1 systemd System and Service Manager
3 CHANGES WITH 247 in spe:
5 * KERNEL API INCOMPATIBILITY: Linux 4.12 introduced two new uevents
6 "bind" and "unbind" to the Linux device model. When this kernel
7 change was made, systemd-udevd was only minimally updated to handle
8 and propagate these new event types. The introduction of these new
9 uevents (which are typically generated for USB devices and devices
10 needing a firmware upload before being functional) resulted in a
11 number of issues which we so far didn't address. We hoped the kernel
12 maintainers would themselves address these issues in some form, but
13 that did not happen. To handle them properly, many (if not most) udev
14 rules files shipped in various packages need updating, and so do many
15 programs that monitor or enumerate devices with libudev or sd-device,
16 or otherwise process uevents. Please note that this incompatibility
17 is not fault of systemd or udev, but caused by an incompatible kernel
18 change that happened back in Linux 4.12, but is becoming more and
19 more visible as the new uevents are generated by more kernel drivers.
21 To minimize issues resulting from this kernel change (but not avoid
22 them entirely) starting with systemd-udevd 247 the udev "tags"
23 concept (which is a concept for marking and filtering devices during
24 enumeration and monitoring) has been reworked: udev tags are now
25 "sticky", meaning that once a tag is assigned to a device it will not
26 be removed from the device again until the device itself is removed
27 (i.e. unplugged). This makes sure that any application monitoring
28 devices that match a specific tag is guaranteed to both see uevents
29 where the device starts being relevant, and those where it stops
30 being relevant (the latter now regularly happening due to the new
31 "unbind" uevent type). The udev tags concept is hence now a concept
32 tied to a *device* instead of a device *event* — unlike for example
33 udev properties whose lifecycle (as before) is generally tied to a
34 device event, meaning that the previously determined properties are
35 forgotten whenever a new uevent is processed.
37 With the newly redefined udev tags concept, sometimes it's necessary
38 to determine which tags are the ones applied by the most recent
39 uevent/database update, in order to discern them from those
40 originating from earlier uevents/database updates of the same
41 device. To accommodate for this a new automatic property CURRENT_TAGS
42 has been added that works similar to the existing TAGS property but
43 only lists tags set by the most recent uevent/database
44 update. Similarly, the libudev/sd-device API has been updated with
45 new functions to enumerate these 'current' tags, in addition to the
46 existing APIs that now enumerate the 'sticky' ones.
48 To properly handle "bind"/"unbind" on Linux 4.12 and newer it is
49 essential that all udev rules files and applications are updated to
50 handle the new events. Specifically:
52 • All rule files that currently use a header guard similar to
53 ACTION!="add|change",GOTO="xyz_end" should be updated to use
54 ACTION=="remove",GOTO="xyz_end" instead, so that the
55 properties/tags they add are also applied whenever "bind" (or
56 "unbind") is seen. (This is most important for all physical device
57 types — those for which "bind" and "unbind" are currently
58 generated, for all other device types this change is still
59 recommended but not as important — but certainly prepares for
60 future kernel uevent type additions).
62 • Similarly, all code monitoring devices that contains an 'if' branch
63 discerning the "add" + "change" uevent actions from all other
64 uevents actions (i.e. considering devices only relevant after "add"
65 or "change", and irrelevant on all other events) should be reworked
66 to instead negatively check for "remove" only (i.e. considering
67 devices relevant after all event types, except for "remove", which
68 invalidates the device). Note that this also means that devices
69 should be considered relevant on "unbind", even though conceptually
70 this — in some form — invalidates the device. Since the precise
71 effect of "unbind" is not generically defined, devices should be
72 considered relevant even after "unbind", however I/O errors
73 accessing the device should then be handled gracefully.
75 • Any code that uses device tags for deciding whether a device is
76 relevant or not most likely needs to be updated to use the new
77 udev_device_has_current_tag() API (or sd_device_has_current_tag()
78 in case sd-device is used), to check whether the tag is set at the
79 moment an uevent is seen (as opposed to the existing
80 udev_device_has_tag() API which checks if the tag ever existed on
81 the device, following the API concept redefinition explained
84 We are very sorry for this breakage and the requirement to update
85 packages using these interfaces. We'd again like to underline that
86 this is not caused by systemd/udev changes, but result of a kernel
89 * The MountAPIVFS= service file setting now defaults to on if
90 RootImage= and RootDirectory= are used, which means that with those
91 two settings /proc/, /sys/ and /dev/ are automatically properly set
92 up for services. Previous behaviour may be restored by explicitly
93 setting MountAPIVFS=off.
95 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
96 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
97 latter it takes precedence over the former, similar to how most of
98 systemd's own configuration is handled. Given that PAM stack
99 definitions are primarily put together by OS vendors/distributions
100 (though possibly overridden by users), this systemd release moves its
101 own PAM stack configuration for the "systemd-user" PAM service (i.e.
102 for the PAM session invoked by the per-user user@.service instance)
103 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
104 packages' vendor versions of their PAM stack definitions from
105 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
106 desired the location to which systemd installs its PAM stack
107 configuration may be changed via the -Dpamconfdir Meson option.
109 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
110 libpwquality and libcryptsetup have been changed to be based on
111 dlopen(): instead of regular dynamic library dependencies declared in
112 the binary ELF headers, these libraries are now loaded on demand
113 only, if they are available. If the libraries cannot be found the
114 relevant operations will fail gracefully, or a suitable fallback
115 logic is chosen. This is supposed to be useful for general purpose
116 distributions, as it allows minimizing the list of dependencies the
117 systemd packages pull in, permitting building of more minimal OS
118 images, while still making use of these "weak" dependencies should
119 they be installed. Since many package managers automatically
120 synthesize package dependencies from ELF shared library dependencies,
121 some additional manual packaging work has to be done now to replace
122 those (slightly downgraded from "required" to "recommended" or
123 whatever is conceptually suitable for the package manager). Note that
124 this change does not alter build-time behaviour: as before the
125 build-time dependencies have to be installed during build, even if
126 they now are optional during runtime.
128 * sd-event.h gained a new call sd_event_add_time_relative() for
129 installing timers relative to the current time. This is mostly a
130 convenience wrapper around the pre-existing sd_event_add_time() call
131 which installs absolute timers.
133 * sd-event event sources may now be placed in a new "exit-on-failure"
134 mode, which may be controlled via the new
135 sd_event_source_get_exit_on_failure() and
136 sd_event_source_set_exit_on_failure() functions. If enabled, any
137 failure returned by the event source handler functions will result in
138 exiting the event loop (unlike the default behaviour of just
139 disabling the event source but continuing with the event loop). This
140 feature is useful to set for all event sources that define "primary"
141 program behaviour (where failure should be fatal) in contrast to
142 "auxiliary" behaviour (where failure should remain local).
144 * Most event source types sd-event supports now accept a NULL handler
145 function, in which case the event loop is exited once the event
146 source is to be dispatched, using the userdata pointer — converted to
147 a signed integer — as exit code of the event loop. Previously this
148 was supported for IO and signal event sources already. Exit event
149 sources still do not support this (simply because it makes little
150 sense there, as the event loop is already exiting when they are
153 * A new per-unit setting RootImageOptions= has been added which allows
154 tweaking the mount options for any file system mounted as effect of
155 the RootImage= setting.
157 * Another new per-unit setting MountImages= has been added, that allows
158 mounting additional disk images into the file system tree accessible
161 * systemd-repart now generates JSON output when requested with the new
164 * systemd-machined's OpenMachineShell() bus call will now pass
165 additional policy metadata data fields to the PolicyKit
166 authentication request.
168 * systemd-tmpfiles gained a new -E switch, which is equivalent to
169 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
170 --exclude=/sys. It's particularly useful in combination with --root=,
171 when operating on OS trees that do not have any of these four runtime
172 directories mounted, as this means no files below these subtrees are
173 created or modified, since those mount points should probably remain
176 * systemd-tmpfiles gained a new --image= switch which is like --root=,
177 but takes a disk image instead of a directory as argument. The
178 specified disk image is mounted inside a temporary mount namespace
179 and the tmpfiles.d/ drop-ins stored in the image are executed and
180 applied to the image. systemd-sysusers similarly gained a new
181 --image= switch, that allows the sysusers.d/ drop-ins stored in the
182 image to be applied onto the image.
184 * Similarly, the journalctl command also gained an --image= switch,
185 which is a quick one-step solution to look at the log data included
188 * journalctl's --output=cat option (which outputs the log content
189 without any metadata, just the pure text messages) will now make use
190 of terminal colors when run on a suitable terminal, similarly to the
193 * JSON group records now support a "description" string that may be
194 used to add a human-readable textual description to such groups. This
195 is supposed to match the user's GECOS field which traditionally
196 didn't have a counterpart for group records.
198 * The "systemd-dissect" tool that may be used to inspect OS disk images
199 and that was previously installed to /usr/lib/systemd/ has now been
200 moved to /usr/bin/, reflecting its updated status of an officially
201 supported tool with a stable interface. It gained support for a new
202 --mkdir switch which when combined with --mount has the effect of
203 creating the directory to mount the image to if it is missing
204 first. It also gained two new commands --copy-from and --copy-to for
205 copying files and directories in and out of an OS image without the
206 need to manually mount it. It also acquired support for a new option
207 --json= to generate JSON output when inspecting an OS image.
209 * The cgroup2 file system is now mounted with the
210 "memory_recursiveprot" mount option, supported since kernel 5.7. This
211 means that the MemoryLow= and MemoryMin= unit file settings now apply
212 recursively to whole subtrees.
214 * systemd-homed now defaults to using the btrfs file system — if
215 available — when creating home directories in LUKS volumes. This may
216 be changed with the DefaultFileSystemType= setting in homed.conf.
217 It's now the default file system in various major distributions and
218 has the major benefit for homed that it can be grown and shrunk while
219 mounted, unlike the other contenders ext4 and xfs, which can both be
220 grown online, but not shrunk (in fact xfs is the technically most
221 limited option here, as it cannot be shrunk at all).
223 * JSON user records managed by systemd-homed gained support for
224 "recovery keys". These are basically secondary passphrases that can
225 unlock user accounts/home directories. They are computer-generated
226 rather than user-chosen, and typically have greater entropy.
227 homectl's --recovery-key= option may be used to add a recovery key to
228 a user account. The generated recovery key is displayed as a QR code,
229 so that it can be scanned to be kept in a safe place. This feature is
230 particularly useful in combination with systemd-homed's support for
231 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
232 security tokens are lost. Recovery keys may be entered wherever the
233 system asks for a password.
235 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
236 home directory which indicates that a home directory has not been
237 deactivated cleanly when offline. This flag is useful to identify
238 home directories for which the offline discard logic did not run when
239 offlining, and where it would be a good idea to log in again to catch
242 * systemctl gained a new parameter --timestamp= which may be used to
243 change the style in which timestamps are output, i.e. whether to show
244 them in local timezone or UTC, or whether to show µs granularity.
246 * Alibaba's "pouch" container manager is now detected by
247 systemd-detect-virt, ConditionVirtualization= and similar
248 constructs. Similar, they now also recognize IBM PowerVM machine
251 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
252 place to use for propagating external mounts into the
253 container. Similarly /run/host/notify is now used as the socket path
254 for container payloads to communicate with the container manager
255 using sd_notify(). The container manager now uses the
256 /run/host/inaccessible/ directory to place "inaccessible" file nodes
257 of all relevant types which may be used by the container payload as
258 bind mount source to over-mount inodes to make them inaccessible.
259 /run/host/container-manager will now be initialized with the same
260 string as the $container environment variable passed to the
261 container's PID 1. /run/host/container-uuid will be initialized with
262 the same string as $container_uuid. This means the /run/host/
263 hierarchy is now the primary way to make host resources available to
264 the container. The Container Interface documents these new files and
267 https://systemd.io/CONTAINER_INTERFACE
269 * Support for the "ConditionNull=" unit file condition has been
270 deprecated and undocumented for 6 years. systemd started to warn
271 about its use 1.5 years ago. It has now been removed entirely.
273 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
274 a sd_bus_error struct and a list of error names, and checks if the
275 error matches one of these names. It's a convenience wrapper that is
276 useful in cases where multiple errors shall be handled the same way.
278 * A new system call filter list "@known" has been added, that contains
279 all system calls known at the time systemd was built.
281 * Behaviour of system call filter allow lists has changed slightly:
282 system calls that are contained in @known will result in a EPERM by
283 default, while those not contained in it result in ENOSYS. This
284 should improve compatibility because known system calls will thus be
285 communicated as prohibited, while unknown (and thus newer ones) will
286 be communicated as not implemented, which hopefully has the greatest
287 chance of triggering the right fallback code paths in client
290 * "systemd-analyze syscall-filter" will now show two separate sections
291 at the bottom of the output: system calls known during systemd build
292 time but not included in any of the filter groups shown above, and
293 system calls defined on the local kernel but known during systemd
296 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
297 systemd-nspawn all system call filter violations will be logged by
298 the kernel (audit). This is useful for tracking down system calls
299 invoked by container payloads that are prohibited by the container's
300 system call filter policy.
302 * Two new unit file settings ProtectProc= and ProcSubset= have been
303 added that expose the hidepid= and subset= mount options of procfs.
304 All processes of the unit will only see processes in /proc that are
305 are owned by the unit's user. This is an important new sandboxing
306 option that is recommended to be set on all system services. All
307 long-running system services that are included in systemd itself set
308 this option now. This option is only supported on kernel 5.8 and
309 above, since the hidepid= option supported on older kernels was not a
310 per-mount option but actually applied to the whole PID namespace.
312 * Socket units gained a new boolean setting FlushPending=. If enabled
313 all pending socket data/connections are flushed whenever the socket
314 unit enters the "listening" state, i.e. after the associated service
317 * The unit file setting NUMAMask= gained a new "all" value: when used,
318 all existing NUMA nodes are added to the NUMA mask.
320 * A new "credentials" logic has been added to system services. This is
321 a simple mechanism to pass privileged data to services in a safe and
322 secure way. It's supposed to be used to pass per-service secret data
323 such as passwords or cryptographic keys but also associated less
324 private information such as user names, certificates, and similar to
325 system services. Each credential is identified by a short user-chosen
326 name and may contain arbitrary binary data. Two new unit file
327 settings have been added: SetCredential= and LoadCredential=. The
328 former allows setting a credential to a literal string, the latter
329 sets a credential to the contents of a file (or data read from a
330 user-chosen AF_UNIX stream socket). Credentials are passed to the
331 service via a special credentials directory, one file for each
332 credential. The path to the credentials directory is passed in a new
333 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
334 are passed in the file system they may be easily referenced in
335 ExecStart= command lines too, thus no explicit support for the
336 credentials logic in daemons is required (though ideally daemons
337 would look for the bits they need in $CREDENTIALS_DIRECTORY
338 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
339 backed by unswappable memory if privileges allow it, immutable if
340 privileges allow it, is accessible only to the service's UID, and is
341 automatically destroyed when the service stops.
343 * systemd-nspawn supports the same credentials logic. It can both
344 consume credentials passed to it via the aforementioned
345 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
346 to its payload. The service manager/PID 1 has been updated to match
347 this: it can also accept credentials from the container manager that
348 invokes it (in fact: any process that invokes it), and passes them on
349 to its services. Thus, credentials can be propagated recursively down
350 the tree: from a system's service manager to a systemd-nspawn
351 service, to the service manager that runs as container payload and to
352 the service it runs below. Credentials may also be added on the
353 systemd-nspawn command line, using new --set-credential= and
354 --load-credential= command line switches that match the
355 aforementioned service settings.
357 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
358 the partition drop-ins which may be used to format/LUKS
359 encrypt/populate any created partitions. The partitions are
360 encrypted/formatted/populated before they are registered in the
361 partition table, so that they appear atomically: either the
362 partitions do not exist yet or they exist fully encrypted, formatted,
363 and populated — there is no time window where they are
364 "half-initialized". Thus the system is robust to abrupt shutdown: if
365 the tool is terminated half-way during its operations on next boot it
366 will start from the beginning.
368 * systemd-repart's --size= operation gained a new "auto" value. If
369 specified, and operating on a loopback file it is automatically sized
370 to the minimal size the size constraints permit. This is useful to
371 use "systemd-repart" as an image builder for minimally sized images.
373 * systemd-resolved now gained a third IPC interface for requesting name
374 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
375 interface is now supported. The nss-resolve NSS module has been
376 modified to use this new interface instead of D-Bus. Using Varlink
377 has a major benefit over D-Bus: it works without a broker service,
378 and thus already during earliest boot, before the dbus daemon has
379 been started. This means name resolution via systemd-resolved now
380 works at the same time systemd-networkd operates: from earliest boot
381 on, including in the initrd.
383 * systemd-resolved gained support for a new DNSStubListenerExtra=
384 configuration file setting which may be used to specify additional IP
385 addresses the built-in DNS stub shall listen on, in addition to the
386 main one on 127.0.0.53:53.
388 * Name lookups issued via systemd-resolved's D-Bus and Varlink
389 interfaces (and thus also via glibc NSS if nss-resolve is used) will
390 now honour a trailing dot in the hostname: if specified the search
391 path logic is turned off. Thus "resolvectl query foo." is now
392 equivalent to "resolvectl query --search=off foo.".
394 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
395 exposes how /etc/resolv.conf is currently managed: by resolved (and
396 in which mode if so) or another subsystem. "resolvctl" will display
397 this property in its status output.
399 * The resolv.conf snippets systemd-resolved provides will now set "."
400 as the search domain if no other search domain is known. This turns
401 off the derivation of an implicit search domain by nss-dns for the
402 hostname, when the hostname is set to an FQDN. This change is done to
403 make nss-dns using resolv.conf provided by systemd-resolved behave
404 more similarly to nss-resolve.
406 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
407 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
408 "birth" time (btime) of a file in addition to the atime, mtime, and
411 * systemd-analyze gained a new verb "capability" that lists all known
412 capabilities by the systemd build and by the kernel.
414 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
415 advance the system clock to it at boot if it is noticed to be before
416 that time. Previously, PID 1 would only advance the time to an epoch
417 time that is set during build-time. With this new file OS builders
418 can change this epoch timestamp on individual OS images without
419 having to rebuild systemd.
421 * systemd-logind will now listen to the KEY_RESTART key from the Linux
422 input layer and reboot the system if it is pressed, similarly to how
423 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
424 was originally defined in the Multimedia context (to restart playback
425 of a song or film), but is now primarily used in various embedded
426 devices for "Reboot" buttons. Accordingly, systemd-logind will now
427 honour it as such. This may configured in more detail via the new
428 HandleRebootKey= and RebootKeyIgnoreInhibited=.
430 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
431 copying OS trees, for example in "systemd-nspawn --ephemeral",
432 "systemd-nspawn --template=", "machinectl clone" and similar. This is
433 useful when operating with OSTree images, which use hardlinks heavily
434 throughout, and where such copies previously resulting in "exploding"
437 * systemd-nspawn's --console= setting gained support for a new
438 "autopipe" value, which is identical to "interactive" when invoked on
439 a TTY, and "pipe" otherwise.
441 * systemd-networkd's .network files gained support for explicitly
442 configuring the multicast membership entries of bridge devices in the
443 [BridgeMDB] section. It also gained support for the PIE queuing
444 discipline in the [FlowQueuePIE] sections.
446 * systemd-networkd's .netdev files may now be used to create "BareUDP"
447 tunnels, configured in the new [BareUDP] setting. VXLAN tunnels may
448 now be marked to be independent of any underlying network interface
449 via the new Independent= boolean setting.
451 * systemd-networkd's Gateway= setting in .network files now accepts the
452 special values _dhcp4 and _ipv6ra to configure additional, locally
453 defined, explicit routes to the gateway acquired via DHCP or IPv6
454 Router Advertisements.
456 * systemctl gained support for two new verbs: "service-log-level" and
457 "service-log-target" may be used on services that implement the
458 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
459 adjust the log level and target. All of systemd's long-running
460 services support this now, but ideally all system services would
461 implement this interface to make the system more uniformly
464 * The SystemCallErrorNumber= unit file setting now accepts the new
465 "kill" and "log" actions, in addition to arbitrary error number
466 specifications as before. If "kill" the the processes are killed on
467 the event, if "log" the offending system call is audit logged.
469 * A new SystemCallLog= unit file setting has been added that accepts a
470 list of system calls that shall be logged about (audit).
472 * The OS image dissection logic (as used by RootImage= in unit files or
473 systemd-nspawn's --image= switch) has gained support for identifying
474 and mounting explicit /usr/ partitions, which are now defined in the
475 discoverable partition specification. This should be useful for
476 environments where the root file system is
477 generated/formatted/populated dynamically on first boot and combined
478 with an immutable /usr/ tree that is supplied by the vendor.
480 * In the final phase of shutdown, within the systemd-shutdown binary
481 we'll now try to detach MD devices (i.e software RAID) in addition to
482 loopback block devices and DM devices as before. This is supposed to
483 be a safety net only, in order to increase robustness if things go
484 wrong. Storage subsystems are expected to properly detach their
485 storage volumes during regular shutdown already (or in case of
486 storage backing the root file system: in the initrd hook we return to
489 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
490 will now log the thread ID in their log output. This is useful when
491 working with heavily threaded programs.
493 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
494 not use the RDRAND CPU instruction. This is useful in environments
495 such as replay debuggers where non-deterministic behaviour is not
498 * The autopaging logic in systemd's various tools (such as systemctl)
499 has been updated to turn on "secure" mode in "less"
500 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
501 detected. This disables invoking external programs from the pager,
502 via the pipe logic. This behaviour may be overridden via the new
503 $SYSTEMD_PAGERSECURE environment variable.
505 * Units which have resource limits (.service, .mount, .swap, .slice,
506 .socket, and .slice) gained new configuration settings
507 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
508 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
509 limits and optional action taken by systemd-oomd.
511 * A new service systemd-oomd has been added. It monitors resource
512 contention for selected parts of the unit hierarchy using the PSI
513 information reported by the kernel, and kills processes when memory
514 or swap pressure is above configured limits. This service is only
515 enabled in developer mode (see below) and should be considered a
516 preview in this release. Behaviour details and option names are
517 subject to change without the usual backwards-compatibility promises.
519 * A new helper oomctl has been added to introspect systemd-oomd state.
520 If also is only available in developer mode and should be considered
521 a preview without the usual backwards-compatibility promises.
523 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
524 enabled, systemd reads the system UID boundaries from /etc/login.defs
525 at runtime, instead of using the built-in values selected during
526 build. This is an option to improve compatibility for upgrades from
527 old systems. It's strongly recommended not to make use of this
528 functionality on new systems (or even enable it during build), as it
529 makes something runtime-configurable that is mostly an implementation
530 detail of the OS, and permits avoidable differences in deployments
531 that create all kinds of problems in the long run.
533 * New meson option '-Dmode=developer|release' has been added. When
534 'developer', additional checks and features are enabled that are
535 relevant during upstream development, e.g. verification that
536 semi-automatically-generated documentation has been properly updated
537 following API changes. Those checks are considered hints for
538 developers and are not actionable in downstream builds. In addition,
539 extra features that are not ready for general consumption may be
540 enabled in developer mode. It is thus recommended to set
541 '-Dmode=release' in end-user and distro builds.
543 * systemd-cryptsetup gained support for processing detached LUKS
544 headers specified on the kernel command line via the header=
545 parameter of the luks.options= kernel command line option. The same
546 device/path syntax as for key files is supported for header files
549 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
550 slot index data for devices that are connected through a PCI bridge
551 where the _SUN index is associated with the bridge instead of the
552 network device itself. Previously this would create ambiguous device
553 naming if multiple network interfaces were connected to the same PCI
554 bridge. Since this is a naming scheme incompatibility on systems that
555 possess hardware like this it has been introduced as new naming
556 scheme "v247". The previous scheme can be selected via the
557 "net.naming-scheme=v245" kernel command line parameter.
559 * ConditionFirstBoot= semantics have been modified to be safe towards
560 abnormal system power-off during first boot. Specifically, the
561 "systemd-machine-id-commit.service" service now acts as boot
562 milestone indicating when the first boot process is sufficiently
563 complete in order to not consider the next following boot also a
564 first boot. If the system is reset before this unit is reached the
565 first time, the next boot will still be considered a first boot; once
566 it has been reached, no further boots will be considered a first
567 boot. The "first-boot-complete.target" unit now acts as official hook
568 point to order against this. If a service shall be run on every boot
569 until the first boot fully succeeds it may thus be ordered before
570 this target unit (and pull it in) and carry ConditionFirstBoot=
573 * bootctl's set-default and set-oneshot commands now accept the three
574 special strings "@default", "@oneshot", "@current" in place of a boot
575 entry id. These strings are resolved to the current default and
576 oneshot boot loader entry, as well as the currently booted one. Thus
577 a command "bootctl set-default @current" may be used to make the
578 currently boot menu item the new default for all subsequent boots.
580 * A new generic target unit "initrd-cryptsetup.target" has been added
581 that is supposed to pull in all encrypted volumes that shall be set
582 up during the initrd phase. It takes the place of "cryptsetup.target"
583 and "remote-cryptsetup.target" that exist during the host boot
584 phase. In other words, the new "initrd-cryptsetup.target" is supposed
585 to take the role for "initrd-fs.target", but for encrypted volumes.
587 * "systemctl edit" has been updated to show the original effective unit
588 contents in commented form in the text editor.
592 * The service manager gained basic support for cgroup v2 freezer. Units
593 can now be suspended or resumed either using new systemctl verbs,
594 freeze and thaw respectively, or via D-Bus.
596 * PID 1 may now automatically load pre-compiled AppArmor policies from
597 /etc/apparmor/earlypolicy during early boot.
599 * The CPUAffinity= setting in service unit files now supports a new
600 special value "numa" that causes the CPU affinity masked to be set
601 based on the NUMA mask.
603 * systemd will now log about all left-over processes remaining in a
604 unit when the unit is stopped. It will now warn about services using
605 KillMode=none, as this is generally an unsafe thing to make use of.
607 * Two new unit file settings
608 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
609 added. They may be used to check whether a specific file system path
610 resides on a block device that is encrypted on the block level
611 (i.e. using dm-crypt/LUKS).
613 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
614 has been added that may be used for simple environment checks. This
615 is particularly useful when passing in environment variables from a
616 container manager (or from PAM in case of the systemd --user
619 * .service unit files now accept a new setting CoredumpFilter= which
620 allows configuration of the memory sections coredumps of the
621 service's processes shall include.
623 * .mount units gained a new ReadWriteOnly= boolean option. If set
624 it will not be attempted to mount a file system read-only if mounting
625 in read-write mode doesn't succeed. An option x-systemd.rw-only is
626 available in /etc/fstab to control the same.
628 * .socket units gained a new boolean setting PassPacketInfo=. If
629 enabled, the kernel will attach additional per-packet metadata to all
630 packets read from the socket, as an ancillary message. This controls
631 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
632 depending on socket type.
634 * .service units gained a new setting RootHash= which may be used to
635 specify the root hash for verity enabled disk images which are
636 specified in RootImage=. RootVerity= may be used to specify a path to
637 the Verity data matching a RootImage= file system. (The latter is
638 only useful for images that do not contain the Verity data embedded
639 into the same image that carries a GPT partition table following the
640 Discoverable Partition Specification). Similarly, systemd-nspawn
641 gained a new switch --verity-data= that takes a path to a file with
642 the verity data of the disk image supplied in --image=, if the image
643 doesn't contain the verity data itself.
645 * .service units gained a new setting RootHashSignature= which takes
646 either a base64 encoded PKCS#7 signature of the root hash specified
647 with RootHash=, or a path to a file to read the signature from. This
648 allows validation of the root hash against public keys available in
649 the kernel keyring, and is only supported on recent kernels
650 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
651 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
652 this mechanism has also been added to systemd-veritysetup.
654 * .service unit files gained two new options
655 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
656 tune behaviour if a start or stop timeout is hit, i.e. whether to
657 terminate the service with SIGTERM, SIGABRT or SIGKILL.
659 * Most options in systemd that accept hexadecimal values prefixed with
660 0x in additional to the usual decimal notation now also support octal
661 notation when the 0o prefix is used and binary notation if the 0b
664 * Various command line parameters and configuration file settings that
665 configure key or certificate files now optionally take paths to
666 AF_UNIX sockets in the file system. If configured that way a stream
667 connection is made to the socket and the required data read from
668 it. This is a simple and natural extension to the existing regular
669 file logic, and permits other software to provide keys or
670 certificates via simple IPC services, for example when unencrypted
671 storage on disk is not desired. Specifically, systemd-networkd's
672 Wireguard and MACSEC key file settings as well as
673 systemd-journal-gatewayd's and systemd-journal-remote's PEM
674 key/certificate parameters support this now.
676 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
677 configuration files that support specifier expansion learnt six new
678 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
679 resolve to the various ID fields from /etc/os-release, %l resolves to
680 the "short" hostname of the system, i.e. the hostname configured in
681 the kernel truncated at the first dot.
683 * Support for the .include syntax in unit files has been removed. The
684 concept has been obsolete for 6 years and we started warning about
685 its pending removal 2 years ago (also see NEWS file below). It's
688 * StandardError= and StandardOutput= in unit files no longer support
689 the "syslog" and "syslog-console" switches. They were long removed
690 from the documentation, but will now result in warnings when used,
691 and be converted to "journal" and "journal+console" automatically.
693 * If the service setting User= is set to the "nobody" user, a warning
694 message is now written to the logs (but the value is nonetheless
695 accepted). Setting User=nobody is unsafe, since the primary purpose
696 of the "nobody" user is to own all files whose owner cannot be mapped
697 locally. It's in particular used by the NFS subsystem and in user
698 namespacing. By running a service under this user's UID it might get
699 read and even write access to all these otherwise unmappable files,
700 which is quite likely a major security problem.
702 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
703 and others) now have a size and inode limits applied (50% of RAM for
704 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
705 that the implicit kernel default is 50% too, so there is no change
706 in the size limit for /tmp and /dev/shm.
708 * nss-mymachines lost support for resolution of users and groups, and
709 now only does resolution of hostnames. This functionality is now
710 provided by nss-systemd. Thus, the 'mymachines' entry should be
711 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
712 (and 'systemd' added if it is not already there).
714 * A new kernel command line option systemd.hostname= has been added
715 that allows controlling the hostname that is initialized early during
718 * A kernel command line option "udev.blockdev_read_only" has been
719 added. If specified all hardware block devices that show up are
720 immediately marked as read-only by udev. This option is useful for
721 making sure that a specific boot under no circumstances modifies data
722 on disk. Use "blockdev --setrw" to undo the effect of this, per
725 * A new boolean kernel command line option systemd.swap= has been
726 added, which may be used to turn off automatic activation of swap
727 devices listed in /etc/fstab.
729 * New kernel command line options systemd.condition-needs-update= and
730 systemd.condition-first-boot= have been added, which override the
731 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
734 * A new kernel command line option systemd.clock-usec= has been added
735 that allows setting the system clock to the specified time in µs
736 since Jan 1st, 1970 early during boot. This is in particular useful
737 in order to make test cases more reliable.
739 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
740 systemd-coredump to save core files for suid processes. When saving
741 the core file, systemd-coredump will use the effective uid and gid of
742 the process that faulted.
744 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
745 now automatically set to "Y" at boot, in order to enable pstore
746 generation for collection with systemd-pstore.
748 * We provide a set of udev rules to enable auto-suspend on PCI and USB
749 devices that were tested to correctly support it. Previously, this
750 was distributed as a set of udev rules, but has now been replaced by
751 by a set of hwdb entries (and a much shorter udev rule to take action
752 if the device modalias matches one of the new hwdb entries).
754 As before, entries are periodically imported from the database
755 maintained by the ChromiumOS project. If you have a device that
756 supports auto-suspend correctly and where it should be enabled by
757 default, please submit a patch that adds it to the database (see
758 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
760 * systemd-udevd gained the new configuration option timeout_signal= as well
761 as a corresponding kernel command line option udev.timeout_signal=.
762 The option can be used to configure the UNIX signal that the main
763 daemon sends to the worker processes on timeout. Setting the signal
764 to SIGABRT is useful for debugging.
766 * .link files managed by systemd-udevd gained options RxFlowControl=,
767 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
768 order to configure various flow control parameters. They also gained
769 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
770 frame ring buffer sizes.
772 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
773 enabled systemd-networkd manages all routes configured by other tools.
775 * .network files managed by systemd-networkd gained a new section
776 [SR-IOV], in order to configure SR-IOV capable network devices.
778 * systemd-networkd's [IPv6Prefix] section in .network files gained a
779 new boolean setting Assign=. If enabled an address from the prefix is
780 automatically assigned to the interface.
782 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
783 controls delegated prefixes assigned by DHCPv6 client. The section
784 has three settings: SubnetID=, Assign=, and Token=. The setting
785 SubnetID= allows explicit configuration of the preferred subnet that
786 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
787 Assign= is enabled (which is the default) an address from any acquired
788 delegated prefix is automatically chosen and assigned to the
789 interface. The setting Token= specifies an optional address generation
792 * systemd-networkd's [Network] section gained a new setting
793 IPv4AcceptLocal=. If enabled the interface accepts packets with local
796 * systemd-networkd gained support for configuring the HTB queuing
797 discipline in the [HierarchyTokenBucket] and
798 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
799 be configured in the [PFIFO] section, "GRED" in
800 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
801 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
802 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
803 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
804 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
805 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
807 * systemd-networkd gained support for a new Termination= setting in the
808 [CAN] section for configuring the termination resistor. It also
809 gained a new ListenOnly= setting for controlling whether to only
810 listen on CAN interfaces, without interfering with traffic otherwise
811 (which is useful for debugging/monitoring CAN network
812 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
813 been added to configure various CAN-FD aspects.
815 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
816 When enabled, DHCPv6 will be attempted right-away without requiring an
817 Router Advertisement packet suggesting it first (i.e. without the 'M'
818 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
819 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
820 the RA packets suggest it.
822 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
823 which may be used to turn off use of the gateway information provided
824 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
825 used to configure how to process leases that lack a lifetime option.
827 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
828 setting SendVendorOption= allowing configuration of additional vendor
829 options to send in the DHCP requests/responses. The [DHCPv6] section
830 gained a new SendOption= setting for sending arbitrary DHCP
831 options. RequestOptions= has been added to request arbitrary options
832 from the server. UserClass= has been added to set the DHCP user class
835 * systemd-networkd's [DHCPServer] section gained a new set of options
836 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
837 information about these three protocols in the DHCP lease. It also
838 gained support for including "MUD" URLs ("Manufacturer Usage
839 Description"). Support for "MUD" URLs was also added to the LLDP
840 stack, configurable in the [LLDP] section in .network files.
842 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
843 mode. Also, the sections now support a new setting SourceMACAddress=.
845 * systemd-networkd's .netdev files now support a new setting
846 VLANProtocol= in the [Bridge] section that allows configuration of
847 the VLAN protocol to use.
849 * systemd-networkd supports a new Group= setting in the [Link] section
850 of the .network files, to control the link group.
852 * systemd-networkd's [Network] section gained a new
853 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
854 link local address is generated.
856 * A new default .network file is now shipped that matches TUN/TAP
857 devices that begin with "vt-" in their name. Such interfaces will
858 have IP routing onto the host links set up automatically. This is
859 supposed to be used by VM managers to trivially acquire a network
860 interface which is fully set up for host communication, simply by
861 carefully picking an interface name to use.
863 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
864 which sets the route priority for routes specified by the DHCP server.
866 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
867 which configures the vendor class information sent to DHCP server.
869 * The BlackList= settings in .network files' [DHCPv4] and
870 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
871 are still understood to provide compatibility.
873 * networkctl gained the new "forcerenew" command for forcing all DHCP
874 server clients to renew their lease. The interface "status" output
875 will now show numerous additional fields of information about an
876 interface. There are new "up" and "down" commands to bring specific
877 interfaces up or down.
879 * systemd-resolved's DNS= configuration option now optionally accepts a
880 port number (after ":") and a host name (after "#"). When the host
881 name is specified, the DNS-over-TLS certificate is validated to match
882 the specified hostname. Additionally, in case of IPv6 addresses, an
883 interface may be specified (after "%").
885 * systemd-resolved may be configured to forward single-label DNS names.
886 This is not standard-conformant, but may make sense in setups where
887 public DNS servers are not used.
889 * systemd-resolved's DNS-over-TLS support gained SNI validation.
891 * systemd-nspawn's --resolv-conf= switch gained a number of new
892 supported values. Specifically, options starting with "replace-" are
893 like those prefixed "copy-" but replace any existing resolv.conf
894 file. And options ending in "-uplink" and "-stub" can now be used to
895 propagate other flavours of resolv.conf into the container (as
896 defined by systemd-resolved).
898 * The various programs included in systemd can now optionally output
899 their log messages on stderr prefixed with a timestamp, controlled by
900 the $SYSTEMD_LOG_TIME environment variable.
902 * systemctl gained a new "-P" switch that is a shortcut for "--value
905 * "systemctl list-units" and "systemctl list-machines" no longer hide
906 their first output column with --no-legend. To hide the first column,
909 * "systemctl reboot" takes the option "--reboot-argument=".
910 The optional positional argument to "systemctl reboot" is now
911 being deprecated in favor of this option.
913 * systemd-run gained a new switch --slice-inherit. If specified the
914 unit it generates is placed in the same slice as the systemd-run
917 * systemd-journald gained support for zstd compression of large fields
918 in journal files. The hash tables in journal files have been hardened
919 against hash collisions. This is an incompatible change and means
920 that journal files created with new systemd versions are not readable
921 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
922 environment variable for systemd-journald.service is set to 0 this
923 new hardening functionality may be turned off, so that generated
924 journal files remain compatible with older journalctl
927 * journalctl will now include a clickable link in the default output for
928 each log message for which an URL with further documentation is
929 known. This is only supported on terminal emulators that support
930 clickable hyperlinks, and is turned off if a pager is used (since
931 "less" still doesn't support hyperlinks,
932 unfortunately). Documentation URLs may be included in log messages
933 either by including a DOCUMENTATION= journal field in it, or by
934 associating a journal message catalog entry with the log message's
935 MESSAGE_ID, which then carries a "Documentation:" tag.
937 * journald.conf gained a new boolean setting Audit= that may be used to
938 control whether systemd-journald will enable audit during
941 * when systemd-journald's log stream is broken up into multiple lines
942 because the PID of the sender changed this is indicated in the
943 generated log records via the _LINE_BREAK=pid-change field.
945 * journalctl's "-o cat" output mode will now show one or more journal
946 fields specified with --output-fields= instead of unconditionally
947 MESSAGE=. This is useful to retrieve a very specific set of fields
948 without any decoration.
950 * The sd-journal.h API gained two new functions:
951 sd_journal_enumerate_available_unique() and
952 sd_journal_enumerate_available_data() that operate like their
953 counterparts that lack the _available_ in the name, but skip items
954 that cannot be read and processed by the local implementation
955 (i.e. are compressed in an unsupported format or such),
957 * coredumpctl gained a new --file= switch, matching the same one in
958 journalctl: a specific journal file may be specified to read the
961 * coredumps collected by systemd-coredump may now be compressed using
964 * systemd-binfmt gained a new switch --unregister for unregistering all
965 registered entries at once. This is now invoked automatically at
966 shutdown, so that binary formats registered with the "F" flag will
967 not block clean file system unmounting.
969 * systemd-notify's --pid= switch gained new values: "parent", "self",
970 "auto" for controlling which PID to send to the service manager: the
971 systemd-notify process' PID, or the one of the process invoking it.
973 * systemd-logind's Session bus object learnt a new method call
974 SetType() for temporarily updating the session type of an already
975 allocated session. This is useful for upgrading tty sessions to
976 graphical ones once a compositor is invoked.
978 * systemd-socket-proxy gained a new switch --exit-idle-time= for
979 configuring an exit-on-idle time.
981 * systemd-repart's --empty= setting gained a new value "create". If
982 specified a new empty regular disk image file is created under the
983 specified name. Its size may be specified with the new --size=
984 option. The latter is also supported without the "create" mode, in
985 order to grow existing disk image files to the specified size. These
986 two new options are useful when creating or manipulating disk images
987 instead of operating on actual block devices.
989 * systemd-repart drop-ins now support a new UUID= setting to control
990 the UUID to assign to a newly created partition.
992 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
995 * systemd-repart's Label= setting now support the usual, simple
998 * systemd-homed's LUKS backend gained the ability to discard empty file
999 system blocks automatically when the user logs out. This is enabled
1000 by default to ensure that home directories take minimal space when
1001 logged out but get full size guarantees when logged in. This may be
1002 controlled with the new --luks-offline-discard= switch to homectl.
1004 * If systemd-homed detects that /home/ is encrypted as a whole it will
1005 now default to the directory or subvolume backends instead of the
1006 LUKS backend, in order to avoid double encryption. The default
1007 storage and file system may now be configured explicitly, too, via
1008 the new /etc/systemd/homed.conf configuration file.
1010 * systemd-homed now supports unlocking home directories with FIDO2
1011 security tokens that support the 'hmac-secret' extension, in addition
1012 to the existing support for PKCS#11 security token unlocking
1013 support. Note that many recent hardware security tokens support both
1014 interfaces. The FIDO2 support is accessible via homectl's
1015 --fido2-device= option.
1017 * homectl's --pkcs11-uri= setting now accepts two special parameters:
1018 if "auto" is specified and only one suitable PKCS#11 security token
1019 is plugged in, its URL is automatically determined and enrolled for
1020 unlocking the home directory. If "list" is specified a brief table of
1021 suitable PKCS#11 security tokens is shown. Similar, the new
1022 --fido2-device= option also supports these two special values, for
1023 automatically selecting and listing suitable FIDO2 devices.
1025 * The /etc/crypttab tmp option now optionally takes an argument
1026 selecting the file system to use. Moreover, the default is now
1027 changed from ext2 to ext4.
1029 * There's a new /etc/crypttab option "keyfile-erase". If specified the
1030 key file listed in the same line is removed after use, regardless if
1031 volume activation was successful or not. This is useful if the key
1032 file is only acquired transiently at runtime and shall be erased
1033 before the system continues to boot.
1035 * There's also a new /etc/crypttab option "try-empty-password". If
1036 specified, before asking the user for a password it is attempted to
1037 unlock the volume with an empty password. This is useful for
1038 installing encrypted images whose password shall be set on first boot
1039 instead of at installation time.
1041 * systemd-cryptsetup will now attempt to load the keys to unlock
1042 volumes with automatically from files in
1043 /etc/cryptsetup-keys.d/<volume>.key and
1044 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
1046 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
1047 /etc/crypttab, during boot.
1049 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
1050 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
1053 * A new generator systemd-xdg-autostart-generator has been added. It
1054 generates systemd unit files from XDG autostart .desktop files, and
1055 may be used to let the systemd user instance manage services that are
1056 started automatically as part of the desktop session.
1058 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
1059 to query and change the firmware's 'reboot into firmware' setup flag.
1061 * systemd-firstboot gained a new switch --kernel-command-line= that may
1062 be used to initialize the /etc/kernel/cmdline file of the image. It
1063 also gained a new switch --root-password-hashed= which is like
1064 --root-password= but accepts a pre-hashed UNIX password as
1065 argument. The new option --delete-root-password may be used to unset
1066 any password for the root user (dangerous!). The --root-shell= switch
1067 may be used to control the shell to use for the root account. A new
1068 --force option may be used to override any already set settings with
1069 the parameters specified on the command line (by default, the tool
1070 will not override what has already been set before, i.e. is purely
1073 * systemd-firstboot gained support for a new --image= switch, which is
1074 similar to --root= but accepts the path to a disk image file, on
1075 which it then operates.
1077 * A new sd-path.h API has been added to libsystemd. It provides a
1078 simple API for retrieving various search paths and primary
1079 directories for various resources.
1081 * A new call sd_notify_barrier() has been added to the sd-daemon.h
1082 API. The call will block until all previously sent sd_notify()
1083 messages have been processed by the service manager. This is useful
1084 to remove races caused by a process already having disappeared at the
1085 time a notification message is processed by the service manager,
1086 making correct attribution impossible. The systemd-notify tool will
1087 now make use of this call implicitly, but this can be turned off again
1088 via the new --no-block switch.
1090 * When sending a file descriptor (fd) to the service manager to keep
1091 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
1092 may be specified. If passed the service manager will refrain from
1093 poll()ing on the file descriptor. Traditionally (and when the
1094 parameter is not specified), the service manager will poll it for
1095 POLLHUP or POLLERR events, and immediately close the fds in that
1098 * The service manager (PID1) gained a new D-Bus method call
1099 SetShowStatus() which may be used to control whether it shall show
1100 boot-time status output on the console. This method has a similar
1101 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
1103 * The sd-bus API gained a number of convenience functions that take
1104 va_list arguments rather than "...". For example, there's now
1105 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
1106 it easier to build wrappers that accept variadic arguments and want
1107 to pass a ready va_list structure to sd-bus.
1109 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
1110 flag which alters how the userdata pointer to pass to the callbacks
1111 is determined. When the flag is set, the offset field is converted
1112 as-is into a pointer, without adding it to the object pointer the
1113 vtable is associated with.
1115 * sd-bus now exposes four new functions:
1116 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
1117 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
1118 validate strings to check if they qualify as various D-Bus concepts.
1120 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
1121 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
1122 that simplify adding argument names to D-Bus methods and signals.
1124 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
1126 * Various D-Bus APIs of systemd daemons now have man pages that
1127 document the methods, signals and properties.
1129 * The expectations on user/group name syntax are now documented in
1130 detail; documentation on how classic home directories may be
1131 converted into home directories managed by homed has been added;
1132 documentation regarding integration of homed/userdb functionality in
1133 desktops has been added:
1135 https://systemd.io/USER_NAMES
1136 https://systemd.io/CONVERTING_TO_HOMED
1137 https://systemd.io/USERDB_AND_DESKTOPS
1139 * Documentation for the on-disk Journal file format has been updated
1140 and has now moved to:
1142 https://systemd.io/JOURNAL_FILE_FORMAT
1144 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
1145 has been extended by a set of environment variables that expose
1146 select fields from the host's os-release file to the container
1147 payload. Similarly, host's os-release files can be mounted into the
1148 container underneath /run/host. Together, those mechanisms provide a
1149 standardized way to expose information about the host to the
1150 container payload. Both interfaces are implemented in systemd-nspawn.
1152 * All D-Bus services shipped in systemd now implement the generic
1153 LogControl1 D-Bus API which allows clients to change log level +
1154 target of the service during runtime.
1156 * Only relevant for developers: the mkosi.default symlink has been
1157 dropped from version control. Please create a symlink to one of the
1158 distribution-specific defaults in .mkosi/ based on your preference.
1160 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
1161 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
1162 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
1163 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
1164 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
1165 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
1166 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
1167 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
1168 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
1169 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
1170 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
1171 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
1172 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
1173 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
1174 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
1175 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
1176 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
1177 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
1178 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
1179 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
1180 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
1181 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
1182 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
1183 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
1184 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
1185 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
1186 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
1187 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
1188 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
1189 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
1190 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
1191 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
1192 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
1193 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
1194 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
1195 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
1196 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
1197 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
1199 – Warsaw, 2020-07-30
1203 * A new tool "systemd-repart" has been added, that operates as an
1204 idempotent declarative repartitioner for GPT partition tables.
1205 Specifically, a set of partitions that must or may exist can be
1206 configured via drop-in files, and during every boot the partition
1207 table on disk is compared with these files, creating missing
1208 partitions or growing existing ones based on configurable relative
1209 and absolute size constraints. The tool is strictly incremental,
1210 i.e. does not delete, shrink or move partitions, but only adds and
1211 grows them. The primary use-case is OS images that ship in minimized
1212 form, that on first boot are grown to the size of the underlying
1213 block device or augmented with additional partitions. For example,
1214 the root partition could be extended to cover the whole disk, or a
1215 swap or /home partitions could be added on first boot. It can also be
1216 used for systems that use an A/B update scheme but ship images with
1217 just the A partition, with B added on first boot. The tool is
1218 primarily intended to be run in the initrd, shortly before
1219 transitioning into the host OS, but can also be run after the
1220 transition took place. It automatically discovers the disk backing
1221 the root file system, and should hence not require any additional
1222 configuration besides the partition definition drop-ins. If no
1223 configuration drop-ins are present, no action is taken.
1225 * A new component "userdb" has been added, along with a small daemon
1226 "systemd-userdb.service" and a client tool "userdbctl". The framework
1227 allows defining rich user and group records in a JSON format,
1228 extending on the classic "struct passwd" and "struct group"
1229 structures. Various components in systemd have been updated to
1230 process records in this format, including systemd-logind and
1231 pam-systemd. The user records are intended to be extensible, and
1232 allow setting various resource management, security and runtime
1233 parameters that shall be applied to processes and sessions of the
1234 user as they log in. This facility is intended to allow associating
1235 such metadata directly with user/group records so that they can be
1236 produced, extended and consumed in unified form. We hope that
1237 eventually frameworks such as sssd will generate records this way, so
1238 that for the first time resource management and various other
1239 per-user settings can be configured in LDAP directories and then
1240 provided to systemd (specifically to systemd-logind and pam-system)
1241 to apply on login. For further details see:
1243 https://systemd.io/USER_RECORD
1244 https://systemd.io/GROUP_RECORD
1245 https://systemd.io/USER_GROUP_API
1247 * A small new service systemd-homed.service has been added, that may be
1248 used to securely manage home directories with built-in encryption.
1249 The complete user record data is unified with the home directory,
1250 thus making home directories naturally migratable. Its primary
1251 back-end is based on LUKS volumes, but fscrypt, plain directories,
1252 and other storage schemes are also supported. This solves a couple of
1253 problems we saw with traditional ways to manage home directories, in
1254 particular when it comes to encryption. For further discussion of
1255 this, see the video of Lennart's talk at AllSystemsGo! 2019:
1257 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
1259 For further details about the format and expectations on home
1260 directories this new daemon makes, see:
1262 https://systemd.io/HOME_DIRECTORY
1264 * systemd-journald is now multi-instantiable. In addition to the main
1265 instance systemd-journald.service there's now a template unit
1266 systemd-journald@.service, with each instance defining a new named
1267 log 'namespace' (whose name is specified via the instance part of the
1268 unit name). A new unit file setting LogNamespace= has been added,
1269 taking such a namespace name, that assigns services to the specified
1270 log namespaces. As each log namespace is serviced by its own
1271 independent journal daemon, this functionality may be used to improve
1272 performance and increase isolation of applications, at the price of
1273 losing global message ordering. Each instance of journald has a
1274 separate set of configuration files, with possibly different disk
1275 usage limitations and other settings.
1277 journalctl now takes a new option --namespace= to show logs from a
1278 specific log namespace. The sd-journal.h API gained
1279 sd_journal_open_namespace() for opening the log stream of a specific
1280 log namespace. systemd-journald also gained the ability to exit on
1281 idle, which is useful in the context of log namespaces, as this means
1282 log daemons for log namespaces can be activated automatically on
1283 demand and will stop automatically when no longer used, minimizing
1286 * When systemd-tmpfiles copies a file tree using the 'C' line type it
1287 will now label every copied file according to the SELinux database.
1289 * When systemd/PID 1 detects it is used in the initrd it will now boot
1290 into initrd.target rather than default.target by default. This should
1291 make it simpler to build initrds with systemd as for many cases the
1292 only difference between a host OS image and an initrd image now is
1293 the presence of the /etc/initrd-release file.
1295 * A new kernel command line option systemd.cpu_affinity= is now
1296 understood. It's equivalent to the CPUAffinity= option in
1297 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
1298 itself and the default for all other processes.
1300 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
1301 equivalent), the SELinux database is now reloaded, ensuring that
1302 sockets and other file system objects are generated taking the new
1303 database into account.
1305 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
1306 "quiet" has been changed to imply that instead of
1307 "systemd.show-status=auto". In this mode, only messages about errors
1308 and significant delays in boot are shown on the console.
1310 * The sd-event.h API gained native support for the new Linux "pidfd"
1311 concept. This permits watching processes using file descriptors
1312 instead of PID numbers, which fixes a number of races and makes
1313 process supervision more robust and efficient. All of systemd's
1314 components will now use pidfds if the kernel supports it for process
1315 watching, with the exception of PID 1 itself, unfortunately. We hope
1316 to move PID 1 to exclusively using pidfds too eventually, but this
1317 requires some more kernel work first. (Background: PID 1 watches
1318 processes using waitid() with the P_ALL flag, and that does not play
1319 together nicely with pidfds yet.)
1321 * Closely related to this, the sd-event.h API gained two new calls
1322 sd_event_source_send_child_signal() (for sending a signal to a
1323 watched process) and sd_event_source_get_child_process_own() (for
1324 marking a process so that it is killed automatically whenever the
1325 event source watching it is freed).
1327 * systemd-networkd gained support for configuring Token Bucket Filter
1328 (TBF) parameters in its qdisc configuration support. Similarly,
1329 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
1330 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
1332 * systemd-networkd gained support for Intermediate Functional Block
1333 (IFB) network devices.
1335 * systemd-networkd gained support for configuring multi-path IP routes,
1336 using the new MultiPathRoute= setting in the [Route] section.
1338 * systemd-networkd's DHCPv4 client has been updated to support a new
1339 SendDecline= option. If enabled, duplicate address detection is done
1340 after a DHCP offer is received from the server. If a conflict is
1341 detected, the address is declined. The DHCPv4 client also gained
1342 support for a new RouteMTUBytes= setting that allows to configure the
1343 MTU size to be used for routes generated from DHCPv4 leases.
1345 * The PrefixRoute= setting in systemd-networkd's [Address] section of
1346 .network files has been deprecated, and replaced by AddPrefixRoute=,
1347 with its sense inverted.
1349 * The Gateway= setting of [Route] sections of .network files gained
1350 support for a special new value "_dhcp". If set, the configured
1351 static route uses the gateway host configured via DHCP.
1353 * New User= and SuppressPrefixLength= settings have been implemented
1354 for the [RoutingPolicyRule] section of .network files to configure
1355 source routing based on UID ranges and prefix length, respectively.
1357 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
1358 D-Bus message object as "sensitive". Those objects are erased from
1359 memory when they are freed. This concept is intended to be used for
1360 messages that contain security sensitive data. A new flag
1361 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
1362 in sd-bus vtables, causing any incoming and outgoing messages of
1363 those methods to be implicitly marked as "sensitive".
1365 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
1366 contents of a message (or parts thereof) to standard output for
1369 * systemd-sysusers gained support for creating users with the primary
1370 group named differently than the user.
1372 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
1373 gained support for growing XFS partitions. Previously it supported
1374 only ext4 and btrfs partitions.
1376 * The support for /etc/crypttab gained a new x-initrd.attach option. If
1377 set, the specified encrypted volume is unlocked already in the
1378 initrd. This concept corresponds to the x-initrd.mount option in
1381 * systemd-cryptsetup gained native support for unlocking encrypted
1382 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
1383 encryption of volumes to YubiKeys. This is exposed in the new
1384 pkcs11-uri= option in /etc/crypttab.
1386 * The /etc/fstab support in systemd now supports two new mount options
1387 x-systemd.{required,wanted}-by=, for explicitly configuring the units
1388 that the specified mount shall be pulled in by, in place of
1389 the usual local-fs.target/remote-fs.target.
1391 * The https://systemd.io/ web site has been relaunched, directly
1392 populated with most of the documentation included in the systemd
1393 repository. systemd also acquired a new logo, thanks to Tobias
1396 * systemd-udevd gained support for managing "alternative" network
1397 interface names, as supported by new Linux kernels. For the first
1398 time this permits assigning multiple (and longer!) names to a network
1399 interface. systemd-udevd will now by default assign the names
1400 generated via all supported naming schemes to each interface. This
1401 may be further tweaked with .link files and the AlternativeName= and
1402 AlternativeNamesPolicy= settings. Other components of systemd have
1403 been updated to support the new alternative names wherever
1404 appropriate. For example, systemd-nspawn will now generate
1405 alternative interface names for the host-facing side of container
1406 veth links based on the full container name without truncation.
1408 * systemd-nspawn interface naming logic has been updated in another way
1409 too: if the main interface name (i.e. as opposed to new-style
1410 "alternative" names) based on the container name is truncated, a
1411 simple hashing scheme is used to give different interface names to
1412 multiple containers whose names all begin with the same prefix. Since
1413 this changes the primary interface names pointing to containers if
1414 truncation happens, the old scheme may still be requested by
1415 selecting an older naming scheme, via the net.naming-scheme= kernel
1416 command line option.
1418 * PrivateUsers= in service files now works in services run by the
1419 systemd --user per-user instance of the service manager.
1421 * A new per-service sandboxing option ProtectClock= has been added that
1422 locks down write access to the system clock. It takes away device
1423 node access to /dev/rtc as well as the system calls that set the
1424 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
1425 Note that this option does not affect access to auxiliary services
1426 that allow changing the clock, for example access to
1429 * The systemd-id128 tool gained a new "show" verb for listing or
1430 resolving a number of well-known UUIDs/128bit IDs, currently mostly
1431 GPT partition table types.
1433 * The Discoverable Partitions Specification has been updated to support
1434 /var and /var/tmp partition discovery. Support for this has been
1435 added to systemd-gpt-auto-generator. For details see:
1437 https://systemd.io/DISCOVERABLE_PARTITIONS
1439 * "systemctl list-unit-files" has been updated to show a new column
1440 with the suggested enablement state based on the vendor preset files
1441 for the respective units.
1443 * "systemctl" gained a new option "--with-dependencies". If specified
1444 commands such as "systemctl status" or "systemctl cat" will now show
1445 all specified units along with all units they depend on.
1447 * networkctl gained support for showing per-interface logs in its
1450 * systemd-networkd-wait-online gained support for specifying the maximum
1451 operational state to wait for, and to wait for interfaces to
1454 * The [Match] section of .link and .network files now supports a new
1455 option PermanentMACAddress= which may be used to check against the
1456 permanent MAC address of a network device even if a randomized MAC
1459 * The [TrafficControlQueueingDiscipline] section in .network files has
1460 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
1461 dropped from the individual setting names.
1463 * Any .link and .network files that have an empty [Match] section (this
1464 also includes empty and commented-out files) will now be
1465 rejected. systemd-udev and systemd-networkd started warning about
1466 such files in version 243.
1468 * systemd-logind will now validate access to the operation of changing
1469 the virtual terminal via a polkit action. By default, only users
1470 with at least one session on a local VT are granted permission.
1472 * When systemd sets up PAM sessions that invoked service processes
1473 shall run in, the pam_setcred() API is now invoked, thus permitting
1474 PAM modules to set additional credentials for the processes.
1476 * portablectl attach/detach verbs now accept --now and --enable options
1477 to combine attachment with enablement and invocation, or detachment
1478 with stopping and disablement.
1480 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
1481 fixed, which in turn exposed bugs in unit configuration of services
1482 which have Type=oneshot and should only run once, but do not have
1483 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
1484 service may be started again after exiting successfully, for example
1485 as a dependency in another transaction. Affected services included
1486 some internal systemd services (most notably
1487 systemd-vconsole-setup.service, which was updated to have
1488 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
1489 plymouth has been suitably updated or patched before upgrading to
1490 this systemd release. See
1491 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
1492 additional discussion.
1494 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
1495 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
1496 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
1497 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
1498 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
1499 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
1500 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
1501 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
1502 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
1503 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
1504 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
1505 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
1506 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
1507 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
1508 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
1509 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
1510 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
1511 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
1512 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
1513 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
1514 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
1515 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
1516 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
1517 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
1520 – Warsaw, 2020-03-06
1524 * Support for the cpuset cgroups v2 controller has been added.
1525 Processes may be restricted to specific CPUs using the new
1526 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
1527 AllowedMemoryNodes= setting.
1529 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
1530 now be configured using a new RestartKillSignal= setting. This
1531 allows units which signals to request termination to implement
1532 different behaviour when stopping in preparation for a restart.
1534 * "systemctl clean" may now be used also for socket, mount, and swap
1537 * systemd will also read configuration options from the EFI variable
1538 SystemdOptions. This may be used to configure systemd behaviour when
1539 modifying the kernel command line is inconvenient, but configuration
1540 on disk is read too late, for example for the options related to
1541 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
1542 set the EFI variable.
1544 * systemd will now disable printk ratelimits in early boot. This should
1545 allow us to capture more logs from the early boot phase where normal
1546 storage is not available and the kernel ring buffer is used for
1547 logging. Configuration on the kernel command line has higher priority
1548 and overrides the systemd setting.
1550 systemd programs which log to /dev/kmsg directly use internal
1551 ratelimits to prevent runaway logging. (Normally this is only used
1552 during early boot, so in practice this change has very little
1555 * Unit files now support top level dropin directories of the form
1556 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
1557 that affects all corresponding unit files.
1559 * systemctl gained support for 'stop --job-mode=triggering' which will
1560 stop the specified unit and any units which could trigger it.
1562 * Unit status display now includes units triggering and triggered by
1563 the unit being shown.
1565 * The RuntimeMaxSec= setting is now supported by scopes, not just
1566 .service units. This is particularly useful for PAM sessions which
1567 create a scope unit for the user login. systemd.runtime_max_sec=
1568 setting may used with the pam_systemd module to limit the duration
1569 of the PAM session, for example for time-limited logins.
1571 * A new @pkey system call group is now defined to make it easier to
1572 allow-list memory protection syscalls for containers and services
1573 which need to use them.
1575 * systemd-udevd: removed the 30s timeout for killing stale workers on
1576 exit. systemd-udevd now waits for workers to finish. The hard-coded
1577 exit timeout of 30s was too short for some large installations, where
1578 driver initialization could be prematurely interrupted during initrd
1579 processing if the root file system had been mounted and init was
1580 preparing to switch root. If udevd is run without systemd and workers
1581 are hanging while udevd receives an exit signal, udevd will now exit
1582 when udev.event_timeout is reached for the last hanging worker. With
1583 systemd, the exit timeout can additionally be configured using
1584 TimeoutStopSec= in systemd-udevd.service.
1586 * udev now provides a program (fido_id) that identifies FIDO CTAP1
1587 ("U2F")/CTAP2 security tokens based on the usage declared in their
1588 report and descriptor and outputs suitable environment variables.
1589 This replaces the externally maintained allow lists of all known
1590 security tokens that were used previously.
1592 * Automatically generated autosuspend udev rules for allow-listed
1593 devices have been imported from the Chromium OS project. This should
1594 improve power saving with many more devices.
1596 * udev gained a new "CONST{key}=value" setting that allows matching
1597 against system-wide constants without forking a helper binary.
1598 Currently "arch" and "virt" keys are supported.
1600 * udev now opens CDROMs in non-exclusive mode when querying their
1601 capabilities. This should fix issues where other programs trying to
1602 use the CDROM cannot gain access to it, but carries a risk of
1603 interfering with programs writing to the disk, if they did not open
1604 the device in exclusive mode as they should.
1606 * systemd-networkd does not create a default route for IPv4 link local
1607 addressing anymore. The creation of the route was unexpected and was
1608 breaking routing in various cases, but people who rely on it being
1609 created implicitly will need to adjust. Such a route may be requested
1610 with DefaultRouteOnDevice=yes.
1612 Similarly, systemd-networkd will not assign a link-local IPv6 address
1613 when IPv6 link-local routing is not enabled.
1615 * Receive and transmit buffers may now be configured on links with
1616 the new RxBufferSize= and TxBufferSize= settings.
1618 * systemd-networkd may now advertise additional IPv6 routes. A new
1619 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
1622 * systemd-networkd may now configure "next hop" routes using the
1623 [NextHop] section and Gateway= and Id= settings.
1625 * systemd-networkd will now retain DHCP config on restarts by default
1626 (but this may be overridden using the KeepConfiguration= setting).
1627 The default for SendRelease= has been changed to true.
1629 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
1630 received from the server.
1632 The client will use the received SIP server list if UseSIP=yes is
1635 The client may be configured to request specific options from the
1636 server using a new RequestOptions= setting.
1638 The client may be configured to send arbitrary options to the server
1639 using a new SendOption= setting.
1641 A new IPServiceType= setting has been added to configure the "IP
1642 service type" value used by the client.
1644 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
1645 request prefix hints in the DHCPv6 solicitation.
1647 * The DHCPv4 server may be configured to send arbitrary options using
1648 a new SendOption= setting.
1650 * The DHCPv4 server may now be configured to emit SIP server list using
1651 the new EmitSIP= and SIP= settings.
1653 * systemd-networkd and networkctl may now renew DHCP leases on demand.
1654 networkctl has a new 'networkctl renew' verb.
1656 * systemd-networkd may now reconfigure links on demand. networkctl
1657 gained two new verbs: "reload" will reload the configuration, and
1658 "reconfigure DEVICE…" will reconfigure one or more devices.
1660 * .network files may now match on SSID and BSSID of a wireless network,
1661 i.e. the access point name and hardware address using the new SSID=
1662 and BSSID= options. networkctl will display the current SSID and
1663 BSSID for wireless links.
1665 .network files may also match on the wireless network type using the
1666 new WLANInterfaceType= option.
1668 * systemd-networkd now includes default configuration that enables
1669 link-local addressing when connected to an ad-hoc wireless network.
1671 * systemd-networkd may configure the Traffic Control queueing
1672 disciplines in the kernel using the new
1673 [TrafficControlQueueingDiscipline] section and Parent=,
1674 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
1675 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
1676 NetworkEmulatorDuplicateRate= settings.
1678 * systemd-tmpfiles gained a new w+ setting to append to files.
1680 * systemd-analyze dump will now report when the memory configuration in
1681 the kernel does not match what systemd has configured (usually,
1682 because some external program has modified the kernel configuration
1685 * systemd-analyze gained a new --base-time= switch instructs the
1686 'calendar' verb to resolve times relative to that timestamp instead
1687 of the present time.
1689 * journalctl --update-catalog now produces deterministic output (making
1690 reproducible image builds easier).
1692 * A new devicetree-overlay setting is now documented in the Boot Loader
1695 * The default value of the WatchdogSec= setting used in systemd
1696 services (the ones bundled with the project itself) may be set at
1697 configuration time using the -Dservice-watchdog= setting. If set to
1698 empty, the watchdogs will be disabled.
1700 * systemd-resolved validates IP addresses in certificates now when GnuTLS
1703 * libcryptsetup >= 2.0.1 is now required.
1705 * A configuration option -Duser-path= may be used to override the $PATH
1706 used by the user service manager. The default is again to use the same
1707 path as the system manager.
1709 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
1710 outputting the 128bit IDs in UUID format (i.e. in the "canonical
1713 * Service units gained a new sandboxing option ProtectKernelLogs= which
1714 makes sure the program cannot get direct access to the kernel log
1715 buffer anymore, i.e. the syslog() system call (not to be confused
1716 with the API of the same name in libc, which is not affected), the
1717 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
1718 inaccessible to the service. It's recommended to enable this setting
1719 for all services that should not be able to read from or write to the
1720 kernel log buffer, which are probably almost all.
1722 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
1723 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
1724 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
1725 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
1726 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
1727 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
1728 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
1729 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
1730 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
1731 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
1732 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
1733 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
1734 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
1735 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
1736 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
1737 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
1738 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
1739 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
1740 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
1741 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
1742 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
1743 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
1744 Zach Smith, Zbigniew Jędrzejewski-Szmek
1746 – Warsaw, 2019-11-29
1750 * This release enables unprivileged programs (i.e. requiring neither
1751 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
1752 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
1753 kernel for the whole UNIX group range, i.e. all processes. This
1754 change should be reasonably safe, as the kernel support for it was
1755 specifically implemented to allow safe access to ICMP Echo for
1756 processes lacking any privileges. If this is not desirable, it can be
1757 disabled again by setting the parameter to "1 0".
1759 * Previously, filters defined with SystemCallFilter= would have the
1760 effect that any calling of an offending system call would terminate
1761 the calling thread. This behaviour never made much sense, since
1762 killing individual threads of unsuspecting processes is likely to
1763 create more problems than it solves. With this release the default
1764 action changed from killing the thread to killing the whole
1765 process. For this to work correctly both a kernel version (>= 4.14)
1766 and a libseccomp version (>= 2.4.0) supporting this new seccomp
1767 action is required. If an older kernel or libseccomp is used the old
1768 behaviour continues to be used. This change does not affect any
1769 services that have no system call filters defined, or that use
1770 SystemCallErrorNumber= (and thus see EPERM or another error instead
1771 of being killed when calling an offending system call). Note that
1772 systemd documentation always claimed that the whole process is
1773 killed. With this change behaviour is thus adjusted to match the
1776 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
1777 4194304 by default, i.e. the full 22bit range the kernel allows, up
1778 from the old 16bit range. This should improve security and
1779 robustness, as PID collisions are made less likely (though certainly
1780 still possible). There are rumours this might create compatibility
1781 problems, though at this moment no practical ones are known to
1782 us. Downstream distributions are hence advised to undo this change in
1783 their builds if they are concerned about maximum compatibility, but
1784 for everybody else we recommend leaving the value bumped. Besides
1785 improving security and robustness this should also simplify things as
1786 the maximum number of allowed concurrent tasks was previously bounded
1787 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
1788 only a single knob is left ("kernel.threads-max"). There have been
1789 concerns that usability is affected by this change because larger PID
1790 numbers are harder to type, but we believe the change from 5 digits
1791 to 7 digits doesn't hamper usability.
1793 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
1794 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
1795 hierarchically set default memory protection values for a particular
1796 subtree of the unit hierarchy.
1798 * Memory protection directives can now take a value of zero, allowing
1799 explicit opting out of a default value propagated by an ancestor.
1801 * systemd now defaults to the "unified" cgroup hierarchy setup during
1802 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
1803 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
1804 change reflects the fact that cgroupsv2 support has matured
1805 substantially in both systemd and in the kernel, and is clearly the
1806 way forward. Downstream production distributions might want to
1807 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
1808 their builds as unfortunately the popular container managers have not
1809 caught up with the kernel API changes.
1811 * Man pages are not built by default anymore (html pages were already
1812 disabled by default), to make development builds quicker. When
1813 building systemd for a full installation with documentation, meson
1814 should be called with -Dman=true and/or -Dhtml=true as appropriate.
1815 The default was changed based on the assumption that quick one-off or
1816 repeated development builds are much more common than full optimized
1817 builds for installation, and people need to pass various other
1818 options to when doing "proper" builds anyway, so the gain from making
1819 development builds quicker is bigger than the one time disruption for
1822 Two scripts are created in the *build* directory to generate and
1823 preview man and html pages on demand, e.g.:
1825 build/man/man systemctl
1826 build/man/html systemd.index
1828 * libidn2 is used by default if both libidn2 and libidn are installed.
1829 Please use -Dlibidn=true if libidn is preferred.
1831 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
1832 big-endian machines. Before, bytes were written and read in native
1833 machine order as exposed by the native libc __cpu_mask interface.
1834 Now, little-endian order is always used (CPUs 0–7 are described by
1835 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
1836 This change fixes D-Bus calls that cross endianness boundary.
1838 The presentation format used for CPUAffinity= by "systemctl show" and
1839 "systemd-analyze dump" is changed to present CPU indices instead of
1840 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
1841 shown as CPUAffinity=03000000000000000000000000000… (on
1842 little-endian) or CPUAffinity=00000000000000300000000000000… (on
1843 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
1844 input format. The maximum integer that will be printed in the new
1845 format is 8191 (four digits), while the old format always used a very
1846 long number (with the length varying by architecture), so they can be
1847 unambiguously distinguished.
1849 * /usr/sbin/halt.local is no longer supported. Implementation in
1850 distributions was inconsistent and it seems this functionality was
1853 To replace this functionality, users should:
1854 - either define a new unit and make it a dependency of final.target
1855 (systemctl add-wants final.target my-halt-local.service)
1856 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
1857 and ensure that it accepts "halt", "poweroff", "reboot", and
1858 "kexec" as an argument, see the description in systemd-shutdown(8).
1860 * When a [Match] section in .link or .network file is empty (contains
1861 no match patterns), a warning will be emitted. Please add any "match
1862 all" pattern instead, e.g. OriginalName=* or Name=* in case all
1863 interfaces should really be matched.
1865 * A new setting NUMAPolicy= may be used to set process memory
1866 allocation policy. This setting can be specified in
1867 /etc/systemd/system.conf and hence will set the default policy for
1868 PID1. The default policy can be overridden on a per-service
1869 basis. The related setting NUMAMask= is used to specify NUMA node
1870 mask that should be associated with the selected policy.
1872 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
1873 generates when processes it manages are reaching their memory limits,
1874 and will place their units in a special state, and optionally kill or
1875 stop the whole unit.
1877 * The service manager will now expose bus properties for the IO
1878 resources used by units. This information is also shown in "systemctl
1879 status" now (for services that have IOAccounting=yes set). Moreover,
1880 the IO accounting data is included in the resource log message
1881 generated whenever a unit stops.
1883 * Units may now configure an explicit timeout to wait for when killed
1884 with SIGABRT, for example when a service watchdog is hit. Previously,
1885 the regular TimeoutStopSec= timeout was applied in this case too —
1886 now a separate timeout may be set using TimeoutAbortSec=.
1888 * Services may now send a special WATCHDOG=trigger message with
1889 sd_notify() to trigger an immediate "watchdog missed" event, and thus
1890 trigger service termination. This is useful both for testing watchdog
1891 handling, but also for defining error paths in services, that shall
1892 be handled the same way as watchdog events.
1894 * There are two new per-unit settings IPIngressFilterPath= and
1895 IPEgressFilterPath= which allow configuration of a BPF program
1896 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
1897 to apply to the IP packet ingress/egress path of all processes of a
1898 unit. This is useful to allow running systemd services with BPF
1899 programs set up externally.
1901 * systemctl gained a new "clean" verb for removing the state, cache,
1902 runtime or logs directories of a service while it is terminated. The
1903 new verb may also be used to remove the state maintained on disk for
1904 timer units that have Persistent= configured.
1906 * During the last phase of shutdown systemd will now automatically
1907 increase the log level configured in the "kernel.printk" sysctl so
1908 that any relevant loggable events happening during late shutdown are
1909 made visible. Previously, loggable events happening so late during
1910 shutdown were generally lost if the "kernel.printk" sysctl was set to
1911 high thresholds, as regular logging daemons are terminated at that
1912 time and thus nothing is written to disk.
1914 * If processes terminated during the last phase of shutdown do not exit
1915 quickly systemd will now show their names after a short time, to make
1916 debugging easier. After a longer timeout they are forcibly killed,
1919 * journalctl (and the other tools that display logs) will now highlight
1920 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
1921 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
1922 are now shown in blue color, to separate them visually from regular
1923 logs. References to configuration files are now turned into clickable
1924 links on terminals that support that.
1926 * systemd-journald will now stop logging to /var/log/journal during
1927 shutdown when /var/ is on a separate mount, so that it can be
1928 unmounted safely during shutdown.
1930 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
1932 * systemd-resolved "Cache=" configuration option in resolved.conf has
1933 been extended to also accept the 'no-negative' value. Previously,
1934 only a boolean option was allowed (yes/no), having yes as the
1935 default. If this option is set to 'no-negative', negative answers are
1936 not cached while the old cache heuristics are used positive answers.
1937 The default remains unchanged.
1939 * The predictable naming scheme for network devices now supports
1940 generating predictable names for "netdevsim" devices.
1942 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
1945 Those two changes form a new net.naming-policy-scheme= entry.
1946 Distributions which want to preserve naming stability may want to set
1947 the -Ddefault-net-naming-scheme= configuration option.
1949 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
1950 interfaces natively.
1952 * systemd-networkd's bridge FDB support now allows configuration of a
1953 destination address for each entry (Destination=), as well as the
1954 VXLAN VNI (VNI=), as well as an option to declare what an entry is
1955 associated with (AssociatedWith=).
1957 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
1958 option for configuring the maximum number of DHCP lease requests. It
1959 also learnt a new BlackList= option for deny-listing DHCP servers (a
1960 similar setting has also been added to the IPv6 RA client), as well
1961 as a SendRelease= option for configuring whether to send a DHCP
1962 RELEASE message when terminating.
1964 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
1965 separately in the [DHCPv4] and [DHCPv6] sections.
1967 * systemd-networkd's DHCP support will now optionally create an
1968 implicit host route to the DNS server specified in the DHCP lease, in
1969 addition to the routes listed explicitly in the lease. This should
1970 ensure that in multi-homed systems DNS traffic leaves the systems on
1971 the interface that acquired the DNS server information even if other
1972 routes such as default routes exist. This behaviour may be turned on
1973 with the new RoutesToDNS= option.
1975 * systemd-networkd's VXLAN support gained a new option
1976 GenericProtocolExtension= for enabling VXLAN Generic Protocol
1977 Extension support, as well as IPDoNotFragment= for setting the IP
1978 "Don't fragment" bit on outgoing packets. A similar option has been
1979 added to the GENEVE support.
1981 * In systemd-networkd's [Route] section you may now configure
1982 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
1983 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
1984 propagation. The Type= setting now supports local, broadcast,
1985 anycast, multicast, any, xresolve routes, too.
1987 * systemd-networkd's [Network] section learnt a new option
1988 DefaultRouteOnDevice= for automatically configuring a default route
1989 onto the network device.
1991 * systemd-networkd's bridging support gained two new options ProxyARP=
1992 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
1993 MulticastRouter= for configuring multicast routing behaviour. A new
1994 option MulticastIGMPVersion= may be used to change bridge's multicast
1995 Internet Group Management Protocol (IGMP) version.
1997 * systemd-networkd's FooOverUDP support gained the ability to configure
1998 local and peer IP addresses via Local= and Peer=. A new option
1999 PeerPort= may be used to configure the peer's IP port.
2001 * systemd-networkd's TUN support gained a new setting VnetHeader= for
2002 tweaking Generic Segment Offload support.
2004 * The address family for policy rules may be specified using the new
2005 Family= option in the [RoutingPolicyRule] section.
2007 * networkctl gained a new "delete" command for removing virtual network
2008 devices, as well as a new "--stats" switch for showing device
2011 * networkd.conf gained a new setting SpeedMeter= and
2012 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
2013 measured speed may be shown by 'networkctl status'.
2015 * "networkctl status" now displays MTU and queue lengths, and more
2016 detailed information about VXLAN and bridge devices.
2018 * systemd-networkd's .network and .link files gained a new Property=
2019 setting in the [Match] section, to match against devices with
2020 specific udev properties.
2022 * systemd-networkd's tunnel support gained a new option
2023 AssignToLoopback= for selecting whether to use the loopback device
2024 "lo" as underlying device.
2026 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
2027 been renamed to LinkLayerAddress=, and it now allows configuration of
2030 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
2031 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
2032 IPv6 configuration (static or DHCPv6) was found for a given
2033 interface. It will not touch the sysctl otherwise.
2035 * The order of entries is $PATH used by the user manager instance was
2036 changed to put bin/ entries before the corresponding sbin/ entries.
2037 It is recommended to not rely on this order, and only ever have one
2038 binary with a given name in the system paths under /usr.
2040 * A new tool systemd-network-generator has been added that may generate
2041 .network, .netdev and .link files from IP configuration specified on
2042 the kernel command line in the format used by Dracut.
2044 * The CriticalConnection= setting in .network files is now deprecated,
2045 and replaced by a new KeepConfiguration= setting which allows more
2046 detailed configuration of the IP configuration to keep in place.
2048 * systemd-analyze gained a few new verbs:
2050 - "systemd-analyze timestamp" parses and converts timestamps. This is
2051 similar to the existing "systemd-analyze calendar" command which
2052 does the same for recurring calendar events.
2054 - "systemd-analyze timespan" parses and converts timespans (i.e.
2055 durations as opposed to points in time).
2057 - "systemd-analyze condition" will parse and test ConditionXYZ=
2060 - "systemd-analyze exit-status" will parse and convert exit status
2061 codes to their names and back.
2063 - "systemd-analyze unit-files" will print a list of all unit
2064 file paths and unit aliases.
2066 * SuccessExitStatus=, RestartPreventExitStatus=, and
2067 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
2068 is equivalent to "65"). Those exit status name mappings may be
2069 displayed with the systemd-analyze exit-status verb describe above.
2071 * systemd-logind now exposes a per-session SetBrightness() bus call,
2072 which may be used to securely change the brightness of a kernel
2073 brightness device, if it belongs to the session's seat. By using this
2074 call unprivileged clients can make changes to "backlight" and "leds"
2075 devices securely with strict requirements on session membership.
2076 Desktop environments may use this to generically make brightness
2077 changes to such devices without shipping private SUID binaries or
2078 udev rules for that purpose.
2080 * "udevadm info" gained a --wait-for-initialization switch to wait for
2081 a device to be initialized.
2083 * systemd-hibernate-resume-generator will now look for resumeflags= on
2084 the kernel command line, which is similar to rootflags= and may be
2085 used to configure device timeout for the hibernation device.
2087 * sd-event learnt a new API call sd_event_source_disable_unref() for
2088 disabling and unref'ing an event source in a single function. A
2089 related call sd_event_source_disable_unrefp() has been added for use
2090 with gcc's cleanup extension.
2092 * The sd-id128.h public API gained a new definition
2093 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
2096 * "busctl introspect" gained a new switch --xml-interface for dumping
2097 XML introspection data unmodified.
2099 * PID 1 may now show the unit name instead of the unit description
2100 string in its status output during boot. This may be configured in
2101 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
2102 kernel command line option systemd.status_unit_format=.
2104 * PID 1 now understands a new option KExecWatchdogSec= in
2105 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
2106 Previously watchdog functionality was only available for regular
2107 reboots. The new setting defaults to off, because we don't know in
2108 the general case if the watchdog will be reset after kexec (some
2109 drivers do reset it, but not all), and the new userspace might not be
2110 configured to handle the watchdog.
2112 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
2113 RebootWatchdogSec= to more clearly communicate what it is about. The
2114 old name is still accepted for compatibility.
2116 * The systemd.debug_shell kernel command line option now optionally
2117 takes a tty name to spawn the debug shell on, which allows a
2118 different tty to be selected than the built-in default.
2120 * Service units gained a new ExecCondition= setting which will run
2121 before ExecStartPre= and either continue execution of the unit (for
2122 clean exit codes), stop execution without marking the unit failed
2123 (for exit codes 1 through 254), or stop execution and fail the unit
2124 (for exit code 255 or abnormal termination).
2126 * A new service systemd-pstore.service has been added that pulls data
2127 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
2130 * timedatectl gained new verbs for configuring per-interface NTP
2131 service configuration for systemd-timesyncd.
2133 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
2134 2019. (You can set non-UTF-8 locales though, if you know their name.)
2136 * If variable assignments in sysctl.d/ files are prefixed with "-" any
2137 failures to apply them are now ignored.
2139 * systemd-random-seed.service now optionally credits entropy when
2140 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
2141 true for the service to enable this behaviour, but please consult the
2142 documentation first, since this comes with a couple of caveats.
2144 * systemd-random-seed.service is now a synchronization point for full
2145 initialization of the kernel's entropy pool. Services that require
2146 /dev/urandom to be correctly initialized should be ordered after this
2149 * The systemd-boot boot loader has been updated to optionally maintain
2150 a random seed file in the EFI System Partition (ESP). During the boot
2151 phase, this random seed is read and updated with a new seed
2152 cryptographically derived from it. Another derived seed is passed to
2153 the OS. The latter seed is then credited to the kernel's entropy pool
2154 very early during userspace initialization (from PID 1). This allows
2155 systems to boot up with a fully initialized kernel entropy pool from
2156 earliest boot on, and thus entirely removes all entropy pool
2157 initialization delays from systems using systemd-boot. Special care
2158 is taken to ensure different seeds are derived on system images
2159 replicated to multiple systems. "bootctl status" will show whether
2160 a seed was received from the boot loader.
2162 * bootctl gained two new verbs:
2164 - "bootctl random-seed" will generate the file in ESP and an EFI
2165 variable to allow a random seed to be passed to the OS as described
2168 - "bootctl is-installed" checks whether systemd-boot is currently
2171 * bootctl will warn if it detects that boot entries are misconfigured
2172 (for example if the kernel image was removed without purging the
2175 * A new document has been added describing systemd's use and support
2176 for the kernel's entropy pool subsystem:
2178 https://systemd.io/RANDOM_SEEDS
2180 * When the system is hibernated the swap device to write the
2181 hibernation image to is now automatically picked from all available
2182 swap devices, preferring the swap device with the highest configured
2183 priority over all others, and picking the device with the most free
2184 space if there are multiple devices with the highest priority.
2186 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
2187 option that permits selecting the timeout how long to wait for a
2188 device with an encryption key before asking for the password.
2190 * IOWeight= has learnt to properly set the IO weight when using the
2191 BFQ scheduler officially found in kernels 5.0+.
2193 * A new mailing list has been created for reporting of security issues:
2194 systemd-security@redhat.com. For mode details, see
2195 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
2197 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
2198 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
2199 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
2200 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
2201 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
2202 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
2203 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
2204 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
2205 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
2206 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
2207 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
2208 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
2209 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
2210 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
2211 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
2212 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
2213 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
2214 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
2215 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
2216 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
2217 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
2218 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
2219 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
2220 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
2221 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
2222 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
2223 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
2224 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
2225 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
2226 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
2228 – Camerino, 2019-09-03
2232 * In .link files, MACAddressPolicy=persistent (the default) is changed
2233 to cover more devices. For devices like bridges, tun, tap, bond, and
2234 similar interfaces that do not have other identifying information,
2235 the interface name is used as the basis for persistent seed for MAC
2236 and IPv4LL addresses. The way that devices that were handled
2237 previously is not changed, and this change is about covering more
2238 devices then previously by the "persistent" policy.
2240 MACAddressPolicy=random may be used to force randomized MACs and
2241 IPv4LL addresses for a device if desired.
2243 Hint: the log output from udev (at debug level) was enhanced to
2244 clarify what policy is followed and which attributes are used.
2245 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
2246 may be used to view this.
2248 Hint: if a bridge interface is created without any slaves, and gains
2249 a slave later, then now the bridge does not inherit slave's MAC.
2250 To inherit slave's MAC, for example, create the following file:
2252 # /etc/systemd/network/98-bridge-inherit-mac.link
2257 MACAddressPolicy=none
2260 * The .device units generated by systemd-fstab-generator and other
2261 generators do not automatically pull in the corresponding .mount unit
2262 as a Wants= dependency. This means that simply plugging in the device
2263 will not cause the mount unit to be started automatically. But please
2264 note that the mount unit may be started for other reasons, in
2265 particular if it is part of local-fs.target, and any unit which
2266 (transitively) depends on local-fs.target is started.
2268 * networkctl list/status/lldp now accept globbing wildcards for network
2269 interface names to match against all existing interfaces.
2271 * The $PIDFILE environment variable is set to point the absolute path
2272 configured with PIDFile= for processes of that service.
2274 * The fallback DNS server list was augmented with Cloudflare public DNS
2275 servers. Use `-Ddns-servers=` to set a different fallback.
2277 * A new special target usb-gadget.target will be started automatically
2278 when a USB Device Controller is detected (which means that the system
2279 is a USB peripheral).
2281 * A new unit setting CPUQuotaPeriodSec= assigns the time period
2282 relatively to which the CPU time quota specified by CPUQuota= is
2285 * A new unit setting ProtectHostname= may be used to prevent services
2286 from modifying hostname information (even if they otherwise would
2287 have privileges to do so).
2289 * A new unit setting NetworkNamespacePath= may be used to specify a
2290 namespace for service or socket units through a path referring to a
2291 Linux network namespace pseudo-file.
2293 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
2294 have an effect on .socket units: when used the listening socket is
2295 created within the configured network namespace instead of the host
2298 * ExecStart= command lines in unit files may now be prefixed with ':'
2299 in which case environment variable substitution is
2300 disabled. (Supported for the other ExecXYZ= settings, too.)
2302 * .timer units gained two new boolean settings OnClockChange= and
2303 OnTimezoneChange= which may be used to also trigger a unit when the
2304 system clock is changed or the local timezone is
2305 modified. systemd-run has been updated to make these options easily
2306 accessible from the command line for transient timers.
2308 * Two new conditions for units have been added: ConditionMemory= may be
2309 used to conditionalize a unit based on installed system
2310 RAM. ConditionCPUs= may be used to conditionalize a unit based on
2311 installed CPU cores.
2313 * The @default system call filter group understood by SystemCallFilter=
2314 has been updated to include the new rseq() system call introduced in
2317 * A new time-set.target has been added that indicates that the system
2318 time has been set from a local source (possibly imprecise). The
2319 existing time-sync.target is stronger and indicates that the time has
2320 been synchronized with a precise external source. Services where
2321 approximate time is sufficient should use the new target.
2323 * "systemctl start" (and related commands) learnt a new
2324 --show-transaction option. If specified brief information about all
2325 jobs queued because of the requested operation is shown.
2327 * systemd-networkd recognizes a new operation state 'enslaved', used
2328 (instead of 'degraded' or 'carrier') for interfaces which form a
2329 bridge, bond, or similar, and an new 'degraded-carrier' operational
2330 state used for the bond or bridge master interface when one of the
2331 enslaved devices is not operational.
2333 * .network files learnt the new IgnoreCarrierLoss= option for leaving
2334 networks configured even if the carrier is lost.
2336 * The RequiredForOnline= setting in .network files may now specify a
2337 minimum operational state required for the interface to be considered
2338 "online" by systemd-networkd-wait-online. Related to this
2339 systemd-networkd-wait-online gained a new option --operational-state=
2340 to configure the same, and its --interface= option was updated to
2341 optionally also take an operational state specific for an interface.
2343 * systemd-networkd-wait-online gained a new setting --any for waiting
2344 for only one of the requested interfaces instead of all of them.
2346 * systemd-networkd now implements L2TP tunnels.
2348 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
2349 may be used to cause autonomous and onlink prefixes received in IPv6
2350 Router Advertisements to be ignored.
2352 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
2353 file settings may be used to tweak bridge behaviour.
2355 * The new TripleSampling= option in .network files may be used to
2356 configure CAN triple sampling.
2358 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
2359 used to point to private or preshared key for a WireGuard interface.
2361 * /etc/crypttab now supports the same-cpu-crypt and
2362 submit-from-crypt-cpus options to tweak encryption work scheduling
2365 * systemd-tmpfiles will now take a BSD file lock before operating on a
2366 contents of directory. This may be used to temporarily exclude
2367 directories from aging by taking the same lock (useful for example
2368 when extracting a tarball into /tmp or /var/tmp as a privileged user,
2369 which might create files with really old timestamps, which
2370 nevertheless should not be deleted). For further details, see:
2372 https://systemd.io/TEMPORARY_DIRECTORIES
2374 * systemd-tmpfiles' h line type gained support for the
2375 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
2376 controlling project quota inheritance.
2378 * sd-boot and bootctl now implement support for an Extended Boot Loader
2379 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
2380 addition to the ESP partition mounted to /efi or /boot/efi.
2381 Configuration file fragments, kernels, initrds and other EFI images
2382 to boot will be loaded from both the ESP and XBOOTLDR partitions.
2383 The XBOOTLDR partition was previously described by the Boot Loader
2384 Specification, but implementation was missing in sd-boot. Support for
2385 this concept allows using the sd-boot boot loader in more
2386 conservative scenarios where the boot loader itself is placed in the
2387 ESP but the kernels to boot (and their metadata) in a separate
2390 * A system may now be booted with systemd.volatile=overlay on the
2391 kernel command line, which causes the root file system to be set up
2392 an overlayfs mount combining the root-only root directory with a
2393 writable tmpfs. In this setup, the underlying root device is not
2394 modified, and any changes are lost at reboot.
2396 * Similar, systemd-nspawn can now boot containers with a volatile
2397 overlayfs root with the new --volatile=overlay switch.
2399 * systemd-nspawn can now consume OCI runtime bundles using a new
2400 --oci-bundle= option. This implementation is fully usable, with most
2401 features in the specification implemented, but since this a lot of
2402 new code and functionality, this feature should most likely not
2403 be used in production yet.
2405 * systemd-nspawn now supports various options described by the OCI
2406 runtime specification on the command-line and in .nspawn files:
2407 --inaccessible=/Inaccessible= may be used to mask parts of the file
2408 system tree, --console=/--pipe may be used to configure how standard
2409 input, output, and error are set up.
2411 * busctl learned the `emit` verb to generate D-Bus signals.
2413 * systemd-analyze cat-config may be used to gather and display
2414 configuration spread over multiple files, for example system and user
2415 presets, tmpfiles.d, sysusers.d, udev rules, etc.
2417 * systemd-analyze calendar now takes an optional new parameter
2418 --iterations= which may be used to show a maximum number of iterations
2419 the specified expression will elapse next.
2421 * The sd-bus C API gained support for naming method parameters in the
2424 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
2425 the reboot() system call expects.
2427 * journalctl learnt a new --cursor-file= option that points to a file
2428 from which a cursor should be loaded in the beginning and to which
2429 the updated cursor should be stored at the end.
2431 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
2432 detected by systemd-detect-virt (and may also be used in
2433 ConditionVirtualization=).
2435 * The behaviour of systemd-logind may now be modified with environment
2436 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
2437 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
2438 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
2439 skip the relevant operation completely (when set to false), or to
2440 create a flag file in /run/systemd (when set to true), instead of
2441 actually commencing the real operation when requested. The presence
2442 of /run/systemd/reboot-to-firmware-setup,
2443 /run/systemd/reboot-to-boot-loader-menu, and
2444 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
2445 boot loader implementations to replace some steps logind performs
2446 during reboot with their own operations.
2448 * systemctl can be used to request a reboot into the boot loader menu
2449 or a specific boot loader entry with the new --boot-load-menu= and
2450 --boot-loader-entry= options to a reboot command. (This requires a
2451 boot loader that supports this, for example sd-boot.)
2453 * kernel-install will no longer unconditionally create the output
2454 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
2455 snippets, but will do only if the machine-specific parent directory
2456 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
2457 to create this parent directory during sd-boot installation.
2459 This makes it easier to use kernel-install with plugins which support
2460 a different layout of the bootloader partitions (for example grub2).
2462 * During package installation (with `ninja install`), we would create
2463 symlinks for getty@tty1.service, systemd-networkd.service,
2464 systemd-networkd.socket, systemd-resolved.service,
2465 remote-cryptsetup.target, remote-fs.target,
2466 systemd-networkd-wait-online.service, and systemd-timesyncd.service
2467 in /etc, as if `systemctl enable` was called for those units, to make
2468 the system usable immediately after installation. Now this is not
2469 done anymore, and instead calling `systemctl preset-all` is
2470 recommended after the first installation of systemd.
2472 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
2473 is built on seccomp. When turned on creation of SUID/SGID files is
2476 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
2477 implied if DynamicUser= is turned on for a service. This hardens
2478 these services, so that they neither can benefit from nor create
2479 SUID/SGID executables. This is a minor compatibility breakage, given
2480 that when DynamicUser= was first introduced SUID/SGID behaviour was
2481 unaffected. However, the security benefit of these two options is
2482 substantial, and the setting is still relatively new, hence we opted
2483 to make it mandatory for services with dynamic users.
2485 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
2486 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
2487 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
2488 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
2489 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
2490 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
2491 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
2492 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
2493 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
2494 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
2495 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
2496 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
2497 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
2498 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
2499 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
2500 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
2501 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
2502 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2504 — Warsaw, 2019-04-11
2508 * The default locale can now be configured at compile time. Otherwise,
2509 a suitable default will be selected automatically (one of C.UTF-8,
2510 en_US.UTF-8, and C).
2512 * The version string shown by systemd and other tools now includes the
2513 git commit hash when built from git. An override may be specified
2514 during compilation, which is intended to be used by distributions to
2515 include the package release information.
2517 * systemd-cat can now filter standard input and standard error streams
2518 for different syslog priorities using the new --stderr-priority=
2521 * systemd-journald and systemd-journal-remote reject entries which
2522 contain too many fields (CVE-2018-16865) and set limits on the
2523 process' command line length (CVE-2018-16864).
2525 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
2528 * A new network device NamePolicy "keep" is implemented for link files,
2529 and used by default in 99-default.link (the fallback configuration
2530 provided by systemd). With this policy, if the network device name
2531 was already set by userspace, the device will not be renamed again.
2532 This matches the naming scheme that was implemented before
2533 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
2534 is also enabled by default, even if not specified. Effectively, this
2535 means that if naming-scheme >= 240 is specified, network devices will
2536 be renamed according to the configuration, even if they have been
2537 renamed already, if "keep" is not specified as the naming policy in
2538 the .link file. The 99-default.link file provided by systemd includes
2539 "keep" for backwards compatibility, but it is recommended for user
2540 installed .link files to *not* include it.
2542 The "kernel" policy, which keeps kernel names declared to be
2543 "persistent", now works again as documented.
2545 * kernel-install script now optionally takes the paths to one or more
2546 initrd files, and passes them to all plugins.
2548 * The mincore() system call has been dropped from the @system-service
2549 system call filter group, as it is pretty exotic and may potentially
2550 used for side-channel attacks.
2552 * -fPIE is dropped from compiler and linker options. Please specify
2553 -Db_pie=true option to meson to build position-independent
2554 executables. Note that the meson option is supported since meson-0.49.
2556 * The fs.protected_regular and fs.protected_fifos sysctls, which were
2557 added in Linux 4.19 to make some data spoofing attacks harder, are
2558 now enabled by default. While this will hopefully improve the
2559 security of most installations, it is technically a backwards
2560 incompatible change; to disable these sysctls again, place the
2561 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
2563 fs.protected_regular = 0
2564 fs.protected_fifos = 0
2566 Note that the similar hardlink and symlink protection has been
2567 enabled since v199, and may be disabled likewise.
2569 * The files read from the EnvironmentFile= setting in unit files now
2570 parse backslashes inside quotes literally, matching the behaviour of
2573 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
2574 now automatically become NOPs when run in a chroot() environment.
2576 * The tmpfiles.d/ "C" line type will now copy directory trees not only
2577 when the destination is so far missing, but also if it already exists
2578 as a directory and is empty. This is useful to cater for systems
2579 where directory trees are put together from multiple separate mount
2580 points but otherwise empty.
2582 * A new function sd_bus_close_unref() (and the associated
2583 sd_bus_close_unrefp()) has been added to libsystemd, that combines
2584 sd_bus_close() and sd_bus_unref() in one.
2586 * udevadm control learnt a new option for --ping for testing whether a
2587 systemd-udevd instance is running and reacting.
2589 * udevadm trigger learnt a new option for --wait-daemon for waiting
2590 systemd-udevd daemon to be initialized.
2592 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
2593 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
2594 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
2595 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
2596 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
2597 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
2598 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
2599 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
2600 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
2601 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
2602 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
2603 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
2604 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
2605 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
2606 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
2607 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
2608 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
2610 — Berlin, 2019-02-14
2614 * NoNewPrivileges=yes has been set for all long-running services
2615 implemented by systemd. Previously, this was problematic due to
2616 SELinux (as this would also prohibit the transition from PID1's label
2617 to the service's label). This restriction has since been lifted, but
2618 an SELinux policy update is required.
2619 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
2621 * DynamicUser=yes is dropped from systemd-networkd.service,
2622 systemd-resolved.service and systemd-timesyncd.service, which was
2623 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
2624 and since v236 for systemd-timesyncd.service. The users and groups
2625 systemd-network, systemd-resolve and systemd-timesync are created
2626 by systemd-sysusers again. Distributors or system administrators
2627 may need to create these users and groups if they not exist (or need
2628 to re-enable DynamicUser= for those units) while upgrading systemd.
2629 Also, the clock file for systemd-timesyncd may need to move from
2630 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
2632 * When unit files are loaded from disk, previously systemd would
2633 sometimes (depending on the unit loading order) load units from the
2634 target path of symlinks in .wants/ or .requires/ directories of other
2635 units. This meant that unit could be loaded from different paths
2636 depending on whether the unit was requested explicitly or as a
2637 dependency of another unit, not honouring the priority of directories
2638 in search path. It also meant that it was possible to successfully
2639 load and start units which are not found in the unit search path, as
2640 long as they were requested as a dependency and linked to from
2641 .wants/ or .requires/. The target paths of those symlinks are not
2642 used for loading units anymore and the unit file must be found in
2645 * A new service type has been added: Type=exec. It's very similar to
2646 Type=simple but ensures the service manager will wait for both fork()
2647 and execve() of the main service binary to complete before proceeding
2648 with follow-up units. This is primarily useful so that the manager
2649 propagates any errors in the preparation phase of service execution
2650 back to the job that requested the unit to be started. For example,
2651 consider a service that has ExecStart= set to a file system binary
2652 that doesn't exist. With Type=simple starting the unit would be
2653 considered instantly successful, as only fork() has to complete
2654 successfully and the manager does not wait for execve(), and hence
2655 its failure is seen "too late". With the new Type=exec service type
2656 starting the unit will fail, as the manager will wait for the
2657 execve() and notice its failure, which is then propagated back to the
2660 NOTE: with the next release 241 of systemd we intend to change the
2661 systemd-run tool to default to Type=exec for transient services
2662 started by it. This should be mostly safe, but in specific corner
2663 cases might result in problems, as the systemd-run tool will then
2664 block on NSS calls (such as user name look-ups due to User=) done
2665 between the fork() and execve(), which under specific circumstances
2666 might cause problems. It is recommended to specify "-p Type=simple"
2667 explicitly in the few cases where this applies. For regular,
2668 non-transient services (i.e. those defined with unit files on disk)
2669 we will continue to default to Type=simple.
2671 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
2672 userspace processes is set to 1024 (soft) and 4096
2673 (hard). Previously, systemd passed this on unmodified to all
2674 processes it forked off. With this systemd release the hard limit
2675 systemd passes on is increased to 512K, overriding the kernel's
2676 defaults and substantially increasing the number of simultaneous file
2677 descriptors unprivileged userspace processes can allocate. Note that
2678 the soft limit remains at 1024 for compatibility reasons: the
2679 traditional UNIX select() call cannot deal with file descriptors >=
2680 1024 and increasing the soft limit globally might thus result in
2681 programs unexpectedly allocating a high file descriptor and thus
2682 failing abnormally when attempting to use it with select() (of
2683 course, programs shouldn't use select() anymore, and prefer
2684 poll()/epoll, but the call unfortunately remains undeservedly popular
2685 at this time). This change reflects the fact that file descriptor
2686 handling in the Linux kernel has been optimized in more recent
2687 kernels and allocating large numbers of them should be much cheaper
2688 both in memory and in performance than it used to be. Programs that
2689 want to take benefit of the increased limit have to "opt-in" into
2690 high file descriptors explicitly by raising their soft limit. Of
2691 course, when they do that they must acknowledge that they cannot use
2692 select() anymore (and neither can any shared library they use — or
2693 any shared library used by any shared library they use and so on).
2694 Which default hard limit is most appropriate is of course hard to
2695 decide. However, given reports that ~300K file descriptors are used
2696 in real-life applications we believe 512K is sufficiently high as new
2697 default for now. Note that there are also reports that using very
2698 high hard limits (e.g. 1G) is problematic: some software allocates
2699 large arrays with one element for each potential file descriptor
2700 (Java, …) — a high hard limit thus triggers excessively large memory
2701 allocations in these applications. Hopefully, the new default of 512K
2702 is a good middle ground: higher than what real-life applications
2703 currently need, and low enough for avoid triggering excessively large
2704 allocations in problematic software. (And yes, somebody should fix
2707 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
2708 to the highest possible values, as separate accounting of file
2709 descriptors is no longer necessary, as memcg tracks them correctly as
2710 part of the memory accounting anyway. Thus, from the four limits on
2711 file descriptors currently enforced (fs.file-max, fs.nr_open,
2712 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
2713 and keep only the latter two. A set of build-time options
2714 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
2715 has been added to revert this change in behaviour, which might be
2716 an option for systems that turn off memcg in the kernel.
2718 * When no /etc/locale.conf file exists (and hence no locale settings
2719 are in place), systemd will now use the "C.UTF-8" locale by default,
2720 and set LANG= to it. This locale is supported by various
2721 distributions including Fedora, with clear indications that upstream
2722 glibc is going to make it available too. This locale enables UTF-8
2723 mode by default, which appears appropriate for 2018.
2725 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
2726 default. This effectively switches the RFC3704 Reverse Path filtering
2727 from Strict mode to Loose mode. This is more appropriate for hosts
2728 that have multiple links with routes to the same networks (e.g.
2729 a client with a Wi-Fi and Ethernet both connected to the internet).
2731 Consult the kernel documentation for details on this sysctl:
2732 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
2734 * CPUAccounting=yes no longer enables the CPU controller when using
2735 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
2736 statistics are now provided independently from the CPU controller.
2738 * Support for disabling a particular cgroup controller within a sub-tree
2739 has been added through the DisableControllers= directive.
2741 * cgroup_no_v1=all on the kernel command line now also implies
2742 using the unified cgroup hierarchy, unless one explicitly passes
2743 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
2745 * The new "MemoryMin=" unit file property may now be used to set the
2746 memory usage protection limit of processes invoked by the unit. This
2747 controls the cgroup v2 memory.min attribute. Similarly, the new
2748 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
2749 cgroup v2 io.latency cgroup property for configuring per-service I/O
2752 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
2753 to the cgroup v1 "devices" cgroup controller.
2755 * systemd-escape now is able to combine --unescape with --template. It
2756 also learnt a new option --instance for extracting and unescaping the
2757 instance part of a unit name.
2759 * sd-bus now provides the sd_bus_message_readv() which is similar to
2760 sd_bus_message_read() but takes a va_list object. The pair
2761 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
2762 has been added for configuring the default method call timeout to
2763 use. sd_bus_error_move() may be used to efficiently move the contents
2764 from one sd_bus_error structure to another, invalidating the
2765 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
2766 be used to control whether a bus connection object is automatically
2767 flushed when an sd-event loop is exited.
2769 * When processing classic BSD syslog log messages, journald will now
2770 save the original time-stamp string supplied in the new
2771 SYSLOG_TIMESTAMP= journal field. This permits consumers to
2772 reconstruct the original BSD syslog message more correctly.
2774 * StandardOutput=/StandardError= in service files gained support for
2775 new "append:…" parameters, for connecting STDOUT/STDERR of a service
2776 to a file, and appending to it.
2778 * The signal to use as last step of killing of unit processes is now
2779 configurable. Previously it was hard-coded to SIGKILL, which may now
2780 be overridden with the new KillSignal= setting. Note that this is the
2781 signal used when regular termination (i.e. SIGTERM) does not suffice.
2782 Similarly, the signal used when aborting a program in case of a
2783 watchdog timeout may now be configured too (WatchdogSignal=).
2785 * The XDG_SESSION_DESKTOP environment variable may now be configured in
2786 the pam_systemd argument line, using the new desktop= switch. This is
2787 useful to initialize it properly from a display manager without
2788 having to touch C code.
2790 * Most configuration options that previously accepted percentage values
2791 now also accept permille values with the '‰' suffix (instead of '%').
2793 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
2796 * systemd-resolved's configuration file resolved.conf gained a new
2797 option ReadEtcHosts= which may be used to turn off processing and
2798 honoring /etc/hosts entries.
2800 * The "--wait" switch may now be passed to "systemctl
2801 is-system-running", in which case the tool will synchronously wait
2802 until the system finished start-up.
2804 * hostnamed gained a new bus call to determine the DMI product UUID.
2806 * On x86-64 systemd will now prefer using the RDRAND processor
2807 instruction over /dev/urandom whenever it requires randomness that
2808 neither has to be crypto-grade nor should be reproducible. This
2809 should substantially reduce the amount of entropy systemd requests
2810 from the kernel during initialization on such systems, though not
2811 reduce it to zero. (Why not zero? systemd still needs to allocate
2812 UUIDs and such uniquely, which require high-quality randomness.)
2814 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
2815 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
2816 for forcing the "Other Information" bit in IPv6 RA messages. The
2817 bonding logic gained four new options AdActorSystemPriority=,
2818 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
2819 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
2820 shuffling of flows. The tunnel logic gained a new
2821 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
2822 Deployment. The policy rule logic gained four new options IPProtocol=,
2823 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
2824 support for the MulticastToUnicast= option. networkd also gained
2825 support for configuring static IPv4 ARP or IPv6 neighbor entries.
2827 * .preset files (as read by 'systemctl preset') may now be used to
2828 instantiate services.
2830 * /etc/crypttab now understands the sector-size= option to configure
2831 the sector size for an encrypted partition.
2833 * Key material for encrypted disks may now be placed on a formatted
2834 medium, and referenced from /etc/crypttab by the UUID of the file
2835 system, followed by "=" suffixed by the path to the key file.
2837 * The "collect" udev component has been removed without replacement, as
2838 it is neither used nor maintained.
2840 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
2841 LogsDirectory=, ConfigurationDirectory= settings are used in a
2842 service the executed processes will now receive a set of environment
2843 variables containing the full paths of these directories.
2844 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
2845 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
2846 are used. Note that these options may be used multiple times per
2847 service in which case the resulting paths will be concatenated and
2848 separated by colons.
2850 * Predictable interface naming has been extended to cover InfiniBand
2851 NICs. They will be exposed with an "ib" prefix.
2853 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
2854 which case the respective line failing is ignored.
2856 * .link files may now be used to configure the equivalent to the
2857 "ethtool advertise" commands.
2859 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
2860 alternative to libudev.h. Previously, the latter was just an internal
2861 wrapper around the former, but now these two APIs are exposed
2864 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
2865 which calculates an app-specific boot ID similar to how
2866 sd_id128_get_machine_app_specific() generates an app-specific machine
2869 * A new tool systemd-id128 has been added that can be used to determine
2870 and generate various 128bit IDs.
2872 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
2875 * systemd-hibernate-resume-generator will now honor the "noresume"
2876 kernel command line option, in which case it will bypass resuming
2877 from any hibernated image.
2879 * The systemd-sleep.conf configuration file gained new options
2880 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
2881 AllowHybridSleep= for prohibiting specific sleep modes even if the
2882 kernel exports them.
2884 * portablectl is now officially supported and has thus moved to
2887 * bootctl learnt the two new commands "set-default" and "set-oneshot"
2888 for setting the default boot loader item to boot to (either
2889 persistently or only for the next boot). This is currently only
2890 compatible with sd-boot, but may be implemented on other boot loaders
2891 too, that follow the boot loader interface. The updated interface is
2892 now documented here:
2894 https://systemd.io/BOOT_LOADER_INTERFACE
2896 * A new kernel command line option systemd.early_core_pattern= is now
2897 understood which may be used to influence the core_pattern PID 1
2898 installs during early boot.
2900 * busctl learnt two new options -j and --json= for outputting method
2901 call replies, properties and monitoring output in JSON.
2903 * journalctl's JSON output now supports simple ANSI coloring as well as
2904 a new "json-seq" mode for generating RFC7464 output.
2906 * Unit files now support the %g/%G specifiers that resolve to the UNIX
2907 group/GID of the service manager runs as, similar to the existing
2908 %u/%U specifiers that resolve to the UNIX user/UID.
2910 * systemd-logind learnt a new global configuration option
2911 UserStopDelaySec= that may be set in logind.conf. It specifies how
2912 long the systemd --user instance shall remain started after a user
2913 logs out. This is useful to speed up repetitive re-connections of the
2914 same user, as it means the user's service manager doesn't have to be
2915 stopped/restarted on each iteration, but can be reused between
2916 subsequent options. This setting defaults to 10s. systemd-logind also
2917 exports two new properties on its Manager D-Bus objects indicating
2918 whether the system's lid is currently closed, and whether the system
2921 * systemd gained support for a generic boot counting logic, which
2922 generically permits automatic reverting to older boot loader entries
2923 if newer updated ones don't work. The boot loader side is implemented
2924 in sd-boot, but is kept open for other boot loaders too. For details
2927 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
2929 * The SuccessAction=/FailureAction= unit file settings now learnt two
2930 new parameters: "exit" and "exit-force", which result in immediate
2931 exiting of the service manager, and are only useful in systemd --user
2932 and container environments.
2934 * Unit files gained support for a pair of options
2935 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
2936 exit status to use as service manager exit status when
2937 SuccessAction=/FailureAction= is set to exit or exit-force.
2939 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
2940 options may now be used to configure the log rate limiting applied by
2941 journald per-service.
2943 * systemd-analyze gained a new verb "timespan" for parsing and
2944 normalizing time span values (i.e. strings like "5min 7s 8us").
2946 * systemd-analyze also gained a new verb "security" for analyzing the
2947 security and sand-boxing settings of services in order to determine an
2948 "exposure level" for them, indicating whether a service would benefit
2949 from more sand-boxing options turned on for them.
2951 * "systemd-analyze syscall-filter" will now also show system calls
2952 supported by the local kernel but not included in any of the defined
2955 * .nspawn files now understand the Ephemeral= setting, matching the
2956 --ephemeral command line switch.
2958 * sd-event gained the new APIs sd_event_source_get_floating() and
2959 sd_event_source_set_floating() for controlling whether a specific
2960 event source is "floating", i.e. destroyed along with the even loop
2963 * Unit objects on D-Bus gained a new "Refs" property that lists all
2964 clients that currently have a reference on the unit (to ensure it is
2967 * The JoinControllers= option in system.conf is no longer supported, as
2968 it didn't work correctly, is hard to support properly, is legacy (as
2969 the concept only exists on cgroup v1) and apparently wasn't used.
2971 * Journal messages that are generated whenever a unit enters the failed
2972 state are now tagged with a unique MESSAGE_ID. Similarly, messages
2973 generated whenever a service process exits are now made recognizable,
2974 too. A tagged message is also emitted whenever a unit enters the
2975 "dead" state on success.
2977 * systemd-run gained a new switch --working-directory= for configuring
2978 the working directory of the service to start. A shortcut -d is
2979 equivalent, setting the working directory of the service to the
2980 current working directory of the invoking program. The new --shell
2981 (or just -S) option has been added for invoking the $SHELL of the
2982 caller as a service, and implies --pty --same-dir --wait --collect
2983 --service-type=exec. Or in other words, "systemd-run -S" is now the
2984 quickest way to quickly get an interactive in a fully clean and
2985 well-defined system service context.
2987 * machinectl gained a new verb "import-fs" for importing an OS tree
2988 from a directory. Moreover, when a directory or tarball is imported
2989 and single top-level directory found with the OS itself below the OS
2990 tree is automatically mangled and moved one level up.
2992 * systemd-importd will no longer set up an implicit btrfs loop-back
2993 file system on /var/lib/machines. If one is already set up, it will
2994 continue to be used.
2996 * A new generator "systemd-run-generator" has been added. It will
2997 synthesize a unit from one or more program command lines included in
2998 the kernel command line. This is very useful in container managers
3001 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
3003 This will run "systemd-nspawn" on an image, invoke the specified
3004 command line and immediately shut down the container again, returning
3005 the command line's exit code.
3007 * The block device locking logic is now documented:
3009 https://systemd.io/BLOCK_DEVICE_LOCKING
3011 * loginctl and machinectl now optionally output the various tables in
3012 JSON using the --output= switch. It is our intention to add similar
3013 support to systemctl and all other commands.
3015 * udevadm's query and trigger verb now optionally take a .device unit
3018 * systemd-udevd's network naming logic now understands a new
3019 net.naming-scheme= kernel command line switch, which may be used to
3020 pick a specific version of the naming scheme. This helps stabilizing
3021 interface names even as systemd/udev are updated and the naming logic
3024 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
3025 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
3026 initialize one to all 0xFF.
3028 * After loading the SELinux policy systemd will now recursively relabel
3029 all files and directories listed in
3030 /run/systemd/relabel-extra.d/*.relabel (which should be simple
3031 newline separated lists of paths) in addition to the ones it already
3032 implicitly relabels in /run, /dev and /sys. After the relabelling is
3033 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
3034 removed. This is useful to permit initrds (i.e. code running before
3035 the SELinux policy is in effect) to generate files in the host
3036 filesystem safely and ensure that the correct label is applied during
3037 the transition to the host OS.
3039 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
3040 mknod() handling in user namespaces. Previously mknod() would always
3041 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
3042 but device nodes generated that way cannot be opened, and attempts to
3043 open them result in EPERM. This breaks the "graceful fallback" logic
3044 in systemd's PrivateDevices= sand-boxing option. This option is
3045 implemented defensively, so that when systemd detects it runs in a
3046 restricted environment (such as a user namespace, or an environment
3047 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
3048 where device nodes cannot be created the effect of PrivateDevices= is
3049 bypassed (following the logic that 2nd-level sand-boxing is not
3050 essential if the system systemd runs in is itself already sand-boxed
3051 as a whole). This logic breaks with 4.18 in container managers where
3052 user namespacing is used: suddenly PrivateDevices= succeeds setting
3053 up a private /dev/ file system containing devices nodes — but when
3054 these are opened they don't work.
3056 At this point it is recommended that container managers utilizing
3057 user namespaces that intend to run systemd in the payload explicitly
3058 block mknod() with seccomp or similar, so that the graceful fallback
3061 We are very sorry for the breakage and the requirement to change
3062 container configurations for newer kernels. It's purely caused by an
3063 incompatible kernel change. The relevant kernel developers have been
3064 notified about this userspace breakage quickly, but they chose to
3067 * PermissionsStartOnly= setting is deprecated (but is still supported
3068 for backwards compatibility). The same functionality is provided by
3069 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
3072 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
3073 pam_systemd anymore.
3075 * The naming scheme for network devices was changed to always rename
3076 devices, even if they were already renamed by userspace. The "kernel"
3077 policy was changed to only apply as a fallback, if no other naming
3080 * The requirements to build systemd is bumped to meson-0.46 and
3083 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
3084 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
3085 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
3086 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
3087 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
3088 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
3089 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
3090 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
3091 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
3092 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
3093 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
3094 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
3095 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
3096 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
3097 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
3098 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
3099 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
3100 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
3101 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
3102 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
3103 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
3104 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
3105 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
3106 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
3107 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
3108 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
3109 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
3110 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
3111 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
3112 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
3113 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
3114 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
3115 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
3116 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
3117 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
3118 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
3119 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
3120 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
3121 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
3122 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
3123 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
3124 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
3125 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
3126 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
3127 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
3129 — Warsaw, 2018-12-21
3133 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
3134 builtin will name network interfaces differently than in previous
3135 versions for virtual network interfaces created with SR-IOV and NPAR
3136 and for devices where the PCI network controller device does not have
3137 a slot number associated.
3139 SR-IOV virtual devices are now named based on the name of the parent
3140 interface, with a suffix of "v<N>", where <N> is the virtual device
3141 number. Previously those virtual devices were named as if completely
3144 The ninth and later NPAR virtual devices will be named following the
3145 scheme used for the first eight NPAR partitions. Previously those
3146 devices were not renamed and the kernel default (eth<n>) was used.
3148 "net_id" will also generate names for PCI devices where the PCI
3149 network controller device does not have an associated slot number
3150 itself, but one of its parents does. Previously those devices were
3151 not renamed and the kernel default (eth<n>) was used.
3153 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
3154 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
3155 the unit. So, it is expected that the default behavior of
3156 systemd-logind is not changed. However, if distribution packagers or
3157 administrators disabled or modified IPAddressDeny= setting by a
3158 drop-in config file, then it may be necessary to update the file to
3159 re-enable AF_INET and AF_INET6 to support network user name services,
3162 * When the RestrictNamespaces= unit property is specified multiple
3163 times, then the specified types are merged now. Previously, only the
3164 last assignment was used. So, if distribution packagers or
3165 administrators modified the setting by a drop-in config file, then it
3166 may be necessary to update the file.
3168 * When OnFailure= is used in combination with Restart= on a service
3169 unit, then the specified units will no longer be triggered on
3170 failures that result in restarting. Previously, the specified units
3171 would be activated each time the unit failed, even when the unit was
3172 going to be restarted automatically. This behaviour contradicted the
3173 documentation. With this release the code is adjusted to match the
3176 * systemd-tmpfiles will now print a notice whenever it encounters
3177 tmpfiles.d/ lines referencing the /var/run/ directory. It will
3178 recommend reworking them to use the /run/ directory instead (for
3179 which /var/run/ is simply a symlinked compatibility alias). This way
3180 systemd-tmpfiles can properly detect line conflicts and merge lines
3181 referencing the same file by two paths, without having to access
3184 * systemctl disable/unmask/preset/preset-all cannot be used with
3185 --runtime. Previously this was allowed, but resulted in unintuitive
3186 behaviour that wasn't useful. systemctl disable/unmask will now undo
3187 both runtime and persistent enablement/masking, i.e. it will remove
3188 any relevant symlinks both in /run and /etc.
3190 * Note that all long-running system services shipped with systemd will
3191 now default to a system call allow list (rather than a deny list, as
3192 before). In particular, systemd-udevd will now enforce one too. For
3193 most cases this should be safe, however downstream distributions
3194 which disabled sandboxing of systemd-udevd (specifically the
3195 MountFlags= setting), might want to disable this security feature
3196 too, as the default allow-listing will prohibit all mount, swap,
3197 reboot and clock changing operations from udev rules.
3199 * sd-boot acquired new loader configuration settings to optionally turn
3200 off Windows and MacOS boot partition discovery as well as
3201 reboot-into-firmware menu items. It is also able to pick a better
3202 screen resolution for HiDPI systems, and now provides loader
3203 configuration settings to change the resolution explicitly.
3205 * systemd-resolved now supports DNS-over-TLS. It's still
3206 turned off by default, use DNSOverTLS=opportunistic to turn it on in
3207 resolved.conf. We intend to make this the default as soon as couple
3208 of additional techniques for optimizing the initial latency caused by
3209 establishing a TLS/TCP connection are implemented.
3211 * systemd-resolved.service and systemd-networkd.service now set
3212 DynamicUser=yes. The users systemd-resolve and systemd-network are
3213 not created by systemd-sysusers anymore.
3215 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
3216 that embed a network facing module into any process using getpwuid()
3217 or related call: the dynamic allocation of the user ID for
3218 systemd-resolved.service means the service manager has to check NSS
3219 if the user name is already taken when forking off the service. Since
3220 the user in the common case won't be defined in /etc/passwd the
3221 lookup is likely to trigger nss-ldap which in turn might use NSS to
3222 ask systemd-resolved for hostname lookups. This will hence result in
3223 a deadlock: a user name lookup in order to start
3224 systemd-resolved.service will result in a hostname lookup for which
3225 systemd-resolved.service needs to be started already. There are
3226 multiple ways to work around this problem: pre-allocate the
3227 "systemd-resolve" user on such systems, so that nss-ldap won't be
3228 triggered; or use a different NSS package that doesn't do networking
3229 in-process but provides a local asynchronous name cache; or configure
3230 the NSS package to avoid lookups for UIDs in the range `pkg-config
3231 systemd --variable=dynamicuidmin` … `pkg-config systemd
3232 --variable=dynamicuidmax`, so that it does not consider itself
3233 authoritative for the same UID range systemd allocates dynamic users
3236 * The systemd-resolve tool has been renamed to resolvectl (it also
3237 remains available under the old name, for compatibility), and its
3238 interface is now verb-based, similar in style to the other <xyz>ctl
3239 tools, such as systemctl or loginctl.
3241 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
3242 compatibility. It may be symlinked under the 'resolvconf' name, in
3243 which case it will take arguments and input compatible with the
3244 Debian and FreeBSD resolvconf tool.
3246 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
3247 where the system initially suspends, and after a timeout resumes and
3250 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
3251 set the client will only send a DUID as client identifier.
3253 * The nss-systemd glibc NSS module will now enumerate dynamic users and
3254 groups in effect. Previously, it could resolve UIDs/GIDs to user
3255 names/groups and vice versa, but did not support enumeration.
3257 * journald's Compress= configuration setting now optionally accepts a
3258 byte threshold value. All journal objects larger than this threshold
3259 will be compressed, smaller ones will not. Previously this threshold
3260 was not configurable and set to 512.
3262 * A new system.conf setting NoNewPrivileges= is now available which may
3263 be used to turn off acquisition of new privileges system-wide
3264 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
3265 for all its children). Note that turning this option on means setuid
3266 binaries and file system capabilities lose their special powers.
3267 While turning on this option is a big step towards a more secure
3268 system, doing so is likely to break numerous pre-existing UNIX tools,
3269 in particular su and sudo.
3271 * A new service systemd-time-sync-wait.service has been added. If
3272 enabled it will delay the time-sync.target unit at boot until time
3273 synchronization has been received from the network. This
3274 functionality is useful on systems lacking a local RTC or where it is
3275 acceptable that the boot process shall be delayed by external network
3278 * When hibernating, systemd will now inform the kernel of the image
3279 write offset, on kernels new enough to support this. This means swap
3280 files should work for hibernation now.
3282 * When loading unit files, systemd will now look for drop-in unit files
3283 extensions in additional places. Previously, for a unit file name
3284 "foo-bar-baz.service" it would look for dropin files in
3285 "foo-bar-baz.service.d/*.conf". Now, it will also look in
3286 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
3287 service name truncated after all inner dashes. This scheme allows
3288 writing drop-ins easily that apply to a whole set of unit files at
3289 once. It's particularly useful for mount and slice units (as their
3290 naming is prefix based), but is also useful for service and other
3291 units, for packages that install multiple unit files at once,
3292 following a strict naming regime of beginning the unit file name with
3293 the package's name. Two new specifiers are now supported in unit
3294 files to match this: %j and %J are replaced by the part of the unit
3295 name following the last dash.
3297 * Unit files and other configuration files that support specifier
3298 expansion now understand another three new specifiers: %T and %V will
3299 resolve to /tmp and /var/tmp respectively, or whatever temporary
3300 directory has been set for the calling user. %E will expand to either
3301 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
3303 * The ExecStart= lines of unit files are no longer required to
3304 reference absolute paths. If non-absolute paths are specified the
3305 specified binary name is searched within the service manager's
3306 built-in $PATH, which may be queried with 'systemd-path
3307 search-binaries-default'. It's generally recommended to continue to
3308 use absolute paths for all binaries specified in unit files.
3310 * Units gained a new load state "bad-setting", which is used when a
3311 unit file was loaded, but contained fatal errors which prevent it
3312 from being started (for example, a service unit has been defined
3313 lacking both ExecStart= and ExecStop= lines).
3315 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
3316 support alternative debuggers, for example lldb. The old name
3317 continues to be available however, for compatibility reasons. Use the
3318 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
3319 to pick an alternative debugger instead of the default gdb.
3321 * systemctl and the other tools will now output escape sequences that
3322 generate proper clickable hyperlinks in various terminal emulators
3323 where useful (for example, in the "systemctl status" output you can
3324 now click on the unit file name to quickly open it in the
3325 editor/viewer of your choice). Note that not all terminal emulators
3326 support this functionality yet, but many do. Unfortunately, the
3327 "less" pager doesn't support this yet, hence this functionality is
3328 currently automatically turned off when a pager is started (which
3329 happens quite often due to auto-paging). We hope to remove this
3330 limitation as soon as "less" learns these escape sequences. This new
3331 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
3332 environment variable. For details on these escape sequences see:
3333 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
3335 * networkd's .network files now support a new IPv6MTUBytes= option for
3336 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
3337 option in the [Route] section to configure the MTU to use for
3338 specific routes. It also gained support for configuration of the DHCP
3339 "UserClass" option through the new UserClass= setting. It gained
3340 three new options in the new [CAN] section for configuring CAN
3341 networks. The MULTICAST and ALLMULTI interface flags may now be
3342 controlled explicitly with the new Multicast= and AllMulticast=
3345 * networkd will now automatically make use of the kernel's route
3346 expiration feature, if it is available.
3348 * udevd's .link files now support setting the number of receive and
3349 transmit channels, using the RxChannels=, TxChannels=,
3350 OtherChannels=, CombinedChannels= settings.
3352 * Support for UDPSegmentationOffload= has been removed, given its
3353 limited support in hardware, and waning software support.
3355 * networkd's .netdev files now support creating "netdevsim" interfaces.
3357 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
3358 to query the unit belonging to a specific kernel control group.
3360 * systemd-analyze gained a new verb "cat-config", which may be used to
3361 dump the contents of any configuration file, with all its matching
3362 drop-in files added in, and honouring the usual search and masking
3363 logic applied to systemd configuration files. For example use
3364 "systemd-analyze cat-config systemd/system.conf" to get the complete
3365 system configuration file of systemd how it would be loaded by PID 1
3366 itself. Similar to this, various tools such as systemd-tmpfiles or
3367 systemd-sysusers, gained a new option "--cat-config", which does the
3368 corresponding operation for their own configuration settings. For
3369 example, "systemd-tmpfiles --cat-config" will now output the full
3370 list of tmpfiles.d/ lines in place.
3372 * timedatectl gained three new verbs: "show" shows bus properties of
3373 systemd-timedated, "timesync-status" shows the current NTP
3374 synchronization state of systemd-timesyncd, and "show-timesync"
3375 shows bus properties of systemd-timesyncd.
3377 * systemd-timesyncd gained a bus interface on which it exposes details
3380 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
3381 understood by systemd-timedated. It takes a colon-separated list of
3382 unit names of NTP client services. The list is used by
3383 "timedatectl set-ntp".
3385 * systemd-nspawn gained a new --rlimit= switch for setting initial
3386 resource limits for the container payload. There's a new switch
3387 --hostname= to explicitly override the container's hostname. A new
3388 --no-new-privileges= switch may be used to control the
3389 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
3390 --oom-score-adjust= switch controls the OOM scoring adjustment value
3391 for the payload. The new --cpu-affinity= switch controls the CPU
3392 affinity of the container payload. The new --resolv-conf= switch
3393 allows more detailed control of /etc/resolv.conf handling of the
3394 container. Similarly, the new --timezone= switch allows more detailed
3395 control of /etc/localtime handling of the container.
3397 * systemd-detect-virt gained a new --list switch, which will print a
3398 list of all currently known VM and container environments.
3400 * Support for "Portable Services" has been added, see
3401 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
3402 experimental, but this is expected to change soon. Reflecting this
3403 experimental state, the "portablectl" binary is not installed into
3404 /usr/bin yet. The binary has to be called with the full path
3405 /usr/lib/systemd/portablectl instead.
3407 * journalctl's and systemctl's -o switch now knows a new log output
3408 mode "with-unit". The output it generates is very similar to the
3409 regular "short" mode, but displays the unit name instead of the
3410 syslog tag for each log line. Also, the date is shown with timezone
3411 information. This mode is probably more useful than the classic
3412 "short" output mode for most purposes, except where pixel-perfect
3413 compatibility with classic /var/log/messages formatting is required.
3415 * A new --dump-bus-properties switch has been added to the systemd
3416 binary, which may be used to dump all supported D-Bus properties.
3417 (Options which are still supported, but are deprecated, are *not*
3420 * sd-bus gained a set of new calls:
3421 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
3422 enable/disable the "floating" state of a bus slot object,
3423 i.e. whether the slot object pins the bus it is allocated for into
3424 memory or if the bus slot object gets disconnected when the bus goes
3425 away. sd_bus_open_with_description(),
3426 sd_bus_open_user_with_description(),
3427 sd_bus_open_system_with_description() may be used to allocate bus
3428 objects and set their description string already during allocation.
3430 * sd-event gained support for watching inotify events from the event
3431 loop, in an efficient way, sharing inotify handles between multiple
3432 users. For this a new function sd_event_add_inotify() has been added.
3434 * sd-event and sd-bus gained support for calling special user-supplied
3435 destructor functions for userdata pointers associated with
3436 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
3437 functions sd_bus_slot_set_destroy_callback,
3438 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
3439 sd_bus_track_get_destroy_callback,
3440 sd_event_source_set_destroy_callback,
3441 sd_event_source_get_destroy_callback have been added.
3443 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
3445 * PID 1 will now automatically reschedule .timer units whenever the
3446 local timezone changes. (They previously got rescheduled
3447 automatically when the system clock changed.)
3449 * New documentation has been added to document cgroups delegation,
3450 portable services and the various code quality tools we have set up:
3452 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
3453 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
3454 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
3456 * The Boot Loader Specification has been added to the source tree.
3458 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
3460 While moving it into our source tree we have updated it and further
3461 changes are now accepted through the usual github PR workflow.
3463 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
3464 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
3465 earlier PAM modules. The data in these fields is used to initialize
3466 the session scope's resource properties. Thus external PAM modules
3467 may now configure per-session limits, for example sourced from
3468 external user databases.
3470 * socket units with Accept=yes will now maintain a "refused" counter in
3471 addition to the existing "accepted" counter, counting connections
3472 refused due to the enforced limits.
3474 * The "systemd-path search-binaries-default" command may now be use to
3475 query the default, built-in $PATH PID 1 will pass to the services it
3478 * A new unit file setting PrivateMounts= has been added. It's a boolean
3479 option. If enabled the unit's processes are invoked in their own file
3480 system namespace. Note that this behaviour is also implied if any
3481 other file system namespacing options (such as PrivateTmp=,
3482 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
3483 primarily useful for services that do not use any of the other file
3484 system namespacing options. One such service is systemd-udevd.service
3485 where this is now used by default.
3487 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
3488 when the system is booted in UEFI "secure mode".
3490 * A new unit "system-update-pre.target" is added, which defines an
3491 optional synchronization point for offline system updates, as
3492 implemented by the pre-existing "system-update.target" unit. It
3493 allows ordering services before the service that executes the actual
3494 update process in a generic way.
3496 * Systemd now emits warnings whenever .include syntax is used.
3498 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
3499 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
3500 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
3501 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
3502 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
3503 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
3504 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
3505 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
3506 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
3507 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
3508 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
3509 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
3510 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
3511 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
3512 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
3513 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
3514 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
3515 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
3516 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
3517 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
3518 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
3519 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
3520 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
3521 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
3522 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
3523 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
3524 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
3525 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
3526 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
3528 — Berlin, 2018-06-22
3532 * The MemoryAccounting= unit property now defaults to on. After
3533 discussions with the upstream control group maintainers we learnt
3534 that the negative impact of cgroup memory accounting on current
3535 kernels is finally relatively minimal, so that it should be safe to
3536 enable this by default without affecting system performance. Besides
3537 memory accounting only task accounting is turned on by default, all
3538 other forms of resource accounting (CPU, IO, IP) remain off for now,
3539 because it's not clear yet that their impact is small enough to move
3540 from opt-in to opt-out. We recommend downstreams to leave memory
3541 accounting on by default if kernel 4.14 or higher is primarily
3542 used. On very resource constrained systems or when support for old
3543 kernels is a necessity, -Dmemory-accounting-default=false can be used
3544 to revert this change.
3546 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
3547 %udev_rules_update) and the journal catalog (%journal_catalog_update)
3548 from the upgrade scriptlets of individual packages now do nothing.
3549 Transfiletriggers have been added which will perform those updates
3550 once at the end of the transaction.
3552 Similar transfiletriggers have been added to execute any sysctl.d
3553 and binfmt.d rules. Thus, it should be unnecessary to provide any
3554 scriptlets to execute this configuration from package installation
3557 * systemd-sysusers gained a mode where the configuration to execute is
3558 specified on the command line, but this configuration is not executed
3559 directly, but instead it is merged with the configuration on disk,
3560 and the result is executed. This is useful for package installation
3561 scripts which want to create the user before installing any files on
3562 disk (in case some of those files are owned by that user), while
3563 still allowing local admin overrides.
3565 This functionality is exposed to rpm scriptlets through a new
3566 %sysusers_create_package macro. Old %sysusers_create and
3567 %sysusers_create_inline macros are deprecated.
3569 A transfiletrigger for sysusers.d configuration is now installed,
3570 which means that it should be unnecessary to call systemd-sysusers from
3571 package installation scripts, unless the package installs any files
3572 owned by those newly-created users, in which case
3573 %sysusers_create_package should be used.
3575 * Analogous change has been done for systemd-tmpfiles: it gained a mode
3576 where the command-line configuration is merged with the configuration
3577 on disk. This is exposed as the new %tmpfiles_create_package macro,
3578 and %tmpfiles_create is deprecated. A transfiletrigger is installed
3579 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
3580 from package installation scripts.
3582 * sysusers.d configuration for a user may now also specify the group
3583 number, in addition to the user number ("u username 123:456"), or
3584 without the user number ("u username -:456").
3586 * Configution items for systemd-sysusers can now be specified as
3587 positional arguments when the new --inline switch is used.
3589 * The login shell of users created through sysusers.d may now be
3590 specified (previously, it was always /bin/sh for root and
3591 /sbin/nologin for other users).
3593 * systemd-analyze gained a new --global switch to look at global user
3594 configuration. It also gained a unit-paths verb to list the unit load
3595 paths that are compiled into systemd (which can be used with
3596 --systemd, --user, or --global).
3598 * udevadm trigger gained a new --settle/-w option to wait for any
3599 triggered events to finish (but just those, and not any other events
3600 which are triggered meanwhile).
3602 * The action that systemd-logind takes when the lid is closed and the
3603 machine is connected to external power can now be configured using
3604 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
3605 was determined by HandleLidSwitch=, and, for backwards compatibility,
3606 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
3608 * journalctl will periodically call sd_journal_process() to make it
3609 resilient against inotify queue overruns when journal files are
3610 rotated very quickly.
3612 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
3613 sd_bus_get_n_queued_write — may be used to check the number of
3614 pending bus messages.
3616 * systemd gained a new
3617 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
3618 which can be used to migrate foreign processes to scope and service
3619 units. The primary user for this new API is systemd itself: the
3620 systemd --user instance uses this call of the systemd --system
3621 instance to migrate processes if it itself gets the request to
3622 migrate processes and the kernel refuses this due to access
3623 restrictions. Thanks to this "systemd-run --scope --user …" works
3624 again in pure cgroup v2 environments when invoked from the user
3627 * A new TemporaryFileSystem= setting can be used to mask out part of
3628 the real file system tree with tmpfs mounts. This may be combined
3629 with BindPaths= and BindReadOnlyPaths= to hide files or directories
3630 not relevant to the unit, while still allowing some paths lower in
3631 the tree to be accessed.
3633 ProtectHome=tmpfs may now be used to hide user home and runtime
3634 directories from units, in a way that is mostly equivalent to
3635 "TemporaryFileSystem=/home /run/user /root".
3637 * Non-service units are now started with KeyringMode=shared by default.
3638 This means that mount and swapon and other mount tools have access
3639 to keys in the main keyring.
3641 * /sys/fs/bpf is now mounted automatically.
3643 * QNX virtualization is now detected by systemd-detect-virt and may
3644 be used in ConditionVirtualization=.
3646 * IPAccounting= may now be enabled also for slice units.
3648 * A new -Dsplit-bin= build configuration switch may be used to specify
3649 whether bin and sbin directories are merged, or if they should be
3650 included separately in $PATH and various listings of executable
3651 directories. The build configuration scripts will try to autodetect
3652 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
3653 system, but distributions are encouraged to configure this
3656 * A new -Dok-color= build configuration switch may be used to change
3657 the colour of "OK" status messages.
3659 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
3660 PrivateNetwork=yes was buggy in previous versions of systemd. This
3661 means that after the upgrade and daemon-reexec, any such units must
3664 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
3665 will not exclude read-only files owned by root from cleanup.
3667 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
3668 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
3669 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
3670 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
3671 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
3672 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
3673 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
3674 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
3675 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
3676 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
3677 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
3678 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
3679 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
3680 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
3681 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
3682 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
3684 — Warsaw, 2018-03-05
3688 * Some keyboards come with a zoom see-saw or rocker which until now got
3689 mapped to the Linux "zoomin/out" keys in hwdb. However, these
3690 keycodes are not recognized by any major desktop. They now produce
3691 Up/Down key events so that they can be used for scrolling.
3693 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
3694 slightly: previously, if an argument was specified for lines of this
3695 type (i.e. the right-most column was set) this string was appended to
3696 existing files each time systemd-tmpfiles was run. This behaviour was
3697 different from what the documentation said, and not particularly
3698 useful, as repeated systemd-tmpfiles invocations would not be
3699 idempotent and grow such files without bounds. With this release
3700 behaviour has been altered to match what the documentation says:
3701 lines of this type only have an effect if the indicated files don't
3702 exist yet, and only then the argument string is written to the file.
3704 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
3705 systemd-tmpfiles behaviour: previously, read-only files owned by root
3706 were always excluded from the file "aging" algorithm (i.e. the
3707 automatic clean-up of directories like /tmp based on
3708 atime/mtime/ctime). We intend to drop this restriction, and age files
3709 by default even when owned by root and read-only. This behaviour was
3710 inherited from older tools, but there have been requests to remove
3711 it, and it's not obvious why this restriction was made in the first
3712 place. Please speak up now, if you are aware of software that reqires
3713 this behaviour, otherwise we'll remove the restriction in v238.
3715 * A new environment variable $SYSTEMD_OFFLINE is now understood by
3716 systemctl. It takes a boolean argument. If on, systemctl assumes it
3717 operates on an "offline" OS tree, and will not attempt to talk to the
3718 service manager. Previously, this mode was implicitly enabled if a
3719 chroot() environment was detected, and this new environment variable
3720 now provides explicit control.
3722 * .path and .socket units may now be created transiently, too.
3723 Previously only service, mount, automount and timer units were
3724 supported as transient units. The systemd-run tool has been updated
3725 to expose this new functionality, you may hence use it now to bind
3726 arbitrary commands to path or socket activation on-the-fly from the
3727 command line. Moreover, almost all properties are now exposed for the
3728 unit types that already supported transient operation.
3730 * The systemd-mount command gained support for a new --owner= parameter
3731 which takes a user name, which is then resolved and included in uid=
3732 and gid= mount options string of the file system to mount.
3734 * A new unit condition ConditionControlGroupController= has been added
3735 that checks whether a specific cgroup controller is available.
3737 * Unit files, udev's .link files, and systemd-networkd's .netdev and
3738 .network files all gained support for a new condition
3739 ConditionKernelVersion= for checking against specific kernel
3742 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
3743 support for configuring device flags in the Flags= setting. In the
3744 same files, the [Tunnel] section gained support for configuring
3745 AllowLocalRemote=. The [Route] section in .network files gained
3746 support for configuring InitialCongestionWindow=,
3747 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
3748 understands RapidCommit=.
3750 * systemd-networkd's DHCPv6 support gained support for Prefix
3753 * sd-bus gained support for a new "watch-bind" feature. When this
3754 feature is enabled, an sd_bus connection may be set up to connect to
3755 an AF_UNIX socket in the file system as soon as it is created. This
3756 functionality is useful for writing early-boot services that
3757 automatically connect to the system bus as soon as it is started,
3758 without ugly time-based polling. systemd-networkd and
3759 systemd-resolved have been updated to make use of this
3760 functionality. busctl exposes this functionality in a new
3761 --watch-bind= command line switch.
3763 * sd-bus will now optionally synthesize a local "Connected" signal as
3764 soon as a D-Bus connection is set up fully. This message mirrors the
3765 already existing "Disconnected" signal which is synthesized when the
3766 connection is terminated. This signal is generally useful but
3767 particularly handy in combination with the "watch-bind" feature
3768 described above. Synthesizing of this message has to be requested
3769 explicitly through the new API call sd_bus_set_connected_signal(). In
3770 addition a new call sd_bus_is_ready() has been added that checks
3771 whether a connection is fully set up (i.e. between the "Connected" and
3772 "Disconnected" signals).
3774 * sd-bus gained two new calls sd_bus_request_name_async() and
3775 sd_bus_release_name_async() for asynchronously registering bus
3776 names. Similar, there is now sd_bus_add_match_async() for installing
3777 a signal match asynchronously. All of systemd's own services have
3778 been updated to make use of these calls. Doing these operations
3779 asynchronously has two benefits: it reduces the risk of deadlocks in
3780 case of cyclic dependencies between bus services, and it speeds up
3781 service initialization since synchronization points for bus
3782 round-trips are removed.
3784 * sd-bus gained two new calls sd_bus_match_signal() and
3785 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
3786 and sd_bus_add_match_async() but instead of taking a D-Bus match
3787 string take match fields as normal function parameters.
3789 * sd-bus gained two new calls sd_bus_set_sender() and
3790 sd_bus_message_set_sender() for setting the sender name of outgoing
3791 messages (either for all outgoing messages or for just one specific
3792 one). These calls are only useful in direct connections as on
3793 brokered connections the broker fills in the sender anyway,
3794 overwriting whatever the client filled in.
3796 * sd-event gained a new pseudo-handle that may be specified on all API
3797 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
3798 used this refers to the default event loop object of the calling
3799 thread. Note however that this does not implicitly allocate one —
3800 which has to be done prior by using sd_event_default(). Similarly
3801 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
3802 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
3803 to the default bus of the specified type of the calling thread. Here
3804 too this does not implicitly allocate bus connection objects, this
3805 has to be done prior with sd_bus_default() and friends.
3807 * sd-event gained a new call pair
3808 sd_event_source_{get|set}_io_fd_own(). This may be used to request
3809 automatic closure of the file descriptor an IO event source watches
3810 when the event source is destroyed.
3812 * systemd-networkd gained support for natively configuring WireGuard
3815 * In previous versions systemd synthesized user records both for the
3816 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
3817 internally. In order to simplify distribution-wide renames of the
3818 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
3819 new transitional flag file has been added: if
3820 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
3821 user and group record within the systemd codebase is disabled.
3823 * systemd-notify gained a new --uid= option for selecting the source
3824 user/UID to use for notification messages sent to the service
3827 * journalctl gained a new --grep= option to list only entries in which
3828 the message matches a certain pattern. By default matching is case
3829 insensitive if the pattern is lowercase, and case sensitive
3830 otherwise. Option --case-sensitive=yes|no can be used to override
3831 this an specify case sensitivity or case insensitivity.
3833 * There's now a "systemd-analyze service-watchdogs" command for printing
3834 the current state of the service runtime watchdog, and optionally
3835 enabling or disabling the per-service watchdogs system-wide if given a
3836 boolean argument (i.e. the concept you configure in WatchdogSec=), for
3837 debugging purposes. There's also a kernel command line option
3838 systemd.service_watchdogs= for controlling the same.
3840 * Two new "log-level" and "log-target" options for systemd-analyze were
3841 added that merge the now deprecated get-log-level, set-log-level and
3842 get-log-target, set-log-target pairs. The deprecated options are still
3843 understood for backwards compatibility. The two new options print the
3844 current value when no arguments are given, and set them when a
3845 level/target is given as an argument.
3847 * sysusers.d's "u" lines now optionally accept both a UID and a GID
3848 specification, separated by a ":" character, in order to create users
3849 where UID and GID do not match.
3851 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
3852 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
3853 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
3854 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
3855 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
3856 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
3857 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
3858 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
3859 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
3860 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
3861 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
3862 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
3863 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
3864 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
3865 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
3866 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
3867 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
3868 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
3869 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
3870 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
3877 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
3878 in v235 has been extended to also set the dummy.ko module option
3879 numdummies=0, preventing the kernel from automatically creating
3880 dummy0. All dummy interfaces must now be explicitly created.
3882 * Unknown '%' specifiers in configuration files are now rejected. This
3883 applies to units and tmpfiles.d configuration. Any percent characters
3884 that are followed by a letter or digit that are not supposed to be
3885 interpreted as the beginning of a specifier should be escaped by
3886 doubling ("%%"). (So "size=5%" is still accepted, as well as
3887 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
3888 valid specifiers today.)
3890 * systemd-resolved now maintains a new dynamic
3891 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
3892 recommended to make /etc/resolv.conf a symlink to it. This file
3893 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
3894 includes dynamically acquired search domains, achieving more correct
3895 DNS resolution by software that bypasses local DNS APIs such as NSS.
3897 * The "uaccess" udev tag has been dropped from /dev/kvm and
3898 /dev/dri/renderD*. These devices now have the 0666 permissions by
3899 default (but this may be changed at build-time). /dev/dri/renderD*
3900 will now be owned by the "render" group along with /dev/kfd.
3902 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
3903 systemd-journal-gatewayd.service and
3904 systemd-journal-upload.service. This means "nss-systemd" must be
3905 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
3906 services are resolved properly.
3908 * In /etc/fstab two new mount options are now understood:
3909 x-systemd.makefs and x-systemd.growfs. The former has the effect that
3910 the configured file system is formatted before it is mounted, the
3911 latter that the file system is resized to the full block device size
3912 after it is mounted (i.e. if the file system is smaller than the
3913 partition it resides on, it's grown). This is similar to the fsck
3914 logic in /etc/fstab, and pulls in systemd-makefs@.service and
3915 systemd-growfs@.service as necessary, similar to
3916 systemd-fsck@.service. Resizing is currently only supported on ext4
3919 * In systemd-networkd, the IPv6 RA logic now optionally may announce
3920 DNS server and domain information.
3922 * Support for the LUKS2 on-disk format for encrypted partitions has
3923 been added. This requires libcryptsetup2 during compilation and
3926 * The systemd --user instance will now signal "readiness" when its
3927 basic.target unit has been reached, instead of when the run queue ran
3928 empty for the first time.
3930 * Tmpfiles.d with user configuration are now also supported.
3931 systemd-tmpfiles gained a new --user switch, and snippets placed in
3932 ~/.config/user-tmpfiles.d/ and corresponding directories will be
3933 executed by systemd-tmpfiles --user running in the new
3934 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
3935 running in the user session.
3937 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
3938 %S resolves to the top-level state directory (/var/lib for the system
3939 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
3940 top-level cache directory (/var/cache for the system instance,
3941 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
3942 logs directory (/var/log for the system instance,
3943 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
3944 existing %t specifier, that resolves to the top-level runtime
3945 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
3948 * journalctl learnt a new parameter --output-fields= for limiting the
3949 set of journal fields to output in verbose and JSON output modes.
3951 * systemd-timesyncd's configuration file gained a new option
3952 RootDistanceMaxSec= for setting the maximum root distance of servers
3953 it'll use, as well as the new options PollIntervalMinSec= and
3954 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
3956 * bootctl gained a new command "list" for listing all available boot
3957 menu items on systems that follow the boot loader specification.
3959 * systemctl gained a new --dry-run switch that shows what would be done
3960 instead of doing it, and is currently supported by the shutdown and
3963 * ConditionSecurity= can now detect the TOMOYO security module.
3965 * Unit file [Install] sections are now also respected in unit drop-in
3966 files. This is intended to be used by drop-ins under /usr/lib/.
3968 * systemd-firstboot may now also set the initial keyboard mapping.
3970 * Udev "changed" events for devices which are exposed as systemd
3971 .device units are now propagated to units specified in
3972 ReloadPropagatedFrom= as reload requests.
3974 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
3975 unit template name (i.e. a name in the form of 'foobar@.service',
3976 without the instance component between the '@' and - the '.'), then
3977 the escaped sysfs path of the device is automatically used as the
3980 * SystemCallFilter= in unit files has been extended so that an "errno"
3981 can be specified individually for each system call. Example:
3982 SystemCallFilter=~uname:EILSEQ.
3984 * The cgroup delegation logic has been substantially updated. Delegate=
3985 now optionally takes a list of controllers (instead of a boolean, as
3986 before), which lists the controllers to delegate at least.
3988 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
3990 * A new LogLevelMax= setting configures the maximum log level any
3991 process of the service may log at (i.e. anything with a lesser
3992 priority than what is specified is automatically dropped). A new
3993 LogExtraFields= setting allows configuration of additional journal
3994 fields to attach to all log records generated by any of the unit's
3997 * New StandardInputData= and StandardInputText= settings along with the
3998 new option StandardInput=data may be used to configure textual or
3999 binary data that shall be passed to the executed service process via
4000 standard input, encoded in-line in the unit file.
4002 * StandardInput=, StandardOutput= and StandardError= may now be used to
4003 connect stdin/stdout/stderr of executed processes directly with a
4004 file or AF_UNIX socket in the file system, using the new "file:" option.
4006 * A new unit file option CollectMode= has been added, that allows
4007 tweaking the garbage collection logic for units. It may be used to
4008 tell systemd to garbage collect units that have failed automatically
4009 (normally it only GCs units that exited successfully). systemd-run
4010 and systemd-mount expose this new functionality with a new -G option.
4012 * "machinectl bind" may now be used to bind mount non-directories
4013 (i.e. regularfiles, devices, fifos, sockets).
4015 * systemd-analyze gained a new verb "calendar" for validating and
4016 testing calendar time specifications to use for OnCalendar= in timer
4017 units. Besides validating the expression it will calculate the next
4018 time the specified expression would elapse.
4020 * In addition to the pre-existing FailureAction= unit file setting
4021 there's now SuccessAction=, for configuring a shutdown action to
4022 execute when a unit completes successfully. This is useful in
4023 particular inside containers that shall terminate after some workload
4024 has been completed. Also, both options are now supported for all unit
4025 types, not just services.
4027 * networkds's IP rule support gained two new options
4028 IncomingInterface= and OutgoingInterface= for configuring the incoming
4029 and outgoing interfaces of configured rules. systemd-networkd also
4030 gained support for "vxcan" network devices.
4032 * networkd gained a new setting RequiredForOnline=, taking a
4033 boolean. If set, systemd-wait-online will take it into consideration
4034 when determining that the system is up, otherwise it will ignore the
4035 interface for this purpose.
4037 * The sd_notify() protocol gained support for a new operation: with
4038 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
4039 store again, ahead of POLLHUP or POLLERR when they are removed
4042 * A new document doc/UIDS-GIDS.md has been added to the source tree,
4043 that documents the UID/GID range and assignment assumptions and
4044 requirements of systemd.
4046 * The watchdog device PID 1 will ping may now be configured through the
4047 WatchdogDevice= configuration file setting, or by setting the
4048 systemd.watchdog_service= kernel commandline option.
4050 * systemd-resolved's gained support for registering DNS-SD services on
4051 the local network using MulticastDNS. Services may either be
4052 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
4053 the same dir below /run, /usr/lib), or through its D-Bus API.
4055 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
4056 extend the effective start, runtime, and stop time. The service must
4057 continue to send EXTEND_TIMEOUT_USEC within the period specified to
4058 prevent the service manager from making the service as timedout.
4060 * systemd-resolved's DNSSEC support gained support for RFC 8080
4061 (Ed25519 keys and signatures).
4063 * The systemd-resolve command line tool gained a new set of options
4064 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
4065 --set-nta= and --revert to configure per-interface DNS configuration
4066 dynamically during runtime. It's useful for pushing DNS information
4067 into systemd-resolved from DNS hook scripts that various interface
4068 managing software supports (such as pppd).
4070 * systemd-nspawn gained a new --network-namespace-path= command line
4071 option, which may be used to make a container join an existing
4072 network namespace, by specifying a path to a "netns" file.
4074 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
4075 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
4076 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
4077 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
4078 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
4079 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
4080 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
4081 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
4082 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
4083 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
4084 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
4085 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
4086 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
4087 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
4088 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
4089 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
4090 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
4091 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
4092 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
4093 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
4094 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
4095 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
4096 Jędrzejewski-Szmek, Zeal Jagannatha
4098 — Berlin, 2017-12-14
4102 * INCOMPATIBILITY: systemd-logind.service and other long-running
4103 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
4104 communication with the outside. This generally improves security of
4105 the system, and is in almost all cases a safe and good choice, as
4106 these services do not and should not provide any network-facing
4107 functionality. However, systemd-logind uses the glibc NSS API to
4108 query the user database. This creates problems on systems where NSS
4109 is set up to directly consult network services for user database
4110 lookups. In particular, this creates incompatibilities with the
4111 "nss-nis" module, which attempts to directly contact the NIS/YP
4112 network servers it is configured for, and will now consistently
4113 fail. In such cases, it is possible to turn off IP sandboxing for
4114 systemd-logind.service (set IPAddressDeny= in its [Service] section
4115 to the empty string, via a .d/ unit file drop-in). Downstream
4116 distributions might want to update their nss-nis packaging to include
4117 such a drop-in snippet, accordingly, to hide this incompatibility
4118 from the user. Another option is to make use of glibc's nscd service
4119 to proxy such network requests through a privilege-separated, minimal
4120 local caching daemon, or to switch to more modern technologies such
4121 sssd, whose NSS hook-ups generally do not involve direct network
4122 access. In general, we think it's definitely time to question the
4123 implementation choices of nss-nis, i.e. whether it's a good idea
4124 today to embed a network-facing loadable module into all local
4125 processes that need to query the user database, including the most
4126 trivial and benign ones, such as "ls". For more details about
4127 IPAddressDeny= see below.
4129 * A new modprobe.d drop-in is now shipped by default that sets the
4130 bonding module option max_bonds=0. This overrides the kernel default,
4131 to avoid conflicts and ambiguity as to whether or not bond0 should be
4132 managed by systemd-networkd or not. This resolves multiple issues
4133 with bond0 properties not being applied, when bond0 is configured
4134 with systemd-networkd. Distributors may choose to not package this,
4135 however in that case users will be prevented from correctly managing
4136 bond0 interface using systemd-networkd.
4138 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
4139 which print the logging level and target of the system manager. They
4140 complement the existing "set-log-level" and "set-log-target" verbs
4141 used to change those values.
4143 * journald.conf gained a new boolean setting ReadKMsg= which defaults
4144 to on. If turned off kernel log messages will not be read by
4145 systemd-journald or included in the logs. It also gained a new
4146 setting LineMax= for configuring the maximum line length in
4147 STDOUT/STDERR log streams. The new default for this value is 48K, up
4148 from the previous hardcoded 2048.
4150 * A new unit setting RuntimeDirectoryPreserve= has been added, which
4151 allows more detailed control of what to do with a runtime directory
4152 configured with RuntimeDirectory= (i.e. a directory below /run or
4153 $XDG_RUNTIME_DIR) after a unit is stopped.
4155 * The RuntimeDirectory= setting for units gained support for creating
4156 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
4157 one top-level directory.
4159 * Units gained new options StateDirectory=, CacheDirectory=,
4160 LogsDirectory= and ConfigurationDirectory= which are closely related
4161 to RuntimeDirectory= but manage per-service directories below
4162 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
4163 possible to write unit files which when activated automatically gain
4164 properly owned service specific directories in these locations, thus
4165 making unit files self-contained and increasing compatibility with
4166 stateless systems and factory reset where /etc or /var are
4167 unpopulated at boot. Matching these new settings there's also
4168 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
4169 ConfigurationDirectoryMode= for configuring the access mode of these
4170 directories. These settings are particularly useful in combination
4171 with DynamicUser=yes as they provide secure, properly-owned,
4172 writable, and stateful locations for storage, excluded from the
4173 sandbox that such services live in otherwise.
4175 * Automake support has been removed from this release. systemd is now
4178 * systemd-journald will now aggressively cache client metadata during
4179 runtime, speeding up log write performance under pressure. This comes
4180 at a small price though: as much of the metadata is read
4181 asynchronously from /proc/ (and isn't implicitly attached to log
4182 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
4183 metadata stored alongside a log entry might be slightly
4184 out-of-date. Previously it could only be slightly newer than the log
4185 message. The time window is small however, and given that the kernel
4186 is unlikely to be improved anytime soon in this regard, this appears
4189 * nss-myhostname/systemd-resolved will now by default synthesize an
4190 A/AAAA resource record for the "_gateway" hostname, pointing to the
4191 current default IP gateway. Previously it did that for the "gateway"
4192 name, hampering adoption, as some distributions wanted to leave that
4193 hostname open for local use. The old behaviour may still be
4194 requested at build time.
4196 * systemd-networkd's [Address] section in .network files gained a new
4197 Scope= setting for configuring the IP address scope. The [Network]
4198 section gained a new boolean setting ConfigureWithoutCarrier= that
4199 tells systemd-networkd to ignore link sensing when configuring the
4200 device. The [DHCP] section gained a new Anonymize= boolean option for
4201 turning on a number of options suggested in RFC 7844. A new
4202 [RoutingPolicyRule] section has been added for configuring the IP
4203 routing policy. The [Route] section has gained support for a new
4204 Type= setting which permits configuring
4205 blackhole/unreachable/prohibit routes.
4207 * The [VRF] section in .netdev files gained a new Table= setting for
4208 configuring the routing table to use. The [Tunnel] section gained a
4209 new Independent= boolean field for configuring tunnels independent of
4210 an underlying network interface. The [Bridge] section gained a new
4211 GroupForwardMask= option for configuration of propagation of link
4212 local frames between bridge ports.
4214 * The WakeOnLan= setting in .link files gained support for a number of
4215 new modes. A new TCP6SegmentationOffload= setting has been added for
4216 configuring TCP/IPv6 hardware segmentation offload.
4218 * The IPv6 RA sender implementation may now optionally send out RDNSS
4219 and RDNSSL records to supply DNS configuration to peers.
4221 * systemd-nspawn gained support for a new --system-call-filter= command
4222 line option for adding and removing entries in the default system
4223 call filter it applies. Moreover systemd-nspawn has been changed to
4224 implement a system call allow list instead of a deny list.
4226 * systemd-run gained support for a new --pipe command line option. If
4227 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
4228 are directly passed on to the activated transient service
4229 executable. This allows invoking arbitrary processes as systemd
4230 services (for example to take benefit of dependency management,
4231 accounting management, resource management or log management that is
4232 done automatically for services) — while still allowing them to be
4233 integrated in a classic UNIX shell pipeline.
4235 * When a service sends RELOAD=1 via sd_notify() and reload propagation
4236 using ReloadPropagationTo= is configured, a reload is now propagated
4237 to configured units. (Previously this was only done on explicitly
4238 requested reloads, using "systemctl reload" or an equivalent
4241 * For each service unit a restart counter is now kept: it is increased
4242 each time the service is restarted due to Restart=, and may be
4243 queried using "systemctl show -p NRestarts …".
4245 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
4246 @signal and @timer have been added, for usage with SystemCallFilter=
4247 in unit files and the new --system-call-filter= command line option
4248 of systemd-nspawn (see above).
4250 * ExecStart= lines in unit files gained two new modifiers: when a
4251 command line is prefixed with "!" the command will be executed as
4252 configured, except for the credentials applied by
4253 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
4254 "+", but does still apply namespacing options unlike "+". There's
4255 also "!!" now, which is mostly identical, but becomes a NOP on
4256 systems that support ambient capabilities. This is useful to write
4257 unit files that work with ambient capabilities where possible but
4258 automatically fall back to traditional privilege dropping mechanisms
4259 on systems where this is not supported.
4261 * ListenNetlink= settings in socket units now support RDMA netlink
4264 * A new unit file setting LockPersonality= has been added which permits
4265 locking down the chosen execution domain ("personality") of a service
4268 * A new special target "getty-pre.target" has been added, which is
4269 ordered before all text logins, and may be used to order services
4270 before textual logins acquire access to the console.
4272 * systemd will now attempt to load the virtio-rng.ko kernel module very
4273 early on if a VM environment supporting this is detected. This should
4274 improve entropy during early boot in virtualized environments.
4276 * A _netdev option is now supported in /etc/crypttab that operates in a
4277 similar way as the same option in /etc/fstab: it permits configuring
4278 encrypted devices that need to be ordered after the network is up.
4279 Following this logic, two new special targets
4280 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
4281 added that are to cryptsetup.target what remote-fs.target and
4282 remote-fs-pre.target are to local-fs.target.
4284 * Service units gained a new UnsetEnvironment= setting which permits
4285 unsetting specific environment variables for services that are
4286 normally passed to it (for example in order to mask out locale
4287 settings for specific services that can't deal with it).
4289 * Units acquired a new boolean option IPAccounting=. When turned on, IP
4290 traffic accounting (packet count as well as byte count) is done for
4291 the service, and shown as part of "systemctl status" or "systemd-run
4294 * Service units acquired two new options IPAddressAllow= and
4295 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
4296 for configuring a simple IP access control list for all sockets of
4297 the unit. These options are available also on .slice and .socket
4298 units, permitting flexible access list configuration for individual
4299 services as well as groups of services (as defined by a slice unit),
4300 including system-wide. Note that IP ACLs configured this way are
4301 enforced on every single IPv4 and IPv6 socket created by any process
4302 of the service unit, and apply to ingress as well as egress traffic.
4304 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
4305 structured log message is generated each time the unit is stopped,
4306 containing information about the consumed resources of this
4309 * A new setting KeyringMode= has been added to unit files, which may be
4310 used to control how the kernel keyring is set up for executed
4313 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
4314 "systemctl kexec" and "systemctl exit" are now always asynchronous in
4315 behaviour (that is: these commands return immediately after the
4316 operation was enqueued instead of waiting for the operation to
4317 complete). Previously, "systemctl poweroff" and "systemctl reboot"
4318 were asynchronous on systems using systemd-logind (i.e. almost
4319 always, and like they were on sysvinit), and the other three commands
4320 were unconditionally synchronous. With this release this is cleaned
4321 up, and callers will see the same asynchronous behaviour on all
4322 systems for all five operations.
4324 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
4327 * .timer units now accept calendar specifications in other timezones
4328 than UTC or the local timezone.
4330 * The tmpfiles snippet var.conf has been changed to create
4331 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
4332 the "utmp" group already, and it appears to be generally understood
4333 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
4334 databases. Previously this was implemented correctly for all these
4335 databases excepts btmp, which has been opened up like this now
4336 too. Note that while the other databases are world-readable
4337 (i.e. 0644), btmp is not and remains more restrictive.
4339 * The systemd-resolve tool gained a new --reset-server-features
4340 switch. When invoked like this systemd-resolved will forget
4341 everything it learnt about the features supported by the configured
4342 upstream DNS servers, and restarts the feature probing logic on the
4343 next resolver look-up for them at the highest feature level
4346 * The status dump systemd-resolved sends to the logs upon receiving
4347 SIGUSR1 now also includes information about all DNS servers it is
4348 configured to use, and the features levels it probed for them.
4350 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
4351 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
4352 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
4353 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
4354 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
4355 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
4356 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
4357 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
4358 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
4359 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
4360 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
4361 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
4362 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
4363 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
4364 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
4365 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
4366 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
4367 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
4368 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
4369 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
4371 — Berlin, 2017-10-06
4375 * Meson is now supported as build system in addition to Automake. It is
4376 our plan to remove Automake in one of our next releases, so that
4377 Meson becomes our exclusive build system. Hence, please start using
4378 the Meson build system in your downstream packaging. There's plenty
4379 of documentation around how to use Meson, the extremely brief
4382 ./autogen.sh && ./configure && make && sudo make install
4386 meson build && ninja -C build && sudo ninja -C build install
4388 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
4389 which permits configuring a timeout on the time a job is
4390 running. This is particularly useful for setting timeouts on jobs for
4393 * Unit files gained two new options ConditionUser= and ConditionGroup=
4394 for conditionalizing units based on the identity of the user/group
4395 running a systemd user instance.
4397 * systemd-networkd now understands a new FlowLabel= setting in the
4398 [VXLAN] section of .network files, as well as a Priority= in
4399 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
4400 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
4401 gained support for configuration of GENEVE links, and IPv6 address
4402 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
4404 * .link files now understand a new Port= setting.
4406 * systemd-networkd's DHCP support gained support for DHCP option 119
4407 (domain search list).
4409 * systemd-networkd gained support for serving IPv6 address ranges using
4410 the Router Advertisement protocol. The new .network configuration
4411 section [IPv6Prefix] may be used to configure the ranges to
4412 serve. This is implemented based on a new, minimal, native server
4413 implementation of RA.
4415 * journalctl's --output= switch gained support for a new parameter
4416 "short-iso-precise" for a mode where timestamps are shown as precise
4419 * systemd-udevd's "net_id" builtin may now generate stable network
4420 interface names from IBM PowerVM VIO devices as well as ACPI platform
4423 * MulticastDNS support in systemd-resolved may now be explicitly
4424 enabled/disabled using the new MulticastDNS= configuration file
4427 * systemd-resolved may now optionally use libidn2 instead of the libidn
4428 for processing internationalized domain names. Support for libidn2
4429 should be considered experimental and should not be enabled by
4432 * "machinectl pull-tar" and related call may now do verification of
4433 downloaded images using SUSE-style .sha256 checksum files in addition
4434 to the already existing support for validating using Ubuntu-style
4437 * sd-bus gained support for a new sd_bus_message_appendv() call which
4438 is va_list equivalent of sd_bus_message_append().
4440 * sd-boot gained support for validating images using SHIM/MOK.
4442 * The SMACK code learnt support for "onlycap".
4444 * systemd-mount --umount is now much smarter in figuring out how to
4445 properly unmount a device given its mount or device path.
4447 * The code to call libnss_dns as a fallback from libnss_resolve when
4448 the communication with systemd-resolved fails was removed. This
4449 fallback was redundant and interfered with the [!UNAVAIL=return]
4450 suffix. See nss-resolve(8) for the recommended configuration.
4452 * systemd-logind may now be restarted without losing state. It stores
4453 the file descriptors for devices it manages in the system manager
4454 using the FDSTORE= mechanism. Please note that further changes in
4455 other components may be required to make use of this (for example
4456 Xorg has code to listen for stops of systemd-logind and terminate
4457 itself when logind is stopped or restarted, in order to avoid using
4458 stale file descriptors for graphical devices, which is now
4459 counterproductive and must be reverted in order for restarts of
4460 systemd-logind to be safe. See
4461 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
4463 * All kernel-install plugins are called with the environment variable
4464 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
4465 /etc/machine-id. If the machine ID could not be determined,
4466 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
4467 anything in the entry directory (passed as the second argument) if
4468 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
4469 temporary directory is passed as the entry directory and removed
4470 after all the plugins exit.
4472 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
4473 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
4474 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
4475 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
4476 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
4477 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
4478 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
4479 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
4480 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
4481 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
4482 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
4483 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
4484 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
4485 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
4486 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
4487 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
4488 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
4489 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
4490 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
4491 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
4492 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
4493 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
4494 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
4495 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
4496 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
4497 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
4498 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
4499 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
4500 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
4503 — Berlin, 2017-07-12
4507 * The "hybrid" control group mode has been modified to improve
4508 compatibility with "legacy" cgroups-v1 setups. Specifically, the
4509 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
4510 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
4511 cgroups-v1 hierarchy), the only externally visible change being that
4512 the cgroups-v2 hierarchy is also mounted, to
4513 /sys/fs/cgroup/unified. This should provide a large degree of
4514 compatibility with "legacy" cgroups-v1, while taking benefit of the
4515 better management capabilities of cgroups-v2.
4517 * The default control group setup mode may be selected both a boot-time
4518 via a set of kernel command line parameters (specifically:
4519 systemd.unified_cgroup_hierarchy= and
4520 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
4521 default selected on the configure command line
4522 (--with-default-hierarchy=). The upstream default is "hybrid"
4523 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
4524 this will change in a future systemd version to be "unified" (pure
4525 cgroups-v2 mode). The third option for the compile time option is
4526 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
4527 distributions to default to "hybrid" mode for release distributions,
4528 starting with v233. We recommend "unified" for development
4529 distributions (specifically: distributions such as Fedora's rawhide)
4530 as that's where things are headed in the long run. Use "legacy" for
4531 greatest stability and compatibility only.
4533 * Note one current limitation of "unified" and "hybrid" control group
4534 setup modes: the kernel currently does not permit the systemd --user
4535 instance (i.e. unprivileged code) to migrate processes between two
4536 disconnected cgroup subtrees, even if both are managed and owned by
4537 the user. This effectively means "systemd-run --user --scope" doesn't
4538 work when invoked from outside of any "systemd --user" service or
4539 scope. Specifically, it is not supported from session scopes. We are
4540 working on fixing this in a future systemd version. (See #3388 for
4541 further details about this.)
4543 * DBus policy files are now installed into /usr rather than /etc. Make
4544 sure your system has dbus >= 1.9.18 running before upgrading to this
4545 version, or override the install path with --with-dbuspolicydir= .
4547 * All python scripts shipped with systemd (specifically: the various
4548 tests written in Python) now require Python 3.
4550 * systemd unit tests can now run standalone (without the source or
4551 build directories), and can be installed into /usr/lib/systemd/tests/
4552 with 'make install-tests'.
4554 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
4555 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
4558 * Support for the %c, %r, %R specifiers in unit files has been
4559 removed. Specifiers are not supposed to be dependent on configuration
4560 in the unit file itself (so that they resolve the same regardless
4561 where used in the unit files), but these specifiers were influenced
4562 by the Slice= option.
4564 * The shell invoked by debug-shell.service now defaults to /bin/sh in
4565 all cases. If distributions want to use a different shell for this
4566 purpose (for example Fedora's /sbin/sushell) they need to specify
4567 this explicitly at configure time using --with-debug-shell=.
4569 * The confirmation spawn prompt has been reworked to offer the
4572 (c)ontinue, proceed without asking anymore
4573 (D)ump, show the state of the unit
4574 (f)ail, don't execute the command and pretend it failed
4576 (i)nfo, show a short summary of the unit
4577 (j)obs, show jobs that are in progress
4578 (s)kip, don't execute the command and pretend it succeeded
4579 (y)es, execute the command
4581 The 'n' choice for the confirmation spawn prompt has been removed,
4582 because its meaning was confusing.
4584 The prompt may now also be redirected to an alternative console by
4585 specifying the console as parameter to systemd.confirm_spawn=.
4587 * Services of Type=notify require a READY=1 notification to be sent
4588 during startup. If no such message is sent, the service now fails,
4589 even if the main process exited with a successful exit code.
4591 * Services that fail to start up correctly now always have their
4592 ExecStopPost= commands executed. Previously, they'd enter "failed"
4593 state directly, without executing these commands.
4595 * The option MulticastDNS= of network configuration files has acquired
4596 an actual implementation. With MulticastDNS=yes a host can resolve
4597 names of remote hosts and reply to mDNS A and AAAA requests.
4599 * When units are about to be started an additional check is now done to
4600 ensure that all dependencies of type BindsTo= (when used in
4601 combination with After=) have been started.
4603 * systemd-analyze gained a new verb "syscall-filter" which shows which
4604 system call groups are defined for the SystemCallFilter= unit file
4605 setting, and which system calls they contain.
4607 * A new system call filter group "@filesystem" has been added,
4608 consisting of various file system related system calls. Group
4609 "@reboot" has been added, covering reboot, kexec and shutdown related
4610 calls. Finally, group "@swap" has been added covering swap
4611 configuration related calls.
4613 * A new unit file option RestrictNamespaces= has been added that may be
4614 used to restrict access to the various process namespace types the
4615 Linux kernel provides. Specifically, it may be used to take away the
4616 right for a service unit to create additional file system, network,
4617 user, and other namespaces. This sandboxing option is particularly
4618 relevant due to the high amount of recently discovered namespacing
4619 related vulnerabilities in the kernel.
4621 * systemd-udev's .link files gained support for a new AutoNegotiation=
4622 setting for configuring Ethernet auto-negotiation.
4624 * systemd-networkd's .network files gained support for a new
4625 ListenPort= setting in the [DHCP] section to explicitly configure the
4626 UDP client port the DHCP client shall listen on.
4628 * .network files gained a new Unmanaged= boolean setting for explicitly
4629 excluding one or more interfaces from management by systemd-networkd.
4631 * The systemd-networkd ProxyARP= option has been renamed to
4632 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
4633 renamed to ReduceARPProxy=. The old names continue to be available
4636 * systemd-networkd gained support for configuring IPv6 Proxy NDP
4637 addresses via the new IPv6ProxyNDPAddress= .network file setting.
4639 * systemd-networkd's bonding device support gained support for two new
4640 configuration options ActiveSlave= and PrimarySlave=.
4642 * The various options in the [Match] section of .network files gained
4643 support for negative matching.
4645 * New systemd-specific mount options are now understood in /etc/fstab:
4647 x-systemd.mount-timeout= may be used to configure the maximum
4648 permitted runtime of the mount command.
4650 x-systemd.device-bound may be set to bind a mount point to its
4651 backing device unit, in order to automatically remove a mount point
4652 if its backing device is unplugged. This option may also be
4653 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
4654 on the block device, which is now automatically set for all CDROM
4655 drives, so that mounted CDs are automatically unmounted when they are
4656 removed from the drive.
4658 x-systemd.after= and x-systemd.before= may be used to explicitly
4659 order a mount after or before another unit or mount point.
4661 * Enqueued start jobs for device units are now automatically garbage
4662 collected if there are no jobs waiting for them anymore.
4664 * systemctl list-jobs gained two new switches: with --after, for every
4665 queued job the jobs it's waiting for are shown; with --before the
4666 jobs which it's blocking are shown.
4668 * systemd-nspawn gained support for ephemeral boots from disk images
4669 (or in other words: --ephemeral and --image= may now be
4670 combined). Moreover, ephemeral boots are now supported for normal
4671 directories, even if the backing file system is not btrfs. Of course,
4672 if the file system does not support file system snapshots or
4673 reflinks, the initial copy operation will be relatively expensive, but
4674 this should still be suitable for many use cases.
4676 * Calendar time specifications in .timer units now support
4677 specifications relative to the end of a month by using "~" instead of
4678 "-" as separator between month and day. For example, "*-02~03" means
4679 "the third last day in February". In addition a new syntax for
4680 repeated events has been added using the "/" character. For example,
4681 "9..17/2:00" means "every two hours from 9am to 5pm".
4683 * systemd-socket-proxyd gained a new parameter --connections-max= for
4684 configuring the maximum number of concurrent connections.
4686 * sd-id128 gained a new API for generating unique IDs for the host in a
4687 way that does not leak the machine ID. Specifically,
4688 sd_id128_get_machine_app_specific() derives an ID based on the
4689 machine ID a in well-defined, non-reversible, stable way. This is
4690 useful whenever an identifier for the host is needed but where the
4691 identifier shall not be useful to identify the system beyond the
4692 scope of the application itself. (Internally this uses HMAC-SHA256 as
4693 keyed hash function using the machine ID as input.)
4695 * NotifyAccess= gained a new supported value "exec". When set
4696 notifications are accepted from all processes systemd itself invoked,
4697 including all control processes.
4699 * .nspawn files gained support for defining overlay mounts using the
4700 Overlay= and OverlayReadOnly= options. Previously this functionality
4701 was only available on the systemd-nspawn command line.
4703 * systemd-nspawn's --bind= and --overlay= options gained support for
4704 bind/overlay mounts whose source lies within the container tree by
4705 prefixing the source path with "+".
4707 * systemd-nspawn's --bind= and --overlay= options gained support for
4708 automatically allocating a temporary source directory in /var/tmp
4709 that is removed when the container dies. Specifically, if the source
4710 directory is specified as empty string this mechanism is selected. An
4711 example usage is --overlay=+/var::/var, which creates an overlay
4712 mount based on the original /var contained in the image, overlaid
4713 with a temporary directory in the host's /var/tmp. This way changes
4714 to /var are automatically flushed when the container shuts down.
4716 * systemd-nspawn --image= option does now permit raw file system block
4717 devices (in addition to images containing partition tables, as
4720 * The disk image dissection logic in systemd-nspawn gained support for
4721 automatically setting up LUKS encrypted as well as Verity protected
4722 partitions. When a container is booted from an encrypted image the
4723 passphrase is queried at start-up time. When a container with Verity
4724 data is started, the root hash is search in a ".roothash" file
4725 accompanying the disk image (alternatively, pass the root hash via
4726 the new --root-hash= command line option).
4728 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
4729 be used to dissect disk images the same way as systemd-nspawn does
4730 it, following the Bootable Partition Specification. It may even be
4731 used to mount disk images with complex partition setups (including
4732 LUKS and Verity partitions) to a local host directory, in order to
4733 inspect them. This tool is not considered public API (yet), and is
4734 thus not installed into /usr/bin. Please do not rely on its
4735 existence, since it might go away or be changed in later systemd
4738 * A new generator "systemd-verity-generator" has been added, similar in
4739 style to "systemd-cryptsetup-generator", permitting automatic setup of
4740 Verity root partitions when systemd boots up. In order to make use of
4741 this your partition setup should follow the Discoverable Partitions
4742 Specification, and the GPT partition ID of the root file system
4743 partition should be identical to the upper 128bit of the Verity root
4744 hash. The GPT partition ID of the Verity partition protecting it
4745 should be the lower 128bit of the Verity root hash. If the partition
4746 image follows this model it is sufficient to specify a single
4747 "roothash=" kernel command line argument to both configure which root
4748 image and verity partition to use as well as the root hash for
4749 it. Note that systemd-nspawn's Verity support follows the same
4750 semantics, meaning that disk images with proper Verity data in place
4751 may be booted in containers with systemd-nspawn as well as on
4752 physical systems via the verity generator. Also note that the "mkosi"
4753 tool available at https://github.com/systemd/mkosi has been updated
4754 to generate Verity protected disk images following this scheme. In
4755 fact, it has been updated to generate disk images that optionally
4756 implement a complete UEFI SecureBoot trust chain, involving a signed
4757 kernel and initrd image that incorporates such a root hash as well as
4758 a Verity-enabled root partition.
4760 * The hardware database (hwdb) udev supports has been updated to carry
4761 accelerometer quirks.
4763 * All system services are now run with a fresh kernel keyring set up
4764 for them. The invocation ID is stored by default in it, thus
4765 providing a safe, non-overridable way to determine the invocation
4768 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
4769 options for bind mounting arbitrary paths in a service-specific
4770 way. When these options are used, arbitrary host or service files and
4771 directories may be mounted to arbitrary locations in the service's
4774 * Documentation has been added that lists all of systemd's low-level
4775 environment variables:
4777 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
4779 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
4780 whether a specific socket file descriptor matches a specified socket
4783 * systemd-firstboot has been updated to check for the
4784 systemd.firstboot= kernel command line option. It accepts a boolean
4785 and when set to false the first boot questions are skipped.
4787 * systemd-fstab-generator has been updated to check for the
4788 systemd.volatile= kernel command line option, which either takes an
4789 optional boolean parameter or the special value "state". If used the
4790 system may be booted in a "volatile" boot mode. Specifically,
4791 "systemd.volatile" is used, the root directory will be mounted as
4792 tmpfs, and only /usr is mounted from the actual root file system. If
4793 "systemd.volatile=state" is used, the root directory will be mounted
4794 as usual, but /var is mounted as tmpfs. This concept provides similar
4795 functionality as systemd-nspawn's --volatile= option, but provides it
4796 on physical boots. Use this option for implementing stateless
4797 systems, or testing systems with all state and/or configuration reset
4798 to the defaults. (Note though that many distributions are not
4799 prepared to boot up without a populated /etc or /var, though.)
4801 * systemd-gpt-auto-generator gained support for LUKS encrypted root
4802 partitions. Previously it only supported LUKS encrypted partitions
4803 for all other uses, except for the root partition itself.
4805 * Socket units gained support for listening on AF_VSOCK sockets for
4806 communication in virtualized QEMU environments.
4808 * The "configure" script gained a new option --with-fallback-hostname=
4809 for specifying the fallback hostname to use if none is configured in
4810 /etc/hostname. For example, by specifying
4811 --with-fallback-hostname=fedora it is possible to default to a
4812 hostname of "fedora" on pristine installations.
4814 * systemd-cgls gained support for a new --unit= switch for listing only
4815 the control groups of a specific unit. Similar --user-unit= has been
4816 added for listing only the control groups of a specific user unit.
4818 * systemd-mount gained a new --umount switch for unmounting a mount or
4819 automount point (and all mount/automount points below it).
4821 * systemd will now refuse full configuration reloads (via systemctl
4822 daemon-reload and related calls) unless at least 16MiB of free space
4823 are available in /run. This is a safety precaution in order to ensure
4824 that generators can safely operate after the reload completed.
4826 * A new unit file option RootImage= has been added, which has a similar
4827 effect as RootDirectory= but mounts the service's root directory from
4828 a disk image instead of plain directory. This logic reuses the same
4829 image dissection and mount logic that systemd-nspawn already uses,
4830 and hence supports any disk images systemd-nspawn supports, including
4831 those following the Discoverable Partition Specification, as well as
4832 Verity enabled images. This option enables systemd to run system
4833 services directly off disk images acting as resource bundles,
4834 possibly even including full integrity data.
4836 * A new MountAPIVFS= unit file option has been added, taking a boolean
4837 argument. If enabled /proc, /sys and /dev (collectively called the
4838 "API VFS") will be mounted for the service. This is only relevant if
4839 RootDirectory= or RootImage= is used for the service, as these mounts
4840 are of course in place in the host mount namespace anyway.
4842 * systemd-nspawn gained support for a new --pivot-root= switch. If
4843 specified the root directory within the container image is pivoted to
4844 the specified mount point, while the original root disk is moved to a
4845 different place. This option enables booting of ostree images
4846 directly with systemd-nspawn.
4848 * The systemd build scripts will no longer complain if the NTP server
4849 addresses are not changed from the defaults. Google now supports
4850 these NTP servers officially. We still recommend downstreams to
4851 properly register an NTP pool with the NTP pool project though.
4853 * coredumpctl gained a new "--reverse" option for printing the list
4854 of coredumps in reverse order.
4856 * coredumpctl will now show additional information about truncated and
4857 inaccessible coredumps, as well as coredumps that are still being
4858 processed. It also gained a new --quiet switch for suppressing
4859 additional informational message in its output.
4861 * coredumpctl gained support for only showing coredumps newer and/or
4862 older than specific timestamps, using the new --since= and --until=
4863 options, reminiscent of journalctl's options by the same name.
4865 * The systemd-coredump logic has been improved so that it may be reused
4866 to collect backtraces in non-compiled languages, for example in
4867 scripting languages such as Python.
4869 * machinectl will now show the UID shift of local containers, if user
4870 namespacing is enabled for them.
4872 * systemd will now optionally run "environment generator" binaries at
4873 configuration load time. They may be used to add environment
4874 variables to the environment block passed to services invoked. One
4875 user environment generator is shipped by default that sets up
4876 environment variables based on files dropped into /etc/environment.d
4877 and ~/.config/environment.d/.
4879 * systemd-resolved now includes the new, recently published 2017 DNSSEC
4882 * hostnamed has been updated to report a new chassis type of
4883 "convertible" to cover "foldable" laptops that can both act as a
4884 tablet and as a laptop, such as various Lenovo Yoga devices.
4886 Contributions from: Adrián López, Alexander Galanin, Alexander
4887 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
4888 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
4889 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
4890 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
4891 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
4892 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
4893 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
4894 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
4895 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
4896 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
4897 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
4898 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
4899 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
4900 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
4901 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
4902 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
4903 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
4904 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
4905 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
4906 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
4907 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
4908 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
4909 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
4910 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
4911 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
4912 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
4913 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
4916 — Berlin, 2017-03-01
4920 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
4921 RestrictAddressFamilies= enabled. These sandboxing options should
4922 generally be compatible with the various external udev call-out
4923 binaries we are aware of, however there may be exceptions, in
4924 particular when exotic languages for these call-outs are used. In
4925 this case, consider turning off these settings locally.
4927 * The new RemoveIPC= option can be used to remove IPC objects owned by
4928 the user or group of a service when that service exits.
4930 * The new ProtectKernelModules= option can be used to disable explicit
4931 load and unload operations of kernel modules by a service. In
4932 addition access to /usr/lib/modules is removed if this option is set.
4934 * ProtectSystem= option gained a new value "strict", which causes the
4935 whole file system tree with the exception of /dev, /proc, and /sys,
4936 to be remounted read-only for a service.
4938 * The new ProtectKernelTunables= option can be used to disable
4939 modification of configuration files in /sys and /proc by a service.
4940 Various directories and files are remounted read-only, so access is
4941 restricted even if the file permissions would allow it.
4943 * The new ProtectControlGroups= option can be used to disable write
4944 access by a service to /sys/fs/cgroup.
4946 * Various systemd services have been hardened with
4947 ProtectKernelTunables=yes, ProtectControlGroups=yes,
4948 RestrictAddressFamilies=.
4950 * Support for dynamically creating users for the lifetime of a service
4951 has been added. If DynamicUser=yes is specified, user and group IDs
4952 will be allocated from the range 61184..65519 for the lifetime of the
4953 service. They can be resolved using the new nss-systemd.so NSS
4954 module. The module must be enabled in /etc/nsswitch.conf. Services
4955 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
4956 any resources allocated by the service will be cleaned up when the
4957 service exits. They also have ProtectHome=read-only and
4958 ProtectSystem=strict enabled, so they are not able to make any
4959 permanent modifications to the system.
4961 * The nss-systemd module also always resolves root and nobody, making
4962 it possible to have no /etc/passwd or /etc/group files in minimal
4963 container or chroot environments.
4965 * Services may be started with their own user namespace using the new
4966 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
4967 under which the service is running are mapped. All other users are
4970 * Support for the cgroup namespace has been added to systemd-nspawn. If
4971 supported by kernel, the container system started by systemd-nspawn
4972 will have its own view of the cgroup hierarchy. This new behaviour
4973 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
4975 * The new MemorySwapMax= option can be used to limit the maximum swap
4976 usage under the unified cgroup hierarchy.
4978 * Support for the CPU controller in the unified cgroup hierarchy has
4979 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
4980 options. This controller requires out-of-tree patches for the kernel
4981 and the support is provisional.
4983 * Mount and automount units may now be created transiently
4984 (i.e. dynamically at runtime via the bus API, instead of requiring
4985 unit files in the file system).
4987 * systemd-mount is a new tool which may mount file systems – much like
4988 mount(8), optionally pulling in additional dependencies through
4989 transient .mount and .automount units. For example, this tool
4990 automatically runs fsck on a backing block device before mounting,
4991 and allows the automount logic to be used dynamically from the
4992 command line for establishing mount points. This tool is particularly
4993 useful when dealing with removable media, as it will ensure fsck is
4994 run – if necessary – before the first access and that the file system
4995 is quickly unmounted after each access by utilizing the automount
4996 logic. This maximizes the chance that the file system on the
4997 removable media stays in a clean state, and if it isn't in a clean
4998 state is fixed automatically.
5000 * LazyUnmount=yes option for mount units has been added to expose the
5001 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
5004 * /efi will be used as the mount point of the EFI boot partition, if
5005 the directory is present, and the mount point was not configured
5006 through other means (e.g. fstab). If /efi directory does not exist,
5007 /boot will be used as before. This makes it easier to automatically
5008 mount the EFI partition on systems where /boot is used for something
5011 * When operating on GPT disk images for containers, systemd-nspawn will
5012 now mount the ESP to /boot or /efi according to the same rules as PID
5013 1 running on a host. This allows tools like "bootctl" to operate
5014 correctly within such containers, in order to make container images
5015 bootable on physical systems.
5017 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
5019 * Two new user session targets have been added to support running
5020 graphical sessions under the systemd --user instance:
5021 graphical-session.target and graphical-session-pre.target. See
5022 systemd.special(7) for a description of how those targets should be
5025 * The vconsole initialization code has been significantly reworked to
5026 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
5027 support unicode keymaps. Font and keymap configuration will now be
5028 copied to all allocated virtual consoles.
5030 * FreeBSD's bhyve virtualization is now detected.
5032 * Information recorded in the journal for core dumps now includes the
5033 contents of /proc/mountinfo and the command line of the process at
5034 the top of the process hierarchy (which is usually the init process
5037 * systemd-journal-gatewayd learned the --directory= option to serve
5038 files from the specified location.
5040 * journalctl --root=… can be used to peruse the journal in the
5041 /var/log/ directories inside of a container tree. This is similar to
5042 the existing --machine= option, but does not require the container to
5045 * The hardware database has been extended to support
5046 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
5049 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
5050 specify the click rate for mice which include a horizontal wheel with
5051 a click rate that is different than the one for the vertical wheel.
5053 * systemd-run gained a new --wait option that makes service execution
5054 synchronous. (Specifically, the command will not return until the
5055 specified service binary exited.)
5057 * systemctl gained a new --wait option that causes the start command to
5058 wait until the units being started have terminated again.
5060 * A new journal output mode "short-full" has been added which displays
5061 timestamps with abbreviated English day names and adds a timezone
5062 suffix. Those timestamps include more information than the default
5063 "short" output mode, and can be passed directly to journalctl's
5064 --since= and --until= options.
5066 * /etc/resolv.conf will be bind-mounted into containers started by
5067 systemd-nspawn, if possible, so any changes to resolv.conf contents
5068 are automatically propagated to the container.
5070 * The number of instances for socket-activated services originating
5071 from a single IP address can be limited with
5072 MaxConnectionsPerSource=, extending the existing setting of
5075 * systemd-networkd gained support for vcan ("Virtual CAN") interface
5078 * .netdev and .network configuration can now be extended through
5081 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
5082 Segmentation Offload, Generic Receive Offload, Large Receive Offload
5083 can be enabled and disabled using the new UDPSegmentationOffload=,
5084 TCPSegmentationOffload=, GenericSegmentationOffload=,
5085 GenericReceiveOffload=, LargeReceiveOffload= options in the
5086 [Link] section of .link files.
5088 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
5089 Port VLAN ID can be configured for bridge devices using the new STP=,
5090 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
5091 section of .netdev files.
5093 * The route table to which routes received over DHCP or RA should be
5094 added can be configured with the new RouteTable= option in the [DHCP]
5095 and [IPv6AcceptRA] sections of .network files.
5097 * The Address Resolution Protocol can be disabled on links managed by
5098 systemd-networkd using the ARP=no setting in the [Link] section of
5101 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
5102 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
5103 encode information about the result and exit codes of the current
5104 service runtime cycle.
5106 * systemd-sysctl will now configure kernel parameters in the order
5107 they occur in the configuration files. This matches what sysctl
5108 has been traditionally doing.
5110 * kernel-install "plugins" that are executed to perform various
5111 tasks after a new kernel is added and before an old one is removed
5112 can now return a special value to terminate the procedure and
5113 prevent any later plugins from running.
5115 * Journald's SplitMode=login setting has been deprecated. It has been
5116 removed from documentation, and its use is discouraged. In a future
5117 release it will be completely removed, and made equivalent to current
5118 default of SplitMode=uid.
5120 * Storage=both option setting in /etc/systemd/coredump.conf has been
5121 removed. With fast LZ4 compression storing the core dump twice is not
5124 * The --share-system systemd-nspawn option has been replaced with an
5125 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
5126 this functionality is discouraged. In addition the variables
5127 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
5128 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
5129 individual namespaces.
5131 * "machinectl list" now shows the IP address of running containers in
5132 the output, as well as OS release information.
5134 * "loginctl list" now shows the TTY of each session in the output.
5136 * sd-bus gained new API calls sd_bus_track_set_recursive(),
5137 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
5138 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
5139 tracking objects in a "recursive" mode, where a single client can be
5140 counted multiple times, if it takes multiple references.
5142 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
5143 sd_bus_get_exit_on_disconnect(). They may be used to make a
5144 process using sd-bus automatically exit if the bus connection is
5147 * Bus clients of the service manager may now "pin" loaded units into
5148 memory, by taking an explicit reference on them. This is useful to
5149 ensure the client can retrieve runtime data about the service even
5150 after the service completed execution. Taking such a reference is
5151 available only for privileged clients and should be helpful to watch
5152 running services in a race-free manner, and in particular collect
5153 information about exit statuses and results.
5155 * The nss-resolve module has been changed to strictly return UNAVAIL
5156 when communication via D-Bus with resolved failed, and NOTFOUND when
5157 a lookup completed but was negative. This means it is now possible to
5158 neatly configure fallbacks using nsswitch.conf result checking
5159 expressions. Taking benefit of this, the new recommended
5160 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
5162 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
5164 * A new setting CtrlAltDelBurstAction= has been added to
5165 /etc/systemd/system.conf which may be used to configure the precise
5166 behaviour if the user on the console presses Ctrl-Alt-Del more often
5167 than 7 times in 2s. Previously this would unconditionally result in
5168 an expedited, immediate reboot. With this new setting the precise
5169 operation may be configured in more detail, and also turned off
5172 * In .netdev files two new settings RemoteChecksumTx= and
5173 RemoteChecksumRx= are now understood that permit configuring the
5174 remote checksumming logic for VXLAN networks.
5176 * The service manager learnt a new "invocation ID" concept for invoked
5177 services. Each runtime cycle of a service will get a new invocation
5178 ID (a 128bit random UUID) assigned that identifies the current
5179 run of the service uniquely and globally. A new invocation ID
5180 is generated each time a service starts up. The journal will store
5181 the invocation ID of a service along with any logged messages, thus
5182 making the invocation ID useful for matching the online runtime of a
5183 service with the offline log data it generated in a safe way without
5184 relying on synchronized timestamps. In many ways this new service
5185 invocation ID concept is similar to the kernel's boot ID concept that
5186 uniquely and globally identifies the runtime of each boot. The
5187 invocation ID of a service is passed to the service itself via an
5188 environment variable ($INVOCATION_ID). A new bus call
5189 GetUnitByInvocationID() has been added that is similar to GetUnit()
5190 but instead of retrieving the bus path for a unit by its name
5191 retrieves it by its invocation ID. The returned path is valid only as
5192 long as the passed invocation ID is current.
5194 * systemd-resolved gained a new "DNSStubListener" setting in
5195 resolved.conf. It either takes a boolean value or the special values
5196 "udp" and "tcp", and configures whether to enable the stub DNS
5197 listener on 127.0.0.53:53.
5199 * IP addresses configured via networkd may now carry additional
5200 configuration settings supported by the kernel. New options include:
5201 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
5202 PrefixRoute=, AutoJoin=.
5204 * The PAM configuration fragment file for "user@.service" shipped with
5205 systemd (i.e. the --user instance of systemd) has been stripped to
5206 the minimum necessary to make the system boot. Previously, it
5207 contained Fedora-specific stanzas that did not apply to other
5208 distributions. It is expected that downstream distributions add
5209 additional configuration lines, matching their needs to this file,
5210 using it only as rough template of what systemd itself needs. Note
5211 that this reduced fragment does not even include an invocation of
5212 pam_limits which most distributions probably want to add, even though
5213 systemd itself does not need it. (There's also the new build time
5214 option --with-pamconfdir=no to disable installation of the PAM
5217 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
5218 capability is now also dropped from its set (in addition to
5219 CAP_SYS_MKNOD as before).
5221 * In service unit files it is now possible to connect a specific named
5222 file descriptor with stdin/stdout/stdout of an executed service. The
5223 name may be specified in matching .socket units using the
5224 FileDescriptorName= setting.
5226 * A number of journal settings may now be configured on the kernel
5227 command line. Specifically, the following options are now understood:
5228 systemd.journald.max_level_console=,
5229 systemd.journald.max_level_store=,
5230 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
5231 systemd.journald.max_level_wall=.
5233 * "systemctl is-enabled --full" will now show by which symlinks a unit
5234 file is enabled in the unit dependency tree.
5236 * Support for VeraCrypt encrypted partitions has been added to the
5237 "cryptsetup" logic and /etc/crypttab.
5239 * systemd-detect-virt gained support for a new --private-users switch
5240 that checks whether the invoking processes are running inside a user
5241 namespace. Similar, a new special value "private-users" for the
5242 existing ConditionVirtualization= setting has been added, permitting
5243 skipping of specific units in user namespace environments.
5245 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
5246 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
5247 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
5248 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
5249 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
5250 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
5251 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
5252 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
5253 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
5254 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
5255 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
5256 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
5257 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
5258 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
5259 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
5260 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
5261 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
5262 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
5263 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
5264 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
5265 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
5266 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
5267 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
5268 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
5269 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
5270 Jędrzejewski-Szmek, Zeal Jagannatha
5272 — Santa Fe, 2016-11-03
5276 * In service units the various ExecXYZ= settings have been extended
5277 with an additional special character as first argument of the
5278 assigned value: if the character '+' is used the specified command
5279 line it will be run with full privileges, regardless of User=,
5280 Group=, CapabilityBoundingSet= and similar options. The effect is
5281 similar to the existing PermissionsStartOnly= option, but allows
5282 configuration of this concept for each executed command line
5285 * Services may now alter the service watchdog timeout at runtime by
5286 sending a WATCHDOG_USEC= message via sd_notify().
5288 * MemoryLimit= and related unit settings now optionally take percentage
5289 specifications. The percentage is taken relative to the amount of
5290 physical memory in the system (or in case of containers, the assigned
5291 amount of memory). This allows scaling service resources neatly with
5292 the amount of RAM available on the system. Similarly, systemd-logind's
5293 RuntimeDirectorySize= option now also optionally takes percentage
5296 * In similar fashion TasksMax= takes percentage values now, too. The
5297 value is taken relative to the configured maximum number of processes
5298 on the system. The per-service task maximum has been changed to 15%
5299 using this functionality. (Effectively this is an increase of 512 →
5300 4915 for service units, given the kernel's default pid_max setting.)
5302 * Calendar time specifications in .timer units now understand a ".."
5303 syntax for time ranges. Example: "4..7:10" may now be used for
5304 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
5307 * The InaccessableDirectories=, ReadOnlyDirectories= and
5308 ReadWriteDirectories= unit file settings have been renamed to
5309 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
5310 applied to all kinds of file nodes, and not just directories, with
5311 the exception of symlinks. Specifically these settings may now be
5312 used on block and character device nodes, UNIX sockets and FIFOS as
5313 well as regular files. The old names of these settings remain
5314 available for compatibility.
5316 * systemd will now log about all service processes it kills forcibly
5317 (using SIGKILL) because they remained after the clean shutdown phase
5318 of the service completed. This should help identifying services that
5319 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
5320 systemd-logind's configuration a similar log message is generated for
5321 processes killed at the end of each session due to this setting.
5323 * systemd will now set the $JOURNAL_STREAM environment variable for all
5324 services whose stdout/stderr are connected to the Journal (which
5325 effectively means by default: all services). The variable contains
5326 the device and inode number of the file descriptor used for
5327 stdout/stderr. This may be used by invoked programs to detect whether
5328 their stdout/stderr is connected to the Journal, in which case they
5329 can switch over to direct Journal communication, thus being able to
5330 pass extended, structured metadata along with their log messages. As
5331 one example, this is now used by glib's logging primitives.
5333 * When using systemd's default tmp.mount unit for /tmp, the mount point
5334 will now be established with the "nosuid" and "nodev" options. This
5335 avoids privilege escalation attacks that put traps and exploits into
5336 /tmp. However, this might cause problems if you e. g. put container
5337 images or overlays into /tmp; if you need this, override tmp.mount's
5338 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
5341 * systemd now supports the "memory" cgroup controller also on
5344 * The systemd-cgtop tool now optionally takes a control group path as
5345 command line argument. If specified, the control group list shown is
5346 limited to subgroups of that group.
5348 * The SystemCallFilter= unit file setting gained support for
5349 pre-defined, named system call filter sets. For example
5350 SystemCallFilter=@clock is now an effective way to make all clock
5351 changing-related system calls unavailable to a service. A number of
5352 similar pre-defined groups are defined. Writing system call filters
5353 for system services is simplified substantially with this new
5354 concept. Accordingly, all of systemd's own, long-running services now
5355 enable system call filtering based on this, by default.
5357 * A new service setting MemoryDenyWriteExecute= has been added, taking
5358 a boolean value. If turned on, a service may no longer create memory
5359 mappings that are writable and executable at the same time. This
5360 enhances security for services where this is enabled as it becomes
5361 harder to dynamically write and then execute memory in exploited
5362 service processes. This option has been enabled for all of systemd's
5363 own long-running services.
5365 * A new RestrictRealtime= service setting has been added, taking a
5366 boolean argument. If set the service's processes may no longer
5367 acquire realtime scheduling. This improves security as realtime
5368 scheduling may otherwise be used to easily freeze the system.
5370 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
5371 value. This may be used for requesting that the system manager inside
5372 of the container reports start-up completion to nspawn which then
5373 propagates this notification further to the service manager
5374 supervising nspawn itself. A related option NotifyReady= in .nspawn
5375 files has been added too. This functionality allows ordering of the
5376 start-up of multiple containers using the usual systemd ordering
5379 * machinectl gained a new command "stop" that is an alias for
5382 * systemd-resolved gained support for contacting DNS servers on
5383 link-local IPv6 addresses.
5385 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
5386 its caches. A method call for requesting the same operation has been
5387 added to the bus API too, and is made available via "systemd-resolve
5390 * systemd-resolve gained a new --status switch. If passed a brief
5391 summary of the used DNS configuration with per-interface information
5394 * resolved.conf gained a new Cache= boolean option, defaulting to
5395 on. If turned off local DNS caching is disabled. This comes with a
5396 performance penalty in particular when DNSSEC is enabled. Note that
5397 resolved disables its internal caching implicitly anyway, when the
5398 configured DNS server is on a host-local IP address such as ::1 or
5399 127.0.0.1, thus automatically avoiding double local caching.
5401 * systemd-resolved now listens on the local IP address 127.0.0.53:53
5402 for DNS requests. This improves compatibility with local programs
5403 that do not use the libc NSS or systemd-resolved's bus APIs for name
5404 resolution. This minimal DNS service is only available to local
5405 programs and does not implement the full DNS protocol, but enough to
5406 cover local DNS clients. A new, static resolv.conf file, listing just
5407 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
5408 now recommended to make /etc/resolv.conf a symlink to this file in
5409 order to route all DNS lookups to systemd-resolved, regardless if
5410 done via NSS, the bus API or raw DNS packets. Note that this local
5411 DNS service is not as fully featured as the libc NSS or
5412 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
5413 used to deliver link-local address information (as this implies
5414 sending a local interface index along), LLMNR/mDNS support via this
5415 interface is severely restricted. It is thus strongly recommended for
5416 all applications to use the libc NSS API or native systemd-resolved
5419 * systemd-networkd's bridge support learned a new setting
5420 VLANFiltering= for controlling VLAN filtering. Moreover a new section
5421 in .network files has been added for configuring VLAN bridging in
5422 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
5424 * systemd-networkd's IPv6 Router Advertisement code now makes use of
5425 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
5426 now be acquired without relying on DHCPv6. Two new options
5427 UseDomains= and UseDNS= have been added to configure this behaviour.
5429 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
5430 renamed IPv6AcceptRA=, without altering its behaviour. The old
5431 setting name remains available for compatibility reasons.
5433 * The systemd-networkd VTI/VTI6 tunneling support gained new options
5434 Key=, InputKey= and OutputKey=.
5436 * systemd-networkd gained support for VRF ("Virtual Routing Function")
5437 interface configuration.
5439 * "systemctl edit" may now be used to create new unit files by
5440 specifying the --force switch.
5442 * sd-event gained a new function sd_event_get_iteration() for
5443 requesting the current iteration counter of the event loop. It starts
5444 at zero and is increased by one with each event loop iteration.
5446 * A new rpm macro %systemd_ordering is provided by the macros.systemd
5447 file. It can be used in lieu of %systemd_requires in packages which
5448 don't use any systemd functionality and are intended to be installed
5449 in minimal containers without systemd present. This macro provides
5450 ordering dependencies to ensure that if the package is installed in
5451 the same rpm transaction as systemd, systemd will be installed before
5452 the scriptlets for the package are executed, allowing unit presets
5455 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
5456 been added to simplify packaging of generators.
5458 * The os-release file gained VERSION_CODENAME field for the
5459 distribution nickname (e.g. VERSION_CODENAME=woody).
5461 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
5462 can be set to disable parsing of metadata and the creation
5463 of persistent symlinks for that device.
5465 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
5466 to make them available to logged-in users has been reverted.
5468 * Much of the common code of the various systemd components is now
5469 built into an internal shared library libsystemd-shared-231.so
5470 (incorporating the systemd version number in the name, to be updated
5471 with future releases) that the components link to. This should
5472 decrease systemd footprint both in memory during runtime and on
5473 disk. Note that the shared library is not for public use, and is
5474 neither API nor ABI stable, but is likely to change with every new
5475 released update. Packagers need to make sure that binaries
5476 linking to libsystemd-shared.so are updated in step with the
5479 * Configuration for "mkosi" is now part of the systemd
5480 repository. mkosi is a tool to easily build legacy-free OS images,
5481 and is available on github: https://github.com/systemd/mkosi. If
5482 "mkosi" is invoked in the build tree a new raw OS image is generated
5483 incorporating the systemd sources currently being worked on and a
5484 clean, fresh distribution installation. The generated OS image may be
5485 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
5486 UEFI PC. This functionality is particularly useful to easily test
5487 local changes made to systemd in a pristine, defined environment. See
5488 doc/HACKING for details.
5490 * configure learned the --with-support-url= option to specify the
5491 distribution's bugtracker.
5493 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
5494 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
5495 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
5496 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
5497 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
5498 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
5499 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
5500 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
5501 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
5502 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
5503 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
5504 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
5505 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
5506 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
5507 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
5508 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
5509 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
5510 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
5511 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
5513 — Berlin, 2016-07-25
5517 * DNSSEC is now turned on by default in systemd-resolved (in
5518 "allow-downgrade" mode), but may be turned off during compile time by
5519 passing "--with-default-dnssec=no" to "configure" (and of course,
5520 during runtime with DNSSEC= in resolved.conf). We recommend
5521 downstreams to leave this on at least during development cycles and
5522 report any issues with the DNSSEC logic upstream. We are very
5523 interested in collecting feedback about the DNSSEC validator and its
5524 limitations in the wild. Note however, that DNSSEC support is
5525 probably nothing downstreams should turn on in stable distros just
5526 yet, as it might create incompatibilities with a few DNS servers and
5527 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
5528 automatically whenever we detect such incompatible setups, but there
5529 might be systems we do not cover yet. Hence: please help us testing
5530 the DNSSEC code, leave this on where you can, report back, but then
5531 again don't consider turning this on in your stable, LTS or
5532 production release just yet. (Note that you have to enable
5533 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
5534 and its DNSSEC mode for hostname resolution from local
5537 * systemd-resolve conveniently resolves DANE records with the --tlsa
5538 option and OPENPGPKEY records with the --openpgp option. It also
5539 supports dumping raw DNS record data via the new --raw= switch.
5541 * systemd-logind will now by default terminate user processes that are
5542 part of the user session scope unit (session-XX.scope) when the user
5543 logs out. This behavior is controlled by the KillUserProcesses=
5544 setting in logind.conf, and the previous default of "no" is now
5545 changed to "yes". This means that user sessions will be properly
5546 cleaned up after, but additional steps are necessary to allow
5547 intentionally long-running processes to survive logout.
5549 While the user is logged in at least once, user@.service is running,
5550 and any service that should survive the end of any individual login
5551 session can be started at a user service or scope using systemd-run.
5552 systemd-run(1) man page has been extended with an example which shows
5553 how to run screen in a scope unit underneath user@.service. The same
5554 command works for tmux.
5556 After the user logs out of all sessions, user@.service will be
5557 terminated too, by default, unless the user has "lingering" enabled.
5558 To effectively allow users to run long-term tasks even if they are
5559 logged out, lingering must be enabled for them. See loginctl(1) for
5560 details. The default polkit policy was modified to allow users to
5561 set lingering for themselves without authentication.
5563 Previous defaults can be restored at compile time by the
5564 --without-kill-user-processes option to "configure".
5566 * systemd-logind gained new configuration settings SessionsMax= and
5567 InhibitorsMax=, both with a default of 8192. It will not register new
5568 user sessions or inhibitors above this limit.
5570 * systemd-logind will now reload configuration on SIGHUP.
5572 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
5573 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
5574 enable. Also, support for the "io" cgroup controller in the unified
5575 hierarchy has been added, so that the "memory", "pids" and "io" are
5576 now the controllers that are supported on the unified hierarchy.
5578 WARNING: it is not possible to use previous systemd versions with
5579 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
5580 is necessary to also update systemd in the initramfs if using the
5581 unified hierarchy. An updated SELinux policy is also required.
5583 * LLDP support has been extended, and both passive (receive-only) and
5584 active (sender) modes are supported. Passive mode ("routers-only") is
5585 enabled by default in systemd-networkd. Active LLDP mode is enabled
5586 by default for containers on the internal network. The "networkctl
5587 lldp" command may be used to list information gathered. "networkctl
5588 status" will also show basic LLDP information on connected peers now.
5590 * The IAID and DUID unique identifier sent in DHCP requests may now be
5591 configured for the system and each .network file managed by
5592 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
5594 * systemd-networkd gained support for configuring proxy ARP support for
5595 each interface, via the ProxyArp= setting in .network files. It also
5596 gained support for configuring the multicast querier feature of
5597 bridge devices, via the new MulticastQuerier= setting in .netdev
5598 files. Similarly, snooping on the IGMP traffic can be controlled
5599 via the new setting MulticastSnooping=.
5601 A new setting PreferredLifetime= has been added for addresses
5602 configured in .network file to configure the lifetime intended for an
5605 The systemd-networkd DHCP server gained the option EmitRouter=, which
5606 defaults to yes, to configure whether the DHCP Option 3 (Router)
5609 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
5610 systemd-socket-activate and installed into /usr/bin. It is now fully
5613 * systemd-journald now uses separate threads to flush changes to disk
5614 when closing journal files, thus reducing impact of slow disk I/O on
5615 logging performance.
5617 * The sd-journal API gained two new calls
5618 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
5619 can be used to open journal files using file descriptors instead of
5620 file or directory paths. sd_journal_open_container() has been
5621 deprecated, sd_journal_open_directory_fd() should be used instead
5622 with the flag SD_JOURNAL_OS_ROOT.
5624 * journalctl learned a new output mode "-o short-unix" that outputs log
5625 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
5626 UTC). It also gained support for a new --no-hostname setting to
5627 suppress the hostname column in the family of "short" output modes.
5629 * systemd-ask-password now optionally skips printing of the password to
5630 stdout with --no-output which can be useful in scripts.
5632 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
5633 (devices tagged with ID_MAKER_TOOL) are now tagged with
5634 "uaccess" and are available to logged in users.
5636 * The DeviceAllow= unit setting now supports specifiers (with "%").
5638 * "systemctl show" gained a new --value switch, which allows print a
5639 only the contents of a specific unit property, without also printing
5640 the property's name. Similar support was added to "show*" verbs
5641 of loginctl and machinectl that output "key=value" lists.
5643 * A new unit type "generated" was added for files dynamically generated
5644 by generator tools. Similarly, a new unit type "transient" is used
5645 for unit files created using the runtime API. "systemctl enable" will
5646 refuse to operate on such files.
5648 * A new command "systemctl revert" has been added that may be used to
5649 revert to the vendor version of a unit file, in case local changes
5650 have been made by adding drop-ins or overriding the unit file.
5652 * "machinectl clean" gained a new verb to automatically remove all or
5653 just hidden container images.
5655 * systemd-tmpfiles gained support for a new line type "e" for emptying
5656 directories, if they exist, without creating them if they don't.
5658 * systemd-nspawn gained support for automatically patching the UID/GIDs
5659 of the owners and the ACLs of all files and directories in a
5660 container tree to match the UID/GID user namespacing range selected
5661 for the container invocation. This mode is enabled via the new
5662 --private-users-chown switch. It also gained support for
5663 automatically choosing a free, previously unused UID/GID range when
5664 starting a container, via the new --private-users=pick setting (which
5665 implies --private-users-chown). Together, these options for the first
5666 time make user namespacing for nspawn containers fully automatic and
5667 thus deployable. The systemd-nspawn@.service template unit file has
5668 been changed to use this functionality by default.
5670 * systemd-nspawn gained a new --network-zone= switch, that allows
5671 creating ad-hoc virtual Ethernet links between multiple containers,
5672 that only exist as long as at least one container referencing them is
5673 running. This allows easy connecting of multiple containers with a
5674 common link that implements an Ethernet broadcast domain. Each of
5675 these network "zones" may be named relatively freely by the user, and
5676 may be referenced by any number of containers, but each container may
5677 only reference one of these "zones". On the lower level, this is
5678 implemented by an automatically managed bridge network interface for
5679 each zone, that is created when the first container referencing its
5680 zone is created and removed when the last one referencing its zone
5683 * The default start timeout may now be configured on the kernel command
5684 line via systemd.default_timeout_start_sec=. It was already
5685 configurable via the DefaultTimeoutStartSec= option in
5686 /etc/systemd/system.conf.
5688 * Socket units gained a new TriggerLimitIntervalSec= and
5689 TriggerLimitBurst= setting to configure a limit on the activation
5690 rate of the socket unit.
5692 * The LimitNICE= setting now optionally takes normal UNIX nice values
5693 in addition to the raw integer limit value. If the specified
5694 parameter is prefixed with "+" or "-" and is in the range -20..19 the
5695 value is understood as UNIX nice value. If not prefixed like this it
5696 is understood as raw RLIMIT_NICE limit.
5698 * Note that the effect of the PrivateDevices= unit file setting changed
5699 slightly with this release: the per-device /dev file system will be
5700 mounted read-only from this version on, and will have "noexec"
5701 set. This (minor) change of behavior might cause some (exceptional)
5702 legacy software to break, when PrivateDevices=yes is set for its
5703 service. Please leave PrivateDevices= off if you run into problems
5706 * systemd-bootchart has been split out to a separate repository:
5707 https://github.com/systemd/systemd-bootchart
5709 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
5710 merged into the kernel in its current form.
5712 * The compatibility libraries libsystemd-daemon.so,
5713 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
5714 which have been deprecated since systemd-209 have been removed along
5715 with the corresponding pkg-config files. All symbols provided by
5716 those libraries are provided by libsystemd.so.
5718 * The Capabilities= unit file setting has been removed (it is ignored
5719 for backwards compatibility). AmbientCapabilities= and
5720 CapabilityBoundingSet= should be used instead.
5722 * A new special target has been added, initrd-root-device.target,
5723 which creates a synchronization point for dependencies of the root
5724 device in early userspace. Initramfs builders must ensure that this
5725 target is now included in early userspace.
5727 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
5728 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
5729 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
5730 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
5731 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
5732 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
5733 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
5734 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
5735 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
5736 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
5737 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
5738 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
5739 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
5740 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
5741 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
5742 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
5743 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
5744 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
5745 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
5746 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
5747 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
5748 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
5749 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
5750 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
5751 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
5754 — Fairfax, 2016-05-21
5758 * The systemd-resolved DNS resolver service has gained a substantial
5759 set of new features, most prominently it may now act as a DNSSEC
5760 validating stub resolver. DNSSEC mode is currently turned off by
5761 default, but is expected to be turned on by default in one of the
5762 next releases. For now, we invite everybody to test the DNSSEC logic
5763 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
5764 service also gained a full set of D-Bus interfaces, including calls
5765 to configure DNS and DNSSEC settings per link (for use by external
5766 network management software). systemd-resolved and systemd-networkd
5767 now distinguish between "search" and "routing" domains. The former
5768 are used to qualify single-label names, the latter are used purely
5769 for routing lookups within certain domains to specific links.
5770 resolved now also synthesizes RRs for all entries from /etc/hosts.
5772 * The systemd-resolve tool (which is a client utility for
5773 systemd-resolved) has been improved considerably and is now fully
5774 supported and documented. Hence it has moved from /usr/lib/systemd to
5777 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
5780 * The coredump collection logic has been reworked: when a coredump is
5781 collected it is now written to disk, compressed and processed
5782 (including stacktrace extraction) from a new instantiated service
5783 systemd-coredump@.service, instead of directly from the
5784 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
5785 processing large coredumps can take up a substantial amount of
5786 resources and time, and this previously happened entirely outside of
5787 systemd's service supervision. With the new logic the core_pattern
5788 hook only does minimal metadata collection before passing off control
5789 to the new instantiated service, which is configured with a time
5790 limit, a nice level and other settings to minimize negative impact on
5791 the rest of the system. Also note that the new logic will honour the
5792 RLIMIT_CORE setting of the crashed process, which now allows users
5793 and processes to turn off coredumping for their processes by setting
5796 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
5797 and all forked processes by default. Previously, PID 1 would leave
5798 the setting at "0" for all processes, as set by the kernel. Note that
5799 the resource limit traditionally has no effect on the generated
5800 coredumps on the system if the /proc/sys/kernel/core_pattern hook
5801 logic is used. Since the limit is now honoured (see above) its
5802 default has been changed so that the coredumping logic is enabled by
5803 default for all processes, while allowing specific opt-out.
5805 * When the stacktrace is extracted from processes of system users, this
5806 is now done as "systemd-coredump" user, in order to sandbox this
5807 potentially security sensitive parsing operation. (Note that when
5808 processing coredumps of normal users this is done under the user ID
5809 of process that crashed, as before.) Packagers should take notice
5810 that it is now necessary to create the "systemd-coredump" system user
5811 and group at package installation time.
5813 * The systemd-activate socket activation testing tool gained support
5814 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
5815 and --seqpacket switches. It also has been extended to support both
5816 new-style and inetd-style file descriptor passing. Use the new
5817 --inetd switch to request inetd-style file descriptor passing.
5819 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
5820 variable, which takes a boolean value. If set to false, ANSI color
5821 output is disabled in the tools even when run on a terminal that
5824 * The VXLAN support in networkd now supports two new settings
5825 DestinationPort= and PortRange=.
5827 * A new systemd.machine_id= kernel command line switch has been added,
5828 that may be used to set the machine ID in /etc/machine-id if it is
5829 not initialized yet. This command line option has no effect if the
5830 file is already initialized.
5832 * systemd-nspawn gained a new --as-pid2 switch that invokes any
5833 specified command line as PID 2 rather than PID 1 in the
5834 container. In this mode PID 1 is a minimal stub init process that
5835 implements the special POSIX and Linux semantics of PID 1 regarding
5836 signal and child process management. Note that this stub init process
5837 is implemented in nspawn itself and requires no support from the
5838 container image. This new logic is useful to support running
5839 arbitrary commands in the container, as normal processes are
5840 generally not prepared to run as PID 1.
5842 * systemd-nspawn gained a new --chdir= switch for setting the current
5843 working directory for the process started in the container.
5845 * "journalctl /dev/sda" will now output all kernel log messages for
5846 specified device from the current boot, in addition to all devices
5847 that are parents of it. This should make log output about devices
5848 pretty useful, as long as kernel drivers attach enough metadata to
5849 the log messages. (The usual SATA drivers do.)
5851 * The sd-journal API gained two new calls
5852 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
5853 that report whether log data from /run or /var has been found.
5855 * journalctl gained a new switch "--fields" that prints all journal
5856 record field names currently in use in the journal. This is backed
5857 by two new sd-journal API calls sd_journal_enumerate_fields() and
5858 sd_journal_restart_fields().
5860 * Most configurable timeouts in systemd now expect an argument of
5861 "infinity" to turn them off, instead of "0" as before. The semantics
5862 from now on is that a timeout of "0" means "now", and "infinity"
5863 means "never". To maintain backwards compatibility, "0" continues to
5864 turn off previously existing timeout settings.
5866 * "systemctl reload-or-try-restart" has been renamed to "systemctl
5867 try-reload-or-restart" to clarify what it actually does: the "try"
5868 logic applies to both reloading and restarting, not just restarting.
5869 The old name continues to be accepted for compatibility.
5871 * On boot-up, when PID 1 detects that the system clock is behind the
5872 release date of the systemd version in use, the clock is now set
5873 to the latter. Previously, this was already done in timesyncd, in order
5874 to avoid running with clocks set to the various clock epochs such as
5875 1902, 1938 or 1970. With this change the logic is now done in PID 1
5876 in addition to timesyncd during early boot-up, so that it is enforced
5877 before the first process is spawned by systemd. Note that the logic
5878 in timesyncd remains, as it is more comprehensive and ensures
5879 clock monotonicity by maintaining a persistent timestamp file in
5880 /var. Since /var is generally not available in earliest boot or the
5881 initrd, this part of the logic remains in timesyncd, and is not done
5884 * Support for tweaking details in net_cls.class_id through the
5885 NetClass= configuration directive has been removed, as the kernel
5886 people have decided to deprecate that controller in cgroup v2.
5887 Userspace tools such as nftables are moving over to setting rules
5888 that are specific to the full cgroup path of a task, which obsoletes
5889 these controllers anyway. The NetClass= directive is kept around for
5890 legacy compatibility reasons. For a more in-depth description of the
5891 kernel change, please refer to the respective upstream commit:
5893 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
5895 * A new service setting RuntimeMaxSec= has been added that may be used
5896 to specify a maximum runtime for a service. If the timeout is hit, the
5897 service is terminated and put into a failure state.
5899 * A new service setting AmbientCapabilities= has been added. It allows
5900 configuration of additional Linux process capabilities that are
5901 passed to the activated processes. This is only available on very
5904 * The process resource limit settings in service units may now be used
5905 to configure hard and soft limits individually.
5907 * The various libsystemd APIs such as sd-bus or sd-event now publicly
5908 expose support for gcc's __attribute__((cleanup())) C extension.
5909 Specifically, for many object destructor functions alternative
5910 versions have been added that have names suffixed with "p" and take a
5911 pointer to a pointer to the object to destroy, instead of just a
5912 pointer to the object itself. This is useful because these destructor
5913 functions may be used directly as parameters to the cleanup
5914 construct. Internally, systemd has been a heavy user of this GCC
5915 extension for a long time, and with this change similar support is
5916 now available to consumers of the library outside of systemd. Note
5917 that by using this extension in your sources compatibility with old
5918 and strictly ANSI compatible C compilers is lost. However, all gcc or
5919 LLVM versions of recent years support this extension.
5921 * Timer units gained support for a new setting RandomizedDelaySec= that
5922 allows configuring some additional randomized delay to the configured
5923 time. This is useful to spread out timer events to avoid load peaks in
5924 clusters or larger setups.
5926 * Calendar time specifications now support sub-second accuracy.
5928 * Socket units now support listening on SCTP and UDP-lite protocol
5931 * The sd-event API now comes with a full set of man pages.
5933 * Older versions of systemd contained experimental support for
5934 compressing journal files and coredumps with the LZ4 compressor that
5935 was not compatible with the lz4 binary (due to API limitations of the
5936 lz4 library). This support has been removed; only support for files
5937 compatible with the lz4 binary remains. This LZ4 logic is now
5938 officially supported and no longer considered experimental.
5940 * The dkr image import logic has been removed again from importd. dkr's
5941 micro-services focus doesn't fit into the machine image focus of
5942 importd, and quickly got out of date with the upstream dkr API.
5944 * Creation of the /run/lock/lockdev/ directory was dropped from
5945 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
5946 been available for many years. If you still need this, you need to
5947 create your own tmpfiles.d config file with:
5949 d /run/lock/lockdev 0775 root lock -
5951 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
5952 and RebootArgument= have been moved from the [Service] section of
5953 unit files to [Unit], and they are now supported on all unit types,
5954 not just service units. Of course, systemd will continue to
5955 understand these settings also at the old location, in order to
5956 maintain compatibility.
5958 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
5959 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
5960 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
5961 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
5962 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
5963 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
5964 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
5965 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
5966 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
5967 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
5968 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
5969 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
5970 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
5971 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
5972 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
5973 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
5974 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
5975 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
5976 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
5978 — Berlin, 2016-02-11
5982 * A number of properties previously only settable in unit
5983 files are now also available as properties to set when
5984 creating transient units programmatically via the bus, as it
5985 is exposed with systemd-run's --property=
5986 setting. Specifically, these are: SyslogIdentifier=,
5987 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
5988 EnvironmentFile=, ReadWriteDirectories=,
5989 ReadOnlyDirectories=, InaccessibleDirectories=,
5990 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
5992 * When creating transient services via the bus API it is now
5993 possible to pass in a set of file descriptors to use as
5994 STDIN/STDOUT/STDERR for the invoked process.
5996 * Slice units may now be created transiently via the bus APIs,
5997 similar to the way service and scope units may already be
5998 created transiently.
6000 * Wherever systemd expects a calendar timestamp specification
6001 (like in journalctl's --since= and --until= switches) UTC
6002 timestamps are now supported. Timestamps suffixed with "UTC"
6003 are now considered to be in Universal Time Coordinated
6004 instead of the local timezone. Also, timestamps may now
6005 optionally be specified with sub-second accuracy. Both of
6006 these additions also apply to recurring calendar event
6007 specification, such as OnCalendar= in timer units.
6009 * journalctl gained a new "--sync" switch that asks the
6010 journal daemon to write all so far unwritten log messages to
6011 disk and sync the files, before returning.
6013 * systemd-tmpfiles learned two new line types "q" and "Q" that
6014 operate like "v", but also set up a basic btrfs quota
6015 hierarchy when used on a btrfs file system with quota
6018 * tmpfiles' "v", "q" and "Q" will now create a plain directory
6019 instead of a subvolume (even on a btrfs file system) if the
6020 root directory is a plain directory, and not a
6021 subvolume. This should simplify things with certain chroot()
6022 environments which are not aware of the concept of btrfs
6025 * systemd-detect-virt gained a new --chroot switch to detect
6026 whether execution takes place in a chroot() environment.
6028 * CPUAffinity= now takes CPU index ranges in addition to
6031 * The various memory-related resource limit settings (such as
6032 LimitAS=) now understand the usual K, M, G, ... suffixes to
6033 the base of 1024 (IEC). Similar, the time-related resource
6034 limit settings understand the usual min, h, day, ...
6037 * There's a new system.conf setting DefaultTasksMax= to
6038 control the default TasksMax= setting for services and
6039 scopes running on the system. (TasksMax= is the primary
6040 setting that exposes the "pids" cgroup controller on systemd
6041 and was introduced in the previous systemd release.) The
6042 setting now defaults to 512, which means services that are
6043 not explicitly configured otherwise will only be able to
6044 create 512 processes or threads at maximum, from this
6045 version on. Note that this means that thread- or
6046 process-heavy services might need to be reconfigured to set
6047 TasksMax= to a higher value. It is sufficient to set
6048 TasksMax= in these specific unit files to a higher value, or
6049 even "infinity". Similar, there's now a logind.conf setting
6050 UserTasksMax= that defaults to 4096 and limits the total
6051 number of processes or tasks each user may own
6052 concurrently. nspawn containers also have the TasksMax=
6053 value set by default now, to 8192. Note that all of this
6054 only has an effect if the "pids" cgroup controller is
6055 enabled in the kernel. The general benefit of these changes
6056 should be a more robust and safer system, that provides a
6057 certain amount of per-service fork() bomb protection.
6059 * systemd-nspawn gained the new --network-veth-extra= switch
6060 to define additional and arbitrarily-named virtual Ethernet
6061 links between the host and the container.
6063 * A new service execution setting PassEnvironment= has been
6064 added that allows importing select environment variables
6065 from PID1's environment block into the environment block of
6068 * Timer units gained support for a new RemainAfterElapse=
6069 setting which takes a boolean argument. It defaults to on,
6070 exposing behaviour unchanged to previous releases. If set to
6071 off, timer units are unloaded after they elapsed if they
6072 cannot elapse again. This is particularly useful for
6073 transient timer units, which shall not stay around longer
6074 than until they first elapse.
6076 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
6077 default now (the kernel default is 16). This is beneficial
6078 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
6079 allows substantially larger numbers of queued
6080 datagrams. This should increase the capability of systemd to
6081 parallelize boot-up, as logging and sd_notify() are unlikely
6082 to stall execution anymore. If you need to change the value
6083 from the new defaults, use the usual sysctl.d/ snippets.
6085 * The compression framing format used by the journal or
6086 coredump processing has changed to be in line with what the
6087 official LZ4 tools generate. LZ4 compression support in
6088 systemd was considered unsupported previously, as the format
6089 was not compatible with the normal tools. With this release
6090 this has changed now, and it is hence safe for downstream
6091 distributions to turn it on. While not compressing as well
6092 as the XZ, LZ4 is substantially faster, which makes
6093 it a good default choice for the compression logic in the
6094 journal and in coredump handling.
6096 * Any reference to /etc/mtab has been dropped from
6097 systemd. The file has been obsolete since a while, but
6098 systemd refused to work on systems where it was incorrectly
6099 set up (it should be a symlink or non-existent). Please make
6100 sure to update to util-linux 2.27.1 or newer in conjunction
6101 with this systemd release, which also drops any reference to
6102 /etc/mtab. If you maintain a distribution make sure that no
6103 software you package still references it, as this is a
6104 likely source of bugs. There's also a glibc bug pending,
6105 asking for removal of any reference to this obsolete file:
6107 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
6109 Note that only util-linux versions built with
6110 --enable-libmount-force-mountinfo are supported.
6112 * Support for the ".snapshot" unit type has been removed. This
6113 feature turned out to be little useful and little used, and
6114 has now been removed from the core and from systemctl.
6116 * The dependency types RequiresOverridable= and
6117 RequisiteOverridable= have been removed from systemd. They
6118 have been used only very sparingly to our knowledge and
6119 other options that provide a similar effect (such as
6120 systemctl --mode=ignore-dependencies) are much more useful
6121 and commonly used. Moreover, they were only half-way
6122 implemented as the option to control behaviour regarding
6123 these dependencies was never added to systemctl. By removing
6124 these dependency types the execution engine becomes a bit
6125 simpler. Unit files that use these dependencies should be
6126 changed to use the non-Overridable dependency types
6127 instead. In fact, when parsing unit files with these
6128 options, that's what systemd will automatically convert them
6129 too, but it will also warn, asking users to fix the unit
6130 files accordingly. Removal of these dependency types should
6131 only affect a negligible number of unit files in the wild.
6133 * Behaviour of networkd's IPForward= option changed
6134 (again). It will no longer maintain a per-interface setting,
6135 but propagate one way from interfaces where this is enabled
6136 to the global kernel setting. The global setting will be
6137 enabled when requested by a network that is set up, but
6138 never be disabled again. This change was made to make sure
6139 IPv4 and IPv6 behaviour regarding packet forwarding is
6140 similar (as the Linux IPv6 stack does not support
6141 per-interface control of this setting) and to minimize
6144 * In unit files the behaviour of %u, %U, %h, %s has
6145 changed. These specifiers will now unconditionally resolve
6146 to the various user database fields of the user that the
6147 systemd instance is running as, instead of the user
6148 configured in the specific unit via User=. Note that this
6149 effectively doesn't change much, as resolving of these
6150 specifiers was already turned off in the --system instance
6151 of systemd, as we cannot do NSS lookups from PID 1. In the
6152 --user instance of systemd these specifiers where correctly
6153 resolved, but hardly made any sense, since the user instance
6154 lacks privileges to do user switches anyway, and User= is
6155 hence useless. Moreover, even in the --user instance of
6156 systemd behaviour was awkward as it would only take settings
6157 from User= assignment placed before the specifier into
6158 account. In order to unify and simplify the logic around
6159 this the specifiers will now always resolve to the
6160 credentials of the user invoking the manager (which in case
6161 of PID 1 is the root user).
6163 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
6164 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
6165 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
6166 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
6167 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
6168 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
6169 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
6170 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
6171 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
6172 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
6173 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
6174 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
6175 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
6176 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
6179 — Berlin, 2015-11-18
6183 * systemd now depends on util-linux v2.27. More specifically,
6184 the newly added mount monitor feature in libmount now
6185 replaces systemd's former own implementation.
6187 * libmount mandates /etc/mtab not to be regular file, and
6188 systemd now enforces this condition at early boot.
6189 /etc/mtab has been deprecated and warned about for a very
6190 long time, so systems running systemd should already have
6191 stopped having this file around as anything else than a
6192 symlink to /proc/self/mounts.
6194 * Support for the "pids" cgroup controller has been added. It
6195 allows accounting the number of tasks in a cgroup and
6196 enforcing limits on it. This adds two new setting
6197 TasksAccounting= and TasksMax= to each unit, as well as a
6198 global option DefaultTasksAccounting=.
6200 * Support for the "net_cls" cgroup controller has been added.
6201 It allows assigning a net class ID to each task in the
6202 cgroup, which can then be used in firewall rules and traffic
6203 shaping configurations. Note that the kernel netfilter net
6204 class code does not currently work reliably for ingress
6205 packets on unestablished sockets.
6207 This adds a new config directive called NetClass= to CGroup
6208 enabled units. Allowed values are positive numbers for fixed
6209 assignments and "auto" for picking a free value
6212 * 'systemctl is-system-running' now returns 'offline' if the
6213 system is not booted with systemd. This command can now be
6214 used as a substitute for 'systemd-notify --booted'.
6216 * Watchdog timeouts have been increased to 3 minutes for all
6217 in-tree service files. Apparently, disk IO issues are more
6218 frequent than we hoped, and user reported >1 minute waiting
6221 * 'machine-id-commit' functionality has been merged into
6222 'machine-id-setup --commit'. The separate binary has been
6225 * The WorkingDirectory= directive in unit files may now be set
6226 to the special value '~'. In this case, the working
6227 directory is set to the home directory of the user
6228 configured in User=.
6230 * "machinectl shell" will now open the shell in the home
6231 directory of the selected user by default.
6233 * The CrashChVT= configuration file setting is renamed to
6234 CrashChangeVT=, following our usual logic of not
6235 abbreviating unnecessarily. The old directive is still
6236 supported for compat reasons. Also, this directive now takes
6237 an integer value between 1 and 63, or a boolean value. The
6238 formerly supported '-1' value for disabling stays around for
6241 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
6242 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
6243 RootDirectory= properties can now be set for transient
6246 * The systemd-analyze tool gained a new "set-log-target" verb
6247 to change the logging target the system manager logs to
6248 dynamically during runtime. This is similar to how
6249 "systemd-analyze set-log-level" already changes the log
6252 * In nspawn /sys is now mounted as tmpfs, with only a selected
6253 set of subdirectories mounted in from the real sysfs. This
6254 enhances security slightly, and is useful for ensuring user
6255 namespaces work correctly.
6257 * Support for USB FunctionFS activation has been added. This
6258 allows implementation of USB gadget services that are
6259 activated as soon as they are requested, so that they don't
6260 have to run continuously, similar to classic socket
6263 * The "systemctl exit" command now optionally takes an
6264 additional parameter that sets the exit code to return from
6265 the systemd manager when exiting. This is only relevant when
6266 running the systemd user instance, or when running the
6267 system instance in a container.
6269 * sd-bus gained the new API calls sd_bus_path_encode_many()
6270 and sd_bus_path_decode_many() that allow easy encoding and
6271 decoding of multiple identifier strings inside a D-Bus
6272 object path. Another new call sd_bus_default_flush_close()
6273 has been added to flush and close per-thread default
6276 * systemd-cgtop gained support for a -M/--machine= switch to
6277 show the control groups within a certain container only.
6279 * "systemctl kill" gained support for an optional --fail
6280 switch. If specified the requested operation will fail of no
6281 processes have been killed, because the unit had no
6282 processes attached, or similar.
6284 * A new systemd.crash_reboot=1 kernel command line option has
6285 been added that triggers a reboot after crashing. This can
6286 also be set through CrashReboot= in systemd.conf.
6288 * The RuntimeDirectory= setting now understands unit
6289 specifiers like %i or %f.
6291 * A new (still internal) library API sd-ipv4acd has been added,
6292 that implements address conflict detection for IPv4. It's
6293 based on code from sd-ipv4ll, and will be useful for
6294 detecting DHCP address conflicts.
6296 * File descriptors passed during socket activation may now be
6297 named. A new API sd_listen_fds_with_names() is added to
6298 access the names. The default names may be overridden,
6299 either in the .socket file using the FileDescriptorName=
6300 parameter, or by passing FDNAME= when storing the file
6301 descriptors using sd_notify().
6303 * systemd-networkd gained support for:
6305 - Setting the IPv6 Router Advertisement settings via
6306 IPv6AcceptRouterAdvertisements= in .network files.
6308 - Configuring the HelloTimeSec=, MaxAgeSec= and
6309 ForwardDelaySec= bridge parameters in .netdev files.
6311 - Configuring PreferredSource= for static routes in
6314 * The "ask-password" framework used to query for LUKS harddisk
6315 passwords or SSL passwords during boot gained support for
6316 caching passwords in the kernel keyring, if it is
6317 available. This makes sure that the user only has to type in
6318 a passphrase once if there are multiple objects to unlock
6319 with the same one. Previously, such password caching was
6320 available only when Plymouth was used; this moves the
6321 caching logic into the systemd codebase itself. The
6322 "systemd-ask-password" utility gained a new --keyname=
6323 switch to control which kernel keyring key to use for
6324 caching a password in. This functionality is also useful for
6325 enabling display managers such as gdm to automatically
6326 unlock the user's GNOME keyring if its passphrase, the
6327 user's password and the harddisk password are the same, if
6328 gdm-autologin is used.
6330 * When downloading tar or raw images using "machinectl
6331 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
6332 file is now also downloaded, if it is available and stored
6333 next to the image file.
6335 * Units of type ".socket" gained a new boolean setting
6336 Writable= which is only useful in conjunction with
6337 ListenSpecial=. If true, enables opening the specified
6338 special file in O_RDWR mode rather than O_RDONLY mode.
6340 * systemd-rfkill has been reworked to become a singleton
6341 service that is activated through /dev/rfkill on each rfkill
6342 state change and saves the settings to disk. This way,
6343 systemd-rfkill is now compatible with devices that exist
6344 only intermittendly, and even restores state if the previous
6345 system shutdown was abrupt rather than clean.
6347 * The journal daemon gained support for vacuuming old journal
6348 files controlled by the number of files that shall remain,
6349 in addition to the already existing control by size and by
6350 date. This is useful as journal interleaving performance
6351 degrades with too many separate journal files, and allows
6352 putting an effective limit on them. The new setting defaults
6353 to 100, but this may be changed by setting SystemMaxFiles=
6354 and RuntimeMaxFiles= in journald.conf. Also, the
6355 "journalctl" tool gained the new --vacuum-files= switch to
6356 manually vacuum journal files to leave only the specified
6357 number of files in place.
6359 * udev will now create /dev/disk/by-path links for ATA devices
6360 on kernels where that is supported.
6362 * Galician, Serbian, Turkish and Korean translations were added.
6364 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
6365 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
6366 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
6367 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
6368 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
6369 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
6370 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
6371 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
6372 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
6373 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
6374 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
6375 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
6376 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
6377 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
6378 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
6379 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
6380 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
6381 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
6383 — Berlin, 2015-10-07
6387 * The DHCP implementation of systemd-networkd gained a set of
6390 - The DHCP server now supports emitting DNS and NTP
6391 information. It may be enabled and configured via
6392 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
6393 and NTP information is enabled, but no servers are
6394 configured, the corresponding uplink information (if there
6395 is any) is propagated.
6397 - Server and client now support transmission and reception
6398 of timezone information. It can be configured via the
6399 newly introduced network options UseTimezone=,
6400 EmitTimezone=, and Timezone=. Transmission of timezone
6401 information is enabled between host and containers by
6402 default now: the container will change its local timezone
6403 to what the host has set.
6405 - Lease timeouts can now be configured via
6406 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
6408 - The DHCP server improved on the stability of
6409 leases. Clients are more likely to get the same lease
6410 information back, even if the server loses state.
6412 - The DHCP server supports two new configuration options to
6413 control the lease address pool metrics, PoolOffset= and
6416 * The encapsulation limit of tunnels in systemd-networkd may
6417 now be configured via 'EncapsulationLimit='. It allows
6418 modifying the maximum additional levels of encapsulation
6419 that are permitted to be prepended to a packet.
6421 * systemd now supports the concept of user buses replacing
6422 session buses, if used with dbus-1.10 (and enabled via dbus
6423 --enable-user-session). It previously only supported this on
6424 kdbus-enabled systems, and this release expands this to
6425 'dbus-daemon' systems.
6427 * systemd-networkd now supports predictable interface names
6430 * systemd now optionally supports the new Linux kernel
6431 "unified" control group hierarchy. If enabled via the kernel
6432 command-line option 'systemd.unified_cgroup_hierarchy=1',
6433 systemd will try to mount the unified cgroup hierarchy
6434 directly on /sys/fs/cgroup. If not enabled, or not
6435 available, systemd will fall back to the legacy cgroup
6436 hierarchy setup, as before. Host system and containers can
6437 mix and match legacy and unified hierarchies as they
6438 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
6439 environment variable to individually select the hierarchy to
6440 use for executed containers. By default, nspawn will use the
6441 unified hierarchy for the containers if the host uses the
6442 unified hierarchy, and the legacy hierarchy otherwise.
6443 Please note that at this point the unified hierarchy is an
6444 experimental kernel feature and is likely to change in one
6445 of the next kernel releases. Therefore, it should not be
6446 enabled by default in downstream distributions yet. The
6447 minimum required kernel version for the unified hierarchy to
6448 work is 4.2. Note that when the unified hierarchy is used
6449 for the first time delegated access to controllers is
6450 safe. Because of this systemd-nspawn containers will get
6451 access to controllers now, as will systemd user
6452 sessions. This means containers and user sessions may now
6453 manage their own resources, partitioning up what the system
6456 * A new special scope unit "init.scope" has been introduced
6457 that encapsulates PID 1 of the system. It may be used to
6458 determine resource usage and enforce resource limits on PID
6459 1 itself. PID 1 hence moved out of the root of the control
6462 * The cgtop tool gained support for filtering out kernel
6463 threads when counting tasks in a control group. Also, the
6464 count of processes is now recursively summed up by
6465 default. Two options -k and --recursive= have been added to
6466 revert to old behaviour. The tool has also been updated to
6467 work correctly in containers now.
6469 * systemd-nspawn's --bind= and --bind-ro= options have been
6470 extended to allow creation of non-recursive bind mounts.
6472 * libsystemd gained two new calls sd_pid_get_cgroup() and
6473 sd_peer_get_cgroup() which return the control group path of
6474 a process or peer of a connected AF_UNIX socket. This
6475 function call is particularly useful when implementing
6476 delegated subtrees support in the control group hierarchy.
6478 * The "sd-event" event loop API of libsystemd now supports
6479 correct dequeuing of real-time signals, without losing
6482 * When systemd requests a polkit decision when managing units it
6483 will now add additional fields to the request, including unit
6484 name and desired operation. This enables more powerful polkit
6485 policies, that make decisions depending on these parameters.
6487 * nspawn learnt support for .nspawn settings files, that may
6488 accompany the image files or directories of containers, and
6489 may contain additional settings for the container. This is
6490 an alternative to configuring container parameters via the
6491 nspawn command line.
6493 Contributions from: Cristian Rodríguez, Daniel Mack, David
6494 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
6495 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
6496 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
6497 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
6498 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
6499 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
6500 Andersen, Tom Gundersen, Torstein Husebø
6502 — Berlin, 2015-09-08
6506 * machinectl gained a new verb 'shell' which opens a fresh
6507 shell on the target container or the host. It is similar to
6508 the existing 'login' command of machinectl, but spawns the
6509 shell directly without prompting for username or
6510 password. The pseudo machine '.host' now refers to the local
6511 host and is used by default. Hence, 'machinectl shell' can
6512 be used as replacement for 'su -' which spawns a session as
6513 a fresh systemd unit in a way that is fully isolated from
6514 the originating session.
6516 * systemd-networkd learned to cope with private-zone DHCP
6517 options and allows other programs to query the values.
6519 * SELinux access control when enabling/disabling units is no
6520 longer enforced with this release. The previous implementation
6521 was incorrect, and a new corrected implementation is not yet
6522 available. As unit file operations are still protected via
6523 polkit and D-Bus policy this is not a security problem. Yet,
6524 distributions which care about optimal SELinux support should
6525 probably not stabilize on this release.
6527 * sd-bus gained support for matches of type "arg0has=", that
6528 test for membership of strings in string arrays sent in bus
6531 * systemd-resolved now dumps the contents of its DNS and LLMNR
6532 caches to the logs on reception of the SIGUSR1 signal. This
6533 is useful to debug DNS behaviour.
6535 * The coredumpctl tool gained a new --directory= option to
6536 operate on journal files in a specific directory.
6538 * "systemctl reboot" and related commands gained a new
6539 "--message=" option which may be used to set a free-text
6540 wall message when shutting down or rebooting the
6541 system. This message is also logged, which is useful for
6542 figuring out the reason for a reboot or shutdown a
6545 * The "systemd-resolve-host" tool's -i switch now takes
6546 network interface numbers as alternative to interface names.
6548 * A new unit file setting for services has been introduced:
6549 UtmpMode= allows configuration of how precisely systemd
6550 handles utmp and wtmp entries for the service if this is
6551 enabled. This allows writing services that appear similar to
6552 user sessions in the output of the "w", "who", "last" and
6555 * systemd-resolved will now locally synthesize DNS resource
6556 records for the "localhost" and "gateway" domains as well as
6557 the local hostname. This should ensure that clients querying
6558 RRs via resolved will get similar results as those going via
6559 NSS, if nss-myhostname is enabled.
6561 Contributions from: Alastair Hughes, Alex Crawford, Daniel
6562 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
6563 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
6564 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
6565 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
6566 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
6567 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
6568 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
6569 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
6570 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
6571 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
6572 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
6574 — Berlin, 2015-08-27
6578 * The systemd-efi-boot-generator functionality was merged into
6579 systemd-gpt-auto-generator.
6581 * systemd-networkd now supports Group Policy for vxlan
6582 devices. It can be enabled via the new boolean configuration
6583 option called 'GroupPolicyExtension='.
6585 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
6586 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
6587 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
6589 — Berlin, 2015-07-31
6593 * The python-systemd code has been removed from the systemd repository.
6594 A new repository has been created which accommodates the code from
6595 now on, and we kindly ask distributions to create a separate package
6596 for this: https://github.com/systemd/python-systemd
6598 * The systemd daemon will now reload its main configuration
6599 (/etc/systemd/system.conf) on daemon-reload.
6601 * sd-dhcp now exposes vendor specific extensions via
6602 sd_dhcp_lease_get_vendor_specific().
6604 * systemd-networkd gained a number of new configuration options.
6606 - A new boolean configuration option for TAP devices called
6607 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
6608 device, thus allowing to send and receive GSO packets.
6610 - A new tunnel configuration option called 'CopyDSCP='.
6611 If enabled, the DSCP field of ip6 tunnels is copied into the
6612 decapsulated packet.
6614 - A set of boolean bridge configuration options were added.
6615 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
6616 and 'UnicastFlood=' are now parsed by networkd and applied to the
6617 respective bridge link device via the respective IFLA_BRPORT_*
6620 - A new string configuration option to override the hostname sent
6621 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
6622 is true, networkd will use the configured hostname instead of the
6623 system hostname when sending DHCP requests.
6625 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
6626 networkd will configure the IPv6 flow-label of the tunnel device
6627 according to RFC2460.
6629 - The 'macvtap' virtual network devices are now supported, similar to
6630 the already supported 'macvlan' devices.
6632 * systemd-resolved now implements RFC5452 to improve resilience against
6633 cache poisoning. Additionally, source port randomization is enabled
6634 by default to further protect against DNS spoofing attacks.
6636 * nss-mymachines now supports translating UIDs and GIDs of running
6637 containers with user-namespaces enabled. If a container 'foo'
6638 translates a host uid 'UID' to the container uid 'TUID', then
6639 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
6640 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
6641 mapped as 'vg-foo-TGID'.
6643 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
6644 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
6645 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
6646 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
6647 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
6648 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
6649 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
6650 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
6651 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
6652 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
6654 — Berlin, 2015-07-29
6658 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
6659 There are no known issues with current sysfs, and udev does not need
6660 or should be used to work around such bugs.
6662 * udev does no longer enable USB HID power management. Several reports
6663 indicate, that some devices cannot handle that setting.
6665 * The udev accelerometer helper was removed. The functionality
6666 is now fully included in iio-sensor-proxy. But this means,
6667 older iio-sensor-proxy versions will no longer provide
6668 accelerometer/orientation data with this systemd version.
6669 Please upgrade iio-sensor-proxy to version 1.0.
6671 * networkd gained a new configuration option IPv6PrivacyExtensions=
6672 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
6673 for Stateless Address") on selected networks.
6675 * For the sake of fewer build-time dependencies and less code in the
6676 main repository, the python bindings are about to be removed in the
6677 next release. A new repository has been created which accommodates
6678 the code from now on, and we kindly ask distributions to create a
6679 separate package for this. The removal will take place in v223.
6681 https://github.com/systemd/python-systemd
6683 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
6684 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
6685 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
6686 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
6687 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
6688 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
6689 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
6690 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
6691 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
6692 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
6694 — Berlin, 2015-07-07
6698 * The sd-bus.h and sd-event.h APIs have now been declared
6699 stable and have been added to the official interface of
6700 libsystemd.so. sd-bus implements an alternative D-Bus client
6701 library, that is relatively easy to use, very efficient and
6702 supports both classic D-Bus as well as kdbus as transport
6703 backend. sd-event is a generic event loop abstraction that
6704 is built around Linux epoll, but adds features such as event
6705 prioritization or efficient timer handling. Both APIs are good
6706 choices for C programs looking for a bus and/or event loop
6707 implementation that is minimal and does not have to be
6708 portable to other kernels.
6710 * kdbus support is no longer compile-time optional. It is now
6711 always built-in. However, it can still be disabled at
6712 runtime using the kdbus=0 kernel command line setting, and
6713 that setting may be changed to default to off, by specifying
6714 --disable-kdbus at build-time. Note though that the kernel
6715 command line setting has no effect if the kdbus.ko kernel
6716 module is not installed, in which case kdbus is (obviously)
6717 also disabled. We encourage all downstream distributions to
6718 begin testing kdbus by adding it to the kernel images in the
6719 development distributions, and leaving kdbus support in
6722 * The minimal required util-linux version has been bumped to
6725 * Support for chkconfig (--enable-chkconfig) was removed in
6726 favor of calling an abstraction tool
6727 /lib/systemd/systemd-sysv-install. This needs to be
6728 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
6729 in README for details.
6731 * If there's a systemd unit and a SysV init script for the
6732 same service name, and the user executes "systemctl enable"
6733 for it (or a related call), then this will now enable both
6734 (or execute the related operation on both), not just the
6737 * The libudev API documentation has been converted from gtkdoc
6740 * gudev has been removed from the systemd tree, it is now an
6743 * The systemd-cgtop tool learnt a new --raw switch to generate
6744 "raw" (machine parsable) output.
6746 * networkd's IPForwarding= .network file setting learnt the
6747 new setting "kernel", which ensures that networkd does not
6748 change the IP forwarding sysctl from the default kernel
6751 * The systemd-logind bus API now exposes a new boolean
6752 property "Docked" that reports whether logind considers the
6753 system "docked", i.e. connected to a docking station or not.
6755 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
6756 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
6757 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
6758 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
6759 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
6760 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
6761 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
6762 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
6763 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
6764 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
6765 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
6766 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
6767 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
6768 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
6769 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
6770 Fink, Zbigniew Jędrzejewski-Szmek
6772 — Berlin, 2015-06-19
6776 * The gudev library has been extracted into a separate repository
6777 available at: https://git.gnome.org/browse/libgudev/
6778 It is now managed as part of the Gnome project. Distributions
6779 are recommended to pass --disable-gudev to systemd and use
6780 gudev from the Gnome project instead. gudev is still included
6781 in systemd, for now. It will be removed soon, though. Please
6782 also see the announcement-thread on systemd-devel:
6783 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
6785 * systemd now exposes a CPUUsageNSec= property for each
6786 service unit on the bus, that contains the overall consumed
6787 CPU time of a service (the sum of what each process of the
6788 service consumed). This value is only available if
6789 CPUAccounting= is turned on for a service, and is then shown
6790 in the "systemctl status" output.
6792 * Support for configuring alternative mappings of the old SysV
6793 runlevels to systemd targets has been removed. They are now
6794 hardcoded in a way that runlevels 2, 3, 4 all map to
6795 multi-user.target and 5 to graphical.target (which
6796 previously was already the default behaviour).
6798 * The auto-mounter logic gained support for mount point
6799 expiry, using a new TimeoutIdleSec= setting in .automount
6800 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
6802 * The EFI System Partition (ESP) as mounted to /boot by
6803 systemd-efi-boot-generator will now be unmounted
6804 automatically after 2 minutes of not being used. This should
6805 minimize the risk of ESP corruptions.
6807 * New /etc/fstab options x-systemd.requires= and
6808 x-systemd.requires-mounts-for= are now supported to express
6809 additional dependencies for mounts. This is useful for
6810 journalling file systems that support external journal
6811 devices or overlay file systems that require underlying file
6812 systems to be mounted.
6814 * systemd does not support direct live-upgrades (via systemctl
6815 daemon-reexec) from versions older than v44 anymore. As no
6816 distribution we are aware of shipped such old versions in a
6817 stable release this should not be problematic.
6819 * When systemd forks off a new per-connection service instance
6820 it will now set the $REMOTE_ADDR environment variable to the
6821 remote IP address, and $REMOTE_PORT environment variable to
6822 the remote IP port. This behaviour is similar to the
6823 corresponding environment variables defined by CGI.
6825 * systemd-networkd gained support for uplink failure
6826 detection. The BindCarrier= option allows binding interface
6827 configuration dynamically to the link sense of other
6828 interfaces. This is useful to achieve behaviour like in
6831 * systemd-networkd gained support for configuring the DHCP
6832 client identifier to use when requesting leases.
6834 * systemd-networkd now has a per-network UseNTP= option to
6835 configure whether NTP server information acquired via DHCP
6836 is passed on to services like systemd-timesyncd.
6838 * systemd-networkd gained support for vti6 tunnels.
6840 * Note that systemd-networkd manages the sysctl variable
6841 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
6842 it is configured for since v219. The variable controls IP
6843 forwarding, and is a per-interface alternative to the global
6844 /proc/sys/net/ipv[46]/ip_forward. This setting is
6845 configurable in the IPForward= option, which defaults to
6846 "no". This means if networkd is used for an interface it is
6847 no longer sufficient to set the global sysctl option to turn
6848 on IP forwarding! Instead, the .network file option
6849 IPForward= needs to be turned on! Note that the
6850 implementation of this behaviour was broken in v219 and has
6853 * Many bonding and vxlan options are now configurable in
6856 * systemd-nspawn gained a new --property= setting to set unit
6857 properties for the container scope. This is useful for
6858 setting resource parameters (e.g. "CPUShares=500") on
6859 containers started from the command line.
6861 * systemd-nspawn gained a new --private-users= switch to make
6862 use of user namespacing available on recent Linux kernels.
6864 * systemd-nspawn may now be called as part of a shell pipeline
6865 in which case the pipes used for stdin and stdout are passed
6866 directly to the process invoked in the container, without
6867 indirection via a pseudo tty.
6869 * systemd-nspawn gained a new switch to control the UNIX
6870 signal to use when killing the init process of the container
6873 * systemd-nspawn gained a new --overlay= switch for mounting
6874 overlay file systems into the container using the new kernel
6877 * When a container image is imported via systemd-importd and
6878 the host file system is not btrfs, a loopback block device
6879 file is created in /var/lib/machines.raw with a btrfs file
6880 system inside. It is then mounted to /var/lib/machines to
6881 enable btrfs features for container management. The loopback
6882 file and btrfs file system is grown as needed when container
6883 images are imported via systemd-importd.
6885 * systemd-machined/systemd-importd gained support for btrfs
6886 quota, to enforce container disk space limits on disk. This
6887 is exposed in "machinectl set-limit".
6889 * systemd-importd now can import containers from local .tar,
6890 .raw and .qcow2 images, and export them to .tar and .raw. It
6891 can also import dkr v2 images now from the network (on top
6894 * systemd-importd gained support for verifying downloaded
6895 images with gpg2 (previously only gpg1 was supported).
6897 * systemd-machined, systemd-logind, systemd: most bus calls are
6898 now accessible to unprivileged processes via polkit. Also,
6899 systemd-logind will now allow users to kill their own sessions
6900 without further privileges or authorization.
6902 * systemd-shutdownd has been removed. This service was
6903 previously responsible for implementing scheduled shutdowns
6904 as exposed in /usr/bin/shutdown's time parameter. This
6905 functionality has now been moved into systemd-logind and is
6906 accessible via a bus interface.
6908 * "systemctl reboot" gained a new switch --firmware-setup that
6909 can be used to reboot into the EFI firmware setup, if that
6910 is available. systemd-logind now exposes an API on the bus
6911 to trigger such reboots, in case graphical desktop UIs want
6912 to cover this functionality.
6914 * "systemctl enable", "systemctl disable" and "systemctl mask"
6915 now support a new "--now" switch. If specified the units
6916 that are enabled will also be started, and the ones
6917 disabled/masked also stopped.
6919 * The Gummiboot EFI boot loader tool has been merged into
6920 systemd, and renamed to "systemd-boot". The bootctl tool has been
6921 updated to support systemd-boot.
6923 * An EFI kernel stub has been added that may be used to create
6924 kernel EFI binaries that contain not only the actual kernel,
6925 but also an initrd, boot splash, command line and OS release
6926 information. This combined binary can then be signed as a
6927 single image, so that the firmware can verify it all in one
6928 step. systemd-boot has special support for EFI binaries created
6929 like this and can extract OS release information from them
6930 and show them in the boot menu. This functionality is useful
6931 to implement cryptographically verified boot schemes.
6933 * Optional support has been added to systemd-fsck to pass
6934 fsck's progress report to an AF_UNIX socket in the file
6937 * udev will no longer create device symlinks for all block devices by
6938 default. A deny list for excluding special block devices from this
6939 logic has been turned into a allow list that requires picking block
6940 devices explicitly that require device symlinks.
6942 * A new (currently still internal) API sd-device.h has been
6943 added to libsystemd. This modernized API is supposed to
6944 replace libudev eventually. In fact, already much of libudev
6945 is now just a wrapper around sd-device.h.
6947 * A new hwdb database for storing metadata about pointing
6948 stick devices has been added.
6950 * systemd-tmpfiles gained support for setting file attributes
6951 similar to the "chattr" tool with new 'h' and 'H' lines.
6953 * systemd-journald will no longer unconditionally set the
6954 btrfs NOCOW flag on new journal files. This is instead done
6955 with tmpfiles snippet using the new 'h' line type. This
6956 allows easy disabling of this logic, by masking the
6957 journal-nocow.conf tmpfiles file.
6959 * systemd-journald will now translate audit message types to
6960 human readable identifiers when writing them to the
6961 journal. This should improve readability of audit messages.
6963 * The LUKS logic gained support for the offset= and skip=
6964 options in /etc/crypttab, as previously implemented by
6967 * /usr/lib/os-release gained a new optional field VARIANT= for
6968 distributions that support multiple variants (such as a
6969 desktop edition, a server edition, ...)
6971 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
6972 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
6973 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
6974 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
6975 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
6976 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
6977 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
6978 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
6979 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
6980 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
6981 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
6982 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
6983 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
6984 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
6985 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
6986 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
6987 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
6988 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
6989 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
6990 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
6991 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
6992 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
6993 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
6994 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
6995 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
6996 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
6997 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
6999 — Berlin, 2015-05-22
7003 * Introduce a new API "sd-hwdb.h" for querying the hardware
7004 metadata database. With this minimal interface one can query
7005 and enumerate the udev hwdb, decoupled from the old libudev
7006 library. libudev's interface for this is now only a wrapper
7007 around sd-hwdb. A new tool systemd-hwdb has been added to
7008 interface with and update the database.
7010 * When any of systemd's tools copies files (for example due to
7011 tmpfiles' C lines) a btrfs reflink will attempted first,
7012 before bytewise copying is done.
7014 * systemd-nspawn gained a new --ephemeral switch. When
7015 specified a btrfs snapshot is taken of the container's root
7016 directory, and immediately removed when the container
7017 terminates again. Thus, a container can be started whose
7018 changes never alter the container's root directory, and are
7019 lost on container termination. This switch can also be used
7020 for starting a container off the root file system of the
7021 host without affecting the host OS. This switch is only
7022 available on btrfs file systems.
7024 * systemd-nspawn gained a new --template= switch. It takes the
7025 path to a container tree to use as template for the tree
7026 specified via --directory=, should that directory be
7027 missing. This allows instantiating containers dynamically,
7028 on first run. This switch is only available on btrfs file
7031 * When a .mount unit refers to a mount point on which multiple
7032 mounts are stacked, and the .mount unit is stopped all of
7033 the stacked mount points will now be unmounted until no
7034 mount point remains.
7036 * systemd now has an explicit notion of supported and
7037 unsupported unit types. Jobs enqueued for unsupported unit
7038 types will now fail with an "unsupported" error code. More
7039 specifically .swap, .automount and .device units are not
7040 supported in containers, .busname units are not supported on
7041 non-kdbus systems. .swap and .automount are also not
7042 supported if their respective kernel compile time options
7045 * machinectl gained support for two new "copy-from" and
7046 "copy-to" commands for copying files from a running
7047 container to the host or vice versa.
7049 * machinectl gained support for a new "bind" command to bind
7050 mount host directories into local containers. This is
7051 currently only supported for nspawn containers.
7053 * networkd gained support for configuring bridge forwarding
7054 database entries (fdb) from .network files.
7056 * A new tiny daemon "systemd-importd" has been added that can
7057 download container images in tar, raw, qcow2 or dkr formats,
7058 and make them available locally in /var/lib/machines, so
7059 that they can run as nspawn containers. The daemon can GPG
7060 verify the downloads (not supported for dkr, since it has no
7061 provisions for verifying downloads). It will transparently
7062 decompress bz2, xz, gzip compressed downloads if necessary,
7063 and restore sparse files on disk. The daemon uses privilege
7064 separation to ensure the actual download logic runs with
7065 fewer privileges than the daemon itself. machinectl has
7066 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
7067 make the functionality of importd available to the
7068 user. With this in place the Fedora and Ubuntu "Cloud"
7069 images can be downloaded and booted as containers unmodified
7070 (the Fedora images lack the appropriate GPG signature files
7071 currently, so they cannot be verified, but this will change
7072 soon, hopefully). Note that downloading images is currently
7073 only fully supported on btrfs.
7075 * machinectl is now able to list container images found in
7076 /var/lib/machines, along with some metadata about sizes of
7077 disk and similar. If the directory is located on btrfs and
7078 quota is enabled, this includes quota display. A new command
7079 "image-status" has been added that shows additional
7080 information about images.
7082 * machinectl is now able to clone container images
7083 efficiently, if the underlying file system (btrfs) supports
7084 it, with the new "machinectl clone" command. It also
7085 gained commands for renaming and removing images, as well as
7086 marking them read-only or read-write (supported also on
7087 legacy file systems).
7089 * networkd gained support for collecting LLDP network
7090 announcements, from hardware that supports this. This is
7091 shown in networkctl output.
7093 * systemd-run gained support for a new -t (--pty) switch for
7094 invoking a binary on a pty whose input and output is
7095 connected to the invoking terminal. This allows executing
7096 processes as system services while interactively
7097 communicating with them via the terminal. Most interestingly
7098 this is supported across container boundaries. Invoking
7099 "systemd-run -t /bin/bash" is an alternative to running a
7100 full login session, the difference being that the former
7101 will not register a session, nor go through the PAM session
7104 * tmpfiles gained support for a new "v" line type for creating
7105 btrfs subvolumes. If the underlying file system is a legacy
7106 file system, this automatically degrades to creating a
7107 normal directory. Among others /var/lib/machines is now
7108 created like this at boot, should it be missing.
7110 * The directory /var/lib/containers/ has been deprecated and
7111 been replaced by /var/lib/machines. The term "machines" has
7112 been used in the systemd context as generic term for both
7113 VMs and containers, and hence appears more appropriate for
7114 this, as the directory can also contain raw images bootable
7117 * systemd-nspawn when invoked with -M but without --directory=
7118 or --image= is now capable of searching for the container
7119 root directory, subvolume or disk image automatically, in
7120 /var/lib/machines. systemd-nspawn@.service has been updated
7121 to make use of this, thus allowing it to be used for raw
7124 * A new machines.target unit has been introduced that is
7125 supposed to group all containers/VMs invoked as services on
7126 the system. systemd-nspawn@.service has been updated to
7127 integrate with that.
7129 * machinectl gained a new "start" command, for invoking a
7130 container as a service. "machinectl start foo" is mostly
7131 equivalent to "systemctl start systemd-nspawn@foo.service",
7132 but handles escaping in a nicer way.
7134 * systemd-nspawn will now mount most of the cgroupfs tree
7135 read-only into each container, with the exception of the
7136 container's own subtree in the name=systemd hierarchy.
7138 * journald now sets the special FS_NOCOW file flag for its
7139 journal files. This should improve performance on btrfs, by
7140 avoiding heavy fragmentation when journald's write-pattern
7141 is used on COW file systems. It degrades btrfs' data
7142 integrity guarantees for the files to the same levels as for
7143 ext3/ext4 however. This should be OK though as journald does
7144 its own data integrity checks and all its objects are
7145 checksummed on disk. Also, journald should handle btrfs disk
7146 full events a lot more gracefully now, by processing SIGBUS
7147 errors, and not relying on fallocate() anymore.
7149 * When journald detects that journal files it is writing to
7150 have been deleted it will immediately start new journal
7153 * systemd now provides a way to store file descriptors
7154 per-service in PID 1. This is useful for daemons to ensure
7155 that fds they require are not lost during a daemon
7156 restart. The fds are passed to the daemon on the next
7157 invocation in the same way socket activation fds are
7158 passed. This is now used by journald to ensure that the
7159 various sockets connected to all the system's stdout/stderr
7160 are not lost when journald is restarted. File descriptors
7161 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
7162 an extension to sd_notify(). Note that a limit is enforced
7163 on the number of fds a service can store in PID 1, and it
7164 defaults to 0, so that no fds may be stored, unless this is
7165 explicitly turned on.
7167 * The default TERM variable to use for units connected to a
7168 terminal, when no other value is explicitly is set is now
7169 vt220 rather than vt102. This should be fairly safe still,
7170 but allows PgUp/PgDn work.
7172 * The /etc/crypttab option header= as known from Debian is now
7175 * "loginctl user-status" and "loginctl session-status" will
7176 now show the last 10 lines of log messages of the
7177 user/session following the status output. Similar,
7178 "machinectl status" will show the last 10 log lines
7179 associated with a virtual machine or container
7180 service. (Note that this is usually not the log messages
7181 done in the VM/container itself, but simply what the
7182 container manager logs. For nspawn this includes all console
7185 * "loginctl session-status" without further argument will now
7186 show the status of the session of the caller. Similar,
7187 "lock-session", "unlock-session", "activate",
7188 "enable-linger", "disable-linger" may now be called without
7189 session/user parameter in which case they apply to the
7190 caller's session/user.
7192 * An X11 session scriptlet is now shipped that uploads
7193 $DISPLAY and $XAUTHORITY into the environment of the systemd
7194 --user daemon if a session begins. This should improve
7195 compatibility with X11 enabled applications run as systemd
7198 * Generators are now subject to masking via /etc and /run, the
7199 same way as unit files.
7201 * networkd .network files gained support for configuring
7202 per-link IPv4/IPv6 packet forwarding as well as IPv4
7203 masquerading. This is by default turned on for veth links to
7204 containers, as registered by systemd-nspawn. This means that
7205 nspawn containers run with --network-veth will now get
7206 automatic routed access to the host's networks without any
7207 further configuration or setup, as long as networkd runs on
7210 * systemd-nspawn gained the --port= (-p) switch to expose TCP
7211 or UDP posts of a container on the host. With this in place
7212 it is possible to run containers with private veth links
7213 (--network-veth), and have their functionality exposed on
7214 the host as if their services were running directly on the
7217 * systemd-nspawn's --network-veth switch now gained a short
7218 version "-n", since with the changes above it is now truly
7219 useful out-of-the-box. The systemd-nspawn@.service has been
7220 updated to make use of it too by default.
7222 * systemd-nspawn will now maintain a per-image R/W lock, to
7223 ensure that the same image is not started more than once
7224 writable. (It's OK to run an image multiple times
7225 simultaneously in read-only mode.)
7227 * systemd-nspawn's --image= option is now capable of
7228 dissecting and booting MBR and GPT disk images that contain
7229 only a single active Linux partition. Previously it
7230 supported only GPT disk images with proper GPT type
7231 IDs. This allows running cloud images from major
7232 distributions directly with systemd-nspawn, without
7235 * In addition to collecting mouse dpi data in the udev
7236 hardware database, there's now support for collecting angle
7237 information for mouse scroll wheels. The database is
7238 supposed to guarantee similar scrolling behavior on mice
7239 that it knows about. There's also support for collecting
7240 information about Touchpad types.
7242 * udev's input_id built-in will now also collect touch screen
7243 dimension data and attach it to probed devices.
7245 * /etc/os-release gained support for a Distribution Privacy
7248 * networkd gained support for creating "ipvlan", "gretap",
7249 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
7251 * systemd-tmpfiles gained support for "a" lines for setting
7254 * systemd-nspawn will now mount /tmp in the container to
7255 tmpfs, automatically.
7257 * systemd now exposes the memory.usage_in_bytes cgroup
7258 attribute and shows it for each service in the "systemctl
7259 status" output, if available.
7261 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
7262 immediate reboot is triggered. This useful if shutdown is
7263 hung and is unable to complete, to expedite the
7264 operation. Note that this kind of reboot will still unmount
7265 all file systems, and hence should not result in fsck being
7268 * A .device unit for an optical block device will now be
7269 considered active only when a medium is in the drive. Also,
7270 mount units are now bound to their backing devices thus
7271 triggering automatic unmounting when devices become
7272 unavailable. With this in place systemd will now
7273 automatically unmount left-over mounts when a CD-ROM is
7274 ejected or an USB stick is yanked from the system.
7276 * networkd-wait-online now has support for waiting for
7277 specific interfaces only (with globbing), and for giving up
7278 after a configurable timeout.
7280 * networkd now exits when idle. It will be automatically
7281 restarted as soon as interfaces show up, are removed or
7282 change state. networkd will stay around as long as there is
7283 at least one DHCP state machine or similar around, that keep
7286 * networkd may now configure IPv6 link-local addressing in
7287 addition to IPv4 link-local addressing.
7289 * The IPv6 "token" for use in SLAAC may now be configured for
7290 each .network interface in networkd.
7292 * Routes configured with networkd may now be assigned a scope
7295 * networkd's [Match] sections now support globbing and lists
7296 of multiple space-separated matches per item.
7298 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
7299 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
7300 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
7301 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
7302 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
7303 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
7304 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
7305 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
7306 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
7307 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
7308 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
7309 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
7310 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
7311 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
7312 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
7313 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
7314 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
7315 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
7316 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
7317 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
7318 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
7319 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
7320 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
7321 Hoffmann, Zbigniew Jędrzejewski-Szmek
7323 — Berlin, 2015-02-16
7327 * When querying unit file enablement status (for example via
7328 "systemctl is-enabled"), a new state "indirect" is now known
7329 which indicates that a unit might not be enabled itself, but
7330 another unit listed in its Also= setting might be.
7332 * Similar to the various existing ConditionXYZ= settings for
7333 units, there are now matching AssertXYZ= settings. While
7334 failing conditions cause a unit to be skipped, but its job
7335 to succeed, failing assertions declared like this will cause
7336 a unit start operation and its job to fail.
7338 * hostnamed now knows a new chassis type "embedded".
7340 * systemctl gained a new "edit" command. When used on a unit
7341 file, this allows extending unit files with .d/ drop-in
7342 configuration snippets or editing the full file (after
7343 copying it from /usr/lib to /etc). This will invoke the
7344 user's editor (as configured with $EDITOR), and reload the
7345 modified configuration after editing.
7347 * "systemctl status" now shows the suggested enablement state
7348 for a unit, as declared in the (usually vendor-supplied)
7349 system preset files.
7351 * nss-myhostname will now resolve the single-label hostname
7352 "gateway" to the locally configured default IP routing
7353 gateways, ordered by their metrics. This assigns a stable
7354 name to the used gateways, regardless which ones are
7355 currently configured. Note that the name will only be
7356 resolved after all other name sources (if nss-myhostname is
7357 configured properly) and should hence not negatively impact
7358 systems that use the single-label hostname "gateway" in
7361 * systemd-inhibit now allows filtering by mode when listing
7364 * Scope and service units gained a new "Delegate" boolean
7365 property, which, when set, allows processes running inside the
7366 unit to further partition resources. This is primarily
7367 useful for systemd user instances as well as container
7370 * journald will now pick up audit messages directly from
7371 the kernel, and log them like any other log message. The
7372 audit fields are split up and fully indexed. This means that
7373 journalctl in many ways is now a (nicer!) alternative to
7374 ausearch, the traditional audit client. Note that this
7375 implements only a minimal audit client. If you want the
7376 special audit modes like reboot-on-log-overflow, please use
7377 the traditional auditd instead, which can be used in
7378 parallel to journald.
7380 * The ConditionSecurity= unit file option now understands the
7381 special string "audit" to check whether auditing is
7384 * journalctl gained two new commands --vacuum-size= and
7385 --vacuum-time= to delete old journal files until the
7386 remaining ones take up no more than the specified size on disk,
7387 or are not older than the specified time.
7389 * A new, native PPPoE library has been added to sd-network,
7390 systemd's library of light-weight networking protocols. This
7391 library will be used in a future version of networkd to
7392 enable PPPoE communication without an external pppd daemon.
7394 * The busctl tool now understands a new "capture" verb that
7395 works similar to "monitor", but writes a packet capture
7396 trace to STDOUT that can be redirected to a file which is
7397 compatible with libcap's capture file format. This can then
7398 be loaded in Wireshark and similar tools to inspect bus
7401 * The busctl tool now understands a new "tree" verb that shows
7402 the object trees of a specific service on the bus, or of all
7405 * The busctl tool now understands a new "introspect" verb that
7406 shows all interfaces and members of objects on the bus,
7407 including their signature and values. This is particularly
7408 useful to get more information about bus objects shown by
7409 the new "busctl tree" command.
7411 * The busctl tool now understands new verbs "call",
7412 "set-property" and "get-property" for invoking bus method
7413 calls, setting and getting bus object properties in a
7416 * busctl gained a new --augment-creds= argument that controls
7417 whether the tool shall augment credential information it
7418 gets from the bus with data from /proc, in a possibly
7421 * nspawn's --link-journal= switch gained two new values
7422 "try-guest" and "try-host" that work like "guest" and
7423 "host", but do not fail if the host has no persistent
7424 journalling enabled. -j is now equivalent to
7425 --link-journal=try-guest.
7427 * macvlan network devices created by nspawn will now have
7428 stable MAC addresses.
7430 * A new SmackProcessLabel= unit setting has been added, which
7431 controls the SMACK security label processes forked off by
7432 the respective unit shall use.
7434 * If compiled with --enable-xkbcommon, systemd-localed will
7435 verify x11 keymap settings by compiling the given keymap. It
7436 will spew out warnings if the compilation fails. This
7437 requires libxkbcommon to be installed.
7439 * When a coredump is collected, a larger number of metadata
7440 fields is now collected and included in the journal records
7441 created for it. More specifically, control group membership,
7442 environment variables, memory maps, working directory,
7443 chroot directory, /proc/$PID/status, and a list of open file
7444 descriptors is now stored in the log entry.
7446 * The udev hwdb now contains DPI information for mice. For
7449 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
7451 * All systemd programs that read standalone configuration
7452 files in /etc now also support a corresponding series of
7453 .conf.d configuration directories in /etc/, /run/,
7454 /usr/local/lib/, /usr/lib/, and (if configured with
7455 --enable-split-usr) /lib/. In particular, the following
7456 configuration files now have corresponding configuration
7457 directories: system.conf user.conf, logind.conf,
7458 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
7459 resolved.conf, timesyncd.conf, journal-remote.conf, and
7460 journal-upload.conf. Note that distributions should use the
7461 configuration directories in /usr/lib/; the directories in
7462 /etc/ are reserved for the system administrator.
7464 * systemd-rfkill will no longer take the rfkill device name
7465 into account when storing rfkill state on disk, as the name
7466 might be dynamically assigned and not stable. Instead, the
7467 ID_PATH udev variable combined with the rfkill type (wlan,
7468 bluetooth, ...) is used.
7470 * A new service systemd-machine-id-commit.service has been
7471 added. When used on systems where /etc is read-only during
7472 boot, and /etc/machine-id is not initialized (but an empty
7473 file), this service will copy the temporary machine ID
7474 created as replacement into /etc after the system is fully
7475 booted up. This is useful for systems that are freshly
7476 installed with a non-initialized machine ID, but should get
7477 a fixed machine ID for subsequent boots.
7479 * networkd's .netdev files now provide a large set of
7480 configuration parameters for VXLAN devices. Similarly, the
7481 bridge port cost parameter is now configurable in .network
7482 files. There's also new support for configuring IP source
7483 routing. networkd .link files gained support for a new
7484 OriginalName= match that is useful to match against the
7485 original interface name the kernel assigned. .network files
7486 may include MTU= and MACAddress= fields for altering the MTU
7487 and MAC address while being connected to a specific network
7490 * The LUKS logic gained supported for configuring
7491 UUID-specific key files. There's also new support for naming
7492 LUKS device from the kernel command line, using the new
7493 luks.name= argument.
7495 * Timer units may now be transiently created via the bus API
7496 (this was previously already available for scope and service
7497 units). In addition it is now possible to create multiple
7498 transient units at the same time with a single bus call. The
7499 "systemd-run" tool has been updated to make use of this for
7500 running commands on a specified time, in at(1)-style.
7502 * tmpfiles gained support for "t" lines, for assigning
7503 extended attributes to files. Among other uses this may be
7504 used to assign SMACK labels to files.
7506 Contributions from: Alin Rauta, Alison Chaiken, Andrej
7507 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
7508 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
7509 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
7510 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
7511 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
7512 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
7513 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
7514 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
7515 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
7516 Michael Biebl, Michael Chapman, Michael Marineau, Michal
7517 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
7518 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
7519 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
7520 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
7521 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7522 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
7523 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
7525 — Berlin, 2014-12-10
7529 * journalctl gained the new options -t/--identifier= to match
7530 on the syslog identifier (aka "tag"), as well as --utc to
7531 show log timestamps in the UTC timezone. journalctl now also
7532 accepts -n/--lines=all to disable line capping in a pager.
7534 * journalctl gained a new switch, --flush, that synchronously
7535 flushes logs from /run/log/journal to /var/log/journal if
7536 persistent storage is enabled. systemd-journal-flush.service
7537 now waits until the operation is complete.
7539 * Services can notify the manager before they start a reload
7540 (by sending RELOADING=1) or shutdown (by sending
7541 STOPPING=1). This allows the manager to track and show the
7542 internal state of daemons and closes a race condition when
7543 the process is still running but has closed its D-Bus
7546 * Services with Type=oneshot do not have to have any ExecStart
7549 * User units are now loaded also from
7550 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
7551 /run/systemd/user directory that was already previously
7552 supported, but is under the control of the user.
7554 * Job timeouts (i.e. timeouts on the time a job that is
7555 queued stays in the run queue) can now optionally result in
7556 immediate reboot or power-off actions (JobTimeoutAction= and
7557 JobTimeoutRebootArgument=). This is useful on ".target"
7558 units, to limit the maximum time a target remains
7559 undispatched in the run queue, and to trigger an emergency
7560 operation in such a case. This is now used by default to
7561 turn off the system if boot-up (as defined by everything in
7562 basic.target) hangs and does not complete for at least
7563 15min. Also, if power-off or reboot hang for at least 30min
7564 an immediate power-off/reboot operation is triggered. This
7565 functionality is particularly useful to increase reliability
7566 on embedded devices, but also on laptops which might
7567 accidentally get powered on when carried in a backpack and
7568 whose boot stays stuck in a hard disk encryption passphrase
7571 * systemd-logind can be configured to also handle lid switch
7572 events even when the machine is docked or multiple displays
7573 are attached (HandleLidSwitchDocked= option).
7575 * A helper binary and a service have been added which can be
7576 used to resume from hibernation in the initramfs. A
7577 generator will parse the resume= option on the kernel
7578 command line to trigger resume.
7580 * A user console daemon systemd-consoled has been
7581 added. Currently, it is a preview, and will so far open a
7582 single terminal on each session of the user marked as
7583 Desktop=systemd-console.
7585 * Route metrics can be specified for DHCP routes added by
7588 * The SELinux context of socket-activated services can be set
7589 from the information provided by the networking stack
7590 (SELinuxContextFromNet= option).
7592 * Userspace firmware loading support has been removed and
7593 the minimum supported kernel version is thus bumped to 3.7.
7595 * Timeout for udev workers has been increased from 1 to 3
7596 minutes, but a warning will be printed after 1 minute to
7597 help diagnose kernel modules that take a long time to load.
7599 * Udev rules can now remove tags on devices with TAG-="foobar".
7601 * systemd's readahead implementation has been removed. In many
7602 circumstances it didn't give expected benefits even for
7603 rotational disk drives and was becoming less relevant in the
7604 age of SSDs. As none of the developers has been using
7605 rotating media anymore, and nobody stepped up to actively
7606 maintain this component of systemd it has now been removed.
7608 * Swap units can use Options= to specify discard options.
7609 Discard options specified for swaps in /etc/fstab are now
7612 * Docker containers are now detected as a separate type of
7615 * The Password Agent protocol gained support for queries where
7616 the user input is shown, useful e.g. for user names.
7617 systemd-ask-password gained a new --echo option to turn that
7620 * The default sysctl.d/ snippets will now set:
7622 net.core.default_qdisc = fq_codel
7624 This selects Fair Queuing Controlled Delay as the default
7625 queuing discipline for network interfaces. fq_codel helps
7626 fight the network bufferbloat problem. It is believed to be
7627 a good default with no tuning required for most workloads.
7628 Downstream distributions may override this choice. On 10Gbit
7629 servers that do not do forwarding, "fq" may perform better.
7630 Systems without a good clocksource should use "pfifo_fast".
7632 * If kdbus is enabled during build a new option BusPolicy= is
7633 available for service units, that allows locking all service
7634 processes into a stricter bus policy, in order to limit
7635 access to various bus services, or even hide most of them
7636 from the service's view entirely.
7638 * networkctl will now show the .network and .link file
7639 networkd has applied to a specific interface.
7641 * sd-login gained a new API call sd_session_get_desktop() to
7642 query which desktop environment has been selected for a
7645 * UNIX utmp support is now compile-time optional to support
7646 legacy-free systems.
7648 * systemctl gained two new commands "add-wants" and
7649 "add-requires" for pulling in units from specific targets
7652 * If the word "rescue" is specified on the kernel command line
7653 the system will now boot into rescue mode (aka
7654 rescue.target), which was previously available only by
7655 specifying "1" or "systemd.unit=rescue.target" on the kernel
7656 command line. This new kernel command line option nicely
7657 mirrors the already existing "emergency" kernel command line
7660 * New kernel command line options mount.usr=, mount.usrflags=,
7661 mount.usrfstype= have been added that match root=, rootflags=,
7662 rootfstype= but allow mounting a specific file system to
7665 * The $NOTIFY_SOCKET is now also passed to control processes of
7666 services, not only the main process.
7668 * This version reenables support for fsck's -l switch. This
7669 means at least version v2.25 of util-linux is required for
7670 operation, otherwise dead-locks on device nodes may
7671 occur. Again: you need to update util-linux to at least
7672 v2.25 when updating systemd to v217.
7674 * The "multi-seat-x" tool has been removed from systemd, as
7675 its functionality has been integrated into X servers 1.16,
7676 and the tool is hence redundant. It is recommended to update
7677 display managers invoking this tool to simply invoke X
7678 directly from now on, again.
7680 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
7681 message flag has been added for all of systemd's polkit
7682 authenticated method calls has been added. In particular this
7683 now allows optional interactive authorization via polkit for
7684 many of PID1's privileged operations such as unit file
7685 enabling and disabling.
7687 * "udevadm hwdb --update" learnt a new switch "--usr" for
7688 placing the rebuilt hardware database in /usr instead of
7689 /etc. When used only hardware database entries stored in
7690 /usr will be used, and any user database entries in /etc are
7691 ignored. This functionality is useful for vendors to ship a
7692 pre-built database on systems where local configuration is
7693 unnecessary or unlikely.
7695 * Calendar time specifications in .timer units now also
7696 understand the strings "semi-annually", "quarterly" and
7697 "minutely" as shortcuts (in addition to the preexisting
7698 "annually", "hourly", ...).
7700 * systemd-tmpfiles will now correctly create files in /dev
7701 at boot which are marked for creation only at boot. It is
7702 recommended to always create static device nodes with 'c!'
7703 and 'b!', so that they are created only at boot and not
7704 overwritten at runtime.
7706 * When the watchdog logic is used for a service (WatchdogSec=)
7707 and the watchdog timeout is hit the service will now be
7708 terminated with SIGABRT (instead of just SIGTERM), in order
7709 to make sure a proper coredump and backtrace is
7710 generated. This ensures that hanging services will result in
7711 similar coredump/backtrace behaviour as services that hit a
7714 Contributions from: Andreas Henriksson, Andrei Borzenkov,
7715 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
7716 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
7717 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
7718 Herrmann, David Sommerseth, David Strauss, Emil Renner
7719 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
7720 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
7721 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
7722 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
7723 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
7724 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
7725 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
7726 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
7727 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
7728 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
7729 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
7730 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
7731 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
7732 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
7733 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
7734 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
7737 — Berlin, 2014-10-28
7741 * timedated no longer reads NTP implementation unit names from
7742 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
7743 implementations should add a
7745 Conflicts=systemd-timesyncd.service
7747 to their unit files to take over and replace systemd's NTP
7748 default functionality.
7750 * systemd-sysusers gained a new line type "r" for configuring
7751 which UID/GID ranges to allocate system users/groups
7752 from. Lines of type "u" may now add an additional column
7753 that specifies the home directory for the system user to be
7754 created. Also, systemd-sysusers may now optionally read user
7755 information from STDIN instead of a file. This is useful for
7756 invoking it from RPM preinst scriptlets that need to create
7757 users before the first RPM file is installed since these
7758 files might need to be owned by them. A new
7759 %sysusers_create_inline RPM macro has been introduced to do
7760 just that. systemd-sysusers now updates the shadow files as
7761 well as the user/group databases, which should enhance
7762 compatibility with certain tools like grpck.
7764 * A number of bus APIs of PID 1 now optionally consult polkit to
7765 permit access for otherwise unprivileged clients under certain
7766 conditions. Note that this currently doesn't support
7767 interactive authentication yet, but this is expected to be
7768 added eventually, too.
7770 * /etc/machine-info now has new fields for configuring the
7771 deployment environment of the machine, as well as the
7772 location of the machine. hostnamectl has been updated with
7773 new command to update these fields.
7775 * systemd-timesyncd has been updated to automatically acquire
7776 NTP server information from systemd-networkd, which might
7777 have been discovered via DHCP.
7779 * systemd-resolved now includes a caching DNS stub resolver
7780 and a complete LLMNR name resolution implementation. A new
7781 NSS module "nss-resolve" has been added which can be used
7782 instead of glibc's own "nss-dns" to resolve hostnames via
7783 systemd-resolved. Hostnames, addresses and arbitrary RRs may
7784 be resolved via systemd-resolved D-Bus APIs. In contrast to
7785 the glibc internal resolver systemd-resolved is aware of
7786 multi-homed system, and keeps DNS server and caches separate
7787 and per-interface. Queries are sent simultaneously on all
7788 interfaces that have DNS servers configured, in order to
7789 properly handle VPNs and local LANs which might resolve
7790 separate sets of domain names. systemd-resolved may acquire
7791 DNS server information from systemd-networkd automatically,
7792 which in turn might have discovered them via DHCP. A tool
7793 "systemd-resolve-host" has been added that may be used to
7794 query the DNS logic in resolved. systemd-resolved implements
7795 IDNA and automatically uses IDNA or UTF-8 encoding depending
7796 on whether classic DNS or LLMNR is used as transport. In the
7797 next releases we intend to add a DNSSEC and mDNS/DNS-SD
7798 implementation to systemd-resolved.
7800 * A new NSS module nss-mymachines has been added, that
7801 automatically resolves the names of all local registered
7802 containers to their respective IP addresses.
7804 * A new client tool "networkctl" for systemd-networkd has been
7805 added. It currently is entirely passive and will query
7806 networking configuration from udev, rtnetlink and networkd,
7807 and present it to the user in a very friendly
7808 way. Eventually, we hope to extend it to become a full
7809 control utility for networkd.
7811 * .socket units gained a new DeferAcceptSec= setting that
7812 controls the kernels' TCP_DEFER_ACCEPT sockopt for
7813 TCP. Similarly, support for controlling TCP keep-alive
7814 settings has been added (KeepAliveTimeSec=,
7815 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
7816 turning off Nagle's algorithm on TCP has been added
7819 * logind learned a new session type "web", for use in projects
7820 like Cockpit which register web clients as PAM sessions.
7822 * timer units with at least one OnCalendar= setting will now
7823 be started only after time-sync.target has been
7824 reached. This way they will not elapse before the system
7825 clock has been corrected by a local NTP client or
7826 similar. This is particular useful on RTC-less embedded
7827 machines, that come up with an invalid system clock.
7829 * systemd-nspawn's --network-veth= switch should now result in
7830 stable MAC addresses for both the outer and the inner side
7833 * systemd-nspawn gained a new --volatile= switch for running
7834 container instances with /etc or /var unpopulated.
7836 * The kdbus client code has been updated to use the new Linux
7837 3.17 memfd subsystem instead of the old kdbus-specific one.
7839 * systemd-networkd's DHCP client and server now support
7840 FORCERENEW. There are also new configuration options to
7841 configure the vendor client identifier and broadcast mode
7844 * systemd will no longer inform the kernel about the current
7845 timezone, as this is necessarily incorrect and racy as the
7846 kernel has no understanding of DST and similar
7847 concepts. This hence means FAT timestamps will be always
7848 considered UTC, similar to what Android is already
7849 doing. Also, when the RTC is configured to the local time
7850 (rather than UTC) systemd will never synchronize back to it,
7851 as this might confuse Windows at a later boot.
7853 * systemd-analyze gained a new command "verify" for offline
7854 validation of unit files.
7856 * systemd-networkd gained support for a couple of additional
7857 settings for bonding networking setups. Also, the metric for
7858 statically configured routes may now be configured. For
7859 network interfaces where this is appropriate the peer IP
7860 address may now be configured.
7862 * systemd-networkd's DHCP client will no longer request
7863 broadcasting by default, as this tripped up some networks.
7864 For hardware where broadcast is required the feature should
7865 be switched back on using RequestBroadcast=yes.
7867 * systemd-networkd will now set up IPv4LL addresses (when
7868 enabled) even if DHCP is configured successfully.
7870 * udev will now default to respect network device names given
7871 by the kernel when the kernel indicates that these are
7872 predictable. This behavior can be tweaked by changing
7873 NamePolicy= in the relevant .link file.
7875 * A new library systemd-terminal has been added that
7876 implements full TTY stream parsing and rendering. This
7877 library is supposed to be used later on for implementing a
7878 full userspace VT subsystem, replacing the current kernel
7881 * A new tool systemd-journal-upload has been added to push
7882 journal data to a remote system running
7883 systemd-journal-remote.
7885 * journald will no longer forward all local data to another
7886 running syslog daemon. This change has been made because
7887 rsyslog (which appears to be the most commonly used syslog
7888 implementation these days) no longer makes use of this, and
7889 instead pulls the data out of the journal on its own. Since
7890 forwarding the messages to a non-existent syslog server is
7891 more expensive than we assumed we have now turned this
7892 off. If you run a syslog server that is not a recent rsyslog
7893 version, you have to turn this option on again
7894 (ForwardToSyslog= in journald.conf).
7896 * journald now optionally supports the LZ4 compressor for
7897 larger journal fields. This compressor should perform much
7898 better than XZ which was the previous default.
7900 * machinectl now shows the IP addresses of local containers,
7901 if it knows them, plus the interface name of the container.
7903 * A new tool "systemd-escape" has been added that makes it
7904 easy to escape strings to build unit names and similar.
7906 * sd_notify() messages may now include a new ERRNO= field
7907 which is parsed and collected by systemd and shown among the
7908 "systemctl status" output for a service.
7910 * A new component "systemd-firstboot" has been added that
7911 queries the most basic systemd information (timezone,
7912 hostname, root password) interactively on first
7913 boot. Alternatively it may also be used to provision these
7914 things offline on OS images installed into directories.
7916 * The default sysctl.d/ snippets will now set
7918 net.ipv4.conf.default.promote_secondaries=1
7920 This has the benefit of no flushing secondary IP addresses
7921 when primary addresses are removed.
7923 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
7924 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
7925 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
7926 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
7927 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
7928 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
7929 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
7930 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
7931 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
7932 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
7933 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
7934 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
7935 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
7936 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
7937 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
7939 — Berlin, 2014-08-19
7943 * A new tool systemd-sysusers has been added. This tool
7944 creates system users and groups in /etc/passwd and
7945 /etc/group, based on static declarative system user/group
7946 definitions in /usr/lib/sysusers.d/. This is useful to
7947 enable factory resets and volatile systems that boot up with
7948 an empty /etc directory, and thus need system users and
7949 groups created during early boot. systemd now also ships
7950 with two default sysusers.d/ files for the most basic
7951 users and groups systemd and the core operating system
7954 * A new tmpfiles snippet has been added that rebuilds the
7955 essential files in /etc on boot, should they be missing.
7957 * A directive for ensuring automatic clean-up of
7958 /var/cache/man/ has been removed from the default
7959 configuration. This line should now be shipped by the man
7960 implementation. The necessary change has been made to the
7961 man-db implementation. Note that you need to update your man
7962 implementation to one that ships this line, otherwise no
7963 automatic clean-up of /var/cache/man will take place.
7965 * A new condition ConditionNeedsUpdate= has been added that
7966 may conditionalize services to only run when /etc or /var
7967 are "older" than the vendor operating system resources in
7968 /usr. This is useful for reconstructing or updating /etc
7969 after an offline update of /usr or a factory reset, on the
7970 next reboot. Services that want to run once after such an
7971 update or reset should use this condition and order
7972 themselves before the new systemd-update-done.service, which
7973 will mark the two directories as fully updated. A number of
7974 service files have been added making use of this, to rebuild
7975 the udev hardware database, the journald message catalog and
7976 dynamic loader cache (ldconfig). The systemd-sysusers tool
7977 described above also makes use of this now. With this in
7978 place it is now possible to start up a minimal operating
7979 system with /etc empty cleanly. For more information on the
7980 concepts involved see this recent blog story:
7982 http://0pointer.de/blog/projects/stateless.html
7984 * A new system group "input" has been introduced, and all
7985 input device nodes get this group assigned. This is useful
7986 for system-level software to get access to input devices. It
7987 complements what is already done for "audio" and "video".
7989 * systemd-networkd learnt minimal DHCPv4 server support in
7990 addition to the existing DHCPv4 client support. It also
7991 learnt DHCPv6 client and IPv6 Router Solicitation client
7992 support. The DHCPv4 client gained support for static routes
7993 passed in from the server. Note that the [DHCPv4] section
7994 known in older systemd-networkd versions has been renamed to
7995 [DHCP] and is now also used by the DHCPv6 client. Existing
7996 .network files using settings of this section should be
7997 updated, though compatibility is maintained. Optionally, the
7998 client hostname may now be sent to the DHCP server.
8000 * networkd gained support for vxlan virtual networks as well
8001 as tun/tap and dummy devices.
8003 * networkd gained support for automatic allocation of address
8004 ranges for interfaces from a system-wide pool of
8005 addresses. This is useful for dynamically managing a large
8006 number of interfaces with a single network configuration
8007 file. In particular this is useful to easily assign
8008 appropriate IP addresses to the veth links of a large number
8009 of nspawn instances.
8011 * RPM macros for processing sysusers, sysctl and binfmt
8012 drop-in snippets at package installation time have been
8015 * The /etc/os-release file should now be placed in
8016 /usr/lib/os-release. The old location is automatically
8017 created as symlink. /usr/lib is the more appropriate
8018 location of this file, since it shall actually describe the
8019 vendor operating system shipped in /usr, and not the
8020 configuration stored in /etc.
8022 * .mount units gained a new boolean SloppyOptions= setting
8023 that maps to mount(8)'s -s option which enables permissive
8024 parsing of unknown mount options.
8026 * tmpfiles learnt a new "L+" directive which creates a symlink
8027 but (unlike "L") deletes a pre-existing file first, should
8028 it already exist and not already be the correct
8029 symlink. Similarly, "b+", "c+" and "p+" directives have been
8030 added as well, which create block and character devices, as
8031 well as fifos in the filesystem, possibly removing any
8032 pre-existing files of different types.
8034 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
8035 'argument' field (which so far specified the source to
8036 symlink/copy the files from) is now optional. If omitted the
8037 same file os copied from /usr/share/factory/ suffixed by the
8038 full destination path. This is useful for populating /etc
8039 with essential files, by copying them from vendor defaults
8040 shipped in /usr/share/factory/etc.
8042 * A new command "systemctl preset-all" has been added that
8043 applies the service preset settings to all installed unit
8044 files. A new switch --preset-mode= has been added that
8045 controls whether only enable or only disable operations
8048 * A new command "systemctl is-system-running" has been added
8049 that allows checking the overall state of the system, for
8050 example whether it is fully up and running.
8052 * When the system boots up with an empty /etc, the equivalent
8053 to "systemctl preset-all" is executed during early boot, to
8054 make sure all default services are enabled after a factory
8057 * systemd now contains a minimal preset file that enables the
8058 most basic services systemd ships by default.
8060 * Unit files' [Install] section gained a new DefaultInstance=
8061 field for defining the default instance to create if a
8062 template unit is enabled with no instance specified.
8064 * A new passive target cryptsetup-pre.target has been added
8065 that may be used by services that need to make they run and
8066 finish before the first LUKS cryptographic device is set up.
8068 * The /dev/loop-control and /dev/btrfs-control device nodes
8069 are now owned by the "disk" group by default, opening up
8070 access to this group.
8072 * systemd-coredump will now automatically generate a
8073 stack trace of all core dumps taking place on the system,
8074 based on elfutils' libdw library. This stack trace is logged
8077 * systemd-coredump may now optionally store coredumps directly
8078 on disk (in /var/lib/systemd/coredump, possibly compressed),
8079 instead of storing them unconditionally in the journal. This
8080 mode is the new default. A new configuration file
8081 /etc/systemd/coredump.conf has been added to configure this
8082 and other parameters of systemd-coredump.
8084 * coredumpctl gained a new "info" verb to show details about a
8085 specific coredump. A new switch "-1" has also been added
8086 that makes sure to only show information about the most
8087 recent entry instead of all entries. Also, as the tool is
8088 generally useful now the "systemd-" prefix of the binary
8089 name has been removed. Distributions that want to maintain
8090 compatibility with the old name should add a symlink from
8091 the old name to the new name.
8093 * journald's SplitMode= now defaults to "uid". This makes sure
8094 that unprivileged users can access their own coredumps with
8095 coredumpctl without restrictions.
8097 * New kernel command line options "systemd.wants=" (for
8098 pulling an additional unit during boot), "systemd.mask="
8099 (for masking a specific unit for the boot), and
8100 "systemd.debug-shell" (for enabling the debug shell on tty9)
8101 have been added. This is implemented in the new generator
8102 "systemd-debug-generator".
8104 * systemd-nspawn will now by default filter a couple of
8105 syscalls for containers, among them those required for
8106 kernel module loading, direct x86 IO port access, swap
8107 management, and kexec. Most importantly though
8108 open_by_handle_at() is now prohibited for containers,
8109 closing a hole similar to a recently discussed vulnerability
8110 in docker regarding access to files on file hierarchies the
8111 container should normally not have access to. Note that, for
8112 nspawn, we generally make no security claims anyway (and
8113 this is explicitly documented in the man page), so this is
8114 just a fix for one of the most obvious problems.
8116 * A new man page file-hierarchy(7) has been added that
8117 contains a minimized, modernized version of the file system
8118 layout systemd expects, similar in style to the FHS
8119 specification or hier(5). A new tool systemd-path(1) has
8120 been added to query many of these paths for the local
8123 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
8124 longer done. Since the directory now has a per-user size
8125 limit, and is cleaned on logout this appears unnecessary,
8126 in particular since this now brings the lifecycle of this
8127 directory closer in line with how IPC objects are handled.
8129 * systemd.pc now exports a number of additional directories,
8130 including $libdir (which is useful to identify the library
8131 path for the primary architecture of the system), and a
8132 couple of drop-in directories.
8134 * udev's predictable network interface names now use the dev_port
8135 sysfs attribute, introduced in linux 3.15 instead of dev_id to
8136 distinguish between ports of the same PCI function. dev_id should
8137 only be used for ports using the same HW address, hence the need
8140 * machined has been updated to export the OS version of a
8141 container (read from /etc/os-release and
8142 /usr/lib/os-release) on the bus. This is now shown in
8143 "machinectl status" for a machine.
8145 * A new service setting RestartForceExitStatus= has been
8146 added. If configured to a set of exit signals or process
8147 return values, the service will be restarted when the main
8148 daemon process exits with any of them, regardless of the
8151 * systemctl's -H switch for connecting to remote systemd
8152 machines has been extended so that it may be used to
8153 directly connect to a specific container on the
8154 host. "systemctl -H root@foobar:waldi" will now connect as
8155 user "root" to host "foobar", and then proceed directly to
8156 the container named "waldi". Note that currently you have to
8157 authenticate as user "root" for this to work, as entering
8158 containers is a privileged operation.
8160 Contributions from: Andreas Henriksson, Benjamin Steinwender,
8161 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
8162 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
8163 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
8164 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
8165 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
8166 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
8167 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
8168 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
8169 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
8170 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
8171 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
8173 — Berlin, 2014-07-03
8177 * As an experimental feature, udev now tries to lock the
8178 disk device node (flock(LOCK_SH|LOCK_NB)) while it
8179 executes events for the disk or any of its partitions.
8180 Applications like partitioning programs can lock the
8181 disk device node (flock(LOCK_EX)) and claim temporary
8182 device ownership that way; udev will entirely skip all event
8183 handling for this disk and its partitions. If the disk
8184 was opened for writing, the close will trigger a partition
8185 table rescan in udev's "watch" facility, and if needed
8186 synthesize "change" events for the disk and all its partitions.
8187 This is now unconditionally enabled, and if it turns out to
8188 cause major problems, we might turn it on only for specific
8189 devices, or might need to disable it entirely. Device Mapper
8190 devices are excluded from this logic.
8192 * We temporarily dropped the "-l" switch for fsck invocations,
8193 since they collide with the flock() logic above. util-linux
8194 upstream has been changed already to avoid this conflict,
8195 and we will re-add "-l" as soon as util-linux with this
8196 change has been released.
8198 * The dependency on libattr has been removed. Since a long
8199 time, the extended attribute calls have moved to glibc, and
8200 libattr is thus unnecessary.
8202 * Virtualization detection works without privileges now. This
8203 means the systemd-detect-virt binary no longer requires
8204 CAP_SYS_PTRACE file capabilities, and our daemons can run
8205 with fewer privileges.
8207 * systemd-networkd now runs under its own "systemd-network"
8208 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
8209 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
8210 loses the ability to write to files owned by root this way.
8212 * Similarly, systemd-resolved now runs under its own
8213 "systemd-resolve" user with no capabilities remaining.
8215 * Similarly, systemd-bus-proxyd now runs under its own
8216 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
8218 * systemd-networkd gained support for setting up "veth"
8219 virtual Ethernet devices for container connectivity, as well
8220 as GRE and VTI tunnels.
8222 * systemd-networkd will no longer automatically attempt to
8223 manually load kernel modules necessary for certain tunnel
8224 transports. Instead, it is assumed the kernel loads them
8225 automatically when required. This only works correctly on
8226 very new kernels. On older kernels, please consider adding
8227 the kernel modules to /etc/modules-load.d/ as a work-around.
8229 * The resolv.conf file systemd-resolved generates has been
8230 moved to /run/systemd/resolve/. If you have a symlink from
8231 /etc/resolv.conf, it might be necessary to correct it.
8233 * Two new service settings, ProtectHome= and ProtectSystem=,
8234 have been added. When enabled, they will make the user data
8235 (such as /home) inaccessible or read-only and the system
8236 (such as /usr) read-only, for specific services. This allows
8237 very light-weight per-service sandboxing to avoid
8238 modifications of user data or system files from
8239 services. These two new switches have been enabled for all
8240 of systemd's long-running services, where appropriate.
8242 * Socket units gained new SocketUser= and SocketGroup=
8243 settings to set the owner user and group of AF_UNIX sockets
8244 and FIFOs in the file system.
8246 * Socket units gained a new RemoveOnStop= setting. If enabled,
8247 all FIFOS and sockets in the file system will be removed
8248 when the specific socket unit is stopped.
8250 * Socket units gained a new Symlinks= setting. It takes a list
8251 of symlinks to create to file system sockets or FIFOs
8252 created by the specific Unix sockets. This is useful to
8253 manage symlinks to socket nodes with the same lifecycle as
8256 * The /dev/log socket and /dev/initctl FIFO have been moved to
8257 /run, and have been replaced by symlinks. This allows
8258 connecting to these facilities even if PrivateDevices=yes is
8259 used for a service (which makes /dev/log itself unavailable,
8260 but /run is left). This also has the benefit of ensuring
8261 that /dev only contains device nodes, directories and
8262 symlinks, and nothing else.
8264 * sd-daemon gained two new calls sd_pid_notify() and
8265 sd_pid_notifyf(). They are similar to sd_notify() and
8266 sd_notifyf(), but allow overriding of the source PID of
8267 notification messages if permissions permit this. This is
8268 useful to send notify messages on behalf of a different
8269 process (for example, the parent process). The
8270 systemd-notify tool has been updated to make use of this
8271 when sending messages (so that notification messages now
8272 originate from the shell script invoking systemd-notify and
8273 not the systemd-notify process itself. This should minimize
8274 a race where systemd fails to associate notification
8275 messages to services when the originating process already
8278 * A new "on-abnormal" setting for Restart= has been added. If
8279 set, it will result in automatic restarts on all "abnormal"
8280 reasons for a process to exit, which includes unclean
8281 signals, core dumps, timeouts and watchdog timeouts, but
8282 does not include clean and unclean exit codes or clean
8283 signals. Restart=on-abnormal is an alternative for
8284 Restart=on-failure for services that shall be able to
8285 terminate and avoid restarts on certain errors, by
8286 indicating so with an unclean exit code. Restart=on-failure
8287 or Restart=on-abnormal is now the recommended setting for
8288 all long-running services.
8290 * If the InaccessibleDirectories= service setting points to a
8291 mount point (or if there are any submounts contained within
8292 it), it is now attempted to completely unmount it, to make
8293 the file systems truly unavailable for the respective
8296 * The ReadOnlyDirectories= service setting and
8297 systemd-nspawn's --read-only parameter are now recursively
8298 applied to all submounts, too.
8300 * Mount units may now be created transiently via the bus APIs.
8302 * The support for SysV and LSB init scripts has been removed
8303 from the systemd daemon itself. Instead, it is now
8304 implemented as a generator that creates native systemd units
8305 from these scripts when needed. This enables us to remove a
8306 substantial amount of legacy code from PID 1, following the
8307 fact that many distributions only ship a very small number
8308 of LSB/SysV init scripts nowadays.
8310 * Privileged Xen (dom0) domains are not considered
8311 virtualization anymore by the virtualization detection
8312 logic. After all, they generally have unrestricted access to
8313 the hardware and usually are used to manage the unprivileged
8316 * systemd-tmpfiles gained a new "C" line type, for copying
8317 files or entire directories.
8319 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
8320 lines. So far, they have been non-globbing versions of the
8321 latter, and have thus been redundant. In future, it is
8322 recommended to only use "z". "m" has hence been removed
8323 from the documentation, even though it stays supported.
8325 * A tmpfiles snippet to recreate the most basic structure in
8326 /var has been added. This is enough to create the /var/run →
8327 /run symlink and create a couple of structural
8328 directories. This allows systems to boot up with an empty or
8329 volatile /var. Of course, while with this change, the core OS
8330 now is capable with dealing with a volatile /var, not all
8331 user services are ready for it. However, we hope that sooner
8332 or later, many service daemons will be changed upstream so
8333 that they are able to automatically create their necessary
8334 directories in /var at boot, should they be missing. This is
8335 the first step to allow state-less systems that only require
8336 the vendor image for /usr to boot.
8338 * systemd-nspawn has gained a new --tmpfs= switch to mount an
8339 empty tmpfs instance to a specific directory. This is
8340 particularly useful for making use of the automatic
8341 reconstruction of /var (see above), by passing --tmpfs=/var.
8343 * Access modes specified in tmpfiles snippets may now be
8344 prefixed with "~", which indicates that they shall be masked
8345 by whether the existing file or directory is currently
8346 writable, readable or executable at all. Also, if specified,
8347 the sgid/suid/sticky bits will be masked for all
8350 * A new passive target unit "network-pre.target" has been
8351 added which is useful for services that shall run before any
8352 network is configured, for example firewall scripts.
8354 * The "floppy" group that previously owned the /dev/fd*
8355 devices is no longer used. The "disk" group is now used
8356 instead. Distributions should probably deprecate usage of
8359 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
8360 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
8361 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
8362 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
8363 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
8364 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
8367 — Berlin, 2014-06-11
8371 * A new "systemd-timesyncd" daemon has been added for
8372 synchronizing the system clock across the network. It
8373 implements an SNTP client. In contrast to NTP
8374 implementations such as chrony or the NTP reference server,
8375 this only implements a client side, and does not bother with
8376 the full NTP complexity, focusing only on querying time from
8377 one remote server and synchronizing the local clock to
8378 it. Unless you intend to serve NTP to networked clients or
8379 want to connect to local hardware clocks, this simple NTP
8380 client should be more than appropriate for most
8381 installations. The daemon runs with minimal privileges, and
8382 has been hooked up with networkd to only operate when
8383 network connectivity is available. The daemon saves the
8384 current clock to disk every time a new NTP sync has been
8385 acquired, and uses this to possibly correct the system clock
8386 early at bootup, in order to accommodate for systems that
8387 lack an RTC such as the Raspberry Pi and embedded devices,
8388 and to make sure that time monotonically progresses on these
8389 systems, even if it is not always correct. To make use of
8390 this daemon, a new system user and group "systemd-timesync"
8391 needs to be created on installation of systemd.
8393 * The queue "seqnum" interface of libudev has been disabled, as
8394 it was generally incompatible with device namespacing as
8395 sequence numbers of devices go "missing" if the devices are
8396 part of a different namespace.
8398 * "systemctl list-timers" and "systemctl list-sockets" gained
8399 a --recursive switch for showing units of these types also
8400 for all local containers, similar in style to the already
8401 supported --recursive switch for "systemctl list-units".
8403 * A new RebootArgument= setting has been added for service
8404 units, which may be used to specify a kernel reboot argument
8405 to use when triggering reboots with StartLimitAction=.
8407 * A new FailureAction= setting has been added for service
8408 units which may be used to specify an operation to trigger
8409 when a service fails. This works similarly to
8410 StartLimitAction=, but unlike it, controls what is done
8411 immediately rather than only after several attempts to
8412 restart the service in question.
8414 * hostnamed got updated to also expose the kernel name,
8415 release, and version on the bus. This is useful for
8416 executing commands like hostnamectl with the -H switch.
8417 systemd-analyze makes use of this to properly display
8418 details when running non-locally.
8420 * The bootchart tool can now show cgroup information in the
8421 graphs it generates.
8423 * The CFS CPU quota cgroup attribute is now exposed for
8424 services. The new CPUQuota= switch has been added for this
8425 which takes a percentage value. Setting this will have the
8426 result that a service may never get more CPU time than the
8427 specified percentage, even if the machine is otherwise idle.
8429 * systemd-networkd learned IPIP and SIT tunnel support.
8431 * LSB init scripts exposing a dependency on $network will now
8432 get a dependency on network-online.target rather than simply
8433 network.target. This should bring LSB handling closer to
8434 what it was on SysV systems.
8436 * A new fsck.repair= kernel option has been added to control
8437 how fsck shall deal with unclean file systems at boot.
8439 * The (.ini) configuration file parser will now silently ignore
8440 sections whose names begin with "X-". This may be used to maintain
8441 application-specific extension sections in unit files.
8443 * machined gained a new API to query the IP addresses of
8444 registered containers. "machinectl status" has been updated
8445 to show these addresses in its output.
8447 * A new call sd_uid_get_display() has been added to the
8448 sd-login APIs for querying the "primary" session of a
8449 user. The "primary" session of the user is elected from the
8450 user's sessions and generally a graphical session is
8451 preferred over a text one.
8453 * A minimal systemd-resolved daemon has been added. It
8454 currently simply acts as a companion to systemd-networkd and
8455 manages resolv.conf based on per-interface DNS
8456 configuration, possibly supplied via DHCP. In the long run
8457 we hope to extend this into a local DNSSEC enabled DNS and
8460 * The systemd-networkd-wait-online tool is now enabled by
8461 default. It will delay network-online.target until a network
8462 connection has been configured. The tool primarily integrates
8463 with networkd, but will also make a best effort to make sense
8464 of network configuration performed in some other way.
8466 * Two new service options StartupCPUShares= and
8467 StartupBlockIOWeight= have been added that work similarly to
8468 CPUShares= and BlockIOWeight= however only apply during
8469 system startup. This is useful to prioritize certain services
8470 differently during bootup than during normal runtime.
8472 * hostnamed has been changed to prefer the statically
8473 configured hostname in /etc/hostname (unless set to
8474 'localhost' or empty) over any dynamic one supplied by
8475 dhcp. With this change, the rules for picking the hostname
8476 match more closely the rules of other configuration settings
8477 where the local administrator's configuration in /etc always
8478 overrides any other settings.
8480 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
8481 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
8482 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
8483 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
8484 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
8485 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
8486 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
8487 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
8488 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
8489 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
8490 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
8491 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
8492 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
8493 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
8494 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
8495 Lindskog, WaLyong Cho, Will Woods, Zbigniew
8498 — Beijing, 2014-05-28
8502 * When restoring the screen brightness at boot, stay away from
8503 the darkest setting or from the lowest 5% of the available
8504 range, depending on which is the larger value of both. This
8505 should effectively protect the user from rebooting into a
8506 black screen, should the brightness have been set to minimum
8509 * sd-login gained a new sd_machine_get_class() call to
8510 determine the class ("vm" or "container") of a machine
8511 registered with machined.
8513 * sd-login gained new calls
8514 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
8515 to query the identity of the peer of a local AF_UNIX
8516 connection. They operate similarly to their sd_pid_get_xyz()
8519 * PID 1 will now maintain a system-wide system state engine
8520 with the states "starting", "running", "degraded",
8521 "maintenance", "stopping". These states are bound to system
8522 startup, normal runtime, runtime with at least one failed
8523 service, rescue/emergency mode and system shutdown. This
8524 state is shown in the "systemctl status" output when no unit
8525 name is passed. It is useful to determine system state, in
8526 particularly when doing so for many systems or containers at
8529 * A new command "list-machines" has been added to "systemctl"
8530 that lists all local OS containers and shows their system
8531 state (see above), if systemd runs inside of them.
8533 * systemctl gained a new "-r" switch to recursively enumerate
8534 units on all local containers, when used with the
8535 "list-unit" command (which is the default one that is
8536 executed when no parameters are specified).
8538 * The GPT automatic partition discovery logic will now honour
8539 two GPT partition flags: one may be set on a partition to
8540 cause it to be mounted read-only, and the other may be set
8541 on a partition to ignore it during automatic discovery.
8543 * Two new GPT type UUIDs have been added for automatic root
8544 partition discovery, for 32-bit and 64-bit ARM. This is not
8545 particularly useful for discovering the root directory on
8546 these architectures during bare-metal boots (since UEFI is
8547 not common there), but still very useful to allow booting of
8548 ARM disk images in nspawn with the -i option.
8550 * MAC addresses of interfaces created with nspawn's
8551 --network-interface= switch will now be generated from the
8552 machine name, and thus be stable between multiple invocations
8555 * logind will now automatically remove all IPC objects owned
8556 by a user if she or he fully logs out. This makes sure that
8557 users who are logged out cannot continue to consume IPC
8558 resources. This covers SysV memory, semaphores and message
8559 queues as well as POSIX shared memory and message
8560 queues. Traditionally, SysV and POSIX IPC had no lifecycle
8561 limits. With this functionality, that is corrected. This may
8562 be turned off by using the RemoveIPC= switch of logind.conf.
8564 * The systemd-machine-id-setup and tmpfiles tools gained a
8565 --root= switch to operate on a specific root directory,
8568 * journald can now forward logged messages to the TTYs of all
8569 logged in users ("wall"). This is the default for all
8570 emergency messages now.
8572 * A new tool systemd-journal-remote has been added to stream
8573 journal log messages across the network.
8575 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
8576 controller trees are mounted into it. Note that the
8577 directories mounted beneath it are not read-only. This is a
8578 security measure and is particularly useful because glibc
8579 actually includes a search logic to pick any tmpfs it can
8580 find to implement shm_open() if /dev/shm is not available
8581 (which it might very well be in namespaced setups).
8583 * machinectl gained a new "poweroff" command to cleanly power
8584 down a local OS container.
8586 * The PrivateDevices= unit file setting will now also drop the
8587 CAP_MKNOD capability from the capability bound set, and
8588 imply DevicePolicy=closed.
8590 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
8591 comprehensively on all long-running systemd services where
8592 this is appropriate.
8594 * systemd-udevd will now run in a disassociated mount
8595 namespace. To mount directories from udev rules, make sure to
8596 pull in mount units via SYSTEMD_WANTS properties.
8598 * The kdbus support gained support for uploading policy into
8599 the kernel. sd-bus gained support for creating "monitoring"
8600 connections that can eavesdrop into all bus communication
8601 for debugging purposes.
8603 * Timestamps may now be specified in seconds since the UNIX
8604 epoch Jan 1st, 1970 by specifying "@" followed by the value
8607 * Native tcpwrap support in systemd has been removed. tcpwrap
8608 is old code, not really maintained anymore and has serious
8609 shortcomings, and better options such as firewalls
8610 exist. For setups that require tcpwrap usage, please
8611 consider invoking your socket-activated service via tcpd,
8612 like on traditional inetd.
8614 * A new system.conf configuration option
8615 DefaultTimerAccuracySec= has been added that controls the
8616 default AccuracySec= setting of .timer units.
8618 * Timer units gained a new WakeSystem= switch. If enabled,
8619 timers configured this way will cause the system to resume
8620 from system suspend (if the system supports that, which most
8623 * Timer units gained a new Persistent= switch. If enabled,
8624 timers configured this way will save to disk when they have
8625 been last triggered. This information is then used on next
8626 reboot to possible execute overdue timer events, that
8627 could not take place because the system was powered off.
8628 This enables simple anacron-like behaviour for timer units.
8630 * systemctl's "list-timers" will now also list the time a
8631 timer unit was last triggered in addition to the next time
8632 it will be triggered.
8634 * systemd-networkd will now assign predictable IPv4LL
8635 addresses to its local interfaces.
8637 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
8638 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
8639 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
8640 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
8641 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
8642 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
8643 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
8644 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
8647 — Berlin, 2014-03-25
8651 * A new unit file setting RestrictAddressFamilies= has been
8652 added to restrict which socket address families unit
8653 processes gain access to. This takes address family names
8654 like "AF_INET" or "AF_UNIX", and is useful to minimize the
8655 attack surface of services via exotic protocol stacks. This
8656 is built on seccomp system call filters.
8658 * Two new unit file settings RuntimeDirectory= and
8659 RuntimeDirectoryMode= have been added that may be used to
8660 manage a per-daemon runtime directories below /run. This is
8661 an alternative for setting up directory permissions with
8662 tmpfiles snippets, and has the advantage that the runtime
8663 directory's lifetime is bound to the daemon runtime and that
8664 the daemon starts up with an empty directory each time. This
8665 is particularly useful when writing services that drop
8666 privileges using the User= or Group= setting.
8668 * The DeviceAllow= unit setting now supports globbing for
8669 matching against device group names.
8671 * The systemd configuration file system.conf gained new
8672 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
8673 DefaultMemoryAccounting= to globally turn on/off accounting
8674 for specific resources (cgroups) for all units. These
8675 settings may still be overridden individually in each unit
8678 * systemd-gpt-auto-generator is now able to discover /srv and
8679 root partitions in addition to /home and swap partitions. It
8680 also supports LUKS-encrypted partitions now. With this in
8681 place, automatic discovery of partitions to mount following
8682 the Discoverable Partitions Specification
8683 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
8684 is now a lot more complete. This allows booting without
8685 /etc/fstab and without root= on the kernel command line on
8686 systems prepared appropriately.
8688 * systemd-nspawn gained a new --image= switch which allows
8689 booting up disk images and Linux installations on any block
8690 device that follow the Discoverable Partitions Specification
8691 (see above). This means that installations made with
8692 appropriately updated installers may now be started and
8693 deployed using container managers, completely
8694 unmodified. (We hope that libvirt-lxc will add support for
8695 this feature soon, too.)
8697 * systemd-nspawn gained a new --network-macvlan= setting to
8698 set up a private macvlan interface for the
8699 container. Similarly, systemd-networkd gained a new
8700 Kind=macvlan setting in .netdev files.
8702 * systemd-networkd now supports configuring local addresses
8705 * A new tool systemd-network-wait-online has been added to
8706 synchronously wait for network connectivity using
8709 * The sd-bus.h bus API gained a new sd_bus_track object for
8710 tracking the lifecycle of bus peers. Note that sd-bus.h is
8711 still not a public API though (unless you specify
8712 --enable-kdbus on the configure command line, which however
8713 voids your warranty and you get no API stability guarantee).
8715 * The $XDG_RUNTIME_DIR runtime directories for each user are
8716 now individual tmpfs instances, which has the benefit of
8717 introducing separate pools for each user, with individual
8718 size limits, and thus making sure that unprivileged clients
8719 can no longer negatively impact the system or other users by
8720 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
8721 RuntimeDirectorySize= has been introduced that allows
8722 controlling the default size limit for all users. It
8723 defaults to 10% of the available physical memory. This is no
8724 replacement for quotas on tmpfs though (which the kernel
8725 still does not support), as /dev/shm and /tmp are still
8726 shared resources used by both the system and unprivileged
8729 * logind will now automatically turn off automatic suspending
8730 on laptop lid close when more than one display is
8731 connected. This was previously expected to be implemented
8732 individually in desktop environments (such as GNOME),
8733 however has been added to logind now, in order to fix a
8734 boot-time race where a desktop environment might not have
8735 been started yet and thus not been able to take an inhibitor
8736 lock at the time where logind already suspends the system
8737 due to a closed lid.
8739 * logind will now wait at least 30s after each system
8740 suspend/resume cycle, and 3min after system boot before
8741 suspending the system due to a closed laptop lid. This
8742 should give USB docking stations and similar enough time to
8743 be probed and configured after system resume and boot in
8744 order to then act as suspend blocker.
8746 * systemd-run gained a new --property= setting which allows
8747 initialization of resource control properties (and others)
8748 for the created scope or service unit. Example: "systemd-run
8749 --property=BlockIOWeight=10 updatedb" may be used to run
8750 updatedb at a low block IO scheduling weight.
8752 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
8753 now also work in --scope mode.
8755 * When systemd is compiled with kdbus support, basic support
8756 for enforced policies is now in place. (Note that enabling
8757 kdbus still voids your warranty and no API compatibility
8760 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
8761 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
8762 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
8763 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
8764 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
8765 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
8766 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
8767 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
8768 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
8769 Zbigniew Jędrzejewski-Szmek
8771 — Berlin, 2014-03-12
8775 * systemd will now relabel /dev after loading the SMACK policy
8776 according to SMACK rules.
8778 * A new unit file option AppArmorProfile= has been added to
8779 set the AppArmor profile for the processes of a unit.
8781 * A new condition check ConditionArchitecture= has been added
8782 to conditionalize units based on the system architecture, as
8783 reported by uname()'s "machine" field.
8785 * systemd-networkd now supports matching on the system
8786 virtualization, architecture, kernel command line, hostname
8789 * logind is now a lot more aggressive when suspending the
8790 machine due to a closed laptop lid. Instead of acting only
8791 on the lid close action, it will continuously watch the lid
8792 status and act on it. This is useful for laptops where the
8793 power button is on the outside of the chassis so that it can
8794 be reached without opening the lid (such as the Lenovo
8795 Yoga). On those machines, logind will now immediately
8796 re-suspend the machine if the power button has been
8797 accidentally pressed while the laptop was suspended and in a
8798 backpack or similar.
8800 * logind will now watch SW_DOCK switches and inhibit reaction
8801 to the lid switch if it is pressed. This means that logind
8802 will not suspend the machine anymore if the lid is closed
8803 and the system is docked, if the laptop supports SW_DOCK
8804 notifications via the input layer. Note that ACPI docking
8805 stations do not generate this currently. Also note that this
8806 logic is usually not fully sufficient and Desktop
8807 Environments should take a lid switch inhibitor lock when an
8808 external display is connected, as systemd will not watch
8811 * nspawn will now make use of the devices cgroup controller by
8812 default, and only permit creation of and access to the usual
8813 API device nodes like /dev/null or /dev/random, as well as
8814 access to (but not creation of) the pty devices.
8816 * We will now ship a default .network file for
8817 systemd-networkd that automatically configures DHCP for
8818 network interfaces created by nspawn's --network-veth or
8819 --network-bridge= switches.
8821 * systemd will now understand the usual M, K, G, T suffixes
8822 according to SI conventions (i.e. to the base 1000) when
8823 referring to throughput and hardware metrics. It will stay
8824 with IEC conventions (i.e. to the base 1024) for software
8825 metrics, according to what is customary according to
8826 Wikipedia. We explicitly document which base applies for
8827 each configuration option.
8829 * The DeviceAllow= setting in unit files now supports a syntax to
8830 allow-list an entire group of devices node majors at once, based on
8831 the /proc/devices listing. For example, with the string "char-pts",
8832 it is now possible to allow-list all current and future pseudo-TTYs
8835 * sd-event learned a new "post" event source. Event sources of
8836 this type are triggered by the dispatching of any event
8837 source of a type that is not "post". This is useful for
8838 implementing clean-up and check event sources that are
8839 triggered by other work being done in the program.
8841 * systemd-networkd is no longer statically enabled, but uses
8842 the usual [Install] sections so that it can be
8843 enabled/disabled using systemctl. It still is enabled by
8846 * When creating a veth interface pair with systemd-nspawn, the
8847 host side will now be prefixed with "vb-" if
8848 --network-bridge= is used, and with "ve-" if --network-veth
8849 is used. This way, it is easy to distinguish these cases on
8850 the host, for example to apply different configuration to
8851 them with systemd-networkd.
8853 * The compatibility libraries for libsystemd-journal.so,
8854 libsystem-id128.so, libsystemd-login.so and
8855 libsystemd-daemon.so do not make use of IFUNC
8856 anymore. Instead, we now build libsystemd.so multiple times
8857 under these alternative names. This means that the footprint
8858 is drastically increased, but given that these are
8859 transitional compatibility libraries, this should not matter
8860 much. This change has been made necessary to support the ARM
8861 platform for these compatibility libraries, as the ARM
8862 toolchain is not really at the same level as the toolchain
8863 for other architectures like x86 and does not support
8864 IFUNC. Please make sure to use --enable-compat-libs only
8865 during a transitional period!
8867 * The .include syntax has been deprecated and is not documented
8868 anymore. Drop-in files in .d directories should be used instead.
8870 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
8871 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
8872 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
8873 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
8874 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
8875 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
8876 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
8877 Zbigniew Jędrzejewski-Szmek
8879 — Berlin, 2014-02-24
8883 * A new component "systemd-networkd" has been added that can
8884 be used to configure local network interfaces statically or
8885 via DHCP. It is capable of bringing up bridges, VLANs, and
8886 bonding. Currently, no hook-ups for interactive network
8887 configuration are provided. Use this for your initrd,
8888 container, embedded, or server setup if you need a simple,
8889 yet powerful, network configuration solution. This
8890 configuration subsystem is quite nifty, as it allows wildcard
8891 hotplug matching in interfaces. For example, with a single
8892 configuration snippet, you can configure that all Ethernet
8893 interfaces showing up are automatically added to a bridge,
8894 or similar. It supports link-sensing and more.
8896 * A new tool "systemd-socket-proxyd" has been added which can
8897 act as a bidirectional proxy for TCP sockets. This is
8898 useful for adding socket activation support to services that
8899 do not actually support socket activation, including virtual
8900 machines and the like.
8902 * Add a new tool to save/restore rfkill state on
8905 * Save/restore state of keyboard backlights in addition to
8906 display backlights on shutdown/boot.
8908 * udev learned a new SECLABEL{} construct to label device
8909 nodes with a specific security label when they appear. For
8910 now, only SECLABEL{selinux} is supported, but the syntax is
8911 prepared for additional security frameworks.
8913 * udev gained a new scheme to configure link-level attributes
8914 from files in /etc/systemd/network/*.link. These files can
8915 match against MAC address, device path, driver name and type,
8916 and will apply attributes like the naming policy, link speed,
8917 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
8918 address assignment policy (randomized, ...).
8920 * The configuration of network interface naming rules for
8921 "permanent interface names" has changed: a new NamePolicy=
8922 setting in the [Link] section of .link files determines the
8923 priority of possible naming schemes (onboard, slot, MAC,
8924 path). The default value of this setting is determined by
8925 /usr/lib/net/links/99-default.link. Old
8926 80-net-name-slot.rules udev configuration file has been
8927 removed, so local configuration overriding this file should
8928 be adapted to override 99-default.link instead.
8930 * When the User= switch is used in a unit file, also
8931 initialize $SHELL= based on the user database entry.
8933 * systemd no longer depends on libdbus. All communication is
8934 now done with sd-bus, systemd's low-level bus library
8937 * kdbus support has been added to PID 1 itself. When kdbus is
8938 enabled, this causes PID 1 to set up the system bus and
8939 enable support for a new ".busname" unit type that
8940 encapsulates bus name activation on kdbus. It works a little
8941 bit like ".socket" units, except for bus names. A new
8942 generator has been added that converts classic dbus1 service
8943 activation files automatically into native systemd .busname
8946 * sd-bus: add a light-weight vtable implementation that allows
8947 defining objects on the bus with a simple static const
8948 vtable array of its methods, signals and properties.
8950 * systemd will not generate or install static dbus
8951 introspection data anymore to /usr/share/dbus-1/interfaces,
8952 as the precise format of these files is unclear, and
8953 nothing makes use of it.
8955 * A proxy daemon is now provided to proxy clients connecting
8956 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
8957 compatibility with classic D-Bus.
8959 * A bus driver implementation has been added that supports the
8960 classic D-Bus bus driver calls on kdbus, also for
8961 compatibility purposes.
8963 * A new API "sd-event.h" has been added that implements a
8964 minimal event loop API built around epoll. It provides a
8965 couple of features that direct epoll usage is lacking:
8966 prioritization of events, scales to large numbers of timer
8967 events, per-event timer slack (accuracy), system-wide
8968 coalescing of timer events, exit handlers, watchdog
8969 supervision support using systemd's sd_notify() API, child
8972 * A new API "sd-rntl.h" has been added that provides an API
8973 around the route netlink interface of the kernel, similar in
8974 style to "sd-bus.h".
8976 * A new API "sd-dhcp-client.h" has been added that provides a
8977 small DHCPv4 client-side implementation. This is used by
8980 * There is a new kernel command line option
8981 "systemd.restore_state=0|1". When set to "0", none of the
8982 systemd tools will restore saved runtime state to hardware
8983 devices. More specifically, the rfkill and backlight states
8986 * The FsckPassNo= compatibility option in mount/service units
8987 has been removed. The fstab generator will now add the
8988 necessary dependencies automatically, and does not require
8989 PID1's support for that anymore.
8991 * journalctl gained a new switch, --list-boots, that lists
8992 recent boots with their times and boot IDs.
8994 * The various tools like systemctl, loginctl, timedatectl,
8995 busctl, systemd-run, ... have gained a new switch "-M" to
8996 connect to a specific, local OS container (as direct
8997 connection, without requiring SSH). This works on any
8998 container that is registered with machined, such as those
8999 created by libvirt-lxc or nspawn.
9001 * systemd-run and systemd-analyze also gained support for "-H"
9002 to connect to remote hosts via SSH. This is particularly
9003 useful for systemd-run because it enables queuing of jobs
9004 onto remote systems.
9006 * machinectl gained a new command "login" to open a getty
9007 login in any local container. This works with any container
9008 that is registered with machined (such as those created by
9009 libvirt-lxc or nspawn), and which runs systemd inside.
9011 * machinectl gained a new "reboot" command that may be used to
9012 trigger a reboot on a specific container that is registered
9013 with machined. This works on any container that runs an init
9014 system of some kind.
9016 * systemctl gained a new "list-timers" command to print a nice
9017 listing of installed timer units with the times they elapse
9020 * Alternative reboot() parameters may now be specified on the
9021 "systemctl reboot" command line and are passed to the
9022 reboot() system call.
9024 * systemctl gained a new --job-mode= switch to configure the
9025 mode to queue a job with. This is a more generic version of
9026 --fail, --irreversible, and --ignore-dependencies, which are
9027 still available but not advertised anymore.
9029 * /etc/systemd/system.conf gained new settings to configure
9030 various default timeouts of units, as well as the default
9031 start limit interval and burst. These may still be overridden
9034 * PID1 will now export on the bus profile data of the security
9035 policy upload process (such as the SELinux policy upload to
9038 * journald: when forwarding logs to the console, include
9039 timestamps (following the setting in
9040 /sys/module/printk/parameters/time).
9042 * OnCalendar= in timer units now understands the special
9043 strings "yearly" and "annually". (Both are equivalent)
9045 * The accuracy of timer units is now configurable with the new
9046 AccuracySec= setting. It defaults to 1min.
9048 * A new dependency type JoinsNamespaceOf= has been added that
9049 allows running two services within the same /tmp and network
9050 namespace, if PrivateNetwork= or PrivateTmp= are used.
9052 * A new command "cat" has been added to systemctl. It outputs
9053 the original unit file of a unit, and concatenates the
9054 contents of additional "drop-in" unit file snippets, so that
9055 the full configuration is shown.
9057 * systemctl now supports globbing on the various "list-xyz"
9058 commands, like "list-units" or "list-sockets", as well as on
9059 those commands which take multiple unit names.
9061 * journalctl's --unit= switch gained support for globbing.
9063 * All systemd daemons now make use of the watchdog logic so
9064 that systemd automatically notices when they hang.
9066 * If the $container_ttys environment variable is set,
9067 getty-generator will automatically spawn a getty for each
9068 listed tty. This is useful for container managers to request
9069 login gettys to be spawned on as many ttys as needed.
9071 * %h, %s, %U specifier support is not available anymore when
9072 used in unit files for PID 1. This is because NSS calls are
9073 not safe from PID 1. They stay available for --user
9074 instances of systemd, and as special case for the root user.
9076 * loginctl gained a new "--no-legend" switch to turn off output
9079 * The "sd-login.h" API gained three new calls:
9080 sd_session_is_remote(), sd_session_get_remote_user(),
9081 sd_session_get_remote_host() to query information about
9084 * The udev hardware database now also carries vendor/product
9085 information of SDIO devices.
9087 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
9088 determine whether watchdog notifications are requested by
9091 * Socket-activated per-connection services now include a
9092 short description of the connection parameters in the
9095 * tmpfiles gained a new "--boot" option. When this is not used,
9096 only lines where the command character is not suffixed with
9097 "!" are executed. When this option is specified, those
9098 options are executed too. This partitions tmpfiles
9099 directives into those that can be safely executed at any
9100 time, and those which should be run only at boot (for
9101 example, a line that creates /run/nologin).
9103 * A new API "sd-resolve.h" has been added which provides a simple
9104 asynchronous wrapper around glibc NSS hostname resolution
9105 calls, such as getaddrinfo(). In contrast to glibc's
9106 getaddrinfo_a(), it does not use signals. In contrast to most
9107 other asynchronous name resolution libraries, this one does
9108 not reimplement DNS, but reuses NSS, so that alternate
9109 hostname resolution systems continue to work, such as mDNS,
9110 LDAP, etc. This API is based on libasyncns, but it has been
9111 cleaned up for inclusion in systemd.
9113 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
9114 "sd-daemon.h" are no longer found in individual libraries
9115 libsystemd-journal.so, libsystemd-login.so,
9116 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
9117 merged them into a single library, libsystemd.so, which
9118 provides all symbols. The reason for this is cyclic
9119 dependencies, as these libraries tend to use each other's
9120 symbols. So far, we have managed to workaround that by linking
9121 a copy of a good part of our code into each of these
9122 libraries again and again, which, however, makes certain
9123 things hard to do, like sharing static variables. Also, it
9124 substantially increases footprint. With this change, there
9125 is only one library for the basic APIs systemd
9126 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
9127 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
9128 library as well, however are subject to the --enable-kdbus
9129 switch (see below). Note that "sd-dhcp-client.h" is not part
9130 of this library (this is because it only consumes, never
9131 provides, services of/to other APIs). To make the transition
9132 easy from the separate libraries to the unified one, we
9133 provide the --enable-compat-libs compile-time switch which
9134 will generate stub libraries that are compatible with the
9135 old ones but redirect all calls to the new one.
9137 * All of the kdbus logic and the new APIs "sd-bus.h",
9138 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
9139 and "sd-utf8.h" are compile-time optional via the
9140 "--enable-kdbus" switch, and they are not compiled in by
9141 default. To make use of kdbus, you have to explicitly enable
9142 the switch. Note however, that neither the kernel nor the
9143 userspace API for all of this is considered stable yet. We
9144 want to maintain the freedom to still change the APIs for
9145 now. By specifying this build-time switch, you acknowledge
9146 that you are aware of the instability of the current
9149 * Also, note that while kdbus is pretty much complete,
9150 it lacks one thing: proper policy support. This means you
9151 can build a fully working system with all features; however,
9152 it will be highly insecure. Policy support will be added in
9153 one of the next releases, at the same time that we will
9154 declare the APIs stable.
9156 * When the kernel command line argument "kdbus" is specified,
9157 systemd will automatically load the kdbus.ko kernel module. At
9158 this stage of development, it is only useful for testing kdbus
9159 and should not be used in production. Note: if "--enable-kdbus"
9160 is specified, and the kdbus.ko kernel module is available, and
9161 "kdbus" is added to the kernel command line, the entire system
9162 runs with kdbus instead of dbus-daemon, with the above mentioned
9163 problem of missing the system policy enforcement. Also a future
9164 version of kdbus.ko or a newer systemd will not be compatible with
9165 each other, and will unlikely be able to boot the machine if only
9166 one of them is updated.
9168 * systemctl gained a new "import-environment" command which
9169 uploads the caller's environment (or parts thereof) into the
9170 service manager so that it is inherited by services started
9171 by the manager. This is useful to upload variables like
9172 $DISPLAY into the user service manager.
9174 * A new PrivateDevices= switch has been added to service units
9175 which allows running a service with a namespaced /dev
9176 directory that does not contain any device nodes for
9177 physical devices. More specifically, it only includes devices
9178 such as /dev/null, /dev/urandom, and /dev/zero which are API
9181 * logind has been extended to support behaviour like VT
9182 switching on seats that do not support a VT. This makes
9183 multi-session available on seats that are not the first seat
9184 (seat0), and on systems where kernel support for VTs has
9185 been disabled at compile-time.
9187 * If a process holds a delay lock for system sleep or shutdown
9188 and fails to release it in time, we will now log its
9189 identity. This makes it easier to identify processes that
9190 cause slow suspends or power-offs.
9192 * When parsing /etc/crypttab, support for a new key-slot=
9193 option as supported by Debian is added. It allows indicating
9194 which LUKS slot to use on disk, speeding up key loading.
9196 * The sd_journal_sendv() API call has been checked and
9197 officially declared to be async-signal-safe so that it may
9198 be invoked from signal handlers for logging purposes.
9200 * Boot-time status output is now enabled automatically after a
9201 short timeout if boot does not progress, in order to give
9202 the user an indication what she or he is waiting for.
9204 * The boot-time output has been improved to show how much time
9205 remains until jobs expire.
9207 * The KillMode= switch in service units gained a new possible
9208 value "mixed". If set, and the unit is shut down, then the
9209 initial SIGTERM signal is sent only to the main daemon
9210 process, while the following SIGKILL signal is sent to
9211 all remaining processes of the service.
9213 * When a scope unit is registered, a new property "Controller"
9214 may be set. If set to a valid bus name, systemd will send a
9215 RequestStop() signal to this name when it would like to shut
9216 down the scope. This may be used to hook manager logic into
9217 the shutdown logic of scope units. Also, scope units may now
9218 be put in a special "abandoned" state, in which case the
9219 manager process which created them takes no further
9220 responsibilities for it.
9222 * When reading unit files, systemd will now verify
9223 the access mode of these files, and warn about certain
9224 suspicious combinations. This has been added to make it
9225 easier to track down packaging bugs where unit files are
9226 marked executable or world-writable.
9228 * systemd-nspawn gained a new "--setenv=" switch to set
9229 container-wide environment variables. The similar option in
9230 systemd-activate was renamed from "--environment=" to
9231 "--setenv=" for consistency.
9233 * systemd-nspawn has been updated to create a new kdbus domain
9234 for each container that is invoked, thus allowing each
9235 container to have its own set of system and user buses,
9236 independent of the host.
9238 * systemd-nspawn gained a new --drop-capability= switch to run
9239 the container with less capabilities than the default. Both
9240 --drop-capability= and --capability= now take the special
9241 string "all" for dropping or keeping all capabilities.
9243 * systemd-nspawn gained new switches for executing containers
9244 with specific SELinux labels set.
9246 * systemd-nspawn gained a new --quiet switch to not generate
9247 any additional output but the container's own console
9250 * systemd-nspawn gained a new --share-system switch to run a
9251 container without PID namespacing enabled.
9253 * systemd-nspawn gained a new --register= switch to control
9254 whether the container is registered with systemd-machined or
9255 not. This is useful for containers that do not run full
9256 OS images, but only specific apps.
9258 * systemd-nspawn gained a new --keep-unit which may be used
9259 when invoked as the only program from a service unit, and
9260 results in registration of the unit service itself in
9261 systemd-machined, instead of a newly opened scope unit.
9263 * systemd-nspawn gained a new --network-interface= switch for
9264 moving arbitrary interfaces to the container. The new
9265 --network-veth switch creates a virtual Ethernet connection
9266 between host and container. The new --network-bridge=
9267 switch then allows assigning the host side of this virtual
9268 Ethernet connection to a bridge device.
9270 * systemd-nspawn gained a new --personality= switch for
9271 setting the kernel personality for the container. This is
9272 useful when running a 32-bit container on a 64-bit host. A
9273 similar option Personality= is now also available for service
9276 * logind will now also track a "Desktop" identifier for each
9277 session which encodes the desktop environment of it. This is
9278 useful for desktop environments that want to identify
9279 multiple running sessions of itself easily.
9281 * A new SELinuxContext= setting for service units has been
9282 added that allows setting a specific SELinux execution
9283 context for a service.
9285 * Most systemd client tools will now honour $SYSTEMD_LESS for
9286 settings of the "less" pager. By default, these tools will
9287 override $LESS to allow certain operations to work, such as
9288 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
9289 influence this logic.
9291 * systemd's "seccomp" hook-up has been changed to make use of
9292 the libseccomp library instead of using its own
9293 implementation. This has benefits for portability among
9296 * For usage together with SystemCallFilter=, a new
9297 SystemCallErrorNumber= setting has been introduced that
9298 allows configuration of a system error number to be returned
9299 on filtered system calls, instead of immediately killing the
9300 process. Also, SystemCallArchitectures= has been added to
9301 limit access to system calls of a particular architecture
9302 (in order to turn off support for unused secondary
9303 architectures). There is also a global
9304 SystemCallArchitectures= setting in system.conf now to turn
9305 off support for non-native system calls system-wide.
9307 * systemd requires a kernel with a working name_to_handle_at(),
9308 please see the kernel config requirements in the README file.
9310 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
9311 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
9312 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
9313 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
9314 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
9315 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
9316 Elia Pinto, Florian Weimer, George McCollister, Goffredo
9317 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
9318 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
9319 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
9320 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
9321 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
9322 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
9323 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
9324 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
9325 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
9326 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
9327 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
9328 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
9329 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
9330 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
9331 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
9332 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
9333 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
9335 — Berlin, 2014-02-20
9339 * logind has gained support for facilitating privileged input
9340 and drm device access for unprivileged clients. This work is
9341 useful to allow Wayland display servers (and similar
9342 programs, such as kmscon) to run under the user's ID and
9343 access input and drm devices which are normally
9344 protected. When this is used (and the kernel is new enough)
9345 logind will "mute" IO on the file descriptors passed to
9346 Wayland as long as it is in the background and "unmute" it
9347 if it returns into the foreground. This allows secure
9348 session switching without allowing background sessions to
9349 eavesdrop on input and display data. This also introduces
9350 session switching support if VT support is turned off in the
9351 kernel, and on seats that are not seat0.
9353 * A new kernel command line option luks.options= is understood
9354 now which allows specifying LUKS options for usage for LUKS
9355 encrypted partitions specified with luks.uuid=.
9357 * tmpfiles.d(5) snippets may now use specifier expansion in
9358 path names. More specifically %m, %b, %H, %v, are now
9359 replaced by the local machine id, boot id, hostname, and
9360 kernel version number.
9362 * A new tmpfiles.d(5) command "m" has been introduced which
9363 may be used to change the owner/group/access mode of a file
9364 or directory if it exists, but do nothing if it does not.
9366 * This release removes high-level support for the
9367 MemorySoftLimit= cgroup setting. The underlying kernel
9368 cgroup attribute memory.soft_limit= is currently badly
9369 designed and likely to be removed from the kernel API in its
9370 current form, hence we should not expose it for now.
9372 * The memory.use_hierarchy cgroup attribute is now enabled for
9373 all cgroups systemd creates in the memory cgroup
9374 hierarchy. This option is likely to be come the built-in
9375 default in the kernel anyway, and the non-hierarchical mode
9376 never made much sense in the intrinsically hierarchical
9379 * A new field _SYSTEMD_SLICE= is logged along with all journal
9380 messages containing the slice a message was generated
9381 from. This is useful to allow easy per-customer filtering of
9382 logs among other things.
9384 * systemd-journald will no longer adjust the group of journal
9385 files it creates to the "systemd-journal" group. Instead we
9386 rely on the journal directory to be owned by the
9387 "systemd-journal" group, and its setgid bit set, so that the
9388 kernel file system layer will automatically enforce that
9389 journal files inherit this group assignment. The reason for
9390 this change is that we cannot allow NSS look-ups from
9391 journald which would be necessary to resolve
9392 "systemd-journal" to a numeric GID, because this might
9393 create deadlocks if NSS involves synchronous queries to
9394 other daemons (such as nscd, or sssd) which in turn are
9395 logging clients of journald and might block on it, which
9396 would then dead lock. A tmpfiles.d(5) snippet included in
9397 systemd will make sure the setgid bit and group are
9398 properly set on the journal directory if it exists on every
9399 boot. However, we recommend adjusting it manually after
9400 upgrades too (or from RPM scriptlets), so that the change is
9401 not delayed until next reboot.
9403 * Backlight and random seed files in /var/lib/ have moved into
9404 the /var/lib/systemd/ directory, in order to centralize all
9405 systemd generated files in one directory.
9407 * Boot time performance measurements (as displayed by
9408 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
9409 performance information if that's available to determine how
9410 much time BIOS and boot loader initialization required. With
9411 a sufficiently new BIOS you hence no longer need to boot
9412 with Gummiboot to get access to such information.
9414 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
9415 Cristian Rodríguez, Dave Reisner, David Herrmann, David
9416 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
9417 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
9418 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
9419 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
9420 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9422 — Berlin, 2013-10-02
9426 * The Restart= option for services now understands a new
9427 on-watchdog setting, which will restart the service
9428 automatically if the service stops sending out watchdog keep
9429 alive messages (as configured with WatchdogSec=).
9431 * The getty generator (which is responsible for bringing up a
9432 getty on configured serial consoles) will no longer only
9433 start a getty on the primary kernel console but on all
9434 others, too. This makes the order in which console= is
9435 specified on the kernel command line less important.
9437 * libsystemd-logind gained a new sd_session_get_vt() call to
9438 retrieve the VT number of a session.
9440 * If the option "tries=0" is set for an entry of /etc/crypttab
9441 its passphrase is queried indefinitely instead of any
9442 maximum number of tries.
9444 * If a service with a configure PID file terminates its PID
9445 file will now be removed automatically if it still exists
9446 afterwards. This should put an end to stale PID files.
9448 * systemd-run will now also take relative binary path names
9449 for execution and no longer insists on absolute paths.
9451 * InaccessibleDirectories= and ReadOnlyDirectories= now take
9452 paths that are optionally prefixed with "-" to indicate that
9453 it should not be considered a failure if they do not exist.
9455 * journalctl -o (and similar commands) now understands a new
9456 output mode "short-precise", it is similar to "short" but
9457 shows timestamps with usec accuracy.
9459 * The option "discard" (as known from Debian) is now
9460 synonymous to "allow-discards" in /etc/crypttab. In fact,
9461 "discard" is preferred now (since it is easier to remember
9464 * Some licensing clean-ups were made, so that more code is now
9465 LGPL-2.1 licensed than before.
9467 * A minimal tool to save/restore the display backlight
9468 brightness across reboots has been added. It will store the
9469 backlight setting as late as possible at shutdown, and
9470 restore it as early as possible during reboot.
9472 * A logic to automatically discover and enable home and swap
9473 partitions on GPT disks has been added. With this in place
9474 /etc/fstab becomes optional for many setups as systemd can
9475 discover certain partitions located on the root disk
9476 automatically. Home partitions are recognized under their
9477 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
9478 partitions are recognized under their GPT type ID
9479 0657fd6da4ab43c484e50933c84b4f4f.
9481 * systemd will no longer pass any environment from the kernel
9482 or initrd to system services. If you want to set an
9483 environment for all services, do so via the kernel command
9484 line systemd.setenv= assignment.
9486 * The systemd-sysctl tool no longer natively reads the file
9487 /etc/sysctl.conf. If desired, the file should be symlinked
9488 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
9489 legacy support by a symlink rather than built-in code, it
9490 also makes the otherwise hidden order of application of the
9491 different files visible. (Note that this partly reverts to a
9492 pre-198 application order of sysctl knobs!)
9494 * The "systemctl set-log-level" and "systemctl dump" commands
9495 have been moved to systemd-analyze.
9497 * systemd-run learned the new --remain-after-exit switch,
9498 which causes the scope unit not to be cleaned up
9499 automatically after the process terminated.
9501 * tmpfiles learned a new --exclude-prefix= switch to exclude
9502 certain paths from operation.
9504 * journald will now automatically flush all messages to disk
9505 as soon as a message at the log level CRIT, ALERT or EMERG
9508 Contributions from: Andrew Cook, Brandon Philips, Christian
9509 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
9510 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
9511 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
9512 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
9513 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
9514 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
9515 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
9516 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
9517 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
9518 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
9519 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
9520 William Giokas, Zbigniew Jędrzejewski-Szmek
9522 — Berlin, 2013-09-13
9526 * The documentation has been updated to cover the various new
9527 concepts introduced with 205.
9529 * Unit files now understand the new %v specifier which
9530 resolves to the kernel version string as returned by "uname
9533 * systemctl now supports filtering the unit list output by
9534 load state, active state and sub state, using the new
9537 * "systemctl status" will now show the results of the
9538 condition checks (like ConditionPathExists= and similar) of
9539 the last start attempts of the unit. They are also logged to
9542 * "journalctl -b" may now be used to look for boot output of a
9543 specific boot. Try "journalctl -b -1" for the previous boot,
9544 but the syntax is substantially more powerful.
9546 * "journalctl --show-cursor" has been added which prints the
9547 cursor string the last shown log line. This may then be used
9548 with the new "journalctl --after-cursor=" switch to continue
9549 browsing logs from that point on.
9551 * "journalctl --force" may now be used to force regeneration
9554 * Creation of "dead" device nodes has been moved from udev
9555 into kmod and tmpfiles. Previously, udev would read the kmod
9556 databases to pre-generate dead device nodes based on meta
9557 information contained in kernel modules, so that these would
9558 be auto-loaded on access rather then at boot. As this
9559 does not really have much to do with the exposing actual
9560 kernel devices to userspace this has always been slightly
9561 alien in the udev codebase. Following the new scheme kmod
9562 will now generate a runtime snippet for tmpfiles from the
9563 module meta information and it now is tmpfiles' job to the
9564 create the nodes. This also allows overriding access and
9565 other parameters for the nodes using the usual tmpfiles
9566 facilities. As side effect this allows us to remove the
9567 CAP_SYS_MKNOD capability bit from udevd entirely.
9569 * logind's device ACLs may now be applied to these "dead"
9570 devices nodes too, thus finally allowing managed access to
9571 devices such as /dev/snd/sequencer without loading the
9572 backing module right-away.
9574 * A new RPM macro has been added that may be used to apply
9575 tmpfiles configuration during package installation.
9577 * systemd-detect-virt and ConditionVirtualization= now can
9578 detect User-Mode-Linux machines (UML).
9580 * journald will now implicitly log the effective capabilities
9581 set of processes in the message metadata.
9583 * systemd-cryptsetup has gained support for TrueCrypt volumes.
9585 * The initrd interface has been simplified (more specifically,
9586 support for passing performance data via environment
9587 variables and fsck results via files in /run has been
9588 removed). These features were non-essential, and are
9589 nowadays available in a much nicer way by having systemd in
9590 the initrd serialize its state and have the hosts systemd
9591 deserialize it again.
9593 * The udev "keymap" data files and tools to apply keyboard
9594 specific mappings of scan to key codes, and force-release
9595 scan code lists have been entirely replaced by a udev
9596 "keyboard" builtin and a hwdb data file.
9598 * systemd will now honour the kernel's "quiet" command line
9599 argument also during late shutdown, resulting in a
9600 completely silent shutdown when used.
9602 * There's now an option to control the SO_REUSEPORT socket
9603 option in .socket units.
9605 * Instance units will now automatically get a per-template
9606 subslice of system.slice unless something else is explicitly
9607 configured. For example, instances of sshd@.service will now
9608 implicitly be placed in system-sshd.slice rather than
9609 system.slice as before.
9611 * Test coverage support may now be enabled at build time.
9613 Contributions from: Dave Reisner, Frederic Crozat, Harald
9614 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
9615 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
9616 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
9617 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
9618 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
9619 Giokas, Zbigniew Jędrzejewski-Szmek
9621 — Berlin, 2013-07-23
9625 * Two new unit types have been introduced:
9627 Scope units are very similar to service units, however, are
9628 created out of pre-existing processes — instead of PID 1
9629 forking off the processes. By using scope units it is
9630 possible for system services and applications to group their
9631 own child processes (worker processes) in a powerful way
9632 which then maybe used to organize them, or kill them
9633 together, or apply resource limits on them.
9635 Slice units may be used to partition system resources in an
9636 hierarchical fashion and then assign other units to them. By
9637 default there are now three slices: system.slice (for all
9638 system services), user.slice (for all user sessions),
9639 machine.slice (for VMs and containers).
9641 Slices and scopes have been introduced primarily in
9642 context of the work to move cgroup handling to a
9643 single-writer scheme, where only PID 1
9644 creates/removes/manages cgroups.
9646 * There's a new concept of "transient" units. In contrast to
9647 normal units these units are created via an API at runtime,
9648 not from configuration from disk. More specifically this
9649 means it is now possible to run arbitrary programs as
9650 independent services, with all execution parameters passed
9651 in via bus APIs rather than read from disk. Transient units
9652 make systemd substantially more dynamic then it ever was,
9653 and useful as a general batch manager.
9655 * logind has been updated to make use of scope and slice units
9656 for managing user sessions. As a user logs in he will get
9657 his own private slice unit, to which all sessions are added
9658 as scope units. We also added support for automatically
9659 adding an instance of user@.service for the user into the
9660 slice. Effectively logind will no longer create cgroup
9661 hierarchies on its own now, it will defer entirely to PID 1
9662 for this by means of scope, service and slice units. Since
9663 user sessions this way become entities managed by PID 1
9664 the output of "systemctl" is now a lot more comprehensive.
9666 * A new mini-daemon "systemd-machined" has been added which
9667 may be used by virtualization managers to register local
9668 VMs/containers. nspawn has been updated accordingly, and
9669 libvirt will be updated shortly. machined will collect a bit
9670 of meta information about the VMs/containers, and assign
9671 them their own scope unit (see above). The collected
9672 meta-data is then made available via the "machinectl" tool,
9673 and exposed in "ps" and similar tools. machined/machinectl
9674 is compile-time optional.
9676 * As discussed earlier, the low-level cgroup configuration
9677 options ControlGroup=, ControlGroupModify=,
9678 ControlGroupPersistent=, ControlGroupAttribute= have been
9679 removed. Please use high-level attribute settings instead as
9680 well as slice units.
9682 * A new bus call SetUnitProperties() has been added to alter
9683 various runtime parameters of a unit. This is primarily
9684 useful to alter cgroup parameters dynamically in a nice way,
9685 but will be extended later on to make more properties
9686 modifiable at runtime. systemctl gained a new set-properties
9687 command that wraps this call.
9689 * A new tool "systemd-run" has been added which can be used to
9690 run arbitrary command lines as transient services or scopes,
9691 while configuring a number of settings via the command
9692 line. This tool is currently very basic, however already
9693 very useful. We plan to extend this tool to even allow
9694 queuing of execution jobs with time triggers from the
9695 command line, similar in fashion to "at".
9697 * nspawn will now inform the user explicitly that kernels with
9698 audit enabled break containers, and suggest the user to turn
9701 * Support for detecting the IMA and AppArmor security
9702 frameworks with ConditionSecurity= has been added.
9704 * journalctl gained a new "-k" switch for showing only kernel
9705 messages, mimicking dmesg output; in addition to "--user"
9706 and "--system" switches for showing only user's own logs
9709 * systemd-delta can now show information about drop-in
9710 snippets extending unit files.
9712 * libsystemd-bus has been substantially updated but is still
9713 not available as public API.
9715 * systemd will now look for the "debug" argument on the kernel
9716 command line and enable debug logging, similar to what
9717 "systemd.log_level=debug" already did before.
9719 * "systemctl set-default", "systemctl get-default" has been
9720 added to configure the default.target symlink, which
9721 controls what to boot into by default.
9723 * "systemctl set-log-level" has been added as a convenient
9724 way to raise and lower systemd logging threshold.
9726 * "systemd-analyze plot" will now show the time the various
9727 generators needed for execution, as well as information
9728 about the unit file loading.
9730 * libsystemd-journal gained a new sd_journal_open_files() call
9731 for opening specific journal files. journactl also gained a
9732 new switch to expose this new functionality. Previously we
9733 only supported opening all files from a directory, or all
9734 files from the system, as opening individual files only is
9735 racy due to journal file rotation.
9737 * systemd gained the new DefaultEnvironment= setting in
9738 /etc/systemd/system.conf to set environment variables for
9741 * If a privileged process logs a journal message with the
9742 OBJECT_PID= field set, then journald will automatically
9743 augment this with additional OBJECT_UID=, OBJECT_GID=,
9744 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
9745 system services want to log events about specific client
9746 processes. journactl/systemctl has been updated to make use
9747 of this information if all log messages regarding a specific
9750 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
9751 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
9752 Reisner, David Coppa, David King, David Strauss, Eelco
9753 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
9754 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
9755 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
9756 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
9757 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
9758 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
9759 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
9760 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
9761 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
9762 Łukasz Stelmach, 장동준
9766 * The Python bindings gained some minimal support for the APIs
9767 exposed by libsystemd-logind.
9769 * ConditionSecurity= gained support for detecting SMACK. Since
9770 this condition already supports SELinux and AppArmor we only
9771 miss IMA for this. Patches welcome!
9773 Contributions from: Karol Lewandowski, Lennart Poettering,
9774 Zbigniew Jędrzejewski-Szmek
9778 * systemd-nspawn will now create /etc/resolv.conf if
9779 necessary, before bind-mounting the host's file onto it.
9781 * systemd-nspawn will now store meta information about a
9782 container on the container's cgroup as extended attribute
9783 fields, including the root directory.
9785 * The cgroup hierarchy has been reworked in many ways. All
9786 objects any of the components systemd creates in the cgroup
9787 tree are now suffixed. More specifically, user sessions are
9788 now placed in cgroups suffixed with ".session", users in
9789 cgroups suffixed with ".user", and nspawn containers in
9790 cgroups suffixed with ".nspawn". Furthermore, all cgroup
9791 names are now escaped in a simple scheme to avoid collision
9792 of userspace object names with kernel filenames. This work
9793 is preparation for making these objects relocatable in the
9794 cgroup tree, in order to allow easy resource partitioning of
9795 these objects without causing naming conflicts.
9797 * systemctl list-dependencies gained the new switches
9798 --plain, --reverse, --after and --before.
9800 * systemd-inhibit now shows the process name of processes that
9801 have taken an inhibitor lock.
9803 * nss-myhostname will now also resolve "localhost"
9804 implicitly. This makes /etc/hosts an optional file and
9805 nicely handles that on IPv6 ::1 maps to both "localhost" and
9808 * libsystemd-logind.so gained a new call
9809 sd_get_machine_names() to enumerate running containers and
9810 VMs (currently only supported by very new libvirt and
9811 nspawn). sd_login_monitor can now be used to watch
9812 VMs/containers coming and going.
9814 * .include is not allowed recursively anymore, and only in
9815 unit files. Usually it is better to use drop-in snippets in
9816 .d/*.conf anyway, as introduced with systemd 198.
9818 * systemd-analyze gained a new "critical-chain" command that
9819 determines the slowest chain of units run during system
9820 boot-up. It is very useful for tracking down where
9821 optimizing boot time is the most beneficial.
9823 * systemd will no longer allow manipulating service paths in
9824 the name=systemd:/system cgroup tree using ControlGroup= in
9825 units. (But is still fine with it in all other dirs.)
9827 * There's a new systemd-nspawn@.service service file that may
9828 be used to easily run nspawn containers as system
9829 services. With the container's root directory in
9830 /var/lib/container/foobar it is now sufficient to run
9831 "systemctl start systemd-nspawn@foobar.service" to boot it.
9833 * systemd-cgls gained a new parameter "--machine" to list only
9834 the processes within a certain container.
9836 * ConditionSecurity= now can check for "apparmor". We still
9837 are lacking checks for SMACK and IMA for this condition
9838 check though. Patches welcome!
9840 * A new configuration file /etc/systemd/sleep.conf has been
9841 added that may be used to configure which kernel operation
9842 systemd is supposed to execute when "suspend", "hibernate"
9843 or "hybrid-sleep" is requested. This makes the new kernel
9844 "freeze" state accessible to the user.
9846 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
9847 the passed argument if applicable.
9849 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
9850 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
9851 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
9852 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
9853 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
9854 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
9855 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
9860 * The output of 'systemctl list-jobs' got some polishing. The
9861 '--type=' argument may now be passed more than once. A new
9862 command 'systemctl list-sockets' has been added which shows
9863 a list of kernel sockets systemd is listening on with the
9864 socket units they belong to, plus the units these socket
9867 * The experimental libsystemd-bus library got substantial
9868 updates to work in conjunction with the (also experimental)
9869 kdbus kernel project. It works well enough to exchange
9870 messages with some sophistication. Note that kdbus is not
9871 ready yet, and the library is mostly an elaborate test case
9872 for now, and not installable.
9874 * systemd gained a new unit 'systemd-static-nodes.service'
9875 that generates static device nodes earlier during boot, and
9876 can run in conjunction with udev.
9878 * libsystemd-login gained a new call sd_pid_get_user_unit()
9879 to retrieve the user systemd unit a process is running
9880 in. This is useful for systems where systemd is used as
9883 * systemd-nspawn now places all containers in the new /machine
9884 top-level cgroup directory in the name=systemd
9885 hierarchy. libvirt will soon do the same, so that we get a
9886 uniform separation of /system, /user and /machine for system
9887 services, user processes and containers/virtual
9888 machines. This new cgroup hierarchy is also useful to stick
9889 stable names to specific container instances, which can be
9890 recognized later this way (this name may be controlled
9891 via systemd-nspawn's new -M switch). libsystemd-login also
9892 gained a new call sd_pid_get_machine_name() to retrieve the
9893 name of the container/VM a specific process belongs to.
9895 * bootchart can now store its data in the journal.
9897 * libsystemd-journal gained a new call
9898 sd_journal_add_conjunction() for AND expressions to the
9899 matching logic. This can be used to express more complex
9900 logical expressions.
9902 * journactl can now take multiple --unit= and --user-unit=
9905 * The cryptsetup logic now understands the "luks.key=" kernel
9906 command line switch for specifying a file to read the
9907 decryption key from. Also, if a configured key file is not
9908 found the tool will now automatically fall back to prompting
9911 * Python systemd.journal module was updated to wrap recently
9912 added functions from libsystemd-journal. The interface was
9913 changed to bring the low level interface in s.j._Reader
9914 closer to the C API, and the high level interface in
9915 s.j.Reader was updated to wrap and convert all data about
9918 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
9919 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
9920 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
9921 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
9922 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
9923 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9927 * journalctl --update-catalog now understands a new --root=
9928 option to operate on catalogs found in a different root
9931 * During shutdown after systemd has terminated all running
9932 services a final killing loop kills all remaining left-over
9933 processes. We will now print the name of these processes
9934 when we send SIGKILL to them, since this usually indicates a
9937 * If /etc/crypttab refers to password files stored on
9938 configured mount points automatic dependencies will now be
9939 generated to ensure the specific mount is established first
9940 before the key file is attempted to be read.
9942 * 'systemctl status' will now show information about the
9943 network sockets a socket unit is listening on.
9945 * 'systemctl status' will also shown information about any
9946 drop-in configuration file for units. (Drop-In configuration
9947 files in this context are files such as
9948 /etc/systemd/systemd/foobar.service.d/*.conf)
9950 * systemd-cgtop now optionally shows summed up CPU times of
9951 cgroups. Press '%' while running cgtop to switch between
9952 percentage and absolute mode. This is useful to determine
9953 which cgroups use up the most CPU time over the entire
9954 runtime of the system. systemd-cgtop has also been updated
9955 to be 'pipeable' for processing with further shell tools.
9957 * 'hostnamectl set-hostname' will now allow setting of FQDN
9960 * The formatting and parsing of time span values has been
9961 changed. The parser now understands fractional expressions
9962 such as "5.5h". The formatter will now output fractional
9963 expressions for all time spans under 1min, i.e. "5.123456s"
9964 rather than "5s 123ms 456us". For time spans under 1s
9965 millisecond values are shown, for those under 1ms
9966 microsecond values are shown. This should greatly improve
9967 all time-related output of systemd.
9969 * libsystemd-login and libsystemd-journal gained new
9970 functions for querying the poll() events mask and poll()
9971 timeout value for integration into arbitrary event
9974 * localectl gained the ability to list available X11 keymaps
9975 (models, layouts, variants, options).
9977 * 'systemd-analyze dot' gained the ability to filter for
9978 specific units via shell-style globs, to create smaller,
9979 more useful graphs. I.e. it is now possible to create simple
9980 graphs of all the dependencies between only target units, or
9981 of all units that Avahi has dependencies with.
9983 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
9984 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
9985 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
9986 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
9987 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
9988 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
9989 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
9993 * The boot-time readahead implementation for rotating media
9994 will now read the read-ahead data in multiple passes which
9995 consist of all read requests made in equidistant time
9996 intervals. This means instead of strictly reading read-ahead
9997 data in its physical order on disk we now try to find a
9998 middle ground between physical and access time order.
10000 * /etc/os-release files gained a new BUILD_ID= field for usage
10001 on operating systems that provide continuous builds of OS
10004 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
10005 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
10006 William Douglas, Zbigniew Jędrzejewski-Szmek
10010 * systemd-python gained an API exposing libsystemd-daemon.
10012 * The SMACK setup logic gained support for uploading CIPSO
10015 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
10016 ReadOnlyDirectories= and InaccessibleDirectories= has
10017 changed. The private /tmp and /var/tmp directories are now
10018 shared by all processes of a service (which means
10019 ExecStartPre= may now leave data in /tmp that ExecStart= of
10020 the same service can still access). When a service is
10021 stopped its temporary directories are immediately deleted
10022 (normal clean-up with tmpfiles is still done in addition to
10025 * By default, systemd will now set a couple of sysctl
10026 variables in the kernel: the safe sysrq options are turned
10027 on, IP route verification is turned on, and source routing
10028 disabled. The recently added hardlink and softlink
10029 protection of the kernel is turned on. These settings should
10030 be reasonably safe, and good defaults for all new systems.
10032 * The predictable network naming logic may now be turned off
10033 with a new kernel command line switch: net.ifnames=0.
10035 * A new libsystemd-bus module has been added that implements a
10036 pretty complete D-Bus client library. For details see:
10038 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
10040 * journald will now explicitly flush the journal files to disk
10041 at the latest 5min after each write. The file will then also
10042 be marked offline until the next write. This should increase
10043 reliability in case of a crash. The synchronization delay
10044 can be configured via SyncIntervalSec= in journald.conf.
10046 * There's a new remote-fs-setup.target unit that can be used
10047 to pull in specific services when at least one remote file
10048 system is to be mounted.
10050 * There are new targets timers.target and paths.target as
10051 canonical targets to pull user timer and path units in
10052 from. This complements sockets.target with a similar
10053 purpose for socket units.
10055 * libudev gained a new call udev_device_set_attribute_value()
10056 to set sysfs attributes of a device.
10058 * The udev daemon now sets the default number of worker
10059 processes executed in parallel based on the number of available
10060 CPUs instead of the amount of available RAM. This is supposed
10061 to provide a more reliable default and limit a too aggressive
10062 parallelism for setups with 1000s of devices connected.
10064 Contributions from: Auke Kok, Colin Walters, Cristian
10065 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
10066 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
10067 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
10068 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
10069 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
10070 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
10071 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
10072 Zbigniew Jędrzejewski-Szmek
10076 * Configuration of unit files may now be extended via drop-in
10077 files without having to edit/override the unit files
10078 themselves. More specifically, if the administrator wants to
10079 change one value for a service file foobar.service he can
10080 now do so by dropping in a configuration snippet into
10081 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
10082 will load all these snippets and apply them on top of the
10083 main unit configuration file, possibly extending or
10084 overriding its settings. Using these drop-in snippets is
10085 generally nicer than the two earlier options for changing
10086 unit files locally: copying the files from
10087 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
10088 them there; or creating a new file in /etc/systemd/system/
10089 that incorporates the original one via ".include". Drop-in
10090 snippets into these .d/ directories can be placed in any
10091 directory systemd looks for units in, and the usual
10092 overriding semantics between /usr/lib, /etc and /run apply
10095 * Most unit file settings which take lists of items can now be
10096 reset by assigning the empty string to them. For example,
10097 normally, settings such as Environment=FOO=BAR append a new
10098 environment variable assignment to the environment block,
10099 each time they are used. By assigning Environment= the empty
10100 string the environment block can be reset to empty. This is
10101 particularly useful with the .d/*.conf drop-in snippets
10102 mentioned above, since this adds the ability to reset list
10103 settings from vendor unit files via these drop-ins.
10105 * systemctl gained a new "list-dependencies" command for
10106 listing the dependencies of a unit recursively.
10108 * Inhibitors are now honored and listed by "systemctl
10109 suspend", "systemctl poweroff" (and similar) too, not only
10110 GNOME. These commands will also list active sessions by
10113 * Resource limits (as exposed by the various control group
10114 controllers) can now be controlled dynamically at runtime
10115 for all units. More specifically, you can now use a command
10116 like "systemctl set-cgroup-attr foobar.service cpu.shares
10117 2000" to alter the CPU shares a specific service gets. These
10118 settings are stored persistently on disk, and thus allow the
10119 administrator to easily adjust the resource usage of
10120 services with a few simple commands. This dynamic resource
10121 management logic is also available to other programs via the
10122 bus. Almost any kernel cgroup attribute and controller is
10125 * systemd-vconsole-setup will now copy all font settings to
10126 all allocated VTs, where it previously applied them only to
10129 * libsystemd-login gained the new sd_session_get_tty() API
10132 * This release drops support for a few legacy or
10133 distribution-specific LSB facility names when parsing init
10134 scripts: $x-display-manager, $mail-transfer-agent,
10135 $mail-transport-agent, $mail-transfer-agent, $smtp,
10136 $null. Also, the mail-transfer-agent.target unit backing
10137 this has been removed. Distributions which want to retain
10138 compatibility with this should carry the burden for
10139 supporting this themselves and patch support for these back
10140 in, if they really need to. Also, the facilities $syslog and
10141 $local_fs are now ignored, since systemd does not support
10142 early-boot LSB init scripts anymore, and these facilities
10143 are implied anyway for normal services. syslog.target has
10146 * There are new bus calls on PID1's Manager object for
10147 cancelling jobs, and removing snapshot units. Previously,
10148 both calls were only available on the Job and Snapshot
10149 objects themselves.
10151 * systemd-journal-gatewayd gained SSL support.
10153 * The various "environment" files, such as /etc/locale.conf
10154 now support continuation lines with a backslash ("\") as
10155 last character in the line, similarly in style (but different)
10156 to how this is supported in shells.
10158 * For normal user processes the _SYSTEMD_USER_UNIT= field is
10159 now implicitly appended to every log entry logged. systemctl
10160 has been updated to filter by this field when operating on a
10161 user systemd instance.
10163 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
10164 CAP_AUDIT_CONTROL capabilities to the capabilities set for
10165 the container. This makes it easier to boot unmodified
10166 Fedora systems in a container, which however still requires
10167 audit=0 to be passed on the kernel command line. Auditing in
10168 kernel and userspace is unfortunately still too broken in
10169 context of containers, hence we recommend compiling it out
10170 of the kernel or using audit=0. Hopefully this will be fixed
10171 one day for good in the kernel.
10173 * nspawn gained the new --bind= and --bind-ro= parameters to
10174 bind mount specific directories from the host into the
10177 * nspawn will now mount its own devpts file system instance
10178 into the container, in order not to leak pty devices from
10179 the host into the container.
10181 * systemd will now read the firmware boot time performance
10182 information from the EFI variables, if the used boot loader
10183 supports this, and takes it into account for boot performance
10184 analysis via "systemd-analyze". This is currently supported
10185 only in conjunction with Gummiboot, but could be supported
10186 by other boot loaders too. For details see:
10188 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
10190 * A new generator has been added that automatically mounts the
10191 EFI System Partition (ESP) to /boot, if that directory
10192 exists, is empty, and no other file system has been
10193 configured to be mounted there.
10195 * logind will now send out PrepareForSleep(false) out
10196 unconditionally, after coming back from suspend. This may be
10197 used by applications as asynchronous notification for
10198 system resume events.
10200 * "systemctl unlock-sessions" has been added, that allows
10201 unlocking the screens of all user sessions at once, similar
10202 to how "systemctl lock-sessions" already locked all users
10203 sessions. This is backed by a new D-Bus call UnlockSessions().
10205 * "loginctl seat-status" will now show the master device of a
10206 seat. (i.e. the device of a seat that needs to be around for
10207 the seat to be considered available, usually the graphics
10210 * tmpfiles gained a new "X" line type, that allows
10211 configuration of files and directories (with wildcards) that
10212 shall be excluded from automatic cleanup ("aging").
10214 * udev default rules set the device node permissions now only
10215 at "add" events, and do not change them any longer with a
10216 later "change" event.
10218 * The log messages for lid events and power/sleep keypresses
10219 now carry a message ID.
10221 * We now have a substantially larger unit test suite, but this
10222 continues to be work in progress.
10224 * udevadm hwdb gained a new --root= parameter to change the
10225 root directory to operate relative to.
10227 * logind will now issue a background sync() request to the kernel
10228 early at shutdown, so that dirty buffers are flushed to disk early
10229 instead of at the last moment, in order to optimize shutdown
10232 * A new bootctl tool has been added that is an interface for
10233 certain boot loader operations. This is currently a preview
10234 and is likely to be extended into a small mechanism daemon
10235 like timedated, localed, hostnamed, and can be used by
10236 graphical UIs to enumerate available boot options, and
10237 request boot into firmware operations.
10239 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
10240 the rest of the package. It also has been updated to work
10241 correctly in initrds.
10243 * polkit previously has been runtime optional, and is now also
10244 compile time optional via a configure switch.
10246 * systemd-analyze has been reimplemented in C. Also "systemctl
10247 dot" has moved into systemd-analyze.
10249 * "systemctl status" with no further parameters will now print
10250 the status of all active or failed units.
10252 * Operations such as "systemctl start" can now be executed
10253 with a new mode "--irreversible" which may be used to queue
10254 operations that cannot accidentally be reversed by a later
10255 job queuing. This is by default used to make shutdown
10256 requests more robust.
10258 * The Python API of systemd now gained a new module for
10259 reading journal files.
10261 * A new tool kernel-install has been added that can install
10262 kernel images according to the Boot Loader Specification:
10264 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
10266 * Boot time console output has been improved to provide
10267 animated boot time output for hanging jobs.
10269 * A new tool systemd-activate has been added which can be used
10270 to test socket activation with, directly from the command
10271 line. This should make it much easier to test and debug
10272 socket activation in daemons.
10274 * journalctl gained a new "--reverse" (or -r) option to show
10275 journal output in reverse order (i.e. newest line first).
10277 * journalctl gained a new "--pager-end" (or -e) option to jump
10278 to immediately jump to the end of the journal in the
10279 pager. This is only supported in conjunction with "less".
10281 * journalctl gained a new "--user-unit=" option, that works
10282 similarly to "--unit=" but filters for user units rather than
10285 * A number of unit files to ease adoption of systemd in
10286 initrds has been added. This moves some minimal logic from
10287 the various initrd implementations into systemd proper.
10289 * The journal files are now owned by a new group
10290 "systemd-journal", which exists specifically to allow access
10291 to the journal, and nothing else. Previously, we used the
10292 "adm" group for that, which however possibly covers more
10293 than just journal/log file access. This new group is now
10294 already used by systemd-journal-gatewayd to ensure this
10295 daemon gets access to the journal files and as little else
10296 as possible. Note that "make install" will also set FS ACLs
10297 up for /var/log/journal to give "adm" and "wheel" read
10298 access to it, in addition to "systemd-journal" which owns
10299 the journal files. We recommend that packaging scripts also
10300 add read access to "adm" + "wheel" to /var/log/journal, and
10301 all existing/future journal files. To normal users and
10302 administrators little changes, however packagers need to
10303 ensure to create the "systemd-journal" system group at
10304 package installation time.
10306 * The systemd-journal-gatewayd now runs as unprivileged user
10307 systemd-journal-gateway:systemd-journal-gateway. Packaging
10308 scripts need to create these system user/group at
10311 * timedated now exposes a new boolean property CanNTP that
10312 indicates whether a local NTP service is available or not.
10314 * systemd-detect-virt will now also detect xen PVs
10316 * The pstore file system is now mounted by default, if it is
10319 * In addition to the SELinux and IMA policies we will now also
10320 load SMACK policies at early boot.
10322 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
10323 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
10324 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
10325 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
10326 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
10327 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
10328 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
10329 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
10330 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
10331 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
10332 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
10333 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
10334 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
10335 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
10339 * Timer units now support calendar time events in addition to
10340 monotonic time events. That means you can now trigger a unit
10341 based on a calendar time specification such as "Thu,Fri
10342 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
10343 or fifth day of any month of the year 2013, given that it is
10344 a thursday or friday. This brings timer event support
10345 considerably closer to cron's capabilities. For details on
10346 the supported calendar time specification language see
10349 * udev now supports a number of different naming policies for
10350 network interfaces for predictable names, and a combination
10351 of these policies is now the default. Please see this wiki
10352 document for details:
10354 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
10356 * Auke Kok's bootchart implementation has been added to the
10357 systemd tree. It is an optional component that can graph the
10358 boot in quite some detail. It is one of the best bootchart
10359 implementations around and minimal in its code and
10362 * nss-myhostname has been integrated into the systemd source
10363 tree. nss-myhostname guarantees that the local hostname
10364 always stays resolvable via NSS. It has been a weak
10365 requirement of systemd-hostnamed since a long time, and
10366 since its code is actually trivial we decided to just
10367 include it in systemd's source tree. It can be turned off
10368 with a configure switch.
10370 * The read-ahead logic is now capable of properly detecting
10371 whether a btrfs file system is on SSD or rotating media, in
10372 order to optimize the read-ahead scheme. Previously, it was
10373 only capable of detecting this on traditional file systems
10376 * In udev, additional device properties are now read from the
10377 IAB in addition to the OUI database. Also, Bluetooth company
10378 identities are attached to the devices as well.
10380 * In service files %U may be used as specifier that is
10381 replaced by the configured user name of the service.
10383 * nspawn may now be invoked without a controlling TTY. This
10384 makes it suitable for invocation as its own service. This
10385 may be used to set up a simple containerized server system
10386 using only core OS tools.
10388 * systemd and nspawn can now accept socket file descriptors
10389 when they are started for socket activation. This enables
10390 implementation of socket activated nspawn
10391 containers. i.e. think about autospawning an entire OS image
10392 when the first SSH or HTTP connection is received. We expect
10393 that similar functionality will also be added to libvirt-lxc
10396 * journalctl will now suppress ANSI color codes when
10397 presenting log data.
10399 * systemctl will no longer show control group information for
10400 a unit if the control group is empty anyway.
10402 * logind can now automatically suspend/hibernate/shutdown the
10405 * /etc/machine-info and hostnamed now also expose the chassis
10406 type of the system. This can be used to determine whether
10407 the local system is a laptop, desktop, handset or
10408 tablet. This information may either be configured by the
10409 user/vendor or is automatically determined from ACPI and DMI
10410 information if possible.
10412 * A number of polkit actions are now bound together with "imply"
10413 rules. This should simplify creating UIs because many actions
10414 will now authenticate similar ones as well.
10416 * Unit files learnt a new condition ConditionACPower= which
10417 may be used to conditionalize a unit depending on whether an
10418 AC power source is connected or not, of whether the system
10419 is running on battery power.
10421 * systemctl gained a new "is-failed" verb that may be used in
10422 shell scripts and suchlike to check whether a specific unit
10423 is in the "failed" state.
10425 * The EnvironmentFile= setting in unit files now supports file
10426 globbing, and can hence be used to easily read a number of
10427 environment files at once.
10429 * systemd will no longer detect and recognize specific
10430 distributions. All distribution-specific #ifdeffery has been
10431 removed, systemd is now fully generic and
10432 distribution-agnostic. Effectively, not too much is lost as
10433 a lot of the code is still accessible via explicit configure
10434 switches. However, support for some distribution specific
10435 legacy configuration file formats has been dropped. We
10436 recommend distributions to simply adopt the configuration
10437 files everybody else uses now and convert the old
10438 configuration from packaging scripts. Most distributions
10439 already did that. If that's not possible or desirable,
10440 distributions are welcome to forward port the specific
10441 pieces of code locally from the git history.
10443 * When logging a message about a unit systemd will now always
10444 log the unit name in the message meta data.
10446 * localectl will now also discover system locale data that is
10447 not stored in locale archives, but directly unpacked.
10449 * logind will no longer unconditionally use framebuffer
10450 devices as seat masters, i.e. as devices that are required
10451 to be existing before a seat is considered preset. Instead,
10452 it will now look for all devices that are tagged as
10453 "seat-master" in udev. By default, framebuffer devices will
10454 be marked as such, but depending on local systems, other
10455 devices might be marked as well. This may be used to
10456 integrate graphics cards using closed source drivers (such
10457 as NVidia ones) more nicely into logind. Note however, that
10458 we recommend using the open source NVidia drivers instead,
10459 and no udev rules for the closed-source drivers will be
10460 shipped from us upstream.
10462 Contributions from: Adam Williamson, Alessandro Crismani, Auke
10463 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
10464 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
10465 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
10466 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
10467 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
10468 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
10469 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
10470 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
10471 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
10472 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
10473 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
10478 * udev gained support for loading additional device properties
10479 from an indexed database that is keyed by vendor/product IDs
10480 and similar device identifiers. For the beginning this
10481 "hwdb" is populated with data from the well-known PCI and
10482 USB database, but also includes PNP, ACPI and OID data. In
10483 the longer run this indexed database shall grow into
10484 becoming the one central database for non-essential
10485 userspace device metadata. Previously, data from the PCI/USB
10486 database was only attached to select devices, since the
10487 lookup was a relatively expensive operation due to O(n) time
10488 complexity (with n being the number of entries in the
10489 database). Since this is now O(1), we decided to add in this
10490 data for all devices where this is available, by
10491 default. Note that the indexed database needs to be rebuilt
10492 when new data files are installed. To achieve this you need
10493 to update your packaging scripts to invoke "udevadm hwdb
10494 --update" after installation of hwdb data files. For
10495 RPM-based distributions we introduced the new
10496 %udev_hwdb_update macro for this purpose.
10498 * The Journal gained support for the "Message Catalog", an
10499 indexed database to link up additional information with
10500 journal entries. For further details please check:
10502 https://www.freedesktop.org/wiki/Software/systemd/catalog
10504 The indexed message catalog database also needs to be
10505 rebuilt after installation of message catalog files. Use
10506 "journalctl --update-catalog" for this. For RPM-based
10507 distributions we introduced the %journal_catalog_update
10508 macro for this purpose.
10510 * The Python Journal bindings gained support for the standard
10511 Python logging framework.
10513 * The Journal API gained new functions for checking whether
10514 the underlying file system of a journal file is capable of
10515 properly reporting file change notifications, or whether
10516 applications that want to reflect journal changes "live"
10517 need to recheck journal files continuously in appropriate
10520 * It is now possible to set the "age" field for tmpfiles
10521 entries to 0, indicating that files matching this entry
10522 shall always be removed when the directories are cleaned up.
10524 * coredumpctl gained a new "gdb" verb which invokes gdb
10525 right-away on the selected coredump.
10527 * There's now support for "hybrid sleep" on kernels that
10528 support this, in addition to "suspend" and "hibernate". Use
10529 "systemctl hybrid-sleep" to make use of this.
10531 * logind's HandleSuspendKey= setting (and related settings)
10532 now gained support for a new "lock" setting to simply
10533 request the screen lock on all local sessions, instead of
10534 actually executing a suspend or hibernation.
10536 * systemd will now mount the EFI variables file system by
10539 * Socket units now gained support for configuration of the
10540 SMACK security label.
10542 * timedatectl will now output the time of the last and next
10543 daylight saving change.
10545 * We dropped support for various legacy and distro-specific
10546 concepts, such as insserv, early-boot SysV services
10547 (i.e. those for non-standard runlevels such as 'b' or 'S')
10548 or ArchLinux /etc/rc.conf support. We recommend the
10549 distributions who still need support this to either continue
10550 to maintain the necessary patches downstream, or find a
10551 different solution. (Talk to us if you have questions!)
10553 * Various systemd components will now bypass polkit checks for
10554 root and otherwise handle properly if polkit is not found to
10555 be around. This should fix most issues for polkit-less
10556 systems. Quite frankly this should have been this way since
10557 day one. It is absolutely our intention to make systemd work
10558 fine on polkit-less systems, and we consider it a bug if
10559 something does not work as it should if polkit is not around.
10561 * For embedded systems it is now possible to build udev and
10562 systemd without blkid and/or kmod support.
10564 * "systemctl switch-root" is now capable of switching root
10565 more than once. I.e. in addition to transitions from the
10566 initrd to the host OS it is now possible to transition to
10567 further OS images from the host. This is useful to implement
10568 offline updating tools.
10570 * Various other additions have been made to the RPM macros
10571 shipped with systemd. Use %udev_rules_update() after
10572 installing new udev rules files. %_udevhwdbdir,
10573 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
10574 %_sysctldir are now available which resolve to the right
10575 directories for packages to place various data files in.
10577 * journalctl gained the new --full switch (in addition to
10578 --all, to disable ellipsation for long messages.
10580 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
10581 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
10582 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
10583 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
10584 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
10585 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
10586 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
10587 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
10588 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
10592 * journalctl gained new --since= and --until= switches to
10593 filter by time. It also now supports nice filtering for
10594 units via --unit=/-u.
10596 * Type=oneshot services may use ExecReload= and do the
10599 * The journal daemon now supports time-based rotation and
10600 vacuuming, in addition to the usual disk-space based
10603 * The journal will now index the available field values for
10604 each field name. This enables clients to show pretty drop
10605 downs of available match values when filtering. The bash
10606 completion of journalctl has been updated
10607 accordingly. journalctl gained a new switch -F to list all
10608 values a certain field takes in the journal database.
10610 * More service events are now written as structured messages
10611 to the journal, and made recognizable via message IDs.
10613 * The timedated, localed and hostnamed mini-services which
10614 previously only provided support for changing time, locale
10615 and hostname settings from graphical DEs such as GNOME now
10616 also have a minimal (but very useful) text-based client
10617 utility each. This is probably the nicest way to changing
10618 these settings from the command line now, especially since
10619 it lists available options and is fully integrated with bash
10622 * There's now a new tool "systemd-coredumpctl" to list and
10623 extract coredumps from the journal.
10625 * We now install a README each in /var/log/ and
10626 /etc/rc.d/init.d explaining where the system logs and init
10627 scripts went. This hopefully should help folks who go to
10628 that dirs and look into the otherwise now empty void and
10629 scratch their heads.
10631 * When user-services are invoked (by systemd --user) the
10632 $MANAGERPID env var is set to the PID of systemd.
10634 * SIGRTMIN+24 when sent to a --user instance will now result
10635 in immediate termination of systemd.
10637 * gatewayd received numerous feature additions such as a
10638 "follow" mode, for live syncing and filtering.
10640 * browse.html now allows filtering and showing detailed
10641 information on specific entries. Keyboard navigation and
10642 mouse screen support has been added.
10644 * gatewayd/journalctl now supports HTML5/JSON
10645 Server-Sent-Events as output.
10647 * The SysV init script compatibility logic will now
10648 heuristically determine whether a script supports the
10649 "reload" verb, and only then make this available as
10650 "systemctl reload".
10652 * "systemctl status --follow" has been removed, use "journalctl
10655 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
10656 have been removed since they are hardly useful to be
10659 * And I'd like to take the opportunity to specifically mention
10660 Zbigniew for his great contributions. Zbigniew, you rock!
10662 Contributions from: Andrew Eikum, Christian Hesse, Colin
10663 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
10664 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
10665 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
10666 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
10667 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
10668 Jędrzejewski-Szmek, Сковорода Никита Андреевич
10672 * If /etc/vconsole.conf is non-existent or empty we will no
10673 longer load any console font or key map at boot by
10674 default. Instead the kernel defaults will be left
10675 intact. This is definitely the right thing to do, as no
10676 configuration should mean no configuration, and hard-coding
10677 font names that are different on all archs is probably a bad
10678 idea. Also, the kernel default key map and font should be
10679 good enough for most cases anyway, and mostly identical to
10680 the userspace fonts/key maps we previously overloaded them
10681 with. If distributions want to continue to default to a
10682 non-kernel font or key map they should ship a default
10683 /etc/vconsole.conf with the appropriate contents.
10685 Contributions from: Colin Walters, Daniel J Walsh, Dave
10686 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
10687 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
10691 * journalctl gained a new --cursor= switch to show entries
10692 starting from the specified location in the journal.
10694 * We now enforce a size limit on journal entry fields exported
10695 with "-o json" in journalctl. Fields larger than 4K will be
10696 assigned null. This can be turned off with --all.
10698 * An (optional) journal gateway daemon is now available as
10699 "systemd-journal-gatewayd.service". This service provides
10700 access to the journal via HTTP and JSON. This functionality
10701 will be used to implement live log synchronization in both
10702 pull and push modes, but has various other users too, such
10703 as easy log access for debugging of embedded devices. Right
10704 now it is already useful to retrieve the journal via HTTP:
10706 # systemctl start systemd-journal-gatewayd.service
10707 # wget http://localhost:19531/entries
10709 This will download the journal contents in a
10710 /var/log/messages compatible format. The same as JSON:
10712 # curl -H"Accept: application/json" http://localhost:19531/entries
10714 This service is also accessible via a web browser where a
10715 single static HTML5 app is served that uses the JSON logic
10716 to enable the user to do some basic browsing of the
10717 journal. This will be extended later on. Here's an example
10718 screenshot of this app in its current state:
10720 http://0pointer.de/public/journal-gatewayd
10722 Contributions from: Kay Sievers, Lennart Poettering, Robert
10723 Milasan, Tom Gundersen
10727 * The bash completion logic is now available for journalctl
10730 * We do not mount the "cpuset" controller anymore together with
10731 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
10732 started if no parameters are assigned to it. "cpuset" hence
10733 broke code that assumed it could create "cpu" groups and
10736 * journalctl -f will now subscribe to terminal size changes,
10737 and line break accordingly.
10739 Contributions from: Dave Reisner, Kay Sievers, Lennart
10740 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
10744 * nspawn will now create a symlink /etc/localtime in the
10745 container environment, copying the host's timezone
10746 setting. Previously this has been done via a bind mount, but
10747 since symlinks cannot be bind mounted this has now been
10748 changed to create/update the appropriate symlink.
10750 * journalctl -n's line number argument is now optional, and
10751 will default to 10 if omitted.
10753 * journald will now log the maximum size the journal files may
10754 take up on disk. This is particularly useful if the default
10755 built-in logic of determining this parameter from the file
10756 system size is used. Use "systemctl status
10757 systemd-journald.service" to see this information.
10759 * The multi-seat X wrapper tool has been stripped down. As X
10760 is now capable of enumerating graphics devices via udev in a
10761 seat-aware way the wrapper is not strictly necessary
10762 anymore. A stripped down temporary stop-gap is still shipped
10763 until the upstream display managers have been updated to
10764 fully support the new X logic. Expect this wrapper to be
10765 removed entirely in one of the next releases.
10767 * HandleSleepKey= in logind.conf has been split up into
10768 HandleSuspendKey= and HandleHibernateKey=. The old setting
10769 is not available anymore. X11 and the kernel are
10770 distinguishing between these keys and we should too. This
10771 also means the inhibition lock for these keys has been split
10774 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
10775 Poettering, Lukas Nykryn, Václav Pavlín
10779 * Whenever a unit changes state we will now log this to the
10780 journal and show along the unit's own log output in
10781 "systemctl status".
10783 * ConditionPathIsMountPoint= can now properly detect bind
10784 mount points too. (Previously, a bind mount of one file
10785 system to another place in the same file system could not be
10786 detected as mount, since they shared struct stat's st_dev
10789 * We will now mount the cgroup controllers cpu, cpuacct,
10790 cpuset and the controllers net_cls, net_prio together by
10793 * nspawn containers will now have a virtualized boot
10794 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
10795 over with a randomized ID at container initialization). This
10796 has the effect of making "journalctl -b" do the right thing
10799 * The JSON output journal serialization has been updated not
10800 to generate "endless" list objects anymore, but rather one
10801 JSON object per line. This is more in line how most JSON
10802 parsers expect JSON objects. The new output mode
10803 "json-pretty" has been added to provide similar output, but
10804 neatly aligned for readability by humans.
10806 * We dropped all explicit sync() invocations in the shutdown
10807 code. The kernel does this implicitly anyway in the kernel
10808 reboot() syscall. halt(8)'s -n option is now a compatibility
10811 * We now support virtualized reboot() in containers, as
10812 supported by newer kernels. We will fall back to exit() if
10813 CAP_SYS_REBOOT is not available to the container. Also,
10814 nspawn makes use of this now and will actually reboot the
10815 container if the containerized OS asks for that.
10817 * journalctl will only show local log output by default
10818 now. Use --merge (-m) to show remote log output, too.
10820 * libsystemd-journal gained the new sd_journal_get_usage()
10821 call to determine the current disk usage of all journal
10822 files. This is exposed in the new "journalctl --disk-usage"
10825 * journald gained a new configuration setting SplitMode= in
10826 journald.conf which may be used to control how user journals
10827 are split off. See journald.conf(5) for details.
10829 * A new condition type ConditionFileNotEmpty= has been added.
10831 * tmpfiles' "w" lines now support file globbing, to write
10832 multiple files at once.
10834 * We added Python bindings for the journal submission
10835 APIs. More Python APIs for a number of selected APIs will
10836 likely follow. Note that we intend to add native bindings
10837 only for the Python language, as we consider it common
10838 enough to deserve bindings shipped within systemd. There are
10839 various projects outside of systemd that provide bindings
10840 for languages such as PHP or Lua.
10842 * Many conditions will now resolve specifiers such as %i. In
10843 addition, PathChanged= and related directives of .path units
10844 now support specifiers as well.
10846 * There's now a new RPM macro definition for the system preset
10849 * journald will now warn if it ca not forward a message to the
10850 syslog daemon because its socket is full.
10852 * timedated will no longer write or process /etc/timezone,
10853 except on Debian. As we do not support late mounted /usr
10854 anymore /etc/localtime always being a symlink is now safe,
10855 and hence the information in /etc/timezone is not necessary
10858 * logind will now always reserve one VT for a text getty (VT6
10859 by default). Previously if more than 6 X sessions where
10860 started they took up all the VTs with auto-spawned gettys,
10861 so that no text gettys were available anymore.
10863 * udev will now automatically inform the btrfs kernel logic
10864 about btrfs RAID components showing up. This should make
10865 simple hotplug based btrfs RAID assembly work.
10867 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
10868 (but not for its children which will stay at the kernel
10869 default). This should allow setups with a lot more listening
10872 * systemd will now always pass the configured timezone to the
10873 kernel at boot. timedated will do the same when the timezone
10876 * logind's inhibition logic has been updated. By default,
10877 logind will now handle the lid switch, the power and sleep
10878 keys all the time, even in graphical sessions. If DEs want
10879 to handle these events on their own they should take the new
10880 handle-power-key, handle-sleep-key and handle-lid-switch
10881 inhibitors during their runtime. A simple way to achieve
10882 that is to invoke the DE wrapped in an invocation of:
10884 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
10886 * Access to unit operations is now checked via SELinux taking
10887 the unit file label and client process label into account.
10889 * systemd will now notify the administrator in the journal
10890 when he over-mounts a non-empty directory.
10892 * There are new specifiers that are resolved in unit files,
10893 for the hostname (%H), the machine ID (%m) and the boot ID
10896 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
10897 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
10898 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
10899 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
10900 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
10901 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
10902 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
10906 * Support for reading structured kernel messages from
10907 /dev/kmsg has now been added and is enabled by default.
10909 * Support for reading kernel messages from /proc/kmsg has now
10910 been removed. If you want kernel messages in the journal
10911 make sure to run a recent kernel (>= 3.5) that supports
10912 reading structured messages from /dev/kmsg (see
10913 above). /proc/kmsg is now exclusive property of classic
10914 syslog daemons again.
10916 * The libudev API gained the new
10917 udev_device_new_from_device_id() call.
10919 * The logic for file system namespace (ReadOnlyDirectory=,
10920 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
10921 require pivot_root() anymore. This means fewer temporary
10922 directories are created below /tmp for this feature.
10924 * nspawn containers will now see and receive all submounts
10925 made on the host OS below the root file system of the
10928 * Forward Secure Sealing is now supported for Journal files,
10929 which provide cryptographical sealing of journal files so
10930 that attackers cannot alter log history anymore without this
10931 being detectable. Lennart will soon post a blog story about
10932 this explaining it in more detail.
10934 * There are two new service settings RestartPreventExitStatus=
10935 and SuccessExitStatus= which allow configuration of exit
10936 status (exit code or signal) which will be excepted from the
10937 restart logic, resp. consider successful.
10939 * journalctl gained the new --verify switch that can be used
10940 to check the integrity of the structure of journal files and
10941 (if Forward Secure Sealing is enabled) the contents of
10944 * nspawn containers will now be run with /dev/stdin, /dev/fd/
10945 and similar symlinks pre-created. This makes running shells
10946 as container init process a lot more fun.
10948 * The fstab support can now handle PARTUUID= and PARTLABEL=
10951 * A new ConditionHost= condition has been added to match
10952 against the hostname (with globs) and machine ID. This is
10953 useful for clusters where a single OS image is used to
10954 provision a large number of hosts which shall run slightly
10955 different sets of services.
10957 * Services which hit the restart limit will now be placed in a
10960 Contributions from: Bertram Poettering, Dave Reisner, Huang
10961 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
10962 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
10966 * When running in --user mode systemd will now become a
10967 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
10968 tree a lot more organized.
10970 * A new PartOf= unit dependency type has been introduced that
10971 may be used to group services in a natural way.
10973 * "systemctl enable" may now be used to enable instances of
10976 * journalctl now prints error log levels in red, and
10977 warning/notice log levels in bright white. It also supports
10978 filtering by log level now.
10980 * cgtop gained a new -n switch (similar to top), to configure
10981 the maximum number of iterations to run for. It also gained
10982 -b, to run in batch mode (accepting no input).
10984 * The suffix ".service" may now be omitted on most systemctl
10985 command lines involving service unit names.
10987 * There's a new bus call in logind to lock all sessions, as
10988 well as a loginctl verb for it "lock-sessions".
10990 * libsystemd-logind.so gained a new call sd_journal_perror()
10991 that works similar to libc perror() but logs to the journal
10992 and encodes structured information about the error number.
10994 * /etc/crypttab entries now understand the new keyfile-size=
10997 * shutdown(8) now can send a (configurable) wall message when
10998 a shutdown is cancelled.
11000 * The mount propagation mode for the root file system will now
11001 default to "shared", which is useful to make containers work
11002 nicely out-of-the-box so that they receive new mounts from
11003 the host. This can be undone locally by running "mount
11004 --make-rprivate /" if needed.
11006 * The prefdm.service file has been removed. Distributions
11007 should maintain this unit downstream if they intend to keep
11008 it around. However, we recommend writing normal unit files
11009 for display managers instead.
11011 * Since systemd is a crucial part of the OS we will now
11012 default to a number of compiler switches that improve
11013 security (hardening) such as read-only relocations, stack
11014 protection, and suchlike.
11016 * The TimeoutSec= setting for services is now split into
11017 TimeoutStartSec= and TimeoutStopSec= to allow configuration
11018 of individual time outs for the start and the stop phase of
11021 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
11022 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
11023 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
11024 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
11025 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
11026 Gundersen, Zbigniew Jędrzejewski-Szmek
11030 * The journal and id128 C APIs are now fully documented as man
11033 * Extra safety checks have been added when transitioning from
11034 the initial RAM disk to the main system to avoid accidental
11037 * /etc/crypttab entries now understand the new keyfile-offset=
11040 * systemctl -t can now be used to filter by unit load state.
11042 * The journal C API gained the new sd_journal_wait() call to
11043 make writing synchronous journal clients easier.
11045 * journalctl gained the new -D switch to show journals from a
11046 specific directory.
11048 * journalctl now displays a special marker between log
11049 messages of two different boots.
11051 * The journal is now explicitly flushed to /var via a service
11052 systemd-journal-flush.service, rather than implicitly simply
11053 by seeing /var/log/journal to be writable.
11055 * journalctl (and the journal C APIs) can now match for much
11056 more complex expressions, with alternatives and
11059 * When transitioning from the initial RAM disk to the main
11060 system we will now kill all processes in a killing spree to
11061 ensure no processes stay around by accident.
11063 * Three new specifiers may be used in unit files: %u, %h, %s
11064 resolve to the user name, user home directory resp. user
11065 shell. This is useful for running systemd user instances.
11067 * We now automatically rotate journal files if their data
11068 object hash table gets a fill level > 75%. We also size the
11069 hash table based on the configured maximum file size. This
11070 together should lower hash collisions drastically and thus
11071 speed things up a bit.
11073 * journalctl gained the new "--header" switch to introspect
11074 header data of journal files.
11076 * A new setting SystemCallFilters= has been added to services which may
11077 be used to apply deny lists or allow lists to system calls. This is
11078 based on SECCOMP Mode 2 of Linux 3.5.
11080 * nspawn gained a new --link-journal= switch (and quicker: -j)
11081 to link the container journal with the host. This makes it
11082 very easy to centralize log viewing on the host for all
11083 guests while still keeping the journal files separated.
11085 * Many bugfixes and optimizations
11087 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
11088 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
11089 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
11094 * Several tools now understand kernel command line arguments,
11095 which are only read when run in an initial RAM disk. They
11096 usually follow closely their normal counterparts, but are
11099 * There's a new tool to analyze the readahead files that are
11100 automatically generated at boot. Use:
11102 /usr/lib/systemd/systemd-readahead analyze /.readahead
11104 * We now provide an early debug shell on tty9 if this enabled. Use:
11106 systemctl enable debug-shell.service
11108 * All plymouth related units have been moved into the Plymouth
11109 package. Please make sure to upgrade your Plymouth version
11112 * systemd-tmpfiles now supports getting passed the basename of
11113 a configuration file only, in which case it will look for it
11114 in all appropriate directories automatically.
11116 * udevadm info now takes a /dev or /sys path as argument, and
11117 does the right thing. Example:
11119 udevadm info /dev/sda
11120 udevadm info /sys/class/block/sda
11122 * systemctl now prints a warning if a unit is stopped but a
11123 unit that might trigger it continues to run. Example: a
11124 service is stopped but the socket that activates it is left
11127 * "systemctl status" will now mention if the log output was
11128 shortened due to rotation since a service has been started.
11130 * The journal API now exposes functions to determine the
11131 "cutoff" times due to rotation.
11133 * journald now understands SIGUSR1 and SIGUSR2 for triggering
11134 immediately flushing of runtime logs to /var if possible,
11135 resp. for triggering immediate rotation of the journal
11138 * It is now considered an error if a service is attempted to
11139 be stopped that is not loaded.
11141 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
11143 * systemd-analyze now supports Python 3
11145 * tmpfiles now supports cleaning up directories via aging
11146 where the first level dirs are always kept around but
11147 directories beneath it automatically aged. This is enabled
11148 by prefixing the age field with '~'.
11150 * Seat objects now expose CanGraphical, CanTTY properties
11151 which is required to deal with very fast bootups where the
11152 display manager might be running before the graphics drivers
11153 completed initialization.
11155 * Seat objects now expose a State property.
11157 * We now include RPM macros for service enabling/disabling
11158 based on the preset logic. We recommend RPM based
11159 distributions to make use of these macros if possible. This
11160 makes it simpler to reuse RPM spec files across
11163 * We now make sure that the collected systemd unit name is
11164 always valid when services log to the journal via
11167 * There's a new man page kernel-command-line(7) detailing all
11168 command line options we understand.
11170 * The fstab generator may now be disabled at boot by passing
11171 fstab=0 on the kernel command line.
11173 * A new kernel command line option modules-load= is now understood
11174 to load a specific kernel module statically, early at boot.
11176 * Unit names specified on the systemctl command line are now
11177 automatically escaped as needed. Also, if file system or
11178 device paths are specified they are automatically turned
11179 into the appropriate mount or device unit names. Example:
11181 systemctl status /home
11182 systemctl status /dev/sda
11184 * The SysVConsole= configuration option has been removed from
11185 system.conf parsing.
11187 * The SysV search path is no longer exported on the D-Bus
11190 * The Names= option has been removed from unit file parsing.
11192 * There's a new man page bootup(7) detailing the boot process.
11194 * Every unit and every generator we ship with systemd now
11195 comes with full documentation. The self-explanatory boot is
11198 * A couple of services gained "systemd-" prefixes in their
11199 name if they wrap systemd code, rather than only external
11200 code. Among them fsck@.service which is now
11201 systemd-fsck@.service.
11203 * The HaveWatchdog property has been removed from the D-Bus
11206 * systemd.confirm_spawn= on the kernel command line should now
11209 * There's a new man page crypttab(5) which details all options
11210 we actually understand.
11212 * systemd-nspawn gained a new --capability= switch to pass
11213 additional capabilities to the container.
11215 * timedated will now read known NTP implementation unit names
11216 from /usr/lib/systemd/ntp-units.d/*.list,
11217 systemd-timedated-ntp.target has been removed.
11219 * journalctl gained a new switch "-b" that lists log data of
11220 the current boot only.
11222 * The notify socket is in the abstract namespace again, in
11223 order to support daemons which chroot() at start-up.
11225 * There is a new Storage= configuration option for journald
11226 which allows configuration of where log data should go. This
11227 also provides a way to disable journal logging entirely, so
11228 that data collected is only forwarded to the console, the
11229 kernel log buffer or another syslog implementation.
11231 * Many bugfixes and optimizations
11233 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
11234 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
11235 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
11236 Shawn Landden, Tom Gundersen
11240 * "systemctl help <unit>" now shows the man page if one is
11243 * Several new man pages have been added.
11245 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
11246 MaxLevelConsole= can now be specified in
11247 journald.conf. These options allow reducing the amount of
11248 data stored on disk or forwarded by the log level.
11250 * TimerSlackNSec= can now be specified in system.conf for
11251 PID1. This allows system-wide power savings.
11253 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
11254 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
11259 * logind is now capable of (optionally) handling power and
11260 sleep keys as well as the lid switch.
11262 * journalctl now understands the syntax "journalctl
11263 /usr/bin/avahi-daemon" to get all log output of a specific
11266 * CapabilityBoundingSet= in system.conf now also influences
11267 the capability bound set of usermode helpers of the kernel.
11269 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
11270 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
11271 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
11272 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
11276 * Note that we skipped 139 releases here in order to set the
11277 new version to something that is greater than both udev's
11278 and systemd's most recent version number.
11280 * udev: all udev sources are merged into the systemd source tree now.
11281 All future udev development will happen in the systemd tree. It
11282 is still fully supported to use the udev daemon and tools without
11283 systemd running, like in initramfs or other init systems. Building
11284 udev though, will require the *build* of the systemd tree, but
11285 udev can be properly *run* without systemd.
11287 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
11288 should be used to create dead device nodes as workarounds for broken
11291 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
11292 no longer supported. udev_monitor_new_from_netlink() needs to be
11293 used to subscribe to events.
11295 * udev: when udevd is started by systemd, processes which are left
11296 behind by forking them off of udev rules, are unconditionally cleaned
11297 up and killed now after the event handling has finished. Services or
11298 daemons must be started as systemd services. Services can be
11299 pulled-in by udev to get started, but they can no longer be directly
11300 forked by udev rules.
11302 * udev: the daemon binary is called systemd-udevd now and installed
11303 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
11304 to adapt to that, create symlink, or rename the binary after building
11307 * libudev no longer provides these symbols:
11308 udev_monitor_from_socket()
11309 udev_queue_get_failed_list_entry()
11310 udev_get_{dev,sys,run}_path()
11311 The versions number was bumped and symbol versioning introduced.
11313 * systemd-loginctl and systemd-journalctl have been renamed
11314 to loginctl and journalctl to match systemctl.
11316 * The config files: /etc/systemd/systemd-logind.conf and
11317 /etc/systemd/systemd-journald.conf have been renamed to
11318 logind.conf and journald.conf. Package updates should rename
11319 the files to the new names on upgrade.
11321 * For almost all files the license is now LGPL2.1+, changed
11322 from the previous GPL2.0+. Exceptions are some minor stuff
11323 of udev (which will be changed to LGPL2.1 eventually, too),
11324 and the MIT licensed sd-daemon.[ch] library that is suitable
11325 to be used as drop-in files.
11327 * systemd and logind now handle system sleep states, in
11328 particular suspending and hibernating.
11330 * logind now implements a sleep/shutdown/idle inhibiting logic
11331 suitable for a variety of uses. Soonishly Lennart will blog
11332 about this in more detail.
11334 * var-run.mount and var-lock.mount are no longer provided
11335 (which previously bind mounted these directories to their new
11336 places). Distributions which have not converted these
11337 directories to symlinks should consider stealing these files
11338 from git history and add them downstream.
11340 * We introduced the Documentation= field for units and added
11341 this to all our shipped units. This is useful to make it
11342 easier to explore the boot and the purpose of the various
11345 * All smaller setup units (such as
11346 systemd-vconsole-setup.service) now detect properly if they
11347 are run in a container and are skipped when
11348 appropriate. This guarantees an entirely noise-free boot in
11349 Linux container environments such as systemd-nspawn.
11351 * A framework for implementing offline system updates is now
11352 integrated, for details see:
11353 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
11355 * A new service type Type=idle is available now which helps us
11356 avoiding ugly interleaving of getty output and boot status
11359 * There's now a system-wide CapabilityBoundingSet= option to
11360 globally reduce the set of capabilities for the
11361 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
11362 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
11363 even CAP_NET_ADMIN system-wide for secure systems.
11365 * There are now system-wide DefaultLimitXXX= options to
11366 globally change the defaults of the various resource limits
11367 for all units started by PID 1.
11369 * Harald Hoyer's systemd test suite has been integrated into
11370 systemd which allows easy testing of systemd builds in qemu
11371 and nspawn. (This is really awesome! Ask us for details!)
11373 * The fstab parser is now implemented as generator, not inside
11376 * systemctl will now warn you if .mount units generated from
11377 /etc/fstab are out of date due to changes in fstab that
11378 have not been read by systemd yet.
11380 * systemd is now suitable for usage in initrds. Dracut has
11381 already been updated to make use of this. With this in place
11382 initrds get a slight bit faster but primarily are much
11383 easier to introspect and debug since "systemctl status" in
11384 the host system can be used to introspect initrd services,
11385 and the journal from the initrd is kept around too.
11387 * systemd-delta has been added, a tool to explore differences
11388 between user/admin configuration and vendor defaults.
11390 * PrivateTmp= now affects both /tmp and /var/tmp.
11392 * Boot time status messages are now much prettier and feature
11393 proper english language. Booting up systemd has never been
11396 * Read-ahead pack files now include the inode number of all
11397 files to pre-cache. When the inode changes the pre-caching
11398 is not attempted. This should be nicer to deal with updated
11399 packages which might result in changes of read-ahead
11402 * We now temporaritly lower the kernel's read_ahead_kb variable
11403 when collecting read-ahead data to ensure the kernel's
11404 built-in read-ahead does not add noise to our measurements
11405 of necessary blocks to pre-cache.
11407 * There's now RequiresMountsFor= to add automatic dependencies
11408 for all mounts necessary for a specific file system path.
11410 * MountAuto= and SwapAuto= have been removed from
11411 system.conf. Mounting file systems at boot has to take place
11414 * nspawn now learned a new switch --uuid= to set the machine
11415 ID on the command line.
11417 * nspawn now learned the -b switch to automatically search
11418 for an init system.
11420 * vt102 is now the default TERM for serial TTYs, upgraded from
11423 * systemd-logind now works on VT-less systems.
11425 * The build tree has been reorganized. The individual
11426 components now have directories of their own.
11428 * A new condition type ConditionPathIsReadWrite= is now available.
11430 * nspawn learned the new -C switch to create cgroups for the
11431 container in other hierarchies.
11433 * We now have support for hardware watchdogs, configurable in
11436 * The scheduled shutdown logic now has a public API.
11438 * We now mount /tmp as tmpfs by default, but this can be
11439 masked and /etc/fstab can override it.
11441 * Since udisks does not make use of /media anymore we are not
11442 mounting a tmpfs on it anymore.
11444 * journalctl gained a new --local switch to only interleave
11445 locally generated journal files.
11447 * We can now load the IMA policy at boot automatically.
11449 * The GTK tools have been split off into a systemd-ui.
11451 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
11452 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
11453 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
11454 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
11455 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
11456 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
11457 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
11458 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
11459 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
11464 * This is mostly a bugfix release
11466 * Support optional initialization of the machine ID from the
11467 KVM or container configured UUID.
11469 * Support immediate reboots with "systemctl reboot -ff"
11471 * Show /etc/os-release data in systemd-analyze output
11473 * Many bugfixes for the journal, including endianness fixes and
11474 ensuring that disk space enforcement works
11476 * sd-login.h is C++ compatible again
11478 * Extend the /etc/os-release format on request of the Debian
11481 * We now refuse non-UTF8 strings used in various configuration
11482 and unit files. This is done to ensure we do not pass invalid
11483 data over D-Bus or expose it elsewhere.
11485 * Register Mimo USB Screens as suitable for automatic seat
11488 * Read SELinux client context from journal clients in a race
11491 * Reorder configuration file lookup order. /etc now always
11492 overrides /run in order to allow the administrator to always
11493 and unconditionally override vendor-supplied or
11494 automatically generated data.
11496 * The various user visible bits of the journal now have man
11497 pages. We still lack man pages for the journal API calls
11500 * We now ship all man pages in HTML format again in the
11503 Contributions from: Dave Reisner, Dirk Eibach, Frederic
11504 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
11505 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
11510 * This is mostly a bugfix release
11512 * systems lacking /etc/os-release are no longer supported.
11514 * Various functionality updates to libsystemd-login.so
11516 * Track class of PAM logins to distinguish greeters from
11517 normal user logins.
11519 Contributions from: Kay Sievers, Lennart Poettering, Michael
11524 * This is an important bugfix release for v41.
11526 * Building man pages is now optional which should be useful
11527 for those building systemd from git but unwilling to install
11530 * Watchdog support for supervising services is now usable. In
11531 a future release support for hardware watchdogs
11532 (i.e. /dev/watchdog) will be added building on this.
11534 * Service start rate limiting is now configurable and can be
11535 turned off per service. When a start rate limit is hit a
11536 reboot can automatically be triggered.
11538 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
11540 Contributions from: Benjamin Franzke, Bill Nottingham,
11541 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
11542 Schmidt, Michał Górny, Piotr Drąg
11546 * The systemd binary is installed /usr/lib/systemd/systemd now;
11547 An existing /sbin/init symlink needs to be adapted with the
11550 * The code that loads kernel modules has been ported to invoke
11551 libkmod directly, instead of modprobe. This means we do not
11552 support systems with module-init-tools anymore.
11554 * Watchdog support is now already useful, but still not
11557 * A new kernel command line option systemd.setenv= is
11558 understood to set system wide environment variables
11559 dynamically at boot.
11561 * We now limit the set of capabilities of systemd-journald.
11563 * We now set SIGPIPE to ignore by default, since it only is
11564 useful in shell pipelines, and has little use in general
11565 code. This can be disabled with IgnoreSIPIPE=no in unit
11568 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
11569 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
11574 * This is mostly a bugfix release
11576 * We now expose the reason why a service failed in the
11577 "Result" D-Bus property.
11579 * Rudimentary service watchdog support (will be completed over
11580 the next few releases.)
11582 * When systemd forks off in order execute some service we will
11583 now immediately changes its argv[0] to reflect which process
11584 it will execute. This is useful to minimize the time window
11585 with a generic argv[0], which makes bootcharts more useful
11587 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
11588 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
11589 Mike Kazantsev, Ray Strode
11593 * This is mostly a test release, but incorporates many
11596 * New systemd-cgtop tool to show control groups by their
11599 * Linking against libacl for ACLs is optional again. If
11600 disabled, support tracking device access for active logins
11601 goes becomes unavailable, and so does access to the user
11602 journals by the respective users.
11604 * If a group "adm" exists, journal files are automatically
11605 owned by them, thus allow members of this group full access
11606 to the system journal as well as all user journals.
11608 * The journal now stores the SELinux context of the logging
11609 client for all entries.
11611 * Add C++ inclusion guards to all public headers
11613 * New output mode "cat" in the journal to print only text
11614 messages, without any meta data like date or time.
11616 * Include tiny X server wrapper as a temporary stop-gap to
11617 teach XOrg udev display enumeration. This is used by display
11618 managers such as gdm, and will go away as soon as XOrg
11619 learned native udev hotplugging for display devices.
11621 * Add new systemd-cat tool for executing arbitrary programs
11622 with STDERR/STDOUT connected to the journal. Can also act as
11623 BSD logger replacement, and does so by default.
11625 * Optionally store all locally generated coredumps in the
11626 journal along with meta data.
11628 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
11629 writing short strings to files (for usage for /sys), and for
11630 creating symlinks, character and block device nodes.
11632 * New unit file option ControlGroupPersistent= to make cgroups
11633 persistent, following the mechanisms outlined in
11634 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
11636 * Support multiple local RTCs in a sane way
11638 * No longer monopolize IO when replaying readahead data on
11639 rotating disks, since we might starve non-file-system IO to
11640 death, since fanotify() will not see accesses done by blkid,
11643 * Do not show kernel threads in systemd-cgls anymore, unless
11644 requested with new -k switch.
11646 Contributions from: Dan Horák, Kay Sievers, Lennart
11647 Poettering, Michal Schmidt
11651 * This is mostly a test release, but incorporates many
11654 * The git repository moved to:
11655 git://anongit.freedesktop.org/systemd/systemd
11656 ssh://git.freedesktop.org/git/systemd/systemd
11658 * First release with the journal
11659 http://0pointer.de/blog/projects/the-journal.html
11661 * The journal replaces both systemd-kmsg-syslogd and
11662 systemd-stdout-bridge.
11664 * New sd_pid_get_unit() API call in libsystemd-logind
11666 * Many systemadm clean-ups
11668 * Introduce remote-fs-pre.target which is ordered before all
11669 remote mounts and may be used to start services before all
11672 * Added Mageia support
11674 * Add bash completion for systemd-loginctl
11676 * Actively monitor PID file creation for daemons which exit in
11677 the parent process before having finished writing the PID
11678 file in the daemon process. Daemons which do this need to be
11679 fixed (i.e. PID file creation must have finished before the
11680 parent exits), but we now react a bit more gracefully to them.
11682 * Add colourful boot output, mimicking the well-known output
11683 of existing distributions.
11685 * New option PassCredentials= for socket units, for
11686 compatibility with a recent kernel ABI breakage.
11688 * /etc/rc.local is now hooked in via a generator binary, and
11689 thus will no longer act as synchronization point during
11692 * systemctl list-unit-files now supports --root=.
11694 * systemd-tmpfiles now understands two new commands: z, Z for
11695 relabelling files according to the SELinux database. This is
11696 useful to apply SELinux labels to specific files in /sys,
11697 among other things.
11699 * Output of SysV services is now forwarded to both the console
11700 and the journal by default, not only just the console.
11702 * New man pages for all APIs from libsystemd-login.
11704 * The build tree got reorganized and the build system is a
11705 lot more modular allowing embedded setups to specifically
11706 select the components of systemd they are interested in.
11708 * Support for Linux systems lacking the kernel VT subsystem is
11711 * configure's --with-rootdir= got renamed to
11712 --with-rootprefix= to follow the naming used by udev and
11715 * Unless specified otherwise we will now install to /usr instead
11716 of /usr/local by default.
11718 * Processes with '@' in argv[0][0] are now excluded from the
11719 final shut-down killing spree, following the logic explained
11721 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
11723 * All processes remaining in a service cgroup when we enter
11724 the START or START_PRE states are now killed with
11725 SIGKILL. That means it is no longer possible to spawn
11726 background processes from ExecStart= lines (which was never
11727 supported anyway, and bad style).
11729 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
11730 reloading of units together.
11732 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
11733 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
11734 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
11735 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
11736 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek