacked, and the .mount unit is stopped all of
4787 the stacked mount points will now be unmounted until no
4788 mount point remains.
4790 * systemd now has an explicit notion of supported and
4791 unsupported unit types. Jobs enqueued for unsupported unit
4792 types will now fail with an "unsupported" error code. More
4793 specifically .swap, .automount and .device units are not
4794 supported in containers, .busname units are not supported on
4795 non-kdbus systems. .swap and .automount are also not
4796 supported if their respective kernel compile time options
4799 * machinectl gained support for two new "copy-from" and
4800 "copy-to" commands for copying files from a running
4801 container to the host or vice versa.
4803 * machinectl gained support for a new "bind" command to bind
4804 mount host directories into local containers. This is
4805 currently only supported for nspawn containers.
4807 * networkd gained support for configuring bridge forwarding
4808 database entries (fdb) from .network files.
4810 * A new tiny daemon "systemd-importd" has been added that can
4811 download container images in tar, raw, qcow2 or dkr formats,
4812 and make them available locally in /var/lib/machines, so
4813 that they can run as nspawn containers. The daemon can GPG
4814 verify the downloads (not supported for dkr, since it has no
4815 provisions for verifying downloads). It will transparently
4816 decompress bz2, xz, gzip compressed downloads if necessary,
4817 and restore sparse files on disk. The daemon uses privilege
4818 separation to ensure the actual download logic runs with
4819 fewer privileges than the daemon itself. machinectl has
4820 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
4821 make the functionality of importd available to the
4822 user. With this in place the Fedora and Ubuntu "Cloud"
4823 images can be downloaded and booted as containers unmodified
4824 (the Fedora images lack the appropriate GPG signature files
4825 currently, so they cannot be verified, but this will change
4826 soon, hopefully). Note that downloading images is currently
4827 only fully supported on btrfs.
4829 * machinectl is now able to list container images found in
4830 /var/lib/machines, along with some metadata about sizes of
4831 disk and similar. If the directory is located on btrfs and
4832 quota is enabled, this includes quota display. A new command
4833 "image-status" has been added that shows additional
4834 information about images.
4836 * machinectl is now able to clone container images
4837 efficiently, if the underlying file system (btrfs) supports
4838 it, with the new "machinectl clone" command. It also
4839 gained commands for renaming and removing images, as well as
4840 marking them read-only or read-write (supported also on
4841 legacy file systems).
4843 * networkd gained support for collecting LLDP network
4844 announcements, from hardware that supports this. This is
4845 shown in networkctl output.
4847 * systemd-run gained support for a new -t (--pty) switch for
4848 invoking a binary on a pty whose input and output is
4849 connected to the invoking terminal. This allows executing
4850 processes as system services while interactively
4851 communicating with them via the terminal. Most interestingly
4852 this is supported across container boundaries. Invoking
4853 "systemd-run -t /bin/bash" is an alternative to running a
4854 full login session, the difference being that the former
4855 will not register a session, nor go through the PAM session
4858 * tmpfiles gained support for a new "v" line type for creating
4859 btrfs subvolumes. If the underlying file system is a legacy
4860 file system, this automatically degrades to creating a
4861 normal directory. Among others /var/lib/machines is now
4862 created like this at boot, should it be missing.
4864 * The directory /var/lib/containers/ has been deprecated and
4865 been replaced by /var/lib/machines. The term "machines" has
4866 been used in the systemd context as generic term for both
4867 VMs and containers, and hence appears more appropriate for
4868 this, as the directory can also contain raw images bootable
4871 * systemd-nspawn when invoked with -M but without --directory=
4872 or --image= is now capable of searching for the container
4873 root directory, subvolume or disk image automatically, in
4874 /var/lib/machines. systemd-nspawn@.service has been updated
4875 to make use of this, thus allowing it to be used for raw
4878 * A new machines.target unit has been introduced that is
4879 supposed to group all containers/VMs invoked as services on
4880 the system. systemd-nspawn@.service has been updated to
4881 integrate with that.
4883 * machinectl gained a new "start" command, for invoking a
4884 container as a service. "machinectl start foo" is mostly
4885 equivalent to "systemctl start systemd-nspawn@foo.service",
4886 but handles escaping in a nicer way.
4888 * systemd-nspawn will now mount most of the cgroupfs tree
4889 read-only into each container, with the exception of the
4890 container's own subtree in the name=systemd hierarchy.
4892 * journald now sets the special FS_NOCOW file flag for its
4893 journal files. This should improve performance on btrfs, by
4894 avoiding heavy fragmentation when journald's write-pattern
4895 is used on COW file systems. It degrades btrfs' data
4896 integrity guarantees for the files to the same levels as for
4897 ext3/ext4 however. This should be OK though as journald does
4898 its own data integrity checks and all its objects are
4899 checksummed on disk. Also, journald should handle btrfs disk
4900 full events a lot more gracefully now, by processing SIGBUS
4901 errors, and not relying on fallocate() anymore.
4903 * When journald detects that journal files it is writing to
4904 have been deleted it will immediately start new journal
4907 * systemd now provides a way to store file descriptors
4908 per-service in PID 1. This is useful for daemons to ensure
4909 that fds they require are not lost during a daemon
4910 restart. The fds are passed to the daemon on the next
4911 invocation in the same way socket activation fds are
4912 passed. This is now used by journald to ensure that the
4913 various sockets connected to all the system's stdout/stderr
4914 are not lost when journald is restarted. File descriptors
4915 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
4916 an extension to sd_notify(). Note that a limit is enforced
4917 on the number of fds a service can store in PID 1, and it
4918 defaults to 0, so that no fds may be stored, unless this is
4919 explicitly turned on.
4921 * The default TERM variable to use for units connected to a
4922 terminal, when no other value is explicitly is set is now
4923 vt220 rather than vt102. This should be fairly safe still,
4924 but allows PgUp/PgDn work.
4926 * The /etc/crypttab option header= as known from Debian is now
4929 * "loginctl user-status" and "loginctl session-status" will
4930 now show the last 10 lines of log messages of the
4931 user/session following the status output. Similar,
4932 "machinectl status" will show the last 10 log lines
4933 associated with a virtual machine or container
4934 service. (Note that this is usually not the log messages
4935 done in the VM/container itself, but simply what the
4936 container manager logs. For nspawn this includes all console
4939 * "loginctl session-status" without further argument will now
4940 show the status of the session of the caller. Similar,
4941 "lock-session", "unlock-session", "activate",
4942 "enable-linger", "disable-linger" may now be called without
4943 session/user parameter in which case they apply to the
4944 caller's session/user.
4946 * An X11 session scriptlet is now shipped that uploads
4947 $DISPLAY and $XAUTHORITY into the environment of the systemd
4948 --user daemon if a session begins. This should improve
4949 compatibility with X11 enabled applications run as systemd
4952 * Generators are now subject to masking via /etc and /run, the
4953 same way as unit files.
4955 * networkd .network files gained support for configuring
4956 per-link IPv4/IPv6 packet forwarding as well as IPv4
4957 masquerading. This is by default turned on for veth links to
4958 containers, as registered by systemd-nspawn. This means that
4959 nspawn containers run with --network-veth will now get
4960 automatic routed access to the host's networks without any
4961 further configuration or setup, as long as networkd runs on
4964 * systemd-nspawn gained the --port= (-p) switch to expose TCP
4965 or UDP posts of a container on the host. With this in place
4966 it is possible to run containers with private veth links
4967 (--network-veth), and have their functionality exposed on
4968 the host as if their services were running directly on the
4971 * systemd-nspawn's --network-veth switch now gained a short
4972 version "-n", since with the changes above it is now truly
4973 useful out-of-the-box. The systemd-nspawn@.service has been
4974 updated to make use of it too by default.
4976 * systemd-nspawn will now maintain a per-image R/W lock, to
4977 ensure that the same image is not started more than once
4978 writable. (It's OK to run an image multiple times
4979 simultaneously in read-only mode.)
4981 * systemd-nspawn's --image= option is now capable of
4982 dissecting and booting MBR and GPT disk images that contain
4983 only a single active Linux partition. Previously it
4984 supported only GPT disk images with proper GPT type
4985 IDs. This allows running cloud images from major
4986 distributions directly with systemd-nspawn, without
4989 * In addition to collecting mouse dpi data in the udev
4990 hardware database, there's now support for collecting angle
4991 information for mouse scroll wheels. The database is
4992 supposed to guarantee similar scrolling behavior on mice
4993 that it knows about. There's also support for collecting
4994 information about Touchpad types.
4996 * udev's input_id built-in will now also collect touch screen
4997 dimension data and attach it to probed devices.
4999 * /etc/os-release gained support for a Distribution Privacy
5002 * networkd gained support for creating "ipvlan", "gretap",
5003 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
5005 * systemd-tmpfiles gained support for "a" lines for setting
5008 * systemd-nspawn will now mount /tmp in the container to
5009 tmpfs, automatically.
5011 * systemd now exposes the memory.usage_in_bytes cgroup
5012 attribute and shows it for each service in the "systemctl
5013 status" output, if available.
5015 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
5016 immediate reboot is triggered. This useful if shutdown is
5017 hung and is unable to complete, to expedite the
5018 operation. Note that this kind of reboot will still unmount
5019 all file systems, and hence should not result in fsck being
5022 * A .device unit for an optical block device will now be
5023 considered active only when a medium is in the drive. Also,
5024 mount units are now bound to their backing devices thus
5025 triggering automatic unmounting when devices become
5026 unavailable. With this in place systemd will now
5027 automatically unmount left-over mounts when a CD-ROM is
5028 ejected or an USB stick is yanked from the system.
5030 * networkd-wait-online now has support for waiting for
5031 specific interfaces only (with globbing), and for giving up
5032 after a configurable timeout.
5034 * networkd now exits when idle. It will be automatically
5035 restarted as soon as interfaces show up, are removed or
5036 change state. networkd will stay around as long as there is
5037 at least one DHCP state machine or similar around, that keep
5040 * networkd may now configure IPv6 link-local addressing in
5041 addition to IPv4 link-local addressing.
5043 * The IPv6 "token" for use in SLAAC may now be configured for
5044 each .network interface in networkd.
5046 * Routes configured with networkd may now be assigned a scope
5049 * networkd's [Match] sections now support globbing and lists
5050 of multiple space-separated matches per item.
5052 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
5053 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
5054 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
5055 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
5056 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
5057 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
5058 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
5059 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
5060 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
5061 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
5062 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
5063 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
5064 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
5065 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
5066 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
5067 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
5068 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
5069 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
5070 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
5071 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
5072 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
5073 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
5074 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
5075 Hoffmann, Zbigniew Jędrzejewski-Szmek
5077 — Berlin, 2015-02-16
5081 * When querying unit file enablement status (for example via
5082 "systemctl is-enabled"), a new state "indirect" is now known
5083 which indicates that a unit might not be enabled itself, but
5084 another unit listed in its Also= setting might be.
5086 * Similar to the various existing ConditionXYZ= settings for
5087 units, there are now matching AssertXYZ= settings. While
5088 failing conditions cause a unit to be skipped, but its job
5089 to succeed, failing assertions declared like this will cause
5090 a unit start operation and its job to fail.
5092 * hostnamed now knows a new chassis type "embedded".
5094 * systemctl gained a new "edit" command. When used on a unit
5095 file, this allows extending unit files with .d/ drop-in
5096 configuration snippets or editing the full file (after
5097 copying it from /usr/lib to /etc). This will invoke the
5098 user's editor (as configured with $EDITOR), and reload the
5099 modified configuration after editing.
5101 * "systemctl status" now shows the suggested enablement state
5102 for a unit, as declared in the (usually vendor-supplied)
5103 system preset files.
5105 * nss-myhostname will now resolve the single-label host name
5106 "gateway" to the locally configured default IP routing
5107 gateways, ordered by their metrics. This assigns a stable
5108 name to the used gateways, regardless which ones are
5109 currently configured. Note that the name will only be
5110 resolved after all other name sources (if nss-myhostname is
5111 configured properly) and should hence not negatively impact
5112 systems that use the single-label host name "gateway" in
5115 * systemd-inhibit now allows filtering by mode when listing
5118 * Scope and service units gained a new "Delegate" boolean
5119 property, which, when set, allows processes running inside the
5120 unit to further partition resources. This is primarily
5121 useful for systemd user instances as well as container
5124 * journald will now pick up audit messages directly from
5125 the kernel, and log them like any other log message. The
5126 audit fields are split up and fully indexed. This means that
5127 journalctl in many ways is now a (nicer!) alternative to
5128 ausearch, the traditional audit client. Note that this
5129 implements only a minimal audit client. If you want the
5130 special audit modes like reboot-on-log-overflow, please use
5131 the traditional auditd instead, which can be used in
5132 parallel to journald.
5134 * The ConditionSecurity= unit file option now understands the
5135 special string "audit" to check whether auditing is
5138 * journalctl gained two new commands --vacuum-size= and
5139 --vacuum-time= to delete old journal files until the
5140 remaining ones take up no more than the specified size on disk,
5141 or are not older than the specified time.
5143 * A new, native PPPoE library has been added to sd-network,
5144 systemd's library of light-weight networking protocols. This
5145 library will be used in a future version of networkd to
5146 enable PPPoE communication without an external pppd daemon.
5148 * The busctl tool now understands a new "capture" verb that
5149 works similar to "monitor", but writes a packet capture
5150 trace to STDOUT that can be redirected to a file which is
5151 compatible with libcap's capture file format. This can then
5152 be loaded in Wireshark and similar tools to inspect bus
5155 * The busctl tool now understands a new "tree" verb that shows
5156 the object trees of a specific service on the bus, or of all
5159 * The busctl tool now understands a new "introspect" verb that
5160 shows all interfaces and members of objects on the bus,
5161 including their signature and values. This is particularly
5162 useful to get more information about bus objects shown by
5163 the new "busctl tree" command.
5165 * The busctl tool now understands new verbs "call",
5166 "set-property" and "get-property" for invoking bus method
5167 calls, setting and getting bus object properties in a
5170 * busctl gained a new --augment-creds= argument that controls
5171 whether the tool shall augment credential information it
5172 gets from the bus with data from /proc, in a possibly
5175 * nspawn's --link-journal= switch gained two new values
5176 "try-guest" and "try-host" that work like "guest" and
5177 "host", but do not fail if the host has no persistent
5178 journalling enabled. -j is now equivalent to
5179 --link-journal=try-guest.
5181 * macvlan network devices created by nspawn will now have
5182 stable MAC addresses.
5184 * A new SmackProcessLabel= unit setting has been added, which
5185 controls the SMACK security label processes forked off by
5186 the respective unit shall use.
5188 * If compiled with --enable-xkbcommon, systemd-localed will
5189 verify x11 keymap settings by compiling the given keymap. It
5190 will spew out warnings if the compilation fails. This
5191 requires libxkbcommon to be installed.
5193 * When a coredump is collected, a larger number of metadata
5194 fields is now collected and included in the journal records
5195 created for it. More specifically, control group membership,
5196 environment variables, memory maps, working directory,
5197 chroot directory, /proc/$PID/status, and a list of open file
5198 descriptors is now stored in the log entry.
5200 * The udev hwdb now contains DPI information for mice. For
5203 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
5205 * All systemd programs that read standalone configuration
5206 files in /etc now also support a corresponding series of
5207 .conf.d configuration directories in /etc/, /run/,
5208 /usr/local/lib/, /usr/lib/, and (if configured with
5209 --enable-split-usr) /lib/. In particular, the following
5210 configuration files now have corresponding configuration
5211 directories: system.conf user.conf, logind.conf,
5212 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
5213 resolved.conf, timesyncd.conf, journal-remote.conf, and
5214 journal-upload.conf. Note that distributions should use the
5215 configuration directories in /usr/lib/; the directories in
5216 /etc/ are reserved for the system administrator.
5218 * systemd-rfkill will no longer take the rfkill device name
5219 into account when storing rfkill state on disk, as the name
5220 might be dynamically assigned and not stable. Instead, the
5221 ID_PATH udev variable combined with the rfkill type (wlan,
5222 bluetooth, ...) is used.
5224 * A new service systemd-machine-id-commit.service has been
5225 added. When used on systems where /etc is read-only during
5226 boot, and /etc/machine-id is not initialized (but an empty
5227 file), this service will copy the temporary machine ID
5228 created as replacement into /etc after the system is fully
5229 booted up. This is useful for systems that are freshly
5230 installed with a non-initialized machine ID, but should get
5231 a fixed machine ID for subsequent boots.
5233 * networkd's .netdev files now provide a large set of
5234 configuration parameters for VXLAN devices. Similarly, the
5235 bridge port cost parameter is now configurable in .network
5236 files. There's also new support for configuring IP source
5237 routing. networkd .link files gained support for a new
5238 OriginalName= match that is useful to match against the
5239 original interface name the kernel assigned. .network files
5240 may include MTU= and MACAddress= fields for altering the MTU
5241 and MAC address while being connected to a specific network
5244 * The LUKS logic gained supported for configuring
5245 UUID-specific key files. There's also new support for naming
5246 LUKS device from the kernel command line, using the new
5247 luks.name= argument.
5249 * Timer units may now be transiently created via the bus API
5250 (this was previously already available for scope and service
5251 units). In addition it is now possible to create multiple
5252 transient units at the same time with a single bus call. The
5253 "systemd-run" tool has been updated to make use of this for
5254 running commands on a specified time, in at(1)-style.
5256 * tmpfiles gained support for "t" lines, for assigning
5257 extended attributes to files. Among other uses this may be
5258 used to assign SMACK labels to files.
5260 Contributions from: Alin Rauta, Alison Chaiken, Andrej
5261 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
5262 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
5263 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
5264 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
5265 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
5266 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
5267 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
5268 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
5269 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
5270 Michael Biebl, Michael Chapman, Michael Marineau, Michal
5271 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
5272 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
5273 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
5274 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
5275 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5276 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
5277 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
5279 — Berlin, 2014-12-10
5283 * journalctl gained the new options -t/--identifier= to match
5284 on the syslog identifier (aka "tag"), as well as --utc to
5285 show log timestamps in the UTC timezone. journalctl now also
5286 accepts -n/--lines=all to disable line capping in a pager.
5288 * journalctl gained a new switch, --flush, that synchronously
5289 flushes logs from /run/log/journal to /var/log/journal if
5290 persistent storage is enabled. systemd-journal-flush.service
5291 now waits until the operation is complete.
5293 * Services can notify the manager before they start a reload
5294 (by sending RELOADING=1) or shutdown (by sending
5295 STOPPING=1). This allows the manager to track and show the
5296 internal state of daemons and closes a race condition when
5297 the process is still running but has closed its D-Bus
5300 * Services with Type=oneshot do not have to have any ExecStart
5303 * User units are now loaded also from
5304 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
5305 /run/systemd/user directory that was already previously
5306 supported, but is under the control of the user.
5308 * Job timeouts (i.e. timeouts on the time a job that is
5309 queued stays in the run queue) can now optionally result in
5310 immediate reboot or power-off actions (JobTimeoutAction= and
5311 JobTimeoutRebootArgument=). This is useful on ".target"
5312 units, to limit the maximum time a target remains
5313 undispatched in the run queue, and to trigger an emergency
5314 operation in such a case. This is now used by default to
5315 turn off the system if boot-up (as defined by everything in
5316 basic.target) hangs and does not complete for at least
5317 15min. Also, if power-off or reboot hang for at least 30min
5318 an immediate power-off/reboot operation is triggered. This
5319 functionality is particularly useful to increase reliability
5320 on embedded devices, but also on laptops which might
5321 accidentally get powered on when carried in a backpack and
5322 whose boot stays stuck in a hard disk encryption passphrase
5325 * systemd-logind can be configured to also handle lid switch
5326 events even when the machine is docked or multiple displays
5327 are attached (HandleLidSwitchDocked= option).
5329 * A helper binary and a service have been added which can be
5330 used to resume from hibernation in the initramfs. A
5331 generator will parse the resume= option on the kernel
5332 command line to trigger resume.
5334 * A user console daemon systemd-consoled has been
5335 added. Currently, it is a preview, and will so far open a
5336 single terminal on each session of the user marked as
5337 Desktop=systemd-console.
5339 * Route metrics can be specified for DHCP routes added by
5342 * The SELinux context of socket-activated services can be set
5343 from the information provided by the networking stack
5344 (SELinuxContextFromNet= option).
5346 * Userspace firmware loading support has been removed and
5347 the minimum supported kernel version is thus bumped to 3.7.
5349 * Timeout for udev workers has been increased from 1 to 3
5350 minutes, but a warning will be printed after 1 minute to
5351 help diagnose kernel modules that take a long time to load.
5353 * Udev rules can now remove tags on devices with TAG-="foobar".
5355 * systemd's readahead implementation has been removed. In many
5356 circumstances it didn't give expected benefits even for
5357 rotational disk drives and was becoming less relevant in the
5358 age of SSDs. As none of the developers has been using
5359 rotating media anymore, and nobody stepped up to actively
5360 maintain this component of systemd it has now been removed.
5362 * Swap units can use Options= to specify discard options.
5363 Discard options specified for swaps in /etc/fstab are now
5366 * Docker containers are now detected as a separate type of
5369 * The Password Agent protocol gained support for queries where
5370 the user input is shown, useful e.g. for user names.
5371 systemd-ask-password gained a new --echo option to turn that
5374 * The default sysctl.d/ snippets will now set:
5376 net.core.default_qdisc = fq_codel
5378 This selects Fair Queuing Controlled Delay as the default
5379 queuing discipline for network interfaces. fq_codel helps
5380 fight the network bufferbloat problem. It is believed to be
5381 a good default with no tuning required for most workloads.
5382 Downstream distributions may override this choice. On 10Gbit
5383 servers that do not do forwarding, "fq" may perform better.
5384 Systems without a good clocksource should use "pfifo_fast".
5386 * If kdbus is enabled during build a new option BusPolicy= is
5387 available for service units, that allows locking all service
5388 processes into a stricter bus policy, in order to limit
5389 access to various bus services, or even hide most of them
5390 from the service's view entirely.
5392 * networkctl will now show the .network and .link file
5393 networkd has applied to a specific interface.
5395 * sd-login gained a new API call sd_session_get_desktop() to
5396 query which desktop environment has been selected for a
5399 * UNIX utmp support is now compile-time optional to support
5400 legacy-free systems.
5402 * systemctl gained two new commands "add-wants" and
5403 "add-requires" for pulling in units from specific targets
5406 * If the word "rescue" is specified on the kernel command line
5407 the system will now boot into rescue mode (aka
5408 rescue.target), which was previously available only by
5409 specifying "1" or "systemd.unit=rescue.target" on the kernel
5410 command line. This new kernel command line option nicely
5411 mirrors the already existing "emergency" kernel command line
5414 * New kernel command line options mount.usr=, mount.usrflags=,
5415 mount.usrfstype= have been added that match root=, rootflags=,
5416 rootfstype= but allow mounting a specific file system to
5419 * The $NOTIFY_SOCKET is now also passed to control processes of
5420 services, not only the main process.
5422 * This version reenables support for fsck's -l switch. This
5423 means at least version v2.25 of util-linux is required for
5424 operation, otherwise dead-locks on device nodes may
5425 occur. Again: you need to update util-linux to at least
5426 v2.25 when updating systemd to v217.
5428 * The "multi-seat-x" tool has been removed from systemd, as
5429 its functionality has been integrated into X servers 1.16,
5430 and the tool is hence redundant. It is recommended to update
5431 display managers invoking this tool to simply invoke X
5432 directly from now on, again.
5434 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
5435 message flag has been added for all of systemd's polkit
5436 authenticated method calls has been added. In particular this
5437 now allows optional interactive authorization via polkit for
5438 many of PID1's privileged operations such as unit file
5439 enabling and disabling.
5441 * "udevadm hwdb --update" learnt a new switch "--usr" for
5442 placing the rebuilt hardware database in /usr instead of
5443 /etc. When used only hardware database entries stored in
5444 /usr will be used, and any user database entries in /etc are
5445 ignored. This functionality is useful for vendors to ship a
5446 pre-built database on systems where local configuration is
5447 unnecessary or unlikely.
5449 * Calendar time specifications in .timer units now also
5450 understand the strings "semi-annually", "quarterly" and
5451 "minutely" as shortcuts (in addition to the preexisting
5452 "anually", "hourly", ...).
5454 * systemd-tmpfiles will now correctly create files in /dev
5455 at boot which are marked for creation only at boot. It is
5456 recommended to always create static device nodes with 'c!'
5457 and 'b!', so that they are created only at boot and not
5458 overwritten at runtime.
5460 * When the watchdog logic is used for a service (WatchdogSec=)
5461 and the watchdog timeout is hit the service will now be
5462 terminated with SIGABRT (instead of just SIGTERM), in order
5463 to make sure a proper coredump and backtrace is
5464 generated. This ensures that hanging services will result in
5465 similar coredump/backtrace behaviour as services that hit a
5468 Contributions from: Andreas Henriksson, Andrei Borzenkov,
5469 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
5470 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
5471 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
5472 Herrmann, David Sommerseth, David Strauss, Emil Renner
5473 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
5474 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
5475 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
5476 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
5477 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
5478 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
5479 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
5480 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
5481 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
5482 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
5483 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
5484 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
5485 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
5486 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
5487 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
5488 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
5491 — Berlin, 2014-10-28
5495 * timedated no longer reads NTP implementation unit names from
5496 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
5497 implementations should add a
5499 Conflicts=systemd-timesyncd.service
5501 to their unit files to take over and replace systemd's NTP
5502 default functionality.
5504 * systemd-sysusers gained a new line type "r" for configuring
5505 which UID/GID ranges to allocate system users/groups
5506 from. Lines of type "u" may now add an additional column
5507 that specifies the home directory for the system user to be
5508 created. Also, systemd-sysusers may now optionally read user
5509 information from STDIN instead of a file. This is useful for
5510 invoking it from RPM preinst scriptlets that need to create
5511 users before the first RPM file is installed since these
5512 files might need to be owned by them. A new
5513 %sysusers_create_inline RPM macro has been introduced to do
5514 just that. systemd-sysusers now updates the shadow files as
5515 well as the user/group databases, which should enhance
5516 compatibility with certain tools like grpck.
5518 * A number of bus APIs of PID 1 now optionally consult polkit to
5519 permit access for otherwise unprivileged clients under certain
5520 conditions. Note that this currently doesn't support
5521 interactive authentication yet, but this is expected to be
5522 added eventually, too.
5524 * /etc/machine-info now has new fields for configuring the
5525 deployment environment of the machine, as well as the
5526 location of the machine. hostnamectl has been updated with
5527 new command to update these fields.
5529 * systemd-timesyncd has been updated to automatically acquire
5530 NTP server information from systemd-networkd, which might
5531 have been discovered via DHCP.
5533 * systemd-resolved now includes a caching DNS stub resolver
5534 and a complete LLMNR name resolution implementation. A new
5535 NSS module "nss-resolve" has been added which can be used
5536 instead of glibc's own "nss-dns" to resolve hostnames via
5537 systemd-resolved. Hostnames, addresses and arbitrary RRs may
5538 be resolved via systemd-resolved D-Bus APIs. In contrast to
5539 the glibc internal resolver systemd-resolved is aware of
5540 multi-homed system, and keeps DNS server and caches separate
5541 and per-interface. Queries are sent simultaneously on all
5542 interfaces that have DNS servers configured, in order to
5543 properly handle VPNs and local LANs which might resolve
5544 separate sets of domain names. systemd-resolved may acquire
5545 DNS server information from systemd-networkd automatically,
5546 which in turn might have discovered them via DHCP. A tool
5547 "systemd-resolve-host" has been added that may be used to
5548 query the DNS logic in resolved. systemd-resolved implements
5549 IDNA and automatically uses IDNA or UTF-8 encoding depending
5550 on whether classic DNS or LLMNR is used as transport. In the
5551 next releases we intend to add a DNSSEC and mDNS/DNS-SD
5552 implementation to systemd-resolved.
5554 * A new NSS module nss-mymachines has been added, that
5555 automatically resolves the names of all local registered
5556 containers to their respective IP addresses.
5558 * A new client tool "networkctl" for systemd-networkd has been
5559 added. It currently is entirely passive and will query
5560 networking configuration from udev, rtnetlink and networkd,
5561 and present it to the user in a very friendly
5562 way. Eventually, we hope to extend it to become a full
5563 control utility for networkd.
5565 * .socket units gained a new DeferAcceptSec= setting that
5566 controls the kernels' TCP_DEFER_ACCEPT sockopt for
5567 TCP. Similarly, support for controlling TCP keep-alive
5568 settings has been added (KeepAliveTimeSec=,
5569 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
5570 turning off Nagle's algorithm on TCP has been added
5573 * logind learned a new session type "web", for use in projects
5574 like Cockpit which register web clients as PAM sessions.
5576 * timer units with at least one OnCalendar= setting will now
5577 be started only after time-sync.target has been
5578 reached. This way they will not elapse before the system
5579 clock has been corrected by a local NTP client or
5580 similar. This is particular useful on RTC-less embedded
5581 machines, that come up with an invalid system clock.
5583 * systemd-nspawn's --network-veth= switch should now result in
5584 stable MAC addresses for both the outer and the inner side
5587 * systemd-nspawn gained a new --volatile= switch for running
5588 container instances with /etc or /var unpopulated.
5590 * The kdbus client code has been updated to use the new Linux
5591 3.17 memfd subsystem instead of the old kdbus-specific one.
5593 * systemd-networkd's DHCP client and server now support
5594 FORCERENEW. There are also new configuration options to
5595 configure the vendor client identifier and broadcast mode
5598 * systemd will no longer inform the kernel about the current
5599 timezone, as this is necessarily incorrect and racy as the
5600 kernel has no understanding of DST and similar
5601 concepts. This hence means FAT timestamps will be always
5602 considered UTC, similar to what Android is already
5603 doing. Also, when the RTC is configured to the local time
5604 (rather than UTC) systemd will never synchronize back to it,
5605 as this might confuse Windows at a later boot.
5607 * systemd-analyze gained a new command "verify" for offline
5608 validation of unit files.
5610 * systemd-networkd gained support for a couple of additional
5611 settings for bonding networking setups. Also, the metric for
5612 statically configured routes may now be configured. For
5613 network interfaces where this is appropriate the peer IP
5614 address may now be configured.
5616 * systemd-networkd's DHCP client will no longer request
5617 broadcasting by default, as this tripped up some networks.
5618 For hardware where broadcast is required the feature should
5619 be switched back on using RequestBroadcast=yes.
5621 * systemd-networkd will now set up IPv4LL addresses (when
5622 enabled) even if DHCP is configured successfully.
5624 * udev will now default to respect network device names given
5625 by the kernel when the kernel indicates that these are
5626 predictable. This behavior can be tweaked by changing
5627 NamePolicy= in the relevant .link file.
5629 * A new library systemd-terminal has been added that
5630 implements full TTY stream parsing and rendering. This
5631 library is supposed to be used later on for implementing a
5632 full userspace VT subsystem, replacing the current kernel
5635 * A new tool systemd-journal-upload has been added to push
5636 journal data to a remote system running
5637 systemd-journal-remote.
5639 * journald will no longer forward all local data to another
5640 running syslog daemon. This change has been made because
5641 rsyslog (which appears to be the most commonly used syslog
5642 implementation these days) no longer makes use of this, and
5643 instead pulls the data out of the journal on its own. Since
5644 forwarding the messages to a non-existent syslog server is
5645 more expensive than we assumed we have now turned this
5646 off. If you run a syslog server that is not a recent rsyslog
5647 version, you have to turn this option on again
5648 (ForwardToSyslog= in journald.conf).
5650 * journald now optionally supports the LZ4 compressor for
5651 larger journal fields. This compressor should perform much
5652 better than XZ which was the previous default.
5654 * machinectl now shows the IP addresses of local containers,
5655 if it knows them, plus the interface name of the container.
5657 * A new tool "systemd-escape" has been added that makes it
5658 easy to escape strings to build unit names and similar.
5660 * sd_notify() messages may now include a new ERRNO= field
5661 which is parsed and collected by systemd and shown among the
5662 "systemctl status" output for a service.
5664 * A new component "systemd-firstboot" has been added that
5665 queries the most basic systemd information (timezone,
5666 hostname, root password) interactively on first
5667 boot. Alternatively it may also be used to provision these
5668 things offline on OS images installed into directories.
5670 * The default sysctl.d/ snippets will now set
5672 net.ipv4.conf.default.promote_secondaries=1
5674 This has the benefit of no flushing secondary IP addresses
5675 when primary addresses are removed.
5677 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
5678 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
5679 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
5680 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
5681 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
5682 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
5683 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
5684 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
5685 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
5686 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
5687 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
5688 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
5689 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
5690 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
5691 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
5693 — Berlin, 2014-08-19
5697 * A new tool systemd-sysusers has been added. This tool
5698 creates system users and groups in /etc/passwd and
5699 /etc/group, based on static declarative system user/group
5700 definitions in /usr/lib/sysusers.d/. This is useful to
5701 enable factory resets and volatile systems that boot up with
5702 an empty /etc directory, and thus need system users and
5703 groups created during early boot. systemd now also ships
5704 with two default sysusers.d/ files for the most basic
5705 users and groups systemd and the core operating system
5708 * A new tmpfiles snippet has been added that rebuilds the
5709 essential files in /etc on boot, should they be missing.
5711 * A directive for ensuring automatic clean-up of
5712 /var/cache/man/ has been removed from the default
5713 configuration. This line should now be shipped by the man
5714 implementation. The necessary change has been made to the
5715 man-db implementation. Note that you need to update your man
5716 implementation to one that ships this line, otherwise no
5717 automatic clean-up of /var/cache/man will take place.
5719 * A new condition ConditionNeedsUpdate= has been added that
5720 may conditionalize services to only run when /etc or /var
5721 are "older" than the vendor operating system resources in
5722 /usr. This is useful for reconstructing or updating /etc
5723 after an offline update of /usr or a factory reset, on the
5724 next reboot. Services that want to run once after such an
5725 update or reset should use this condition and order
5726 themselves before the new systemd-update-done.service, which
5727 will mark the two directories as fully updated. A number of
5728 service files have been added making use of this, to rebuild
5729 the udev hardware database, the journald message catalog and
5730 dynamic loader cache (ldconfig). The systemd-sysusers tool
5731 described above also makes use of this now. With this in
5732 place it is now possible to start up a minimal operating
5733 system with /etc empty cleanly. For more information on the
5734 concepts involved see this recent blog story:
5736 http://0pointer.de/blog/projects/stateless.html
5738 * A new system group "input" has been introduced, and all
5739 input device nodes get this group assigned. This is useful
5740 for system-level software to get access to input devices. It
5741 complements what is already done for "audio" and "video".
5743 * systemd-networkd learnt minimal DHCPv4 server support in
5744 addition to the existing DHCPv4 client support. It also
5745 learnt DHCPv6 client and IPv6 Router Solicitation client
5746 support. The DHCPv4 client gained support for static routes
5747 passed in from the server. Note that the [DHCPv4] section
5748 known in older systemd-networkd versions has been renamed to
5749 [DHCP] and is now also used by the DHCPv6 client. Existing
5750 .network files using settings of this section should be
5751 updated, though compatibility is maintained. Optionally, the
5752 client hostname may now be sent to the DHCP server.
5754 * networkd gained support for vxlan virtual networks as well
5755 as tun/tap and dummy devices.
5757 * networkd gained support for automatic allocation of address
5758 ranges for interfaces from a system-wide pool of
5759 addresses. This is useful for dynamically managing a large
5760 number of interfaces with a single network configuration
5761 file. In particular this is useful to easily assign
5762 appropriate IP addresses to the veth links of a large number
5763 of nspawn instances.
5765 * RPM macros for processing sysusers, sysctl and binfmt
5766 drop-in snippets at package installation time have been
5769 * The /etc/os-release file should now be placed in
5770 /usr/lib/os-release. The old location is automatically
5771 created as symlink. /usr/lib is the more appropriate
5772 location of this file, since it shall actually describe the
5773 vendor operating system shipped in /usr, and not the
5774 configuration stored in /etc.
5776 * .mount units gained a new boolean SloppyOptions= setting
5777 that maps to mount(8)'s -s option which enables permissive
5778 parsing of unknown mount options.
5780 * tmpfiles learnt a new "L+" directive which creates a symlink
5781 but (unlike "L") deletes a pre-existing file first, should
5782 it already exist and not already be the correct
5783 symlink. Similarly, "b+", "c+" and "p+" directives have been
5784 added as well, which create block and character devices, as
5785 well as fifos in the filesystem, possibly removing any
5786 pre-existing files of different types.
5788 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
5789 'argument' field (which so far specified the source to
5790 symlink/copy the files from) is now optional. If omitted the
5791 same file os copied from /usr/share/factory/ suffixed by the
5792 full destination path. This is useful for populating /etc
5793 with essential files, by copying them from vendor defaults
5794 shipped in /usr/share/factory/etc.
5796 * A new command "systemctl preset-all" has been added that
5797 applies the service preset settings to all installed unit
5798 files. A new switch --preset-mode= has been added that
5799 controls whether only enable or only disable operations
5802 * A new command "systemctl is-system-running" has been added
5803 that allows checking the overall state of the system, for
5804 example whether it is fully up and running.
5806 * When the system boots up with an empty /etc, the equivalent
5807 to "systemctl preset-all" is executed during early boot, to
5808 make sure all default services are enabled after a factory
5811 * systemd now contains a minimal preset file that enables the
5812 most basic services systemd ships by default.
5814 * Unit files' [Install] section gained a new DefaultInstance=
5815 field for defining the default instance to create if a
5816 template unit is enabled with no instance specified.
5818 * A new passive target cryptsetup-pre.target has been added
5819 that may be used by services that need to make they run and
5820 finish before the first LUKS cryptographic device is set up.
5822 * The /dev/loop-control and /dev/btrfs-control device nodes
5823 are now owned by the "disk" group by default, opening up
5824 access to this group.
5826 * systemd-coredump will now automatically generate a
5827 stack trace of all core dumps taking place on the system,
5828 based on elfutils' libdw library. This stack trace is logged
5831 * systemd-coredump may now optionally store coredumps directly
5832 on disk (in /var/lib/systemd/coredump, possibly compressed),
5833 instead of storing them unconditionally in the journal. This
5834 mode is the new default. A new configuration file
5835 /etc/systemd/coredump.conf has been added to configure this
5836 and other parameters of systemd-coredump.
5838 * coredumpctl gained a new "info" verb to show details about a
5839 specific coredump. A new switch "-1" has also been added
5840 that makes sure to only show information about the most
5841 recent entry instead of all entries. Also, as the tool is
5842 generally useful now the "systemd-" prefix of the binary
5843 name has been removed. Distributions that want to maintain
5844 compatibility with the old name should add a symlink from
5845 the old name to the new name.
5847 * journald's SplitMode= now defaults to "uid". This makes sure
5848 that unprivileged users can access their own coredumps with
5849 coredumpctl without restrictions.
5851 * New kernel command line options "systemd.wants=" (for
5852 pulling an additional unit during boot), "systemd.mask="
5853 (for masking a specific unit for the boot), and
5854 "systemd.debug-shell" (for enabling the debug shell on tty9)
5855 have been added. This is implemented in the new generator
5856 "systemd-debug-generator".
5858 * systemd-nspawn will now by default filter a couple of
5859 syscalls for containers, among them those required for
5860 kernel module loading, direct x86 IO port access, swap
5861 management, and kexec. Most importantly though
5862 open_by_handle_at() is now prohibited for containers,
5863 closing a hole similar to a recently discussed vulnerability
5864 in docker regarding access to files on file hierarchies the
5865 container should normally not have access to. Note that, for
5866 nspawn, we generally make no security claims anyway (and
5867 this is explicitly documented in the man page), so this is
5868 just a fix for one of the most obvious problems.
5870 * A new man page file-hierarchy(7) has been added that
5871 contains a minimized, modernized version of the file system
5872 layout systemd expects, similar in style to the FHS
5873 specification or hier(5). A new tool systemd-path(1) has
5874 been added to query many of these paths for the local
5877 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
5878 longer done. Since the directory now has a per-user size
5879 limit, and is cleaned on logout this appears unnecessary,
5880 in particular since this now brings the lifecycle of this
5881 directory closer in line with how IPC objects are handled.
5883 * systemd.pc now exports a number of additional directories,
5884 including $libdir (which is useful to identify the library
5885 path for the primary architecture of the system), and a
5886 couple of drop-in directories.
5888 * udev's predictable network interface names now use the dev_port
5889 sysfs attribute, introduced in linux 3.15 instead of dev_id to
5890 distinguish between ports of the same PCI function. dev_id should
5891 only be used for ports using the same HW address, hence the need
5894 * machined has been updated to export the OS version of a
5895 container (read from /etc/os-release and
5896 /usr/lib/os-release) on the bus. This is now shown in
5897 "machinectl status" for a machine.
5899 * A new service setting RestartForceExitStatus= has been
5900 added. If configured to a set of exit signals or process
5901 return values, the service will be restarted when the main
5902 daemon process exits with any of them, regardless of the
5905 * systemctl's -H switch for connecting to remote systemd
5906 machines has been extended so that it may be used to
5907 directly connect to a specific container on the
5908 host. "systemctl -H root@foobar:waldi" will now connect as
5909 user "root" to host "foobar", and then proceed directly to
5910 the container named "waldi". Note that currently you have to
5911 authenticate as user "root" for this to work, as entering
5912 containers is a privileged operation.
5914 Contributions from: Andreas Henriksson, Benjamin Steinwender,
5915 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
5916 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
5917 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
5918 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
5919 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
5920 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
5921 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
5922 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
5923 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
5924 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
5925 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
5927 — Berlin, 2014-07-03
5931 * As an experimental feature, udev now tries to lock the
5932 disk device node (flock(LOCK_SH|LOCK_NB)) while it
5933 executes events for the disk or any of its partitions.
5934 Applications like partitioning programs can lock the
5935 disk device node (flock(LOCK_EX)) and claim temporary
5936 device ownership that way; udev will entirely skip all event
5937 handling for this disk and its partitions. If the disk
5938 was opened for writing, the close will trigger a partition
5939 table rescan in udev's "watch" facility, and if needed
5940 synthesize "change" events for the disk and all its partitions.
5941 This is now unconditionally enabled, and if it turns out to
5942 cause major problems, we might turn it on only for specific
5943 devices, or might need to disable it entirely. Device Mapper
5944 devices are excluded from this logic.
5946 * We temporarily dropped the "-l" switch for fsck invocations,
5947 since they collide with the flock() logic above. util-linux
5948 upstream has been changed already to avoid this conflict,
5949 and we will readd "-l" as soon as util-linux with this
5950 change has been released.
5952 * The dependency on libattr has been removed. Since a long
5953 time, the extended attribute calls have moved to glibc, and
5954 libattr is thus unnecessary.
5956 * Virtualization detection works without privileges now. This
5957 means the systemd-detect-virt binary no longer requires
5958 CAP_SYS_PTRACE file capabilities, and our daemons can run
5959 with fewer privileges.
5961 * systemd-networkd now runs under its own "systemd-network"
5962 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
5963 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
5964 loses the ability to write to files owned by root this way.
5966 * Similarly, systemd-resolved now runs under its own
5967 "systemd-resolve" user with no capabilities remaining.
5969 * Similarly, systemd-bus-proxyd now runs under its own
5970 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
5972 * systemd-networkd gained support for setting up "veth"
5973 virtual Ethernet devices for container connectivity, as well
5974 as GRE and VTI tunnels.
5976 * systemd-networkd will no longer automatically attempt to
5977 manually load kernel modules necessary for certain tunnel
5978 transports. Instead, it is assumed the kernel loads them
5979 automatically when required. This only works correctly on
5980 very new kernels. On older kernels, please consider adding
5981 the kernel modules to /etc/modules-load.d/ as a work-around.
5983 * The resolv.conf file systemd-resolved generates has been
5984 moved to /run/systemd/resolve/. If you have a symlink from
5985 /etc/resolv.conf, it might be necessary to correct it.
5987 * Two new service settings, ProtectHome= and ProtectSystem=,
5988 have been added. When enabled, they will make the user data
5989 (such as /home) inaccessible or read-only and the system
5990 (such as /usr) read-only, for specific services. This allows
5991 very light-weight per-service sandboxing to avoid
5992 modifications of user data or system files from
5993 services. These two new switches have been enabled for all
5994 of systemd's long-running services, where appropriate.
5996 * Socket units gained new SocketUser= and SocketGroup=
5997 settings to set the owner user and group of AF_UNIX sockets
5998 and FIFOs in the file system.
6000 * Socket units gained a new RemoveOnStop= setting. If enabled,
6001 all FIFOS and sockets in the file system will be removed
6002 when the specific socket unit is stopped.
6004 * Socket units gained a new Symlinks= setting. It takes a list
6005 of symlinks to create to file system sockets or FIFOs
6006 created by the specific Unix sockets. This is useful to
6007 manage symlinks to socket nodes with the same lifecycle as
6010 * The /dev/log socket and /dev/initctl FIFO have been moved to
6011 /run, and have been replaced by symlinks. This allows
6012 connecting to these facilities even if PrivateDevices=yes is
6013 used for a service (which makes /dev/log itself unavailable,
6014 but /run is left). This also has the benefit of ensuring
6015 that /dev only contains device nodes, directories and
6016 symlinks, and nothing else.
6018 * sd-daemon gained two new calls sd_pid_notify() and
6019 sd_pid_notifyf(). They are similar to sd_notify() and
6020 sd_notifyf(), but allow overriding of the source PID of
6021 notification messages if permissions permit this. This is
6022 useful to send notify messages on behalf of a different
6023 process (for example, the parent process). The
6024 systemd-notify tool has been updated to make use of this
6025 when sending messages (so that notification messages now
6026 originate from the shell script invoking systemd-notify and
6027 not the systemd-notify process itself. This should minimize
6028 a race where systemd fails to associate notification
6029 messages to services when the originating process already
6032 * A new "on-abnormal" setting for Restart= has been added. If
6033 set, it will result in automatic restarts on all "abnormal"
6034 reasons for a process to exit, which includes unclean
6035 signals, core dumps, timeouts and watchdog timeouts, but
6036 does not include clean and unclean exit codes or clean
6037 signals. Restart=on-abnormal is an alternative for
6038 Restart=on-failure for services that shall be able to
6039 terminate and avoid restarts on certain errors, by
6040 indicating so with an unclean exit code. Restart=on-failure
6041 or Restart=on-abnormal is now the recommended setting for
6042 all long-running services.
6044 * If the InaccessibleDirectories= service setting points to a
6045 mount point (or if there are any submounts contained within
6046 it), it is now attempted to completely unmount it, to make
6047 the file systems truly unavailable for the respective
6050 * The ReadOnlyDirectories= service setting and
6051 systemd-nspawn's --read-only parameter are now recursively
6052 applied to all submounts, too.
6054 * Mount units may now be created transiently via the bus APIs.
6056 * The support for SysV and LSB init scripts has been removed
6057 from the systemd daemon itself. Instead, it is now
6058 implemented as a generator that creates native systemd units
6059 from these scripts when needed. This enables us to remove a
6060 substantial amount of legacy code from PID 1, following the
6061 fact that many distributions only ship a very small number
6062 of LSB/SysV init scripts nowadays.
6064 * Privileged Xen (dom0) domains are not considered
6065 virtualization anymore by the virtualization detection
6066 logic. After all, they generally have unrestricted access to
6067 the hardware and usually are used to manage the unprivileged
6070 * systemd-tmpfiles gained a new "C" line type, for copying
6071 files or entire directories.
6073 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
6074 lines. So far, they have been non-globbing versions of the
6075 latter, and have thus been redundant. In future, it is
6076 recommended to only use "z". "m" has hence been removed
6077 from the documentation, even though it stays supported.
6079 * A tmpfiles snippet to recreate the most basic structure in
6080 /var has been added. This is enough to create the /var/run →
6081 /run symlink and create a couple of structural
6082 directories. This allows systems to boot up with an empty or
6083 volatile /var. Of course, while with this change, the core OS
6084 now is capable with dealing with a volatile /var, not all
6085 user services are ready for it. However, we hope that sooner
6086 or later, many service daemons will be changed upstream so
6087 that they are able to automatically create their necessary
6088 directories in /var at boot, should they be missing. This is
6089 the first step to allow state-less systems that only require
6090 the vendor image for /usr to boot.
6092 * systemd-nspawn has gained a new --tmpfs= switch to mount an
6093 empty tmpfs instance to a specific directory. This is
6094 particularly useful for making use of the automatic
6095 reconstruction of /var (see above), by passing --tmpfs=/var.
6097 * Access modes specified in tmpfiles snippets may now be
6098 prefixed with "~", which indicates that they shall be masked
6099 by whether the existing file or directory is currently
6100 writable, readable or executable at all. Also, if specified,
6101 the sgid/suid/sticky bits will be masked for all
6104 * A new passive target unit "network-pre.target" has been
6105 added which is useful for services that shall run before any
6106 network is configured, for example firewall scripts.
6108 * The "floppy" group that previously owned the /dev/fd*
6109 devices is no longer used. The "disk" group is now used
6110 instead. Distributions should probably deprecate usage of
6113 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
6114 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
6115 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
6116 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
6117 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
6118 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
6121 — Berlin, 2014-06-11
6125 * A new "systemd-timesyncd" daemon has been added for
6126 synchronizing the system clock across the network. It
6127 implements an SNTP client. In contrast to NTP
6128 implementations such as chrony or the NTP reference server,
6129 this only implements a client side, and does not bother with
6130 the full NTP complexity, focusing only on querying time from
6131 one remote server and synchronizing the local clock to
6132 it. Unless you intend to serve NTP to networked clients or
6133 want to connect to local hardware clocks, this simple NTP
6134 client should be more than appropriate for most
6135 installations. The daemon runs with minimal privileges, and
6136 has been hooked up with networkd to only operate when
6137 network connectivity is available. The daemon saves the
6138 current clock to disk every time a new NTP sync has been
6139 acquired, and uses this to possibly correct the system clock
6140 early at bootup, in order to accommodate for systems that
6141 lack an RTC such as the Raspberry Pi and embedded devices,
6142 and to make sure that time monotonically progresses on these
6143 systems, even if it is not always correct. To make use of
6144 this daemon, a new system user and group "systemd-timesync"
6145 needs to be created on installation of systemd.
6147 * The queue "seqnum" interface of libudev has been disabled, as
6148 it was generally incompatible with device namespacing as
6149 sequence numbers of devices go "missing" if the devices are
6150 part of a different namespace.
6152 * "systemctl list-timers" and "systemctl list-sockets" gained
6153 a --recursive switch for showing units of these types also
6154 for all local containers, similar in style to the already
6155 supported --recursive switch for "systemctl list-units".
6157 * A new RebootArgument= setting has been added for service
6158 units, which may be used to specify a kernel reboot argument
6159 to use when triggering reboots with StartLimitAction=.
6161 * A new FailureAction= setting has been added for service
6162 units which may be used to specify an operation to trigger
6163 when a service fails. This works similarly to
6164 StartLimitAction=, but unlike it, controls what is done
6165 immediately rather than only after several attempts to
6166 restart the service in question.
6168 * hostnamed got updated to also expose the kernel name,
6169 release, and version on the bus. This is useful for
6170 executing commands like hostnamectl with the -H switch.
6171 systemd-analyze makes use of this to properly display
6172 details when running non-locally.
6174 * The bootchart tool can now show cgroup information in the
6175 graphs it generates.
6177 * The CFS CPU quota cgroup attribute is now exposed for
6178 services. The new CPUQuota= switch has been added for this
6179 which takes a percentage value. Setting this will have the
6180 result that a service may never get more CPU time than the
6181 specified percentage, even if the machine is otherwise idle.
6183 * systemd-networkd learned IPIP and SIT tunnel support.
6185 * LSB init scripts exposing a dependency on $network will now
6186 get a dependency on network-online.target rather than simply
6187 network.target. This should bring LSB handling closer to
6188 what it was on SysV systems.
6190 * A new fsck.repair= kernel option has been added to control
6191 how fsck shall deal with unclean file systems at boot.
6193 * The (.ini) configuration file parser will now silently
6194 ignore sections whose name begins with "X-". This may be
6195 used to maintain application-specific extension sections in unit
6198 * machined gained a new API to query the IP addresses of
6199 registered containers. "machinectl status" has been updated
6200 to show these addresses in its output.
6202 * A new call sd_uid_get_display() has been added to the
6203 sd-login APIs for querying the "primary" session of a
6204 user. The "primary" session of the user is elected from the
6205 user's sessions and generally a graphical session is
6206 preferred over a text one.
6208 * A minimal systemd-resolved daemon has been added. It
6209 currently simply acts as a companion to systemd-networkd and
6210 manages resolv.conf based on per-interface DNS
6211 configuration, possibly supplied via DHCP. In the long run
6212 we hope to extend this into a local DNSSEC enabled DNS and
6215 * The systemd-networkd-wait-online tool is now enabled by
6216 default. It will delay network-online.target until a network
6217 connection has been configured. The tool primarily integrates
6218 with networkd, but will also make a best effort to make sense
6219 of network configuration performed in some other way.
6221 * Two new service options StartupCPUShares= and
6222 StartupBlockIOWeight= have been added that work similarly to
6223 CPUShares= and BlockIOWeight= however only apply during
6224 system startup. This is useful to prioritize certain services
6225 differently during bootup than during normal runtime.
6227 * hostnamed has been changed to prefer the statically
6228 configured hostname in /etc/hostname (unless set to
6229 'localhost' or empty) over any dynamic one supplied by
6230 dhcp. With this change, the rules for picking the hostname
6231 match more closely the rules of other configuration settings
6232 where the local administrator's configuration in /etc always
6233 overrides any other settings.
6235 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
6236 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
6237 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
6238 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
6239 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
6240 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
6241 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
6242 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
6243 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
6244 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
6245 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
6246 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
6247 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
6248 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
6249 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
6250 Lindskog, WaLyong Cho, Will Woods, Zbigniew
6253 — Beijing, 2014-05-28
6257 * When restoring the screen brightness at boot, stay away from
6258 the darkest setting or from the lowest 5% of the available
6259 range, depending on which is the larger value of both. This
6260 should effectively protect the user from rebooting into a
6261 black screen, should the brightness have been set to minimum
6264 * sd-login gained a new sd_machine_get_class() call to
6265 determine the class ("vm" or "container") of a machine
6266 registered with machined.
6268 * sd-login gained new calls
6269 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
6270 to query the identity of the peer of a local AF_UNIX
6271 connection. They operate similarly to their sd_pid_get_xyz()
6274 * PID 1 will now maintain a system-wide system state engine
6275 with the states "starting", "running", "degraded",
6276 "maintenance", "stopping". These states are bound to system
6277 startup, normal runtime, runtime with at least one failed
6278 service, rescue/emergency mode and system shutdown. This
6279 state is shown in the "systemctl status" output when no unit
6280 name is passed. It is useful to determine system state, in
6281 particularly when doing so for many systems or containers at
6284 * A new command "list-machines" has been added to "systemctl"
6285 that lists all local OS containers and shows their system
6286 state (see above), if systemd runs inside of them.
6288 * systemctl gained a new "-r" switch to recursively enumerate
6289 units on all local containers, when used with the
6290 "list-unit" command (which is the default one that is
6291 executed when no parameters are specified).
6293 * The GPT automatic partition discovery logic will now honour
6294 two GPT partition flags: one may be set on a partition to
6295 cause it to be mounted read-only, and the other may be set
6296 on a partition to ignore it during automatic discovery.
6298 * Two new GPT type UUIDs have been added for automatic root
6299 partition discovery, for 32-bit and 64-bit ARM. This is not
6300 particularly useful for discovering the root directory on
6301 these architectures during bare-metal boots (since UEFI is
6302 not common there), but still very useful to allow booting of
6303 ARM disk images in nspawn with the -i option.
6305 * MAC addresses of interfaces created with nspawn's
6306 --network-interface= switch will now be generated from the
6307 machine name, and thus be stable between multiple invocations
6310 * logind will now automatically remove all IPC objects owned
6311 by a user if she or he fully logs out. This makes sure that
6312 users who are logged out cannot continue to consume IPC
6313 resources. This covers SysV memory, semaphores and message
6314 queues as well as POSIX shared memory and message
6315 queues. Traditionally, SysV and POSIX IPC had no lifecycle
6316 limits. With this functionality, that is corrected. This may
6317 be turned off by using the RemoveIPC= switch of logind.conf.
6319 * The systemd-machine-id-setup and tmpfiles tools gained a
6320 --root= switch to operate on a specific root directory,
6323 * journald can now forward logged messages to the TTYs of all
6324 logged in users ("wall"). This is the default for all
6325 emergency messages now.
6327 * A new tool systemd-journal-remote has been added to stream
6328 journal log messages across the network.
6330 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
6331 controller trees are mounted into it. Note that the
6332 directories mounted beneath it are not read-only. This is a
6333 security measure and is particularly useful because glibc
6334 actually includes a search logic to pick any tmpfs it can
6335 find to implement shm_open() if /dev/shm is not available
6336 (which it might very well be in namespaced setups).
6338 * machinectl gained a new "poweroff" command to cleanly power
6339 down a local OS container.
6341 * The PrivateDevices= unit file setting will now also drop the
6342 CAP_MKNOD capability from the capability bound set, and
6343 imply DevicePolicy=closed.
6345 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
6346 comprehensively on all long-running systemd services where
6347 this is appropriate.
6349 * systemd-udevd will now run in a disassociated mount
6350 namespace. To mount directories from udev rules, make sure to
6351 pull in mount units via SYSTEMD_WANTS properties.
6353 * The kdbus support gained support for uploading policy into
6354 the kernel. sd-bus gained support for creating "monitoring"
6355 connections that can eavesdrop into all bus communication
6356 for debugging purposes.
6358 * Timestamps may now be specified in seconds since the UNIX
6359 epoch Jan 1st, 1970 by specifying "@" followed by the value
6362 * Native tcpwrap support in systemd has been removed. tcpwrap
6363 is old code, not really maintained anymore and has serious
6364 shortcomings, and better options such as firewalls
6365 exist. For setups that require tcpwrap usage, please
6366 consider invoking your socket-activated service via tcpd,
6367 like on traditional inetd.
6369 * A new system.conf configuration option
6370 DefaultTimerAccuracySec= has been added that controls the
6371 default AccuracySec= setting of .timer units.
6373 * Timer units gained a new WakeSystem= switch. If enabled,
6374 timers configured this way will cause the system to resume
6375 from system suspend (if the system supports that, which most
6378 * Timer units gained a new Persistent= switch. If enabled,
6379 timers configured this way will save to disk when they have
6380 been last triggered. This information is then used on next
6381 reboot to possible execute overdue timer events, that
6382 could not take place because the system was powered off.
6383 This enables simple anacron-like behaviour for timer units.
6385 * systemctl's "list-timers" will now also list the time a
6386 timer unit was last triggered in addition to the next time
6387 it will be triggered.
6389 * systemd-networkd will now assign predictable IPv4LL
6390 addresses to its local interfaces.
6392 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
6393 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
6394 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
6395 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
6396 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
6397 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
6398 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
6399 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
6402 — Berlin, 2014-03-25
6406 * A new unit file setting RestrictAddressFamilies= has been
6407 added to restrict which socket address families unit
6408 processes gain access to. This takes address family names
6409 like "AF_INET" or "AF_UNIX", and is useful to minimize the
6410 attack surface of services via exotic protocol stacks. This
6411 is built on seccomp system call filters.
6413 * Two new unit file settings RuntimeDirectory= and
6414 RuntimeDirectoryMode= have been added that may be used to
6415 manage a per-daemon runtime directories below /run. This is
6416 an alternative for setting up directory permissions with
6417 tmpfiles snippets, and has the advantage that the runtime
6418 directory's lifetime is bound to the daemon runtime and that
6419 the daemon starts up with an empty directory each time. This
6420 is particularly useful when writing services that drop
6421 privileges using the User= or Group= setting.
6423 * The DeviceAllow= unit setting now supports globbing for
6424 matching against device group names.
6426 * The systemd configuration file system.conf gained new
6427 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
6428 DefaultMemoryAccounting= to globally turn on/off accounting
6429 for specific resources (cgroups) for all units. These
6430 settings may still be overridden individually in each unit
6433 * systemd-gpt-auto-generator is now able to discover /srv and
6434 root partitions in addition to /home and swap partitions. It
6435 also supports LUKS-encrypted partitions now. With this in
6436 place, automatic discovery of partitions to mount following
6437 the Discoverable Partitions Specification
6438 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
6439 is now a lot more complete. This allows booting without
6440 /etc/fstab and without root= on the kernel command line on
6441 systems prepared appropriately.
6443 * systemd-nspawn gained a new --image= switch which allows
6444 booting up disk images and Linux installations on any block
6445 device that follow the Discoverable Partitions Specification
6446 (see above). This means that installations made with
6447 appropriately updated installers may now be started and
6448 deployed using container managers, completely
6449 unmodified. (We hope that libvirt-lxc will add support for
6450 this feature soon, too.)
6452 * systemd-nspawn gained a new --network-macvlan= setting to
6453 set up a private macvlan interface for the
6454 container. Similarly, systemd-networkd gained a new
6455 Kind=macvlan setting in .netdev files.
6457 * systemd-networkd now supports configuring local addresses
6460 * A new tool systemd-network-wait-online has been added to
6461 synchronously wait for network connectivity using
6464 * The sd-bus.h bus API gained a new sd_bus_track object for
6465 tracking the lifecycle of bus peers. Note that sd-bus.h is
6466 still not a public API though (unless you specify
6467 --enable-kdbus on the configure command line, which however
6468 voids your warranty and you get no API stability guarantee).
6470 * The $XDG_RUNTIME_DIR runtime directories for each user are
6471 now individual tmpfs instances, which has the benefit of
6472 introducing separate pools for each user, with individual
6473 size limits, and thus making sure that unprivileged clients
6474 can no longer negatively impact the system or other users by
6475 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
6476 RuntimeDirectorySize= has been introduced that allows
6477 controlling the default size limit for all users. It
6478 defaults to 10% of the available physical memory. This is no
6479 replacement for quotas on tmpfs though (which the kernel
6480 still does not support), as /dev/shm and /tmp are still
6481 shared resources used by both the system and unprivileged
6484 * logind will now automatically turn off automatic suspending
6485 on laptop lid close when more than one display is
6486 connected. This was previously expected to be implemented
6487 individually in desktop environments (such as GNOME),
6488 however has been added to logind now, in order to fix a
6489 boot-time race where a desktop environment might not have
6490 been started yet and thus not been able to take an inhibitor
6491 lock at the time where logind already suspends the system
6492 due to a closed lid.
6494 * logind will now wait at least 30s after each system
6495 suspend/resume cycle, and 3min after system boot before
6496 suspending the system due to a closed laptop lid. This
6497 should give USB docking stations and similar enough time to
6498 be probed and configured after system resume and boot in
6499 order to then act as suspend blocker.
6501 * systemd-run gained a new --property= setting which allows
6502 initialization of resource control properties (and others)
6503 for the created scope or service unit. Example: "systemd-run
6504 --property=BlockIOWeight=10 updatedb" may be used to run
6505 updatedb at a low block IO scheduling weight.
6507 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
6508 now also work in --scope mode.
6510 * When systemd is compiled with kdbus support, basic support
6511 for enforced policies is now in place. (Note that enabling
6512 kdbus still voids your warranty and no API compatibility
6515 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
6516 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
6517 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
6518 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
6519 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
6520 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
6521 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
6522 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
6523 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
6524 Zbigniew Jędrzejewski-Szmek
6526 — Berlin, 2014-03-12
6530 * systemd will now relabel /dev after loading the SMACK policy
6531 according to SMACK rules.
6533 * A new unit file option AppArmorProfile= has been added to
6534 set the AppArmor profile for the processes of a unit.
6536 * A new condition check ConditionArchitecture= has been added
6537 to conditionalize units based on the system architecture, as
6538 reported by uname()'s "machine" field.
6540 * systemd-networkd now supports matching on the system
6541 virtualization, architecture, kernel command line, host name
6544 * logind is now a lot more aggressive when suspending the
6545 machine due to a closed laptop lid. Instead of acting only
6546 on the lid close action, it will continuously watch the lid
6547 status and act on it. This is useful for laptops where the
6548 power button is on the outside of the chassis so that it can
6549 be reached without opening the lid (such as the Lenovo
6550 Yoga). On those machines, logind will now immediately
6551 re-suspend the machine if the power button has been
6552 accidentally pressed while the laptop was suspended and in a
6553 backpack or similar.
6555 * logind will now watch SW_DOCK switches and inhibit reaction
6556 to the lid switch if it is pressed. This means that logind
6557 will not suspend the machine anymore if the lid is closed
6558 and the system is docked, if the laptop supports SW_DOCK
6559 notifications via the input layer. Note that ACPI docking
6560 stations do not generate this currently. Also note that this
6561 logic is usually not fully sufficient and Desktop
6562 Environments should take a lid switch inhibitor lock when an
6563 external display is connected, as systemd will not watch
6566 * nspawn will now make use of the devices cgroup controller by
6567 default, and only permit creation of and access to the usual
6568 API device nodes like /dev/null or /dev/random, as well as
6569 access to (but not creation of) the pty devices.
6571 * We will now ship a default .network file for
6572 systemd-networkd that automatically configures DHCP for
6573 network interfaces created by nspawn's --network-veth or
6574 --network-bridge= switches.
6576 * systemd will now understand the usual M, K, G, T suffixes
6577 according to SI conventions (i.e. to the base 1000) when
6578 referring to throughput and hardware metrics. It will stay
6579 with IEC conventions (i.e. to the base 1024) for software
6580 metrics, according to what is customary according to
6581 Wikipedia. We explicitly document which base applies for
6582 each configuration option.
6584 * The DeviceAllow= setting in unit files now supports a syntax
6585 to whitelist an entire group of devices node majors at once,
6586 based on the /proc/devices listing. For example, with the
6587 string "char-pts", it is now possible to whitelist all
6588 current and future pseudo-TTYs at once.
6590 * sd-event learned a new "post" event source. Event sources of
6591 this type are triggered by the dispatching of any event
6592 source of a type that is not "post". This is useful for
6593 implementing clean-up and check event sources that are
6594 triggered by other work being done in the program.
6596 * systemd-networkd is no longer statically enabled, but uses
6597 the usual [Install] sections so that it can be
6598 enabled/disabled using systemctl. It still is enabled by
6601 * When creating a veth interface pair with systemd-nspawn, the
6602 host side will now be prefixed with "vb-" if
6603 --network-bridge= is used, and with "ve-" if --network-veth
6604 is used. This way, it is easy to distinguish these cases on
6605 the host, for example to apply different configuration to
6606 them with systemd-networkd.
6608 * The compatibility libraries for libsystemd-journal.so,
6609 libsystem-id128.so, libsystemd-login.so and
6610 libsystemd-daemon.so do not make use of IFUNC
6611 anymore. Instead, we now build libsystemd.so multiple times
6612 under these alternative names. This means that the footprint
6613 is drastically increased, but given that these are
6614 transitional compatibility libraries, this should not matter
6615 much. This change has been made necessary to support the ARM
6616 platform for these compatibility libraries, as the ARM
6617 toolchain is not really at the same level as the toolchain
6618 for other architectures like x86 and does not support
6619 IFUNC. Please make sure to use --enable-compat-libs only
6620 during a transitional period!
6622 * The .include syntax has been deprecated and is not documented
6623 anymore. Drop-in files in .d directories should be used instead.
6625 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
6626 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
6627 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
6628 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
6629 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
6630 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
6631 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
6632 Zbigniew Jędrzejewski-Szmek
6634 — Berlin, 2014-02-24
6638 * A new component "systemd-networkd" has been added that can
6639 be used to configure local network interfaces statically or
6640 via DHCP. It is capable of bringing up bridges, VLANs, and
6641 bonding. Currently, no hook-ups for interactive network
6642 configuration are provided. Use this for your initrd,
6643 container, embedded, or server setup if you need a simple,
6644 yet powerful, network configuration solution. This
6645 configuration subsystem is quite nifty, as it allows wildcard
6646 hotplug matching in interfaces. For example, with a single
6647 configuration snippet, you can configure that all Ethernet
6648 interfaces showing up are automatically added to a bridge,
6649 or similar. It supports link-sensing and more.
6651 * A new tool "systemd-socket-proxyd" has been added which can
6652 act as a bidirectional proxy for TCP sockets. This is
6653 useful for adding socket activation support to services that
6654 do not actually support socket activation, including virtual
6655 machines and the like.
6657 * Add a new tool to save/restore rfkill state on
6660 * Save/restore state of keyboard backlights in addition to
6661 display backlights on shutdown/boot.
6663 * udev learned a new SECLABEL{} construct to label device
6664 nodes with a specific security label when they appear. For
6665 now, only SECLABEL{selinux} is supported, but the syntax is
6666 prepared for additional security frameworks.
6668 * udev gained a new scheme to configure link-level attributes
6669 from files in /etc/systemd/network/*.link. These files can
6670 match against MAC address, device path, driver name and type,
6671 and will apply attributes like the naming policy, link speed,
6672 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
6673 address assignment policy (randomized, ...).
6675 * The configuration of network interface naming rules for
6676 "permanent interface names" has changed: a new NamePolicy=
6677 setting in the [Link] section of .link files determines the
6678 priority of possible naming schemes (onboard, slot, MAC,
6679 path). The default value of this setting is determined by
6680 /usr/lib/net/links/99-default.link. Old
6681 80-net-name-slot.rules udev configuration file has been
6682 removed, so local configuration overriding this file should
6683 be adapted to override 99-default.link instead.
6685 * When the User= switch is used in a unit file, also
6686 initialize $SHELL= based on the user database entry.
6688 * systemd no longer depends on libdbus. All communication is
6689 now done with sd-bus, systemd's low-level bus library
6692 * kdbus support has been added to PID 1 itself. When kdbus is
6693 enabled, this causes PID 1 to set up the system bus and
6694 enable support for a new ".busname" unit type that
6695 encapsulates bus name activation on kdbus. It works a little
6696 bit like ".socket" units, except for bus names. A new
6697 generator has been added that converts classic dbus1 service
6698 activation files automatically into native systemd .busname
6701 * sd-bus: add a light-weight vtable implementation that allows
6702 defining objects on the bus with a simple static const
6703 vtable array of its methods, signals and properties.
6705 * systemd will not generate or install static dbus
6706 introspection data anymore to /usr/share/dbus-1/interfaces,
6707 as the precise format of these files is unclear, and
6708 nothing makes use of it.
6710 * A proxy daemon is now provided to proxy clients connecting
6711 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
6712 compatibility with classic D-Bus.
6714 * A bus driver implementation has been added that supports the
6715 classic D-Bus bus driver calls on kdbus, also for
6716 compatibility purposes.
6718 * A new API "sd-event.h" has been added that implements a
6719 minimal event loop API built around epoll. It provides a
6720 couple of features that direct epoll usage is lacking:
6721 prioritization of events, scales to large numbers of timer
6722 events, per-event timer slack (accuracy), system-wide
6723 coalescing of timer events, exit handlers, watchdog
6724 supervision support using systemd's sd_notify() API, child
6727 * A new API "sd-rntl.h" has been added that provides an API
6728 around the route netlink interface of the kernel, similar in
6729 style to "sd-bus.h".
6731 * A new API "sd-dhcp-client.h" has been added that provides a
6732 small DHCPv4 client-side implementation. This is used by
6735 * There is a new kernel command line option
6736 "systemd.restore_state=0|1". When set to "0", none of the
6737 systemd tools will restore saved runtime state to hardware
6738 devices. More specifically, the rfkill and backlight states
6741 * The FsckPassNo= compatibility option in mount/service units
6742 has been removed. The fstab generator will now add the
6743 necessary dependencies automatically, and does not require
6744 PID1's support for that anymore.
6746 * journalctl gained a new switch, --list-boots, that lists
6747 recent boots with their times and boot IDs.
6749 * The various tools like systemctl, loginctl, timedatectl,
6750 busctl, systemd-run, ... have gained a new switch "-M" to
6751 connect to a specific, local OS container (as direct
6752 connection, without requiring SSH). This works on any
6753 container that is registered with machined, such as those
6754 created by libvirt-lxc or nspawn.
6756 * systemd-run and systemd-analyze also gained support for "-H"
6757 to connect to remote hosts via SSH. This is particularly
6758 useful for systemd-run because it enables queuing of jobs
6759 onto remote systems.
6761 * machinectl gained a new command "login" to open a getty
6762 login in any local container. This works with any container
6763 that is registered with machined (such as those created by
6764 libvirt-lxc or nspawn), and which runs systemd inside.
6766 * machinectl gained a new "reboot" command that may be used to
6767 trigger a reboot on a specific container that is registered
6768 with machined. This works on any container that runs an init
6769 system of some kind.
6771 * systemctl gained a new "list-timers" command to print a nice
6772 listing of installed timer units with the times they elapse
6775 * Alternative reboot() parameters may now be specified on the
6776 "systemctl reboot" command line and are passed to the
6777 reboot() system call.
6779 * systemctl gained a new --job-mode= switch to configure the
6780 mode to queue a job with. This is a more generic version of
6781 --fail, --irreversible, and --ignore-dependencies, which are
6782 still available but not advertised anymore.
6784 * /etc/systemd/system.conf gained new settings to configure
6785 various default timeouts of units, as well as the default
6786 start limit interval and burst. These may still be overridden
6789 * PID1 will now export on the bus profile data of the security
6790 policy upload process (such as the SELinux policy upload to
6793 * journald: when forwarding logs to the console, include
6794 timestamps (following the setting in
6795 /sys/module/printk/parameters/time).
6797 * OnCalendar= in timer units now understands the special
6798 strings "yearly" and "annually". (Both are equivalent)
6800 * The accuracy of timer units is now configurable with the new
6801 AccuracySec= setting. It defaults to 1min.
6803 * A new dependency type JoinsNamespaceOf= has been added that
6804 allows running two services within the same /tmp and network
6805 namespace, if PrivateNetwork= or PrivateTmp= are used.
6807 * A new command "cat" has been added to systemctl. It outputs
6808 the original unit file of a unit, and concatenates the
6809 contents of additional "drop-in" unit file snippets, so that
6810 the full configuration is shown.
6812 * systemctl now supports globbing on the various "list-xyz"
6813 commands, like "list-units" or "list-sockets", as well as on
6814 those commands which take multiple unit names.
6816 * journalctl's --unit= switch gained support for globbing.
6818 * All systemd daemons now make use of the watchdog logic so
6819 that systemd automatically notices when they hang.
6821 * If the $container_ttys environment variable is set,
6822 getty-generator will automatically spawn a getty for each
6823 listed tty. This is useful for container managers to request
6824 login gettys to be spawned on as many ttys as needed.
6826 * %h, %s, %U specifier support is not available anymore when
6827 used in unit files for PID 1. This is because NSS calls are
6828 not safe from PID 1. They stay available for --user
6829 instances of systemd, and as special case for the root user.
6831 * loginctl gained a new "--no-legend" switch to turn off output
6834 * The "sd-login.h" API gained three new calls:
6835 sd_session_is_remote(), sd_session_get_remote_user(),
6836 sd_session_get_remote_host() to query information about
6839 * The udev hardware database now also carries vendor/product
6840 information of SDIO devices.
6842 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
6843 determine whether watchdog notifications are requested by
6846 * Socket-activated per-connection services now include a
6847 short description of the connection parameters in the
6850 * tmpfiles gained a new "--boot" option. When this is not used,
6851 only lines where the command character is not suffixed with
6852 "!" are executed. When this option is specified, those
6853 options are executed too. This partitions tmpfiles
6854 directives into those that can be safely executed at any
6855 time, and those which should be run only at boot (for
6856 example, a line that creates /run/nologin).
6858 * A new API "sd-resolve.h" has been added which provides a simple
6859 asynchronous wrapper around glibc NSS host name resolution
6860 calls, such as getaddrinfo(). In contrast to glibc's
6861 getaddrinfo_a(), it does not use signals. In contrast to most
6862 other asynchronous name resolution libraries, this one does
6863 not reimplement DNS, but reuses NSS, so that alternate
6864 host name resolution systems continue to work, such as mDNS,
6865 LDAP, etc. This API is based on libasyncns, but it has been
6866 cleaned up for inclusion in systemd.
6868 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
6869 "sd-daemon.h" are no longer found in individual libraries
6870 libsystemd-journal.so, libsystemd-login.so,
6871 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
6872 merged them into a single library, libsystemd.so, which
6873 provides all symbols. The reason for this is cyclic
6874 dependencies, as these libraries tend to use each other's
6875 symbols. So far, we have managed to workaround that by linking
6876 a copy of a good part of our code into each of these
6877 libraries again and again, which, however, makes certain
6878 things hard to do, like sharing static variables. Also, it
6879 substantially increases footprint. With this change, there
6880 is only one library for the basic APIs systemd
6881 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
6882 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
6883 library as well, however are subject to the --enable-kdbus
6884 switch (see below). Note that "sd-dhcp-client.h" is not part
6885 of this library (this is because it only consumes, never
6886 provides, services of/to other APIs). To make the transition
6887 easy from the separate libraries to the unified one, we
6888 provide the --enable-compat-libs compile-time switch which
6889 will generate stub libraries that are compatible with the
6890 old ones but redirect all calls to the new one.
6892 * All of the kdbus logic and the new APIs "sd-bus.h",
6893 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
6894 and "sd-utf8.h" are compile-time optional via the
6895 "--enable-kdbus" switch, and they are not compiled in by
6896 default. To make use of kdbus, you have to explicitly enable
6897 the switch. Note however, that neither the kernel nor the
6898 userspace API for all of this is considered stable yet. We
6899 want to maintain the freedom to still change the APIs for
6900 now. By specifying this build-time switch, you acknowledge
6901 that you are aware of the instability of the current
6904 * Also, note that while kdbus is pretty much complete,
6905 it lacks one thing: proper policy support. This means you
6906 can build a fully working system with all features; however,
6907 it will be highly insecure. Policy support will be added in
6908 one of the next releases, at the same time that we will
6909 declare the APIs stable.
6911 * When the kernel command line argument "kdbus" is specified,
6912 systemd will automatically load the kdbus.ko kernel module. At
6913 this stage of development, it is only useful for testing kdbus
6914 and should not be used in production. Note: if "--enable-kdbus"
6915 is specified, and the kdbus.ko kernel module is available, and
6916 "kdbus" is added to the kernel command line, the entire system
6917 runs with kdbus instead of dbus-daemon, with the above mentioned
6918 problem of missing the system policy enforcement. Also a future
6919 version of kdbus.ko or a newer systemd will not be compatible with
6920 each other, and will unlikely be able to boot the machine if only
6921 one of them is updated.
6923 * systemctl gained a new "import-environment" command which
6924 uploads the caller's environment (or parts thereof) into the
6925 service manager so that it is inherited by services started
6926 by the manager. This is useful to upload variables like
6927 $DISPLAY into the user service manager.
6929 * A new PrivateDevices= switch has been added to service units
6930 which allows running a service with a namespaced /dev
6931 directory that does not contain any device nodes for
6932 physical devices. More specifically, it only includes devices
6933 such as /dev/null, /dev/urandom, and /dev/zero which are API
6936 * logind has been extended to support behaviour like VT
6937 switching on seats that do not support a VT. This makes
6938 multi-session available on seats that are not the first seat
6939 (seat0), and on systems where kernel support for VTs has
6940 been disabled at compile-time.
6942 * If a process holds a delay lock for system sleep or shutdown
6943 and fails to release it in time, we will now log its
6944 identity. This makes it easier to identify processes that
6945 cause slow suspends or power-offs.
6947 * When parsing /etc/crypttab, support for a new key-slot=
6948 option as supported by Debian is added. It allows indicating
6949 which LUKS slot to use on disk, speeding up key loading.
6951 * The sd_journal_sendv() API call has been checked and
6952 officially declared to be async-signal-safe so that it may
6953 be invoked from signal handlers for logging purposes.
6955 * Boot-time status output is now enabled automatically after a
6956 short timeout if boot does not progress, in order to give
6957 the user an indication what she or he is waiting for.
6959 * The boot-time output has been improved to show how much time
6960 remains until jobs expire.
6962 * The KillMode= switch in service units gained a new possible
6963 value "mixed". If set, and the unit is shut down, then the
6964 initial SIGTERM signal is sent only to the main daemon
6965 process, while the following SIGKILL signal is sent to
6966 all remaining processes of the service.
6968 * When a scope unit is registered, a new property "Controller"
6969 may be set. If set to a valid bus name, systemd will send a
6970 RequestStop() signal to this name when it would like to shut
6971 down the scope. This may be used to hook manager logic into
6972 the shutdown logic of scope units. Also, scope units may now
6973 be put in a special "abandoned" state, in which case the
6974 manager process which created them takes no further
6975 responsibilities for it.
6977 * When reading unit files, systemd will now verify
6978 the access mode of these files, and warn about certain
6979 suspicious combinations. This has been added to make it
6980 easier to track down packaging bugs where unit files are
6981 marked executable or world-writable.
6983 * systemd-nspawn gained a new "--setenv=" switch to set
6984 container-wide environment variables. The similar option in
6985 systemd-activate was renamed from "--environment=" to
6986 "--setenv=" for consistency.
6988 * systemd-nspawn has been updated to create a new kdbus domain
6989 for each container that is invoked, thus allowing each
6990 container to have its own set of system and user buses,
6991 independent of the host.
6993 * systemd-nspawn gained a new --drop-capability= switch to run
6994 the container with less capabilities than the default. Both
6995 --drop-capability= and --capability= now take the special
6996 string "all" for dropping or keeping all capabilities.
6998 * systemd-nspawn gained new switches for executing containers
6999 with specific SELinux labels set.
7001 * systemd-nspawn gained a new --quiet switch to not generate
7002 any additional output but the container's own console
7005 * systemd-nspawn gained a new --share-system switch to run a
7006 container without PID namespacing enabled.
7008 * systemd-nspawn gained a new --register= switch to control
7009 whether the container is registered with systemd-machined or
7010 not. This is useful for containers that do not run full
7011 OS images, but only specific apps.
7013 * systemd-nspawn gained a new --keep-unit which may be used
7014 when invoked as the only program from a service unit, and
7015 results in registration of the unit service itself in
7016 systemd-machined, instead of a newly opened scope unit.
7018 * systemd-nspawn gained a new --network-interface= switch for
7019 moving arbitrary interfaces to the container. The new
7020 --network-veth switch creates a virtual Ethernet connection
7021 between host and container. The new --network-bridge=
7022 switch then allows assigning the host side of this virtual
7023 Ethernet connection to a bridge device.
7025 * systemd-nspawn gained a new --personality= switch for
7026 setting the kernel personality for the container. This is
7027 useful when running a 32-bit container on a 64-bit host. A
7028 similar option Personality= is now also available for service
7031 * logind will now also track a "Desktop" identifier for each
7032 session which encodes the desktop environment of it. This is
7033 useful for desktop environments that want to identify
7034 multiple running sessions of itself easily.
7036 * A new SELinuxContext= setting for service units has been
7037 added that allows setting a specific SELinux execution
7038 context for a service.
7040 * Most systemd client tools will now honour $SYSTEMD_LESS for
7041 settings of the "less" pager. By default, these tools will
7042 override $LESS to allow certain operations to work, such as
7043 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
7044 influence this logic.
7046 * systemd's "seccomp" hook-up has been changed to make use of
7047 the libseccomp library instead of using its own
7048 implementation. This has benefits for portability among
7051 * For usage together with SystemCallFilter=, a new
7052 SystemCallErrorNumber= setting has been introduced that
7053 allows configuration of a system error number to be returned
7054 on filtered system calls, instead of immediately killing the
7055 process. Also, SystemCallArchitectures= has been added to
7056 limit access to system calls of a particular architecture
7057 (in order to turn off support for unused secondary
7058 architectures). There is also a global
7059 SystemCallArchitectures= setting in system.conf now to turn
7060 off support for non-native system calls system-wide.
7062 * systemd requires a kernel with a working name_to_handle_at(),
7063 please see the kernel config requirements in the README file.
7065 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
7066 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
7067 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
7068 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
7069 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
7070 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
7071 Elia Pinto, Florian Weimer, George McCollister, Goffredo
7072 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
7073 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
7074 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
7075 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
7076 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
7077 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
7078 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
7079 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
7080 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
7081 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
7082 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
7083 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
7084 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
7085 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
7086 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
7087 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
7088 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
7090 — Berlin, 2014-02-20
7094 * logind has gained support for facilitating privileged input
7095 and drm device access for unprivileged clients. This work is
7096 useful to allow Wayland display servers (and similar
7097 programs, such as kmscon) to run under the user's ID and
7098 access input and drm devices which are normally
7099 protected. When this is used (and the kernel is new enough)
7100 logind will "mute" IO on the file descriptors passed to
7101 Wayland as long as it is in the background and "unmute" it
7102 if it returns into the foreground. This allows secure
7103 session switching without allowing background sessions to
7104 eavesdrop on input and display data. This also introduces
7105 session switching support if VT support is turned off in the
7106 kernel, and on seats that are not seat0.
7108 * A new kernel command line option luks.options= is understood
7109 now which allows specifying LUKS options for usage for LUKS
7110 encrypted partitions specified with luks.uuid=.
7112 * tmpfiles.d(5) snippets may now use specifier expansion in
7113 path names. More specifically %m, %b, %H, %v, are now
7114 replaced by the local machine id, boot id, hostname, and
7115 kernel version number.
7117 * A new tmpfiles.d(5) command "m" has been introduced which
7118 may be used to change the owner/group/access mode of a file
7119 or directory if it exists, but do nothing if it does not.
7121 * This release removes high-level support for the
7122 MemorySoftLimit= cgroup setting. The underlying kernel
7123 cgroup attribute memory.soft_limit= is currently badly
7124 designed and likely to be removed from the kernel API in its
7125 current form, hence we should not expose it for now.
7127 * The memory.use_hierarchy cgroup attribute is now enabled for
7128 all cgroups systemd creates in the memory cgroup
7129 hierarchy. This option is likely to be come the built-in
7130 default in the kernel anyway, and the non-hierarchical mode
7131 never made much sense in the intrinsically hierarchical
7134 * A new field _SYSTEMD_SLICE= is logged along with all journal
7135 messages containing the slice a message was generated
7136 from. This is useful to allow easy per-customer filtering of
7137 logs among other things.
7139 * systemd-journald will no longer adjust the group of journal
7140 files it creates to the "systemd-journal" group. Instead we
7141 rely on the journal directory to be owned by the
7142 "systemd-journal" group, and its setgid bit set, so that the
7143 kernel file system layer will automatically enforce that
7144 journal files inherit this group assignment. The reason for
7145 this change is that we cannot allow NSS look-ups from
7146 journald which would be necessary to resolve
7147 "systemd-journal" to a numeric GID, because this might
7148 create deadlocks if NSS involves synchronous queries to
7149 other daemons (such as nscd, or sssd) which in turn are
7150 logging clients of journald and might block on it, which
7151 would then dead lock. A tmpfiles.d(5) snippet included in
7152 systemd will make sure the setgid bit and group are
7153 properly set on the journal directory if it exists on every
7154 boot. However, we recommend adjusting it manually after
7155 upgrades too (or from RPM scriptlets), so that the change is
7156 not delayed until next reboot.
7158 * Backlight and random seed files in /var/lib/ have moved into
7159 the /var/lib/systemd/ directory, in order to centralize all
7160 systemd generated files in one directory.
7162 * Boot time performance measurements (as displayed by
7163 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
7164 performance information if that's available to determine how
7165 much time BIOS and boot loader initialization required. With
7166 a sufficiently new BIOS you hence no longer need to boot
7167 with Gummiboot to get access to such information.
7169 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
7170 Cristian Rodríguez, Dave Reisner, David Herrmann, David
7171 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
7172 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
7173 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
7174 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
7175 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7177 — Berlin, 2013-10-02
7181 * The Restart= option for services now understands a new
7182 on-watchdog setting, which will restart the service
7183 automatically if the service stops sending out watchdog keep
7184 alive messages (as configured with WatchdogSec=).
7186 * The getty generator (which is responsible for bringing up a
7187 getty on configured serial consoles) will no longer only
7188 start a getty on the primary kernel console but on all
7189 others, too. This makes the order in which console= is
7190 specified on the kernel command line less important.
7192 * libsystemd-logind gained a new sd_session_get_vt() call to
7193 retrieve the VT number of a session.
7195 * If the option "tries=0" is set for an entry of /etc/crypttab
7196 its passphrase is queried indefinitely instead of any
7197 maximum number of tries.
7199 * If a service with a configure PID file terminates its PID
7200 file will now be removed automatically if it still exists
7201 afterwards. This should put an end to stale PID files.
7203 * systemd-run will now also take relative binary path names
7204 for execution and no longer insists on absolute paths.
7206 * InaccessibleDirectories= and ReadOnlyDirectories= now take
7207 paths that are optionally prefixed with "-" to indicate that
7208 it should not be considered a failure if they do not exist.
7210 * journalctl -o (and similar commands) now understands a new
7211 output mode "short-precise", it is similar to "short" but
7212 shows timestamps with usec accuracy.
7214 * The option "discard" (as known from Debian) is now
7215 synonymous to "allow-discards" in /etc/crypttab. In fact,
7216 "discard" is preferred now (since it is easier to remember
7219 * Some licensing clean-ups were made, so that more code is now
7220 LGPL-2.1 licensed than before.
7222 * A minimal tool to save/restore the display backlight
7223 brightness across reboots has been added. It will store the
7224 backlight setting as late as possible at shutdown, and
7225 restore it as early as possible during reboot.
7227 * A logic to automatically discover and enable home and swap
7228 partitions on GPT disks has been added. With this in place
7229 /etc/fstab becomes optional for many setups as systemd can
7230 discover certain partitions located on the root disk
7231 automatically. Home partitions are recognized under their
7232 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
7233 partitions are recognized under their GPT type ID
7234 0657fd6da4ab43c484e50933c84b4f4f.
7236 * systemd will no longer pass any environment from the kernel
7237 or initrd to system services. If you want to set an
7238 environment for all services, do so via the kernel command
7239 line systemd.setenv= assignment.
7241 * The systemd-sysctl tool no longer natively reads the file
7242 /etc/sysctl.conf. If desired, the file should be symlinked
7243 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
7244 legacy support by a symlink rather than built-in code, it
7245 also makes the otherwise hidden order of application of the
7246 different files visible. (Note that this partly reverts to a
7247 pre-198 application order of sysctl knobs!)
7249 * The "systemctl set-log-level" and "systemctl dump" commands
7250 have been moved to systemd-analyze.
7252 * systemd-run learned the new --remain-after-exit switch,
7253 which causes the scope unit not to be cleaned up
7254 automatically after the process terminated.
7256 * tmpfiles learned a new --exclude-prefix= switch to exclude
7257 certain paths from operation.
7259 * journald will now automatically flush all messages to disk
7260 as soon as a message at the log level CRIT, ALERT or EMERG
7263 Contributions from: Andrew Cook, Brandon Philips, Christian
7264 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
7265 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
7266 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
7267 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
7268 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
7269 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
7270 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
7271 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
7272 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
7273 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
7274 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
7275 William Giokas, Zbigniew Jędrzejewski-Szmek
7277 — Berlin, 2013-09-13
7281 * The documentation has been updated to cover the various new
7282 concepts introduced with 205.
7284 * Unit files now understand the new %v specifier which
7285 resolves to the kernel version string as returned by "uname
7288 * systemctl now supports filtering the unit list output by
7289 load state, active state and sub state, using the new
7292 * "systemctl status" will now show the results of the
7293 condition checks (like ConditionPathExists= and similar) of
7294 the last start attempts of the unit. They are also logged to
7297 * "journalctl -b" may now be used to look for boot output of a
7298 specific boot. Try "journalctl -b -1" for the previous boot,
7299 but the syntax is substantially more powerful.
7301 * "journalctl --show-cursor" has been added which prints the
7302 cursor string the last shown log line. This may then be used
7303 with the new "journalctl --after-cursor=" switch to continue
7304 browsing logs from that point on.
7306 * "journalctl --force" may now be used to force regeneration
7309 * Creation of "dead" device nodes has been moved from udev
7310 into kmod and tmpfiles. Previously, udev would read the kmod
7311 databases to pre-generate dead device nodes based on meta
7312 information contained in kernel modules, so that these would
7313 be auto-loaded on access rather then at boot. As this
7314 does not really have much to do with the exposing actual
7315 kernel devices to userspace this has always been slightly
7316 alien in the udev codebase. Following the new scheme kmod
7317 will now generate a runtime snippet for tmpfiles from the
7318 module meta information and it now is tmpfiles' job to the
7319 create the nodes. This also allows overriding access and
7320 other parameters for the nodes using the usual tmpfiles
7321 facilities. As side effect this allows us to remove the
7322 CAP_SYS_MKNOD capability bit from udevd entirely.
7324 * logind's device ACLs may now be applied to these "dead"
7325 devices nodes too, thus finally allowing managed access to
7326 devices such as /dev/snd/sequencer without loading the
7327 backing module right-away.
7329 * A new RPM macro has been added that may be used to apply
7330 tmpfiles configuration during package installation.
7332 * systemd-detect-virt and ConditionVirtualization= now can
7333 detect User-Mode-Linux machines (UML).
7335 * journald will now implicitly log the effective capabilities
7336 set of processes in the message metadata.
7338 * systemd-cryptsetup has gained support for TrueCrypt volumes.
7340 * The initrd interface has been simplified (more specifically,
7341 support for passing performance data via environment
7342 variables and fsck results via files in /run has been
7343 removed). These features were non-essential, and are
7344 nowadays available in a much nicer way by having systemd in
7345 the initrd serialize its state and have the hosts systemd
7346 deserialize it again.
7348 * The udev "keymap" data files and tools to apply keyboard
7349 specific mappings of scan to key codes, and force-release
7350 scan code lists have been entirely replaced by a udev
7351 "keyboard" builtin and a hwdb data file.
7353 * systemd will now honour the kernel's "quiet" command line
7354 argument also during late shutdown, resulting in a
7355 completely silent shutdown when used.
7357 * There's now an option to control the SO_REUSEPORT socket
7358 option in .socket units.
7360 * Instance units will now automatically get a per-template
7361 subslice of system.slice unless something else is explicitly
7362 configured. For example, instances of sshd@.service will now
7363 implicitly be placed in system-sshd.slice rather than
7364 system.slice as before.
7366 * Test coverage support may now be enabled at build time.
7368 Contributions from: Dave Reisner, Frederic Crozat, Harald
7369 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
7370 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
7371 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
7372 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
7373 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
7374 Giokas, Zbigniew Jędrzejewski-Szmek
7376 — Berlin, 2013-07-23
7380 * Two new unit types have been introduced:
7382 Scope units are very similar to service units, however, are
7383 created out of pre-existing processes — instead of PID 1
7384 forking off the processes. By using scope units it is
7385 possible for system services and applications to group their
7386 own child processes (worker processes) in a powerful way
7387 which then maybe used to organize them, or kill them
7388 together, or apply resource limits on them.
7390 Slice units may be used to partition system resources in an
7391 hierarchical fashion and then assign other units to them. By
7392 default there are now three slices: system.slice (for all
7393 system services), user.slice (for all user sessions),
7394 machine.slice (for VMs and containers).
7396 Slices and scopes have been introduced primarily in
7397 context of the work to move cgroup handling to a
7398 single-writer scheme, where only PID 1
7399 creates/removes/manages cgroups.
7401 * There's a new concept of "transient" units. In contrast to
7402 normal units these units are created via an API at runtime,
7403 not from configuration from disk. More specifically this
7404 means it is now possible to run arbitrary programs as
7405 independent services, with all execution parameters passed
7406 in via bus APIs rather than read from disk. Transient units
7407 make systemd substantially more dynamic then it ever was,
7408 and useful as a general batch manager.
7410 * logind has been updated to make use of scope and slice units
7411 for managing user sessions. As a user logs in he will get
7412 his own private slice unit, to which all sessions are added
7413 as scope units. We also added support for automatically
7414 adding an instance of user@.service for the user into the
7415 slice. Effectively logind will no longer create cgroup
7416 hierarchies on its own now, it will defer entirely to PID 1
7417 for this by means of scope, service and slice units. Since
7418 user sessions this way become entities managed by PID 1
7419 the output of "systemctl" is now a lot more comprehensive.
7421 * A new mini-daemon "systemd-machined" has been added which
7422 may be used by virtualization managers to register local
7423 VMs/containers. nspawn has been updated accordingly, and
7424 libvirt will be updated shortly. machined will collect a bit
7425 of meta information about the VMs/containers, and assign
7426 them their own scope unit (see above). The collected
7427 meta-data is then made available via the "machinectl" tool,
7428 and exposed in "ps" and similar tools. machined/machinectl
7429 is compile-time optional.
7431 * As discussed earlier, the low-level cgroup configuration
7432 options ControlGroup=, ControlGroupModify=,
7433 ControlGroupPersistent=, ControlGroupAttribute= have been
7434 removed. Please use high-level attribute settings instead as
7435 well as slice units.
7437 * A new bus call SetUnitProperties() has been added to alter
7438 various runtime parameters of a unit. This is primarily
7439 useful to alter cgroup parameters dynamically in a nice way,
7440 but will be extended later on to make more properties
7441 modifiable at runtime. systemctl gained a new set-properties
7442 command that wraps this call.
7444 * A new tool "systemd-run" has been added which can be used to
7445 run arbitrary command lines as transient services or scopes,
7446 while configuring a number of settings via the command
7447 line. This tool is currently very basic, however already
7448 very useful. We plan to extend this tool to even allow
7449 queuing of execution jobs with time triggers from the
7450 command line, similar in fashion to "at".
7452 * nspawn will now inform the user explicitly that kernels with
7453 audit enabled break containers, and suggest the user to turn
7456 * Support for detecting the IMA and AppArmor security
7457 frameworks with ConditionSecurity= has been added.
7459 * journalctl gained a new "-k" switch for showing only kernel
7460 messages, mimicking dmesg output; in addition to "--user"
7461 and "--system" switches for showing only user's own logs
7464 * systemd-delta can now show information about drop-in
7465 snippets extending unit files.
7467 * libsystemd-bus has been substantially updated but is still
7468 not available as public API.
7470 * systemd will now look for the "debug" argument on the kernel
7471 command line and enable debug logging, similar to what
7472 "systemd.log_level=debug" already did before.
7474 * "systemctl set-default", "systemctl get-default" has been
7475 added to configure the default.target symlink, which
7476 controls what to boot into by default.
7478 * "systemctl set-log-level" has been added as a convenient
7479 way to raise and lower systemd logging threshold.
7481 * "systemd-analyze plot" will now show the time the various
7482 generators needed for execution, as well as information
7483 about the unit file loading.
7485 * libsystemd-journal gained a new sd_journal_open_files() call
7486 for opening specific journal files. journactl also gained a
7487 new switch to expose this new functionality. Previously we
7488 only supported opening all files from a directory, or all
7489 files from the system, as opening individual files only is
7490 racy due to journal file rotation.
7492 * systemd gained the new DefaultEnvironment= setting in
7493 /etc/systemd/system.conf to set environment variables for
7496 * If a privileged process logs a journal message with the
7497 OBJECT_PID= field set, then journald will automatically
7498 augment this with additional OBJECT_UID=, OBJECT_GID=,
7499 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
7500 system services want to log events about specific client
7501 processes. journactl/systemctl has been updated to make use
7502 of this information if all log messages regarding a specific
7505 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
7506 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
7507 Reisner, David Coppa, David King, David Strauss, Eelco
7508 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
7509 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
7510 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
7511 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
7512 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
7513 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
7514 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
7515 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
7516 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
7517 Łukasz Stelmach, 장동준
7521 * The Python bindings gained some minimal support for the APIs
7522 exposed by libsystemd-logind.
7524 * ConditionSecurity= gained support for detecting SMACK. Since
7525 this condition already supports SELinux and AppArmor we only
7526 miss IMA for this. Patches welcome!
7528 Contributions from: Karol Lewandowski, Lennart Poettering,
7529 Zbigniew Jędrzejewski-Szmek
7533 * systemd-nspawn will now create /etc/resolv.conf if
7534 necessary, before bind-mounting the host's file onto it.
7536 * systemd-nspawn will now store meta information about a
7537 container on the container's cgroup as extended attribute
7538 fields, including the root directory.
7540 * The cgroup hierarchy has been reworked in many ways. All
7541 objects any of the components systemd creates in the cgroup
7542 tree are now suffixed. More specifically, user sessions are
7543 now placed in cgroups suffixed with ".session", users in
7544 cgroups suffixed with ".user", and nspawn containers in
7545 cgroups suffixed with ".nspawn". Furthermore, all cgroup
7546 names are now escaped in a simple scheme to avoid collision
7547 of userspace object names with kernel filenames. This work
7548 is preparation for making these objects relocatable in the
7549 cgroup tree, in order to allow easy resource partitioning of
7550 these objects without causing naming conflicts.
7552 * systemctl list-dependencies gained the new switches
7553 --plain, --reverse, --after and --before.
7555 * systemd-inhibit now shows the process name of processes that
7556 have taken an inhibitor lock.
7558 * nss-myhostname will now also resolve "localhost"
7559 implicitly. This makes /etc/hosts an optional file and
7560 nicely handles that on IPv6 ::1 maps to both "localhost" and
7563 * libsystemd-logind.so gained a new call
7564 sd_get_machine_names() to enumerate running containers and
7565 VMs (currently only supported by very new libvirt and
7566 nspawn). sd_login_monitor can now be used to watch
7567 VMs/containers coming and going.
7569 * .include is not allowed recursively anymore, and only in
7570 unit files. Usually it is better to use drop-in snippets in
7571 .d/*.conf anyway, as introduced with systemd 198.
7573 * systemd-analyze gained a new "critical-chain" command that
7574 determines the slowest chain of units run during system
7575 boot-up. It is very useful for tracking down where
7576 optimizing boot time is the most beneficial.
7578 * systemd will no longer allow manipulating service paths in
7579 the name=systemd:/system cgroup tree using ControlGroup= in
7580 units. (But is still fine with it in all other dirs.)
7582 * There's a new systemd-nspawn@.service service file that may
7583 be used to easily run nspawn containers as system
7584 services. With the container's root directory in
7585 /var/lib/container/foobar it is now sufficient to run
7586 "systemctl start systemd-nspawn@foobar.service" to boot it.
7588 * systemd-cgls gained a new parameter "--machine" to list only
7589 the processes within a certain container.
7591 * ConditionSecurity= now can check for "apparmor". We still
7592 are lacking checks for SMACK and IMA for this condition
7593 check though. Patches welcome!
7595 * A new configuration file /etc/systemd/sleep.conf has been
7596 added that may be used to configure which kernel operation
7597 systemd is supposed to execute when "suspend", "hibernate"
7598 or "hybrid-sleep" is requested. This makes the new kernel
7599 "freeze" state accessible to the user.
7601 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
7602 the passed argument if applicable.
7604 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
7605 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
7606 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
7607 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
7608 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
7609 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
7610 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
7615 * The output of 'systemctl list-jobs' got some polishing. The
7616 '--type=' argument may now be passed more than once. A new
7617 command 'systemctl list-sockets' has been added which shows
7618 a list of kernel sockets systemd is listening on with the
7619 socket units they belong to, plus the units these socket
7622 * The experimental libsystemd-bus library got substantial
7623 updates to work in conjunction with the (also experimental)
7624 kdbus kernel project. It works well enough to exchange
7625 messages with some sophistication. Note that kdbus is not
7626 ready yet, and the library is mostly an elaborate test case
7627 for now, and not installable.
7629 * systemd gained a new unit 'systemd-static-nodes.service'
7630 that generates static device nodes earlier during boot, and
7631 can run in conjunction with udev.
7633 * libsystemd-login gained a new call sd_pid_get_user_unit()
7634 to retrieve the user systemd unit a process is running
7635 in. This is useful for systems where systemd is used as
7638 * systemd-nspawn now places all containers in the new /machine
7639 top-level cgroup directory in the name=systemd
7640 hierarchy. libvirt will soon do the same, so that we get a
7641 uniform separation of /system, /user and /machine for system
7642 services, user processes and containers/virtual
7643 machines. This new cgroup hierarchy is also useful to stick
7644 stable names to specific container instances, which can be
7645 recognized later this way (this name may be controlled
7646 via systemd-nspawn's new -M switch). libsystemd-login also
7647 gained a new call sd_pid_get_machine_name() to retrieve the
7648 name of the container/VM a specific process belongs to.
7650 * bootchart can now store its data in the journal.
7652 * libsystemd-journal gained a new call
7653 sd_journal_add_conjunction() for AND expressions to the
7654 matching logic. This can be used to express more complex
7655 logical expressions.
7657 * journactl can now take multiple --unit= and --user-unit=
7660 * The cryptsetup logic now understands the "luks.key=" kernel
7661 command line switch for specifying a file to read the
7662 decryption key from. Also, if a configured key file is not
7663 found the tool will now automatically fall back to prompting
7666 * Python systemd.journal module was updated to wrap recently
7667 added functions from libsystemd-journal. The interface was
7668 changed to bring the low level interface in s.j._Reader
7669 closer to the C API, and the high level interface in
7670 s.j.Reader was updated to wrap and convert all data about
7673 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
7674 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
7675 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
7676 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
7677 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
7678 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7682 * journalctl --update-catalog now understands a new --root=
7683 option to operate on catalogs found in a different root
7686 * During shutdown after systemd has terminated all running
7687 services a final killing loop kills all remaining left-over
7688 processes. We will now print the name of these processes
7689 when we send SIGKILL to them, since this usually indicates a
7692 * If /etc/crypttab refers to password files stored on
7693 configured mount points automatic dependencies will now be
7694 generated to ensure the specific mount is established first
7695 before the key file is attempted to be read.
7697 * 'systemctl status' will now show information about the
7698 network sockets a socket unit is listening on.
7700 * 'systemctl status' will also shown information about any
7701 drop-in configuration file for units. (Drop-In configuration
7702 files in this context are files such as
7703 /etc/systemd/systemd/foobar.service.d/*.conf)
7705 * systemd-cgtop now optionally shows summed up CPU times of
7706 cgroups. Press '%' while running cgtop to switch between
7707 percentage and absolute mode. This is useful to determine
7708 which cgroups use up the most CPU time over the entire
7709 runtime of the system. systemd-cgtop has also been updated
7710 to be 'pipeable' for processing with further shell tools.
7712 * 'hostnamectl set-hostname' will now allow setting of FQDN
7715 * The formatting and parsing of time span values has been
7716 changed. The parser now understands fractional expressions
7717 such as "5.5h". The formatter will now output fractional
7718 expressions for all time spans under 1min, i.e. "5.123456s"
7719 rather than "5s 123ms 456us". For time spans under 1s
7720 millisecond values are shown, for those under 1ms
7721 microsecond values are shown. This should greatly improve
7722 all time-related output of systemd.
7724 * libsystemd-login and libsystemd-journal gained new
7725 functions for querying the poll() events mask and poll()
7726 timeout value for integration into arbitrary event
7729 * localectl gained the ability to list available X11 keymaps
7730 (models, layouts, variants, options).
7732 * 'systemd-analyze dot' gained the ability to filter for
7733 specific units via shell-style globs, to create smaller,
7734 more useful graphs. I.e. it is now possible to create simple
7735 graphs of all the dependencies between only target units, or
7736 of all units that Avahi has dependencies with.
7738 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
7739 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
7740 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
7741 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
7742 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
7743 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
7744 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
7748 * The boot-time readahead implementation for rotating media
7749 will now read the read-ahead data in multiple passes which
7750 consist of all read requests made in equidistant time
7751 intervals. This means instead of strictly reading read-ahead
7752 data in its physical order on disk we now try to find a
7753 middle ground between physical and access time order.
7755 * /etc/os-release files gained a new BUILD_ID= field for usage
7756 on operating systems that provide continuous builds of OS
7759 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
7760 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
7761 William Douglas, Zbigniew Jędrzejewski-Szmek
7765 * systemd-python gained an API exposing libsystemd-daemon.
7767 * The SMACK setup logic gained support for uploading CIPSO
7770 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
7771 ReadOnlyDirectories= and InaccessibleDirectories= has
7772 changed. The private /tmp and /var/tmp directories are now
7773 shared by all processes of a service (which means
7774 ExecStartPre= may now leave data in /tmp that ExecStart= of
7775 the same service can still access). When a service is
7776 stopped its temporary directories are immediately deleted
7777 (normal clean-up with tmpfiles is still done in addition to
7780 * By default, systemd will now set a couple of sysctl
7781 variables in the kernel: the safe sysrq options are turned
7782 on, IP route verification is turned on, and source routing
7783 disabled. The recently added hardlink and softlink
7784 protection of the kernel is turned on. These settings should
7785 be reasonably safe, and good defaults for all new systems.
7787 * The predictable network naming logic may now be turned off
7788 with a new kernel command line switch: net.ifnames=0.
7790 * A new libsystemd-bus module has been added that implements a
7791 pretty complete D-Bus client library. For details see:
7793 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
7795 * journald will now explicitly flush the journal files to disk
7796 at the latest 5min after each write. The file will then also
7797 be marked offline until the next write. This should increase
7798 reliability in case of a crash. The synchronization delay
7799 can be configured via SyncIntervalSec= in journald.conf.
7801 * There's a new remote-fs-setup.target unit that can be used
7802 to pull in specific services when at least one remote file
7803 system is to be mounted.
7805 * There are new targets timers.target and paths.target as
7806 canonical targets to pull user timer and path units in
7807 from. This complements sockets.target with a similar
7808 purpose for socket units.
7810 * libudev gained a new call udev_device_set_attribute_value()
7811 to set sysfs attributes of a device.
7813 * The udev daemon now sets the default number of worker
7814 processes executed in parallel based on the number of available
7815 CPUs instead of the amount of available RAM. This is supposed
7816 to provide a more reliable default and limit a too aggressive
7817 parallelism for setups with 1000s of devices connected.
7819 Contributions from: Auke Kok, Colin Walters, Cristian
7820 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
7821 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
7822 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
7823 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
7824 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
7825 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
7826 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
7827 Zbigniew Jędrzejewski-Szmek
7831 * Configuration of unit files may now be extended via drop-in
7832 files without having to edit/override the unit files
7833 themselves. More specifically, if the administrator wants to
7834 change one value for a service file foobar.service he can
7835 now do so by dropping in a configuration snippet into
7836 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
7837 will load all these snippets and apply them on top of the
7838 main unit configuration file, possibly extending or
7839 overriding its settings. Using these drop-in snippets is
7840 generally nicer than the two earlier options for changing
7841 unit files locally: copying the files from
7842 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
7843 them there; or creating a new file in /etc/systemd/system/
7844 that incorporates the original one via ".include". Drop-in
7845 snippets into these .d/ directories can be placed in any
7846 directory systemd looks for units in, and the usual
7847 overriding semantics between /usr/lib, /etc and /run apply
7850 * Most unit file settings which take lists of items can now be
7851 reset by assigning the empty string to them. For example,
7852 normally, settings such as Environment=FOO=BAR append a new
7853 environment variable assignment to the environment block,
7854 each time they are used. By assigning Environment= the empty
7855 string the environment block can be reset to empty. This is
7856 particularly useful with the .d/*.conf drop-in snippets
7857 mentioned above, since this adds the ability to reset list
7858 settings from vendor unit files via these drop-ins.
7860 * systemctl gained a new "list-dependencies" command for
7861 listing the dependencies of a unit recursively.
7863 * Inhibitors are now honored and listed by "systemctl
7864 suspend", "systemctl poweroff" (and similar) too, not only
7865 GNOME. These commands will also list active sessions by
7868 * Resource limits (as exposed by the various control group
7869 controllers) can now be controlled dynamically at runtime
7870 for all units. More specifically, you can now use a command
7871 like "systemctl set-cgroup-attr foobar.service cpu.shares
7872 2000" to alter the CPU shares a specific service gets. These
7873 settings are stored persistently on disk, and thus allow the
7874 administrator to easily adjust the resource usage of
7875 services with a few simple commands. This dynamic resource
7876 management logic is also available to other programs via the
7877 bus. Almost any kernel cgroup attribute and controller is
7880 * systemd-vconsole-setup will now copy all font settings to
7881 all allocated VTs, where it previously applied them only to
7884 * libsystemd-login gained the new sd_session_get_tty() API
7887 * This release drops support for a few legacy or
7888 distribution-specific LSB facility names when parsing init
7889 scripts: $x-display-manager, $mail-transfer-agent,
7890 $mail-transport-agent, $mail-transfer-agent, $smtp,
7891 $null. Also, the mail-transfer-agent.target unit backing
7892 this has been removed. Distributions which want to retain
7893 compatibility with this should carry the burden for
7894 supporting this themselves and patch support for these back
7895 in, if they really need to. Also, the facilities $syslog and
7896 $local_fs are now ignored, since systemd does not support
7897 early-boot LSB init scripts anymore, and these facilities
7898 are implied anyway for normal services. syslog.target has
7901 * There are new bus calls on PID1's Manager object for
7902 cancelling jobs, and removing snapshot units. Previously,
7903 both calls were only available on the Job and Snapshot
7906 * systemd-journal-gatewayd gained SSL support.
7908 * The various "environment" files, such as /etc/locale.conf
7909 now support continuation lines with a backslash ("\") as
7910 last character in the line, similarly in style (but different)
7911 to how this is supported in shells.
7913 * For normal user processes the _SYSTEMD_USER_UNIT= field is
7914 now implicitly appended to every log entry logged. systemctl
7915 has been updated to filter by this field when operating on a
7916 user systemd instance.
7918 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
7919 CAP_AUDIT_CONTROL capabilities to the capabilities set for
7920 the container. This makes it easier to boot unmodified
7921 Fedora systems in a container, which however still requires
7922 audit=0 to be passed on the kernel command line. Auditing in
7923 kernel and userspace is unfortunately still too broken in
7924 context of containers, hence we recommend compiling it out
7925 of the kernel or using audit=0. Hopefully this will be fixed
7926 one day for good in the kernel.
7928 * nspawn gained the new --bind= and --bind-ro= parameters to
7929 bind mount specific directories from the host into the
7932 * nspawn will now mount its own devpts file system instance
7933 into the container, in order not to leak pty devices from
7934 the host into the container.
7936 * systemd will now read the firmware boot time performance
7937 information from the EFI variables, if the used boot loader
7938 supports this, and takes it into account for boot performance
7939 analysis via "systemd-analyze". This is currently supported
7940 only in conjunction with Gummiboot, but could be supported
7941 by other boot loaders too. For details see:
7943 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
7945 * A new generator has been added that automatically mounts the
7946 EFI System Partition (ESP) to /boot, if that directory
7947 exists, is empty, and no other file system has been
7948 configured to be mounted there.
7950 * logind will now send out PrepareForSleep(false) out
7951 unconditionally, after coming back from suspend. This may be
7952 used by applications as asynchronous notification for
7953 system resume events.
7955 * "systemctl unlock-sessions" has been added, that allows
7956 unlocking the screens of all user sessions at once, similar
7957 to how "systemctl lock-sessions" already locked all users
7958 sessions. This is backed by a new D-Bus call UnlockSessions().
7960 * "loginctl seat-status" will now show the master device of a
7961 seat. (i.e. the device of a seat that needs to be around for
7962 the seat to be considered available, usually the graphics
7965 * tmpfiles gained a new "X" line type, that allows
7966 configuration of files and directories (with wildcards) that
7967 shall be excluded from automatic cleanup ("aging").
7969 * udev default rules set the device node permissions now only
7970 at "add" events, and do not change them any longer with a
7971 later "change" event.
7973 * The log messages for lid events and power/sleep keypresses
7974 now carry a message ID.
7976 * We now have a substantially larger unit test suite, but this
7977 continues to be work in progress.
7979 * udevadm hwdb gained a new --root= parameter to change the
7980 root directory to operate relative to.
7982 * logind will now issue a background sync() request to the kernel
7983 early at shutdown, so that dirty buffers are flushed to disk early
7984 instead of at the last moment, in order to optimize shutdown
7987 * A new bootctl tool has been added that is an interface for
7988 certain boot loader operations. This is currently a preview
7989 and is likely to be extended into a small mechanism daemon
7990 like timedated, localed, hostnamed, and can be used by
7991 graphical UIs to enumerate available boot options, and
7992 request boot into firmware operations.
7994 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
7995 the rest of the package. It also has been updated to work
7996 correctly in initrds.
7998 * polkit previously has been runtime optional, and is now also
7999 compile time optional via a configure switch.
8001 * systemd-analyze has been reimplemented in C. Also "systemctl
8002 dot" has moved into systemd-analyze.
8004 * "systemctl status" with no further parameters will now print
8005 the status of all active or failed units.
8007 * Operations such as "systemctl start" can now be executed
8008 with a new mode "--irreversible" which may be used to queue
8009 operations that cannot accidentally be reversed by a later
8010 job queuing. This is by default used to make shutdown
8011 requests more robust.
8013 * The Python API of systemd now gained a new module for
8014 reading journal files.
8016 * A new tool kernel-install has been added that can install
8017 kernel images according to the Boot Loader Specification:
8019 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
8021 * Boot time console output has been improved to provide
8022 animated boot time output for hanging jobs.
8024 * A new tool systemd-activate has been added which can be used
8025 to test socket activation with, directly from the command
8026 line. This should make it much easier to test and debug
8027 socket activation in daemons.
8029 * journalctl gained a new "--reverse" (or -r) option to show
8030 journal output in reverse order (i.e. newest line first).
8032 * journalctl gained a new "--pager-end" (or -e) option to jump
8033 to immediately jump to the end of the journal in the
8034 pager. This is only supported in conjunction with "less".
8036 * journalctl gained a new "--user-unit=" option, that works
8037 similarly to "--unit=" but filters for user units rather than
8040 * A number of unit files to ease adoption of systemd in
8041 initrds has been added. This moves some minimal logic from
8042 the various initrd implementations into systemd proper.
8044 * The journal files are now owned by a new group
8045 "systemd-journal", which exists specifically to allow access
8046 to the journal, and nothing else. Previously, we used the
8047 "adm" group for that, which however possibly covers more
8048 than just journal/log file access. This new group is now
8049 already used by systemd-journal-gatewayd to ensure this
8050 daemon gets access to the journal files and as little else
8051 as possible. Note that "make install" will also set FS ACLs
8052 up for /var/log/journal to give "adm" and "wheel" read
8053 access to it, in addition to "systemd-journal" which owns
8054 the journal files. We recommend that packaging scripts also
8055 add read access to "adm" + "wheel" to /var/log/journal, and
8056 all existing/future journal files. To normal users and
8057 administrators little changes, however packagers need to
8058 ensure to create the "systemd-journal" system group at
8059 package installation time.
8061 * The systemd-journal-gatewayd now runs as unprivileged user
8062 systemd-journal-gateway:systemd-journal-gateway. Packaging
8063 scripts need to create these system user/group at
8066 * timedated now exposes a new boolean property CanNTP that
8067 indicates whether a local NTP service is available or not.
8069 * systemd-detect-virt will now also detect xen PVs
8071 * The pstore file system is now mounted by default, if it is
8074 * In addition to the SELinux and IMA policies we will now also
8075 load SMACK policies at early boot.
8077 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
8078 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
8079 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
8080 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
8081 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
8082 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
8083 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
8084 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
8085 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
8086 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
8087 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
8088 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
8089 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
8090 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
8094 * Timer units now support calendar time events in addition to
8095 monotonic time events. That means you can now trigger a unit
8096 based on a calendar time specification such as "Thu,Fri
8097 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
8098 or fifth day of any month of the year 2013, given that it is
8099 a thursday or friday. This brings timer event support
8100 considerably closer to cron's capabilities. For details on
8101 the supported calendar time specification language see
8104 * udev now supports a number of different naming policies for
8105 network interfaces for predictable names, and a combination
8106 of these policies is now the default. Please see this wiki
8107 document for details:
8109 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
8111 * Auke Kok's bootchart implementation has been added to the
8112 systemd tree. It is an optional component that can graph the
8113 boot in quite some detail. It is one of the best bootchart
8114 implementations around and minimal in its code and
8117 * nss-myhostname has been integrated into the systemd source
8118 tree. nss-myhostname guarantees that the local hostname
8119 always stays resolvable via NSS. It has been a weak
8120 requirement of systemd-hostnamed since a long time, and
8121 since its code is actually trivial we decided to just
8122 include it in systemd's source tree. It can be turned off
8123 with a configure switch.
8125 * The read-ahead logic is now capable of properly detecting
8126 whether a btrfs file system is on SSD or rotating media, in
8127 order to optimize the read-ahead scheme. Previously, it was
8128 only capable of detecting this on traditional file systems
8131 * In udev, additional device properties are now read from the
8132 IAB in addition to the OUI database. Also, Bluetooth company
8133 identities are attached to the devices as well.
8135 * In service files %U may be used as specifier that is
8136 replaced by the configured user name of the service.
8138 * nspawn may now be invoked without a controlling TTY. This
8139 makes it suitable for invocation as its own service. This
8140 may be used to set up a simple containerized server system
8141 using only core OS tools.
8143 * systemd and nspawn can now accept socket file descriptors
8144 when they are started for socket activation. This enables
8145 implementation of socket activated nspawn
8146 containers. i.e. think about autospawning an entire OS image
8147 when the first SSH or HTTP connection is received. We expect
8148 that similar functionality will also be added to libvirt-lxc
8151 * journalctl will now suppress ANSI color codes when
8152 presenting log data.
8154 * systemctl will no longer show control group information for
8155 a unit if the control group is empty anyway.
8157 * logind can now automatically suspend/hibernate/shutdown the
8160 * /etc/machine-info and hostnamed now also expose the chassis
8161 type of the system. This can be used to determine whether
8162 the local system is a laptop, desktop, handset or
8163 tablet. This information may either be configured by the
8164 user/vendor or is automatically determined from ACPI and DMI
8165 information if possible.
8167 * A number of polkit actions are now bound together with "imply"
8168 rules. This should simplify creating UIs because many actions
8169 will now authenticate similar ones as well.
8171 * Unit files learnt a new condition ConditionACPower= which
8172 may be used to conditionalize a unit depending on whether an
8173 AC power source is connected or not, of whether the system
8174 is running on battery power.
8176 * systemctl gained a new "is-failed" verb that may be used in
8177 shell scripts and suchlike to check whether a specific unit
8178 is in the "failed" state.
8180 * The EnvironmentFile= setting in unit files now supports file
8181 globbing, and can hence be used to easily read a number of
8182 environment files at once.
8184 * systemd will no longer detect and recognize specific
8185 distributions. All distribution-specific #ifdeffery has been
8186 removed, systemd is now fully generic and
8187 distribution-agnostic. Effectively, not too much is lost as
8188 a lot of the code is still accessible via explicit configure
8189 switches. However, support for some distribution specific
8190 legacy configuration file formats has been dropped. We
8191 recommend distributions to simply adopt the configuration
8192 files everybody else uses now and convert the old
8193 configuration from packaging scripts. Most distributions
8194 already did that. If that's not possible or desirable,
8195 distributions are welcome to forward port the specific
8196 pieces of code locally from the git history.
8198 * When logging a message about a unit systemd will now always
8199 log the unit name in the message meta data.
8201 * localectl will now also discover system locale data that is
8202 not stored in locale archives, but directly unpacked.
8204 * logind will no longer unconditionally use framebuffer
8205 devices as seat masters, i.e. as devices that are required
8206 to be existing before a seat is considered preset. Instead,
8207 it will now look for all devices that are tagged as
8208 "seat-master" in udev. By default, framebuffer devices will
8209 be marked as such, but depending on local systems, other
8210 devices might be marked as well. This may be used to
8211 integrate graphics cards using closed source drivers (such
8212 as NVidia ones) more nicely into logind. Note however, that
8213 we recommend using the open source NVidia drivers instead,
8214 and no udev rules for the closed-source drivers will be
8215 shipped from us upstream.
8217 Contributions from: Adam Williamson, Alessandro Crismani, Auke
8218 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
8219 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
8220 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
8221 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
8222 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
8223 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
8224 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
8225 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
8226 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
8227 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
8228 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
8233 * udev gained support for loading additional device properties
8234 from an indexed database that is keyed by vendor/product IDs
8235 and similar device identifiers. For the beginning this
8236 "hwdb" is populated with data from the well-known PCI and
8237 USB database, but also includes PNP, ACPI and OID data. In
8238 the longer run this indexed database shall grow into
8239 becoming the one central database for non-essential
8240 userspace device metadata. Previously, data from the PCI/USB
8241 database was only attached to select devices, since the
8242 lookup was a relatively expensive operation due to O(n) time
8243 complexity (with n being the number of entries in the
8244 database). Since this is now O(1), we decided to add in this
8245 data for all devices where this is available, by
8246 default. Note that the indexed database needs to be rebuilt
8247 when new data files are installed. To achieve this you need
8248 to update your packaging scripts to invoke "udevadm hwdb
8249 --update" after installation of hwdb data files. For
8250 RPM-based distributions we introduced the new
8251 %udev_hwdb_update macro for this purpose.
8253 * The Journal gained support for the "Message Catalog", an
8254 indexed database to link up additional information with
8255 journal entries. For further details please check:
8257 https://www.freedesktop.org/wiki/Software/systemd/catalog
8259 The indexed message catalog database also needs to be
8260 rebuilt after installation of message catalog files. Use
8261 "journalctl --update-catalog" for this. For RPM-based
8262 distributions we introduced the %journal_catalog_update
8263 macro for this purpose.
8265 * The Python Journal bindings gained support for the standard
8266 Python logging framework.
8268 * The Journal API gained new functions for checking whether
8269 the underlying file system of a journal file is capable of
8270 properly reporting file change notifications, or whether
8271 applications that want to reflect journal changes "live"
8272 need to recheck journal files continuously in appropriate
8275 * It is now possible to set the "age" field for tmpfiles
8276 entries to 0, indicating that files matching this entry
8277 shall always be removed when the directories are cleaned up.
8279 * coredumpctl gained a new "gdb" verb which invokes gdb
8280 right-away on the selected coredump.
8282 * There's now support for "hybrid sleep" on kernels that
8283 support this, in addition to "suspend" and "hibernate". Use
8284 "systemctl hybrid-sleep" to make use of this.
8286 * logind's HandleSuspendKey= setting (and related settings)
8287 now gained support for a new "lock" setting to simply
8288 request the screen lock on all local sessions, instead of
8289 actually executing a suspend or hibernation.
8291 * systemd will now mount the EFI variables file system by
8294 * Socket units now gained support for configuration of the
8295 SMACK security label.
8297 * timedatectl will now output the time of the last and next
8298 daylight saving change.
8300 * We dropped support for various legacy and distro-specific
8301 concepts, such as insserv, early-boot SysV services
8302 (i.e. those for non-standard runlevels such as 'b' or 'S')
8303 or ArchLinux /etc/rc.conf support. We recommend the
8304 distributions who still need support this to either continue
8305 to maintain the necessary patches downstream, or find a
8306 different solution. (Talk to us if you have questions!)
8308 * Various systemd components will now bypass polkit checks for
8309 root and otherwise handle properly if polkit is not found to
8310 be around. This should fix most issues for polkit-less
8311 systems. Quite frankly this should have been this way since
8312 day one. It is absolutely our intention to make systemd work
8313 fine on polkit-less systems, and we consider it a bug if
8314 something does not work as it should if polkit is not around.
8316 * For embedded systems it is now possible to build udev and
8317 systemd without blkid and/or kmod support.
8319 * "systemctl switch-root" is now capable of switching root
8320 more than once. I.e. in addition to transitions from the
8321 initrd to the host OS it is now possible to transition to
8322 further OS images from the host. This is useful to implement
8323 offline updating tools.
8325 * Various other additions have been made to the RPM macros
8326 shipped with systemd. Use %udev_rules_update() after
8327 installing new udev rules files. %_udevhwdbdir,
8328 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
8329 %_sysctldir are now available which resolve to the right
8330 directories for packages to place various data files in.
8332 * journalctl gained the new --full switch (in addition to
8333 --all, to disable ellipsation for long messages.
8335 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
8336 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
8337 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
8338 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
8339 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
8340 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
8341 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
8342 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
8343 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
8347 * journalctl gained new --since= and --until= switches to
8348 filter by time. It also now supports nice filtering for
8349 units via --unit=/-u.
8351 * Type=oneshot services may use ExecReload= and do the
8354 * The journal daemon now supports time-based rotation and
8355 vacuuming, in addition to the usual disk-space based
8358 * The journal will now index the available field values for
8359 each field name. This enables clients to show pretty drop
8360 downs of available match values when filtering. The bash
8361 completion of journalctl has been updated
8362 accordingly. journalctl gained a new switch -F to list all
8363 values a certain field takes in the journal database.
8365 * More service events are now written as structured messages
8366 to the journal, and made recognizable via message IDs.
8368 * The timedated, localed and hostnamed mini-services which
8369 previously only provided support for changing time, locale
8370 and hostname settings from graphical DEs such as GNOME now
8371 also have a minimal (but very useful) text-based client
8372 utility each. This is probably the nicest way to changing
8373 these settings from the command line now, especially since
8374 it lists available options and is fully integrated with bash
8377 * There's now a new tool "systemd-coredumpctl" to list and
8378 extract coredumps from the journal.
8380 * We now install a README each in /var/log/ and
8381 /etc/rc.d/init.d explaining where the system logs and init
8382 scripts went. This hopefully should help folks who go to
8383 that dirs and look into the otherwise now empty void and
8384 scratch their heads.
8386 * When user-services are invoked (by systemd --user) the
8387 $MANAGERPID env var is set to the PID of systemd.
8389 * SIGRTMIN+24 when sent to a --user instance will now result
8390 in immediate termination of systemd.
8392 * gatewayd received numerous feature additions such as a
8393 "follow" mode, for live syncing and filtering.
8395 * browse.html now allows filtering and showing detailed
8396 information on specific entries. Keyboard navigation and
8397 mouse screen support has been added.
8399 * gatewayd/journalctl now supports HTML5/JSON
8400 Server-Sent-Events as output.
8402 * The SysV init script compatibility logic will now
8403 heuristically determine whether a script supports the
8404 "reload" verb, and only then make this available as
8407 * "systemctl status --follow" has been removed, use "journalctl
8410 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
8411 have been removed since they are hardly useful to be
8414 * And I'd like to take the opportunity to specifically mention
8415 Zbigniew for his great contributions. Zbigniew, you rock!
8417 Contributions from: Andrew Eikum, Christian Hesse, Colin
8418 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
8419 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
8420 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
8421 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
8422 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
8423 Jędrzejewski-Szmek, Сковорода Никита Андреевич
8427 * If /etc/vconsole.conf is non-existent or empty we will no
8428 longer load any console font or key map at boot by
8429 default. Instead the kernel defaults will be left
8430 intact. This is definitely the right thing to do, as no
8431 configuration should mean no configuration, and hard-coding
8432 font names that are different on all archs is probably a bad
8433 idea. Also, the kernel default key map and font should be
8434 good enough for most cases anyway, and mostly identical to
8435 the userspace fonts/key maps we previously overloaded them
8436 with. If distributions want to continue to default to a
8437 non-kernel font or key map they should ship a default
8438 /etc/vconsole.conf with the appropriate contents.
8440 Contributions from: Colin Walters, Daniel J Walsh, Dave
8441 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
8442 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8446 * journalctl gained a new --cursor= switch to show entries
8447 starting from the specified location in the journal.
8449 * We now enforce a size limit on journal entry fields exported
8450 with "-o json" in journalctl. Fields larger than 4K will be
8451 assigned null. This can be turned off with --all.
8453 * An (optional) journal gateway daemon is now available as
8454 "systemd-journal-gatewayd.service". This service provides
8455 access to the journal via HTTP and JSON. This functionality
8456 will be used to implement live log synchronization in both
8457 pull and push modes, but has various other users too, such
8458 as easy log access for debugging of embedded devices. Right
8459 now it is already useful to retrieve the journal via HTTP:
8461 # systemctl start systemd-journal-gatewayd.service
8462 # wget http://localhost:19531/entries
8464 This will download the journal contents in a
8465 /var/log/messages compatible format. The same as JSON:
8467 # curl -H"Accept: application/json" http://localhost:19531/entries
8469 This service is also accessible via a web browser where a
8470 single static HTML5 app is served that uses the JSON logic
8471 to enable the user to do some basic browsing of the
8472 journal. This will be extended later on. Here's an example
8473 screenshot of this app in its current state:
8475 http://0pointer.de/public/journal-gatewayd
8477 Contributions from: Kay Sievers, Lennart Poettering, Robert
8478 Milasan, Tom Gundersen
8482 * The bash completion logic is now available for journalctl
8485 * We do not mount the "cpuset" controller anymore together with
8486 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
8487 started if no parameters are assigned to it. "cpuset" hence
8488 broke code that assumed it could create "cpu" groups and
8491 * journalctl -f will now subscribe to terminal size changes,
8492 and line break accordingly.
8494 Contributions from: Dave Reisner, Kay Sievers, Lennart
8495 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
8499 * nspawn will now create a symlink /etc/localtime in the
8500 container environment, copying the host's timezone
8501 setting. Previously this has been done via a bind mount, but
8502 since symlinks cannot be bind mounted this has now been
8503 changed to create/update the appropriate symlink.
8505 * journalctl -n's line number argument is now optional, and
8506 will default to 10 if omitted.
8508 * journald will now log the maximum size the journal files may
8509 take up on disk. This is particularly useful if the default
8510 built-in logic of determining this parameter from the file
8511 system size is used. Use "systemctl status
8512 systemd-journald.service" to see this information.
8514 * The multi-seat X wrapper tool has been stripped down. As X
8515 is now capable of enumerating graphics devices via udev in a
8516 seat-aware way the wrapper is not strictly necessary
8517 anymore. A stripped down temporary stop-gap is still shipped
8518 until the upstream display managers have been updated to
8519 fully support the new X logic. Expect this wrapper to be
8520 removed entirely in one of the next releases.
8522 * HandleSleepKey= in logind.conf has been split up into
8523 HandleSuspendKey= and HandleHibernateKey=. The old setting
8524 is not available anymore. X11 and the kernel are
8525 distinguishing between these keys and we should too. This
8526 also means the inhibition lock for these keys has been split
8529 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
8530 Poettering, Lukas Nykryn, Václav Pavlín
8534 * Whenever a unit changes state we will now log this to the
8535 journal and show along the unit's own log output in
8538 * ConditionPathIsMountPoint= can now properly detect bind
8539 mount points too. (Previously, a bind mount of one file
8540 system to another place in the same file system could not be
8541 detected as mount, since they shared struct stat's st_dev
8544 * We will now mount the cgroup controllers cpu, cpuacct,
8545 cpuset and the controllers net_cls, net_prio together by
8548 * nspawn containers will now have a virtualized boot
8549 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
8550 over with a randomized ID at container initialization). This
8551 has the effect of making "journalctl -b" do the right thing
8554 * The JSON output journal serialization has been updated not
8555 to generate "endless" list objects anymore, but rather one
8556 JSON object per line. This is more in line how most JSON
8557 parsers expect JSON objects. The new output mode
8558 "json-pretty" has been added to provide similar output, but
8559 neatly aligned for readability by humans.
8561 * We dropped all explicit sync() invocations in the shutdown
8562 code. The kernel does this implicitly anyway in the kernel
8563 reboot() syscall. halt(8)'s -n option is now a compatibility
8566 * We now support virtualized reboot() in containers, as
8567 supported by newer kernels. We will fall back to exit() if
8568 CAP_SYS_REBOOT is not available to the container. Also,
8569 nspawn makes use of this now and will actually reboot the
8570 container if the containerized OS asks for that.
8572 * journalctl will only show local log output by default
8573 now. Use --merge (-m) to show remote log output, too.
8575 * libsystemd-journal gained the new sd_journal_get_usage()
8576 call to determine the current disk usage of all journal
8577 files. This is exposed in the new "journalctl --disk-usage"
8580 * journald gained a new configuration setting SplitMode= in
8581 journald.conf which may be used to control how user journals
8582 are split off. See journald.conf(5) for details.
8584 * A new condition type ConditionFileNotEmpty= has been added.
8586 * tmpfiles' "w" lines now support file globbing, to write
8587 multiple files at once.
8589 * We added Python bindings for the journal submission
8590 APIs. More Python APIs for a number of selected APIs will
8591 likely follow. Note that we intend to add native bindings
8592 only for the Python language, as we consider it common
8593 enough to deserve bindings shipped within systemd. There are
8594 various projects outside of systemd that provide bindings
8595 for languages such as PHP or Lua.
8597 * Many conditions will now resolve specifiers such as %i. In
8598 addition, PathChanged= and related directives of .path units
8599 now support specifiers as well.
8601 * There's now a new RPM macro definition for the system preset
8604 * journald will now warn if it ca not forward a message to the
8605 syslog daemon because its socket is full.
8607 * timedated will no longer write or process /etc/timezone,
8608 except on Debian. As we do not support late mounted /usr
8609 anymore /etc/localtime always being a symlink is now safe,
8610 and hence the information in /etc/timezone is not necessary
8613 * logind will now always reserve one VT for a text getty (VT6
8614 by default). Previously if more than 6 X sessions where
8615 started they took up all the VTs with auto-spawned gettys,
8616 so that no text gettys were available anymore.
8618 * udev will now automatically inform the btrfs kernel logic
8619 about btrfs RAID components showing up. This should make
8620 simple hotplug based btrfs RAID assembly work.
8622 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
8623 (but not for its children which will stay at the kernel
8624 default). This should allow setups with a lot more listening
8627 * systemd will now always pass the configured timezone to the
8628 kernel at boot. timedated will do the same when the timezone
8631 * logind's inhibition logic has been updated. By default,
8632 logind will now handle the lid switch, the power and sleep
8633 keys all the time, even in graphical sessions. If DEs want
8634 to handle these events on their own they should take the new
8635 handle-power-key, handle-sleep-key and handle-lid-switch
8636 inhibitors during their runtime. A simple way to achieve
8637 that is to invoke the DE wrapped in an invocation of:
8639 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
8641 * Access to unit operations is now checked via SELinux taking
8642 the unit file label and client process label into account.
8644 * systemd will now notify the administrator in the journal
8645 when he over-mounts a non-empty directory.
8647 * There are new specifiers that are resolved in unit files,
8648 for the host name (%H), the machine ID (%m) and the boot ID
8651 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
8652 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
8653 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
8654 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
8655 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
8656 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
8657 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
8661 * Support for reading structured kernel messages from
8662 /dev/kmsg has now been added and is enabled by default.
8664 * Support for reading kernel messages from /proc/kmsg has now
8665 been removed. If you want kernel messages in the journal
8666 make sure to run a recent kernel (>= 3.5) that supports
8667 reading structured messages from /dev/kmsg (see
8668 above). /proc/kmsg is now exclusive property of classic
8669 syslog daemons again.
8671 * The libudev API gained the new
8672 udev_device_new_from_device_id() call.
8674 * The logic for file system namespace (ReadOnlyDirectory=,
8675 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
8676 require pivot_root() anymore. This means fewer temporary
8677 directories are created below /tmp for this feature.
8679 * nspawn containers will now see and receive all submounts
8680 made on the host OS below the root file system of the
8683 * Forward Secure Sealing is now supported for Journal files,
8684 which provide cryptographical sealing of journal files so
8685 that attackers cannot alter log history anymore without this
8686 being detectable. Lennart will soon post a blog story about
8687 this explaining it in more detail.
8689 * There are two new service settings RestartPreventExitStatus=
8690 and SuccessExitStatus= which allow configuration of exit
8691 status (exit code or signal) which will be excepted from the
8692 restart logic, resp. consider successful.
8694 * journalctl gained the new --verify switch that can be used
8695 to check the integrity of the structure of journal files and
8696 (if Forward Secure Sealing is enabled) the contents of
8699 * nspawn containers will now be run with /dev/stdin, /dev/fd/
8700 and similar symlinks pre-created. This makes running shells
8701 as container init process a lot more fun.
8703 * The fstab support can now handle PARTUUID= and PARTLABEL=
8706 * A new ConditionHost= condition has been added to match
8707 against the hostname (with globs) and machine ID. This is
8708 useful for clusters where a single OS image is used to
8709 provision a large number of hosts which shall run slightly
8710 different sets of services.
8712 * Services which hit the restart limit will now be placed in a
8715 Contributions from: Bertram Poettering, Dave Reisner, Huang
8716 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
8717 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
8721 * When running in --user mode systemd will now become a
8722 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
8723 tree a lot more organized.
8725 * A new PartOf= unit dependency type has been introduced that
8726 may be used to group services in a natural way.
8728 * "systemctl enable" may now be used to enable instances of
8731 * journalctl now prints error log levels in red, and
8732 warning/notice log levels in bright white. It also supports
8733 filtering by log level now.
8735 * cgtop gained a new -n switch (similar to top), to configure
8736 the maximum number of iterations to run for. It also gained
8737 -b, to run in batch mode (accepting no input).
8739 * The suffix ".service" may now be omitted on most systemctl
8740 command lines involving service unit names.
8742 * There's a new bus call in logind to lock all sessions, as
8743 well as a loginctl verb for it "lock-sessions".
8745 * libsystemd-logind.so gained a new call sd_journal_perror()
8746 that works similar to libc perror() but logs to the journal
8747 and encodes structured information about the error number.
8749 * /etc/crypttab entries now understand the new keyfile-size=
8752 * shutdown(8) now can send a (configurable) wall message when
8753 a shutdown is cancelled.
8755 * The mount propagation mode for the root file system will now
8756 default to "shared", which is useful to make containers work
8757 nicely out-of-the-box so that they receive new mounts from
8758 the host. This can be undone locally by running "mount
8759 --make-rprivate /" if needed.
8761 * The prefdm.service file has been removed. Distributions
8762 should maintain this unit downstream if they intend to keep
8763 it around. However, we recommend writing normal unit files
8764 for display managers instead.
8766 * Since systemd is a crucial part of the OS we will now
8767 default to a number of compiler switches that improve
8768 security (hardening) such as read-only relocations, stack
8769 protection, and suchlike.
8771 * The TimeoutSec= setting for services is now split into
8772 TimeoutStartSec= and TimeoutStopSec= to allow configuration
8773 of individual time outs for the start and the stop phase of
8776 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
8777 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
8778 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
8779 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
8780 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
8781 Gundersen, Zbigniew Jędrzejewski-Szmek
8785 * The journal and id128 C APIs are now fully documented as man
8788 * Extra safety checks have been added when transitioning from
8789 the initial RAM disk to the main system to avoid accidental
8792 * /etc/crypttab entries now understand the new keyfile-offset=
8795 * systemctl -t can now be used to filter by unit load state.
8797 * The journal C API gained the new sd_journal_wait() call to
8798 make writing synchronous journal clients easier.
8800 * journalctl gained the new -D switch to show journals from a
8803 * journalctl now displays a special marker between log
8804 messages of two different boots.
8806 * The journal is now explicitly flushed to /var via a service
8807 systemd-journal-flush.service, rather than implicitly simply
8808 by seeing /var/log/journal to be writable.
8810 * journalctl (and the journal C APIs) can now match for much
8811 more complex expressions, with alternatives and
8814 * When transitioning from the initial RAM disk to the main
8815 system we will now kill all processes in a killing spree to
8816 ensure no processes stay around by accident.
8818 * Three new specifiers may be used in unit files: %u, %h, %s
8819 resolve to the user name, user home directory resp. user
8820 shell. This is useful for running systemd user instances.
8822 * We now automatically rotate journal files if their data
8823 object hash table gets a fill level > 75%. We also size the
8824 hash table based on the configured maximum file size. This
8825 together should lower hash collisions drastically and thus
8826 speed things up a bit.
8828 * journalctl gained the new "--header" switch to introspect
8829 header data of journal files.
8831 * A new setting SystemCallFilters= has been added to services
8832 which may be used to apply blacklists or whitelists to
8833 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
8835 * nspawn gained a new --link-journal= switch (and quicker: -j)
8836 to link the container journal with the host. This makes it
8837 very easy to centralize log viewing on the host for all
8838 guests while still keeping the journal files separated.
8840 * Many bugfixes and optimizations
8842 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
8843 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
8844 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
8849 * Several tools now understand kernel command line arguments,
8850 which are only read when run in an initial RAM disk. They
8851 usually follow closely their normal counterparts, but are
8854 * There's a new tool to analyze the readahead files that are
8855 automatically generated at boot. Use:
8857 /usr/lib/systemd/systemd-readahead analyze /.readahead
8859 * We now provide an early debug shell on tty9 if this enabled. Use:
8861 systemctl enable debug-shell.service
8863 * All plymouth related units have been moved into the Plymouth
8864 package. Please make sure to upgrade your Plymouth version
8867 * systemd-tmpfiles now supports getting passed the basename of
8868 a configuration file only, in which case it will look for it
8869 in all appropriate directories automatically.
8871 * udevadm info now takes a /dev or /sys path as argument, and
8872 does the right thing. Example:
8874 udevadm info /dev/sda
8875 udevadm info /sys/class/block/sda
8877 * systemctl now prints a warning if a unit is stopped but a
8878 unit that might trigger it continues to run. Example: a
8879 service is stopped but the socket that activates it is left
8882 * "systemctl status" will now mention if the log output was
8883 shortened due to rotation since a service has been started.
8885 * The journal API now exposes functions to determine the
8886 "cutoff" times due to rotation.
8888 * journald now understands SIGUSR1 and SIGUSR2 for triggering
8889 immediately flushing of runtime logs to /var if possible,
8890 resp. for triggering immediate rotation of the journal
8893 * It is now considered an error if a service is attempted to
8894 be stopped that is not loaded.
8896 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
8898 * systemd-analyze now supports Python 3
8900 * tmpfiles now supports cleaning up directories via aging
8901 where the first level dirs are always kept around but
8902 directories beneath it automatically aged. This is enabled
8903 by prefixing the age field with '~'.
8905 * Seat objects now expose CanGraphical, CanTTY properties
8906 which is required to deal with very fast bootups where the
8907 display manager might be running before the graphics drivers
8908 completed initialization.
8910 * Seat objects now expose a State property.
8912 * We now include RPM macros for service enabling/disabling
8913 based on the preset logic. We recommend RPM based
8914 distributions to make use of these macros if possible. This
8915 makes it simpler to reuse RPM spec files across
8918 * We now make sure that the collected systemd unit name is
8919 always valid when services log to the journal via
8922 * There's a new man page kernel-command-line(7) detailing all
8923 command line options we understand.
8925 * The fstab generator may now be disabled at boot by passing
8926 fstab=0 on the kernel command line.
8928 * A new kernel command line option modules-load= is now understood
8929 to load a specific kernel module statically, early at boot.
8931 * Unit names specified on the systemctl command line are now
8932 automatically escaped as needed. Also, if file system or
8933 device paths are specified they are automatically turned
8934 into the appropriate mount or device unit names. Example:
8936 systemctl status /home
8937 systemctl status /dev/sda
8939 * The SysVConsole= configuration option has been removed from
8940 system.conf parsing.
8942 * The SysV search path is no longer exported on the D-Bus
8945 * The Names= option has been removed from unit file parsing.
8947 * There's a new man page bootup(7) detailing the boot process.
8949 * Every unit and every generator we ship with systemd now
8950 comes with full documentation. The self-explanatory boot is
8953 * A couple of services gained "systemd-" prefixes in their
8954 name if they wrap systemd code, rather than only external
8955 code. Among them fsck@.service which is now
8956 systemd-fsck@.service.
8958 * The HaveWatchdog property has been removed from the D-Bus
8961 * systemd.confirm_spawn= on the kernel command line should now
8964 * There's a new man page crypttab(5) which details all options
8965 we actually understand.
8967 * systemd-nspawn gained a new --capability= switch to pass
8968 additional capabilities to the container.
8970 * timedated will now read known NTP implementation unit names
8971 from /usr/lib/systemd/ntp-units.d/*.list,
8972 systemd-timedated-ntp.target has been removed.
8974 * journalctl gained a new switch "-b" that lists log data of
8975 the current boot only.
8977 * The notify socket is in the abstract namespace again, in
8978 order to support daemons which chroot() at start-up.
8980 * There is a new Storage= configuration option for journald
8981 which allows configuration of where log data should go. This
8982 also provides a way to disable journal logging entirely, so
8983 that data collected is only forwarded to the console, the
8984 kernel log buffer or another syslog implementation.
8986 * Many bugfixes and optimizations
8988 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
8989 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
8990 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
8991 Shawn Landden, Tom Gundersen
8995 * "systemctl help <unit>" now shows the man page if one is
8998 * Several new man pages have been added.
9000 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
9001 MaxLevelConsole= can now be specified in
9002 journald.conf. These options allow reducing the amount of
9003 data stored on disk or forwarded by the log level.
9005 * TimerSlackNSec= can now be specified in system.conf for
9006 PID1. This allows system-wide power savings.
9008 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
9009 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
9014 * logind is now capable of (optionally) handling power and
9015 sleep keys as well as the lid switch.
9017 * journalctl now understands the syntax "journalctl
9018 /usr/bin/avahi-daemon" to get all log output of a specific
9021 * CapabilityBoundingSet= in system.conf now also influences
9022 the capability bound set of usermode helpers of the kernel.
9024 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
9025 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
9026 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
9027 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
9031 * Note that we skipped 139 releases here in order to set the
9032 new version to something that is greater than both udev's
9033 and systemd's most recent version number.
9035 * udev: all udev sources are merged into the systemd source tree now.
9036 All future udev development will happen in the systemd tree. It
9037 is still fully supported to use the udev daemon and tools without
9038 systemd running, like in initramfs or other init systems. Building
9039 udev though, will require the *build* of the systemd tree, but
9040 udev can be properly *run* without systemd.
9042 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
9043 should be used to create dead device nodes as workarounds for broken
9046 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
9047 no longer supported. udev_monitor_new_from_netlink() needs to be
9048 used to subscribe to events.
9050 * udev: when udevd is started by systemd, processes which are left
9051 behind by forking them off of udev rules, are unconditionally cleaned
9052 up and killed now after the event handling has finished. Services or
9053 daemons must be started as systemd services. Services can be
9054 pulled-in by udev to get started, but they can no longer be directly
9055 forked by udev rules.
9057 * udev: the daemon binary is called systemd-udevd now and installed
9058 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
9059 to adapt to that, create symlink, or rename the binary after building
9062 * libudev no longer provides these symbols:
9063 udev_monitor_from_socket()
9064 udev_queue_get_failed_list_entry()
9065 udev_get_{dev,sys,run}_path()
9066 The versions number was bumped and symbol versioning introduced.
9068 * systemd-loginctl and systemd-journalctl have been renamed
9069 to loginctl and journalctl to match systemctl.
9071 * The config files: /etc/systemd/systemd-logind.conf and
9072 /etc/systemd/systemd-journald.conf have been renamed to
9073 logind.conf and journald.conf. Package updates should rename
9074 the files to the new names on upgrade.
9076 * For almost all files the license is now LGPL2.1+, changed
9077 from the previous GPL2.0+. Exceptions are some minor stuff
9078 of udev (which will be changed to LGPL2.1 eventually, too),
9079 and the MIT licensed sd-daemon.[ch] library that is suitable
9080 to be used as drop-in files.
9082 * systemd and logind now handle system sleep states, in
9083 particular suspending and hibernating.
9085 * logind now implements a sleep/shutdown/idle inhibiting logic
9086 suitable for a variety of uses. Soonishly Lennart will blog
9087 about this in more detail.
9089 * var-run.mount and var-lock.mount are no longer provided
9090 (which previously bind mounted these directories to their new
9091 places). Distributions which have not converted these
9092 directories to symlinks should consider stealing these files
9093 from git history and add them downstream.
9095 * We introduced the Documentation= field for units and added
9096 this to all our shipped units. This is useful to make it
9097 easier to explore the boot and the purpose of the various
9100 * All smaller setup units (such as
9101 systemd-vconsole-setup.service) now detect properly if they
9102 are run in a container and are skipped when
9103 appropriate. This guarantees an entirely noise-free boot in
9104 Linux container environments such as systemd-nspawn.
9106 * A framework for implementing offline system updates is now
9107 integrated, for details see:
9108 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
9110 * A new service type Type=idle is available now which helps us
9111 avoiding ugly interleaving of getty output and boot status
9114 * There's now a system-wide CapabilityBoundingSet= option to
9115 globally reduce the set of capabilities for the
9116 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
9117 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
9118 even CAP_NET_ADMIN system-wide for secure systems.
9120 * There are now system-wide DefaultLimitXXX= options to
9121 globally change the defaults of the various resource limits
9122 for all units started by PID 1.
9124 * Harald Hoyer's systemd test suite has been integrated into
9125 systemd which allows easy testing of systemd builds in qemu
9126 and nspawn. (This is really awesome! Ask us for details!)
9128 * The fstab parser is now implemented as generator, not inside
9131 * systemctl will now warn you if .mount units generated from
9132 /etc/fstab are out of date due to changes in fstab that
9133 have not been read by systemd yet.
9135 * systemd is now suitable for usage in initrds. Dracut has
9136 already been updated to make use of this. With this in place
9137 initrds get a slight bit faster but primarily are much
9138 easier to introspect and debug since "systemctl status" in
9139 the host system can be used to introspect initrd services,
9140 and the journal from the initrd is kept around too.
9142 * systemd-delta has been added, a tool to explore differences
9143 between user/admin configuration and vendor defaults.
9145 * PrivateTmp= now affects both /tmp and /var/tmp.
9147 * Boot time status messages are now much prettier and feature
9148 proper english language. Booting up systemd has never been
9151 * Read-ahead pack files now include the inode number of all
9152 files to pre-cache. When the inode changes the pre-caching
9153 is not attempted. This should be nicer to deal with updated
9154 packages which might result in changes of read-ahead
9157 * We now temporaritly lower the kernel's read_ahead_kb variable
9158 when collecting read-ahead data to ensure the kernel's
9159 built-in read-ahead does not add noise to our measurements
9160 of necessary blocks to pre-cache.
9162 * There's now RequiresMountsFor= to add automatic dependencies
9163 for all mounts necessary for a specific file system path.
9165 * MountAuto= and SwapAuto= have been removed from
9166 system.conf. Mounting file systems at boot has to take place
9169 * nspawn now learned a new switch --uuid= to set the machine
9170 ID on the command line.
9172 * nspawn now learned the -b switch to automatically search
9175 * vt102 is now the default TERM for serial TTYs, upgraded from
9178 * systemd-logind now works on VT-less systems.
9180 * The build tree has been reorganized. The individual
9181 components now have directories of their own.
9183 * A new condition type ConditionPathIsReadWrite= is now available.
9185 * nspawn learned the new -C switch to create cgroups for the
9186 container in other hierarchies.
9188 * We now have support for hardware watchdogs, configurable in
9191 * The scheduled shutdown logic now has a public API.
9193 * We now mount /tmp as tmpfs by default, but this can be
9194 masked and /etc/fstab can override it.
9196 * Since udisks does not make use of /media anymore we are not
9197 mounting a tmpfs on it anymore.
9199 * journalctl gained a new --local switch to only interleave
9200 locally generated journal files.
9202 * We can now load the IMA policy at boot automatically.
9204 * The GTK tools have been split off into a systemd-ui.
9206 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
9207 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
9208 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
9209 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
9210 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
9211 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
9212 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
9213 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
9214 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
9219 * This is mostly a bugfix release
9221 * Support optional initialization of the machine ID from the
9222 KVM or container configured UUID.
9224 * Support immediate reboots with "systemctl reboot -ff"
9226 * Show /etc/os-release data in systemd-analyze output
9228 * Many bugfixes for the journal, including endianness fixes and
9229 ensuring that disk space enforcement works
9231 * sd-login.h is C++ compatible again
9233 * Extend the /etc/os-release format on request of the Debian
9236 * We now refuse non-UTF8 strings used in various configuration
9237 and unit files. This is done to ensure we do not pass invalid
9238 data over D-Bus or expose it elsewhere.
9240 * Register Mimo USB Screens as suitable for automatic seat
9243 * Read SELinux client context from journal clients in a race
9246 * Reorder configuration file lookup order. /etc now always
9247 overrides /run in order to allow the administrator to always
9248 and unconditionally override vendor-supplied or
9249 automatically generated data.
9251 * The various user visible bits of the journal now have man
9252 pages. We still lack man pages for the journal API calls
9255 * We now ship all man pages in HTML format again in the
9258 Contributions from: Dave Reisner, Dirk Eibach, Frederic
9259 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
9260 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
9265 * This is mostly a bugfix release
9267 * systems lacking /etc/os-release are no longer supported.
9269 * Various functionality updates to libsystemd-login.so
9271 * Track class of PAM logins to distinguish greeters from
9274 Contributions from: Kay Sievers, Lennart Poettering, Michael
9279 * This is an important bugfix release for v41.
9281 * Building man pages is now optional which should be useful
9282 for those building systemd from git but unwilling to install
9285 * Watchdog support for supervising services is now usable. In
9286 a future release support for hardware watchdogs
9287 (i.e. /dev/watchdog) will be added building on this.
9289 * Service start rate limiting is now configurable and can be
9290 turned off per service. When a start rate limit is hit a
9291 reboot can automatically be triggered.
9293 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
9295 Contributions from: Benjamin Franzke, Bill Nottingham,
9296 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
9297 Schmidt, Michał Górny, Piotr Drąg
9301 * The systemd binary is installed /usr/lib/systemd/systemd now;
9302 An existing /sbin/init symlink needs to be adapted with the
9305 * The code that loads kernel modules has been ported to invoke
9306 libkmod directly, instead of modprobe. This means we do not
9307 support systems with module-init-tools anymore.
9309 * Watchdog support is now already useful, but still not
9312 * A new kernel command line option systemd.setenv= is
9313 understood to set system wide environment variables
9314 dynamically at boot.
9316 * We now limit the set of capabilities of systemd-journald.
9318 * We now set SIGPIPE to ignore by default, since it only is
9319 useful in shell pipelines, and has little use in general
9320 code. This can be disabled with IgnoreSIPIPE=no in unit
9323 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
9324 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
9329 * This is mostly a bugfix release
9331 * We now expose the reason why a service failed in the
9332 "Result" D-Bus property.
9334 * Rudimentary service watchdog support (will be completed over
9335 the next few releases.)
9337 * When systemd forks off in order execute some service we will
9338 now immediately changes its argv[0] to reflect which process
9339 it will execute. This is useful to minimize the time window
9340 with a generic argv[0], which makes bootcharts more useful
9342 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
9343 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
9344 Mike Kazantsev, Ray Strode
9348 * This is mostly a test release, but incorporates many
9351 * New systemd-cgtop tool to show control groups by their
9354 * Linking against libacl for ACLs is optional again. If
9355 disabled, support tracking device access for active logins
9356 goes becomes unavailable, and so does access to the user
9357 journals by the respective users.
9359 * If a group "adm" exists, journal files are automatically
9360 owned by them, thus allow members of this group full access
9361 to the system journal as well as all user journals.
9363 * The journal now stores the SELinux context of the logging
9364 client for all entries.
9366 * Add C++ inclusion guards to all public headers
9368 * New output mode "cat" in the journal to print only text
9369 messages, without any meta data like date or time.
9371 * Include tiny X server wrapper as a temporary stop-gap to
9372 teach XOrg udev display enumeration. This is used by display
9373 managers such as gdm, and will go away as soon as XOrg
9374 learned native udev hotplugging for display devices.
9376 * Add new systemd-cat tool for executing arbitrary programs
9377 with STDERR/STDOUT connected to the journal. Can also act as
9378 BSD logger replacement, and does so by default.
9380 * Optionally store all locally generated coredumps in the
9381 journal along with meta data.
9383 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
9384 writing short strings to files (for usage for /sys), and for
9385 creating symlinks, character and block device nodes.
9387 * New unit file option ControlGroupPersistent= to make cgroups
9388 persistent, following the mechanisms outlined in
9389 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
9391 * Support multiple local RTCs in a sane way
9393 * No longer monopolize IO when replaying readahead data on
9394 rotating disks, since we might starve non-file-system IO to
9395 death, since fanotify() will not see accesses done by blkid,
9398 * Do not show kernel threads in systemd-cgls anymore, unless
9399 requested with new -k switch.
9401 Contributions from: Dan Horák, Kay Sievers, Lennart
9402 Poettering, Michal Schmidt
9406 * This is mostly a test release, but incorporates many
9409 * The git repository moved to:
9410 git://anongit.freedesktop.org/systemd/systemd
9411 ssh://git.freedesktop.org/git/systemd/systemd
9413 * First release with the journal
9414 http://0pointer.de/blog/projects/the-journal.html
9416 * The journal replaces both systemd-kmsg-syslogd and
9417 systemd-stdout-bridge.
9419 * New sd_pid_get_unit() API call in libsystemd-logind
9421 * Many systemadm clean-ups
9423 * Introduce remote-fs-pre.target which is ordered before all
9424 remote mounts and may be used to start services before all
9427 * Added Mageia support
9429 * Add bash completion for systemd-loginctl
9431 * Actively monitor PID file creation for daemons which exit in
9432 the parent process before having finished writing the PID
9433 file in the daemon process. Daemons which do this need to be
9434 fixed (i.e. PID file creation must have finished before the
9435 parent exits), but we now react a bit more gracefully to them.
9437 * Add colourful boot output, mimicking the well-known output
9438 of existing distributions.
9440 * New option PassCredentials= for socket units, for
9441 compatibility with a recent kernel ABI breakage.
9443 * /etc/rc.local is now hooked in via a generator binary, and
9444 thus will no longer act as synchronization point during
9447 * systemctl list-unit-files now supports --root=.
9449 * systemd-tmpfiles now understands two new commands: z, Z for
9450 relabelling files according to the SELinux database. This is
9451 useful to apply SELinux labels to specific files in /sys,
9454 * Output of SysV services is now forwarded to both the console
9455 and the journal by default, not only just the console.
9457 * New man pages for all APIs from libsystemd-login.
9459 * The build tree got reorganized and the build system is a
9460 lot more modular allowing embedded setups to specifically
9461 select the components of systemd they are interested in.
9463 * Support for Linux systems lacking the kernel VT subsystem is
9466 * configure's --with-rootdir= got renamed to
9467 --with-rootprefix= to follow the naming used by udev and
9470 * Unless specified otherwise we will now install to /usr instead
9471 of /usr/local by default.
9473 * Processes with '@' in argv[0][0] are now excluded from the
9474 final shut-down killing spree, following the logic explained
9476 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
9478 * All processes remaining in a service cgroup when we enter
9479 the START or START_PRE states are now killed with
9480 SIGKILL. That means it is no longer possible to spawn
9481 background processes from ExecStart= lines (which was never
9482 supported anyway, and bad style).
9484 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
9485 reloading of units together.
9487 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
9488 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
9489 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
9490 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
9491 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek