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 * Timer units gained a new FixedRandomDelay= boolean setting. If
162 enabled, the random delay configured with RandomizedDelaySec= is
163 selected in a way that is stable on a given system (though still
164 different for different units).
166 * Socket units gained a new setting Timestamping= that takes "us", "ns"
167 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
170 * systemd-repart now generates JSON output when requested with the new
173 * systemd-machined's OpenMachineShell() bus call will now pass
174 additional policy metadata data fields to the PolicyKit
175 authentication request.
177 * systemd-tmpfiles gained a new -E switch, which is equivalent to
178 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
179 --exclude=/sys. It's particularly useful in combination with --root=,
180 when operating on OS trees that do not have any of these four runtime
181 directories mounted, as this means no files below these subtrees are
182 created or modified, since those mount points should probably remain
185 * systemd-tmpfiles gained a new --image= switch which is like --root=,
186 but takes a disk image instead of a directory as argument. The
187 specified disk image is mounted inside a temporary mount namespace
188 and the tmpfiles.d/ drop-ins stored in the image are executed and
189 applied to the image. systemd-sysusers similarly gained a new
190 --image= switch, that allows the sysusers.d/ drop-ins stored in the
191 image to be applied onto the image.
193 * Similarly, the journalctl command also gained an --image= switch,
194 which is a quick one-step solution to look at the log data included
197 * journalctl's --output=cat option (which outputs the log content
198 without any metadata, just the pure text messages) will now make use
199 of terminal colors when run on a suitable terminal, similarly to the
202 * JSON group records now support a "description" string that may be
203 used to add a human-readable textual description to such groups. This
204 is supposed to match the user's GECOS field which traditionally
205 didn't have a counterpart for group records.
207 * The "systemd-dissect" tool that may be used to inspect OS disk images
208 and that was previously installed to /usr/lib/systemd/ has now been
209 moved to /usr/bin/, reflecting its updated status of an officially
210 supported tool with a stable interface. It gained support for a new
211 --mkdir switch which when combined with --mount has the effect of
212 creating the directory to mount the image to if it is missing
213 first. It also gained two new commands --copy-from and --copy-to for
214 copying files and directories in and out of an OS image without the
215 need to manually mount it. It also acquired support for a new option
216 --json= to generate JSON output when inspecting an OS image.
218 * The cgroup2 file system is now mounted with the
219 "memory_recursiveprot" mount option, supported since kernel 5.7. This
220 means that the MemoryLow= and MemoryMin= unit file settings now apply
221 recursively to whole subtrees.
223 * systemd-homed now defaults to using the btrfs file system — if
224 available — when creating home directories in LUKS volumes. This may
225 be changed with the DefaultFileSystemType= setting in homed.conf.
226 It's now the default file system in various major distributions and
227 has the major benefit for homed that it can be grown and shrunk while
228 mounted, unlike the other contenders ext4 and xfs, which can both be
229 grown online, but not shrunk (in fact xfs is the technically most
230 limited option here, as it cannot be shrunk at all).
232 * JSON user records managed by systemd-homed gained support for
233 "recovery keys". These are basically secondary passphrases that can
234 unlock user accounts/home directories. They are computer-generated
235 rather than user-chosen, and typically have greater entropy.
236 homectl's --recovery-key= option may be used to add a recovery key to
237 a user account. The generated recovery key is displayed as a QR code,
238 so that it can be scanned to be kept in a safe place. This feature is
239 particularly useful in combination with systemd-homed's support for
240 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
241 security tokens are lost. Recovery keys may be entered wherever the
242 system asks for a password.
244 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
245 home directory which indicates that a home directory has not been
246 deactivated cleanly when offline. This flag is useful to identify
247 home directories for which the offline discard logic did not run when
248 offlining, and where it would be a good idea to log in again to catch
251 * systemctl gained a new parameter --timestamp= which may be used to
252 change the style in which timestamps are output, i.e. whether to show
253 them in local timezone or UTC, or whether to show µs granularity.
255 * Alibaba's "pouch" container manager is now detected by
256 systemd-detect-virt, ConditionVirtualization= and similar
257 constructs. Similar, they now also recognize IBM PowerVM machine
260 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
261 place to use for propagating external mounts into the
262 container. Similarly /run/host/notify is now used as the socket path
263 for container payloads to communicate with the container manager
264 using sd_notify(). The container manager now uses the
265 /run/host/inaccessible/ directory to place "inaccessible" file nodes
266 of all relevant types which may be used by the container payload as
267 bind mount source to over-mount inodes to make them inaccessible.
268 /run/host/container-manager will now be initialized with the same
269 string as the $container environment variable passed to the
270 container's PID 1. /run/host/container-uuid will be initialized with
271 the same string as $container_uuid. This means the /run/host/
272 hierarchy is now the primary way to make host resources available to
273 the container. The Container Interface documents these new files and
276 https://systemd.io/CONTAINER_INTERFACE
278 * Support for the "ConditionNull=" unit file condition has been
279 deprecated and undocumented for 6 years. systemd started to warn
280 about its use 1.5 years ago. It has now been removed entirely.
282 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
283 a sd_bus_error struct and a list of error names, and checks if the
284 error matches one of these names. It's a convenience wrapper that is
285 useful in cases where multiple errors shall be handled the same way.
287 * A new system call filter list "@known" has been added, that contains
288 all system calls known at the time systemd was built.
290 * Behaviour of system call filter allow lists has changed slightly:
291 system calls that are contained in @known will result in a EPERM by
292 default, while those not contained in it result in ENOSYS. This
293 should improve compatibility because known system calls will thus be
294 communicated as prohibited, while unknown (and thus newer ones) will
295 be communicated as not implemented, which hopefully has the greatest
296 chance of triggering the right fallback code paths in client
299 * "systemd-analyze syscall-filter" will now show two separate sections
300 at the bottom of the output: system calls known during systemd build
301 time but not included in any of the filter groups shown above, and
302 system calls defined on the local kernel but known during systemd
305 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
306 systemd-nspawn all system call filter violations will be logged by
307 the kernel (audit). This is useful for tracking down system calls
308 invoked by container payloads that are prohibited by the container's
309 system call filter policy.
311 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
312 systemd-nspawn (and other programs that use seccomp) all seccomp
313 filtering is turned off.
315 * Two new unit file settings ProtectProc= and ProcSubset= have been
316 added that expose the hidepid= and subset= mount options of procfs.
317 All processes of the unit will only see processes in /proc that are
318 are owned by the unit's user. This is an important new sandboxing
319 option that is recommended to be set on all system services. All
320 long-running system services that are included in systemd itself set
321 this option now. This option is only supported on kernel 5.8 and
322 above, since the hidepid= option supported on older kernels was not a
323 per-mount option but actually applied to the whole PID namespace.
325 * Socket units gained a new boolean setting FlushPending=. If enabled
326 all pending socket data/connections are flushed whenever the socket
327 unit enters the "listening" state, i.e. after the associated service
330 * The unit file setting NUMAMask= gained a new "all" value: when used,
331 all existing NUMA nodes are added to the NUMA mask.
333 * A new "credentials" logic has been added to system services. This is
334 a simple mechanism to pass privileged data to services in a safe and
335 secure way. It's supposed to be used to pass per-service secret data
336 such as passwords or cryptographic keys but also associated less
337 private information such as user names, certificates, and similar to
338 system services. Each credential is identified by a short user-chosen
339 name and may contain arbitrary binary data. Two new unit file
340 settings have been added: SetCredential= and LoadCredential=. The
341 former allows setting a credential to a literal string, the latter
342 sets a credential to the contents of a file (or data read from a
343 user-chosen AF_UNIX stream socket). Credentials are passed to the
344 service via a special credentials directory, one file for each
345 credential. The path to the credentials directory is passed in a new
346 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
347 are passed in the file system they may be easily referenced in
348 ExecStart= command lines too, thus no explicit support for the
349 credentials logic in daemons is required (though ideally daemons
350 would look for the bits they need in $CREDENTIALS_DIRECTORY
351 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
352 backed by unswappable memory if privileges allow it, immutable if
353 privileges allow it, is accessible only to the service's UID, and is
354 automatically destroyed when the service stops.
356 * systemd-nspawn supports the same credentials logic. It can both
357 consume credentials passed to it via the aforementioned
358 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
359 to its payload. The service manager/PID 1 has been updated to match
360 this: it can also accept credentials from the container manager that
361 invokes it (in fact: any process that invokes it), and passes them on
362 to its services. Thus, credentials can be propagated recursively down
363 the tree: from a system's service manager to a systemd-nspawn
364 service, to the service manager that runs as container payload and to
365 the service it runs below. Credentials may also be added on the
366 systemd-nspawn command line, using new --set-credential= and
367 --load-credential= command line switches that match the
368 aforementioned service settings.
370 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
371 the partition drop-ins which may be used to format/LUKS
372 encrypt/populate any created partitions. The partitions are
373 encrypted/formatted/populated before they are registered in the
374 partition table, so that they appear atomically: either the
375 partitions do not exist yet or they exist fully encrypted, formatted,
376 and populated — there is no time window where they are
377 "half-initialized". Thus the system is robust to abrupt shutdown: if
378 the tool is terminated half-way during its operations on next boot it
379 will start from the beginning.
381 * systemd-repart's --size= operation gained a new "auto" value. If
382 specified, and operating on a loopback file it is automatically sized
383 to the minimal size the size constraints permit. This is useful to
384 use "systemd-repart" as an image builder for minimally sized images.
386 * systemd-resolved now gained a third IPC interface for requesting name
387 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
388 interface is now supported. The nss-resolve NSS module has been
389 modified to use this new interface instead of D-Bus. Using Varlink
390 has a major benefit over D-Bus: it works without a broker service,
391 and thus already during earliest boot, before the dbus daemon has
392 been started. This means name resolution via systemd-resolved now
393 works at the same time systemd-networkd operates: from earliest boot
394 on, including in the initrd.
396 * systemd-resolved gained support for a new DNSStubListenerExtra=
397 configuration file setting which may be used to specify additional IP
398 addresses the built-in DNS stub shall listen on, in addition to the
399 main one on 127.0.0.53:53.
401 * Name lookups issued via systemd-resolved's D-Bus and Varlink
402 interfaces (and thus also via glibc NSS if nss-resolve is used) will
403 now honour a trailing dot in the hostname: if specified the search
404 path logic is turned off. Thus "resolvectl query foo." is now
405 equivalent to "resolvectl query --search=off foo.".
407 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
408 exposes how /etc/resolv.conf is currently managed: by resolved (and
409 in which mode if so) or another subsystem. "resolvctl" will display
410 this property in its status output.
412 * The resolv.conf snippets systemd-resolved provides will now set "."
413 as the search domain if no other search domain is known. This turns
414 off the derivation of an implicit search domain by nss-dns for the
415 hostname, when the hostname is set to an FQDN. This change is done to
416 make nss-dns using resolv.conf provided by systemd-resolved behave
417 more similarly to nss-resolve.
419 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
420 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
421 "birth" time (btime) of a file in addition to the atime, mtime, and
424 * systemd-analyze gained a new verb "capability" that lists all known
425 capabilities by the systemd build and by the kernel.
427 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
428 advance the system clock to it at boot if it is noticed to be before
429 that time. Previously, PID 1 would only advance the time to an epoch
430 time that is set during build-time. With this new file OS builders
431 can change this epoch timestamp on individual OS images without
432 having to rebuild systemd.
434 * systemd-logind will now listen to the KEY_RESTART key from the Linux
435 input layer and reboot the system if it is pressed, similarly to how
436 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
437 was originally defined in the Multimedia context (to restart playback
438 of a song or film), but is now primarily used in various embedded
439 devices for "Reboot" buttons. Accordingly, systemd-logind will now
440 honour it as such. This may configured in more detail via the new
441 HandleRebootKey= and RebootKeyIgnoreInhibited=.
443 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
444 copying OS trees, for example in "systemd-nspawn --ephemeral",
445 "systemd-nspawn --template=", "machinectl clone" and similar. This is
446 useful when operating with OSTree images, which use hardlinks heavily
447 throughout, and where such copies previously resulting in "exploding"
450 * systemd-nspawn's --console= setting gained support for a new
451 "autopipe" value, which is identical to "interactive" when invoked on
452 a TTY, and "pipe" otherwise.
454 * systemd-networkd's .network files gained support for explicitly
455 configuring the multicast membership entries of bridge devices in the
456 [BridgeMDB] section. It also gained support for the PIE queuing
457 discipline in the [FlowQueuePIE] sections.
459 * systemd-networkd's .netdev files may now be used to create "BareUDP"
460 tunnels, configured in the new [BareUDP] setting. VXLAN tunnels may
461 now be marked to be independent of any underlying network interface
462 via the new Independent= boolean setting.
464 * systemd-networkd's Gateway= setting in .network files now accepts the
465 special values _dhcp4 and _ipv6ra to configure additional, locally
466 defined, explicit routes to the gateway acquired via DHCP or IPv6
467 Router Advertisements.
469 * systemd-networkd's [IPv6PrefixDelegation] section and
470 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
471 IPv6SendRA= (the old names are still accepted for backwards
474 * systemctl gained support for two new verbs: "service-log-level" and
475 "service-log-target" may be used on services that implement the
476 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
477 adjust the log level and target. All of systemd's long-running
478 services support this now, but ideally all system services would
479 implement this interface to make the system more uniformly
482 * The SystemCallErrorNumber= unit file setting now accepts the new
483 "kill" and "log" actions, in addition to arbitrary error number
484 specifications as before. If "kill" the the processes are killed on
485 the event, if "log" the offending system call is audit logged.
487 * A new SystemCallLog= unit file setting has been added that accepts a
488 list of system calls that shall be logged about (audit).
490 * The OS image dissection logic (as used by RootImage= in unit files or
491 systemd-nspawn's --image= switch) has gained support for identifying
492 and mounting explicit /usr/ partitions, which are now defined in the
493 discoverable partition specification. This should be useful for
494 environments where the root file system is
495 generated/formatted/populated dynamically on first boot and combined
496 with an immutable /usr/ tree that is supplied by the vendor.
498 * In the final phase of shutdown, within the systemd-shutdown binary
499 we'll now try to detach MD devices (i.e software RAID) in addition to
500 loopback block devices and DM devices as before. This is supposed to
501 be a safety net only, in order to increase robustness if things go
502 wrong. Storage subsystems are expected to properly detach their
503 storage volumes during regular shutdown already (or in case of
504 storage backing the root file system: in the initrd hook we return to
507 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
508 will now log the thread ID in their log output. This is useful when
509 working with heavily threaded programs.
511 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
512 not use the RDRAND CPU instruction. This is useful in environments
513 such as replay debuggers where non-deterministic behaviour is not
516 * The autopaging logic in systemd's various tools (such as systemctl)
517 has been updated to turn on "secure" mode in "less"
518 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
519 detected. This disables invoking external programs from the pager,
520 via the pipe logic. This behaviour may be overridden via the new
521 $SYSTEMD_PAGERSECURE environment variable.
523 * Units which have resource limits (.service, .mount, .swap, .slice,
524 .socket, and .slice) gained new configuration settings
525 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
526 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
527 limits and optional action taken by systemd-oomd.
529 * A new service systemd-oomd has been added. It monitors resource
530 contention for selected parts of the unit hierarchy using the PSI
531 information reported by the kernel, and kills processes when memory
532 or swap pressure is above configured limits. This service is only
533 enabled in developer mode (see below) and should be considered a
534 preview in this release. Behaviour details and option names are
535 subject to change without the usual backwards-compatibility promises.
537 * A new helper oomctl has been added to introspect systemd-oomd state.
538 If also is only available in developer mode and should be considered
539 a preview without the usual backwards-compatibility promises.
541 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
542 enabled, systemd reads the system UID boundaries from /etc/login.defs
543 at runtime, instead of using the built-in values selected during
544 build. This is an option to improve compatibility for upgrades from
545 old systems. It's strongly recommended not to make use of this
546 functionality on new systems (or even enable it during build), as it
547 makes something runtime-configurable that is mostly an implementation
548 detail of the OS, and permits avoidable differences in deployments
549 that create all kinds of problems in the long run.
551 * New meson option '-Dmode=developer|release' has been added. When
552 'developer', additional checks and features are enabled that are
553 relevant during upstream development, e.g. verification that
554 semi-automatically-generated documentation has been properly updated
555 following API changes. Those checks are considered hints for
556 developers and are not actionable in downstream builds. In addition,
557 extra features that are not ready for general consumption may be
558 enabled in developer mode. It is thus recommended to set
559 '-Dmode=release' in end-user and distro builds.
561 * systemd-cryptsetup gained support for processing detached LUKS
562 headers specified on the kernel command line via the header=
563 parameter of the luks.options= kernel command line option. The same
564 device/path syntax as for key files is supported for header files
567 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
568 slot index data for devices that are connected through a PCI bridge
569 where the _SUN index is associated with the bridge instead of the
570 network device itself. Previously this would create ambiguous device
571 naming if multiple network interfaces were connected to the same PCI
572 bridge. Since this is a naming scheme incompatibility on systems that
573 possess hardware like this it has been introduced as new naming
574 scheme "v247". The previous scheme can be selected via the
575 "net.naming-scheme=v245" kernel command line parameter.
577 * ConditionFirstBoot= semantics have been modified to be safe towards
578 abnormal system power-off during first boot. Specifically, the
579 "systemd-machine-id-commit.service" service now acts as boot
580 milestone indicating when the first boot process is sufficiently
581 complete in order to not consider the next following boot also a
582 first boot. If the system is reset before this unit is reached the
583 first time, the next boot will still be considered a first boot; once
584 it has been reached, no further boots will be considered a first
585 boot. The "first-boot-complete.target" unit now acts as official hook
586 point to order against this. If a service shall be run on every boot
587 until the first boot fully succeeds it may thus be ordered before
588 this target unit (and pull it in) and carry ConditionFirstBoot=
591 * bootctl's set-default and set-oneshot commands now accept the three
592 special strings "@default", "@oneshot", "@current" in place of a boot
593 entry id. These strings are resolved to the current default and
594 oneshot boot loader entry, as well as the currently booted one. Thus
595 a command "bootctl set-default @current" may be used to make the
596 currently boot menu item the new default for all subsequent boots.
598 * "systemctl edit" has been updated to show the original effective unit
599 contents in commented form in the text editor.
601 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
602 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
603 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
604 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
605 Dey, Charles Lee, Chris Down, Christian Göttsche, Clemens Gruber, Daan
606 De Meyer, Daniele Medri, Daniel Mack, Daniel Rusek, Dan Streetman,
607 David Tardon, Dimitri John Ledkov, Dmitry Borodaenko, Elias Probst,
608 Elisei Roca, ErrantSpore, Etienne Doms, Fabrice Fontaine, fangxiuning,
609 Felix Riemann, Florian Klink, Franck Bui, Frantisek Sumsal, fwSmit,
610 George Rawlinson, germanztz, Gibeom Gwon, Glen Whitney, Gogo Gogsi,
611 Göran Uddeborg, Grant Mathews, Hans de Goede, Hans Ulrich Niedermann,
612 Haochen Tong, Harald Seiler, huangyong, Hubert Kario, Ikey Doherty, Jan
613 Chren, Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann,
614 Jonathan Lebon, Josh Brobst, Juergen Hoetzel, Julien Humbert, Kai-Chuan
615 Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan Gjoshev, Kyle
616 Huey, Kyle Russell, Lennart Poettering, lichangze, Luca Boccassi, Lucas
617 Werkmeister, Luca Weiss, Marc Kleine-Budde, Marco Wang, Martin Wilck,
618 Marti Raudsepp, masmullin2000, Máté Pozsgay, Matt Fenwick, Michael
619 Biebl, Michael Scherer, Michal Koutný, Michal Sekletár, Michal
620 Suchanek, Mikael Szreder, Milo Casagrande, mirabilos, Mitsuha_QuQ,
621 mog422, Muhammet Kara, Nazar Vinnichuk, Nicholas Narsing, Nicolas
622 Fella, Njibhu, nl6720, Oğuz Ersen, Olivier Le Moal, Ondrej Kozina,
623 onlybugreports, Pass Automated Testing Suite, Pat Coulthard, Pedro
624 Ruiz, Peter Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan
625 C, Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert
626 Marko, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer, Samanta
627 Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd, Susant
628 Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
629 Miettinen, Vito Caputo, Weblate, Wen Yang, williamvds, Yu, Li-Yu, Yuri
630 Chornoivan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Zmicer Turok,
637 * The service manager gained basic support for cgroup v2 freezer. Units
638 can now be suspended or resumed either using new systemctl verbs,
639 freeze and thaw respectively, or via D-Bus.
641 * PID 1 may now automatically load pre-compiled AppArmor policies from
642 /etc/apparmor/earlypolicy during early boot.
644 * The CPUAffinity= setting in service unit files now supports a new
645 special value "numa" that causes the CPU affinity masked to be set
646 based on the NUMA mask.
648 * systemd will now log about all left-over processes remaining in a
649 unit when the unit is stopped. It will now warn about services using
650 KillMode=none, as this is generally an unsafe thing to make use of.
652 * Two new unit file settings
653 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
654 added. They may be used to check whether a specific file system path
655 resides on a block device that is encrypted on the block level
656 (i.e. using dm-crypt/LUKS).
658 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
659 has been added that may be used for simple environment checks. This
660 is particularly useful when passing in environment variables from a
661 container manager (or from PAM in case of the systemd --user
664 * .service unit files now accept a new setting CoredumpFilter= which
665 allows configuration of the memory sections coredumps of the
666 service's processes shall include.
668 * .mount units gained a new ReadWriteOnly= boolean option. If set
669 it will not be attempted to mount a file system read-only if mounting
670 in read-write mode doesn't succeed. An option x-systemd.rw-only is
671 available in /etc/fstab to control the same.
673 * .socket units gained a new boolean setting PassPacketInfo=. If
674 enabled, the kernel will attach additional per-packet metadata to all
675 packets read from the socket, as an ancillary message. This controls
676 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
677 depending on socket type.
679 * .service units gained a new setting RootHash= which may be used to
680 specify the root hash for verity enabled disk images which are
681 specified in RootImage=. RootVerity= may be used to specify a path to
682 the Verity data matching a RootImage= file system. (The latter is
683 only useful for images that do not contain the Verity data embedded
684 into the same image that carries a GPT partition table following the
685 Discoverable Partition Specification). Similarly, systemd-nspawn
686 gained a new switch --verity-data= that takes a path to a file with
687 the verity data of the disk image supplied in --image=, if the image
688 doesn't contain the verity data itself.
690 * .service units gained a new setting RootHashSignature= which takes
691 either a base64 encoded PKCS#7 signature of the root hash specified
692 with RootHash=, or a path to a file to read the signature from. This
693 allows validation of the root hash against public keys available in
694 the kernel keyring, and is only supported on recent kernels
695 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
696 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
697 this mechanism has also been added to systemd-veritysetup.
699 * .service unit files gained two new options
700 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
701 tune behaviour if a start or stop timeout is hit, i.e. whether to
702 terminate the service with SIGTERM, SIGABRT or SIGKILL.
704 * Most options in systemd that accept hexadecimal values prefixed with
705 0x in additional to the usual decimal notation now also support octal
706 notation when the 0o prefix is used and binary notation if the 0b
709 * Various command line parameters and configuration file settings that
710 configure key or certificate files now optionally take paths to
711 AF_UNIX sockets in the file system. If configured that way a stream
712 connection is made to the socket and the required data read from
713 it. This is a simple and natural extension to the existing regular
714 file logic, and permits other software to provide keys or
715 certificates via simple IPC services, for example when unencrypted
716 storage on disk is not desired. Specifically, systemd-networkd's
717 Wireguard and MACSEC key file settings as well as
718 systemd-journal-gatewayd's and systemd-journal-remote's PEM
719 key/certificate parameters support this now.
721 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
722 configuration files that support specifier expansion learnt six new
723 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
724 resolve to the various ID fields from /etc/os-release, %l resolves to
725 the "short" hostname of the system, i.e. the hostname configured in
726 the kernel truncated at the first dot.
728 * Support for the .include syntax in unit files has been removed. The
729 concept has been obsolete for 6 years and we started warning about
730 its pending removal 2 years ago (also see NEWS file below). It's
733 * StandardError= and StandardOutput= in unit files no longer support
734 the "syslog" and "syslog-console" switches. They were long removed
735 from the documentation, but will now result in warnings when used,
736 and be converted to "journal" and "journal+console" automatically.
738 * If the service setting User= is set to the "nobody" user, a warning
739 message is now written to the logs (but the value is nonetheless
740 accepted). Setting User=nobody is unsafe, since the primary purpose
741 of the "nobody" user is to own all files whose owner cannot be mapped
742 locally. It's in particular used by the NFS subsystem and in user
743 namespacing. By running a service under this user's UID it might get
744 read and even write access to all these otherwise unmappable files,
745 which is quite likely a major security problem.
747 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
748 and others) now have a size and inode limits applied (50% of RAM for
749 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
750 that the implicit kernel default is 50% too, so there is no change
751 in the size limit for /tmp and /dev/shm.
753 * nss-mymachines lost support for resolution of users and groups, and
754 now only does resolution of hostnames. This functionality is now
755 provided by nss-systemd. Thus, the 'mymachines' entry should be
756 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
757 (and 'systemd' added if it is not already there).
759 * A new kernel command line option systemd.hostname= has been added
760 that allows controlling the hostname that is initialized early during
763 * A kernel command line option "udev.blockdev_read_only" has been
764 added. If specified all hardware block devices that show up are
765 immediately marked as read-only by udev. This option is useful for
766 making sure that a specific boot under no circumstances modifies data
767 on disk. Use "blockdev --setrw" to undo the effect of this, per
770 * A new boolean kernel command line option systemd.swap= has been
771 added, which may be used to turn off automatic activation of swap
772 devices listed in /etc/fstab.
774 * New kernel command line options systemd.condition-needs-update= and
775 systemd.condition-first-boot= have been added, which override the
776 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
779 * A new kernel command line option systemd.clock-usec= has been added
780 that allows setting the system clock to the specified time in µs
781 since Jan 1st, 1970 early during boot. This is in particular useful
782 in order to make test cases more reliable.
784 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
785 systemd-coredump to save core files for suid processes. When saving
786 the core file, systemd-coredump will use the effective uid and gid of
787 the process that faulted.
789 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
790 now automatically set to "Y" at boot, in order to enable pstore
791 generation for collection with systemd-pstore.
793 * We provide a set of udev rules to enable auto-suspend on PCI and USB
794 devices that were tested to correctly support it. Previously, this
795 was distributed as a set of udev rules, but has now been replaced by
796 by a set of hwdb entries (and a much shorter udev rule to take action
797 if the device modalias matches one of the new hwdb entries).
799 As before, entries are periodically imported from the database
800 maintained by the ChromiumOS project. If you have a device that
801 supports auto-suspend correctly and where it should be enabled by
802 default, please submit a patch that adds it to the database (see
803 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
805 * systemd-udevd gained the new configuration option timeout_signal= as well
806 as a corresponding kernel command line option udev.timeout_signal=.
807 The option can be used to configure the UNIX signal that the main
808 daemon sends to the worker processes on timeout. Setting the signal
809 to SIGABRT is useful for debugging.
811 * .link files managed by systemd-udevd gained options RxFlowControl=,
812 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
813 order to configure various flow control parameters. They also gained
814 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
815 frame ring buffer sizes.
817 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
818 enabled systemd-networkd manages all routes configured by other tools.
820 * .network files managed by systemd-networkd gained a new section
821 [SR-IOV], in order to configure SR-IOV capable network devices.
823 * systemd-networkd's [IPv6Prefix] section in .network files gained a
824 new boolean setting Assign=. If enabled an address from the prefix is
825 automatically assigned to the interface.
827 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
828 controls delegated prefixes assigned by DHCPv6 client. The section
829 has three settings: SubnetID=, Assign=, and Token=. The setting
830 SubnetID= allows explicit configuration of the preferred subnet that
831 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
832 Assign= is enabled (which is the default) an address from any acquired
833 delegated prefix is automatically chosen and assigned to the
834 interface. The setting Token= specifies an optional address generation
837 * systemd-networkd's [Network] section gained a new setting
838 IPv4AcceptLocal=. If enabled the interface accepts packets with local
841 * systemd-networkd gained support for configuring the HTB queuing
842 discipline in the [HierarchyTokenBucket] and
843 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
844 be configured in the [PFIFO] section, "GRED" in
845 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
846 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
847 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
848 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
849 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
850 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
852 * systemd-networkd gained support for a new Termination= setting in the
853 [CAN] section for configuring the termination resistor. It also
854 gained a new ListenOnly= setting for controlling whether to only
855 listen on CAN interfaces, without interfering with traffic otherwise
856 (which is useful for debugging/monitoring CAN network
857 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
858 been added to configure various CAN-FD aspects.
860 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
861 When enabled, DHCPv6 will be attempted right-away without requiring an
862 Router Advertisement packet suggesting it first (i.e. without the 'M'
863 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
864 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
865 the RA packets suggest it.
867 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
868 which may be used to turn off use of the gateway information provided
869 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
870 used to configure how to process leases that lack a lifetime option.
872 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
873 setting SendVendorOption= allowing configuration of additional vendor
874 options to send in the DHCP requests/responses. The [DHCPv6] section
875 gained a new SendOption= setting for sending arbitrary DHCP
876 options. RequestOptions= has been added to request arbitrary options
877 from the server. UserClass= has been added to set the DHCP user class
880 * systemd-networkd's [DHCPServer] section gained a new set of options
881 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
882 information about these three protocols in the DHCP lease. It also
883 gained support for including "MUD" URLs ("Manufacturer Usage
884 Description"). Support for "MUD" URLs was also added to the LLDP
885 stack, configurable in the [LLDP] section in .network files.
887 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
888 mode. Also, the sections now support a new setting SourceMACAddress=.
890 * systemd-networkd's .netdev files now support a new setting
891 VLANProtocol= in the [Bridge] section that allows configuration of
892 the VLAN protocol to use.
894 * systemd-networkd supports a new Group= setting in the [Link] section
895 of the .network files, to control the link group.
897 * systemd-networkd's [Network] section gained a new
898 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
899 link local address is generated.
901 * A new default .network file is now shipped that matches TUN/TAP
902 devices that begin with "vt-" in their name. Such interfaces will
903 have IP routing onto the host links set up automatically. This is
904 supposed to be used by VM managers to trivially acquire a network
905 interface which is fully set up for host communication, simply by
906 carefully picking an interface name to use.
908 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
909 which sets the route priority for routes specified by the DHCP server.
911 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
912 which configures the vendor class information sent to DHCP server.
914 * The BlackList= settings in .network files' [DHCPv4] and
915 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
916 are still understood to provide compatibility.
918 * networkctl gained the new "forcerenew" command for forcing all DHCP
919 server clients to renew their lease. The interface "status" output
920 will now show numerous additional fields of information about an
921 interface. There are new "up" and "down" commands to bring specific
922 interfaces up or down.
924 * systemd-resolved's DNS= configuration option now optionally accepts a
925 port number (after ":") and a host name (after "#"). When the host
926 name is specified, the DNS-over-TLS certificate is validated to match
927 the specified hostname. Additionally, in case of IPv6 addresses, an
928 interface may be specified (after "%").
930 * systemd-resolved may be configured to forward single-label DNS names.
931 This is not standard-conformant, but may make sense in setups where
932 public DNS servers are not used.
934 * systemd-resolved's DNS-over-TLS support gained SNI validation.
936 * systemd-nspawn's --resolv-conf= switch gained a number of new
937 supported values. Specifically, options starting with "replace-" are
938 like those prefixed "copy-" but replace any existing resolv.conf
939 file. And options ending in "-uplink" and "-stub" can now be used to
940 propagate other flavours of resolv.conf into the container (as
941 defined by systemd-resolved).
943 * The various programs included in systemd can now optionally output
944 their log messages on stderr prefixed with a timestamp, controlled by
945 the $SYSTEMD_LOG_TIME environment variable.
947 * systemctl gained a new "-P" switch that is a shortcut for "--value
950 * "systemctl list-units" and "systemctl list-machines" no longer hide
951 their first output column with --no-legend. To hide the first column,
954 * "systemctl reboot" takes the option "--reboot-argument=".
955 The optional positional argument to "systemctl reboot" is now
956 being deprecated in favor of this option.
958 * systemd-run gained a new switch --slice-inherit. If specified the
959 unit it generates is placed in the same slice as the systemd-run
962 * systemd-journald gained support for zstd compression of large fields
963 in journal files. The hash tables in journal files have been hardened
964 against hash collisions. This is an incompatible change and means
965 that journal files created with new systemd versions are not readable
966 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
967 environment variable for systemd-journald.service is set to 0 this
968 new hardening functionality may be turned off, so that generated
969 journal files remain compatible with older journalctl
972 * journalctl will now include a clickable link in the default output for
973 each log message for which an URL with further documentation is
974 known. This is only supported on terminal emulators that support
975 clickable hyperlinks, and is turned off if a pager is used (since
976 "less" still doesn't support hyperlinks,
977 unfortunately). Documentation URLs may be included in log messages
978 either by including a DOCUMENTATION= journal field in it, or by
979 associating a journal message catalog entry with the log message's
980 MESSAGE_ID, which then carries a "Documentation:" tag.
982 * journald.conf gained a new boolean setting Audit= that may be used to
983 control whether systemd-journald will enable audit during
986 * when systemd-journald's log stream is broken up into multiple lines
987 because the PID of the sender changed this is indicated in the
988 generated log records via the _LINE_BREAK=pid-change field.
990 * journalctl's "-o cat" output mode will now show one or more journal
991 fields specified with --output-fields= instead of unconditionally
992 MESSAGE=. This is useful to retrieve a very specific set of fields
993 without any decoration.
995 * The sd-journal.h API gained two new functions:
996 sd_journal_enumerate_available_unique() and
997 sd_journal_enumerate_available_data() that operate like their
998 counterparts that lack the _available_ in the name, but skip items
999 that cannot be read and processed by the local implementation
1000 (i.e. are compressed in an unsupported format or such),
1002 * coredumpctl gained a new --file= switch, matching the same one in
1003 journalctl: a specific journal file may be specified to read the
1006 * coredumps collected by systemd-coredump may now be compressed using
1009 * systemd-binfmt gained a new switch --unregister for unregistering all
1010 registered entries at once. This is now invoked automatically at
1011 shutdown, so that binary formats registered with the "F" flag will
1012 not block clean file system unmounting.
1014 * systemd-notify's --pid= switch gained new values: "parent", "self",
1015 "auto" for controlling which PID to send to the service manager: the
1016 systemd-notify process' PID, or the one of the process invoking it.
1018 * systemd-logind's Session bus object learnt a new method call
1019 SetType() for temporarily updating the session type of an already
1020 allocated session. This is useful for upgrading tty sessions to
1021 graphical ones once a compositor is invoked.
1023 * systemd-socket-proxy gained a new switch --exit-idle-time= for
1024 configuring an exit-on-idle time.
1026 * systemd-repart's --empty= setting gained a new value "create". If
1027 specified a new empty regular disk image file is created under the
1028 specified name. Its size may be specified with the new --size=
1029 option. The latter is also supported without the "create" mode, in
1030 order to grow existing disk image files to the specified size. These
1031 two new options are useful when creating or manipulating disk images
1032 instead of operating on actual block devices.
1034 * systemd-repart drop-ins now support a new UUID= setting to control
1035 the UUID to assign to a newly created partition.
1037 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
1040 * systemd-repart's Label= setting now support the usual, simple
1041 specifier expansion.
1043 * systemd-homed's LUKS backend gained the ability to discard empty file
1044 system blocks automatically when the user logs out. This is enabled
1045 by default to ensure that home directories take minimal space when
1046 logged out but get full size guarantees when logged in. This may be
1047 controlled with the new --luks-offline-discard= switch to homectl.
1049 * If systemd-homed detects that /home/ is encrypted as a whole it will
1050 now default to the directory or subvolume backends instead of the
1051 LUKS backend, in order to avoid double encryption. The default
1052 storage and file system may now be configured explicitly, too, via
1053 the new /etc/systemd/homed.conf configuration file.
1055 * systemd-homed now supports unlocking home directories with FIDO2
1056 security tokens that support the 'hmac-secret' extension, in addition
1057 to the existing support for PKCS#11 security token unlocking
1058 support. Note that many recent hardware security tokens support both
1059 interfaces. The FIDO2 support is accessible via homectl's
1060 --fido2-device= option.
1062 * homectl's --pkcs11-uri= setting now accepts two special parameters:
1063 if "auto" is specified and only one suitable PKCS#11 security token
1064 is plugged in, its URL is automatically determined and enrolled for
1065 unlocking the home directory. If "list" is specified a brief table of
1066 suitable PKCS#11 security tokens is shown. Similar, the new
1067 --fido2-device= option also supports these two special values, for
1068 automatically selecting and listing suitable FIDO2 devices.
1070 * The /etc/crypttab tmp option now optionally takes an argument
1071 selecting the file system to use. Moreover, the default is now
1072 changed from ext2 to ext4.
1074 * There's a new /etc/crypttab option "keyfile-erase". If specified the
1075 key file listed in the same line is removed after use, regardless if
1076 volume activation was successful or not. This is useful if the key
1077 file is only acquired transiently at runtime and shall be erased
1078 before the system continues to boot.
1080 * There's also a new /etc/crypttab option "try-empty-password". If
1081 specified, before asking the user for a password it is attempted to
1082 unlock the volume with an empty password. This is useful for
1083 installing encrypted images whose password shall be set on first boot
1084 instead of at installation time.
1086 * systemd-cryptsetup will now attempt to load the keys to unlock
1087 volumes with automatically from files in
1088 /etc/cryptsetup-keys.d/<volume>.key and
1089 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
1091 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
1092 /etc/crypttab, during boot.
1094 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
1095 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
1098 * A new generator systemd-xdg-autostart-generator has been added. It
1099 generates systemd unit files from XDG autostart .desktop files, and
1100 may be used to let the systemd user instance manage services that are
1101 started automatically as part of the desktop session.
1103 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
1104 to query and change the firmware's 'reboot into firmware' setup flag.
1106 * systemd-firstboot gained a new switch --kernel-command-line= that may
1107 be used to initialize the /etc/kernel/cmdline file of the image. It
1108 also gained a new switch --root-password-hashed= which is like
1109 --root-password= but accepts a pre-hashed UNIX password as
1110 argument. The new option --delete-root-password may be used to unset
1111 any password for the root user (dangerous!). The --root-shell= switch
1112 may be used to control the shell to use for the root account. A new
1113 --force option may be used to override any already set settings with
1114 the parameters specified on the command line (by default, the tool
1115 will not override what has already been set before, i.e. is purely
1118 * systemd-firstboot gained support for a new --image= switch, which is
1119 similar to --root= but accepts the path to a disk image file, on
1120 which it then operates.
1122 * A new sd-path.h API has been added to libsystemd. It provides a
1123 simple API for retrieving various search paths and primary
1124 directories for various resources.
1126 * A new call sd_notify_barrier() has been added to the sd-daemon.h
1127 API. The call will block until all previously sent sd_notify()
1128 messages have been processed by the service manager. This is useful
1129 to remove races caused by a process already having disappeared at the
1130 time a notification message is processed by the service manager,
1131 making correct attribution impossible. The systemd-notify tool will
1132 now make use of this call implicitly, but this can be turned off again
1133 via the new --no-block switch.
1135 * When sending a file descriptor (fd) to the service manager to keep
1136 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
1137 may be specified. If passed the service manager will refrain from
1138 poll()ing on the file descriptor. Traditionally (and when the
1139 parameter is not specified), the service manager will poll it for
1140 POLLHUP or POLLERR events, and immediately close the fds in that
1143 * The service manager (PID1) gained a new D-Bus method call
1144 SetShowStatus() which may be used to control whether it shall show
1145 boot-time status output on the console. This method has a similar
1146 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
1148 * The sd-bus API gained a number of convenience functions that take
1149 va_list arguments rather than "...". For example, there's now
1150 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
1151 it easier to build wrappers that accept variadic arguments and want
1152 to pass a ready va_list structure to sd-bus.
1154 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
1155 flag which alters how the userdata pointer to pass to the callbacks
1156 is determined. When the flag is set, the offset field is converted
1157 as-is into a pointer, without adding it to the object pointer the
1158 vtable is associated with.
1160 * sd-bus now exposes four new functions:
1161 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
1162 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
1163 validate strings to check if they qualify as various D-Bus concepts.
1165 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
1166 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
1167 that simplify adding argument names to D-Bus methods and signals.
1169 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
1171 * Various D-Bus APIs of systemd daemons now have man pages that
1172 document the methods, signals and properties.
1174 * The expectations on user/group name syntax are now documented in
1175 detail; documentation on how classic home directories may be
1176 converted into home directories managed by homed has been added;
1177 documentation regarding integration of homed/userdb functionality in
1178 desktops has been added:
1180 https://systemd.io/USER_NAMES
1181 https://systemd.io/CONVERTING_TO_HOMED
1182 https://systemd.io/USERDB_AND_DESKTOPS
1184 * Documentation for the on-disk Journal file format has been updated
1185 and has now moved to:
1187 https://systemd.io/JOURNAL_FILE_FORMAT
1189 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
1190 has been extended by a set of environment variables that expose
1191 select fields from the host's os-release file to the container
1192 payload. Similarly, host's os-release files can be mounted into the
1193 container underneath /run/host. Together, those mechanisms provide a
1194 standardized way to expose information about the host to the
1195 container payload. Both interfaces are implemented in systemd-nspawn.
1197 * All D-Bus services shipped in systemd now implement the generic
1198 LogControl1 D-Bus API which allows clients to change log level +
1199 target of the service during runtime.
1201 * Only relevant for developers: the mkosi.default symlink has been
1202 dropped from version control. Please create a symlink to one of the
1203 distribution-specific defaults in .mkosi/ based on your preference.
1205 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
1206 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
1207 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
1208 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
1209 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
1210 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
1211 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
1212 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
1213 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
1214 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
1215 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
1216 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
1217 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
1218 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
1219 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
1220 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
1221 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
1222 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
1223 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
1224 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
1225 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
1226 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
1227 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
1228 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
1229 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
1230 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
1231 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
1232 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
1233 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
1234 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
1235 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
1236 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
1237 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
1238 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
1239 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
1240 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
1241 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
1242 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
1244 – Warsaw, 2020-07-30
1248 * A new tool "systemd-repart" has been added, that operates as an
1249 idempotent declarative repartitioner for GPT partition tables.
1250 Specifically, a set of partitions that must or may exist can be
1251 configured via drop-in files, and during every boot the partition
1252 table on disk is compared with these files, creating missing
1253 partitions or growing existing ones based on configurable relative
1254 and absolute size constraints. The tool is strictly incremental,
1255 i.e. does not delete, shrink or move partitions, but only adds and
1256 grows them. The primary use-case is OS images that ship in minimized
1257 form, that on first boot are grown to the size of the underlying
1258 block device or augmented with additional partitions. For example,
1259 the root partition could be extended to cover the whole disk, or a
1260 swap or /home partitions could be added on first boot. It can also be
1261 used for systems that use an A/B update scheme but ship images with
1262 just the A partition, with B added on first boot. The tool is
1263 primarily intended to be run in the initrd, shortly before
1264 transitioning into the host OS, but can also be run after the
1265 transition took place. It automatically discovers the disk backing
1266 the root file system, and should hence not require any additional
1267 configuration besides the partition definition drop-ins. If no
1268 configuration drop-ins are present, no action is taken.
1270 * A new component "userdb" has been added, along with a small daemon
1271 "systemd-userdbd.service" and a client tool "userdbctl". The framework
1272 allows defining rich user and group records in a JSON format,
1273 extending on the classic "struct passwd" and "struct group"
1274 structures. Various components in systemd have been updated to
1275 process records in this format, including systemd-logind and
1276 pam-systemd. The user records are intended to be extensible, and
1277 allow setting various resource management, security and runtime
1278 parameters that shall be applied to processes and sessions of the
1279 user as they log in. This facility is intended to allow associating
1280 such metadata directly with user/group records so that they can be
1281 produced, extended and consumed in unified form. We hope that
1282 eventually frameworks such as sssd will generate records this way, so
1283 that for the first time resource management and various other
1284 per-user settings can be configured in LDAP directories and then
1285 provided to systemd (specifically to systemd-logind and pam-system)
1286 to apply on login. For further details see:
1288 https://systemd.io/USER_RECORD
1289 https://systemd.io/GROUP_RECORD
1290 https://systemd.io/USER_GROUP_API
1292 * A small new service systemd-homed.service has been added, that may be
1293 used to securely manage home directories with built-in encryption.
1294 The complete user record data is unified with the home directory,
1295 thus making home directories naturally migratable. Its primary
1296 back-end is based on LUKS volumes, but fscrypt, plain directories,
1297 and other storage schemes are also supported. This solves a couple of
1298 problems we saw with traditional ways to manage home directories, in
1299 particular when it comes to encryption. For further discussion of
1300 this, see the video of Lennart's talk at AllSystemsGo! 2019:
1302 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
1304 For further details about the format and expectations on home
1305 directories this new daemon makes, see:
1307 https://systemd.io/HOME_DIRECTORY
1309 * systemd-journald is now multi-instantiable. In addition to the main
1310 instance systemd-journald.service there's now a template unit
1311 systemd-journald@.service, with each instance defining a new named
1312 log 'namespace' (whose name is specified via the instance part of the
1313 unit name). A new unit file setting LogNamespace= has been added,
1314 taking such a namespace name, that assigns services to the specified
1315 log namespaces. As each log namespace is serviced by its own
1316 independent journal daemon, this functionality may be used to improve
1317 performance and increase isolation of applications, at the price of
1318 losing global message ordering. Each instance of journald has a
1319 separate set of configuration files, with possibly different disk
1320 usage limitations and other settings.
1322 journalctl now takes a new option --namespace= to show logs from a
1323 specific log namespace. The sd-journal.h API gained
1324 sd_journal_open_namespace() for opening the log stream of a specific
1325 log namespace. systemd-journald also gained the ability to exit on
1326 idle, which is useful in the context of log namespaces, as this means
1327 log daemons for log namespaces can be activated automatically on
1328 demand and will stop automatically when no longer used, minimizing
1331 * When systemd-tmpfiles copies a file tree using the 'C' line type it
1332 will now label every copied file according to the SELinux database.
1334 * When systemd/PID 1 detects it is used in the initrd it will now boot
1335 into initrd.target rather than default.target by default. This should
1336 make it simpler to build initrds with systemd as for many cases the
1337 only difference between a host OS image and an initrd image now is
1338 the presence of the /etc/initrd-release file.
1340 * A new kernel command line option systemd.cpu_affinity= is now
1341 understood. It's equivalent to the CPUAffinity= option in
1342 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
1343 itself and the default for all other processes.
1345 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
1346 equivalent), the SELinux database is now reloaded, ensuring that
1347 sockets and other file system objects are generated taking the new
1348 database into account.
1350 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
1351 "quiet" has been changed to imply that instead of
1352 "systemd.show-status=auto". In this mode, only messages about errors
1353 and significant delays in boot are shown on the console.
1355 * The sd-event.h API gained native support for the new Linux "pidfd"
1356 concept. This permits watching processes using file descriptors
1357 instead of PID numbers, which fixes a number of races and makes
1358 process supervision more robust and efficient. All of systemd's
1359 components will now use pidfds if the kernel supports it for process
1360 watching, with the exception of PID 1 itself, unfortunately. We hope
1361 to move PID 1 to exclusively using pidfds too eventually, but this
1362 requires some more kernel work first. (Background: PID 1 watches
1363 processes using waitid() with the P_ALL flag, and that does not play
1364 together nicely with pidfds yet.)
1366 * Closely related to this, the sd-event.h API gained two new calls
1367 sd_event_source_send_child_signal() (for sending a signal to a
1368 watched process) and sd_event_source_get_child_process_own() (for
1369 marking a process so that it is killed automatically whenever the
1370 event source watching it is freed).
1372 * systemd-networkd gained support for configuring Token Bucket Filter
1373 (TBF) parameters in its qdisc configuration support. Similarly,
1374 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
1375 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
1377 * systemd-networkd gained support for Intermediate Functional Block
1378 (IFB) network devices.
1380 * systemd-networkd gained support for configuring multi-path IP routes,
1381 using the new MultiPathRoute= setting in the [Route] section.
1383 * systemd-networkd's DHCPv4 client has been updated to support a new
1384 SendDecline= option. If enabled, duplicate address detection is done
1385 after a DHCP offer is received from the server. If a conflict is
1386 detected, the address is declined. The DHCPv4 client also gained
1387 support for a new RouteMTUBytes= setting that allows to configure the
1388 MTU size to be used for routes generated from DHCPv4 leases.
1390 * The PrefixRoute= setting in systemd-networkd's [Address] section of
1391 .network files has been deprecated, and replaced by AddPrefixRoute=,
1392 with its sense inverted.
1394 * The Gateway= setting of [Route] sections of .network files gained
1395 support for a special new value "_dhcp". If set, the configured
1396 static route uses the gateway host configured via DHCP.
1398 * New User= and SuppressPrefixLength= settings have been implemented
1399 for the [RoutingPolicyRule] section of .network files to configure
1400 source routing based on UID ranges and prefix length, respectively.
1402 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
1403 D-Bus message object as "sensitive". Those objects are erased from
1404 memory when they are freed. This concept is intended to be used for
1405 messages that contain security sensitive data. A new flag
1406 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
1407 in sd-bus vtables, causing any incoming and outgoing messages of
1408 those methods to be implicitly marked as "sensitive".
1410 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
1411 contents of a message (or parts thereof) to standard output for
1414 * systemd-sysusers gained support for creating users with the primary
1415 group named differently than the user.
1417 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
1418 gained support for growing XFS partitions. Previously it supported
1419 only ext4 and btrfs partitions.
1421 * The support for /etc/crypttab gained a new x-initrd.attach option. If
1422 set, the specified encrypted volume is unlocked already in the
1423 initrd. This concept corresponds to the x-initrd.mount option in
1426 * systemd-cryptsetup gained native support for unlocking encrypted
1427 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
1428 encryption of volumes to YubiKeys. This is exposed in the new
1429 pkcs11-uri= option in /etc/crypttab.
1431 * The /etc/fstab support in systemd now supports two new mount options
1432 x-systemd.{required,wanted}-by=, for explicitly configuring the units
1433 that the specified mount shall be pulled in by, in place of
1434 the usual local-fs.target/remote-fs.target.
1436 * The https://systemd.io/ web site has been relaunched, directly
1437 populated with most of the documentation included in the systemd
1438 repository. systemd also acquired a new logo, thanks to Tobias
1441 * systemd-udevd gained support for managing "alternative" network
1442 interface names, as supported by new Linux kernels. For the first
1443 time this permits assigning multiple (and longer!) names to a network
1444 interface. systemd-udevd will now by default assign the names
1445 generated via all supported naming schemes to each interface. This
1446 may be further tweaked with .link files and the AlternativeName= and
1447 AlternativeNamesPolicy= settings. Other components of systemd have
1448 been updated to support the new alternative names wherever
1449 appropriate. For example, systemd-nspawn will now generate
1450 alternative interface names for the host-facing side of container
1451 veth links based on the full container name without truncation.
1453 * systemd-nspawn interface naming logic has been updated in another way
1454 too: if the main interface name (i.e. as opposed to new-style
1455 "alternative" names) based on the container name is truncated, a
1456 simple hashing scheme is used to give different interface names to
1457 multiple containers whose names all begin with the same prefix. Since
1458 this changes the primary interface names pointing to containers if
1459 truncation happens, the old scheme may still be requested by
1460 selecting an older naming scheme, via the net.naming-scheme= kernel
1461 command line option.
1463 * PrivateUsers= in service files now works in services run by the
1464 systemd --user per-user instance of the service manager.
1466 * A new per-service sandboxing option ProtectClock= has been added that
1467 locks down write access to the system clock. It takes away device
1468 node access to /dev/rtc as well as the system calls that set the
1469 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
1470 Note that this option does not affect access to auxiliary services
1471 that allow changing the clock, for example access to
1474 * The systemd-id128 tool gained a new "show" verb for listing or
1475 resolving a number of well-known UUIDs/128bit IDs, currently mostly
1476 GPT partition table types.
1478 * The Discoverable Partitions Specification has been updated to support
1479 /var and /var/tmp partition discovery. Support for this has been
1480 added to systemd-gpt-auto-generator. For details see:
1482 https://systemd.io/DISCOVERABLE_PARTITIONS
1484 * "systemctl list-unit-files" has been updated to show a new column
1485 with the suggested enablement state based on the vendor preset files
1486 for the respective units.
1488 * "systemctl" gained a new option "--with-dependencies". If specified
1489 commands such as "systemctl status" or "systemctl cat" will now show
1490 all specified units along with all units they depend on.
1492 * networkctl gained support for showing per-interface logs in its
1495 * systemd-networkd-wait-online gained support for specifying the maximum
1496 operational state to wait for, and to wait for interfaces to
1499 * The [Match] section of .link and .network files now supports a new
1500 option PermanentMACAddress= which may be used to check against the
1501 permanent MAC address of a network device even if a randomized MAC
1504 * The [TrafficControlQueueingDiscipline] section in .network files has
1505 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
1506 dropped from the individual setting names.
1508 * Any .link and .network files that have an empty [Match] section (this
1509 also includes empty and commented-out files) will now be
1510 rejected. systemd-udev and systemd-networkd started warning about
1511 such files in version 243.
1513 * systemd-logind will now validate access to the operation of changing
1514 the virtual terminal via a polkit action. By default, only users
1515 with at least one session on a local VT are granted permission.
1517 * When systemd sets up PAM sessions that invoked service processes
1518 shall run in, the pam_setcred() API is now invoked, thus permitting
1519 PAM modules to set additional credentials for the processes.
1521 * portablectl attach/detach verbs now accept --now and --enable options
1522 to combine attachment with enablement and invocation, or detachment
1523 with stopping and disablement.
1525 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
1526 fixed, which in turn exposed bugs in unit configuration of services
1527 which have Type=oneshot and should only run once, but do not have
1528 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
1529 service may be started again after exiting successfully, for example
1530 as a dependency in another transaction. Affected services included
1531 some internal systemd services (most notably
1532 systemd-vconsole-setup.service, which was updated to have
1533 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
1534 plymouth has been suitably updated or patched before upgrading to
1535 this systemd release. See
1536 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
1537 additional discussion.
1539 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
1540 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
1541 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
1542 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
1543 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
1544 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
1545 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
1546 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
1547 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
1548 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
1549 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
1550 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
1551 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
1552 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
1553 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
1554 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
1555 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
1556 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
1557 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
1558 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
1559 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
1560 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
1561 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
1562 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
1565 – Warsaw, 2020-03-06
1569 * Support for the cpuset cgroups v2 controller has been added.
1570 Processes may be restricted to specific CPUs using the new
1571 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
1572 AllowedMemoryNodes= setting.
1574 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
1575 now be configured using a new RestartKillSignal= setting. This
1576 allows units which signals to request termination to implement
1577 different behaviour when stopping in preparation for a restart.
1579 * "systemctl clean" may now be used also for socket, mount, and swap
1582 * systemd will also read configuration options from the EFI variable
1583 SystemdOptions. This may be used to configure systemd behaviour when
1584 modifying the kernel command line is inconvenient, but configuration
1585 on disk is read too late, for example for the options related to
1586 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
1587 set the EFI variable.
1589 * systemd will now disable printk ratelimits in early boot. This should
1590 allow us to capture more logs from the early boot phase where normal
1591 storage is not available and the kernel ring buffer is used for
1592 logging. Configuration on the kernel command line has higher priority
1593 and overrides the systemd setting.
1595 systemd programs which log to /dev/kmsg directly use internal
1596 ratelimits to prevent runaway logging. (Normally this is only used
1597 during early boot, so in practice this change has very little
1600 * Unit files now support top level dropin directories of the form
1601 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
1602 that affects all corresponding unit files.
1604 * systemctl gained support for 'stop --job-mode=triggering' which will
1605 stop the specified unit and any units which could trigger it.
1607 * Unit status display now includes units triggering and triggered by
1608 the unit being shown.
1610 * The RuntimeMaxSec= setting is now supported by scopes, not just
1611 .service units. This is particularly useful for PAM sessions which
1612 create a scope unit for the user login. systemd.runtime_max_sec=
1613 setting may used with the pam_systemd module to limit the duration
1614 of the PAM session, for example for time-limited logins.
1616 * A new @pkey system call group is now defined to make it easier to
1617 allow-list memory protection syscalls for containers and services
1618 which need to use them.
1620 * systemd-udevd: removed the 30s timeout for killing stale workers on
1621 exit. systemd-udevd now waits for workers to finish. The hard-coded
1622 exit timeout of 30s was too short for some large installations, where
1623 driver initialization could be prematurely interrupted during initrd
1624 processing if the root file system had been mounted and init was
1625 preparing to switch root. If udevd is run without systemd and workers
1626 are hanging while udevd receives an exit signal, udevd will now exit
1627 when udev.event_timeout is reached for the last hanging worker. With
1628 systemd, the exit timeout can additionally be configured using
1629 TimeoutStopSec= in systemd-udevd.service.
1631 * udev now provides a program (fido_id) that identifies FIDO CTAP1
1632 ("U2F")/CTAP2 security tokens based on the usage declared in their
1633 report and descriptor and outputs suitable environment variables.
1634 This replaces the externally maintained allow lists of all known
1635 security tokens that were used previously.
1637 * Automatically generated autosuspend udev rules for allow-listed
1638 devices have been imported from the Chromium OS project. This should
1639 improve power saving with many more devices.
1641 * udev gained a new "CONST{key}=value" setting that allows matching
1642 against system-wide constants without forking a helper binary.
1643 Currently "arch" and "virt" keys are supported.
1645 * udev now opens CDROMs in non-exclusive mode when querying their
1646 capabilities. This should fix issues where other programs trying to
1647 use the CDROM cannot gain access to it, but carries a risk of
1648 interfering with programs writing to the disk, if they did not open
1649 the device in exclusive mode as they should.
1651 * systemd-networkd does not create a default route for IPv4 link local
1652 addressing anymore. The creation of the route was unexpected and was
1653 breaking routing in various cases, but people who rely on it being
1654 created implicitly will need to adjust. Such a route may be requested
1655 with DefaultRouteOnDevice=yes.
1657 Similarly, systemd-networkd will not assign a link-local IPv6 address
1658 when IPv6 link-local routing is not enabled.
1660 * Receive and transmit buffers may now be configured on links with
1661 the new RxBufferSize= and TxBufferSize= settings.
1663 * systemd-networkd may now advertise additional IPv6 routes. A new
1664 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
1667 * systemd-networkd may now configure "next hop" routes using the
1668 [NextHop] section and Gateway= and Id= settings.
1670 * systemd-networkd will now retain DHCP config on restarts by default
1671 (but this may be overridden using the KeepConfiguration= setting).
1672 The default for SendRelease= has been changed to true.
1674 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
1675 received from the server.
1677 The client will use the received SIP server list if UseSIP=yes is
1680 The client may be configured to request specific options from the
1681 server using a new RequestOptions= setting.
1683 The client may be configured to send arbitrary options to the server
1684 using a new SendOption= setting.
1686 A new IPServiceType= setting has been added to configure the "IP
1687 service type" value used by the client.
1689 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
1690 request prefix hints in the DHCPv6 solicitation.
1692 * The DHCPv4 server may be configured to send arbitrary options using
1693 a new SendOption= setting.
1695 * The DHCPv4 server may now be configured to emit SIP server list using
1696 the new EmitSIP= and SIP= settings.
1698 * systemd-networkd and networkctl may now renew DHCP leases on demand.
1699 networkctl has a new 'networkctl renew' verb.
1701 * systemd-networkd may now reconfigure links on demand. networkctl
1702 gained two new verbs: "reload" will reload the configuration, and
1703 "reconfigure DEVICE…" will reconfigure one or more devices.
1705 * .network files may now match on SSID and BSSID of a wireless network,
1706 i.e. the access point name and hardware address using the new SSID=
1707 and BSSID= options. networkctl will display the current SSID and
1708 BSSID for wireless links.
1710 .network files may also match on the wireless network type using the
1711 new WLANInterfaceType= option.
1713 * systemd-networkd now includes default configuration that enables
1714 link-local addressing when connected to an ad-hoc wireless network.
1716 * systemd-networkd may configure the Traffic Control queueing
1717 disciplines in the kernel using the new
1718 [TrafficControlQueueingDiscipline] section and Parent=,
1719 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
1720 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
1721 NetworkEmulatorDuplicateRate= settings.
1723 * systemd-tmpfiles gained a new w+ setting to append to files.
1725 * systemd-analyze dump will now report when the memory configuration in
1726 the kernel does not match what systemd has configured (usually,
1727 because some external program has modified the kernel configuration
1730 * systemd-analyze gained a new --base-time= switch instructs the
1731 'calendar' verb to resolve times relative to that timestamp instead
1732 of the present time.
1734 * journalctl --update-catalog now produces deterministic output (making
1735 reproducible image builds easier).
1737 * A new devicetree-overlay setting is now documented in the Boot Loader
1740 * The default value of the WatchdogSec= setting used in systemd
1741 services (the ones bundled with the project itself) may be set at
1742 configuration time using the -Dservice-watchdog= setting. If set to
1743 empty, the watchdogs will be disabled.
1745 * systemd-resolved validates IP addresses in certificates now when GnuTLS
1748 * libcryptsetup >= 2.0.1 is now required.
1750 * A configuration option -Duser-path= may be used to override the $PATH
1751 used by the user service manager. The default is again to use the same
1752 path as the system manager.
1754 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
1755 outputting the 128bit IDs in UUID format (i.e. in the "canonical
1758 * Service units gained a new sandboxing option ProtectKernelLogs= which
1759 makes sure the program cannot get direct access to the kernel log
1760 buffer anymore, i.e. the syslog() system call (not to be confused
1761 with the API of the same name in libc, which is not affected), the
1762 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
1763 inaccessible to the service. It's recommended to enable this setting
1764 for all services that should not be able to read from or write to the
1765 kernel log buffer, which are probably almost all.
1767 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
1768 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
1769 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
1770 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
1771 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
1772 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
1773 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
1774 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
1775 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
1776 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
1777 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
1778 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
1779 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
1780 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
1781 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
1782 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
1783 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
1784 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
1785 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
1786 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
1787 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
1788 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
1789 Zach Smith, Zbigniew Jędrzejewski-Szmek
1791 – Warsaw, 2019-11-29
1795 * This release enables unprivileged programs (i.e. requiring neither
1796 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
1797 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
1798 kernel for the whole UNIX group range, i.e. all processes. This
1799 change should be reasonably safe, as the kernel support for it was
1800 specifically implemented to allow safe access to ICMP Echo for
1801 processes lacking any privileges. If this is not desirable, it can be
1802 disabled again by setting the parameter to "1 0".
1804 * Previously, filters defined with SystemCallFilter= would have the
1805 effect that any calling of an offending system call would terminate
1806 the calling thread. This behaviour never made much sense, since
1807 killing individual threads of unsuspecting processes is likely to
1808 create more problems than it solves. With this release the default
1809 action changed from killing the thread to killing the whole
1810 process. For this to work correctly both a kernel version (>= 4.14)
1811 and a libseccomp version (>= 2.4.0) supporting this new seccomp
1812 action is required. If an older kernel or libseccomp is used the old
1813 behaviour continues to be used. This change does not affect any
1814 services that have no system call filters defined, or that use
1815 SystemCallErrorNumber= (and thus see EPERM or another error instead
1816 of being killed when calling an offending system call). Note that
1817 systemd documentation always claimed that the whole process is
1818 killed. With this change behaviour is thus adjusted to match the
1821 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
1822 4194304 by default, i.e. the full 22bit range the kernel allows, up
1823 from the old 16bit range. This should improve security and
1824 robustness, as PID collisions are made less likely (though certainly
1825 still possible). There are rumours this might create compatibility
1826 problems, though at this moment no practical ones are known to
1827 us. Downstream distributions are hence advised to undo this change in
1828 their builds if they are concerned about maximum compatibility, but
1829 for everybody else we recommend leaving the value bumped. Besides
1830 improving security and robustness this should also simplify things as
1831 the maximum number of allowed concurrent tasks was previously bounded
1832 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
1833 only a single knob is left ("kernel.threads-max"). There have been
1834 concerns that usability is affected by this change because larger PID
1835 numbers are harder to type, but we believe the change from 5 digits
1836 to 7 digits doesn't hamper usability.
1838 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
1839 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
1840 hierarchically set default memory protection values for a particular
1841 subtree of the unit hierarchy.
1843 * Memory protection directives can now take a value of zero, allowing
1844 explicit opting out of a default value propagated by an ancestor.
1846 * systemd now defaults to the "unified" cgroup hierarchy setup during
1847 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
1848 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
1849 change reflects the fact that cgroupsv2 support has matured
1850 substantially in both systemd and in the kernel, and is clearly the
1851 way forward. Downstream production distributions might want to
1852 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
1853 their builds as unfortunately the popular container managers have not
1854 caught up with the kernel API changes.
1856 * Man pages are not built by default anymore (html pages were already
1857 disabled by default), to make development builds quicker. When
1858 building systemd for a full installation with documentation, meson
1859 should be called with -Dman=true and/or -Dhtml=true as appropriate.
1860 The default was changed based on the assumption that quick one-off or
1861 repeated development builds are much more common than full optimized
1862 builds for installation, and people need to pass various other
1863 options to when doing "proper" builds anyway, so the gain from making
1864 development builds quicker is bigger than the one time disruption for
1867 Two scripts are created in the *build* directory to generate and
1868 preview man and html pages on demand, e.g.:
1870 build/man/man systemctl
1871 build/man/html systemd.index
1873 * libidn2 is used by default if both libidn2 and libidn are installed.
1874 Please use -Dlibidn=true if libidn is preferred.
1876 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
1877 big-endian machines. Before, bytes were written and read in native
1878 machine order as exposed by the native libc __cpu_mask interface.
1879 Now, little-endian order is always used (CPUs 0–7 are described by
1880 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
1881 This change fixes D-Bus calls that cross endianness boundary.
1883 The presentation format used for CPUAffinity= by "systemctl show" and
1884 "systemd-analyze dump" is changed to present CPU indices instead of
1885 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
1886 shown as CPUAffinity=03000000000000000000000000000… (on
1887 little-endian) or CPUAffinity=00000000000000300000000000000… (on
1888 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
1889 input format. The maximum integer that will be printed in the new
1890 format is 8191 (four digits), while the old format always used a very
1891 long number (with the length varying by architecture), so they can be
1892 unambiguously distinguished.
1894 * /usr/sbin/halt.local is no longer supported. Implementation in
1895 distributions was inconsistent and it seems this functionality was
1898 To replace this functionality, users should:
1899 - either define a new unit and make it a dependency of final.target
1900 (systemctl add-wants final.target my-halt-local.service)
1901 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
1902 and ensure that it accepts "halt", "poweroff", "reboot", and
1903 "kexec" as an argument, see the description in systemd-shutdown(8).
1905 * When a [Match] section in .link or .network file is empty (contains
1906 no match patterns), a warning will be emitted. Please add any "match
1907 all" pattern instead, e.g. OriginalName=* or Name=* in case all
1908 interfaces should really be matched.
1910 * A new setting NUMAPolicy= may be used to set process memory
1911 allocation policy. This setting can be specified in
1912 /etc/systemd/system.conf and hence will set the default policy for
1913 PID1. The default policy can be overridden on a per-service
1914 basis. The related setting NUMAMask= is used to specify NUMA node
1915 mask that should be associated with the selected policy.
1917 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
1918 generates when processes it manages are reaching their memory limits,
1919 and will place their units in a special state, and optionally kill or
1920 stop the whole unit.
1922 * The service manager will now expose bus properties for the IO
1923 resources used by units. This information is also shown in "systemctl
1924 status" now (for services that have IOAccounting=yes set). Moreover,
1925 the IO accounting data is included in the resource log message
1926 generated whenever a unit stops.
1928 * Units may now configure an explicit timeout to wait for when killed
1929 with SIGABRT, for example when a service watchdog is hit. Previously,
1930 the regular TimeoutStopSec= timeout was applied in this case too —
1931 now a separate timeout may be set using TimeoutAbortSec=.
1933 * Services may now send a special WATCHDOG=trigger message with
1934 sd_notify() to trigger an immediate "watchdog missed" event, and thus
1935 trigger service termination. This is useful both for testing watchdog
1936 handling, but also for defining error paths in services, that shall
1937 be handled the same way as watchdog events.
1939 * There are two new per-unit settings IPIngressFilterPath= and
1940 IPEgressFilterPath= which allow configuration of a BPF program
1941 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
1942 to apply to the IP packet ingress/egress path of all processes of a
1943 unit. This is useful to allow running systemd services with BPF
1944 programs set up externally.
1946 * systemctl gained a new "clean" verb for removing the state, cache,
1947 runtime or logs directories of a service while it is terminated. The
1948 new verb may also be used to remove the state maintained on disk for
1949 timer units that have Persistent= configured.
1951 * During the last phase of shutdown systemd will now automatically
1952 increase the log level configured in the "kernel.printk" sysctl so
1953 that any relevant loggable events happening during late shutdown are
1954 made visible. Previously, loggable events happening so late during
1955 shutdown were generally lost if the "kernel.printk" sysctl was set to
1956 high thresholds, as regular logging daemons are terminated at that
1957 time and thus nothing is written to disk.
1959 * If processes terminated during the last phase of shutdown do not exit
1960 quickly systemd will now show their names after a short time, to make
1961 debugging easier. After a longer timeout they are forcibly killed,
1964 * journalctl (and the other tools that display logs) will now highlight
1965 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
1966 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
1967 are now shown in blue color, to separate them visually from regular
1968 logs. References to configuration files are now turned into clickable
1969 links on terminals that support that.
1971 * systemd-journald will now stop logging to /var/log/journal during
1972 shutdown when /var/ is on a separate mount, so that it can be
1973 unmounted safely during shutdown.
1975 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
1977 * systemd-resolved "Cache=" configuration option in resolved.conf has
1978 been extended to also accept the 'no-negative' value. Previously,
1979 only a boolean option was allowed (yes/no), having yes as the
1980 default. If this option is set to 'no-negative', negative answers are
1981 not cached while the old cache heuristics are used positive answers.
1982 The default remains unchanged.
1984 * The predictable naming scheme for network devices now supports
1985 generating predictable names for "netdevsim" devices.
1987 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
1990 Those two changes form a new net.naming-policy-scheme= entry.
1991 Distributions which want to preserve naming stability may want to set
1992 the -Ddefault-net-naming-scheme= configuration option.
1994 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
1995 interfaces natively.
1997 * systemd-networkd's bridge FDB support now allows configuration of a
1998 destination address for each entry (Destination=), as well as the
1999 VXLAN VNI (VNI=), as well as an option to declare what an entry is
2000 associated with (AssociatedWith=).
2002 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
2003 option for configuring the maximum number of DHCP lease requests. It
2004 also learnt a new BlackList= option for deny-listing DHCP servers (a
2005 similar setting has also been added to the IPv6 RA client), as well
2006 as a SendRelease= option for configuring whether to send a DHCP
2007 RELEASE message when terminating.
2009 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
2010 separately in the [DHCPv4] and [DHCPv6] sections.
2012 * systemd-networkd's DHCP support will now optionally create an
2013 implicit host route to the DNS server specified in the DHCP lease, in
2014 addition to the routes listed explicitly in the lease. This should
2015 ensure that in multi-homed systems DNS traffic leaves the systems on
2016 the interface that acquired the DNS server information even if other
2017 routes such as default routes exist. This behaviour may be turned on
2018 with the new RoutesToDNS= option.
2020 * systemd-networkd's VXLAN support gained a new option
2021 GenericProtocolExtension= for enabling VXLAN Generic Protocol
2022 Extension support, as well as IPDoNotFragment= for setting the IP
2023 "Don't fragment" bit on outgoing packets. A similar option has been
2024 added to the GENEVE support.
2026 * In systemd-networkd's [Route] section you may now configure
2027 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
2028 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
2029 propagation. The Type= setting now supports local, broadcast,
2030 anycast, multicast, any, xresolve routes, too.
2032 * systemd-networkd's [Network] section learnt a new option
2033 DefaultRouteOnDevice= for automatically configuring a default route
2034 onto the network device.
2036 * systemd-networkd's bridging support gained two new options ProxyARP=
2037 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
2038 MulticastRouter= for configuring multicast routing behaviour. A new
2039 option MulticastIGMPVersion= may be used to change bridge's multicast
2040 Internet Group Management Protocol (IGMP) version.
2042 * systemd-networkd's FooOverUDP support gained the ability to configure
2043 local and peer IP addresses via Local= and Peer=. A new option
2044 PeerPort= may be used to configure the peer's IP port.
2046 * systemd-networkd's TUN support gained a new setting VnetHeader= for
2047 tweaking Generic Segment Offload support.
2049 * The address family for policy rules may be specified using the new
2050 Family= option in the [RoutingPolicyRule] section.
2052 * networkctl gained a new "delete" command for removing virtual network
2053 devices, as well as a new "--stats" switch for showing device
2056 * networkd.conf gained a new setting SpeedMeter= and
2057 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
2058 measured speed may be shown by 'networkctl status'.
2060 * "networkctl status" now displays MTU and queue lengths, and more
2061 detailed information about VXLAN and bridge devices.
2063 * systemd-networkd's .network and .link files gained a new Property=
2064 setting in the [Match] section, to match against devices with
2065 specific udev properties.
2067 * systemd-networkd's tunnel support gained a new option
2068 AssignToLoopback= for selecting whether to use the loopback device
2069 "lo" as underlying device.
2071 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
2072 been renamed to LinkLayerAddress=, and it now allows configuration of
2075 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
2076 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
2077 IPv6 configuration (static or DHCPv6) was found for a given
2078 interface. It will not touch the sysctl otherwise.
2080 * The order of entries is $PATH used by the user manager instance was
2081 changed to put bin/ entries before the corresponding sbin/ entries.
2082 It is recommended to not rely on this order, and only ever have one
2083 binary with a given name in the system paths under /usr.
2085 * A new tool systemd-network-generator has been added that may generate
2086 .network, .netdev and .link files from IP configuration specified on
2087 the kernel command line in the format used by Dracut.
2089 * The CriticalConnection= setting in .network files is now deprecated,
2090 and replaced by a new KeepConfiguration= setting which allows more
2091 detailed configuration of the IP configuration to keep in place.
2093 * systemd-analyze gained a few new verbs:
2095 - "systemd-analyze timestamp" parses and converts timestamps. This is
2096 similar to the existing "systemd-analyze calendar" command which
2097 does the same for recurring calendar events.
2099 - "systemd-analyze timespan" parses and converts timespans (i.e.
2100 durations as opposed to points in time).
2102 - "systemd-analyze condition" will parse and test ConditionXYZ=
2105 - "systemd-analyze exit-status" will parse and convert exit status
2106 codes to their names and back.
2108 - "systemd-analyze unit-files" will print a list of all unit
2109 file paths and unit aliases.
2111 * SuccessExitStatus=, RestartPreventExitStatus=, and
2112 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
2113 is equivalent to "65"). Those exit status name mappings may be
2114 displayed with the systemd-analyze exit-status verb describe above.
2116 * systemd-logind now exposes a per-session SetBrightness() bus call,
2117 which may be used to securely change the brightness of a kernel
2118 brightness device, if it belongs to the session's seat. By using this
2119 call unprivileged clients can make changes to "backlight" and "leds"
2120 devices securely with strict requirements on session membership.
2121 Desktop environments may use this to generically make brightness
2122 changes to such devices without shipping private SUID binaries or
2123 udev rules for that purpose.
2125 * "udevadm info" gained a --wait-for-initialization switch to wait for
2126 a device to be initialized.
2128 * systemd-hibernate-resume-generator will now look for resumeflags= on
2129 the kernel command line, which is similar to rootflags= and may be
2130 used to configure device timeout for the hibernation device.
2132 * sd-event learnt a new API call sd_event_source_disable_unref() for
2133 disabling and unref'ing an event source in a single function. A
2134 related call sd_event_source_disable_unrefp() has been added for use
2135 with gcc's cleanup extension.
2137 * The sd-id128.h public API gained a new definition
2138 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
2141 * "busctl introspect" gained a new switch --xml-interface for dumping
2142 XML introspection data unmodified.
2144 * PID 1 may now show the unit name instead of the unit description
2145 string in its status output during boot. This may be configured in
2146 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
2147 kernel command line option systemd.status_unit_format=.
2149 * PID 1 now understands a new option KExecWatchdogSec= in
2150 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
2151 Previously watchdog functionality was only available for regular
2152 reboots. The new setting defaults to off, because we don't know in
2153 the general case if the watchdog will be reset after kexec (some
2154 drivers do reset it, but not all), and the new userspace might not be
2155 configured to handle the watchdog.
2157 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
2158 RebootWatchdogSec= to more clearly communicate what it is about. The
2159 old name is still accepted for compatibility.
2161 * The systemd.debug_shell kernel command line option now optionally
2162 takes a tty name to spawn the debug shell on, which allows a
2163 different tty to be selected than the built-in default.
2165 * Service units gained a new ExecCondition= setting which will run
2166 before ExecStartPre= and either continue execution of the unit (for
2167 clean exit codes), stop execution without marking the unit failed
2168 (for exit codes 1 through 254), or stop execution and fail the unit
2169 (for exit code 255 or abnormal termination).
2171 * A new service systemd-pstore.service has been added that pulls data
2172 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
2175 * timedatectl gained new verbs for configuring per-interface NTP
2176 service configuration for systemd-timesyncd.
2178 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
2179 2019. (You can set non-UTF-8 locales though, if you know their name.)
2181 * If variable assignments in sysctl.d/ files are prefixed with "-" any
2182 failures to apply them are now ignored.
2184 * systemd-random-seed.service now optionally credits entropy when
2185 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
2186 true for the service to enable this behaviour, but please consult the
2187 documentation first, since this comes with a couple of caveats.
2189 * systemd-random-seed.service is now a synchronization point for full
2190 initialization of the kernel's entropy pool. Services that require
2191 /dev/urandom to be correctly initialized should be ordered after this
2194 * The systemd-boot boot loader has been updated to optionally maintain
2195 a random seed file in the EFI System Partition (ESP). During the boot
2196 phase, this random seed is read and updated with a new seed
2197 cryptographically derived from it. Another derived seed is passed to
2198 the OS. The latter seed is then credited to the kernel's entropy pool
2199 very early during userspace initialization (from PID 1). This allows
2200 systems to boot up with a fully initialized kernel entropy pool from
2201 earliest boot on, and thus entirely removes all entropy pool
2202 initialization delays from systems using systemd-boot. Special care
2203 is taken to ensure different seeds are derived on system images
2204 replicated to multiple systems. "bootctl status" will show whether
2205 a seed was received from the boot loader.
2207 * bootctl gained two new verbs:
2209 - "bootctl random-seed" will generate the file in ESP and an EFI
2210 variable to allow a random seed to be passed to the OS as described
2213 - "bootctl is-installed" checks whether systemd-boot is currently
2216 * bootctl will warn if it detects that boot entries are misconfigured
2217 (for example if the kernel image was removed without purging the
2220 * A new document has been added describing systemd's use and support
2221 for the kernel's entropy pool subsystem:
2223 https://systemd.io/RANDOM_SEEDS
2225 * When the system is hibernated the swap device to write the
2226 hibernation image to is now automatically picked from all available
2227 swap devices, preferring the swap device with the highest configured
2228 priority over all others, and picking the device with the most free
2229 space if there are multiple devices with the highest priority.
2231 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
2232 option that permits selecting the timeout how long to wait for a
2233 device with an encryption key before asking for the password.
2235 * IOWeight= has learnt to properly set the IO weight when using the
2236 BFQ scheduler officially found in kernels 5.0+.
2238 * A new mailing list has been created for reporting of security issues:
2239 systemd-security@redhat.com. For mode details, see
2240 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
2242 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
2243 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
2244 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
2245 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
2246 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
2247 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
2248 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
2249 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
2250 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
2251 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
2252 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
2253 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
2254 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
2255 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
2256 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
2257 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
2258 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
2259 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
2260 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
2261 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
2262 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
2263 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
2264 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
2265 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
2266 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
2267 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
2268 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
2269 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
2270 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
2271 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
2273 – Camerino, 2019-09-03
2277 * In .link files, MACAddressPolicy=persistent (the default) is changed
2278 to cover more devices. For devices like bridges, tun, tap, bond, and
2279 similar interfaces that do not have other identifying information,
2280 the interface name is used as the basis for persistent seed for MAC
2281 and IPv4LL addresses. The way that devices that were handled
2282 previously is not changed, and this change is about covering more
2283 devices then previously by the "persistent" policy.
2285 MACAddressPolicy=random may be used to force randomized MACs and
2286 IPv4LL addresses for a device if desired.
2288 Hint: the log output from udev (at debug level) was enhanced to
2289 clarify what policy is followed and which attributes are used.
2290 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
2291 may be used to view this.
2293 Hint: if a bridge interface is created without any slaves, and gains
2294 a slave later, then now the bridge does not inherit slave's MAC.
2295 To inherit slave's MAC, for example, create the following file:
2297 # /etc/systemd/network/98-bridge-inherit-mac.link
2302 MACAddressPolicy=none
2305 * The .device units generated by systemd-fstab-generator and other
2306 generators do not automatically pull in the corresponding .mount unit
2307 as a Wants= dependency. This means that simply plugging in the device
2308 will not cause the mount unit to be started automatically. But please
2309 note that the mount unit may be started for other reasons, in
2310 particular if it is part of local-fs.target, and any unit which
2311 (transitively) depends on local-fs.target is started.
2313 * networkctl list/status/lldp now accept globbing wildcards for network
2314 interface names to match against all existing interfaces.
2316 * The $PIDFILE environment variable is set to point the absolute path
2317 configured with PIDFile= for processes of that service.
2319 * The fallback DNS server list was augmented with Cloudflare public DNS
2320 servers. Use `-Ddns-servers=` to set a different fallback.
2322 * A new special target usb-gadget.target will be started automatically
2323 when a USB Device Controller is detected (which means that the system
2324 is a USB peripheral).
2326 * A new unit setting CPUQuotaPeriodSec= assigns the time period
2327 relatively to which the CPU time quota specified by CPUQuota= is
2330 * A new unit setting ProtectHostname= may be used to prevent services
2331 from modifying hostname information (even if they otherwise would
2332 have privileges to do so).
2334 * A new unit setting NetworkNamespacePath= may be used to specify a
2335 namespace for service or socket units through a path referring to a
2336 Linux network namespace pseudo-file.
2338 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
2339 have an effect on .socket units: when used the listening socket is
2340 created within the configured network namespace instead of the host
2343 * ExecStart= command lines in unit files may now be prefixed with ':'
2344 in which case environment variable substitution is
2345 disabled. (Supported for the other ExecXYZ= settings, too.)
2347 * .timer units gained two new boolean settings OnClockChange= and
2348 OnTimezoneChange= which may be used to also trigger a unit when the
2349 system clock is changed or the local timezone is
2350 modified. systemd-run has been updated to make these options easily
2351 accessible from the command line for transient timers.
2353 * Two new conditions for units have been added: ConditionMemory= may be
2354 used to conditionalize a unit based on installed system
2355 RAM. ConditionCPUs= may be used to conditionalize a unit based on
2356 installed CPU cores.
2358 * The @default system call filter group understood by SystemCallFilter=
2359 has been updated to include the new rseq() system call introduced in
2362 * A new time-set.target has been added that indicates that the system
2363 time has been set from a local source (possibly imprecise). The
2364 existing time-sync.target is stronger and indicates that the time has
2365 been synchronized with a precise external source. Services where
2366 approximate time is sufficient should use the new target.
2368 * "systemctl start" (and related commands) learnt a new
2369 --show-transaction option. If specified brief information about all
2370 jobs queued because of the requested operation is shown.
2372 * systemd-networkd recognizes a new operation state 'enslaved', used
2373 (instead of 'degraded' or 'carrier') for interfaces which form a
2374 bridge, bond, or similar, and an new 'degraded-carrier' operational
2375 state used for the bond or bridge master interface when one of the
2376 enslaved devices is not operational.
2378 * .network files learnt the new IgnoreCarrierLoss= option for leaving
2379 networks configured even if the carrier is lost.
2381 * The RequiredForOnline= setting in .network files may now specify a
2382 minimum operational state required for the interface to be considered
2383 "online" by systemd-networkd-wait-online. Related to this
2384 systemd-networkd-wait-online gained a new option --operational-state=
2385 to configure the same, and its --interface= option was updated to
2386 optionally also take an operational state specific for an interface.
2388 * systemd-networkd-wait-online gained a new setting --any for waiting
2389 for only one of the requested interfaces instead of all of them.
2391 * systemd-networkd now implements L2TP tunnels.
2393 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
2394 may be used to cause autonomous and onlink prefixes received in IPv6
2395 Router Advertisements to be ignored.
2397 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
2398 file settings may be used to tweak bridge behaviour.
2400 * The new TripleSampling= option in .network files may be used to
2401 configure CAN triple sampling.
2403 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
2404 used to point to private or preshared key for a WireGuard interface.
2406 * /etc/crypttab now supports the same-cpu-crypt and
2407 submit-from-crypt-cpus options to tweak encryption work scheduling
2410 * systemd-tmpfiles will now take a BSD file lock before operating on a
2411 contents of directory. This may be used to temporarily exclude
2412 directories from aging by taking the same lock (useful for example
2413 when extracting a tarball into /tmp or /var/tmp as a privileged user,
2414 which might create files with really old timestamps, which
2415 nevertheless should not be deleted). For further details, see:
2417 https://systemd.io/TEMPORARY_DIRECTORIES
2419 * systemd-tmpfiles' h line type gained support for the
2420 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
2421 controlling project quota inheritance.
2423 * sd-boot and bootctl now implement support for an Extended Boot Loader
2424 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
2425 addition to the ESP partition mounted to /efi or /boot/efi.
2426 Configuration file fragments, kernels, initrds and other EFI images
2427 to boot will be loaded from both the ESP and XBOOTLDR partitions.
2428 The XBOOTLDR partition was previously described by the Boot Loader
2429 Specification, but implementation was missing in sd-boot. Support for
2430 this concept allows using the sd-boot boot loader in more
2431 conservative scenarios where the boot loader itself is placed in the
2432 ESP but the kernels to boot (and their metadata) in a separate
2435 * A system may now be booted with systemd.volatile=overlay on the
2436 kernel command line, which causes the root file system to be set up
2437 an overlayfs mount combining the root-only root directory with a
2438 writable tmpfs. In this setup, the underlying root device is not
2439 modified, and any changes are lost at reboot.
2441 * Similar, systemd-nspawn can now boot containers with a volatile
2442 overlayfs root with the new --volatile=overlay switch.
2444 * systemd-nspawn can now consume OCI runtime bundles using a new
2445 --oci-bundle= option. This implementation is fully usable, with most
2446 features in the specification implemented, but since this a lot of
2447 new code and functionality, this feature should most likely not
2448 be used in production yet.
2450 * systemd-nspawn now supports various options described by the OCI
2451 runtime specification on the command-line and in .nspawn files:
2452 --inaccessible=/Inaccessible= may be used to mask parts of the file
2453 system tree, --console=/--pipe may be used to configure how standard
2454 input, output, and error are set up.
2456 * busctl learned the `emit` verb to generate D-Bus signals.
2458 * systemd-analyze cat-config may be used to gather and display
2459 configuration spread over multiple files, for example system and user
2460 presets, tmpfiles.d, sysusers.d, udev rules, etc.
2462 * systemd-analyze calendar now takes an optional new parameter
2463 --iterations= which may be used to show a maximum number of iterations
2464 the specified expression will elapse next.
2466 * The sd-bus C API gained support for naming method parameters in the
2469 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
2470 the reboot() system call expects.
2472 * journalctl learnt a new --cursor-file= option that points to a file
2473 from which a cursor should be loaded in the beginning and to which
2474 the updated cursor should be stored at the end.
2476 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
2477 detected by systemd-detect-virt (and may also be used in
2478 ConditionVirtualization=).
2480 * The behaviour of systemd-logind may now be modified with environment
2481 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
2482 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
2483 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
2484 skip the relevant operation completely (when set to false), or to
2485 create a flag file in /run/systemd (when set to true), instead of
2486 actually commencing the real operation when requested. The presence
2487 of /run/systemd/reboot-to-firmware-setup,
2488 /run/systemd/reboot-to-boot-loader-menu, and
2489 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
2490 boot loader implementations to replace some steps logind performs
2491 during reboot with their own operations.
2493 * systemctl can be used to request a reboot into the boot loader menu
2494 or a specific boot loader entry with the new --boot-load-menu= and
2495 --boot-loader-entry= options to a reboot command. (This requires a
2496 boot loader that supports this, for example sd-boot.)
2498 * kernel-install will no longer unconditionally create the output
2499 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
2500 snippets, but will do only if the machine-specific parent directory
2501 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
2502 to create this parent directory during sd-boot installation.
2504 This makes it easier to use kernel-install with plugins which support
2505 a different layout of the bootloader partitions (for example grub2).
2507 * During package installation (with `ninja install`), we would create
2508 symlinks for getty@tty1.service, systemd-networkd.service,
2509 systemd-networkd.socket, systemd-resolved.service,
2510 remote-cryptsetup.target, remote-fs.target,
2511 systemd-networkd-wait-online.service, and systemd-timesyncd.service
2512 in /etc, as if `systemctl enable` was called for those units, to make
2513 the system usable immediately after installation. Now this is not
2514 done anymore, and instead calling `systemctl preset-all` is
2515 recommended after the first installation of systemd.
2517 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
2518 is built on seccomp. When turned on creation of SUID/SGID files is
2521 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
2522 implied if DynamicUser= is turned on for a service. This hardens
2523 these services, so that they neither can benefit from nor create
2524 SUID/SGID executables. This is a minor compatibility breakage, given
2525 that when DynamicUser= was first introduced SUID/SGID behaviour was
2526 unaffected. However, the security benefit of these two options is
2527 substantial, and the setting is still relatively new, hence we opted
2528 to make it mandatory for services with dynamic users.
2530 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
2531 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
2532 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
2533 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
2534 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
2535 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
2536 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
2537 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
2538 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
2539 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
2540 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
2541 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
2542 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
2543 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
2544 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
2545 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
2546 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
2547 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2549 — Warsaw, 2019-04-11
2553 * The default locale can now be configured at compile time. Otherwise,
2554 a suitable default will be selected automatically (one of C.UTF-8,
2555 en_US.UTF-8, and C).
2557 * The version string shown by systemd and other tools now includes the
2558 git commit hash when built from git. An override may be specified
2559 during compilation, which is intended to be used by distributions to
2560 include the package release information.
2562 * systemd-cat can now filter standard input and standard error streams
2563 for different syslog priorities using the new --stderr-priority=
2566 * systemd-journald and systemd-journal-remote reject entries which
2567 contain too many fields (CVE-2018-16865) and set limits on the
2568 process' command line length (CVE-2018-16864).
2570 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
2573 * A new network device NamePolicy "keep" is implemented for link files,
2574 and used by default in 99-default.link (the fallback configuration
2575 provided by systemd). With this policy, if the network device name
2576 was already set by userspace, the device will not be renamed again.
2577 This matches the naming scheme that was implemented before
2578 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
2579 is also enabled by default, even if not specified. Effectively, this
2580 means that if naming-scheme >= 240 is specified, network devices will
2581 be renamed according to the configuration, even if they have been
2582 renamed already, if "keep" is not specified as the naming policy in
2583 the .link file. The 99-default.link file provided by systemd includes
2584 "keep" for backwards compatibility, but it is recommended for user
2585 installed .link files to *not* include it.
2587 The "kernel" policy, which keeps kernel names declared to be
2588 "persistent", now works again as documented.
2590 * kernel-install script now optionally takes the paths to one or more
2591 initrd files, and passes them to all plugins.
2593 * The mincore() system call has been dropped from the @system-service
2594 system call filter group, as it is pretty exotic and may potentially
2595 used for side-channel attacks.
2597 * -fPIE is dropped from compiler and linker options. Please specify
2598 -Db_pie=true option to meson to build position-independent
2599 executables. Note that the meson option is supported since meson-0.49.
2601 * The fs.protected_regular and fs.protected_fifos sysctls, which were
2602 added in Linux 4.19 to make some data spoofing attacks harder, are
2603 now enabled by default. While this will hopefully improve the
2604 security of most installations, it is technically a backwards
2605 incompatible change; to disable these sysctls again, place the
2606 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
2608 fs.protected_regular = 0
2609 fs.protected_fifos = 0
2611 Note that the similar hardlink and symlink protection has been
2612 enabled since v199, and may be disabled likewise.
2614 * The files read from the EnvironmentFile= setting in unit files now
2615 parse backslashes inside quotes literally, matching the behaviour of
2618 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
2619 now automatically become NOPs when run in a chroot() environment.
2621 * The tmpfiles.d/ "C" line type will now copy directory trees not only
2622 when the destination is so far missing, but also if it already exists
2623 as a directory and is empty. This is useful to cater for systems
2624 where directory trees are put together from multiple separate mount
2625 points but otherwise empty.
2627 * A new function sd_bus_close_unref() (and the associated
2628 sd_bus_close_unrefp()) has been added to libsystemd, that combines
2629 sd_bus_close() and sd_bus_unref() in one.
2631 * udevadm control learnt a new option for --ping for testing whether a
2632 systemd-udevd instance is running and reacting.
2634 * udevadm trigger learnt a new option for --wait-daemon for waiting
2635 systemd-udevd daemon to be initialized.
2637 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
2638 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
2639 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
2640 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
2641 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
2642 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
2643 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
2644 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
2645 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
2646 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
2647 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
2648 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
2649 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
2650 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
2651 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
2652 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
2653 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
2655 — Berlin, 2019-02-14
2659 * NoNewPrivileges=yes has been set for all long-running services
2660 implemented by systemd. Previously, this was problematic due to
2661 SELinux (as this would also prohibit the transition from PID1's label
2662 to the service's label). This restriction has since been lifted, but
2663 an SELinux policy update is required.
2664 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
2666 * DynamicUser=yes is dropped from systemd-networkd.service,
2667 systemd-resolved.service and systemd-timesyncd.service, which was
2668 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
2669 and since v236 for systemd-timesyncd.service. The users and groups
2670 systemd-network, systemd-resolve and systemd-timesync are created
2671 by systemd-sysusers again. Distributors or system administrators
2672 may need to create these users and groups if they not exist (or need
2673 to re-enable DynamicUser= for those units) while upgrading systemd.
2674 Also, the clock file for systemd-timesyncd may need to move from
2675 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
2677 * When unit files are loaded from disk, previously systemd would
2678 sometimes (depending on the unit loading order) load units from the
2679 target path of symlinks in .wants/ or .requires/ directories of other
2680 units. This meant that unit could be loaded from different paths
2681 depending on whether the unit was requested explicitly or as a
2682 dependency of another unit, not honouring the priority of directories
2683 in search path. It also meant that it was possible to successfully
2684 load and start units which are not found in the unit search path, as
2685 long as they were requested as a dependency and linked to from
2686 .wants/ or .requires/. The target paths of those symlinks are not
2687 used for loading units anymore and the unit file must be found in
2690 * A new service type has been added: Type=exec. It's very similar to
2691 Type=simple but ensures the service manager will wait for both fork()
2692 and execve() of the main service binary to complete before proceeding
2693 with follow-up units. This is primarily useful so that the manager
2694 propagates any errors in the preparation phase of service execution
2695 back to the job that requested the unit to be started. For example,
2696 consider a service that has ExecStart= set to a file system binary
2697 that doesn't exist. With Type=simple starting the unit would be
2698 considered instantly successful, as only fork() has to complete
2699 successfully and the manager does not wait for execve(), and hence
2700 its failure is seen "too late". With the new Type=exec service type
2701 starting the unit will fail, as the manager will wait for the
2702 execve() and notice its failure, which is then propagated back to the
2705 NOTE: with the next release 241 of systemd we intend to change the
2706 systemd-run tool to default to Type=exec for transient services
2707 started by it. This should be mostly safe, but in specific corner
2708 cases might result in problems, as the systemd-run tool will then
2709 block on NSS calls (such as user name look-ups due to User=) done
2710 between the fork() and execve(), which under specific circumstances
2711 might cause problems. It is recommended to specify "-p Type=simple"
2712 explicitly in the few cases where this applies. For regular,
2713 non-transient services (i.e. those defined with unit files on disk)
2714 we will continue to default to Type=simple.
2716 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
2717 userspace processes is set to 1024 (soft) and 4096
2718 (hard). Previously, systemd passed this on unmodified to all
2719 processes it forked off. With this systemd release the hard limit
2720 systemd passes on is increased to 512K, overriding the kernel's
2721 defaults and substantially increasing the number of simultaneous file
2722 descriptors unprivileged userspace processes can allocate. Note that
2723 the soft limit remains at 1024 for compatibility reasons: the
2724 traditional UNIX select() call cannot deal with file descriptors >=
2725 1024 and increasing the soft limit globally might thus result in
2726 programs unexpectedly allocating a high file descriptor and thus
2727 failing abnormally when attempting to use it with select() (of
2728 course, programs shouldn't use select() anymore, and prefer
2729 poll()/epoll, but the call unfortunately remains undeservedly popular
2730 at this time). This change reflects the fact that file descriptor
2731 handling in the Linux kernel has been optimized in more recent
2732 kernels and allocating large numbers of them should be much cheaper
2733 both in memory and in performance than it used to be. Programs that
2734 want to take benefit of the increased limit have to "opt-in" into
2735 high file descriptors explicitly by raising their soft limit. Of
2736 course, when they do that they must acknowledge that they cannot use
2737 select() anymore (and neither can any shared library they use — or
2738 any shared library used by any shared library they use and so on).
2739 Which default hard limit is most appropriate is of course hard to
2740 decide. However, given reports that ~300K file descriptors are used
2741 in real-life applications we believe 512K is sufficiently high as new
2742 default for now. Note that there are also reports that using very
2743 high hard limits (e.g. 1G) is problematic: some software allocates
2744 large arrays with one element for each potential file descriptor
2745 (Java, …) — a high hard limit thus triggers excessively large memory
2746 allocations in these applications. Hopefully, the new default of 512K
2747 is a good middle ground: higher than what real-life applications
2748 currently need, and low enough for avoid triggering excessively large
2749 allocations in problematic software. (And yes, somebody should fix
2752 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
2753 to the highest possible values, as separate accounting of file
2754 descriptors is no longer necessary, as memcg tracks them correctly as
2755 part of the memory accounting anyway. Thus, from the four limits on
2756 file descriptors currently enforced (fs.file-max, fs.nr_open,
2757 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
2758 and keep only the latter two. A set of build-time options
2759 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
2760 has been added to revert this change in behaviour, which might be
2761 an option for systems that turn off memcg in the kernel.
2763 * When no /etc/locale.conf file exists (and hence no locale settings
2764 are in place), systemd will now use the "C.UTF-8" locale by default,
2765 and set LANG= to it. This locale is supported by various
2766 distributions including Fedora, with clear indications that upstream
2767 glibc is going to make it available too. This locale enables UTF-8
2768 mode by default, which appears appropriate for 2018.
2770 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
2771 default. This effectively switches the RFC3704 Reverse Path filtering
2772 from Strict mode to Loose mode. This is more appropriate for hosts
2773 that have multiple links with routes to the same networks (e.g.
2774 a client with a Wi-Fi and Ethernet both connected to the internet).
2776 Consult the kernel documentation for details on this sysctl:
2777 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
2779 * CPUAccounting=yes no longer enables the CPU controller when using
2780 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
2781 statistics are now provided independently from the CPU controller.
2783 * Support for disabling a particular cgroup controller within a sub-tree
2784 has been added through the DisableControllers= directive.
2786 * cgroup_no_v1=all on the kernel command line now also implies
2787 using the unified cgroup hierarchy, unless one explicitly passes
2788 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
2790 * The new "MemoryMin=" unit file property may now be used to set the
2791 memory usage protection limit of processes invoked by the unit. This
2792 controls the cgroup v2 memory.min attribute. Similarly, the new
2793 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
2794 cgroup v2 io.latency cgroup property for configuring per-service I/O
2797 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
2798 to the cgroup v1 "devices" cgroup controller.
2800 * systemd-escape now is able to combine --unescape with --template. It
2801 also learnt a new option --instance for extracting and unescaping the
2802 instance part of a unit name.
2804 * sd-bus now provides the sd_bus_message_readv() which is similar to
2805 sd_bus_message_read() but takes a va_list object. The pair
2806 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
2807 has been added for configuring the default method call timeout to
2808 use. sd_bus_error_move() may be used to efficiently move the contents
2809 from one sd_bus_error structure to another, invalidating the
2810 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
2811 be used to control whether a bus connection object is automatically
2812 flushed when an sd-event loop is exited.
2814 * When processing classic BSD syslog log messages, journald will now
2815 save the original time-stamp string supplied in the new
2816 SYSLOG_TIMESTAMP= journal field. This permits consumers to
2817 reconstruct the original BSD syslog message more correctly.
2819 * StandardOutput=/StandardError= in service files gained support for
2820 new "append:…" parameters, for connecting STDOUT/STDERR of a service
2821 to a file, and appending to it.
2823 * The signal to use as last step of killing of unit processes is now
2824 configurable. Previously it was hard-coded to SIGKILL, which may now
2825 be overridden with the new KillSignal= setting. Note that this is the
2826 signal used when regular termination (i.e. SIGTERM) does not suffice.
2827 Similarly, the signal used when aborting a program in case of a
2828 watchdog timeout may now be configured too (WatchdogSignal=).
2830 * The XDG_SESSION_DESKTOP environment variable may now be configured in
2831 the pam_systemd argument line, using the new desktop= switch. This is
2832 useful to initialize it properly from a display manager without
2833 having to touch C code.
2835 * Most configuration options that previously accepted percentage values
2836 now also accept permille values with the '‰' suffix (instead of '%').
2838 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
2841 * systemd-resolved's configuration file resolved.conf gained a new
2842 option ReadEtcHosts= which may be used to turn off processing and
2843 honoring /etc/hosts entries.
2845 * The "--wait" switch may now be passed to "systemctl
2846 is-system-running", in which case the tool will synchronously wait
2847 until the system finished start-up.
2849 * hostnamed gained a new bus call to determine the DMI product UUID.
2851 * On x86-64 systemd will now prefer using the RDRAND processor
2852 instruction over /dev/urandom whenever it requires randomness that
2853 neither has to be crypto-grade nor should be reproducible. This
2854 should substantially reduce the amount of entropy systemd requests
2855 from the kernel during initialization on such systems, though not
2856 reduce it to zero. (Why not zero? systemd still needs to allocate
2857 UUIDs and such uniquely, which require high-quality randomness.)
2859 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
2860 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
2861 for forcing the "Other Information" bit in IPv6 RA messages. The
2862 bonding logic gained four new options AdActorSystemPriority=,
2863 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
2864 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
2865 shuffling of flows. The tunnel logic gained a new
2866 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
2867 Deployment. The policy rule logic gained four new options IPProtocol=,
2868 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
2869 support for the MulticastToUnicast= option. networkd also gained
2870 support for configuring static IPv4 ARP or IPv6 neighbor entries.
2872 * .preset files (as read by 'systemctl preset') may now be used to
2873 instantiate services.
2875 * /etc/crypttab now understands the sector-size= option to configure
2876 the sector size for an encrypted partition.
2878 * Key material for encrypted disks may now be placed on a formatted
2879 medium, and referenced from /etc/crypttab by the UUID of the file
2880 system, followed by "=" suffixed by the path to the key file.
2882 * The "collect" udev component has been removed without replacement, as
2883 it is neither used nor maintained.
2885 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
2886 LogsDirectory=, ConfigurationDirectory= settings are used in a
2887 service the executed processes will now receive a set of environment
2888 variables containing the full paths of these directories.
2889 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
2890 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
2891 are used. Note that these options may be used multiple times per
2892 service in which case the resulting paths will be concatenated and
2893 separated by colons.
2895 * Predictable interface naming has been extended to cover InfiniBand
2896 NICs. They will be exposed with an "ib" prefix.
2898 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
2899 which case the respective line failing is ignored.
2901 * .link files may now be used to configure the equivalent to the
2902 "ethtool advertise" commands.
2904 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
2905 alternative to libudev.h. Previously, the latter was just an internal
2906 wrapper around the former, but now these two APIs are exposed
2909 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
2910 which calculates an app-specific boot ID similar to how
2911 sd_id128_get_machine_app_specific() generates an app-specific machine
2914 * A new tool systemd-id128 has been added that can be used to determine
2915 and generate various 128bit IDs.
2917 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
2920 * systemd-hibernate-resume-generator will now honor the "noresume"
2921 kernel command line option, in which case it will bypass resuming
2922 from any hibernated image.
2924 * The systemd-sleep.conf configuration file gained new options
2925 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
2926 AllowHybridSleep= for prohibiting specific sleep modes even if the
2927 kernel exports them.
2929 * portablectl is now officially supported and has thus moved to
2932 * bootctl learnt the two new commands "set-default" and "set-oneshot"
2933 for setting the default boot loader item to boot to (either
2934 persistently or only for the next boot). This is currently only
2935 compatible with sd-boot, but may be implemented on other boot loaders
2936 too, that follow the boot loader interface. The updated interface is
2937 now documented here:
2939 https://systemd.io/BOOT_LOADER_INTERFACE
2941 * A new kernel command line option systemd.early_core_pattern= is now
2942 understood which may be used to influence the core_pattern PID 1
2943 installs during early boot.
2945 * busctl learnt two new options -j and --json= for outputting method
2946 call replies, properties and monitoring output in JSON.
2948 * journalctl's JSON output now supports simple ANSI coloring as well as
2949 a new "json-seq" mode for generating RFC7464 output.
2951 * Unit files now support the %g/%G specifiers that resolve to the UNIX
2952 group/GID of the service manager runs as, similar to the existing
2953 %u/%U specifiers that resolve to the UNIX user/UID.
2955 * systemd-logind learnt a new global configuration option
2956 UserStopDelaySec= that may be set in logind.conf. It specifies how
2957 long the systemd --user instance shall remain started after a user
2958 logs out. This is useful to speed up repetitive re-connections of the
2959 same user, as it means the user's service manager doesn't have to be
2960 stopped/restarted on each iteration, but can be reused between
2961 subsequent options. This setting defaults to 10s. systemd-logind also
2962 exports two new properties on its Manager D-Bus objects indicating
2963 whether the system's lid is currently closed, and whether the system
2966 * systemd gained support for a generic boot counting logic, which
2967 generically permits automatic reverting to older boot loader entries
2968 if newer updated ones don't work. The boot loader side is implemented
2969 in sd-boot, but is kept open for other boot loaders too. For details
2972 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
2974 * The SuccessAction=/FailureAction= unit file settings now learnt two
2975 new parameters: "exit" and "exit-force", which result in immediate
2976 exiting of the service manager, and are only useful in systemd --user
2977 and container environments.
2979 * Unit files gained support for a pair of options
2980 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
2981 exit status to use as service manager exit status when
2982 SuccessAction=/FailureAction= is set to exit or exit-force.
2984 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
2985 options may now be used to configure the log rate limiting applied by
2986 journald per-service.
2988 * systemd-analyze gained a new verb "timespan" for parsing and
2989 normalizing time span values (i.e. strings like "5min 7s 8us").
2991 * systemd-analyze also gained a new verb "security" for analyzing the
2992 security and sand-boxing settings of services in order to determine an
2993 "exposure level" for them, indicating whether a service would benefit
2994 from more sand-boxing options turned on for them.
2996 * "systemd-analyze syscall-filter" will now also show system calls
2997 supported by the local kernel but not included in any of the defined
3000 * .nspawn files now understand the Ephemeral= setting, matching the
3001 --ephemeral command line switch.
3003 * sd-event gained the new APIs sd_event_source_get_floating() and
3004 sd_event_source_set_floating() for controlling whether a specific
3005 event source is "floating", i.e. destroyed along with the even loop
3008 * Unit objects on D-Bus gained a new "Refs" property that lists all
3009 clients that currently have a reference on the unit (to ensure it is
3012 * The JoinControllers= option in system.conf is no longer supported, as
3013 it didn't work correctly, is hard to support properly, is legacy (as
3014 the concept only exists on cgroup v1) and apparently wasn't used.
3016 * Journal messages that are generated whenever a unit enters the failed
3017 state are now tagged with a unique MESSAGE_ID. Similarly, messages
3018 generated whenever a service process exits are now made recognizable,
3019 too. A tagged message is also emitted whenever a unit enters the
3020 "dead" state on success.
3022 * systemd-run gained a new switch --working-directory= for configuring
3023 the working directory of the service to start. A shortcut -d is
3024 equivalent, setting the working directory of the service to the
3025 current working directory of the invoking program. The new --shell
3026 (or just -S) option has been added for invoking the $SHELL of the
3027 caller as a service, and implies --pty --same-dir --wait --collect
3028 --service-type=exec. Or in other words, "systemd-run -S" is now the
3029 quickest way to quickly get an interactive in a fully clean and
3030 well-defined system service context.
3032 * machinectl gained a new verb "import-fs" for importing an OS tree
3033 from a directory. Moreover, when a directory or tarball is imported
3034 and single top-level directory found with the OS itself below the OS
3035 tree is automatically mangled and moved one level up.
3037 * systemd-importd will no longer set up an implicit btrfs loop-back
3038 file system on /var/lib/machines. If one is already set up, it will
3039 continue to be used.
3041 * A new generator "systemd-run-generator" has been added. It will
3042 synthesize a unit from one or more program command lines included in
3043 the kernel command line. This is very useful in container managers
3046 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
3048 This will run "systemd-nspawn" on an image, invoke the specified
3049 command line and immediately shut down the container again, returning
3050 the command line's exit code.
3052 * The block device locking logic is now documented:
3054 https://systemd.io/BLOCK_DEVICE_LOCKING
3056 * loginctl and machinectl now optionally output the various tables in
3057 JSON using the --output= switch. It is our intention to add similar
3058 support to systemctl and all other commands.
3060 * udevadm's query and trigger verb now optionally take a .device unit
3063 * systemd-udevd's network naming logic now understands a new
3064 net.naming-scheme= kernel command line switch, which may be used to
3065 pick a specific version of the naming scheme. This helps stabilizing
3066 interface names even as systemd/udev are updated and the naming logic
3069 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
3070 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
3071 initialize one to all 0xFF.
3073 * After loading the SELinux policy systemd will now recursively relabel
3074 all files and directories listed in
3075 /run/systemd/relabel-extra.d/*.relabel (which should be simple
3076 newline separated lists of paths) in addition to the ones it already
3077 implicitly relabels in /run, /dev and /sys. After the relabelling is
3078 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
3079 removed. This is useful to permit initrds (i.e. code running before
3080 the SELinux policy is in effect) to generate files in the host
3081 filesystem safely and ensure that the correct label is applied during
3082 the transition to the host OS.
3084 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
3085 mknod() handling in user namespaces. Previously mknod() would always
3086 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
3087 but device nodes generated that way cannot be opened, and attempts to
3088 open them result in EPERM. This breaks the "graceful fallback" logic
3089 in systemd's PrivateDevices= sand-boxing option. This option is
3090 implemented defensively, so that when systemd detects it runs in a
3091 restricted environment (such as a user namespace, or an environment
3092 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
3093 where device nodes cannot be created the effect of PrivateDevices= is
3094 bypassed (following the logic that 2nd-level sand-boxing is not
3095 essential if the system systemd runs in is itself already sand-boxed
3096 as a whole). This logic breaks with 4.18 in container managers where
3097 user namespacing is used: suddenly PrivateDevices= succeeds setting
3098 up a private /dev/ file system containing devices nodes — but when
3099 these are opened they don't work.
3101 At this point it is recommended that container managers utilizing
3102 user namespaces that intend to run systemd in the payload explicitly
3103 block mknod() with seccomp or similar, so that the graceful fallback
3106 We are very sorry for the breakage and the requirement to change
3107 container configurations for newer kernels. It's purely caused by an
3108 incompatible kernel change. The relevant kernel developers have been
3109 notified about this userspace breakage quickly, but they chose to
3112 * PermissionsStartOnly= setting is deprecated (but is still supported
3113 for backwards compatibility). The same functionality is provided by
3114 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
3117 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
3118 pam_systemd anymore.
3120 * The naming scheme for network devices was changed to always rename
3121 devices, even if they were already renamed by userspace. The "kernel"
3122 policy was changed to only apply as a fallback, if no other naming
3125 * The requirements to build systemd is bumped to meson-0.46 and
3128 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
3129 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
3130 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
3131 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
3132 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
3133 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
3134 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
3135 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
3136 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
3137 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
3138 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
3139 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
3140 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
3141 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
3142 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
3143 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
3144 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
3145 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
3146 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
3147 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
3148 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
3149 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
3150 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
3151 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
3152 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
3153 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
3154 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
3155 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
3156 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
3157 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
3158 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
3159 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
3160 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
3161 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
3162 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
3163 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
3164 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
3165 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
3166 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
3167 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
3168 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
3169 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
3170 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
3171 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
3172 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
3174 — Warsaw, 2018-12-21
3178 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
3179 builtin will name network interfaces differently than in previous
3180 versions for virtual network interfaces created with SR-IOV and NPAR
3181 and for devices where the PCI network controller device does not have
3182 a slot number associated.
3184 SR-IOV virtual devices are now named based on the name of the parent
3185 interface, with a suffix of "v<N>", where <N> is the virtual device
3186 number. Previously those virtual devices were named as if completely
3189 The ninth and later NPAR virtual devices will be named following the
3190 scheme used for the first eight NPAR partitions. Previously those
3191 devices were not renamed and the kernel default (eth<n>) was used.
3193 "net_id" will also generate names for PCI devices where the PCI
3194 network controller device does not have an associated slot number
3195 itself, but one of its parents does. Previously those devices were
3196 not renamed and the kernel default (eth<n>) was used.
3198 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
3199 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
3200 the unit. So, it is expected that the default behavior of
3201 systemd-logind is not changed. However, if distribution packagers or
3202 administrators disabled or modified IPAddressDeny= setting by a
3203 drop-in config file, then it may be necessary to update the file to
3204 re-enable AF_INET and AF_INET6 to support network user name services,
3207 * When the RestrictNamespaces= unit property is specified multiple
3208 times, then the specified types are merged now. Previously, only the
3209 last assignment was used. So, if distribution packagers or
3210 administrators modified the setting by a drop-in config file, then it
3211 may be necessary to update the file.
3213 * When OnFailure= is used in combination with Restart= on a service
3214 unit, then the specified units will no longer be triggered on
3215 failures that result in restarting. Previously, the specified units
3216 would be activated each time the unit failed, even when the unit was
3217 going to be restarted automatically. This behaviour contradicted the
3218 documentation. With this release the code is adjusted to match the
3221 * systemd-tmpfiles will now print a notice whenever it encounters
3222 tmpfiles.d/ lines referencing the /var/run/ directory. It will
3223 recommend reworking them to use the /run/ directory instead (for
3224 which /var/run/ is simply a symlinked compatibility alias). This way
3225 systemd-tmpfiles can properly detect line conflicts and merge lines
3226 referencing the same file by two paths, without having to access
3229 * systemctl disable/unmask/preset/preset-all cannot be used with
3230 --runtime. Previously this was allowed, but resulted in unintuitive
3231 behaviour that wasn't useful. systemctl disable/unmask will now undo
3232 both runtime and persistent enablement/masking, i.e. it will remove
3233 any relevant symlinks both in /run and /etc.
3235 * Note that all long-running system services shipped with systemd will
3236 now default to a system call allow list (rather than a deny list, as
3237 before). In particular, systemd-udevd will now enforce one too. For
3238 most cases this should be safe, however downstream distributions
3239 which disabled sandboxing of systemd-udevd (specifically the
3240 MountFlags= setting), might want to disable this security feature
3241 too, as the default allow-listing will prohibit all mount, swap,
3242 reboot and clock changing operations from udev rules.
3244 * sd-boot acquired new loader configuration settings to optionally turn
3245 off Windows and MacOS boot partition discovery as well as
3246 reboot-into-firmware menu items. It is also able to pick a better
3247 screen resolution for HiDPI systems, and now provides loader
3248 configuration settings to change the resolution explicitly.
3250 * systemd-resolved now supports DNS-over-TLS. It's still
3251 turned off by default, use DNSOverTLS=opportunistic to turn it on in
3252 resolved.conf. We intend to make this the default as soon as couple
3253 of additional techniques for optimizing the initial latency caused by
3254 establishing a TLS/TCP connection are implemented.
3256 * systemd-resolved.service and systemd-networkd.service now set
3257 DynamicUser=yes. The users systemd-resolve and systemd-network are
3258 not created by systemd-sysusers anymore.
3260 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
3261 that embed a network facing module into any process using getpwuid()
3262 or related call: the dynamic allocation of the user ID for
3263 systemd-resolved.service means the service manager has to check NSS
3264 if the user name is already taken when forking off the service. Since
3265 the user in the common case won't be defined in /etc/passwd the
3266 lookup is likely to trigger nss-ldap which in turn might use NSS to
3267 ask systemd-resolved for hostname lookups. This will hence result in
3268 a deadlock: a user name lookup in order to start
3269 systemd-resolved.service will result in a hostname lookup for which
3270 systemd-resolved.service needs to be started already. There are
3271 multiple ways to work around this problem: pre-allocate the
3272 "systemd-resolve" user on such systems, so that nss-ldap won't be
3273 triggered; or use a different NSS package that doesn't do networking
3274 in-process but provides a local asynchronous name cache; or configure
3275 the NSS package to avoid lookups for UIDs in the range `pkg-config
3276 systemd --variable=dynamicuidmin` … `pkg-config systemd
3277 --variable=dynamicuidmax`, so that it does not consider itself
3278 authoritative for the same UID range systemd allocates dynamic users
3281 * The systemd-resolve tool has been renamed to resolvectl (it also
3282 remains available under the old name, for compatibility), and its
3283 interface is now verb-based, similar in style to the other <xyz>ctl
3284 tools, such as systemctl or loginctl.
3286 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
3287 compatibility. It may be symlinked under the 'resolvconf' name, in
3288 which case it will take arguments and input compatible with the
3289 Debian and FreeBSD resolvconf tool.
3291 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
3292 where the system initially suspends, and after a timeout resumes and
3295 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
3296 set the client will only send a DUID as client identifier.
3298 * The nss-systemd glibc NSS module will now enumerate dynamic users and
3299 groups in effect. Previously, it could resolve UIDs/GIDs to user
3300 names/groups and vice versa, but did not support enumeration.
3302 * journald's Compress= configuration setting now optionally accepts a
3303 byte threshold value. All journal objects larger than this threshold
3304 will be compressed, smaller ones will not. Previously this threshold
3305 was not configurable and set to 512.
3307 * A new system.conf setting NoNewPrivileges= is now available which may
3308 be used to turn off acquisition of new privileges system-wide
3309 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
3310 for all its children). Note that turning this option on means setuid
3311 binaries and file system capabilities lose their special powers.
3312 While turning on this option is a big step towards a more secure
3313 system, doing so is likely to break numerous pre-existing UNIX tools,
3314 in particular su and sudo.
3316 * A new service systemd-time-sync-wait.service has been added. If
3317 enabled it will delay the time-sync.target unit at boot until time
3318 synchronization has been received from the network. This
3319 functionality is useful on systems lacking a local RTC or where it is
3320 acceptable that the boot process shall be delayed by external network
3323 * When hibernating, systemd will now inform the kernel of the image
3324 write offset, on kernels new enough to support this. This means swap
3325 files should work for hibernation now.
3327 * When loading unit files, systemd will now look for drop-in unit files
3328 extensions in additional places. Previously, for a unit file name
3329 "foo-bar-baz.service" it would look for dropin files in
3330 "foo-bar-baz.service.d/*.conf". Now, it will also look in
3331 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
3332 service name truncated after all inner dashes. This scheme allows
3333 writing drop-ins easily that apply to a whole set of unit files at
3334 once. It's particularly useful for mount and slice units (as their
3335 naming is prefix based), but is also useful for service and other
3336 units, for packages that install multiple unit files at once,
3337 following a strict naming regime of beginning the unit file name with
3338 the package's name. Two new specifiers are now supported in unit
3339 files to match this: %j and %J are replaced by the part of the unit
3340 name following the last dash.
3342 * Unit files and other configuration files that support specifier
3343 expansion now understand another three new specifiers: %T and %V will
3344 resolve to /tmp and /var/tmp respectively, or whatever temporary
3345 directory has been set for the calling user. %E will expand to either
3346 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
3348 * The ExecStart= lines of unit files are no longer required to
3349 reference absolute paths. If non-absolute paths are specified the
3350 specified binary name is searched within the service manager's
3351 built-in $PATH, which may be queried with 'systemd-path
3352 search-binaries-default'. It's generally recommended to continue to
3353 use absolute paths for all binaries specified in unit files.
3355 * Units gained a new load state "bad-setting", which is used when a
3356 unit file was loaded, but contained fatal errors which prevent it
3357 from being started (for example, a service unit has been defined
3358 lacking both ExecStart= and ExecStop= lines).
3360 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
3361 support alternative debuggers, for example lldb. The old name
3362 continues to be available however, for compatibility reasons. Use the
3363 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
3364 to pick an alternative debugger instead of the default gdb.
3366 * systemctl and the other tools will now output escape sequences that
3367 generate proper clickable hyperlinks in various terminal emulators
3368 where useful (for example, in the "systemctl status" output you can
3369 now click on the unit file name to quickly open it in the
3370 editor/viewer of your choice). Note that not all terminal emulators
3371 support this functionality yet, but many do. Unfortunately, the
3372 "less" pager doesn't support this yet, hence this functionality is
3373 currently automatically turned off when a pager is started (which
3374 happens quite often due to auto-paging). We hope to remove this
3375 limitation as soon as "less" learns these escape sequences. This new
3376 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
3377 environment variable. For details on these escape sequences see:
3378 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
3380 * networkd's .network files now support a new IPv6MTUBytes= option for
3381 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
3382 option in the [Route] section to configure the MTU to use for
3383 specific routes. It also gained support for configuration of the DHCP
3384 "UserClass" option through the new UserClass= setting. It gained
3385 three new options in the new [CAN] section for configuring CAN
3386 networks. The MULTICAST and ALLMULTI interface flags may now be
3387 controlled explicitly with the new Multicast= and AllMulticast=
3390 * networkd will now automatically make use of the kernel's route
3391 expiration feature, if it is available.
3393 * udevd's .link files now support setting the number of receive and
3394 transmit channels, using the RxChannels=, TxChannels=,
3395 OtherChannels=, CombinedChannels= settings.
3397 * Support for UDPSegmentationOffload= has been removed, given its
3398 limited support in hardware, and waning software support.
3400 * networkd's .netdev files now support creating "netdevsim" interfaces.
3402 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
3403 to query the unit belonging to a specific kernel control group.
3405 * systemd-analyze gained a new verb "cat-config", which may be used to
3406 dump the contents of any configuration file, with all its matching
3407 drop-in files added in, and honouring the usual search and masking
3408 logic applied to systemd configuration files. For example use
3409 "systemd-analyze cat-config systemd/system.conf" to get the complete
3410 system configuration file of systemd how it would be loaded by PID 1
3411 itself. Similar to this, various tools such as systemd-tmpfiles or
3412 systemd-sysusers, gained a new option "--cat-config", which does the
3413 corresponding operation for their own configuration settings. For
3414 example, "systemd-tmpfiles --cat-config" will now output the full
3415 list of tmpfiles.d/ lines in place.
3417 * timedatectl gained three new verbs: "show" shows bus properties of
3418 systemd-timedated, "timesync-status" shows the current NTP
3419 synchronization state of systemd-timesyncd, and "show-timesync"
3420 shows bus properties of systemd-timesyncd.
3422 * systemd-timesyncd gained a bus interface on which it exposes details
3425 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
3426 understood by systemd-timedated. It takes a colon-separated list of
3427 unit names of NTP client services. The list is used by
3428 "timedatectl set-ntp".
3430 * systemd-nspawn gained a new --rlimit= switch for setting initial
3431 resource limits for the container payload. There's a new switch
3432 --hostname= to explicitly override the container's hostname. A new
3433 --no-new-privileges= switch may be used to control the
3434 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
3435 --oom-score-adjust= switch controls the OOM scoring adjustment value
3436 for the payload. The new --cpu-affinity= switch controls the CPU
3437 affinity of the container payload. The new --resolv-conf= switch
3438 allows more detailed control of /etc/resolv.conf handling of the
3439 container. Similarly, the new --timezone= switch allows more detailed
3440 control of /etc/localtime handling of the container.
3442 * systemd-detect-virt gained a new --list switch, which will print a
3443 list of all currently known VM and container environments.
3445 * Support for "Portable Services" has been added, see
3446 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
3447 experimental, but this is expected to change soon. Reflecting this
3448 experimental state, the "portablectl" binary is not installed into
3449 /usr/bin yet. The binary has to be called with the full path
3450 /usr/lib/systemd/portablectl instead.
3452 * journalctl's and systemctl's -o switch now knows a new log output
3453 mode "with-unit". The output it generates is very similar to the
3454 regular "short" mode, but displays the unit name instead of the
3455 syslog tag for each log line. Also, the date is shown with timezone
3456 information. This mode is probably more useful than the classic
3457 "short" output mode for most purposes, except where pixel-perfect
3458 compatibility with classic /var/log/messages formatting is required.
3460 * A new --dump-bus-properties switch has been added to the systemd
3461 binary, which may be used to dump all supported D-Bus properties.
3462 (Options which are still supported, but are deprecated, are *not*
3465 * sd-bus gained a set of new calls:
3466 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
3467 enable/disable the "floating" state of a bus slot object,
3468 i.e. whether the slot object pins the bus it is allocated for into
3469 memory or if the bus slot object gets disconnected when the bus goes
3470 away. sd_bus_open_with_description(),
3471 sd_bus_open_user_with_description(),
3472 sd_bus_open_system_with_description() may be used to allocate bus
3473 objects and set their description string already during allocation.
3475 * sd-event gained support for watching inotify events from the event
3476 loop, in an efficient way, sharing inotify handles between multiple
3477 users. For this a new function sd_event_add_inotify() has been added.
3479 * sd-event and sd-bus gained support for calling special user-supplied
3480 destructor functions for userdata pointers associated with
3481 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
3482 functions sd_bus_slot_set_destroy_callback,
3483 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
3484 sd_bus_track_get_destroy_callback,
3485 sd_event_source_set_destroy_callback,
3486 sd_event_source_get_destroy_callback have been added.
3488 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
3490 * PID 1 will now automatically reschedule .timer units whenever the
3491 local timezone changes. (They previously got rescheduled
3492 automatically when the system clock changed.)
3494 * New documentation has been added to document cgroups delegation,
3495 portable services and the various code quality tools we have set up:
3497 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
3498 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
3499 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
3501 * The Boot Loader Specification has been added to the source tree.
3503 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
3505 While moving it into our source tree we have updated it and further
3506 changes are now accepted through the usual github PR workflow.
3508 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
3509 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
3510 earlier PAM modules. The data in these fields is used to initialize
3511 the session scope's resource properties. Thus external PAM modules
3512 may now configure per-session limits, for example sourced from
3513 external user databases.
3515 * socket units with Accept=yes will now maintain a "refused" counter in
3516 addition to the existing "accepted" counter, counting connections
3517 refused due to the enforced limits.
3519 * The "systemd-path search-binaries-default" command may now be use to
3520 query the default, built-in $PATH PID 1 will pass to the services it
3523 * A new unit file setting PrivateMounts= has been added. It's a boolean
3524 option. If enabled the unit's processes are invoked in their own file
3525 system namespace. Note that this behaviour is also implied if any
3526 other file system namespacing options (such as PrivateTmp=,
3527 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
3528 primarily useful for services that do not use any of the other file
3529 system namespacing options. One such service is systemd-udevd.service
3530 where this is now used by default.
3532 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
3533 when the system is booted in UEFI "secure mode".
3535 * A new unit "system-update-pre.target" is added, which defines an
3536 optional synchronization point for offline system updates, as
3537 implemented by the pre-existing "system-update.target" unit. It
3538 allows ordering services before the service that executes the actual
3539 update process in a generic way.
3541 * Systemd now emits warnings whenever .include syntax is used.
3543 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
3544 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
3545 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
3546 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
3547 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
3548 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
3549 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
3550 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
3551 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
3552 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
3553 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
3554 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
3555 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
3556 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
3557 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
3558 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
3559 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
3560 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
3561 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
3562 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
3563 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
3564 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
3565 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
3566 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
3567 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
3568 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
3569 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
3570 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
3571 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
3573 — Berlin, 2018-06-22
3577 * The MemoryAccounting= unit property now defaults to on. After
3578 discussions with the upstream control group maintainers we learnt
3579 that the negative impact of cgroup memory accounting on current
3580 kernels is finally relatively minimal, so that it should be safe to
3581 enable this by default without affecting system performance. Besides
3582 memory accounting only task accounting is turned on by default, all
3583 other forms of resource accounting (CPU, IO, IP) remain off for now,
3584 because it's not clear yet that their impact is small enough to move
3585 from opt-in to opt-out. We recommend downstreams to leave memory
3586 accounting on by default if kernel 4.14 or higher is primarily
3587 used. On very resource constrained systems or when support for old
3588 kernels is a necessity, -Dmemory-accounting-default=false can be used
3589 to revert this change.
3591 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
3592 %udev_rules_update) and the journal catalog (%journal_catalog_update)
3593 from the upgrade scriptlets of individual packages now do nothing.
3594 Transfiletriggers have been added which will perform those updates
3595 once at the end of the transaction.
3597 Similar transfiletriggers have been added to execute any sysctl.d
3598 and binfmt.d rules. Thus, it should be unnecessary to provide any
3599 scriptlets to execute this configuration from package installation
3602 * systemd-sysusers gained a mode where the configuration to execute is
3603 specified on the command line, but this configuration is not executed
3604 directly, but instead it is merged with the configuration on disk,
3605 and the result is executed. This is useful for package installation
3606 scripts which want to create the user before installing any files on
3607 disk (in case some of those files are owned by that user), while
3608 still allowing local admin overrides.
3610 This functionality is exposed to rpm scriptlets through a new
3611 %sysusers_create_package macro. Old %sysusers_create and
3612 %sysusers_create_inline macros are deprecated.
3614 A transfiletrigger for sysusers.d configuration is now installed,
3615 which means that it should be unnecessary to call systemd-sysusers from
3616 package installation scripts, unless the package installs any files
3617 owned by those newly-created users, in which case
3618 %sysusers_create_package should be used.
3620 * Analogous change has been done for systemd-tmpfiles: it gained a mode
3621 where the command-line configuration is merged with the configuration
3622 on disk. This is exposed as the new %tmpfiles_create_package macro,
3623 and %tmpfiles_create is deprecated. A transfiletrigger is installed
3624 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
3625 from package installation scripts.
3627 * sysusers.d configuration for a user may now also specify the group
3628 number, in addition to the user number ("u username 123:456"), or
3629 without the user number ("u username -:456").
3631 * Configution items for systemd-sysusers can now be specified as
3632 positional arguments when the new --inline switch is used.
3634 * The login shell of users created through sysusers.d may now be
3635 specified (previously, it was always /bin/sh for root and
3636 /sbin/nologin for other users).
3638 * systemd-analyze gained a new --global switch to look at global user
3639 configuration. It also gained a unit-paths verb to list the unit load
3640 paths that are compiled into systemd (which can be used with
3641 --systemd, --user, or --global).
3643 * udevadm trigger gained a new --settle/-w option to wait for any
3644 triggered events to finish (but just those, and not any other events
3645 which are triggered meanwhile).
3647 * The action that systemd-logind takes when the lid is closed and the
3648 machine is connected to external power can now be configured using
3649 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
3650 was determined by HandleLidSwitch=, and, for backwards compatibility,
3651 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
3653 * journalctl will periodically call sd_journal_process() to make it
3654 resilient against inotify queue overruns when journal files are
3655 rotated very quickly.
3657 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
3658 sd_bus_get_n_queued_write — may be used to check the number of
3659 pending bus messages.
3661 * systemd gained a new
3662 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
3663 which can be used to migrate foreign processes to scope and service
3664 units. The primary user for this new API is systemd itself: the
3665 systemd --user instance uses this call of the systemd --system
3666 instance to migrate processes if it itself gets the request to
3667 migrate processes and the kernel refuses this due to access
3668 restrictions. Thanks to this "systemd-run --scope --user …" works
3669 again in pure cgroup v2 environments when invoked from the user
3672 * A new TemporaryFileSystem= setting can be used to mask out part of
3673 the real file system tree with tmpfs mounts. This may be combined
3674 with BindPaths= and BindReadOnlyPaths= to hide files or directories
3675 not relevant to the unit, while still allowing some paths lower in
3676 the tree to be accessed.
3678 ProtectHome=tmpfs may now be used to hide user home and runtime
3679 directories from units, in a way that is mostly equivalent to
3680 "TemporaryFileSystem=/home /run/user /root".
3682 * Non-service units are now started with KeyringMode=shared by default.
3683 This means that mount and swapon and other mount tools have access
3684 to keys in the main keyring.
3686 * /sys/fs/bpf is now mounted automatically.
3688 * QNX virtualization is now detected by systemd-detect-virt and may
3689 be used in ConditionVirtualization=.
3691 * IPAccounting= may now be enabled also for slice units.
3693 * A new -Dsplit-bin= build configuration switch may be used to specify
3694 whether bin and sbin directories are merged, or if they should be
3695 included separately in $PATH and various listings of executable
3696 directories. The build configuration scripts will try to autodetect
3697 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
3698 system, but distributions are encouraged to configure this
3701 * A new -Dok-color= build configuration switch may be used to change
3702 the colour of "OK" status messages.
3704 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
3705 PrivateNetwork=yes was buggy in previous versions of systemd. This
3706 means that after the upgrade and daemon-reexec, any such units must
3709 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
3710 will not exclude read-only files owned by root from cleanup.
3712 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
3713 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
3714 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
3715 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
3716 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
3717 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
3718 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
3719 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
3720 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
3721 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
3722 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
3723 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
3724 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
3725 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
3726 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
3727 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
3729 — Warsaw, 2018-03-05
3733 * Some keyboards come with a zoom see-saw or rocker which until now got
3734 mapped to the Linux "zoomin/out" keys in hwdb. However, these
3735 keycodes are not recognized by any major desktop. They now produce
3736 Up/Down key events so that they can be used for scrolling.
3738 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
3739 slightly: previously, if an argument was specified for lines of this
3740 type (i.e. the right-most column was set) this string was appended to
3741 existing files each time systemd-tmpfiles was run. This behaviour was
3742 different from what the documentation said, and not particularly
3743 useful, as repeated systemd-tmpfiles invocations would not be
3744 idempotent and grow such files without bounds. With this release
3745 behaviour has been altered to match what the documentation says:
3746 lines of this type only have an effect if the indicated files don't
3747 exist yet, and only then the argument string is written to the file.
3749 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
3750 systemd-tmpfiles behaviour: previously, read-only files owned by root
3751 were always excluded from the file "aging" algorithm (i.e. the
3752 automatic clean-up of directories like /tmp based on
3753 atime/mtime/ctime). We intend to drop this restriction, and age files
3754 by default even when owned by root and read-only. This behaviour was
3755 inherited from older tools, but there have been requests to remove
3756 it, and it's not obvious why this restriction was made in the first
3757 place. Please speak up now, if you are aware of software that reqires
3758 this behaviour, otherwise we'll remove the restriction in v238.
3760 * A new environment variable $SYSTEMD_OFFLINE is now understood by
3761 systemctl. It takes a boolean argument. If on, systemctl assumes it
3762 operates on an "offline" OS tree, and will not attempt to talk to the
3763 service manager. Previously, this mode was implicitly enabled if a
3764 chroot() environment was detected, and this new environment variable
3765 now provides explicit control.
3767 * .path and .socket units may now be created transiently, too.
3768 Previously only service, mount, automount and timer units were
3769 supported as transient units. The systemd-run tool has been updated
3770 to expose this new functionality, you may hence use it now to bind
3771 arbitrary commands to path or socket activation on-the-fly from the
3772 command line. Moreover, almost all properties are now exposed for the
3773 unit types that already supported transient operation.
3775 * The systemd-mount command gained support for a new --owner= parameter
3776 which takes a user name, which is then resolved and included in uid=
3777 and gid= mount options string of the file system to mount.
3779 * A new unit condition ConditionControlGroupController= has been added
3780 that checks whether a specific cgroup controller is available.
3782 * Unit files, udev's .link files, and systemd-networkd's .netdev and
3783 .network files all gained support for a new condition
3784 ConditionKernelVersion= for checking against specific kernel
3787 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
3788 support for configuring device flags in the Flags= setting. In the
3789 same files, the [Tunnel] section gained support for configuring
3790 AllowLocalRemote=. The [Route] section in .network files gained
3791 support for configuring InitialCongestionWindow=,
3792 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
3793 understands RapidCommit=.
3795 * systemd-networkd's DHCPv6 support gained support for Prefix
3798 * sd-bus gained support for a new "watch-bind" feature. When this
3799 feature is enabled, an sd_bus connection may be set up to connect to
3800 an AF_UNIX socket in the file system as soon as it is created. This
3801 functionality is useful for writing early-boot services that
3802 automatically connect to the system bus as soon as it is started,
3803 without ugly time-based polling. systemd-networkd and
3804 systemd-resolved have been updated to make use of this
3805 functionality. busctl exposes this functionality in a new
3806 --watch-bind= command line switch.
3808 * sd-bus will now optionally synthesize a local "Connected" signal as
3809 soon as a D-Bus connection is set up fully. This message mirrors the
3810 already existing "Disconnected" signal which is synthesized when the
3811 connection is terminated. This signal is generally useful but
3812 particularly handy in combination with the "watch-bind" feature
3813 described above. Synthesizing of this message has to be requested
3814 explicitly through the new API call sd_bus_set_connected_signal(). In
3815 addition a new call sd_bus_is_ready() has been added that checks
3816 whether a connection is fully set up (i.e. between the "Connected" and
3817 "Disconnected" signals).
3819 * sd-bus gained two new calls sd_bus_request_name_async() and
3820 sd_bus_release_name_async() for asynchronously registering bus
3821 names. Similar, there is now sd_bus_add_match_async() for installing
3822 a signal match asynchronously. All of systemd's own services have
3823 been updated to make use of these calls. Doing these operations
3824 asynchronously has two benefits: it reduces the risk of deadlocks in
3825 case of cyclic dependencies between bus services, and it speeds up
3826 service initialization since synchronization points for bus
3827 round-trips are removed.
3829 * sd-bus gained two new calls sd_bus_match_signal() and
3830 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
3831 and sd_bus_add_match_async() but instead of taking a D-Bus match
3832 string take match fields as normal function parameters.
3834 * sd-bus gained two new calls sd_bus_set_sender() and
3835 sd_bus_message_set_sender() for setting the sender name of outgoing
3836 messages (either for all outgoing messages or for just one specific
3837 one). These calls are only useful in direct connections as on
3838 brokered connections the broker fills in the sender anyway,
3839 overwriting whatever the client filled in.
3841 * sd-event gained a new pseudo-handle that may be specified on all API
3842 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
3843 used this refers to the default event loop object of the calling
3844 thread. Note however that this does not implicitly allocate one —
3845 which has to be done prior by using sd_event_default(). Similarly
3846 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
3847 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
3848 to the default bus of the specified type of the calling thread. Here
3849 too this does not implicitly allocate bus connection objects, this
3850 has to be done prior with sd_bus_default() and friends.
3852 * sd-event gained a new call pair
3853 sd_event_source_{get|set}_io_fd_own(). This may be used to request
3854 automatic closure of the file descriptor an IO event source watches
3855 when the event source is destroyed.
3857 * systemd-networkd gained support for natively configuring WireGuard
3860 * In previous versions systemd synthesized user records both for the
3861 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
3862 internally. In order to simplify distribution-wide renames of the
3863 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
3864 new transitional flag file has been added: if
3865 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
3866 user and group record within the systemd codebase is disabled.
3868 * systemd-notify gained a new --uid= option for selecting the source
3869 user/UID to use for notification messages sent to the service
3872 * journalctl gained a new --grep= option to list only entries in which
3873 the message matches a certain pattern. By default matching is case
3874 insensitive if the pattern is lowercase, and case sensitive
3875 otherwise. Option --case-sensitive=yes|no can be used to override
3876 this an specify case sensitivity or case insensitivity.
3878 * There's now a "systemd-analyze service-watchdogs" command for printing
3879 the current state of the service runtime watchdog, and optionally
3880 enabling or disabling the per-service watchdogs system-wide if given a
3881 boolean argument (i.e. the concept you configure in WatchdogSec=), for
3882 debugging purposes. There's also a kernel command line option
3883 systemd.service_watchdogs= for controlling the same.
3885 * Two new "log-level" and "log-target" options for systemd-analyze were
3886 added that merge the now deprecated get-log-level, set-log-level and
3887 get-log-target, set-log-target pairs. The deprecated options are still
3888 understood for backwards compatibility. The two new options print the
3889 current value when no arguments are given, and set them when a
3890 level/target is given as an argument.
3892 * sysusers.d's "u" lines now optionally accept both a UID and a GID
3893 specification, separated by a ":" character, in order to create users
3894 where UID and GID do not match.
3896 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
3897 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
3898 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
3899 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
3900 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
3901 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
3902 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
3903 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
3904 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
3905 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
3906 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
3907 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
3908 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
3909 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
3910 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
3911 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
3912 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
3913 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
3914 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
3915 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
3922 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
3923 in v235 has been extended to also set the dummy.ko module option
3924 numdummies=0, preventing the kernel from automatically creating
3925 dummy0. All dummy interfaces must now be explicitly created.
3927 * Unknown '%' specifiers in configuration files are now rejected. This
3928 applies to units and tmpfiles.d configuration. Any percent characters
3929 that are followed by a letter or digit that are not supposed to be
3930 interpreted as the beginning of a specifier should be escaped by
3931 doubling ("%%"). (So "size=5%" is still accepted, as well as
3932 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
3933 valid specifiers today.)
3935 * systemd-resolved now maintains a new dynamic
3936 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
3937 recommended to make /etc/resolv.conf a symlink to it. This file
3938 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
3939 includes dynamically acquired search domains, achieving more correct
3940 DNS resolution by software that bypasses local DNS APIs such as NSS.
3942 * The "uaccess" udev tag has been dropped from /dev/kvm and
3943 /dev/dri/renderD*. These devices now have the 0666 permissions by
3944 default (but this may be changed at build-time). /dev/dri/renderD*
3945 will now be owned by the "render" group along with /dev/kfd.
3947 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
3948 systemd-journal-gatewayd.service and
3949 systemd-journal-upload.service. This means "nss-systemd" must be
3950 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
3951 services are resolved properly.
3953 * In /etc/fstab two new mount options are now understood:
3954 x-systemd.makefs and x-systemd.growfs. The former has the effect that
3955 the configured file system is formatted before it is mounted, the
3956 latter that the file system is resized to the full block device size
3957 after it is mounted (i.e. if the file system is smaller than the
3958 partition it resides on, it's grown). This is similar to the fsck
3959 logic in /etc/fstab, and pulls in systemd-makefs@.service and
3960 systemd-growfs@.service as necessary, similar to
3961 systemd-fsck@.service. Resizing is currently only supported on ext4
3964 * In systemd-networkd, the IPv6 RA logic now optionally may announce
3965 DNS server and domain information.
3967 * Support for the LUKS2 on-disk format for encrypted partitions has
3968 been added. This requires libcryptsetup2 during compilation and
3971 * The systemd --user instance will now signal "readiness" when its
3972 basic.target unit has been reached, instead of when the run queue ran
3973 empty for the first time.
3975 * Tmpfiles.d with user configuration are now also supported.
3976 systemd-tmpfiles gained a new --user switch, and snippets placed in
3977 ~/.config/user-tmpfiles.d/ and corresponding directories will be
3978 executed by systemd-tmpfiles --user running in the new
3979 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
3980 running in the user session.
3982 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
3983 %S resolves to the top-level state directory (/var/lib for the system
3984 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
3985 top-level cache directory (/var/cache for the system instance,
3986 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
3987 logs directory (/var/log for the system instance,
3988 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
3989 existing %t specifier, that resolves to the top-level runtime
3990 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
3993 * journalctl learnt a new parameter --output-fields= for limiting the
3994 set of journal fields to output in verbose and JSON output modes.
3996 * systemd-timesyncd's configuration file gained a new option
3997 RootDistanceMaxSec= for setting the maximum root distance of servers
3998 it'll use, as well as the new options PollIntervalMinSec= and
3999 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
4001 * bootctl gained a new command "list" for listing all available boot
4002 menu items on systems that follow the boot loader specification.
4004 * systemctl gained a new --dry-run switch that shows what would be done
4005 instead of doing it, and is currently supported by the shutdown and
4008 * ConditionSecurity= can now detect the TOMOYO security module.
4010 * Unit file [Install] sections are now also respected in unit drop-in
4011 files. This is intended to be used by drop-ins under /usr/lib/.
4013 * systemd-firstboot may now also set the initial keyboard mapping.
4015 * Udev "changed" events for devices which are exposed as systemd
4016 .device units are now propagated to units specified in
4017 ReloadPropagatedFrom= as reload requests.
4019 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
4020 unit template name (i.e. a name in the form of 'foobar@.service',
4021 without the instance component between the '@' and - the '.'), then
4022 the escaped sysfs path of the device is automatically used as the
4025 * SystemCallFilter= in unit files has been extended so that an "errno"
4026 can be specified individually for each system call. Example:
4027 SystemCallFilter=~uname:EILSEQ.
4029 * The cgroup delegation logic has been substantially updated. Delegate=
4030 now optionally takes a list of controllers (instead of a boolean, as
4031 before), which lists the controllers to delegate at least.
4033 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
4035 * A new LogLevelMax= setting configures the maximum log level any
4036 process of the service may log at (i.e. anything with a lesser
4037 priority than what is specified is automatically dropped). A new
4038 LogExtraFields= setting allows configuration of additional journal
4039 fields to attach to all log records generated by any of the unit's
4042 * New StandardInputData= and StandardInputText= settings along with the
4043 new option StandardInput=data may be used to configure textual or
4044 binary data that shall be passed to the executed service process via
4045 standard input, encoded in-line in the unit file.
4047 * StandardInput=, StandardOutput= and StandardError= may now be used to
4048 connect stdin/stdout/stderr of executed processes directly with a
4049 file or AF_UNIX socket in the file system, using the new "file:" option.
4051 * A new unit file option CollectMode= has been added, that allows
4052 tweaking the garbage collection logic for units. It may be used to
4053 tell systemd to garbage collect units that have failed automatically
4054 (normally it only GCs units that exited successfully). systemd-run
4055 and systemd-mount expose this new functionality with a new -G option.
4057 * "machinectl bind" may now be used to bind mount non-directories
4058 (i.e. regularfiles, devices, fifos, sockets).
4060 * systemd-analyze gained a new verb "calendar" for validating and
4061 testing calendar time specifications to use for OnCalendar= in timer
4062 units. Besides validating the expression it will calculate the next
4063 time the specified expression would elapse.
4065 * In addition to the pre-existing FailureAction= unit file setting
4066 there's now SuccessAction=, for configuring a shutdown action to
4067 execute when a unit completes successfully. This is useful in
4068 particular inside containers that shall terminate after some workload
4069 has been completed. Also, both options are now supported for all unit
4070 types, not just services.
4072 * networkds's IP rule support gained two new options
4073 IncomingInterface= and OutgoingInterface= for configuring the incoming
4074 and outgoing interfaces of configured rules. systemd-networkd also
4075 gained support for "vxcan" network devices.
4077 * networkd gained a new setting RequiredForOnline=, taking a
4078 boolean. If set, systemd-wait-online will take it into consideration
4079 when determining that the system is up, otherwise it will ignore the
4080 interface for this purpose.
4082 * The sd_notify() protocol gained support for a new operation: with
4083 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
4084 store again, ahead of POLLHUP or POLLERR when they are removed
4087 * A new document doc/UIDS-GIDS.md has been added to the source tree,
4088 that documents the UID/GID range and assignment assumptions and
4089 requirements of systemd.
4091 * The watchdog device PID 1 will ping may now be configured through the
4092 WatchdogDevice= configuration file setting, or by setting the
4093 systemd.watchdog_service= kernel commandline option.
4095 * systemd-resolved's gained support for registering DNS-SD services on
4096 the local network using MulticastDNS. Services may either be
4097 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
4098 the same dir below /run, /usr/lib), or through its D-Bus API.
4100 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
4101 extend the effective start, runtime, and stop time. The service must
4102 continue to send EXTEND_TIMEOUT_USEC within the period specified to
4103 prevent the service manager from making the service as timedout.
4105 * systemd-resolved's DNSSEC support gained support for RFC 8080
4106 (Ed25519 keys and signatures).
4108 * The systemd-resolve command line tool gained a new set of options
4109 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
4110 --set-nta= and --revert to configure per-interface DNS configuration
4111 dynamically during runtime. It's useful for pushing DNS information
4112 into systemd-resolved from DNS hook scripts that various interface
4113 managing software supports (such as pppd).
4115 * systemd-nspawn gained a new --network-namespace-path= command line
4116 option, which may be used to make a container join an existing
4117 network namespace, by specifying a path to a "netns" file.
4119 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
4120 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
4121 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
4122 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
4123 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
4124 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
4125 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
4126 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
4127 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
4128 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
4129 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
4130 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
4131 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
4132 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
4133 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
4134 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
4135 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
4136 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
4137 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
4138 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
4139 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
4140 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
4141 Jędrzejewski-Szmek, Zeal Jagannatha
4143 — Berlin, 2017-12-14
4147 * INCOMPATIBILITY: systemd-logind.service and other long-running
4148 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
4149 communication with the outside. This generally improves security of
4150 the system, and is in almost all cases a safe and good choice, as
4151 these services do not and should not provide any network-facing
4152 functionality. However, systemd-logind uses the glibc NSS API to
4153 query the user database. This creates problems on systems where NSS
4154 is set up to directly consult network services for user database
4155 lookups. In particular, this creates incompatibilities with the
4156 "nss-nis" module, which attempts to directly contact the NIS/YP
4157 network servers it is configured for, and will now consistently
4158 fail. In such cases, it is possible to turn off IP sandboxing for
4159 systemd-logind.service (set IPAddressDeny= in its [Service] section
4160 to the empty string, via a .d/ unit file drop-in). Downstream
4161 distributions might want to update their nss-nis packaging to include
4162 such a drop-in snippet, accordingly, to hide this incompatibility
4163 from the user. Another option is to make use of glibc's nscd service
4164 to proxy such network requests through a privilege-separated, minimal
4165 local caching daemon, or to switch to more modern technologies such
4166 sssd, whose NSS hook-ups generally do not involve direct network
4167 access. In general, we think it's definitely time to question the
4168 implementation choices of nss-nis, i.e. whether it's a good idea
4169 today to embed a network-facing loadable module into all local
4170 processes that need to query the user database, including the most
4171 trivial and benign ones, such as "ls". For more details about
4172 IPAddressDeny= see below.
4174 * A new modprobe.d drop-in is now shipped by default that sets the
4175 bonding module option max_bonds=0. This overrides the kernel default,
4176 to avoid conflicts and ambiguity as to whether or not bond0 should be
4177 managed by systemd-networkd or not. This resolves multiple issues
4178 with bond0 properties not being applied, when bond0 is configured
4179 with systemd-networkd. Distributors may choose to not package this,
4180 however in that case users will be prevented from correctly managing
4181 bond0 interface using systemd-networkd.
4183 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
4184 which print the logging level and target of the system manager. They
4185 complement the existing "set-log-level" and "set-log-target" verbs
4186 used to change those values.
4188 * journald.conf gained a new boolean setting ReadKMsg= which defaults
4189 to on. If turned off kernel log messages will not be read by
4190 systemd-journald or included in the logs. It also gained a new
4191 setting LineMax= for configuring the maximum line length in
4192 STDOUT/STDERR log streams. The new default for this value is 48K, up
4193 from the previous hardcoded 2048.
4195 * A new unit setting RuntimeDirectoryPreserve= has been added, which
4196 allows more detailed control of what to do with a runtime directory
4197 configured with RuntimeDirectory= (i.e. a directory below /run or
4198 $XDG_RUNTIME_DIR) after a unit is stopped.
4200 * The RuntimeDirectory= setting for units gained support for creating
4201 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
4202 one top-level directory.
4204 * Units gained new options StateDirectory=, CacheDirectory=,
4205 LogsDirectory= and ConfigurationDirectory= which are closely related
4206 to RuntimeDirectory= but manage per-service directories below
4207 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
4208 possible to write unit files which when activated automatically gain
4209 properly owned service specific directories in these locations, thus
4210 making unit files self-contained and increasing compatibility with
4211 stateless systems and factory reset where /etc or /var are
4212 unpopulated at boot. Matching these new settings there's also
4213 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
4214 ConfigurationDirectoryMode= for configuring the access mode of these
4215 directories. These settings are particularly useful in combination
4216 with DynamicUser=yes as they provide secure, properly-owned,
4217 writable, and stateful locations for storage, excluded from the
4218 sandbox that such services live in otherwise.
4220 * Automake support has been removed from this release. systemd is now
4223 * systemd-journald will now aggressively cache client metadata during
4224 runtime, speeding up log write performance under pressure. This comes
4225 at a small price though: as much of the metadata is read
4226 asynchronously from /proc/ (and isn't implicitly attached to log
4227 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
4228 metadata stored alongside a log entry might be slightly
4229 out-of-date. Previously it could only be slightly newer than the log
4230 message. The time window is small however, and given that the kernel
4231 is unlikely to be improved anytime soon in this regard, this appears
4234 * nss-myhostname/systemd-resolved will now by default synthesize an
4235 A/AAAA resource record for the "_gateway" hostname, pointing to the
4236 current default IP gateway. Previously it did that for the "gateway"
4237 name, hampering adoption, as some distributions wanted to leave that
4238 hostname open for local use. The old behaviour may still be
4239 requested at build time.
4241 * systemd-networkd's [Address] section in .network files gained a new
4242 Scope= setting for configuring the IP address scope. The [Network]
4243 section gained a new boolean setting ConfigureWithoutCarrier= that
4244 tells systemd-networkd to ignore link sensing when configuring the
4245 device. The [DHCP] section gained a new Anonymize= boolean option for
4246 turning on a number of options suggested in RFC 7844. A new
4247 [RoutingPolicyRule] section has been added for configuring the IP
4248 routing policy. The [Route] section has gained support for a new
4249 Type= setting which permits configuring
4250 blackhole/unreachable/prohibit routes.
4252 * The [VRF] section in .netdev files gained a new Table= setting for
4253 configuring the routing table to use. The [Tunnel] section gained a
4254 new Independent= boolean field for configuring tunnels independent of
4255 an underlying network interface. The [Bridge] section gained a new
4256 GroupForwardMask= option for configuration of propagation of link
4257 local frames between bridge ports.
4259 * The WakeOnLan= setting in .link files gained support for a number of
4260 new modes. A new TCP6SegmentationOffload= setting has been added for
4261 configuring TCP/IPv6 hardware segmentation offload.
4263 * The IPv6 RA sender implementation may now optionally send out RDNSS
4264 and RDNSSL records to supply DNS configuration to peers.
4266 * systemd-nspawn gained support for a new --system-call-filter= command
4267 line option for adding and removing entries in the default system
4268 call filter it applies. Moreover systemd-nspawn has been changed to
4269 implement a system call allow list instead of a deny list.
4271 * systemd-run gained support for a new --pipe command line option. If
4272 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
4273 are directly passed on to the activated transient service
4274 executable. This allows invoking arbitrary processes as systemd
4275 services (for example to take benefit of dependency management,
4276 accounting management, resource management or log management that is
4277 done automatically for services) — while still allowing them to be
4278 integrated in a classic UNIX shell pipeline.
4280 * When a service sends RELOAD=1 via sd_notify() and reload propagation
4281 using ReloadPropagationTo= is configured, a reload is now propagated
4282 to configured units. (Previously this was only done on explicitly
4283 requested reloads, using "systemctl reload" or an equivalent
4286 * For each service unit a restart counter is now kept: it is increased
4287 each time the service is restarted due to Restart=, and may be
4288 queried using "systemctl show -p NRestarts …".
4290 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
4291 @signal and @timer have been added, for usage with SystemCallFilter=
4292 in unit files and the new --system-call-filter= command line option
4293 of systemd-nspawn (see above).
4295 * ExecStart= lines in unit files gained two new modifiers: when a
4296 command line is prefixed with "!" the command will be executed as
4297 configured, except for the credentials applied by
4298 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
4299 "+", but does still apply namespacing options unlike "+". There's
4300 also "!!" now, which is mostly identical, but becomes a NOP on
4301 systems that support ambient capabilities. This is useful to write
4302 unit files that work with ambient capabilities where possible but
4303 automatically fall back to traditional privilege dropping mechanisms
4304 on systems where this is not supported.
4306 * ListenNetlink= settings in socket units now support RDMA netlink
4309 * A new unit file setting LockPersonality= has been added which permits
4310 locking down the chosen execution domain ("personality") of a service
4313 * A new special target "getty-pre.target" has been added, which is
4314 ordered before all text logins, and may be used to order services
4315 before textual logins acquire access to the console.
4317 * systemd will now attempt to load the virtio-rng.ko kernel module very
4318 early on if a VM environment supporting this is detected. This should
4319 improve entropy during early boot in virtualized environments.
4321 * A _netdev option is now supported in /etc/crypttab that operates in a
4322 similar way as the same option in /etc/fstab: it permits configuring
4323 encrypted devices that need to be ordered after the network is up.
4324 Following this logic, two new special targets
4325 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
4326 added that are to cryptsetup.target what remote-fs.target and
4327 remote-fs-pre.target are to local-fs.target.
4329 * Service units gained a new UnsetEnvironment= setting which permits
4330 unsetting specific environment variables for services that are
4331 normally passed to it (for example in order to mask out locale
4332 settings for specific services that can't deal with it).
4334 * Units acquired a new boolean option IPAccounting=. When turned on, IP
4335 traffic accounting (packet count as well as byte count) is done for
4336 the service, and shown as part of "systemctl status" or "systemd-run
4339 * Service units acquired two new options IPAddressAllow= and
4340 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
4341 for configuring a simple IP access control list for all sockets of
4342 the unit. These options are available also on .slice and .socket
4343 units, permitting flexible access list configuration for individual
4344 services as well as groups of services (as defined by a slice unit),
4345 including system-wide. Note that IP ACLs configured this way are
4346 enforced on every single IPv4 and IPv6 socket created by any process
4347 of the service unit, and apply to ingress as well as egress traffic.
4349 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
4350 structured log message is generated each time the unit is stopped,
4351 containing information about the consumed resources of this
4354 * A new setting KeyringMode= has been added to unit files, which may be
4355 used to control how the kernel keyring is set up for executed
4358 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
4359 "systemctl kexec" and "systemctl exit" are now always asynchronous in
4360 behaviour (that is: these commands return immediately after the
4361 operation was enqueued instead of waiting for the operation to
4362 complete). Previously, "systemctl poweroff" and "systemctl reboot"
4363 were asynchronous on systems using systemd-logind (i.e. almost
4364 always, and like they were on sysvinit), and the other three commands
4365 were unconditionally synchronous. With this release this is cleaned
4366 up, and callers will see the same asynchronous behaviour on all
4367 systems for all five operations.
4369 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
4372 * .timer units now accept calendar specifications in other timezones
4373 than UTC or the local timezone.
4375 * The tmpfiles snippet var.conf has been changed to create
4376 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
4377 the "utmp" group already, and it appears to be generally understood
4378 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
4379 databases. Previously this was implemented correctly for all these
4380 databases excepts btmp, which has been opened up like this now
4381 too. Note that while the other databases are world-readable
4382 (i.e. 0644), btmp is not and remains more restrictive.
4384 * The systemd-resolve tool gained a new --reset-server-features
4385 switch. When invoked like this systemd-resolved will forget
4386 everything it learnt about the features supported by the configured
4387 upstream DNS servers, and restarts the feature probing logic on the
4388 next resolver look-up for them at the highest feature level
4391 * The status dump systemd-resolved sends to the logs upon receiving
4392 SIGUSR1 now also includes information about all DNS servers it is
4393 configured to use, and the features levels it probed for them.
4395 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
4396 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
4397 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
4398 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
4399 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
4400 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
4401 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
4402 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
4403 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
4404 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
4405 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
4406 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
4407 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
4408 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
4409 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
4410 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
4411 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
4412 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
4413 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
4414 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
4416 — Berlin, 2017-10-06
4420 * Meson is now supported as build system in addition to Automake. It is
4421 our plan to remove Automake in one of our next releases, so that
4422 Meson becomes our exclusive build system. Hence, please start using
4423 the Meson build system in your downstream packaging. There's plenty
4424 of documentation around how to use Meson, the extremely brief
4427 ./autogen.sh && ./configure && make && sudo make install
4431 meson build && ninja -C build && sudo ninja -C build install
4433 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
4434 which permits configuring a timeout on the time a job is
4435 running. This is particularly useful for setting timeouts on jobs for
4438 * Unit files gained two new options ConditionUser= and ConditionGroup=
4439 for conditionalizing units based on the identity of the user/group
4440 running a systemd user instance.
4442 * systemd-networkd now understands a new FlowLabel= setting in the
4443 [VXLAN] section of .network files, as well as a Priority= in
4444 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
4445 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
4446 gained support for configuration of GENEVE links, and IPv6 address
4447 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
4449 * .link files now understand a new Port= setting.
4451 * systemd-networkd's DHCP support gained support for DHCP option 119
4452 (domain search list).
4454 * systemd-networkd gained support for serving IPv6 address ranges using
4455 the Router Advertisement protocol. The new .network configuration
4456 section [IPv6Prefix] may be used to configure the ranges to
4457 serve. This is implemented based on a new, minimal, native server
4458 implementation of RA.
4460 * journalctl's --output= switch gained support for a new parameter
4461 "short-iso-precise" for a mode where timestamps are shown as precise
4464 * systemd-udevd's "net_id" builtin may now generate stable network
4465 interface names from IBM PowerVM VIO devices as well as ACPI platform
4468 * MulticastDNS support in systemd-resolved may now be explicitly
4469 enabled/disabled using the new MulticastDNS= configuration file
4472 * systemd-resolved may now optionally use libidn2 instead of the libidn
4473 for processing internationalized domain names. Support for libidn2
4474 should be considered experimental and should not be enabled by
4477 * "machinectl pull-tar" and related call may now do verification of
4478 downloaded images using SUSE-style .sha256 checksum files in addition
4479 to the already existing support for validating using Ubuntu-style
4482 * sd-bus gained support for a new sd_bus_message_appendv() call which
4483 is va_list equivalent of sd_bus_message_append().
4485 * sd-boot gained support for validating images using SHIM/MOK.
4487 * The SMACK code learnt support for "onlycap".
4489 * systemd-mount --umount is now much smarter in figuring out how to
4490 properly unmount a device given its mount or device path.
4492 * The code to call libnss_dns as a fallback from libnss_resolve when
4493 the communication with systemd-resolved fails was removed. This
4494 fallback was redundant and interfered with the [!UNAVAIL=return]
4495 suffix. See nss-resolve(8) for the recommended configuration.
4497 * systemd-logind may now be restarted without losing state. It stores
4498 the file descriptors for devices it manages in the system manager
4499 using the FDSTORE= mechanism. Please note that further changes in
4500 other components may be required to make use of this (for example
4501 Xorg has code to listen for stops of systemd-logind and terminate
4502 itself when logind is stopped or restarted, in order to avoid using
4503 stale file descriptors for graphical devices, which is now
4504 counterproductive and must be reverted in order for restarts of
4505 systemd-logind to be safe. See
4506 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
4508 * All kernel-install plugins are called with the environment variable
4509 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
4510 /etc/machine-id. If the machine ID could not be determined,
4511 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
4512 anything in the entry directory (passed as the second argument) if
4513 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
4514 temporary directory is passed as the entry directory and removed
4515 after all the plugins exit.
4517 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
4518 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
4519 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
4520 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
4521 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
4522 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
4523 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
4524 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
4525 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
4526 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
4527 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
4528 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
4529 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
4530 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
4531 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
4532 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
4533 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
4534 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
4535 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
4536 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
4537 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
4538 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
4539 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
4540 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
4541 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
4542 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
4543 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
4544 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
4545 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
4548 — Berlin, 2017-07-12
4552 * The "hybrid" control group mode has been modified to improve
4553 compatibility with "legacy" cgroups-v1 setups. Specifically, the
4554 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
4555 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
4556 cgroups-v1 hierarchy), the only externally visible change being that
4557 the cgroups-v2 hierarchy is also mounted, to
4558 /sys/fs/cgroup/unified. This should provide a large degree of
4559 compatibility with "legacy" cgroups-v1, while taking benefit of the
4560 better management capabilities of cgroups-v2.
4562 * The default control group setup mode may be selected both a boot-time
4563 via a set of kernel command line parameters (specifically:
4564 systemd.unified_cgroup_hierarchy= and
4565 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
4566 default selected on the configure command line
4567 (--with-default-hierarchy=). The upstream default is "hybrid"
4568 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
4569 this will change in a future systemd version to be "unified" (pure
4570 cgroups-v2 mode). The third option for the compile time option is
4571 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
4572 distributions to default to "hybrid" mode for release distributions,
4573 starting with v233. We recommend "unified" for development
4574 distributions (specifically: distributions such as Fedora's rawhide)
4575 as that's where things are headed in the long run. Use "legacy" for
4576 greatest stability and compatibility only.
4578 * Note one current limitation of "unified" and "hybrid" control group
4579 setup modes: the kernel currently does not permit the systemd --user
4580 instance (i.e. unprivileged code) to migrate processes between two
4581 disconnected cgroup subtrees, even if both are managed and owned by
4582 the user. This effectively means "systemd-run --user --scope" doesn't
4583 work when invoked from outside of any "systemd --user" service or
4584 scope. Specifically, it is not supported from session scopes. We are
4585 working on fixing this in a future systemd version. (See #3388 for
4586 further details about this.)
4588 * DBus policy files are now installed into /usr rather than /etc. Make
4589 sure your system has dbus >= 1.9.18 running before upgrading to this
4590 version, or override the install path with --with-dbuspolicydir= .
4592 * All python scripts shipped with systemd (specifically: the various
4593 tests written in Python) now require Python 3.
4595 * systemd unit tests can now run standalone (without the source or
4596 build directories), and can be installed into /usr/lib/systemd/tests/
4597 with 'make install-tests'.
4599 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
4600 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
4603 * Support for the %c, %r, %R specifiers in unit files has been
4604 removed. Specifiers are not supposed to be dependent on configuration
4605 in the unit file itself (so that they resolve the same regardless
4606 where used in the unit files), but these specifiers were influenced
4607 by the Slice= option.
4609 * The shell invoked by debug-shell.service now defaults to /bin/sh in
4610 all cases. If distributions want to use a different shell for this
4611 purpose (for example Fedora's /sbin/sushell) they need to specify
4612 this explicitly at configure time using --with-debug-shell=.
4614 * The confirmation spawn prompt has been reworked to offer the
4617 (c)ontinue, proceed without asking anymore
4618 (D)ump, show the state of the unit
4619 (f)ail, don't execute the command and pretend it failed
4621 (i)nfo, show a short summary of the unit
4622 (j)obs, show jobs that are in progress
4623 (s)kip, don't execute the command and pretend it succeeded
4624 (y)es, execute the command
4626 The 'n' choice for the confirmation spawn prompt has been removed,
4627 because its meaning was confusing.
4629 The prompt may now also be redirected to an alternative console by
4630 specifying the console as parameter to systemd.confirm_spawn=.
4632 * Services of Type=notify require a READY=1 notification to be sent
4633 during startup. If no such message is sent, the service now fails,
4634 even if the main process exited with a successful exit code.
4636 * Services that fail to start up correctly now always have their
4637 ExecStopPost= commands executed. Previously, they'd enter "failed"
4638 state directly, without executing these commands.
4640 * The option MulticastDNS= of network configuration files has acquired
4641 an actual implementation. With MulticastDNS=yes a host can resolve
4642 names of remote hosts and reply to mDNS A and AAAA requests.
4644 * When units are about to be started an additional check is now done to
4645 ensure that all dependencies of type BindsTo= (when used in
4646 combination with After=) have been started.
4648 * systemd-analyze gained a new verb "syscall-filter" which shows which
4649 system call groups are defined for the SystemCallFilter= unit file
4650 setting, and which system calls they contain.
4652 * A new system call filter group "@filesystem" has been added,
4653 consisting of various file system related system calls. Group
4654 "@reboot" has been added, covering reboot, kexec and shutdown related
4655 calls. Finally, group "@swap" has been added covering swap
4656 configuration related calls.
4658 * A new unit file option RestrictNamespaces= has been added that may be
4659 used to restrict access to the various process namespace types the
4660 Linux kernel provides. Specifically, it may be used to take away the
4661 right for a service unit to create additional file system, network,
4662 user, and other namespaces. This sandboxing option is particularly
4663 relevant due to the high amount of recently discovered namespacing
4664 related vulnerabilities in the kernel.
4666 * systemd-udev's .link files gained support for a new AutoNegotiation=
4667 setting for configuring Ethernet auto-negotiation.
4669 * systemd-networkd's .network files gained support for a new
4670 ListenPort= setting in the [DHCP] section to explicitly configure the
4671 UDP client port the DHCP client shall listen on.
4673 * .network files gained a new Unmanaged= boolean setting for explicitly
4674 excluding one or more interfaces from management by systemd-networkd.
4676 * The systemd-networkd ProxyARP= option has been renamed to
4677 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
4678 renamed to ReduceARPProxy=. The old names continue to be available
4681 * systemd-networkd gained support for configuring IPv6 Proxy NDP
4682 addresses via the new IPv6ProxyNDPAddress= .network file setting.
4684 * systemd-networkd's bonding device support gained support for two new
4685 configuration options ActiveSlave= and PrimarySlave=.
4687 * The various options in the [Match] section of .network files gained
4688 support for negative matching.
4690 * New systemd-specific mount options are now understood in /etc/fstab:
4692 x-systemd.mount-timeout= may be used to configure the maximum
4693 permitted runtime of the mount command.
4695 x-systemd.device-bound may be set to bind a mount point to its
4696 backing device unit, in order to automatically remove a mount point
4697 if its backing device is unplugged. This option may also be
4698 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
4699 on the block device, which is now automatically set for all CDROM
4700 drives, so that mounted CDs are automatically unmounted when they are
4701 removed from the drive.
4703 x-systemd.after= and x-systemd.before= may be used to explicitly
4704 order a mount after or before another unit or mount point.
4706 * Enqueued start jobs for device units are now automatically garbage
4707 collected if there are no jobs waiting for them anymore.
4709 * systemctl list-jobs gained two new switches: with --after, for every
4710 queued job the jobs it's waiting for are shown; with --before the
4711 jobs which it's blocking are shown.
4713 * systemd-nspawn gained support for ephemeral boots from disk images
4714 (or in other words: --ephemeral and --image= may now be
4715 combined). Moreover, ephemeral boots are now supported for normal
4716 directories, even if the backing file system is not btrfs. Of course,
4717 if the file system does not support file system snapshots or
4718 reflinks, the initial copy operation will be relatively expensive, but
4719 this should still be suitable for many use cases.
4721 * Calendar time specifications in .timer units now support
4722 specifications relative to the end of a month by using "~" instead of
4723 "-" as separator between month and day. For example, "*-02~03" means
4724 "the third last day in February". In addition a new syntax for
4725 repeated events has been added using the "/" character. For example,
4726 "9..17/2:00" means "every two hours from 9am to 5pm".
4728 * systemd-socket-proxyd gained a new parameter --connections-max= for
4729 configuring the maximum number of concurrent connections.
4731 * sd-id128 gained a new API for generating unique IDs for the host in a
4732 way that does not leak the machine ID. Specifically,
4733 sd_id128_get_machine_app_specific() derives an ID based on the
4734 machine ID a in well-defined, non-reversible, stable way. This is
4735 useful whenever an identifier for the host is needed but where the
4736 identifier shall not be useful to identify the system beyond the
4737 scope of the application itself. (Internally this uses HMAC-SHA256 as
4738 keyed hash function using the machine ID as input.)
4740 * NotifyAccess= gained a new supported value "exec". When set
4741 notifications are accepted from all processes systemd itself invoked,
4742 including all control processes.
4744 * .nspawn files gained support for defining overlay mounts using the
4745 Overlay= and OverlayReadOnly= options. Previously this functionality
4746 was only available on the systemd-nspawn command line.
4748 * systemd-nspawn's --bind= and --overlay= options gained support for
4749 bind/overlay mounts whose source lies within the container tree by
4750 prefixing the source path with "+".
4752 * systemd-nspawn's --bind= and --overlay= options gained support for
4753 automatically allocating a temporary source directory in /var/tmp
4754 that is removed when the container dies. Specifically, if the source
4755 directory is specified as empty string this mechanism is selected. An
4756 example usage is --overlay=+/var::/var, which creates an overlay
4757 mount based on the original /var contained in the image, overlaid
4758 with a temporary directory in the host's /var/tmp. This way changes
4759 to /var are automatically flushed when the container shuts down.
4761 * systemd-nspawn --image= option does now permit raw file system block
4762 devices (in addition to images containing partition tables, as
4765 * The disk image dissection logic in systemd-nspawn gained support for
4766 automatically setting up LUKS encrypted as well as Verity protected
4767 partitions. When a container is booted from an encrypted image the
4768 passphrase is queried at start-up time. When a container with Verity
4769 data is started, the root hash is search in a ".roothash" file
4770 accompanying the disk image (alternatively, pass the root hash via
4771 the new --root-hash= command line option).
4773 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
4774 be used to dissect disk images the same way as systemd-nspawn does
4775 it, following the Bootable Partition Specification. It may even be
4776 used to mount disk images with complex partition setups (including
4777 LUKS and Verity partitions) to a local host directory, in order to
4778 inspect them. This tool is not considered public API (yet), and is
4779 thus not installed into /usr/bin. Please do not rely on its
4780 existence, since it might go away or be changed in later systemd
4783 * A new generator "systemd-verity-generator" has been added, similar in
4784 style to "systemd-cryptsetup-generator", permitting automatic setup of
4785 Verity root partitions when systemd boots up. In order to make use of
4786 this your partition setup should follow the Discoverable Partitions
4787 Specification, and the GPT partition ID of the root file system
4788 partition should be identical to the upper 128bit of the Verity root
4789 hash. The GPT partition ID of the Verity partition protecting it
4790 should be the lower 128bit of the Verity root hash. If the partition
4791 image follows this model it is sufficient to specify a single
4792 "roothash=" kernel command line argument to both configure which root
4793 image and verity partition to use as well as the root hash for
4794 it. Note that systemd-nspawn's Verity support follows the same
4795 semantics, meaning that disk images with proper Verity data in place
4796 may be booted in containers with systemd-nspawn as well as on
4797 physical systems via the verity generator. Also note that the "mkosi"
4798 tool available at https://github.com/systemd/mkosi has been updated
4799 to generate Verity protected disk images following this scheme. In
4800 fact, it has been updated to generate disk images that optionally
4801 implement a complete UEFI SecureBoot trust chain, involving a signed
4802 kernel and initrd image that incorporates such a root hash as well as
4803 a Verity-enabled root partition.
4805 * The hardware database (hwdb) udev supports has been updated to carry
4806 accelerometer quirks.
4808 * All system services are now run with a fresh kernel keyring set up
4809 for them. The invocation ID is stored by default in it, thus
4810 providing a safe, non-overridable way to determine the invocation
4813 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
4814 options for bind mounting arbitrary paths in a service-specific
4815 way. When these options are used, arbitrary host or service files and
4816 directories may be mounted to arbitrary locations in the service's
4819 * Documentation has been added that lists all of systemd's low-level
4820 environment variables:
4822 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
4824 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
4825 whether a specific socket file descriptor matches a specified socket
4828 * systemd-firstboot has been updated to check for the
4829 systemd.firstboot= kernel command line option. It accepts a boolean
4830 and when set to false the first boot questions are skipped.
4832 * systemd-fstab-generator has been updated to check for the
4833 systemd.volatile= kernel command line option, which either takes an
4834 optional boolean parameter or the special value "state". If used the
4835 system may be booted in a "volatile" boot mode. Specifically,
4836 "systemd.volatile" is used, the root directory will be mounted as
4837 tmpfs, and only /usr is mounted from the actual root file system. If
4838 "systemd.volatile=state" is used, the root directory will be mounted
4839 as usual, but /var is mounted as tmpfs. This concept provides similar
4840 functionality as systemd-nspawn's --volatile= option, but provides it
4841 on physical boots. Use this option for implementing stateless
4842 systems, or testing systems with all state and/or configuration reset
4843 to the defaults. (Note though that many distributions are not
4844 prepared to boot up without a populated /etc or /var, though.)
4846 * systemd-gpt-auto-generator gained support for LUKS encrypted root
4847 partitions. Previously it only supported LUKS encrypted partitions
4848 for all other uses, except for the root partition itself.
4850 * Socket units gained support for listening on AF_VSOCK sockets for
4851 communication in virtualized QEMU environments.
4853 * The "configure" script gained a new option --with-fallback-hostname=
4854 for specifying the fallback hostname to use if none is configured in
4855 /etc/hostname. For example, by specifying
4856 --with-fallback-hostname=fedora it is possible to default to a
4857 hostname of "fedora" on pristine installations.
4859 * systemd-cgls gained support for a new --unit= switch for listing only
4860 the control groups of a specific unit. Similar --user-unit= has been
4861 added for listing only the control groups of a specific user unit.
4863 * systemd-mount gained a new --umount switch for unmounting a mount or
4864 automount point (and all mount/automount points below it).
4866 * systemd will now refuse full configuration reloads (via systemctl
4867 daemon-reload and related calls) unless at least 16MiB of free space
4868 are available in /run. This is a safety precaution in order to ensure
4869 that generators can safely operate after the reload completed.
4871 * A new unit file option RootImage= has been added, which has a similar
4872 effect as RootDirectory= but mounts the service's root directory from
4873 a disk image instead of plain directory. This logic reuses the same
4874 image dissection and mount logic that systemd-nspawn already uses,
4875 and hence supports any disk images systemd-nspawn supports, including
4876 those following the Discoverable Partition Specification, as well as
4877 Verity enabled images. This option enables systemd to run system
4878 services directly off disk images acting as resource bundles,
4879 possibly even including full integrity data.
4881 * A new MountAPIVFS= unit file option has been added, taking a boolean
4882 argument. If enabled /proc, /sys and /dev (collectively called the
4883 "API VFS") will be mounted for the service. This is only relevant if
4884 RootDirectory= or RootImage= is used for the service, as these mounts
4885 are of course in place in the host mount namespace anyway.
4887 * systemd-nspawn gained support for a new --pivot-root= switch. If
4888 specified the root directory within the container image is pivoted to
4889 the specified mount point, while the original root disk is moved to a
4890 different place. This option enables booting of ostree images
4891 directly with systemd-nspawn.
4893 * The systemd build scripts will no longer complain if the NTP server
4894 addresses are not changed from the defaults. Google now supports
4895 these NTP servers officially. We still recommend downstreams to
4896 properly register an NTP pool with the NTP pool project though.
4898 * coredumpctl gained a new "--reverse" option for printing the list
4899 of coredumps in reverse order.
4901 * coredumpctl will now show additional information about truncated and
4902 inaccessible coredumps, as well as coredumps that are still being
4903 processed. It also gained a new --quiet switch for suppressing
4904 additional informational message in its output.
4906 * coredumpctl gained support for only showing coredumps newer and/or
4907 older than specific timestamps, using the new --since= and --until=
4908 options, reminiscent of journalctl's options by the same name.
4910 * The systemd-coredump logic has been improved so that it may be reused
4911 to collect backtraces in non-compiled languages, for example in
4912 scripting languages such as Python.
4914 * machinectl will now show the UID shift of local containers, if user
4915 namespacing is enabled for them.
4917 * systemd will now optionally run "environment generator" binaries at
4918 configuration load time. They may be used to add environment
4919 variables to the environment block passed to services invoked. One
4920 user environment generator is shipped by default that sets up
4921 environment variables based on files dropped into /etc/environment.d
4922 and ~/.config/environment.d/.
4924 * systemd-resolved now includes the new, recently published 2017 DNSSEC
4927 * hostnamed has been updated to report a new chassis type of
4928 "convertible" to cover "foldable" laptops that can both act as a
4929 tablet and as a laptop, such as various Lenovo Yoga devices.
4931 Contributions from: Adrián López, Alexander Galanin, Alexander
4932 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
4933 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
4934 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
4935 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
4936 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
4937 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
4938 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
4939 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
4940 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
4941 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
4942 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
4943 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
4944 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
4945 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
4946 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
4947 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
4948 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
4949 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
4950 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
4951 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
4952 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
4953 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
4954 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
4955 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
4956 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
4957 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
4958 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
4961 — Berlin, 2017-03-01
4965 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
4966 RestrictAddressFamilies= enabled. These sandboxing options should
4967 generally be compatible with the various external udev call-out
4968 binaries we are aware of, however there may be exceptions, in
4969 particular when exotic languages for these call-outs are used. In
4970 this case, consider turning off these settings locally.
4972 * The new RemoveIPC= option can be used to remove IPC objects owned by
4973 the user or group of a service when that service exits.
4975 * The new ProtectKernelModules= option can be used to disable explicit
4976 load and unload operations of kernel modules by a service. In
4977 addition access to /usr/lib/modules is removed if this option is set.
4979 * ProtectSystem= option gained a new value "strict", which causes the
4980 whole file system tree with the exception of /dev, /proc, and /sys,
4981 to be remounted read-only for a service.
4983 * The new ProtectKernelTunables= option can be used to disable
4984 modification of configuration files in /sys and /proc by a service.
4985 Various directories and files are remounted read-only, so access is
4986 restricted even if the file permissions would allow it.
4988 * The new ProtectControlGroups= option can be used to disable write
4989 access by a service to /sys/fs/cgroup.
4991 * Various systemd services have been hardened with
4992 ProtectKernelTunables=yes, ProtectControlGroups=yes,
4993 RestrictAddressFamilies=.
4995 * Support for dynamically creating users for the lifetime of a service
4996 has been added. If DynamicUser=yes is specified, user and group IDs
4997 will be allocated from the range 61184..65519 for the lifetime of the
4998 service. They can be resolved using the new nss-systemd.so NSS
4999 module. The module must be enabled in /etc/nsswitch.conf. Services
5000 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
5001 any resources allocated by the service will be cleaned up when the
5002 service exits. They also have ProtectHome=read-only and
5003 ProtectSystem=strict enabled, so they are not able to make any
5004 permanent modifications to the system.
5006 * The nss-systemd module also always resolves root and nobody, making
5007 it possible to have no /etc/passwd or /etc/group files in minimal
5008 container or chroot environments.
5010 * Services may be started with their own user namespace using the new
5011 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
5012 under which the service is running are mapped. All other users are
5015 * Support for the cgroup namespace has been added to systemd-nspawn. If
5016 supported by kernel, the container system started by systemd-nspawn
5017 will have its own view of the cgroup hierarchy. This new behaviour
5018 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
5020 * The new MemorySwapMax= option can be used to limit the maximum swap
5021 usage under the unified cgroup hierarchy.
5023 * Support for the CPU controller in the unified cgroup hierarchy has
5024 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
5025 options. This controller requires out-of-tree patches for the kernel
5026 and the support is provisional.
5028 * Mount and automount units may now be created transiently
5029 (i.e. dynamically at runtime via the bus API, instead of requiring
5030 unit files in the file system).
5032 * systemd-mount is a new tool which may mount file systems – much like
5033 mount(8), optionally pulling in additional dependencies through
5034 transient .mount and .automount units. For example, this tool
5035 automatically runs fsck on a backing block device before mounting,
5036 and allows the automount logic to be used dynamically from the
5037 command line for establishing mount points. This tool is particularly
5038 useful when dealing with removable media, as it will ensure fsck is
5039 run – if necessary – before the first access and that the file system
5040 is quickly unmounted after each access by utilizing the automount
5041 logic. This maximizes the chance that the file system on the
5042 removable media stays in a clean state, and if it isn't in a clean
5043 state is fixed automatically.
5045 * LazyUnmount=yes option for mount units has been added to expose the
5046 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
5049 * /efi will be used as the mount point of the EFI boot partition, if
5050 the directory is present, and the mount point was not configured
5051 through other means (e.g. fstab). If /efi directory does not exist,
5052 /boot will be used as before. This makes it easier to automatically
5053 mount the EFI partition on systems where /boot is used for something
5056 * When operating on GPT disk images for containers, systemd-nspawn will
5057 now mount the ESP to /boot or /efi according to the same rules as PID
5058 1 running on a host. This allows tools like "bootctl" to operate
5059 correctly within such containers, in order to make container images
5060 bootable on physical systems.
5062 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
5064 * Two new user session targets have been added to support running
5065 graphical sessions under the systemd --user instance:
5066 graphical-session.target and graphical-session-pre.target. See
5067 systemd.special(7) for a description of how those targets should be
5070 * The vconsole initialization code has been significantly reworked to
5071 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
5072 support unicode keymaps. Font and keymap configuration will now be
5073 copied to all allocated virtual consoles.
5075 * FreeBSD's bhyve virtualization is now detected.
5077 * Information recorded in the journal for core dumps now includes the
5078 contents of /proc/mountinfo and the command line of the process at
5079 the top of the process hierarchy (which is usually the init process
5082 * systemd-journal-gatewayd learned the --directory= option to serve
5083 files from the specified location.
5085 * journalctl --root=… can be used to peruse the journal in the
5086 /var/log/ directories inside of a container tree. This is similar to
5087 the existing --machine= option, but does not require the container to
5090 * The hardware database has been extended to support
5091 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
5094 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
5095 specify the click rate for mice which include a horizontal wheel with
5096 a click rate that is different than the one for the vertical wheel.
5098 * systemd-run gained a new --wait option that makes service execution
5099 synchronous. (Specifically, the command will not return until the
5100 specified service binary exited.)
5102 * systemctl gained a new --wait option that causes the start command to
5103 wait until the units being started have terminated again.
5105 * A new journal output mode "short-full" has been added which displays
5106 timestamps with abbreviated English day names and adds a timezone
5107 suffix. Those timestamps include more information than the default
5108 "short" output mode, and can be passed directly to journalctl's
5109 --since= and --until= options.
5111 * /etc/resolv.conf will be bind-mounted into containers started by
5112 systemd-nspawn, if possible, so any changes to resolv.conf contents
5113 are automatically propagated to the container.
5115 * The number of instances for socket-activated services originating
5116 from a single IP address can be limited with
5117 MaxConnectionsPerSource=, extending the existing setting of
5120 * systemd-networkd gained support for vcan ("Virtual CAN") interface
5123 * .netdev and .network configuration can now be extended through
5126 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
5127 Segmentation Offload, Generic Receive Offload, Large Receive Offload
5128 can be enabled and disabled using the new UDPSegmentationOffload=,
5129 TCPSegmentationOffload=, GenericSegmentationOffload=,
5130 GenericReceiveOffload=, LargeReceiveOffload= options in the
5131 [Link] section of .link files.
5133 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
5134 Port VLAN ID can be configured for bridge devices using the new STP=,
5135 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
5136 section of .netdev files.
5138 * The route table to which routes received over DHCP or RA should be
5139 added can be configured with the new RouteTable= option in the [DHCP]
5140 and [IPv6AcceptRA] sections of .network files.
5142 * The Address Resolution Protocol can be disabled on links managed by
5143 systemd-networkd using the ARP=no setting in the [Link] section of
5146 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
5147 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
5148 encode information about the result and exit codes of the current
5149 service runtime cycle.
5151 * systemd-sysctl will now configure kernel parameters in the order
5152 they occur in the configuration files. This matches what sysctl
5153 has been traditionally doing.
5155 * kernel-install "plugins" that are executed to perform various
5156 tasks after a new kernel is added and before an old one is removed
5157 can now return a special value to terminate the procedure and
5158 prevent any later plugins from running.
5160 * Journald's SplitMode=login setting has been deprecated. It has been
5161 removed from documentation, and its use is discouraged. In a future
5162 release it will be completely removed, and made equivalent to current
5163 default of SplitMode=uid.
5165 * Storage=both option setting in /etc/systemd/coredump.conf has been
5166 removed. With fast LZ4 compression storing the core dump twice is not
5169 * The --share-system systemd-nspawn option has been replaced with an
5170 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
5171 this functionality is discouraged. In addition the variables
5172 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
5173 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
5174 individual namespaces.
5176 * "machinectl list" now shows the IP address of running containers in
5177 the output, as well as OS release information.
5179 * "loginctl list" now shows the TTY of each session in the output.
5181 * sd-bus gained new API calls sd_bus_track_set_recursive(),
5182 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
5183 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
5184 tracking objects in a "recursive" mode, where a single client can be
5185 counted multiple times, if it takes multiple references.
5187 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
5188 sd_bus_get_exit_on_disconnect(). They may be used to make a
5189 process using sd-bus automatically exit if the bus connection is
5192 * Bus clients of the service manager may now "pin" loaded units into
5193 memory, by taking an explicit reference on them. This is useful to
5194 ensure the client can retrieve runtime data about the service even
5195 after the service completed execution. Taking such a reference is
5196 available only for privileged clients and should be helpful to watch
5197 running services in a race-free manner, and in particular collect
5198 information about exit statuses and results.
5200 * The nss-resolve module has been changed to strictly return UNAVAIL
5201 when communication via D-Bus with resolved failed, and NOTFOUND when
5202 a lookup completed but was negative. This means it is now possible to
5203 neatly configure fallbacks using nsswitch.conf result checking
5204 expressions. Taking benefit of this, the new recommended
5205 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
5207 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
5209 * A new setting CtrlAltDelBurstAction= has been added to
5210 /etc/systemd/system.conf which may be used to configure the precise
5211 behaviour if the user on the console presses Ctrl-Alt-Del more often
5212 than 7 times in 2s. Previously this would unconditionally result in
5213 an expedited, immediate reboot. With this new setting the precise
5214 operation may be configured in more detail, and also turned off
5217 * In .netdev files two new settings RemoteChecksumTx= and
5218 RemoteChecksumRx= are now understood that permit configuring the
5219 remote checksumming logic for VXLAN networks.
5221 * The service manager learnt a new "invocation ID" concept for invoked
5222 services. Each runtime cycle of a service will get a new invocation
5223 ID (a 128bit random UUID) assigned that identifies the current
5224 run of the service uniquely and globally. A new invocation ID
5225 is generated each time a service starts up. The journal will store
5226 the invocation ID of a service along with any logged messages, thus
5227 making the invocation ID useful for matching the online runtime of a
5228 service with the offline log data it generated in a safe way without
5229 relying on synchronized timestamps. In many ways this new service
5230 invocation ID concept is similar to the kernel's boot ID concept that
5231 uniquely and globally identifies the runtime of each boot. The
5232 invocation ID of a service is passed to the service itself via an
5233 environment variable ($INVOCATION_ID). A new bus call
5234 GetUnitByInvocationID() has been added that is similar to GetUnit()
5235 but instead of retrieving the bus path for a unit by its name
5236 retrieves it by its invocation ID. The returned path is valid only as
5237 long as the passed invocation ID is current.
5239 * systemd-resolved gained a new "DNSStubListener" setting in
5240 resolved.conf. It either takes a boolean value or the special values
5241 "udp" and "tcp", and configures whether to enable the stub DNS
5242 listener on 127.0.0.53:53.
5244 * IP addresses configured via networkd may now carry additional
5245 configuration settings supported by the kernel. New options include:
5246 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
5247 PrefixRoute=, AutoJoin=.
5249 * The PAM configuration fragment file for "user@.service" shipped with
5250 systemd (i.e. the --user instance of systemd) has been stripped to
5251 the minimum necessary to make the system boot. Previously, it
5252 contained Fedora-specific stanzas that did not apply to other
5253 distributions. It is expected that downstream distributions add
5254 additional configuration lines, matching their needs to this file,
5255 using it only as rough template of what systemd itself needs. Note
5256 that this reduced fragment does not even include an invocation of
5257 pam_limits which most distributions probably want to add, even though
5258 systemd itself does not need it. (There's also the new build time
5259 option --with-pamconfdir=no to disable installation of the PAM
5262 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
5263 capability is now also dropped from its set (in addition to
5264 CAP_SYS_MKNOD as before).
5266 * In service unit files it is now possible to connect a specific named
5267 file descriptor with stdin/stdout/stdout of an executed service. The
5268 name may be specified in matching .socket units using the
5269 FileDescriptorName= setting.
5271 * A number of journal settings may now be configured on the kernel
5272 command line. Specifically, the following options are now understood:
5273 systemd.journald.max_level_console=,
5274 systemd.journald.max_level_store=,
5275 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
5276 systemd.journald.max_level_wall=.
5278 * "systemctl is-enabled --full" will now show by which symlinks a unit
5279 file is enabled in the unit dependency tree.
5281 * Support for VeraCrypt encrypted partitions has been added to the
5282 "cryptsetup" logic and /etc/crypttab.
5284 * systemd-detect-virt gained support for a new --private-users switch
5285 that checks whether the invoking processes are running inside a user
5286 namespace. Similar, a new special value "private-users" for the
5287 existing ConditionVirtualization= setting has been added, permitting
5288 skipping of specific units in user namespace environments.
5290 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
5291 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
5292 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
5293 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
5294 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
5295 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
5296 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
5297 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
5298 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
5299 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
5300 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
5301 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
5302 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
5303 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
5304 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
5305 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
5306 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
5307 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
5308 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
5309 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
5310 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
5311 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
5312 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
5313 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
5314 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
5315 Jędrzejewski-Szmek, Zeal Jagannatha
5317 — Santa Fe, 2016-11-03
5321 * In service units the various ExecXYZ= settings have been extended
5322 with an additional special character as first argument of the
5323 assigned value: if the character '+' is used the specified command
5324 line it will be run with full privileges, regardless of User=,
5325 Group=, CapabilityBoundingSet= and similar options. The effect is
5326 similar to the existing PermissionsStartOnly= option, but allows
5327 configuration of this concept for each executed command line
5330 * Services may now alter the service watchdog timeout at runtime by
5331 sending a WATCHDOG_USEC= message via sd_notify().
5333 * MemoryLimit= and related unit settings now optionally take percentage
5334 specifications. The percentage is taken relative to the amount of
5335 physical memory in the system (or in case of containers, the assigned
5336 amount of memory). This allows scaling service resources neatly with
5337 the amount of RAM available on the system. Similarly, systemd-logind's
5338 RuntimeDirectorySize= option now also optionally takes percentage
5341 * In similar fashion TasksMax= takes percentage values now, too. The
5342 value is taken relative to the configured maximum number of processes
5343 on the system. The per-service task maximum has been changed to 15%
5344 using this functionality. (Effectively this is an increase of 512 →
5345 4915 for service units, given the kernel's default pid_max setting.)
5347 * Calendar time specifications in .timer units now understand a ".."
5348 syntax for time ranges. Example: "4..7:10" may now be used for
5349 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
5352 * The InaccessableDirectories=, ReadOnlyDirectories= and
5353 ReadWriteDirectories= unit file settings have been renamed to
5354 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
5355 applied to all kinds of file nodes, and not just directories, with
5356 the exception of symlinks. Specifically these settings may now be
5357 used on block and character device nodes, UNIX sockets and FIFOS as
5358 well as regular files. The old names of these settings remain
5359 available for compatibility.
5361 * systemd will now log about all service processes it kills forcibly
5362 (using SIGKILL) because they remained after the clean shutdown phase
5363 of the service completed. This should help identifying services that
5364 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
5365 systemd-logind's configuration a similar log message is generated for
5366 processes killed at the end of each session due to this setting.
5368 * systemd will now set the $JOURNAL_STREAM environment variable for all
5369 services whose stdout/stderr are connected to the Journal (which
5370 effectively means by default: all services). The variable contains
5371 the device and inode number of the file descriptor used for
5372 stdout/stderr. This may be used by invoked programs to detect whether
5373 their stdout/stderr is connected to the Journal, in which case they
5374 can switch over to direct Journal communication, thus being able to
5375 pass extended, structured metadata along with their log messages. As
5376 one example, this is now used by glib's logging primitives.
5378 * When using systemd's default tmp.mount unit for /tmp, the mount point
5379 will now be established with the "nosuid" and "nodev" options. This
5380 avoids privilege escalation attacks that put traps and exploits into
5381 /tmp. However, this might cause problems if you e. g. put container
5382 images or overlays into /tmp; if you need this, override tmp.mount's
5383 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
5386 * systemd now supports the "memory" cgroup controller also on
5389 * The systemd-cgtop tool now optionally takes a control group path as
5390 command line argument. If specified, the control group list shown is
5391 limited to subgroups of that group.
5393 * The SystemCallFilter= unit file setting gained support for
5394 pre-defined, named system call filter sets. For example
5395 SystemCallFilter=@clock is now an effective way to make all clock
5396 changing-related system calls unavailable to a service. A number of
5397 similar pre-defined groups are defined. Writing system call filters
5398 for system services is simplified substantially with this new
5399 concept. Accordingly, all of systemd's own, long-running services now
5400 enable system call filtering based on this, by default.
5402 * A new service setting MemoryDenyWriteExecute= has been added, taking
5403 a boolean value. If turned on, a service may no longer create memory
5404 mappings that are writable and executable at the same time. This
5405 enhances security for services where this is enabled as it becomes
5406 harder to dynamically write and then execute memory in exploited
5407 service processes. This option has been enabled for all of systemd's
5408 own long-running services.
5410 * A new RestrictRealtime= service setting has been added, taking a
5411 boolean argument. If set the service's processes may no longer
5412 acquire realtime scheduling. This improves security as realtime
5413 scheduling may otherwise be used to easily freeze the system.
5415 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
5416 value. This may be used for requesting that the system manager inside
5417 of the container reports start-up completion to nspawn which then
5418 propagates this notification further to the service manager
5419 supervising nspawn itself. A related option NotifyReady= in .nspawn
5420 files has been added too. This functionality allows ordering of the
5421 start-up of multiple containers using the usual systemd ordering
5424 * machinectl gained a new command "stop" that is an alias for
5427 * systemd-resolved gained support for contacting DNS servers on
5428 link-local IPv6 addresses.
5430 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
5431 its caches. A method call for requesting the same operation has been
5432 added to the bus API too, and is made available via "systemd-resolve
5435 * systemd-resolve gained a new --status switch. If passed a brief
5436 summary of the used DNS configuration with per-interface information
5439 * resolved.conf gained a new Cache= boolean option, defaulting to
5440 on. If turned off local DNS caching is disabled. This comes with a
5441 performance penalty in particular when DNSSEC is enabled. Note that
5442 resolved disables its internal caching implicitly anyway, when the
5443 configured DNS server is on a host-local IP address such as ::1 or
5444 127.0.0.1, thus automatically avoiding double local caching.
5446 * systemd-resolved now listens on the local IP address 127.0.0.53:53
5447 for DNS requests. This improves compatibility with local programs
5448 that do not use the libc NSS or systemd-resolved's bus APIs for name
5449 resolution. This minimal DNS service is only available to local
5450 programs and does not implement the full DNS protocol, but enough to
5451 cover local DNS clients. A new, static resolv.conf file, listing just
5452 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
5453 now recommended to make /etc/resolv.conf a symlink to this file in
5454 order to route all DNS lookups to systemd-resolved, regardless if
5455 done via NSS, the bus API or raw DNS packets. Note that this local
5456 DNS service is not as fully featured as the libc NSS or
5457 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
5458 used to deliver link-local address information (as this implies
5459 sending a local interface index along), LLMNR/mDNS support via this
5460 interface is severely restricted. It is thus strongly recommended for
5461 all applications to use the libc NSS API or native systemd-resolved
5464 * systemd-networkd's bridge support learned a new setting
5465 VLANFiltering= for controlling VLAN filtering. Moreover a new section
5466 in .network files has been added for configuring VLAN bridging in
5467 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
5469 * systemd-networkd's IPv6 Router Advertisement code now makes use of
5470 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
5471 now be acquired without relying on DHCPv6. Two new options
5472 UseDomains= and UseDNS= have been added to configure this behaviour.
5474 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
5475 renamed IPv6AcceptRA=, without altering its behaviour. The old
5476 setting name remains available for compatibility reasons.
5478 * The systemd-networkd VTI/VTI6 tunneling support gained new options
5479 Key=, InputKey= and OutputKey=.
5481 * systemd-networkd gained support for VRF ("Virtual Routing Function")
5482 interface configuration.
5484 * "systemctl edit" may now be used to create new unit files by
5485 specifying the --force switch.
5487 * sd-event gained a new function sd_event_get_iteration() for
5488 requesting the current iteration counter of the event loop. It starts
5489 at zero and is increased by one with each event loop iteration.
5491 * A new rpm macro %systemd_ordering is provided by the macros.systemd
5492 file. It can be used in lieu of %systemd_requires in packages which
5493 don't use any systemd functionality and are intended to be installed
5494 in minimal containers without systemd present. This macro provides
5495 ordering dependencies to ensure that if the package is installed in
5496 the same rpm transaction as systemd, systemd will be installed before
5497 the scriptlets for the package are executed, allowing unit presets
5500 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
5501 been added to simplify packaging of generators.
5503 * The os-release file gained VERSION_CODENAME field for the
5504 distribution nickname (e.g. VERSION_CODENAME=woody).
5506 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
5507 can be set to disable parsing of metadata and the creation
5508 of persistent symlinks for that device.
5510 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
5511 to make them available to logged-in users has been reverted.
5513 * Much of the common code of the various systemd components is now
5514 built into an internal shared library libsystemd-shared-231.so
5515 (incorporating the systemd version number in the name, to be updated
5516 with future releases) that the components link to. This should
5517 decrease systemd footprint both in memory during runtime and on
5518 disk. Note that the shared library is not for public use, and is
5519 neither API nor ABI stable, but is likely to change with every new
5520 released update. Packagers need to make sure that binaries
5521 linking to libsystemd-shared.so are updated in step with the
5524 * Configuration for "mkosi" is now part of the systemd
5525 repository. mkosi is a tool to easily build legacy-free OS images,
5526 and is available on github: https://github.com/systemd/mkosi. If
5527 "mkosi" is invoked in the build tree a new raw OS image is generated
5528 incorporating the systemd sources currently being worked on and a
5529 clean, fresh distribution installation. The generated OS image may be
5530 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
5531 UEFI PC. This functionality is particularly useful to easily test
5532 local changes made to systemd in a pristine, defined environment. See
5533 doc/HACKING for details.
5535 * configure learned the --with-support-url= option to specify the
5536 distribution's bugtracker.
5538 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
5539 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
5540 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
5541 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
5542 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
5543 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
5544 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
5545 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
5546 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
5547 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
5548 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
5549 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
5550 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
5551 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
5552 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
5553 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
5554 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
5555 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
5556 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
5558 — Berlin, 2016-07-25
5562 * DNSSEC is now turned on by default in systemd-resolved (in
5563 "allow-downgrade" mode), but may be turned off during compile time by
5564 passing "--with-default-dnssec=no" to "configure" (and of course,
5565 during runtime with DNSSEC= in resolved.conf). We recommend
5566 downstreams to leave this on at least during development cycles and
5567 report any issues with the DNSSEC logic upstream. We are very
5568 interested in collecting feedback about the DNSSEC validator and its
5569 limitations in the wild. Note however, that DNSSEC support is
5570 probably nothing downstreams should turn on in stable distros just
5571 yet, as it might create incompatibilities with a few DNS servers and
5572 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
5573 automatically whenever we detect such incompatible setups, but there
5574 might be systems we do not cover yet. Hence: please help us testing
5575 the DNSSEC code, leave this on where you can, report back, but then
5576 again don't consider turning this on in your stable, LTS or
5577 production release just yet. (Note that you have to enable
5578 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
5579 and its DNSSEC mode for hostname resolution from local
5582 * systemd-resolve conveniently resolves DANE records with the --tlsa
5583 option and OPENPGPKEY records with the --openpgp option. It also
5584 supports dumping raw DNS record data via the new --raw= switch.
5586 * systemd-logind will now by default terminate user processes that are
5587 part of the user session scope unit (session-XX.scope) when the user
5588 logs out. This behavior is controlled by the KillUserProcesses=
5589 setting in logind.conf, and the previous default of "no" is now
5590 changed to "yes". This means that user sessions will be properly
5591 cleaned up after, but additional steps are necessary to allow
5592 intentionally long-running processes to survive logout.
5594 While the user is logged in at least once, user@.service is running,
5595 and any service that should survive the end of any individual login
5596 session can be started at a user service or scope using systemd-run.
5597 systemd-run(1) man page has been extended with an example which shows
5598 how to run screen in a scope unit underneath user@.service. The same
5599 command works for tmux.
5601 After the user logs out of all sessions, user@.service will be
5602 terminated too, by default, unless the user has "lingering" enabled.
5603 To effectively allow users to run long-term tasks even if they are
5604 logged out, lingering must be enabled for them. See loginctl(1) for
5605 details. The default polkit policy was modified to allow users to
5606 set lingering for themselves without authentication.
5608 Previous defaults can be restored at compile time by the
5609 --without-kill-user-processes option to "configure".
5611 * systemd-logind gained new configuration settings SessionsMax= and
5612 InhibitorsMax=, both with a default of 8192. It will not register new
5613 user sessions or inhibitors above this limit.
5615 * systemd-logind will now reload configuration on SIGHUP.
5617 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
5618 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
5619 enable. Also, support for the "io" cgroup controller in the unified
5620 hierarchy has been added, so that the "memory", "pids" and "io" are
5621 now the controllers that are supported on the unified hierarchy.
5623 WARNING: it is not possible to use previous systemd versions with
5624 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
5625 is necessary to also update systemd in the initramfs if using the
5626 unified hierarchy. An updated SELinux policy is also required.
5628 * LLDP support has been extended, and both passive (receive-only) and
5629 active (sender) modes are supported. Passive mode ("routers-only") is
5630 enabled by default in systemd-networkd. Active LLDP mode is enabled
5631 by default for containers on the internal network. The "networkctl
5632 lldp" command may be used to list information gathered. "networkctl
5633 status" will also show basic LLDP information on connected peers now.
5635 * The IAID and DUID unique identifier sent in DHCP requests may now be
5636 configured for the system and each .network file managed by
5637 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
5639 * systemd-networkd gained support for configuring proxy ARP support for
5640 each interface, via the ProxyArp= setting in .network files. It also
5641 gained support for configuring the multicast querier feature of
5642 bridge devices, via the new MulticastQuerier= setting in .netdev
5643 files. Similarly, snooping on the IGMP traffic can be controlled
5644 via the new setting MulticastSnooping=.
5646 A new setting PreferredLifetime= has been added for addresses
5647 configured in .network file to configure the lifetime intended for an
5650 The systemd-networkd DHCP server gained the option EmitRouter=, which
5651 defaults to yes, to configure whether the DHCP Option 3 (Router)
5654 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
5655 systemd-socket-activate and installed into /usr/bin. It is now fully
5658 * systemd-journald now uses separate threads to flush changes to disk
5659 when closing journal files, thus reducing impact of slow disk I/O on
5660 logging performance.
5662 * The sd-journal API gained two new calls
5663 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
5664 can be used to open journal files using file descriptors instead of
5665 file or directory paths. sd_journal_open_container() has been
5666 deprecated, sd_journal_open_directory_fd() should be used instead
5667 with the flag SD_JOURNAL_OS_ROOT.
5669 * journalctl learned a new output mode "-o short-unix" that outputs log
5670 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
5671 UTC). It also gained support for a new --no-hostname setting to
5672 suppress the hostname column in the family of "short" output modes.
5674 * systemd-ask-password now optionally skips printing of the password to
5675 stdout with --no-output which can be useful in scripts.
5677 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
5678 (devices tagged with ID_MAKER_TOOL) are now tagged with
5679 "uaccess" and are available to logged in users.
5681 * The DeviceAllow= unit setting now supports specifiers (with "%").
5683 * "systemctl show" gained a new --value switch, which allows print a
5684 only the contents of a specific unit property, without also printing
5685 the property's name. Similar support was added to "show*" verbs
5686 of loginctl and machinectl that output "key=value" lists.
5688 * A new unit type "generated" was added for files dynamically generated
5689 by generator tools. Similarly, a new unit type "transient" is used
5690 for unit files created using the runtime API. "systemctl enable" will
5691 refuse to operate on such files.
5693 * A new command "systemctl revert" has been added that may be used to
5694 revert to the vendor version of a unit file, in case local changes
5695 have been made by adding drop-ins or overriding the unit file.
5697 * "machinectl clean" gained a new verb to automatically remove all or
5698 just hidden container images.
5700 * systemd-tmpfiles gained support for a new line type "e" for emptying
5701 directories, if they exist, without creating them if they don't.
5703 * systemd-nspawn gained support for automatically patching the UID/GIDs
5704 of the owners and the ACLs of all files and directories in a
5705 container tree to match the UID/GID user namespacing range selected
5706 for the container invocation. This mode is enabled via the new
5707 --private-users-chown switch. It also gained support for
5708 automatically choosing a free, previously unused UID/GID range when
5709 starting a container, via the new --private-users=pick setting (which
5710 implies --private-users-chown). Together, these options for the first
5711 time make user namespacing for nspawn containers fully automatic and
5712 thus deployable. The systemd-nspawn@.service template unit file has
5713 been changed to use this functionality by default.
5715 * systemd-nspawn gained a new --network-zone= switch, that allows
5716 creating ad-hoc virtual Ethernet links between multiple containers,
5717 that only exist as long as at least one container referencing them is
5718 running. This allows easy connecting of multiple containers with a
5719 common link that implements an Ethernet broadcast domain. Each of
5720 these network "zones" may be named relatively freely by the user, and
5721 may be referenced by any number of containers, but each container may
5722 only reference one of these "zones". On the lower level, this is
5723 implemented by an automatically managed bridge network interface for
5724 each zone, that is created when the first container referencing its
5725 zone is created and removed when the last one referencing its zone
5728 * The default start timeout may now be configured on the kernel command
5729 line via systemd.default_timeout_start_sec=. It was already
5730 configurable via the DefaultTimeoutStartSec= option in
5731 /etc/systemd/system.conf.
5733 * Socket units gained a new TriggerLimitIntervalSec= and
5734 TriggerLimitBurst= setting to configure a limit on the activation
5735 rate of the socket unit.
5737 * The LimitNICE= setting now optionally takes normal UNIX nice values
5738 in addition to the raw integer limit value. If the specified
5739 parameter is prefixed with "+" or "-" and is in the range -20..19 the
5740 value is understood as UNIX nice value. If not prefixed like this it
5741 is understood as raw RLIMIT_NICE limit.
5743 * Note that the effect of the PrivateDevices= unit file setting changed
5744 slightly with this release: the per-device /dev file system will be
5745 mounted read-only from this version on, and will have "noexec"
5746 set. This (minor) change of behavior might cause some (exceptional)
5747 legacy software to break, when PrivateDevices=yes is set for its
5748 service. Please leave PrivateDevices= off if you run into problems
5751 * systemd-bootchart has been split out to a separate repository:
5752 https://github.com/systemd/systemd-bootchart
5754 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
5755 merged into the kernel in its current form.
5757 * The compatibility libraries libsystemd-daemon.so,
5758 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
5759 which have been deprecated since systemd-209 have been removed along
5760 with the corresponding pkg-config files. All symbols provided by
5761 those libraries are provided by libsystemd.so.
5763 * The Capabilities= unit file setting has been removed (it is ignored
5764 for backwards compatibility). AmbientCapabilities= and
5765 CapabilityBoundingSet= should be used instead.
5767 * A new special target has been added, initrd-root-device.target,
5768 which creates a synchronization point for dependencies of the root
5769 device in early userspace. Initramfs builders must ensure that this
5770 target is now included in early userspace.
5772 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
5773 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
5774 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
5775 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
5776 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
5777 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
5778 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
5779 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
5780 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
5781 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
5782 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
5783 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
5784 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
5785 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
5786 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
5787 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
5788 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
5789 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
5790 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
5791 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
5792 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
5793 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
5794 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
5795 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
5796 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
5799 — Fairfax, 2016-05-21
5803 * The systemd-resolved DNS resolver service has gained a substantial
5804 set of new features, most prominently it may now act as a DNSSEC
5805 validating stub resolver. DNSSEC mode is currently turned off by
5806 default, but is expected to be turned on by default in one of the
5807 next releases. For now, we invite everybody to test the DNSSEC logic
5808 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
5809 service also gained a full set of D-Bus interfaces, including calls
5810 to configure DNS and DNSSEC settings per link (for use by external
5811 network management software). systemd-resolved and systemd-networkd
5812 now distinguish between "search" and "routing" domains. The former
5813 are used to qualify single-label names, the latter are used purely
5814 for routing lookups within certain domains to specific links.
5815 resolved now also synthesizes RRs for all entries from /etc/hosts.
5817 * The systemd-resolve tool (which is a client utility for
5818 systemd-resolved) has been improved considerably and is now fully
5819 supported and documented. Hence it has moved from /usr/lib/systemd to
5822 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
5825 * The coredump collection logic has been reworked: when a coredump is
5826 collected it is now written to disk, compressed and processed
5827 (including stacktrace extraction) from a new instantiated service
5828 systemd-coredump@.service, instead of directly from the
5829 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
5830 processing large coredumps can take up a substantial amount of
5831 resources and time, and this previously happened entirely outside of
5832 systemd's service supervision. With the new logic the core_pattern
5833 hook only does minimal metadata collection before passing off control
5834 to the new instantiated service, which is configured with a time
5835 limit, a nice level and other settings to minimize negative impact on
5836 the rest of the system. Also note that the new logic will honour the
5837 RLIMIT_CORE setting of the crashed process, which now allows users
5838 and processes to turn off coredumping for their processes by setting
5841 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
5842 and all forked processes by default. Previously, PID 1 would leave
5843 the setting at "0" for all processes, as set by the kernel. Note that
5844 the resource limit traditionally has no effect on the generated
5845 coredumps on the system if the /proc/sys/kernel/core_pattern hook
5846 logic is used. Since the limit is now honoured (see above) its
5847 default has been changed so that the coredumping logic is enabled by
5848 default for all processes, while allowing specific opt-out.
5850 * When the stacktrace is extracted from processes of system users, this
5851 is now done as "systemd-coredump" user, in order to sandbox this
5852 potentially security sensitive parsing operation. (Note that when
5853 processing coredumps of normal users this is done under the user ID
5854 of process that crashed, as before.) Packagers should take notice
5855 that it is now necessary to create the "systemd-coredump" system user
5856 and group at package installation time.
5858 * The systemd-activate socket activation testing tool gained support
5859 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
5860 and --seqpacket switches. It also has been extended to support both
5861 new-style and inetd-style file descriptor passing. Use the new
5862 --inetd switch to request inetd-style file descriptor passing.
5864 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
5865 variable, which takes a boolean value. If set to false, ANSI color
5866 output is disabled in the tools even when run on a terminal that
5869 * The VXLAN support in networkd now supports two new settings
5870 DestinationPort= and PortRange=.
5872 * A new systemd.machine_id= kernel command line switch has been added,
5873 that may be used to set the machine ID in /etc/machine-id if it is
5874 not initialized yet. This command line option has no effect if the
5875 file is already initialized.
5877 * systemd-nspawn gained a new --as-pid2 switch that invokes any
5878 specified command line as PID 2 rather than PID 1 in the
5879 container. In this mode PID 1 is a minimal stub init process that
5880 implements the special POSIX and Linux semantics of PID 1 regarding
5881 signal and child process management. Note that this stub init process
5882 is implemented in nspawn itself and requires no support from the
5883 container image. This new logic is useful to support running
5884 arbitrary commands in the container, as normal processes are
5885 generally not prepared to run as PID 1.
5887 * systemd-nspawn gained a new --chdir= switch for setting the current
5888 working directory for the process started in the container.
5890 * "journalctl /dev/sda" will now output all kernel log messages for
5891 specified device from the current boot, in addition to all devices
5892 that are parents of it. This should make log output about devices
5893 pretty useful, as long as kernel drivers attach enough metadata to
5894 the log messages. (The usual SATA drivers do.)
5896 * The sd-journal API gained two new calls
5897 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
5898 that report whether log data from /run or /var has been found.
5900 * journalctl gained a new switch "--fields" that prints all journal
5901 record field names currently in use in the journal. This is backed
5902 by two new sd-journal API calls sd_journal_enumerate_fields() and
5903 sd_journal_restart_fields().
5905 * Most configurable timeouts in systemd now expect an argument of
5906 "infinity" to turn them off, instead of "0" as before. The semantics
5907 from now on is that a timeout of "0" means "now", and "infinity"
5908 means "never". To maintain backwards compatibility, "0" continues to
5909 turn off previously existing timeout settings.
5911 * "systemctl reload-or-try-restart" has been renamed to "systemctl
5912 try-reload-or-restart" to clarify what it actually does: the "try"
5913 logic applies to both reloading and restarting, not just restarting.
5914 The old name continues to be accepted for compatibility.
5916 * On boot-up, when PID 1 detects that the system clock is behind the
5917 release date of the systemd version in use, the clock is now set
5918 to the latter. Previously, this was already done in timesyncd, in order
5919 to avoid running with clocks set to the various clock epochs such as
5920 1902, 1938 or 1970. With this change the logic is now done in PID 1
5921 in addition to timesyncd during early boot-up, so that it is enforced
5922 before the first process is spawned by systemd. Note that the logic
5923 in timesyncd remains, as it is more comprehensive and ensures
5924 clock monotonicity by maintaining a persistent timestamp file in
5925 /var. Since /var is generally not available in earliest boot or the
5926 initrd, this part of the logic remains in timesyncd, and is not done
5929 * Support for tweaking details in net_cls.class_id through the
5930 NetClass= configuration directive has been removed, as the kernel
5931 people have decided to deprecate that controller in cgroup v2.
5932 Userspace tools such as nftables are moving over to setting rules
5933 that are specific to the full cgroup path of a task, which obsoletes
5934 these controllers anyway. The NetClass= directive is kept around for
5935 legacy compatibility reasons. For a more in-depth description of the
5936 kernel change, please refer to the respective upstream commit:
5938 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
5940 * A new service setting RuntimeMaxSec= has been added that may be used
5941 to specify a maximum runtime for a service. If the timeout is hit, the
5942 service is terminated and put into a failure state.
5944 * A new service setting AmbientCapabilities= has been added. It allows
5945 configuration of additional Linux process capabilities that are
5946 passed to the activated processes. This is only available on very
5949 * The process resource limit settings in service units may now be used
5950 to configure hard and soft limits individually.
5952 * The various libsystemd APIs such as sd-bus or sd-event now publicly
5953 expose support for gcc's __attribute__((cleanup())) C extension.
5954 Specifically, for many object destructor functions alternative
5955 versions have been added that have names suffixed with "p" and take a
5956 pointer to a pointer to the object to destroy, instead of just a
5957 pointer to the object itself. This is useful because these destructor
5958 functions may be used directly as parameters to the cleanup
5959 construct. Internally, systemd has been a heavy user of this GCC
5960 extension for a long time, and with this change similar support is
5961 now available to consumers of the library outside of systemd. Note
5962 that by using this extension in your sources compatibility with old
5963 and strictly ANSI compatible C compilers is lost. However, all gcc or
5964 LLVM versions of recent years support this extension.
5966 * Timer units gained support for a new setting RandomizedDelaySec= that
5967 allows configuring some additional randomized delay to the configured
5968 time. This is useful to spread out timer events to avoid load peaks in
5969 clusters or larger setups.
5971 * Calendar time specifications now support sub-second accuracy.
5973 * Socket units now support listening on SCTP and UDP-lite protocol
5976 * The sd-event API now comes with a full set of man pages.
5978 * Older versions of systemd contained experimental support for
5979 compressing journal files and coredumps with the LZ4 compressor that
5980 was not compatible with the lz4 binary (due to API limitations of the
5981 lz4 library). This support has been removed; only support for files
5982 compatible with the lz4 binary remains. This LZ4 logic is now
5983 officially supported and no longer considered experimental.
5985 * The dkr image import logic has been removed again from importd. dkr's
5986 micro-services focus doesn't fit into the machine image focus of
5987 importd, and quickly got out of date with the upstream dkr API.
5989 * Creation of the /run/lock/lockdev/ directory was dropped from
5990 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
5991 been available for many years. If you still need this, you need to
5992 create your own tmpfiles.d config file with:
5994 d /run/lock/lockdev 0775 root lock -
5996 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
5997 and RebootArgument= have been moved from the [Service] section of
5998 unit files to [Unit], and they are now supported on all unit types,
5999 not just service units. Of course, systemd will continue to
6000 understand these settings also at the old location, in order to
6001 maintain compatibility.
6003 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
6004 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
6005 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
6006 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
6007 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
6008 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
6009 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
6010 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
6011 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
6012 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
6013 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
6014 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
6015 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
6016 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
6017 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
6018 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
6019 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
6020 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
6021 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
6023 — Berlin, 2016-02-11
6027 * A number of properties previously only settable in unit
6028 files are now also available as properties to set when
6029 creating transient units programmatically via the bus, as it
6030 is exposed with systemd-run's --property=
6031 setting. Specifically, these are: SyslogIdentifier=,
6032 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
6033 EnvironmentFile=, ReadWriteDirectories=,
6034 ReadOnlyDirectories=, InaccessibleDirectories=,
6035 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
6037 * When creating transient services via the bus API it is now
6038 possible to pass in a set of file descriptors to use as
6039 STDIN/STDOUT/STDERR for the invoked process.
6041 * Slice units may now be created transiently via the bus APIs,
6042 similar to the way service and scope units may already be
6043 created transiently.
6045 * Wherever systemd expects a calendar timestamp specification
6046 (like in journalctl's --since= and --until= switches) UTC
6047 timestamps are now supported. Timestamps suffixed with "UTC"
6048 are now considered to be in Universal Time Coordinated
6049 instead of the local timezone. Also, timestamps may now
6050 optionally be specified with sub-second accuracy. Both of
6051 these additions also apply to recurring calendar event
6052 specification, such as OnCalendar= in timer units.
6054 * journalctl gained a new "--sync" switch that asks the
6055 journal daemon to write all so far unwritten log messages to
6056 disk and sync the files, before returning.
6058 * systemd-tmpfiles learned two new line types "q" and "Q" that
6059 operate like "v", but also set up a basic btrfs quota
6060 hierarchy when used on a btrfs file system with quota
6063 * tmpfiles' "v", "q" and "Q" will now create a plain directory
6064 instead of a subvolume (even on a btrfs file system) if the
6065 root directory is a plain directory, and not a
6066 subvolume. This should simplify things with certain chroot()
6067 environments which are not aware of the concept of btrfs
6070 * systemd-detect-virt gained a new --chroot switch to detect
6071 whether execution takes place in a chroot() environment.
6073 * CPUAffinity= now takes CPU index ranges in addition to
6076 * The various memory-related resource limit settings (such as
6077 LimitAS=) now understand the usual K, M, G, ... suffixes to
6078 the base of 1024 (IEC). Similar, the time-related resource
6079 limit settings understand the usual min, h, day, ...
6082 * There's a new system.conf setting DefaultTasksMax= to
6083 control the default TasksMax= setting for services and
6084 scopes running on the system. (TasksMax= is the primary
6085 setting that exposes the "pids" cgroup controller on systemd
6086 and was introduced in the previous systemd release.) The
6087 setting now defaults to 512, which means services that are
6088 not explicitly configured otherwise will only be able to
6089 create 512 processes or threads at maximum, from this
6090 version on. Note that this means that thread- or
6091 process-heavy services might need to be reconfigured to set
6092 TasksMax= to a higher value. It is sufficient to set
6093 TasksMax= in these specific unit files to a higher value, or
6094 even "infinity". Similar, there's now a logind.conf setting
6095 UserTasksMax= that defaults to 4096 and limits the total
6096 number of processes or tasks each user may own
6097 concurrently. nspawn containers also have the TasksMax=
6098 value set by default now, to 8192. Note that all of this
6099 only has an effect if the "pids" cgroup controller is
6100 enabled in the kernel. The general benefit of these changes
6101 should be a more robust and safer system, that provides a
6102 certain amount of per-service fork() bomb protection.
6104 * systemd-nspawn gained the new --network-veth-extra= switch
6105 to define additional and arbitrarily-named virtual Ethernet
6106 links between the host and the container.
6108 * A new service execution setting PassEnvironment= has been
6109 added that allows importing select environment variables
6110 from PID1's environment block into the environment block of
6113 * Timer units gained support for a new RemainAfterElapse=
6114 setting which takes a boolean argument. It defaults to on,
6115 exposing behaviour unchanged to previous releases. If set to
6116 off, timer units are unloaded after they elapsed if they
6117 cannot elapse again. This is particularly useful for
6118 transient timer units, which shall not stay around longer
6119 than until they first elapse.
6121 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
6122 default now (the kernel default is 16). This is beneficial
6123 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
6124 allows substantially larger numbers of queued
6125 datagrams. This should increase the capability of systemd to
6126 parallelize boot-up, as logging and sd_notify() are unlikely
6127 to stall execution anymore. If you need to change the value
6128 from the new defaults, use the usual sysctl.d/ snippets.
6130 * The compression framing format used by the journal or
6131 coredump processing has changed to be in line with what the
6132 official LZ4 tools generate. LZ4 compression support in
6133 systemd was considered unsupported previously, as the format
6134 was not compatible with the normal tools. With this release
6135 this has changed now, and it is hence safe for downstream
6136 distributions to turn it on. While not compressing as well
6137 as the XZ, LZ4 is substantially faster, which makes
6138 it a good default choice for the compression logic in the
6139 journal and in coredump handling.
6141 * Any reference to /etc/mtab has been dropped from
6142 systemd. The file has been obsolete since a while, but
6143 systemd refused to work on systems where it was incorrectly
6144 set up (it should be a symlink or non-existent). Please make
6145 sure to update to util-linux 2.27.1 or newer in conjunction
6146 with this systemd release, which also drops any reference to
6147 /etc/mtab. If you maintain a distribution make sure that no
6148 software you package still references it, as this is a
6149 likely source of bugs. There's also a glibc bug pending,
6150 asking for removal of any reference to this obsolete file:
6152 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
6154 Note that only util-linux versions built with
6155 --enable-libmount-force-mountinfo are supported.
6157 * Support for the ".snapshot" unit type has been removed. This
6158 feature turned out to be little useful and little used, and
6159 has now been removed from the core and from systemctl.
6161 * The dependency types RequiresOverridable= and
6162 RequisiteOverridable= have been removed from systemd. They
6163 have been used only very sparingly to our knowledge and
6164 other options that provide a similar effect (such as
6165 systemctl --mode=ignore-dependencies) are much more useful
6166 and commonly used. Moreover, they were only half-way
6167 implemented as the option to control behaviour regarding
6168 these dependencies was never added to systemctl. By removing
6169 these dependency types the execution engine becomes a bit
6170 simpler. Unit files that use these dependencies should be
6171 changed to use the non-Overridable dependency types
6172 instead. In fact, when parsing unit files with these
6173 options, that's what systemd will automatically convert them
6174 too, but it will also warn, asking users to fix the unit
6175 files accordingly. Removal of these dependency types should
6176 only affect a negligible number of unit files in the wild.
6178 * Behaviour of networkd's IPForward= option changed
6179 (again). It will no longer maintain a per-interface setting,
6180 but propagate one way from interfaces where this is enabled
6181 to the global kernel setting. The global setting will be
6182 enabled when requested by a network that is set up, but
6183 never be disabled again. This change was made to make sure
6184 IPv4 and IPv6 behaviour regarding packet forwarding is
6185 similar (as the Linux IPv6 stack does not support
6186 per-interface control of this setting) and to minimize
6189 * In unit files the behaviour of %u, %U, %h, %s has
6190 changed. These specifiers will now unconditionally resolve
6191 to the various user database fields of the user that the
6192 systemd instance is running as, instead of the user
6193 configured in the specific unit via User=. Note that this
6194 effectively doesn't change much, as resolving of these
6195 specifiers was already turned off in the --system instance
6196 of systemd, as we cannot do NSS lookups from PID 1. In the
6197 --user instance of systemd these specifiers where correctly
6198 resolved, but hardly made any sense, since the user instance
6199 lacks privileges to do user switches anyway, and User= is
6200 hence useless. Moreover, even in the --user instance of
6201 systemd behaviour was awkward as it would only take settings
6202 from User= assignment placed before the specifier into
6203 account. In order to unify and simplify the logic around
6204 this the specifiers will now always resolve to the
6205 credentials of the user invoking the manager (which in case
6206 of PID 1 is the root user).
6208 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
6209 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
6210 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
6211 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
6212 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
6213 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
6214 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
6215 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
6216 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
6217 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
6218 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
6219 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
6220 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
6221 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
6224 — Berlin, 2015-11-18
6228 * systemd now depends on util-linux v2.27. More specifically,
6229 the newly added mount monitor feature in libmount now
6230 replaces systemd's former own implementation.
6232 * libmount mandates /etc/mtab not to be regular file, and
6233 systemd now enforces this condition at early boot.
6234 /etc/mtab has been deprecated and warned about for a very
6235 long time, so systems running systemd should already have
6236 stopped having this file around as anything else than a
6237 symlink to /proc/self/mounts.
6239 * Support for the "pids" cgroup controller has been added. It
6240 allows accounting the number of tasks in a cgroup and
6241 enforcing limits on it. This adds two new setting
6242 TasksAccounting= and TasksMax= to each unit, as well as a
6243 global option DefaultTasksAccounting=.
6245 * Support for the "net_cls" cgroup controller has been added.
6246 It allows assigning a net class ID to each task in the
6247 cgroup, which can then be used in firewall rules and traffic
6248 shaping configurations. Note that the kernel netfilter net
6249 class code does not currently work reliably for ingress
6250 packets on unestablished sockets.
6252 This adds a new config directive called NetClass= to CGroup
6253 enabled units. Allowed values are positive numbers for fixed
6254 assignments and "auto" for picking a free value
6257 * 'systemctl is-system-running' now returns 'offline' if the
6258 system is not booted with systemd. This command can now be
6259 used as a substitute for 'systemd-notify --booted'.
6261 * Watchdog timeouts have been increased to 3 minutes for all
6262 in-tree service files. Apparently, disk IO issues are more
6263 frequent than we hoped, and user reported >1 minute waiting
6266 * 'machine-id-commit' functionality has been merged into
6267 'machine-id-setup --commit'. The separate binary has been
6270 * The WorkingDirectory= directive in unit files may now be set
6271 to the special value '~'. In this case, the working
6272 directory is set to the home directory of the user
6273 configured in User=.
6275 * "machinectl shell" will now open the shell in the home
6276 directory of the selected user by default.
6278 * The CrashChVT= configuration file setting is renamed to
6279 CrashChangeVT=, following our usual logic of not
6280 abbreviating unnecessarily. The old directive is still
6281 supported for compat reasons. Also, this directive now takes
6282 an integer value between 1 and 63, or a boolean value. The
6283 formerly supported '-1' value for disabling stays around for
6286 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
6287 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
6288 RootDirectory= properties can now be set for transient
6291 * The systemd-analyze tool gained a new "set-log-target" verb
6292 to change the logging target the system manager logs to
6293 dynamically during runtime. This is similar to how
6294 "systemd-analyze set-log-level" already changes the log
6297 * In nspawn /sys is now mounted as tmpfs, with only a selected
6298 set of subdirectories mounted in from the real sysfs. This
6299 enhances security slightly, and is useful for ensuring user
6300 namespaces work correctly.
6302 * Support for USB FunctionFS activation has been added. This
6303 allows implementation of USB gadget services that are
6304 activated as soon as they are requested, so that they don't
6305 have to run continuously, similar to classic socket
6308 * The "systemctl exit" command now optionally takes an
6309 additional parameter that sets the exit code to return from
6310 the systemd manager when exiting. This is only relevant when
6311 running the systemd user instance, or when running the
6312 system instance in a container.
6314 * sd-bus gained the new API calls sd_bus_path_encode_many()
6315 and sd_bus_path_decode_many() that allow easy encoding and
6316 decoding of multiple identifier strings inside a D-Bus
6317 object path. Another new call sd_bus_default_flush_close()
6318 has been added to flush and close per-thread default
6321 * systemd-cgtop gained support for a -M/--machine= switch to
6322 show the control groups within a certain container only.
6324 * "systemctl kill" gained support for an optional --fail
6325 switch. If specified the requested operation will fail of no
6326 processes have been killed, because the unit had no
6327 processes attached, or similar.
6329 * A new systemd.crash_reboot=1 kernel command line option has
6330 been added that triggers a reboot after crashing. This can
6331 also be set through CrashReboot= in systemd.conf.
6333 * The RuntimeDirectory= setting now understands unit
6334 specifiers like %i or %f.
6336 * A new (still internal) library API sd-ipv4acd has been added,
6337 that implements address conflict detection for IPv4. It's
6338 based on code from sd-ipv4ll, and will be useful for
6339 detecting DHCP address conflicts.
6341 * File descriptors passed during socket activation may now be
6342 named. A new API sd_listen_fds_with_names() is added to
6343 access the names. The default names may be overridden,
6344 either in the .socket file using the FileDescriptorName=
6345 parameter, or by passing FDNAME= when storing the file
6346 descriptors using sd_notify().
6348 * systemd-networkd gained support for:
6350 - Setting the IPv6 Router Advertisement settings via
6351 IPv6AcceptRouterAdvertisements= in .network files.
6353 - Configuring the HelloTimeSec=, MaxAgeSec= and
6354 ForwardDelaySec= bridge parameters in .netdev files.
6356 - Configuring PreferredSource= for static routes in
6359 * The "ask-password" framework used to query for LUKS harddisk
6360 passwords or SSL passwords during boot gained support for
6361 caching passwords in the kernel keyring, if it is
6362 available. This makes sure that the user only has to type in
6363 a passphrase once if there are multiple objects to unlock
6364 with the same one. Previously, such password caching was
6365 available only when Plymouth was used; this moves the
6366 caching logic into the systemd codebase itself. The
6367 "systemd-ask-password" utility gained a new --keyname=
6368 switch to control which kernel keyring key to use for
6369 caching a password in. This functionality is also useful for
6370 enabling display managers such as gdm to automatically
6371 unlock the user's GNOME keyring if its passphrase, the
6372 user's password and the harddisk password are the same, if
6373 gdm-autologin is used.
6375 * When downloading tar or raw images using "machinectl
6376 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
6377 file is now also downloaded, if it is available and stored
6378 next to the image file.
6380 * Units of type ".socket" gained a new boolean setting
6381 Writable= which is only useful in conjunction with
6382 ListenSpecial=. If true, enables opening the specified
6383 special file in O_RDWR mode rather than O_RDONLY mode.
6385 * systemd-rfkill has been reworked to become a singleton
6386 service that is activated through /dev/rfkill on each rfkill
6387 state change and saves the settings to disk. This way,
6388 systemd-rfkill is now compatible with devices that exist
6389 only intermittendly, and even restores state if the previous
6390 system shutdown was abrupt rather than clean.
6392 * The journal daemon gained support for vacuuming old journal
6393 files controlled by the number of files that shall remain,
6394 in addition to the already existing control by size and by
6395 date. This is useful as journal interleaving performance
6396 degrades with too many separate journal files, and allows
6397 putting an effective limit on them. The new setting defaults
6398 to 100, but this may be changed by setting SystemMaxFiles=
6399 and RuntimeMaxFiles= in journald.conf. Also, the
6400 "journalctl" tool gained the new --vacuum-files= switch to
6401 manually vacuum journal files to leave only the specified
6402 number of files in place.
6404 * udev will now create /dev/disk/by-path links for ATA devices
6405 on kernels where that is supported.
6407 * Galician, Serbian, Turkish and Korean translations were added.
6409 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
6410 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
6411 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
6412 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
6413 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
6414 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
6415 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
6416 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
6417 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
6418 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
6419 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
6420 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
6421 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
6422 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
6423 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
6424 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
6425 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
6426 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
6428 — Berlin, 2015-10-07
6432 * The DHCP implementation of systemd-networkd gained a set of
6435 - The DHCP server now supports emitting DNS and NTP
6436 information. It may be enabled and configured via
6437 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
6438 and NTP information is enabled, but no servers are
6439 configured, the corresponding uplink information (if there
6440 is any) is propagated.
6442 - Server and client now support transmission and reception
6443 of timezone information. It can be configured via the
6444 newly introduced network options UseTimezone=,
6445 EmitTimezone=, and Timezone=. Transmission of timezone
6446 information is enabled between host and containers by
6447 default now: the container will change its local timezone
6448 to what the host has set.
6450 - Lease timeouts can now be configured via
6451 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
6453 - The DHCP server improved on the stability of
6454 leases. Clients are more likely to get the same lease
6455 information back, even if the server loses state.
6457 - The DHCP server supports two new configuration options to
6458 control the lease address pool metrics, PoolOffset= and
6461 * The encapsulation limit of tunnels in systemd-networkd may
6462 now be configured via 'EncapsulationLimit='. It allows
6463 modifying the maximum additional levels of encapsulation
6464 that are permitted to be prepended to a packet.
6466 * systemd now supports the concept of user buses replacing
6467 session buses, if used with dbus-1.10 (and enabled via dbus
6468 --enable-user-session). It previously only supported this on
6469 kdbus-enabled systems, and this release expands this to
6470 'dbus-daemon' systems.
6472 * systemd-networkd now supports predictable interface names
6475 * systemd now optionally supports the new Linux kernel
6476 "unified" control group hierarchy. If enabled via the kernel
6477 command-line option 'systemd.unified_cgroup_hierarchy=1',
6478 systemd will try to mount the unified cgroup hierarchy
6479 directly on /sys/fs/cgroup. If not enabled, or not
6480 available, systemd will fall back to the legacy cgroup
6481 hierarchy setup, as before. Host system and containers can
6482 mix and match legacy and unified hierarchies as they
6483 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
6484 environment variable to individually select the hierarchy to
6485 use for executed containers. By default, nspawn will use the
6486 unified hierarchy for the containers if the host uses the
6487 unified hierarchy, and the legacy hierarchy otherwise.
6488 Please note that at this point the unified hierarchy is an
6489 experimental kernel feature and is likely to change in one
6490 of the next kernel releases. Therefore, it should not be
6491 enabled by default in downstream distributions yet. The
6492 minimum required kernel version for the unified hierarchy to
6493 work is 4.2. Note that when the unified hierarchy is used
6494 for the first time delegated access to controllers is
6495 safe. Because of this systemd-nspawn containers will get
6496 access to controllers now, as will systemd user
6497 sessions. This means containers and user sessions may now
6498 manage their own resources, partitioning up what the system
6501 * A new special scope unit "init.scope" has been introduced
6502 that encapsulates PID 1 of the system. It may be used to
6503 determine resource usage and enforce resource limits on PID
6504 1 itself. PID 1 hence moved out of the root of the control
6507 * The cgtop tool gained support for filtering out kernel
6508 threads when counting tasks in a control group. Also, the
6509 count of processes is now recursively summed up by
6510 default. Two options -k and --recursive= have been added to
6511 revert to old behaviour. The tool has also been updated to
6512 work correctly in containers now.
6514 * systemd-nspawn's --bind= and --bind-ro= options have been
6515 extended to allow creation of non-recursive bind mounts.
6517 * libsystemd gained two new calls sd_pid_get_cgroup() and
6518 sd_peer_get_cgroup() which return the control group path of
6519 a process or peer of a connected AF_UNIX socket. This
6520 function call is particularly useful when implementing
6521 delegated subtrees support in the control group hierarchy.
6523 * The "sd-event" event loop API of libsystemd now supports
6524 correct dequeuing of real-time signals, without losing
6527 * When systemd requests a polkit decision when managing units it
6528 will now add additional fields to the request, including unit
6529 name and desired operation. This enables more powerful polkit
6530 policies, that make decisions depending on these parameters.
6532 * nspawn learnt support for .nspawn settings files, that may
6533 accompany the image files or directories of containers, and
6534 may contain additional settings for the container. This is
6535 an alternative to configuring container parameters via the
6536 nspawn command line.
6538 Contributions from: Cristian Rodríguez, Daniel Mack, David
6539 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
6540 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
6541 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
6542 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
6543 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
6544 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
6545 Andersen, Tom Gundersen, Torstein Husebø
6547 — Berlin, 2015-09-08
6551 * machinectl gained a new verb 'shell' which opens a fresh
6552 shell on the target container or the host. It is similar to
6553 the existing 'login' command of machinectl, but spawns the
6554 shell directly without prompting for username or
6555 password. The pseudo machine '.host' now refers to the local
6556 host and is used by default. Hence, 'machinectl shell' can
6557 be used as replacement for 'su -' which spawns a session as
6558 a fresh systemd unit in a way that is fully isolated from
6559 the originating session.
6561 * systemd-networkd learned to cope with private-zone DHCP
6562 options and allows other programs to query the values.
6564 * SELinux access control when enabling/disabling units is no
6565 longer enforced with this release. The previous implementation
6566 was incorrect, and a new corrected implementation is not yet
6567 available. As unit file operations are still protected via
6568 polkit and D-Bus policy this is not a security problem. Yet,
6569 distributions which care about optimal SELinux support should
6570 probably not stabilize on this release.
6572 * sd-bus gained support for matches of type "arg0has=", that
6573 test for membership of strings in string arrays sent in bus
6576 * systemd-resolved now dumps the contents of its DNS and LLMNR
6577 caches to the logs on reception of the SIGUSR1 signal. This
6578 is useful to debug DNS behaviour.
6580 * The coredumpctl tool gained a new --directory= option to
6581 operate on journal files in a specific directory.
6583 * "systemctl reboot" and related commands gained a new
6584 "--message=" option which may be used to set a free-text
6585 wall message when shutting down or rebooting the
6586 system. This message is also logged, which is useful for
6587 figuring out the reason for a reboot or shutdown a
6590 * The "systemd-resolve-host" tool's -i switch now takes
6591 network interface numbers as alternative to interface names.
6593 * A new unit file setting for services has been introduced:
6594 UtmpMode= allows configuration of how precisely systemd
6595 handles utmp and wtmp entries for the service if this is
6596 enabled. This allows writing services that appear similar to
6597 user sessions in the output of the "w", "who", "last" and
6600 * systemd-resolved will now locally synthesize DNS resource
6601 records for the "localhost" and "gateway" domains as well as
6602 the local hostname. This should ensure that clients querying
6603 RRs via resolved will get similar results as those going via
6604 NSS, if nss-myhostname is enabled.
6606 Contributions from: Alastair Hughes, Alex Crawford, Daniel
6607 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
6608 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
6609 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
6610 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
6611 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
6612 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
6613 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
6614 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
6615 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
6616 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
6617 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
6619 — Berlin, 2015-08-27
6623 * The systemd-efi-boot-generator functionality was merged into
6624 systemd-gpt-auto-generator.
6626 * systemd-networkd now supports Group Policy for vxlan
6627 devices. It can be enabled via the new boolean configuration
6628 option called 'GroupPolicyExtension='.
6630 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
6631 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
6632 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
6634 — Berlin, 2015-07-31
6638 * The python-systemd code has been removed from the systemd repository.
6639 A new repository has been created which accommodates the code from
6640 now on, and we kindly ask distributions to create a separate package
6641 for this: https://github.com/systemd/python-systemd
6643 * The systemd daemon will now reload its main configuration
6644 (/etc/systemd/system.conf) on daemon-reload.
6646 * sd-dhcp now exposes vendor specific extensions via
6647 sd_dhcp_lease_get_vendor_specific().
6649 * systemd-networkd gained a number of new configuration options.
6651 - A new boolean configuration option for TAP devices called
6652 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
6653 device, thus allowing to send and receive GSO packets.
6655 - A new tunnel configuration option called 'CopyDSCP='.
6656 If enabled, the DSCP field of ip6 tunnels is copied into the
6657 decapsulated packet.
6659 - A set of boolean bridge configuration options were added.
6660 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
6661 and 'UnicastFlood=' are now parsed by networkd and applied to the
6662 respective bridge link device via the respective IFLA_BRPORT_*
6665 - A new string configuration option to override the hostname sent
6666 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
6667 is true, networkd will use the configured hostname instead of the
6668 system hostname when sending DHCP requests.
6670 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
6671 networkd will configure the IPv6 flow-label of the tunnel device
6672 according to RFC2460.
6674 - The 'macvtap' virtual network devices are now supported, similar to
6675 the already supported 'macvlan' devices.
6677 * systemd-resolved now implements RFC5452 to improve resilience against
6678 cache poisoning. Additionally, source port randomization is enabled
6679 by default to further protect against DNS spoofing attacks.
6681 * nss-mymachines now supports translating UIDs and GIDs of running
6682 containers with user-namespaces enabled. If a container 'foo'
6683 translates a host uid 'UID' to the container uid 'TUID', then
6684 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
6685 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
6686 mapped as 'vg-foo-TGID'.
6688 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
6689 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
6690 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
6691 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
6692 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
6693 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
6694 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
6695 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
6696 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
6697 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
6699 — Berlin, 2015-07-29
6703 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
6704 There are no known issues with current sysfs, and udev does not need
6705 or should be used to work around such bugs.
6707 * udev does no longer enable USB HID power management. Several reports
6708 indicate, that some devices cannot handle that setting.
6710 * The udev accelerometer helper was removed. The functionality
6711 is now fully included in iio-sensor-proxy. But this means,
6712 older iio-sensor-proxy versions will no longer provide
6713 accelerometer/orientation data with this systemd version.
6714 Please upgrade iio-sensor-proxy to version 1.0.
6716 * networkd gained a new configuration option IPv6PrivacyExtensions=
6717 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
6718 for Stateless Address") on selected networks.
6720 * For the sake of fewer build-time dependencies and less code in the
6721 main repository, the python bindings are about to be removed in the
6722 next release. A new repository has been created which accommodates
6723 the code from now on, and we kindly ask distributions to create a
6724 separate package for this. The removal will take place in v223.
6726 https://github.com/systemd/python-systemd
6728 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
6729 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
6730 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
6731 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
6732 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
6733 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
6734 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
6735 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
6736 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
6737 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
6739 — Berlin, 2015-07-07
6743 * The sd-bus.h and sd-event.h APIs have now been declared
6744 stable and have been added to the official interface of
6745 libsystemd.so. sd-bus implements an alternative D-Bus client
6746 library, that is relatively easy to use, very efficient and
6747 supports both classic D-Bus as well as kdbus as transport
6748 backend. sd-event is a generic event loop abstraction that
6749 is built around Linux epoll, but adds features such as event
6750 prioritization or efficient timer handling. Both APIs are good
6751 choices for C programs looking for a bus and/or event loop
6752 implementation that is minimal and does not have to be
6753 portable to other kernels.
6755 * kdbus support is no longer compile-time optional. It is now
6756 always built-in. However, it can still be disabled at
6757 runtime using the kdbus=0 kernel command line setting, and
6758 that setting may be changed to default to off, by specifying
6759 --disable-kdbus at build-time. Note though that the kernel
6760 command line setting has no effect if the kdbus.ko kernel
6761 module is not installed, in which case kdbus is (obviously)
6762 also disabled. We encourage all downstream distributions to
6763 begin testing kdbus by adding it to the kernel images in the
6764 development distributions, and leaving kdbus support in
6767 * The minimal required util-linux version has been bumped to
6770 * Support for chkconfig (--enable-chkconfig) was removed in
6771 favor of calling an abstraction tool
6772 /lib/systemd/systemd-sysv-install. This needs to be
6773 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
6774 in README for details.
6776 * If there's a systemd unit and a SysV init script for the
6777 same service name, and the user executes "systemctl enable"
6778 for it (or a related call), then this will now enable both
6779 (or execute the related operation on both), not just the
6782 * The libudev API documentation has been converted from gtkdoc
6785 * gudev has been removed from the systemd tree, it is now an
6788 * The systemd-cgtop tool learnt a new --raw switch to generate
6789 "raw" (machine parsable) output.
6791 * networkd's IPForwarding= .network file setting learnt the
6792 new setting "kernel", which ensures that networkd does not
6793 change the IP forwarding sysctl from the default kernel
6796 * The systemd-logind bus API now exposes a new boolean
6797 property "Docked" that reports whether logind considers the
6798 system "docked", i.e. connected to a docking station or not.
6800 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
6801 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
6802 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
6803 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
6804 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
6805 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
6806 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
6807 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
6808 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
6809 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
6810 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
6811 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
6812 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
6813 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
6814 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
6815 Fink, Zbigniew Jędrzejewski-Szmek
6817 — Berlin, 2015-06-19
6821 * The gudev library has been extracted into a separate repository
6822 available at: https://git.gnome.org/browse/libgudev/
6823 It is now managed as part of the Gnome project. Distributions
6824 are recommended to pass --disable-gudev to systemd and use
6825 gudev from the Gnome project instead. gudev is still included
6826 in systemd, for now. It will be removed soon, though. Please
6827 also see the announcement-thread on systemd-devel:
6828 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
6830 * systemd now exposes a CPUUsageNSec= property for each
6831 service unit on the bus, that contains the overall consumed
6832 CPU time of a service (the sum of what each process of the
6833 service consumed). This value is only available if
6834 CPUAccounting= is turned on for a service, and is then shown
6835 in the "systemctl status" output.
6837 * Support for configuring alternative mappings of the old SysV
6838 runlevels to systemd targets has been removed. They are now
6839 hardcoded in a way that runlevels 2, 3, 4 all map to
6840 multi-user.target and 5 to graphical.target (which
6841 previously was already the default behaviour).
6843 * The auto-mounter logic gained support for mount point
6844 expiry, using a new TimeoutIdleSec= setting in .automount
6845 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
6847 * The EFI System Partition (ESP) as mounted to /boot by
6848 systemd-efi-boot-generator will now be unmounted
6849 automatically after 2 minutes of not being used. This should
6850 minimize the risk of ESP corruptions.
6852 * New /etc/fstab options x-systemd.requires= and
6853 x-systemd.requires-mounts-for= are now supported to express
6854 additional dependencies for mounts. This is useful for
6855 journalling file systems that support external journal
6856 devices or overlay file systems that require underlying file
6857 systems to be mounted.
6859 * systemd does not support direct live-upgrades (via systemctl
6860 daemon-reexec) from versions older than v44 anymore. As no
6861 distribution we are aware of shipped such old versions in a
6862 stable release this should not be problematic.
6864 * When systemd forks off a new per-connection service instance
6865 it will now set the $REMOTE_ADDR environment variable to the
6866 remote IP address, and $REMOTE_PORT environment variable to
6867 the remote IP port. This behaviour is similar to the
6868 corresponding environment variables defined by CGI.
6870 * systemd-networkd gained support for uplink failure
6871 detection. The BindCarrier= option allows binding interface
6872 configuration dynamically to the link sense of other
6873 interfaces. This is useful to achieve behaviour like in
6876 * systemd-networkd gained support for configuring the DHCP
6877 client identifier to use when requesting leases.
6879 * systemd-networkd now has a per-network UseNTP= option to
6880 configure whether NTP server information acquired via DHCP
6881 is passed on to services like systemd-timesyncd.
6883 * systemd-networkd gained support for vti6 tunnels.
6885 * Note that systemd-networkd manages the sysctl variable
6886 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
6887 it is configured for since v219. The variable controls IP
6888 forwarding, and is a per-interface alternative to the global
6889 /proc/sys/net/ipv[46]/ip_forward. This setting is
6890 configurable in the IPForward= option, which defaults to
6891 "no". This means if networkd is used for an interface it is
6892 no longer sufficient to set the global sysctl option to turn
6893 on IP forwarding! Instead, the .network file option
6894 IPForward= needs to be turned on! Note that the
6895 implementation of this behaviour was broken in v219 and has
6898 * Many bonding and vxlan options are now configurable in
6901 * systemd-nspawn gained a new --property= setting to set unit
6902 properties for the container scope. This is useful for
6903 setting resource parameters (e.g. "CPUShares=500") on
6904 containers started from the command line.
6906 * systemd-nspawn gained a new --private-users= switch to make
6907 use of user namespacing available on recent Linux kernels.
6909 * systemd-nspawn may now be called as part of a shell pipeline
6910 in which case the pipes used for stdin and stdout are passed
6911 directly to the process invoked in the container, without
6912 indirection via a pseudo tty.
6914 * systemd-nspawn gained a new switch to control the UNIX
6915 signal to use when killing the init process of the container
6918 * systemd-nspawn gained a new --overlay= switch for mounting
6919 overlay file systems into the container using the new kernel
6922 * When a container image is imported via systemd-importd and
6923 the host file system is not btrfs, a loopback block device
6924 file is created in /var/lib/machines.raw with a btrfs file
6925 system inside. It is then mounted to /var/lib/machines to
6926 enable btrfs features for container management. The loopback
6927 file and btrfs file system is grown as needed when container
6928 images are imported via systemd-importd.
6930 * systemd-machined/systemd-importd gained support for btrfs
6931 quota, to enforce container disk space limits on disk. This
6932 is exposed in "machinectl set-limit".
6934 * systemd-importd now can import containers from local .tar,
6935 .raw and .qcow2 images, and export them to .tar and .raw. It
6936 can also import dkr v2 images now from the network (on top
6939 * systemd-importd gained support for verifying downloaded
6940 images with gpg2 (previously only gpg1 was supported).
6942 * systemd-machined, systemd-logind, systemd: most bus calls are
6943 now accessible to unprivileged processes via polkit. Also,
6944 systemd-logind will now allow users to kill their own sessions
6945 without further privileges or authorization.
6947 * systemd-shutdownd has been removed. This service was
6948 previously responsible for implementing scheduled shutdowns
6949 as exposed in /usr/bin/shutdown's time parameter. This
6950 functionality has now been moved into systemd-logind and is
6951 accessible via a bus interface.
6953 * "systemctl reboot" gained a new switch --firmware-setup that
6954 can be used to reboot into the EFI firmware setup, if that
6955 is available. systemd-logind now exposes an API on the bus
6956 to trigger such reboots, in case graphical desktop UIs want
6957 to cover this functionality.
6959 * "systemctl enable", "systemctl disable" and "systemctl mask"
6960 now support a new "--now" switch. If specified the units
6961 that are enabled will also be started, and the ones
6962 disabled/masked also stopped.
6964 * The Gummiboot EFI boot loader tool has been merged into
6965 systemd, and renamed to "systemd-boot". The bootctl tool has been
6966 updated to support systemd-boot.
6968 * An EFI kernel stub has been added that may be used to create
6969 kernel EFI binaries that contain not only the actual kernel,
6970 but also an initrd, boot splash, command line and OS release
6971 information. This combined binary can then be signed as a
6972 single image, so that the firmware can verify it all in one
6973 step. systemd-boot has special support for EFI binaries created
6974 like this and can extract OS release information from them
6975 and show them in the boot menu. This functionality is useful
6976 to implement cryptographically verified boot schemes.
6978 * Optional support has been added to systemd-fsck to pass
6979 fsck's progress report to an AF_UNIX socket in the file
6982 * udev will no longer create device symlinks for all block devices by
6983 default. A deny list for excluding special block devices from this
6984 logic has been turned into a allow list that requires picking block
6985 devices explicitly that require device symlinks.
6987 * A new (currently still internal) API sd-device.h has been
6988 added to libsystemd. This modernized API is supposed to
6989 replace libudev eventually. In fact, already much of libudev
6990 is now just a wrapper around sd-device.h.
6992 * A new hwdb database for storing metadata about pointing
6993 stick devices has been added.
6995 * systemd-tmpfiles gained support for setting file attributes
6996 similar to the "chattr" tool with new 'h' and 'H' lines.
6998 * systemd-journald will no longer unconditionally set the
6999 btrfs NOCOW flag on new journal files. This is instead done
7000 with tmpfiles snippet using the new 'h' line type. This
7001 allows easy disabling of this logic, by masking the
7002 journal-nocow.conf tmpfiles file.
7004 * systemd-journald will now translate audit message types to
7005 human readable identifiers when writing them to the
7006 journal. This should improve readability of audit messages.
7008 * The LUKS logic gained support for the offset= and skip=
7009 options in /etc/crypttab, as previously implemented by
7012 * /usr/lib/os-release gained a new optional field VARIANT= for
7013 distributions that support multiple variants (such as a
7014 desktop edition, a server edition, ...)
7016 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
7017 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
7018 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
7019 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
7020 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
7021 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
7022 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
7023 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
7024 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
7025 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
7026 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
7027 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
7028 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
7029 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
7030 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
7031 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
7032 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
7033 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
7034 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
7035 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
7036 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
7037 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
7038 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
7039 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
7040 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
7041 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
7042 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
7044 — Berlin, 2015-05-22
7048 * Introduce a new API "sd-hwdb.h" for querying the hardware
7049 metadata database. With this minimal interface one can query
7050 and enumerate the udev hwdb, decoupled from the old libudev
7051 library. libudev's interface for this is now only a wrapper
7052 around sd-hwdb. A new tool systemd-hwdb has been added to
7053 interface with and update the database.
7055 * When any of systemd's tools copies files (for example due to
7056 tmpfiles' C lines) a btrfs reflink will attempted first,
7057 before bytewise copying is done.
7059 * systemd-nspawn gained a new --ephemeral switch. When
7060 specified a btrfs snapshot is taken of the container's root
7061 directory, and immediately removed when the container
7062 terminates again. Thus, a container can be started whose
7063 changes never alter the container's root directory, and are
7064 lost on container termination. This switch can also be used
7065 for starting a container off the root file system of the
7066 host without affecting the host OS. This switch is only
7067 available on btrfs file systems.
7069 * systemd-nspawn gained a new --template= switch. It takes the
7070 path to a container tree to use as template for the tree
7071 specified via --directory=, should that directory be
7072 missing. This allows instantiating containers dynamically,
7073 on first run. This switch is only available on btrfs file
7076 * When a .mount unit refers to a mount point on which multiple
7077 mounts are stacked, and the .mount unit is stopped all of
7078 the stacked mount points will now be unmounted until no
7079 mount point remains.
7081 * systemd now has an explicit notion of supported and
7082 unsupported unit types. Jobs enqueued for unsupported unit
7083 types will now fail with an "unsupported" error code. More
7084 specifically .swap, .automount and .device units are not
7085 supported in containers, .busname units are not supported on
7086 non-kdbus systems. .swap and .automount are also not
7087 supported if their respective kernel compile time options
7090 * machinectl gained support for two new "copy-from" and
7091 "copy-to" commands for copying files from a running
7092 container to the host or vice versa.
7094 * machinectl gained support for a new "bind" command to bind
7095 mount host directories into local containers. This is
7096 currently only supported for nspawn containers.
7098 * networkd gained support for configuring bridge forwarding
7099 database entries (fdb) from .network files.
7101 * A new tiny daemon "systemd-importd" has been added that can
7102 download container images in tar, raw, qcow2 or dkr formats,
7103 and make them available locally in /var/lib/machines, so
7104 that they can run as nspawn containers. The daemon can GPG
7105 verify the downloads (not supported for dkr, since it has no
7106 provisions for verifying downloads). It will transparently
7107 decompress bz2, xz, gzip compressed downloads if necessary,
7108 and restore sparse files on disk. The daemon uses privilege
7109 separation to ensure the actual download logic runs with
7110 fewer privileges than the daemon itself. machinectl has
7111 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
7112 make the functionality of importd available to the
7113 user. With this in place the Fedora and Ubuntu "Cloud"
7114 images can be downloaded and booted as containers unmodified
7115 (the Fedora images lack the appropriate GPG signature files
7116 currently, so they cannot be verified, but this will change
7117 soon, hopefully). Note that downloading images is currently
7118 only fully supported on btrfs.
7120 * machinectl is now able to list container images found in
7121 /var/lib/machines, along with some metadata about sizes of
7122 disk and similar. If the directory is located on btrfs and
7123 quota is enabled, this includes quota display. A new command
7124 "image-status" has been added that shows additional
7125 information about images.
7127 * machinectl is now able to clone container images
7128 efficiently, if the underlying file system (btrfs) supports
7129 it, with the new "machinectl clone" command. It also
7130 gained commands for renaming and removing images, as well as
7131 marking them read-only or read-write (supported also on
7132 legacy file systems).
7134 * networkd gained support for collecting LLDP network
7135 announcements, from hardware that supports this. This is
7136 shown in networkctl output.
7138 * systemd-run gained support for a new -t (--pty) switch for
7139 invoking a binary on a pty whose input and output is
7140 connected to the invoking terminal. This allows executing
7141 processes as system services while interactively
7142 communicating with them via the terminal. Most interestingly
7143 this is supported across container boundaries. Invoking
7144 "systemd-run -t /bin/bash" is an alternative to running a
7145 full login session, the difference being that the former
7146 will not register a session, nor go through the PAM session
7149 * tmpfiles gained support for a new "v" line type for creating
7150 btrfs subvolumes. If the underlying file system is a legacy
7151 file system, this automatically degrades to creating a
7152 normal directory. Among others /var/lib/machines is now
7153 created like this at boot, should it be missing.
7155 * The directory /var/lib/containers/ has been deprecated and
7156 been replaced by /var/lib/machines. The term "machines" has
7157 been used in the systemd context as generic term for both
7158 VMs and containers, and hence appears more appropriate for
7159 this, as the directory can also contain raw images bootable
7162 * systemd-nspawn when invoked with -M but without --directory=
7163 or --image= is now capable of searching for the container
7164 root directory, subvolume or disk image automatically, in
7165 /var/lib/machines. systemd-nspawn@.service has been updated
7166 to make use of this, thus allowing it to be used for raw
7169 * A new machines.target unit has been introduced that is
7170 supposed to group all containers/VMs invoked as services on
7171 the system. systemd-nspawn@.service has been updated to
7172 integrate with that.
7174 * machinectl gained a new "start" command, for invoking a
7175 container as a service. "machinectl start foo" is mostly
7176 equivalent to "systemctl start systemd-nspawn@foo.service",
7177 but handles escaping in a nicer way.
7179 * systemd-nspawn will now mount most of the cgroupfs tree
7180 read-only into each container, with the exception of the
7181 container's own subtree in the name=systemd hierarchy.
7183 * journald now sets the special FS_NOCOW file flag for its
7184 journal files. This should improve performance on btrfs, by
7185 avoiding heavy fragmentation when journald's write-pattern
7186 is used on COW file systems. It degrades btrfs' data
7187 integrity guarantees for the files to the same levels as for
7188 ext3/ext4 however. This should be OK though as journald does
7189 its own data integrity checks and all its objects are
7190 checksummed on disk. Also, journald should handle btrfs disk
7191 full events a lot more gracefully now, by processing SIGBUS
7192 errors, and not relying on fallocate() anymore.
7194 * When journald detects that journal files it is writing to
7195 have been deleted it will immediately start new journal
7198 * systemd now provides a way to store file descriptors
7199 per-service in PID 1. This is useful for daemons to ensure
7200 that fds they require are not lost during a daemon
7201 restart. The fds are passed to the daemon on the next
7202 invocation in the same way socket activation fds are
7203 passed. This is now used by journald to ensure that the
7204 various sockets connected to all the system's stdout/stderr
7205 are not lost when journald is restarted. File descriptors
7206 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
7207 an extension to sd_notify(). Note that a limit is enforced
7208 on the number of fds a service can store in PID 1, and it
7209 defaults to 0, so that no fds may be stored, unless this is
7210 explicitly turned on.
7212 * The default TERM variable to use for units connected to a
7213 terminal, when no other value is explicitly is set is now
7214 vt220 rather than vt102. This should be fairly safe still,
7215 but allows PgUp/PgDn work.
7217 * The /etc/crypttab option header= as known from Debian is now
7220 * "loginctl user-status" and "loginctl session-status" will
7221 now show the last 10 lines of log messages of the
7222 user/session following the status output. Similar,
7223 "machinectl status" will show the last 10 log lines
7224 associated with a virtual machine or container
7225 service. (Note that this is usually not the log messages
7226 done in the VM/container itself, but simply what the
7227 container manager logs. For nspawn this includes all console
7230 * "loginctl session-status" without further argument will now
7231 show the status of the session of the caller. Similar,
7232 "lock-session", "unlock-session", "activate",
7233 "enable-linger", "disable-linger" may now be called without
7234 session/user parameter in which case they apply to the
7235 caller's session/user.
7237 * An X11 session scriptlet is now shipped that uploads
7238 $DISPLAY and $XAUTHORITY into the environment of the systemd
7239 --user daemon if a session begins. This should improve
7240 compatibility with X11 enabled applications run as systemd
7243 * Generators are now subject to masking via /etc and /run, the
7244 same way as unit files.
7246 * networkd .network files gained support for configuring
7247 per-link IPv4/IPv6 packet forwarding as well as IPv4
7248 masquerading. This is by default turned on for veth links to
7249 containers, as registered by systemd-nspawn. This means that
7250 nspawn containers run with --network-veth will now get
7251 automatic routed access to the host's networks without any
7252 further configuration or setup, as long as networkd runs on
7255 * systemd-nspawn gained the --port= (-p) switch to expose TCP
7256 or UDP posts of a container on the host. With this in place
7257 it is possible to run containers with private veth links
7258 (--network-veth), and have their functionality exposed on
7259 the host as if their services were running directly on the
7262 * systemd-nspawn's --network-veth switch now gained a short
7263 version "-n", since with the changes above it is now truly
7264 useful out-of-the-box. The systemd-nspawn@.service has been
7265 updated to make use of it too by default.
7267 * systemd-nspawn will now maintain a per-image R/W lock, to
7268 ensure that the same image is not started more than once
7269 writable. (It's OK to run an image multiple times
7270 simultaneously in read-only mode.)
7272 * systemd-nspawn's --image= option is now capable of
7273 dissecting and booting MBR and GPT disk images that contain
7274 only a single active Linux partition. Previously it
7275 supported only GPT disk images with proper GPT type
7276 IDs. This allows running cloud images from major
7277 distributions directly with systemd-nspawn, without
7280 * In addition to collecting mouse dpi data in the udev
7281 hardware database, there's now support for collecting angle
7282 information for mouse scroll wheels. The database is
7283 supposed to guarantee similar scrolling behavior on mice
7284 that it knows about. There's also support for collecting
7285 information about Touchpad types.
7287 * udev's input_id built-in will now also collect touch screen
7288 dimension data and attach it to probed devices.
7290 * /etc/os-release gained support for a Distribution Privacy
7293 * networkd gained support for creating "ipvlan", "gretap",
7294 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
7296 * systemd-tmpfiles gained support for "a" lines for setting
7299 * systemd-nspawn will now mount /tmp in the container to
7300 tmpfs, automatically.
7302 * systemd now exposes the memory.usage_in_bytes cgroup
7303 attribute and shows it for each service in the "systemctl
7304 status" output, if available.
7306 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
7307 immediate reboot is triggered. This useful if shutdown is
7308 hung and is unable to complete, to expedite the
7309 operation. Note that this kind of reboot will still unmount
7310 all file systems, and hence should not result in fsck being
7313 * A .device unit for an optical block device will now be
7314 considered active only when a medium is in the drive. Also,
7315 mount units are now bound to their backing devices thus
7316 triggering automatic unmounting when devices become
7317 unavailable. With this in place systemd will now
7318 automatically unmount left-over mounts when a CD-ROM is
7319 ejected or an USB stick is yanked from the system.
7321 * networkd-wait-online now has support for waiting for
7322 specific interfaces only (with globbing), and for giving up
7323 after a configurable timeout.
7325 * networkd now exits when idle. It will be automatically
7326 restarted as soon as interfaces show up, are removed or
7327 change state. networkd will stay around as long as there is
7328 at least one DHCP state machine or similar around, that keep
7331 * networkd may now configure IPv6 link-local addressing in
7332 addition to IPv4 link-local addressing.
7334 * The IPv6 "token" for use in SLAAC may now be configured for
7335 each .network interface in networkd.
7337 * Routes configured with networkd may now be assigned a scope
7340 * networkd's [Match] sections now support globbing and lists
7341 of multiple space-separated matches per item.
7343 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
7344 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
7345 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
7346 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
7347 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
7348 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
7349 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
7350 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
7351 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
7352 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
7353 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
7354 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
7355 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
7356 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
7357 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
7358 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
7359 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
7360 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
7361 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
7362 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
7363 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
7364 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
7365 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
7366 Hoffmann, Zbigniew Jędrzejewski-Szmek
7368 — Berlin, 2015-02-16
7372 * When querying unit file enablement status (for example via
7373 "systemctl is-enabled"), a new state "indirect" is now known
7374 which indicates that a unit might not be enabled itself, but
7375 another unit listed in its Also= setting might be.
7377 * Similar to the various existing ConditionXYZ= settings for
7378 units, there are now matching AssertXYZ= settings. While
7379 failing conditions cause a unit to be skipped, but its job
7380 to succeed, failing assertions declared like this will cause
7381 a unit start operation and its job to fail.
7383 * hostnamed now knows a new chassis type "embedded".
7385 * systemctl gained a new "edit" command. When used on a unit
7386 file, this allows extending unit files with .d/ drop-in
7387 configuration snippets or editing the full file (after
7388 copying it from /usr/lib to /etc). This will invoke the
7389 user's editor (as configured with $EDITOR), and reload the
7390 modified configuration after editing.
7392 * "systemctl status" now shows the suggested enablement state
7393 for a unit, as declared in the (usually vendor-supplied)
7394 system preset files.
7396 * nss-myhostname will now resolve the single-label hostname
7397 "gateway" to the locally configured default IP routing
7398 gateways, ordered by their metrics. This assigns a stable
7399 name to the used gateways, regardless which ones are
7400 currently configured. Note that the name will only be
7401 resolved after all other name sources (if nss-myhostname is
7402 configured properly) and should hence not negatively impact
7403 systems that use the single-label hostname "gateway" in
7406 * systemd-inhibit now allows filtering by mode when listing
7409 * Scope and service units gained a new "Delegate" boolean
7410 property, which, when set, allows processes running inside the
7411 unit to further partition resources. This is primarily
7412 useful for systemd user instances as well as container
7415 * journald will now pick up audit messages directly from
7416 the kernel, and log them like any other log message. The
7417 audit fields are split up and fully indexed. This means that
7418 journalctl in many ways is now a (nicer!) alternative to
7419 ausearch, the traditional audit client. Note that this
7420 implements only a minimal audit client. If you want the
7421 special audit modes like reboot-on-log-overflow, please use
7422 the traditional auditd instead, which can be used in
7423 parallel to journald.
7425 * The ConditionSecurity= unit file option now understands the
7426 special string "audit" to check whether auditing is
7429 * journalctl gained two new commands --vacuum-size= and
7430 --vacuum-time= to delete old journal files until the
7431 remaining ones take up no more than the specified size on disk,
7432 or are not older than the specified time.
7434 * A new, native PPPoE library has been added to sd-network,
7435 systemd's library of light-weight networking protocols. This
7436 library will be used in a future version of networkd to
7437 enable PPPoE communication without an external pppd daemon.
7439 * The busctl tool now understands a new "capture" verb that
7440 works similar to "monitor", but writes a packet capture
7441 trace to STDOUT that can be redirected to a file which is
7442 compatible with libcap's capture file format. This can then
7443 be loaded in Wireshark and similar tools to inspect bus
7446 * The busctl tool now understands a new "tree" verb that shows
7447 the object trees of a specific service on the bus, or of all
7450 * The busctl tool now understands a new "introspect" verb that
7451 shows all interfaces and members of objects on the bus,
7452 including their signature and values. This is particularly
7453 useful to get more information about bus objects shown by
7454 the new "busctl tree" command.
7456 * The busctl tool now understands new verbs "call",
7457 "set-property" and "get-property" for invoking bus method
7458 calls, setting and getting bus object properties in a
7461 * busctl gained a new --augment-creds= argument that controls
7462 whether the tool shall augment credential information it
7463 gets from the bus with data from /proc, in a possibly
7466 * nspawn's --link-journal= switch gained two new values
7467 "try-guest" and "try-host" that work like "guest" and
7468 "host", but do not fail if the host has no persistent
7469 journalling enabled. -j is now equivalent to
7470 --link-journal=try-guest.
7472 * macvlan network devices created by nspawn will now have
7473 stable MAC addresses.
7475 * A new SmackProcessLabel= unit setting has been added, which
7476 controls the SMACK security label processes forked off by
7477 the respective unit shall use.
7479 * If compiled with --enable-xkbcommon, systemd-localed will
7480 verify x11 keymap settings by compiling the given keymap. It
7481 will spew out warnings if the compilation fails. This
7482 requires libxkbcommon to be installed.
7484 * When a coredump is collected, a larger number of metadata
7485 fields is now collected and included in the journal records
7486 created for it. More specifically, control group membership,
7487 environment variables, memory maps, working directory,
7488 chroot directory, /proc/$PID/status, and a list of open file
7489 descriptors is now stored in the log entry.
7491 * The udev hwdb now contains DPI information for mice. For
7494 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
7496 * All systemd programs that read standalone configuration
7497 files in /etc now also support a corresponding series of
7498 .conf.d configuration directories in /etc/, /run/,
7499 /usr/local/lib/, /usr/lib/, and (if configured with
7500 --enable-split-usr) /lib/. In particular, the following
7501 configuration files now have corresponding configuration
7502 directories: system.conf user.conf, logind.conf,
7503 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
7504 resolved.conf, timesyncd.conf, journal-remote.conf, and
7505 journal-upload.conf. Note that distributions should use the
7506 configuration directories in /usr/lib/; the directories in
7507 /etc/ are reserved for the system administrator.
7509 * systemd-rfkill will no longer take the rfkill device name
7510 into account when storing rfkill state on disk, as the name
7511 might be dynamically assigned and not stable. Instead, the
7512 ID_PATH udev variable combined with the rfkill type (wlan,
7513 bluetooth, ...) is used.
7515 * A new service systemd-machine-id-commit.service has been
7516 added. When used on systems where /etc is read-only during
7517 boot, and /etc/machine-id is not initialized (but an empty
7518 file), this service will copy the temporary machine ID
7519 created as replacement into /etc after the system is fully
7520 booted up. This is useful for systems that are freshly
7521 installed with a non-initialized machine ID, but should get
7522 a fixed machine ID for subsequent boots.
7524 * networkd's .netdev files now provide a large set of
7525 configuration parameters for VXLAN devices. Similarly, the
7526 bridge port cost parameter is now configurable in .network
7527 files. There's also new support for configuring IP source
7528 routing. networkd .link files gained support for a new
7529 OriginalName= match that is useful to match against the
7530 original interface name the kernel assigned. .network files
7531 may include MTU= and MACAddress= fields for altering the MTU
7532 and MAC address while being connected to a specific network
7535 * The LUKS logic gained supported for configuring
7536 UUID-specific key files. There's also new support for naming
7537 LUKS device from the kernel command line, using the new
7538 luks.name= argument.
7540 * Timer units may now be transiently created via the bus API
7541 (this was previously already available for scope and service
7542 units). In addition it is now possible to create multiple
7543 transient units at the same time with a single bus call. The
7544 "systemd-run" tool has been updated to make use of this for
7545 running commands on a specified time, in at(1)-style.
7547 * tmpfiles gained support for "t" lines, for assigning
7548 extended attributes to files. Among other uses this may be
7549 used to assign SMACK labels to files.
7551 Contributions from: Alin Rauta, Alison Chaiken, Andrej
7552 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
7553 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
7554 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
7555 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
7556 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
7557 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
7558 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
7559 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
7560 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
7561 Michael Biebl, Michael Chapman, Michael Marineau, Michal
7562 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
7563 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
7564 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
7565 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
7566 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7567 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
7568 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
7570 — Berlin, 2014-12-10
7574 * journalctl gained the new options -t/--identifier= to match
7575 on the syslog identifier (aka "tag"), as well as --utc to
7576 show log timestamps in the UTC timezone. journalctl now also
7577 accepts -n/--lines=all to disable line capping in a pager.
7579 * journalctl gained a new switch, --flush, that synchronously
7580 flushes logs from /run/log/journal to /var/log/journal if
7581 persistent storage is enabled. systemd-journal-flush.service
7582 now waits until the operation is complete.
7584 * Services can notify the manager before they start a reload
7585 (by sending RELOADING=1) or shutdown (by sending
7586 STOPPING=1). This allows the manager to track and show the
7587 internal state of daemons and closes a race condition when
7588 the process is still running but has closed its D-Bus
7591 * Services with Type=oneshot do not have to have any ExecStart
7594 * User units are now loaded also from
7595 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
7596 /run/systemd/user directory that was already previously
7597 supported, but is under the control of the user.
7599 * Job timeouts (i.e. timeouts on the time a job that is
7600 queued stays in the run queue) can now optionally result in
7601 immediate reboot or power-off actions (JobTimeoutAction= and
7602 JobTimeoutRebootArgument=). This is useful on ".target"
7603 units, to limit the maximum time a target remains
7604 undispatched in the run queue, and to trigger an emergency
7605 operation in such a case. This is now used by default to
7606 turn off the system if boot-up (as defined by everything in
7607 basic.target) hangs and does not complete for at least
7608 15min. Also, if power-off or reboot hang for at least 30min
7609 an immediate power-off/reboot operation is triggered. This
7610 functionality is particularly useful to increase reliability
7611 on embedded devices, but also on laptops which might
7612 accidentally get powered on when carried in a backpack and
7613 whose boot stays stuck in a hard disk encryption passphrase
7616 * systemd-logind can be configured to also handle lid switch
7617 events even when the machine is docked or multiple displays
7618 are attached (HandleLidSwitchDocked= option).
7620 * A helper binary and a service have been added which can be
7621 used to resume from hibernation in the initramfs. A
7622 generator will parse the resume= option on the kernel
7623 command line to trigger resume.
7625 * A user console daemon systemd-consoled has been
7626 added. Currently, it is a preview, and will so far open a
7627 single terminal on each session of the user marked as
7628 Desktop=systemd-console.
7630 * Route metrics can be specified for DHCP routes added by
7633 * The SELinux context of socket-activated services can be set
7634 from the information provided by the networking stack
7635 (SELinuxContextFromNet= option).
7637 * Userspace firmware loading support has been removed and
7638 the minimum supported kernel version is thus bumped to 3.7.
7640 * Timeout for udev workers has been increased from 1 to 3
7641 minutes, but a warning will be printed after 1 minute to
7642 help diagnose kernel modules that take a long time to load.
7644 * Udev rules can now remove tags on devices with TAG-="foobar".
7646 * systemd's readahead implementation has been removed. In many
7647 circumstances it didn't give expected benefits even for
7648 rotational disk drives and was becoming less relevant in the
7649 age of SSDs. As none of the developers has been using
7650 rotating media anymore, and nobody stepped up to actively
7651 maintain this component of systemd it has now been removed.
7653 * Swap units can use Options= to specify discard options.
7654 Discard options specified for swaps in /etc/fstab are now
7657 * Docker containers are now detected as a separate type of
7660 * The Password Agent protocol gained support for queries where
7661 the user input is shown, useful e.g. for user names.
7662 systemd-ask-password gained a new --echo option to turn that
7665 * The default sysctl.d/ snippets will now set:
7667 net.core.default_qdisc = fq_codel
7669 This selects Fair Queuing Controlled Delay as the default
7670 queuing discipline for network interfaces. fq_codel helps
7671 fight the network bufferbloat problem. It is believed to be
7672 a good default with no tuning required for most workloads.
7673 Downstream distributions may override this choice. On 10Gbit
7674 servers that do not do forwarding, "fq" may perform better.
7675 Systems without a good clocksource should use "pfifo_fast".
7677 * If kdbus is enabled during build a new option BusPolicy= is
7678 available for service units, that allows locking all service
7679 processes into a stricter bus policy, in order to limit
7680 access to various bus services, or even hide most of them
7681 from the service's view entirely.
7683 * networkctl will now show the .network and .link file
7684 networkd has applied to a specific interface.
7686 * sd-login gained a new API call sd_session_get_desktop() to
7687 query which desktop environment has been selected for a
7690 * UNIX utmp support is now compile-time optional to support
7691 legacy-free systems.
7693 * systemctl gained two new commands "add-wants" and
7694 "add-requires" for pulling in units from specific targets
7697 * If the word "rescue" is specified on the kernel command line
7698 the system will now boot into rescue mode (aka
7699 rescue.target), which was previously available only by
7700 specifying "1" or "systemd.unit=rescue.target" on the kernel
7701 command line. This new kernel command line option nicely
7702 mirrors the already existing "emergency" kernel command line
7705 * New kernel command line options mount.usr=, mount.usrflags=,
7706 mount.usrfstype= have been added that match root=, rootflags=,
7707 rootfstype= but allow mounting a specific file system to
7710 * The $NOTIFY_SOCKET is now also passed to control processes of
7711 services, not only the main process.
7713 * This version reenables support for fsck's -l switch. This
7714 means at least version v2.25 of util-linux is required for
7715 operation, otherwise dead-locks on device nodes may
7716 occur. Again: you need to update util-linux to at least
7717 v2.25 when updating systemd to v217.
7719 * The "multi-seat-x" tool has been removed from systemd, as
7720 its functionality has been integrated into X servers 1.16,
7721 and the tool is hence redundant. It is recommended to update
7722 display managers invoking this tool to simply invoke X
7723 directly from now on, again.
7725 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
7726 message flag has been added for all of systemd's polkit
7727 authenticated method calls has been added. In particular this
7728 now allows optional interactive authorization via polkit for
7729 many of PID1's privileged operations such as unit file
7730 enabling and disabling.
7732 * "udevadm hwdb --update" learnt a new switch "--usr" for
7733 placing the rebuilt hardware database in /usr instead of
7734 /etc. When used only hardware database entries stored in
7735 /usr will be used, and any user database entries in /etc are
7736 ignored. This functionality is useful for vendors to ship a
7737 pre-built database on systems where local configuration is
7738 unnecessary or unlikely.
7740 * Calendar time specifications in .timer units now also
7741 understand the strings "semi-annually", "quarterly" and
7742 "minutely" as shortcuts (in addition to the preexisting
7743 "annually", "hourly", ...).
7745 * systemd-tmpfiles will now correctly create files in /dev
7746 at boot which are marked for creation only at boot. It is
7747 recommended to always create static device nodes with 'c!'
7748 and 'b!', so that they are created only at boot and not
7749 overwritten at runtime.
7751 * When the watchdog logic is used for a service (WatchdogSec=)
7752 and the watchdog timeout is hit the service will now be
7753 terminated with SIGABRT (instead of just SIGTERM), in order
7754 to make sure a proper coredump and backtrace is
7755 generated. This ensures that hanging services will result in
7756 similar coredump/backtrace behaviour as services that hit a
7759 Contributions from: Andreas Henriksson, Andrei Borzenkov,
7760 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
7761 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
7762 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
7763 Herrmann, David Sommerseth, David Strauss, Emil Renner
7764 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
7765 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
7766 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
7767 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
7768 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
7769 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
7770 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
7771 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
7772 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
7773 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
7774 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
7775 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
7776 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
7777 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
7778 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
7779 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
7782 — Berlin, 2014-10-28
7786 * timedated no longer reads NTP implementation unit names from
7787 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
7788 implementations should add a
7790 Conflicts=systemd-timesyncd.service
7792 to their unit files to take over and replace systemd's NTP
7793 default functionality.
7795 * systemd-sysusers gained a new line type "r" for configuring
7796 which UID/GID ranges to allocate system users/groups
7797 from. Lines of type "u" may now add an additional column
7798 that specifies the home directory for the system user to be
7799 created. Also, systemd-sysusers may now optionally read user
7800 information from STDIN instead of a file. This is useful for
7801 invoking it from RPM preinst scriptlets that need to create
7802 users before the first RPM file is installed since these
7803 files might need to be owned by them. A new
7804 %sysusers_create_inline RPM macro has been introduced to do
7805 just that. systemd-sysusers now updates the shadow files as
7806 well as the user/group databases, which should enhance
7807 compatibility with certain tools like grpck.
7809 * A number of bus APIs of PID 1 now optionally consult polkit to
7810 permit access for otherwise unprivileged clients under certain
7811 conditions. Note that this currently doesn't support
7812 interactive authentication yet, but this is expected to be
7813 added eventually, too.
7815 * /etc/machine-info now has new fields for configuring the
7816 deployment environment of the machine, as well as the
7817 location of the machine. hostnamectl has been updated with
7818 new command to update these fields.
7820 * systemd-timesyncd has been updated to automatically acquire
7821 NTP server information from systemd-networkd, which might
7822 have been discovered via DHCP.
7824 * systemd-resolved now includes a caching DNS stub resolver
7825 and a complete LLMNR name resolution implementation. A new
7826 NSS module "nss-resolve" has been added which can be used
7827 instead of glibc's own "nss-dns" to resolve hostnames via
7828 systemd-resolved. Hostnames, addresses and arbitrary RRs may
7829 be resolved via systemd-resolved D-Bus APIs. In contrast to
7830 the glibc internal resolver systemd-resolved is aware of
7831 multi-homed system, and keeps DNS server and caches separate
7832 and per-interface. Queries are sent simultaneously on all
7833 interfaces that have DNS servers configured, in order to
7834 properly handle VPNs and local LANs which might resolve
7835 separate sets of domain names. systemd-resolved may acquire
7836 DNS server information from systemd-networkd automatically,
7837 which in turn might have discovered them via DHCP. A tool
7838 "systemd-resolve-host" has been added that may be used to
7839 query the DNS logic in resolved. systemd-resolved implements
7840 IDNA and automatically uses IDNA or UTF-8 encoding depending
7841 on whether classic DNS or LLMNR is used as transport. In the
7842 next releases we intend to add a DNSSEC and mDNS/DNS-SD
7843 implementation to systemd-resolved.
7845 * A new NSS module nss-mymachines has been added, that
7846 automatically resolves the names of all local registered
7847 containers to their respective IP addresses.
7849 * A new client tool "networkctl" for systemd-networkd has been
7850 added. It currently is entirely passive and will query
7851 networking configuration from udev, rtnetlink and networkd,
7852 and present it to the user in a very friendly
7853 way. Eventually, we hope to extend it to become a full
7854 control utility for networkd.
7856 * .socket units gained a new DeferAcceptSec= setting that
7857 controls the kernels' TCP_DEFER_ACCEPT sockopt for
7858 TCP. Similarly, support for controlling TCP keep-alive
7859 settings has been added (KeepAliveTimeSec=,
7860 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
7861 turning off Nagle's algorithm on TCP has been added
7864 * logind learned a new session type "web", for use in projects
7865 like Cockpit which register web clients as PAM sessions.
7867 * timer units with at least one OnCalendar= setting will now
7868 be started only after time-sync.target has been
7869 reached. This way they will not elapse before the system
7870 clock has been corrected by a local NTP client or
7871 similar. This is particular useful on RTC-less embedded
7872 machines, that come up with an invalid system clock.
7874 * systemd-nspawn's --network-veth= switch should now result in
7875 stable MAC addresses for both the outer and the inner side
7878 * systemd-nspawn gained a new --volatile= switch for running
7879 container instances with /etc or /var unpopulated.
7881 * The kdbus client code has been updated to use the new Linux
7882 3.17 memfd subsystem instead of the old kdbus-specific one.
7884 * systemd-networkd's DHCP client and server now support
7885 FORCERENEW. There are also new configuration options to
7886 configure the vendor client identifier and broadcast mode
7889 * systemd will no longer inform the kernel about the current
7890 timezone, as this is necessarily incorrect and racy as the
7891 kernel has no understanding of DST and similar
7892 concepts. This hence means FAT timestamps will be always
7893 considered UTC, similar to what Android is already
7894 doing. Also, when the RTC is configured to the local time
7895 (rather than UTC) systemd will never synchronize back to it,
7896 as this might confuse Windows at a later boot.
7898 * systemd-analyze gained a new command "verify" for offline
7899 validation of unit files.
7901 * systemd-networkd gained support for a couple of additional
7902 settings for bonding networking setups. Also, the metric for
7903 statically configured routes may now be configured. For
7904 network interfaces where this is appropriate the peer IP
7905 address may now be configured.
7907 * systemd-networkd's DHCP client will no longer request
7908 broadcasting by default, as this tripped up some networks.
7909 For hardware where broadcast is required the feature should
7910 be switched back on using RequestBroadcast=yes.
7912 * systemd-networkd will now set up IPv4LL addresses (when
7913 enabled) even if DHCP is configured successfully.
7915 * udev will now default to respect network device names given
7916 by the kernel when the kernel indicates that these are
7917 predictable. This behavior can be tweaked by changing
7918 NamePolicy= in the relevant .link file.
7920 * A new library systemd-terminal has been added that
7921 implements full TTY stream parsing and rendering. This
7922 library is supposed to be used later on for implementing a
7923 full userspace VT subsystem, replacing the current kernel
7926 * A new tool systemd-journal-upload has been added to push
7927 journal data to a remote system running
7928 systemd-journal-remote.
7930 * journald will no longer forward all local data to another
7931 running syslog daemon. This change has been made because
7932 rsyslog (which appears to be the most commonly used syslog
7933 implementation these days) no longer makes use of this, and
7934 instead pulls the data out of the journal on its own. Since
7935 forwarding the messages to a non-existent syslog server is
7936 more expensive than we assumed we have now turned this
7937 off. If you run a syslog server that is not a recent rsyslog
7938 version, you have to turn this option on again
7939 (ForwardToSyslog= in journald.conf).
7941 * journald now optionally supports the LZ4 compressor for
7942 larger journal fields. This compressor should perform much
7943 better than XZ which was the previous default.
7945 * machinectl now shows the IP addresses of local containers,
7946 if it knows them, plus the interface name of the container.
7948 * A new tool "systemd-escape" has been added that makes it
7949 easy to escape strings to build unit names and similar.
7951 * sd_notify() messages may now include a new ERRNO= field
7952 which is parsed and collected by systemd and shown among the
7953 "systemctl status" output for a service.
7955 * A new component "systemd-firstboot" has been added that
7956 queries the most basic systemd information (timezone,
7957 hostname, root password) interactively on first
7958 boot. Alternatively it may also be used to provision these
7959 things offline on OS images installed into directories.
7961 * The default sysctl.d/ snippets will now set
7963 net.ipv4.conf.default.promote_secondaries=1
7965 This has the benefit of no flushing secondary IP addresses
7966 when primary addresses are removed.
7968 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
7969 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
7970 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
7971 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
7972 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
7973 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
7974 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
7975 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
7976 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
7977 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
7978 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
7979 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
7980 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
7981 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
7982 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
7984 — Berlin, 2014-08-19
7988 * A new tool systemd-sysusers has been added. This tool
7989 creates system users and groups in /etc/passwd and
7990 /etc/group, based on static declarative system user/group
7991 definitions in /usr/lib/sysusers.d/. This is useful to
7992 enable factory resets and volatile systems that boot up with
7993 an empty /etc directory, and thus need system users and
7994 groups created during early boot. systemd now also ships
7995 with two default sysusers.d/ files for the most basic
7996 users and groups systemd and the core operating system
7999 * A new tmpfiles snippet has been added that rebuilds the
8000 essential files in /etc on boot, should they be missing.
8002 * A directive for ensuring automatic clean-up of
8003 /var/cache/man/ has been removed from the default
8004 configuration. This line should now be shipped by the man
8005 implementation. The necessary change has been made to the
8006 man-db implementation. Note that you need to update your man
8007 implementation to one that ships this line, otherwise no
8008 automatic clean-up of /var/cache/man will take place.
8010 * A new condition ConditionNeedsUpdate= has been added that
8011 may conditionalize services to only run when /etc or /var
8012 are "older" than the vendor operating system resources in
8013 /usr. This is useful for reconstructing or updating /etc
8014 after an offline update of /usr or a factory reset, on the
8015 next reboot. Services that want to run once after such an
8016 update or reset should use this condition and order
8017 themselves before the new systemd-update-done.service, which
8018 will mark the two directories as fully updated. A number of
8019 service files have been added making use of this, to rebuild
8020 the udev hardware database, the journald message catalog and
8021 dynamic loader cache (ldconfig). The systemd-sysusers tool
8022 described above also makes use of this now. With this in
8023 place it is now possible to start up a minimal operating
8024 system with /etc empty cleanly. For more information on the
8025 concepts involved see this recent blog story:
8027 http://0pointer.de/blog/projects/stateless.html
8029 * A new system group "input" has been introduced, and all
8030 input device nodes get this group assigned. This is useful
8031 for system-level software to get access to input devices. It
8032 complements what is already done for "audio" and "video".
8034 * systemd-networkd learnt minimal DHCPv4 server support in
8035 addition to the existing DHCPv4 client support. It also
8036 learnt DHCPv6 client and IPv6 Router Solicitation client
8037 support. The DHCPv4 client gained support for static routes
8038 passed in from the server. Note that the [DHCPv4] section
8039 known in older systemd-networkd versions has been renamed to
8040 [DHCP] and is now also used by the DHCPv6 client. Existing
8041 .network files using settings of this section should be
8042 updated, though compatibility is maintained. Optionally, the
8043 client hostname may now be sent to the DHCP server.
8045 * networkd gained support for vxlan virtual networks as well
8046 as tun/tap and dummy devices.
8048 * networkd gained support for automatic allocation of address
8049 ranges for interfaces from a system-wide pool of
8050 addresses. This is useful for dynamically managing a large
8051 number of interfaces with a single network configuration
8052 file. In particular this is useful to easily assign
8053 appropriate IP addresses to the veth links of a large number
8054 of nspawn instances.
8056 * RPM macros for processing sysusers, sysctl and binfmt
8057 drop-in snippets at package installation time have been
8060 * The /etc/os-release file should now be placed in
8061 /usr/lib/os-release. The old location is automatically
8062 created as symlink. /usr/lib is the more appropriate
8063 location of this file, since it shall actually describe the
8064 vendor operating system shipped in /usr, and not the
8065 configuration stored in /etc.
8067 * .mount units gained a new boolean SloppyOptions= setting
8068 that maps to mount(8)'s -s option which enables permissive
8069 parsing of unknown mount options.
8071 * tmpfiles learnt a new "L+" directive which creates a symlink
8072 but (unlike "L") deletes a pre-existing file first, should
8073 it already exist and not already be the correct
8074 symlink. Similarly, "b+", "c+" and "p+" directives have been
8075 added as well, which create block and character devices, as
8076 well as fifos in the filesystem, possibly removing any
8077 pre-existing files of different types.
8079 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
8080 'argument' field (which so far specified the source to
8081 symlink/copy the files from) is now optional. If omitted the
8082 same file os copied from /usr/share/factory/ suffixed by the
8083 full destination path. This is useful for populating /etc
8084 with essential files, by copying them from vendor defaults
8085 shipped in /usr/share/factory/etc.
8087 * A new command "systemctl preset-all" has been added that
8088 applies the service preset settings to all installed unit
8089 files. A new switch --preset-mode= has been added that
8090 controls whether only enable or only disable operations
8093 * A new command "systemctl is-system-running" has been added
8094 that allows checking the overall state of the system, for
8095 example whether it is fully up and running.
8097 * When the system boots up with an empty /etc, the equivalent
8098 to "systemctl preset-all" is executed during early boot, to
8099 make sure all default services are enabled after a factory
8102 * systemd now contains a minimal preset file that enables the
8103 most basic services systemd ships by default.
8105 * Unit files' [Install] section gained a new DefaultInstance=
8106 field for defining the default instance to create if a
8107 template unit is enabled with no instance specified.
8109 * A new passive target cryptsetup-pre.target has been added
8110 that may be used by services that need to make they run and
8111 finish before the first LUKS cryptographic device is set up.
8113 * The /dev/loop-control and /dev/btrfs-control device nodes
8114 are now owned by the "disk" group by default, opening up
8115 access to this group.
8117 * systemd-coredump will now automatically generate a
8118 stack trace of all core dumps taking place on the system,
8119 based on elfutils' libdw library. This stack trace is logged
8122 * systemd-coredump may now optionally store coredumps directly
8123 on disk (in /var/lib/systemd/coredump, possibly compressed),
8124 instead of storing them unconditionally in the journal. This
8125 mode is the new default. A new configuration file
8126 /etc/systemd/coredump.conf has been added to configure this
8127 and other parameters of systemd-coredump.
8129 * coredumpctl gained a new "info" verb to show details about a
8130 specific coredump. A new switch "-1" has also been added
8131 that makes sure to only show information about the most
8132 recent entry instead of all entries. Also, as the tool is
8133 generally useful now the "systemd-" prefix of the binary
8134 name has been removed. Distributions that want to maintain
8135 compatibility with the old name should add a symlink from
8136 the old name to the new name.
8138 * journald's SplitMode= now defaults to "uid". This makes sure
8139 that unprivileged users can access their own coredumps with
8140 coredumpctl without restrictions.
8142 * New kernel command line options "systemd.wants=" (for
8143 pulling an additional unit during boot), "systemd.mask="
8144 (for masking a specific unit for the boot), and
8145 "systemd.debug-shell" (for enabling the debug shell on tty9)
8146 have been added. This is implemented in the new generator
8147 "systemd-debug-generator".
8149 * systemd-nspawn will now by default filter a couple of
8150 syscalls for containers, among them those required for
8151 kernel module loading, direct x86 IO port access, swap
8152 management, and kexec. Most importantly though
8153 open_by_handle_at() is now prohibited for containers,
8154 closing a hole similar to a recently discussed vulnerability
8155 in docker regarding access to files on file hierarchies the
8156 container should normally not have access to. Note that, for
8157 nspawn, we generally make no security claims anyway (and
8158 this is explicitly documented in the man page), so this is
8159 just a fix for one of the most obvious problems.
8161 * A new man page file-hierarchy(7) has been added that
8162 contains a minimized, modernized version of the file system
8163 layout systemd expects, similar in style to the FHS
8164 specification or hier(5). A new tool systemd-path(1) has
8165 been added to query many of these paths for the local
8168 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
8169 longer done. Since the directory now has a per-user size
8170 limit, and is cleaned on logout this appears unnecessary,
8171 in particular since this now brings the lifecycle of this
8172 directory closer in line with how IPC objects are handled.
8174 * systemd.pc now exports a number of additional directories,
8175 including $libdir (which is useful to identify the library
8176 path for the primary architecture of the system), and a
8177 couple of drop-in directories.
8179 * udev's predictable network interface names now use the dev_port
8180 sysfs attribute, introduced in linux 3.15 instead of dev_id to
8181 distinguish between ports of the same PCI function. dev_id should
8182 only be used for ports using the same HW address, hence the need
8185 * machined has been updated to export the OS version of a
8186 container (read from /etc/os-release and
8187 /usr/lib/os-release) on the bus. This is now shown in
8188 "machinectl status" for a machine.
8190 * A new service setting RestartForceExitStatus= has been
8191 added. If configured to a set of exit signals or process
8192 return values, the service will be restarted when the main
8193 daemon process exits with any of them, regardless of the
8196 * systemctl's -H switch for connecting to remote systemd
8197 machines has been extended so that it may be used to
8198 directly connect to a specific container on the
8199 host. "systemctl -H root@foobar:waldi" will now connect as
8200 user "root" to host "foobar", and then proceed directly to
8201 the container named "waldi". Note that currently you have to
8202 authenticate as user "root" for this to work, as entering
8203 containers is a privileged operation.
8205 Contributions from: Andreas Henriksson, Benjamin Steinwender,
8206 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
8207 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
8208 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
8209 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
8210 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
8211 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
8212 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
8213 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
8214 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
8215 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
8216 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
8218 — Berlin, 2014-07-03
8222 * As an experimental feature, udev now tries to lock the
8223 disk device node (flock(LOCK_SH|LOCK_NB)) while it
8224 executes events for the disk or any of its partitions.
8225 Applications like partitioning programs can lock the
8226 disk device node (flock(LOCK_EX)) and claim temporary
8227 device ownership that way; udev will entirely skip all event
8228 handling for this disk and its partitions. If the disk
8229 was opened for writing, the close will trigger a partition
8230 table rescan in udev's "watch" facility, and if needed
8231 synthesize "change" events for the disk and all its partitions.
8232 This is now unconditionally enabled, and if it turns out to
8233 cause major problems, we might turn it on only for specific
8234 devices, or might need to disable it entirely. Device Mapper
8235 devices are excluded from this logic.
8237 * We temporarily dropped the "-l" switch for fsck invocations,
8238 since they collide with the flock() logic above. util-linux
8239 upstream has been changed already to avoid this conflict,
8240 and we will re-add "-l" as soon as util-linux with this
8241 change has been released.
8243 * The dependency on libattr has been removed. Since a long
8244 time, the extended attribute calls have moved to glibc, and
8245 libattr is thus unnecessary.
8247 * Virtualization detection works without privileges now. This
8248 means the systemd-detect-virt binary no longer requires
8249 CAP_SYS_PTRACE file capabilities, and our daemons can run
8250 with fewer privileges.
8252 * systemd-networkd now runs under its own "systemd-network"
8253 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
8254 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
8255 loses the ability to write to files owned by root this way.
8257 * Similarly, systemd-resolved now runs under its own
8258 "systemd-resolve" user with no capabilities remaining.
8260 * Similarly, systemd-bus-proxyd now runs under its own
8261 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
8263 * systemd-networkd gained support for setting up "veth"
8264 virtual Ethernet devices for container connectivity, as well
8265 as GRE and VTI tunnels.
8267 * systemd-networkd will no longer automatically attempt to
8268 manually load kernel modules necessary for certain tunnel
8269 transports. Instead, it is assumed the kernel loads them
8270 automatically when required. This only works correctly on
8271 very new kernels. On older kernels, please consider adding
8272 the kernel modules to /etc/modules-load.d/ as a work-around.
8274 * The resolv.conf file systemd-resolved generates has been
8275 moved to /run/systemd/resolve/. If you have a symlink from
8276 /etc/resolv.conf, it might be necessary to correct it.
8278 * Two new service settings, ProtectHome= and ProtectSystem=,
8279 have been added. When enabled, they will make the user data
8280 (such as /home) inaccessible or read-only and the system
8281 (such as /usr) read-only, for specific services. This allows
8282 very light-weight per-service sandboxing to avoid
8283 modifications of user data or system files from
8284 services. These two new switches have been enabled for all
8285 of systemd's long-running services, where appropriate.
8287 * Socket units gained new SocketUser= and SocketGroup=
8288 settings to set the owner user and group of AF_UNIX sockets
8289 and FIFOs in the file system.
8291 * Socket units gained a new RemoveOnStop= setting. If enabled,
8292 all FIFOS and sockets in the file system will be removed
8293 when the specific socket unit is stopped.
8295 * Socket units gained a new Symlinks= setting. It takes a list
8296 of symlinks to create to file system sockets or FIFOs
8297 created by the specific Unix sockets. This is useful to
8298 manage symlinks to socket nodes with the same lifecycle as
8301 * The /dev/log socket and /dev/initctl FIFO have been moved to
8302 /run, and have been replaced by symlinks. This allows
8303 connecting to these facilities even if PrivateDevices=yes is
8304 used for a service (which makes /dev/log itself unavailable,
8305 but /run is left). This also has the benefit of ensuring
8306 that /dev only contains device nodes, directories and
8307 symlinks, and nothing else.
8309 * sd-daemon gained two new calls sd_pid_notify() and
8310 sd_pid_notifyf(). They are similar to sd_notify() and
8311 sd_notifyf(), but allow overriding of the source PID of
8312 notification messages if permissions permit this. This is
8313 useful to send notify messages on behalf of a different
8314 process (for example, the parent process). The
8315 systemd-notify tool has been updated to make use of this
8316 when sending messages (so that notification messages now
8317 originate from the shell script invoking systemd-notify and
8318 not the systemd-notify process itself. This should minimize
8319 a race where systemd fails to associate notification
8320 messages to services when the originating process already
8323 * A new "on-abnormal" setting for Restart= has been added. If
8324 set, it will result in automatic restarts on all "abnormal"
8325 reasons for a process to exit, which includes unclean
8326 signals, core dumps, timeouts and watchdog timeouts, but
8327 does not include clean and unclean exit codes or clean
8328 signals. Restart=on-abnormal is an alternative for
8329 Restart=on-failure for services that shall be able to
8330 terminate and avoid restarts on certain errors, by
8331 indicating so with an unclean exit code. Restart=on-failure
8332 or Restart=on-abnormal is now the recommended setting for
8333 all long-running services.
8335 * If the InaccessibleDirectories= service setting points to a
8336 mount point (or if there are any submounts contained within
8337 it), it is now attempted to completely unmount it, to make
8338 the file systems truly unavailable for the respective
8341 * The ReadOnlyDirectories= service setting and
8342 systemd-nspawn's --read-only parameter are now recursively
8343 applied to all submounts, too.
8345 * Mount units may now be created transiently via the bus APIs.
8347 * The support for SysV and LSB init scripts has been removed
8348 from the systemd daemon itself. Instead, it is now
8349 implemented as a generator that creates native systemd units
8350 from these scripts when needed. This enables us to remove a
8351 substantial amount of legacy code from PID 1, following the
8352 fact that many distributions only ship a very small number
8353 of LSB/SysV init scripts nowadays.
8355 * Privileged Xen (dom0) domains are not considered
8356 virtualization anymore by the virtualization detection
8357 logic. After all, they generally have unrestricted access to
8358 the hardware and usually are used to manage the unprivileged
8361 * systemd-tmpfiles gained a new "C" line type, for copying
8362 files or entire directories.
8364 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
8365 lines. So far, they have been non-globbing versions of the
8366 latter, and have thus been redundant. In future, it is
8367 recommended to only use "z". "m" has hence been removed
8368 from the documentation, even though it stays supported.
8370 * A tmpfiles snippet to recreate the most basic structure in
8371 /var has been added. This is enough to create the /var/run →
8372 /run symlink and create a couple of structural
8373 directories. This allows systems to boot up with an empty or
8374 volatile /var. Of course, while with this change, the core OS
8375 now is capable with dealing with a volatile /var, not all
8376 user services are ready for it. However, we hope that sooner
8377 or later, many service daemons will be changed upstream so
8378 that they are able to automatically create their necessary
8379 directories in /var at boot, should they be missing. This is
8380 the first step to allow state-less systems that only require
8381 the vendor image for /usr to boot.
8383 * systemd-nspawn has gained a new --tmpfs= switch to mount an
8384 empty tmpfs instance to a specific directory. This is
8385 particularly useful for making use of the automatic
8386 reconstruction of /var (see above), by passing --tmpfs=/var.
8388 * Access modes specified in tmpfiles snippets may now be
8389 prefixed with "~", which indicates that they shall be masked
8390 by whether the existing file or directory is currently
8391 writable, readable or executable at all. Also, if specified,
8392 the sgid/suid/sticky bits will be masked for all
8395 * A new passive target unit "network-pre.target" has been
8396 added which is useful for services that shall run before any
8397 network is configured, for example firewall scripts.
8399 * The "floppy" group that previously owned the /dev/fd*
8400 devices is no longer used. The "disk" group is now used
8401 instead. Distributions should probably deprecate usage of
8404 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
8405 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
8406 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
8407 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
8408 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
8409 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
8412 — Berlin, 2014-06-11
8416 * A new "systemd-timesyncd" daemon has been added for
8417 synchronizing the system clock across the network. It
8418 implements an SNTP client. In contrast to NTP
8419 implementations such as chrony or the NTP reference server,
8420 this only implements a client side, and does not bother with
8421 the full NTP complexity, focusing only on querying time from
8422 one remote server and synchronizing the local clock to
8423 it. Unless you intend to serve NTP to networked clients or
8424 want to connect to local hardware clocks, this simple NTP
8425 client should be more than appropriate for most
8426 installations. The daemon runs with minimal privileges, and
8427 has been hooked up with networkd to only operate when
8428 network connectivity is available. The daemon saves the
8429 current clock to disk every time a new NTP sync has been
8430 acquired, and uses this to possibly correct the system clock
8431 early at bootup, in order to accommodate for systems that
8432 lack an RTC such as the Raspberry Pi and embedded devices,
8433 and to make sure that time monotonically progresses on these
8434 systems, even if it is not always correct. To make use of
8435 this daemon, a new system user and group "systemd-timesync"
8436 needs to be created on installation of systemd.
8438 * The queue "seqnum" interface of libudev has been disabled, as
8439 it was generally incompatible with device namespacing as
8440 sequence numbers of devices go "missing" if the devices are
8441 part of a different namespace.
8443 * "systemctl list-timers" and "systemctl list-sockets" gained
8444 a --recursive switch for showing units of these types also
8445 for all local containers, similar in style to the already
8446 supported --recursive switch for "systemctl list-units".
8448 * A new RebootArgument= setting has been added for service
8449 units, which may be used to specify a kernel reboot argument
8450 to use when triggering reboots with StartLimitAction=.
8452 * A new FailureAction= setting has been added for service
8453 units which may be used to specify an operation to trigger
8454 when a service fails. This works similarly to
8455 StartLimitAction=, but unlike it, controls what is done
8456 immediately rather than only after several attempts to
8457 restart the service in question.
8459 * hostnamed got updated to also expose the kernel name,
8460 release, and version on the bus. This is useful for
8461 executing commands like hostnamectl with the -H switch.
8462 systemd-analyze makes use of this to properly display
8463 details when running non-locally.
8465 * The bootchart tool can now show cgroup information in the
8466 graphs it generates.
8468 * The CFS CPU quota cgroup attribute is now exposed for
8469 services. The new CPUQuota= switch has been added for this
8470 which takes a percentage value. Setting this will have the
8471 result that a service may never get more CPU time than the
8472 specified percentage, even if the machine is otherwise idle.
8474 * systemd-networkd learned IPIP and SIT tunnel support.
8476 * LSB init scripts exposing a dependency on $network will now
8477 get a dependency on network-online.target rather than simply
8478 network.target. This should bring LSB handling closer to
8479 what it was on SysV systems.
8481 * A new fsck.repair= kernel option has been added to control
8482 how fsck shall deal with unclean file systems at boot.
8484 * The (.ini) configuration file parser will now silently ignore
8485 sections whose names begin with "X-". This may be used to maintain
8486 application-specific extension sections in unit files.
8488 * machined gained a new API to query the IP addresses of
8489 registered containers. "machinectl status" has been updated
8490 to show these addresses in its output.
8492 * A new call sd_uid_get_display() has been added to the
8493 sd-login APIs for querying the "primary" session of a
8494 user. The "primary" session of the user is elected from the
8495 user's sessions and generally a graphical session is
8496 preferred over a text one.
8498 * A minimal systemd-resolved daemon has been added. It
8499 currently simply acts as a companion to systemd-networkd and
8500 manages resolv.conf based on per-interface DNS
8501 configuration, possibly supplied via DHCP. In the long run
8502 we hope to extend this into a local DNSSEC enabled DNS and
8505 * The systemd-networkd-wait-online tool is now enabled by
8506 default. It will delay network-online.target until a network
8507 connection has been configured. The tool primarily integrates
8508 with networkd, but will also make a best effort to make sense
8509 of network configuration performed in some other way.
8511 * Two new service options StartupCPUShares= and
8512 StartupBlockIOWeight= have been added that work similarly to
8513 CPUShares= and BlockIOWeight= however only apply during
8514 system startup. This is useful to prioritize certain services
8515 differently during bootup than during normal runtime.
8517 * hostnamed has been changed to prefer the statically
8518 configured hostname in /etc/hostname (unless set to
8519 'localhost' or empty) over any dynamic one supplied by
8520 dhcp. With this change, the rules for picking the hostname
8521 match more closely the rules of other configuration settings
8522 where the local administrator's configuration in /etc always
8523 overrides any other settings.
8525 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
8526 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
8527 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
8528 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
8529 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
8530 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
8531 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
8532 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
8533 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
8534 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
8535 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
8536 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
8537 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
8538 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
8539 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
8540 Lindskog, WaLyong Cho, Will Woods, Zbigniew
8543 — Beijing, 2014-05-28
8547 * When restoring the screen brightness at boot, stay away from
8548 the darkest setting or from the lowest 5% of the available
8549 range, depending on which is the larger value of both. This
8550 should effectively protect the user from rebooting into a
8551 black screen, should the brightness have been set to minimum
8554 * sd-login gained a new sd_machine_get_class() call to
8555 determine the class ("vm" or "container") of a machine
8556 registered with machined.
8558 * sd-login gained new calls
8559 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
8560 to query the identity of the peer of a local AF_UNIX
8561 connection. They operate similarly to their sd_pid_get_xyz()
8564 * PID 1 will now maintain a system-wide system state engine
8565 with the states "starting", "running", "degraded",
8566 "maintenance", "stopping". These states are bound to system
8567 startup, normal runtime, runtime with at least one failed
8568 service, rescue/emergency mode and system shutdown. This
8569 state is shown in the "systemctl status" output when no unit
8570 name is passed. It is useful to determine system state, in
8571 particularly when doing so for many systems or containers at
8574 * A new command "list-machines" has been added to "systemctl"
8575 that lists all local OS containers and shows their system
8576 state (see above), if systemd runs inside of them.
8578 * systemctl gained a new "-r" switch to recursively enumerate
8579 units on all local containers, when used with the
8580 "list-unit" command (which is the default one that is
8581 executed when no parameters are specified).
8583 * The GPT automatic partition discovery logic will now honour
8584 two GPT partition flags: one may be set on a partition to
8585 cause it to be mounted read-only, and the other may be set
8586 on a partition to ignore it during automatic discovery.
8588 * Two new GPT type UUIDs have been added for automatic root
8589 partition discovery, for 32-bit and 64-bit ARM. This is not
8590 particularly useful for discovering the root directory on
8591 these architectures during bare-metal boots (since UEFI is
8592 not common there), but still very useful to allow booting of
8593 ARM disk images in nspawn with the -i option.
8595 * MAC addresses of interfaces created with nspawn's
8596 --network-interface= switch will now be generated from the
8597 machine name, and thus be stable between multiple invocations
8600 * logind will now automatically remove all IPC objects owned
8601 by a user if she or he fully logs out. This makes sure that
8602 users who are logged out cannot continue to consume IPC
8603 resources. This covers SysV memory, semaphores and message
8604 queues as well as POSIX shared memory and message
8605 queues. Traditionally, SysV and POSIX IPC had no lifecycle
8606 limits. With this functionality, that is corrected. This may
8607 be turned off by using the RemoveIPC= switch of logind.conf.
8609 * The systemd-machine-id-setup and tmpfiles tools gained a
8610 --root= switch to operate on a specific root directory,
8613 * journald can now forward logged messages to the TTYs of all
8614 logged in users ("wall"). This is the default for all
8615 emergency messages now.
8617 * A new tool systemd-journal-remote has been added to stream
8618 journal log messages across the network.
8620 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
8621 controller trees are mounted into it. Note that the
8622 directories mounted beneath it are not read-only. This is a
8623 security measure and is particularly useful because glibc
8624 actually includes a search logic to pick any tmpfs it can
8625 find to implement shm_open() if /dev/shm is not available
8626 (which it might very well be in namespaced setups).
8628 * machinectl gained a new "poweroff" command to cleanly power
8629 down a local OS container.
8631 * The PrivateDevices= unit file setting will now also drop the
8632 CAP_MKNOD capability from the capability bound set, and
8633 imply DevicePolicy=closed.
8635 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
8636 comprehensively on all long-running systemd services where
8637 this is appropriate.
8639 * systemd-udevd will now run in a disassociated mount
8640 namespace. To mount directories from udev rules, make sure to
8641 pull in mount units via SYSTEMD_WANTS properties.
8643 * The kdbus support gained support for uploading policy into
8644 the kernel. sd-bus gained support for creating "monitoring"
8645 connections that can eavesdrop into all bus communication
8646 for debugging purposes.
8648 * Timestamps may now be specified in seconds since the UNIX
8649 epoch Jan 1st, 1970 by specifying "@" followed by the value
8652 * Native tcpwrap support in systemd has been removed. tcpwrap
8653 is old code, not really maintained anymore and has serious
8654 shortcomings, and better options such as firewalls
8655 exist. For setups that require tcpwrap usage, please
8656 consider invoking your socket-activated service via tcpd,
8657 like on traditional inetd.
8659 * A new system.conf configuration option
8660 DefaultTimerAccuracySec= has been added that controls the
8661 default AccuracySec= setting of .timer units.
8663 * Timer units gained a new WakeSystem= switch. If enabled,
8664 timers configured this way will cause the system to resume
8665 from system suspend (if the system supports that, which most
8668 * Timer units gained a new Persistent= switch. If enabled,
8669 timers configured this way will save to disk when they have
8670 been last triggered. This information is then used on next
8671 reboot to possible execute overdue timer events, that
8672 could not take place because the system was powered off.
8673 This enables simple anacron-like behaviour for timer units.
8675 * systemctl's "list-timers" will now also list the time a
8676 timer unit was last triggered in addition to the next time
8677 it will be triggered.
8679 * systemd-networkd will now assign predictable IPv4LL
8680 addresses to its local interfaces.
8682 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
8683 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
8684 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
8685 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
8686 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
8687 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
8688 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
8689 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
8692 — Berlin, 2014-03-25
8696 * A new unit file setting RestrictAddressFamilies= has been
8697 added to restrict which socket address families unit
8698 processes gain access to. This takes address family names
8699 like "AF_INET" or "AF_UNIX", and is useful to minimize the
8700 attack surface of services via exotic protocol stacks. This
8701 is built on seccomp system call filters.
8703 * Two new unit file settings RuntimeDirectory= and
8704 RuntimeDirectoryMode= have been added that may be used to
8705 manage a per-daemon runtime directories below /run. This is
8706 an alternative for setting up directory permissions with
8707 tmpfiles snippets, and has the advantage that the runtime
8708 directory's lifetime is bound to the daemon runtime and that
8709 the daemon starts up with an empty directory each time. This
8710 is particularly useful when writing services that drop
8711 privileges using the User= or Group= setting.
8713 * The DeviceAllow= unit setting now supports globbing for
8714 matching against device group names.
8716 * The systemd configuration file system.conf gained new
8717 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
8718 DefaultMemoryAccounting= to globally turn on/off accounting
8719 for specific resources (cgroups) for all units. These
8720 settings may still be overridden individually in each unit
8723 * systemd-gpt-auto-generator is now able to discover /srv and
8724 root partitions in addition to /home and swap partitions. It
8725 also supports LUKS-encrypted partitions now. With this in
8726 place, automatic discovery of partitions to mount following
8727 the Discoverable Partitions Specification
8728 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
8729 is now a lot more complete. This allows booting without
8730 /etc/fstab and without root= on the kernel command line on
8731 systems prepared appropriately.
8733 * systemd-nspawn gained a new --image= switch which allows
8734 booting up disk images and Linux installations on any block
8735 device that follow the Discoverable Partitions Specification
8736 (see above). This means that installations made with
8737 appropriately updated installers may now be started and
8738 deployed using container managers, completely
8739 unmodified. (We hope that libvirt-lxc will add support for
8740 this feature soon, too.)
8742 * systemd-nspawn gained a new --network-macvlan= setting to
8743 set up a private macvlan interface for the
8744 container. Similarly, systemd-networkd gained a new
8745 Kind=macvlan setting in .netdev files.
8747 * systemd-networkd now supports configuring local addresses
8750 * A new tool systemd-network-wait-online has been added to
8751 synchronously wait for network connectivity using
8754 * The sd-bus.h bus API gained a new sd_bus_track object for
8755 tracking the lifecycle of bus peers. Note that sd-bus.h is
8756 still not a public API though (unless you specify
8757 --enable-kdbus on the configure command line, which however
8758 voids your warranty and you get no API stability guarantee).
8760 * The $XDG_RUNTIME_DIR runtime directories for each user are
8761 now individual tmpfs instances, which has the benefit of
8762 introducing separate pools for each user, with individual
8763 size limits, and thus making sure that unprivileged clients
8764 can no longer negatively impact the system or other users by
8765 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
8766 RuntimeDirectorySize= has been introduced that allows
8767 controlling the default size limit for all users. It
8768 defaults to 10% of the available physical memory. This is no
8769 replacement for quotas on tmpfs though (which the kernel
8770 still does not support), as /dev/shm and /tmp are still
8771 shared resources used by both the system and unprivileged
8774 * logind will now automatically turn off automatic suspending
8775 on laptop lid close when more than one display is
8776 connected. This was previously expected to be implemented
8777 individually in desktop environments (such as GNOME),
8778 however has been added to logind now, in order to fix a
8779 boot-time race where a desktop environment might not have
8780 been started yet and thus not been able to take an inhibitor
8781 lock at the time where logind already suspends the system
8782 due to a closed lid.
8784 * logind will now wait at least 30s after each system
8785 suspend/resume cycle, and 3min after system boot before
8786 suspending the system due to a closed laptop lid. This
8787 should give USB docking stations and similar enough time to
8788 be probed and configured after system resume and boot in
8789 order to then act as suspend blocker.
8791 * systemd-run gained a new --property= setting which allows
8792 initialization of resource control properties (and others)
8793 for the created scope or service unit. Example: "systemd-run
8794 --property=BlockIOWeight=10 updatedb" may be used to run
8795 updatedb at a low block IO scheduling weight.
8797 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
8798 now also work in --scope mode.
8800 * When systemd is compiled with kdbus support, basic support
8801 for enforced policies is now in place. (Note that enabling
8802 kdbus still voids your warranty and no API compatibility
8805 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
8806 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
8807 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
8808 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
8809 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
8810 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
8811 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
8812 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
8813 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
8814 Zbigniew Jędrzejewski-Szmek
8816 — Berlin, 2014-03-12
8820 * systemd will now relabel /dev after loading the SMACK policy
8821 according to SMACK rules.
8823 * A new unit file option AppArmorProfile= has been added to
8824 set the AppArmor profile for the processes of a unit.
8826 * A new condition check ConditionArchitecture= has been added
8827 to conditionalize units based on the system architecture, as
8828 reported by uname()'s "machine" field.
8830 * systemd-networkd now supports matching on the system
8831 virtualization, architecture, kernel command line, hostname
8834 * logind is now a lot more aggressive when suspending the
8835 machine due to a closed laptop lid. Instead of acting only
8836 on the lid close action, it will continuously watch the lid
8837 status and act on it. This is useful for laptops where the
8838 power button is on the outside of the chassis so that it can
8839 be reached without opening the lid (such as the Lenovo
8840 Yoga). On those machines, logind will now immediately
8841 re-suspend the machine if the power button has been
8842 accidentally pressed while the laptop was suspended and in a
8843 backpack or similar.
8845 * logind will now watch SW_DOCK switches and inhibit reaction
8846 to the lid switch if it is pressed. This means that logind
8847 will not suspend the machine anymore if the lid is closed
8848 and the system is docked, if the laptop supports SW_DOCK
8849 notifications via the input layer. Note that ACPI docking
8850 stations do not generate this currently. Also note that this
8851 logic is usually not fully sufficient and Desktop
8852 Environments should take a lid switch inhibitor lock when an
8853 external display is connected, as systemd will not watch
8856 * nspawn will now make use of the devices cgroup controller by
8857 default, and only permit creation of and access to the usual
8858 API device nodes like /dev/null or /dev/random, as well as
8859 access to (but not creation of) the pty devices.
8861 * We will now ship a default .network file for
8862 systemd-networkd that automatically configures DHCP for
8863 network interfaces created by nspawn's --network-veth or
8864 --network-bridge= switches.
8866 * systemd will now understand the usual M, K, G, T suffixes
8867 according to SI conventions (i.e. to the base 1000) when
8868 referring to throughput and hardware metrics. It will stay
8869 with IEC conventions (i.e. to the base 1024) for software
8870 metrics, according to what is customary according to
8871 Wikipedia. We explicitly document which base applies for
8872 each configuration option.
8874 * The DeviceAllow= setting in unit files now supports a syntax to
8875 allow-list an entire group of devices node majors at once, based on
8876 the /proc/devices listing. For example, with the string "char-pts",
8877 it is now possible to allow-list all current and future pseudo-TTYs
8880 * sd-event learned a new "post" event source. Event sources of
8881 this type are triggered by the dispatching of any event
8882 source of a type that is not "post". This is useful for
8883 implementing clean-up and check event sources that are
8884 triggered by other work being done in the program.
8886 * systemd-networkd is no longer statically enabled, but uses
8887 the usual [Install] sections so that it can be
8888 enabled/disabled using systemctl. It still is enabled by
8891 * When creating a veth interface pair with systemd-nspawn, the
8892 host side will now be prefixed with "vb-" if
8893 --network-bridge= is used, and with "ve-" if --network-veth
8894 is used. This way, it is easy to distinguish these cases on
8895 the host, for example to apply different configuration to
8896 them with systemd-networkd.
8898 * The compatibility libraries for libsystemd-journal.so,
8899 libsystem-id128.so, libsystemd-login.so and
8900 libsystemd-daemon.so do not make use of IFUNC
8901 anymore. Instead, we now build libsystemd.so multiple times
8902 under these alternative names. This means that the footprint
8903 is drastically increased, but given that these are
8904 transitional compatibility libraries, this should not matter
8905 much. This change has been made necessary to support the ARM
8906 platform for these compatibility libraries, as the ARM
8907 toolchain is not really at the same level as the toolchain
8908 for other architectures like x86 and does not support
8909 IFUNC. Please make sure to use --enable-compat-libs only
8910 during a transitional period!
8912 * The .include syntax has been deprecated and is not documented
8913 anymore. Drop-in files in .d directories should be used instead.
8915 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
8916 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
8917 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
8918 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
8919 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
8920 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
8921 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
8922 Zbigniew Jędrzejewski-Szmek
8924 — Berlin, 2014-02-24
8928 * A new component "systemd-networkd" has been added that can
8929 be used to configure local network interfaces statically or
8930 via DHCP. It is capable of bringing up bridges, VLANs, and
8931 bonding. Currently, no hook-ups for interactive network
8932 configuration are provided. Use this for your initrd,
8933 container, embedded, or server setup if you need a simple,
8934 yet powerful, network configuration solution. This
8935 configuration subsystem is quite nifty, as it allows wildcard
8936 hotplug matching in interfaces. For example, with a single
8937 configuration snippet, you can configure that all Ethernet
8938 interfaces showing up are automatically added to a bridge,
8939 or similar. It supports link-sensing and more.
8941 * A new tool "systemd-socket-proxyd" has been added which can
8942 act as a bidirectional proxy for TCP sockets. This is
8943 useful for adding socket activation support to services that
8944 do not actually support socket activation, including virtual
8945 machines and the like.
8947 * Add a new tool to save/restore rfkill state on
8950 * Save/restore state of keyboard backlights in addition to
8951 display backlights on shutdown/boot.
8953 * udev learned a new SECLABEL{} construct to label device
8954 nodes with a specific security label when they appear. For
8955 now, only SECLABEL{selinux} is supported, but the syntax is
8956 prepared for additional security frameworks.
8958 * udev gained a new scheme to configure link-level attributes
8959 from files in /etc/systemd/network/*.link. These files can
8960 match against MAC address, device path, driver name and type,
8961 and will apply attributes like the naming policy, link speed,
8962 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
8963 address assignment policy (randomized, ...).
8965 * The configuration of network interface naming rules for
8966 "permanent interface names" has changed: a new NamePolicy=
8967 setting in the [Link] section of .link files determines the
8968 priority of possible naming schemes (onboard, slot, MAC,
8969 path). The default value of this setting is determined by
8970 /usr/lib/net/links/99-default.link. Old
8971 80-net-name-slot.rules udev configuration file has been
8972 removed, so local configuration overriding this file should
8973 be adapted to override 99-default.link instead.
8975 * When the User= switch is used in a unit file, also
8976 initialize $SHELL= based on the user database entry.
8978 * systemd no longer depends on libdbus. All communication is
8979 now done with sd-bus, systemd's low-level bus library
8982 * kdbus support has been added to PID 1 itself. When kdbus is
8983 enabled, this causes PID 1 to set up the system bus and
8984 enable support for a new ".busname" unit type that
8985 encapsulates bus name activation on kdbus. It works a little
8986 bit like ".socket" units, except for bus names. A new
8987 generator has been added that converts classic dbus1 service
8988 activation files automatically into native systemd .busname
8991 * sd-bus: add a light-weight vtable implementation that allows
8992 defining objects on the bus with a simple static const
8993 vtable array of its methods, signals and properties.
8995 * systemd will not generate or install static dbus
8996 introspection data anymore to /usr/share/dbus-1/interfaces,
8997 as the precise format of these files is unclear, and
8998 nothing makes use of it.
9000 * A proxy daemon is now provided to proxy clients connecting
9001 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
9002 compatibility with classic D-Bus.
9004 * A bus driver implementation has been added that supports the
9005 classic D-Bus bus driver calls on kdbus, also for
9006 compatibility purposes.
9008 * A new API "sd-event.h" has been added that implements a
9009 minimal event loop API built around epoll. It provides a
9010 couple of features that direct epoll usage is lacking:
9011 prioritization of events, scales to large numbers of timer
9012 events, per-event timer slack (accuracy), system-wide
9013 coalescing of timer events, exit handlers, watchdog
9014 supervision support using systemd's sd_notify() API, child
9017 * A new API "sd-rntl.h" has been added that provides an API
9018 around the route netlink interface of the kernel, similar in
9019 style to "sd-bus.h".
9021 * A new API "sd-dhcp-client.h" has been added that provides a
9022 small DHCPv4 client-side implementation. This is used by
9025 * There is a new kernel command line option
9026 "systemd.restore_state=0|1". When set to "0", none of the
9027 systemd tools will restore saved runtime state to hardware
9028 devices. More specifically, the rfkill and backlight states
9031 * The FsckPassNo= compatibility option in mount/service units
9032 has been removed. The fstab generator will now add the
9033 necessary dependencies automatically, and does not require
9034 PID1's support for that anymore.
9036 * journalctl gained a new switch, --list-boots, that lists
9037 recent boots with their times and boot IDs.
9039 * The various tools like systemctl, loginctl, timedatectl,
9040 busctl, systemd-run, ... have gained a new switch "-M" to
9041 connect to a specific, local OS container (as direct
9042 connection, without requiring SSH). This works on any
9043 container that is registered with machined, such as those
9044 created by libvirt-lxc or nspawn.
9046 * systemd-run and systemd-analyze also gained support for "-H"
9047 to connect to remote hosts via SSH. This is particularly
9048 useful for systemd-run because it enables queuing of jobs
9049 onto remote systems.
9051 * machinectl gained a new command "login" to open a getty
9052 login in any local container. This works with any container
9053 that is registered with machined (such as those created by
9054 libvirt-lxc or nspawn), and which runs systemd inside.
9056 * machinectl gained a new "reboot" command that may be used to
9057 trigger a reboot on a specific container that is registered
9058 with machined. This works on any container that runs an init
9059 system of some kind.
9061 * systemctl gained a new "list-timers" command to print a nice
9062 listing of installed timer units with the times they elapse
9065 * Alternative reboot() parameters may now be specified on the
9066 "systemctl reboot" command line and are passed to the
9067 reboot() system call.
9069 * systemctl gained a new --job-mode= switch to configure the
9070 mode to queue a job with. This is a more generic version of
9071 --fail, --irreversible, and --ignore-dependencies, which are
9072 still available but not advertised anymore.
9074 * /etc/systemd/system.conf gained new settings to configure
9075 various default timeouts of units, as well as the default
9076 start limit interval and burst. These may still be overridden
9079 * PID1 will now export on the bus profile data of the security
9080 policy upload process (such as the SELinux policy upload to
9083 * journald: when forwarding logs to the console, include
9084 timestamps (following the setting in
9085 /sys/module/printk/parameters/time).
9087 * OnCalendar= in timer units now understands the special
9088 strings "yearly" and "annually". (Both are equivalent)
9090 * The accuracy of timer units is now configurable with the new
9091 AccuracySec= setting. It defaults to 1min.
9093 * A new dependency type JoinsNamespaceOf= has been added that
9094 allows running two services within the same /tmp and network
9095 namespace, if PrivateNetwork= or PrivateTmp= are used.
9097 * A new command "cat" has been added to systemctl. It outputs
9098 the original unit file of a unit, and concatenates the
9099 contents of additional "drop-in" unit file snippets, so that
9100 the full configuration is shown.
9102 * systemctl now supports globbing on the various "list-xyz"
9103 commands, like "list-units" or "list-sockets", as well as on
9104 those commands which take multiple unit names.
9106 * journalctl's --unit= switch gained support for globbing.
9108 * All systemd daemons now make use of the watchdog logic so
9109 that systemd automatically notices when they hang.
9111 * If the $container_ttys environment variable is set,
9112 getty-generator will automatically spawn a getty for each
9113 listed tty. This is useful for container managers to request
9114 login gettys to be spawned on as many ttys as needed.
9116 * %h, %s, %U specifier support is not available anymore when
9117 used in unit files for PID 1. This is because NSS calls are
9118 not safe from PID 1. They stay available for --user
9119 instances of systemd, and as special case for the root user.
9121 * loginctl gained a new "--no-legend" switch to turn off output
9124 * The "sd-login.h" API gained three new calls:
9125 sd_session_is_remote(), sd_session_get_remote_user(),
9126 sd_session_get_remote_host() to query information about
9129 * The udev hardware database now also carries vendor/product
9130 information of SDIO devices.
9132 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
9133 determine whether watchdog notifications are requested by
9136 * Socket-activated per-connection services now include a
9137 short description of the connection parameters in the
9140 * tmpfiles gained a new "--boot" option. When this is not used,
9141 only lines where the command character is not suffixed with
9142 "!" are executed. When this option is specified, those
9143 options are executed too. This partitions tmpfiles
9144 directives into those that can be safely executed at any
9145 time, and those which should be run only at boot (for
9146 example, a line that creates /run/nologin).
9148 * A new API "sd-resolve.h" has been added which provides a simple
9149 asynchronous wrapper around glibc NSS hostname resolution
9150 calls, such as getaddrinfo(). In contrast to glibc's
9151 getaddrinfo_a(), it does not use signals. In contrast to most
9152 other asynchronous name resolution libraries, this one does
9153 not reimplement DNS, but reuses NSS, so that alternate
9154 hostname resolution systems continue to work, such as mDNS,
9155 LDAP, etc. This API is based on libasyncns, but it has been
9156 cleaned up for inclusion in systemd.
9158 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
9159 "sd-daemon.h" are no longer found in individual libraries
9160 libsystemd-journal.so, libsystemd-login.so,
9161 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
9162 merged them into a single library, libsystemd.so, which
9163 provides all symbols. The reason for this is cyclic
9164 dependencies, as these libraries tend to use each other's
9165 symbols. So far, we have managed to workaround that by linking
9166 a copy of a good part of our code into each of these
9167 libraries again and again, which, however, makes certain
9168 things hard to do, like sharing static variables. Also, it
9169 substantially increases footprint. With this change, there
9170 is only one library for the basic APIs systemd
9171 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
9172 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
9173 library as well, however are subject to the --enable-kdbus
9174 switch (see below). Note that "sd-dhcp-client.h" is not part
9175 of this library (this is because it only consumes, never
9176 provides, services of/to other APIs). To make the transition
9177 easy from the separate libraries to the unified one, we
9178 provide the --enable-compat-libs compile-time switch which
9179 will generate stub libraries that are compatible with the
9180 old ones but redirect all calls to the new one.
9182 * All of the kdbus logic and the new APIs "sd-bus.h",
9183 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
9184 and "sd-utf8.h" are compile-time optional via the
9185 "--enable-kdbus" switch, and they are not compiled in by
9186 default. To make use of kdbus, you have to explicitly enable
9187 the switch. Note however, that neither the kernel nor the
9188 userspace API for all of this is considered stable yet. We
9189 want to maintain the freedom to still change the APIs for
9190 now. By specifying this build-time switch, you acknowledge
9191 that you are aware of the instability of the current
9194 * Also, note that while kdbus is pretty much complete,
9195 it lacks one thing: proper policy support. This means you
9196 can build a fully working system with all features; however,
9197 it will be highly insecure. Policy support will be added in
9198 one of the next releases, at the same time that we will
9199 declare the APIs stable.
9201 * When the kernel command line argument "kdbus" is specified,
9202 systemd will automatically load the kdbus.ko kernel module. At
9203 this stage of development, it is only useful for testing kdbus
9204 and should not be used in production. Note: if "--enable-kdbus"
9205 is specified, and the kdbus.ko kernel module is available, and
9206 "kdbus" is added to the kernel command line, the entire system
9207 runs with kdbus instead of dbus-daemon, with the above mentioned
9208 problem of missing the system policy enforcement. Also a future
9209 version of kdbus.ko or a newer systemd will not be compatible with
9210 each other, and will unlikely be able to boot the machine if only
9211 one of them is updated.
9213 * systemctl gained a new "import-environment" command which
9214 uploads the caller's environment (or parts thereof) into the
9215 service manager so that it is inherited by services started
9216 by the manager. This is useful to upload variables like
9217 $DISPLAY into the user service manager.
9219 * A new PrivateDevices= switch has been added to service units
9220 which allows running a service with a namespaced /dev
9221 directory that does not contain any device nodes for
9222 physical devices. More specifically, it only includes devices
9223 such as /dev/null, /dev/urandom, and /dev/zero which are API
9226 * logind has been extended to support behaviour like VT
9227 switching on seats that do not support a VT. This makes
9228 multi-session available on seats that are not the first seat
9229 (seat0), and on systems where kernel support for VTs has
9230 been disabled at compile-time.
9232 * If a process holds a delay lock for system sleep or shutdown
9233 and fails to release it in time, we will now log its
9234 identity. This makes it easier to identify processes that
9235 cause slow suspends or power-offs.
9237 * When parsing /etc/crypttab, support for a new key-slot=
9238 option as supported by Debian is added. It allows indicating
9239 which LUKS slot to use on disk, speeding up key loading.
9241 * The sd_journal_sendv() API call has been checked and
9242 officially declared to be async-signal-safe so that it may
9243 be invoked from signal handlers for logging purposes.
9245 * Boot-time status output is now enabled automatically after a
9246 short timeout if boot does not progress, in order to give
9247 the user an indication what she or he is waiting for.
9249 * The boot-time output has been improved to show how much time
9250 remains until jobs expire.
9252 * The KillMode= switch in service units gained a new possible
9253 value "mixed". If set, and the unit is shut down, then the
9254 initial SIGTERM signal is sent only to the main daemon
9255 process, while the following SIGKILL signal is sent to
9256 all remaining processes of the service.
9258 * When a scope unit is registered, a new property "Controller"
9259 may be set. If set to a valid bus name, systemd will send a
9260 RequestStop() signal to this name when it would like to shut
9261 down the scope. This may be used to hook manager logic into
9262 the shutdown logic of scope units. Also, scope units may now
9263 be put in a special "abandoned" state, in which case the
9264 manager process which created them takes no further
9265 responsibilities for it.
9267 * When reading unit files, systemd will now verify
9268 the access mode of these files, and warn about certain
9269 suspicious combinations. This has been added to make it
9270 easier to track down packaging bugs where unit files are
9271 marked executable or world-writable.
9273 * systemd-nspawn gained a new "--setenv=" switch to set
9274 container-wide environment variables. The similar option in
9275 systemd-activate was renamed from "--environment=" to
9276 "--setenv=" for consistency.
9278 * systemd-nspawn has been updated to create a new kdbus domain
9279 for each container that is invoked, thus allowing each
9280 container to have its own set of system and user buses,
9281 independent of the host.
9283 * systemd-nspawn gained a new --drop-capability= switch to run
9284 the container with less capabilities than the default. Both
9285 --drop-capability= and --capability= now take the special
9286 string "all" for dropping or keeping all capabilities.
9288 * systemd-nspawn gained new switches for executing containers
9289 with specific SELinux labels set.
9291 * systemd-nspawn gained a new --quiet switch to not generate
9292 any additional output but the container's own console
9295 * systemd-nspawn gained a new --share-system switch to run a
9296 container without PID namespacing enabled.
9298 * systemd-nspawn gained a new --register= switch to control
9299 whether the container is registered with systemd-machined or
9300 not. This is useful for containers that do not run full
9301 OS images, but only specific apps.
9303 * systemd-nspawn gained a new --keep-unit which may be used
9304 when invoked as the only program from a service unit, and
9305 results in registration of the unit service itself in
9306 systemd-machined, instead of a newly opened scope unit.
9308 * systemd-nspawn gained a new --network-interface= switch for
9309 moving arbitrary interfaces to the container. The new
9310 --network-veth switch creates a virtual Ethernet connection
9311 between host and container. The new --network-bridge=
9312 switch then allows assigning the host side of this virtual
9313 Ethernet connection to a bridge device.
9315 * systemd-nspawn gained a new --personality= switch for
9316 setting the kernel personality for the container. This is
9317 useful when running a 32-bit container on a 64-bit host. A
9318 similar option Personality= is now also available for service
9321 * logind will now also track a "Desktop" identifier for each
9322 session which encodes the desktop environment of it. This is
9323 useful for desktop environments that want to identify
9324 multiple running sessions of itself easily.
9326 * A new SELinuxContext= setting for service units has been
9327 added that allows setting a specific SELinux execution
9328 context for a service.
9330 * Most systemd client tools will now honour $SYSTEMD_LESS for
9331 settings of the "less" pager. By default, these tools will
9332 override $LESS to allow certain operations to work, such as
9333 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
9334 influence this logic.
9336 * systemd's "seccomp" hook-up has been changed to make use of
9337 the libseccomp library instead of using its own
9338 implementation. This has benefits for portability among
9341 * For usage together with SystemCallFilter=, a new
9342 SystemCallErrorNumber= setting has been introduced that
9343 allows configuration of a system error number to be returned
9344 on filtered system calls, instead of immediately killing the
9345 process. Also, SystemCallArchitectures= has been added to
9346 limit access to system calls of a particular architecture
9347 (in order to turn off support for unused secondary
9348 architectures). There is also a global
9349 SystemCallArchitectures= setting in system.conf now to turn
9350 off support for non-native system calls system-wide.
9352 * systemd requires a kernel with a working name_to_handle_at(),
9353 please see the kernel config requirements in the README file.
9355 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
9356 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
9357 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
9358 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
9359 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
9360 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
9361 Elia Pinto, Florian Weimer, George McCollister, Goffredo
9362 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
9363 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
9364 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
9365 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
9366 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
9367 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
9368 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
9369 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
9370 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
9371 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
9372 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
9373 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
9374 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
9375 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
9376 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
9377 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
9378 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
9380 — Berlin, 2014-02-20
9384 * logind has gained support for facilitating privileged input
9385 and drm device access for unprivileged clients. This work is
9386 useful to allow Wayland display servers (and similar
9387 programs, such as kmscon) to run under the user's ID and
9388 access input and drm devices which are normally
9389 protected. When this is used (and the kernel is new enough)
9390 logind will "mute" IO on the file descriptors passed to
9391 Wayland as long as it is in the background and "unmute" it
9392 if it returns into the foreground. This allows secure
9393 session switching without allowing background sessions to
9394 eavesdrop on input and display data. This also introduces
9395 session switching support if VT support is turned off in the
9396 kernel, and on seats that are not seat0.
9398 * A new kernel command line option luks.options= is understood
9399 now which allows specifying LUKS options for usage for LUKS
9400 encrypted partitions specified with luks.uuid=.
9402 * tmpfiles.d(5) snippets may now use specifier expansion in
9403 path names. More specifically %m, %b, %H, %v, are now
9404 replaced by the local machine id, boot id, hostname, and
9405 kernel version number.
9407 * A new tmpfiles.d(5) command "m" has been introduced which
9408 may be used to change the owner/group/access mode of a file
9409 or directory if it exists, but do nothing if it does not.
9411 * This release removes high-level support for the
9412 MemorySoftLimit= cgroup setting. The underlying kernel
9413 cgroup attribute memory.soft_limit= is currently badly
9414 designed and likely to be removed from the kernel API in its
9415 current form, hence we should not expose it for now.
9417 * The memory.use_hierarchy cgroup attribute is now enabled for
9418 all cgroups systemd creates in the memory cgroup
9419 hierarchy. This option is likely to be come the built-in
9420 default in the kernel anyway, and the non-hierarchical mode
9421 never made much sense in the intrinsically hierarchical
9424 * A new field _SYSTEMD_SLICE= is logged along with all journal
9425 messages containing the slice a message was generated
9426 from. This is useful to allow easy per-customer filtering of
9427 logs among other things.
9429 * systemd-journald will no longer adjust the group of journal
9430 files it creates to the "systemd-journal" group. Instead we
9431 rely on the journal directory to be owned by the
9432 "systemd-journal" group, and its setgid bit set, so that the
9433 kernel file system layer will automatically enforce that
9434 journal files inherit this group assignment. The reason for
9435 this change is that we cannot allow NSS look-ups from
9436 journald which would be necessary to resolve
9437 "systemd-journal" to a numeric GID, because this might
9438 create deadlocks if NSS involves synchronous queries to
9439 other daemons (such as nscd, or sssd) which in turn are
9440 logging clients of journald and might block on it, which
9441 would then dead lock. A tmpfiles.d(5) snippet included in
9442 systemd will make sure the setgid bit and group are
9443 properly set on the journal directory if it exists on every
9444 boot. However, we recommend adjusting it manually after
9445 upgrades too (or from RPM scriptlets), so that the change is
9446 not delayed until next reboot.
9448 * Backlight and random seed files in /var/lib/ have moved into
9449 the /var/lib/systemd/ directory, in order to centralize all
9450 systemd generated files in one directory.
9452 * Boot time performance measurements (as displayed by
9453 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
9454 performance information if that's available to determine how
9455 much time BIOS and boot loader initialization required. With
9456 a sufficiently new BIOS you hence no longer need to boot
9457 with Gummiboot to get access to such information.
9459 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
9460 Cristian Rodríguez, Dave Reisner, David Herrmann, David
9461 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
9462 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
9463 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
9464 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
9465 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9467 — Berlin, 2013-10-02
9471 * The Restart= option for services now understands a new
9472 on-watchdog setting, which will restart the service
9473 automatically if the service stops sending out watchdog keep
9474 alive messages (as configured with WatchdogSec=).
9476 * The getty generator (which is responsible for bringing up a
9477 getty on configured serial consoles) will no longer only
9478 start a getty on the primary kernel console but on all
9479 others, too. This makes the order in which console= is
9480 specified on the kernel command line less important.
9482 * libsystemd-logind gained a new sd_session_get_vt() call to
9483 retrieve the VT number of a session.
9485 * If the option "tries=0" is set for an entry of /etc/crypttab
9486 its passphrase is queried indefinitely instead of any
9487 maximum number of tries.
9489 * If a service with a configure PID file terminates its PID
9490 file will now be removed automatically if it still exists
9491 afterwards. This should put an end to stale PID files.
9493 * systemd-run will now also take relative binary path names
9494 for execution and no longer insists on absolute paths.
9496 * InaccessibleDirectories= and ReadOnlyDirectories= now take
9497 paths that are optionally prefixed with "-" to indicate that
9498 it should not be considered a failure if they do not exist.
9500 * journalctl -o (and similar commands) now understands a new
9501 output mode "short-precise", it is similar to "short" but
9502 shows timestamps with usec accuracy.
9504 * The option "discard" (as known from Debian) is now
9505 synonymous to "allow-discards" in /etc/crypttab. In fact,
9506 "discard" is preferred now (since it is easier to remember
9509 * Some licensing clean-ups were made, so that more code is now
9510 LGPL-2.1 licensed than before.
9512 * A minimal tool to save/restore the display backlight
9513 brightness across reboots has been added. It will store the
9514 backlight setting as late as possible at shutdown, and
9515 restore it as early as possible during reboot.
9517 * A logic to automatically discover and enable home and swap
9518 partitions on GPT disks has been added. With this in place
9519 /etc/fstab becomes optional for many setups as systemd can
9520 discover certain partitions located on the root disk
9521 automatically. Home partitions are recognized under their
9522 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
9523 partitions are recognized under their GPT type ID
9524 0657fd6da4ab43c484e50933c84b4f4f.
9526 * systemd will no longer pass any environment from the kernel
9527 or initrd to system services. If you want to set an
9528 environment for all services, do so via the kernel command
9529 line systemd.setenv= assignment.
9531 * The systemd-sysctl tool no longer natively reads the file
9532 /etc/sysctl.conf. If desired, the file should be symlinked
9533 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
9534 legacy support by a symlink rather than built-in code, it
9535 also makes the otherwise hidden order of application of the
9536 different files visible. (Note that this partly reverts to a
9537 pre-198 application order of sysctl knobs!)
9539 * The "systemctl set-log-level" and "systemctl dump" commands
9540 have been moved to systemd-analyze.
9542 * systemd-run learned the new --remain-after-exit switch,
9543 which causes the scope unit not to be cleaned up
9544 automatically after the process terminated.
9546 * tmpfiles learned a new --exclude-prefix= switch to exclude
9547 certain paths from operation.
9549 * journald will now automatically flush all messages to disk
9550 as soon as a message at the log level CRIT, ALERT or EMERG
9553 Contributions from: Andrew Cook, Brandon Philips, Christian
9554 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
9555 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
9556 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
9557 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
9558 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
9559 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
9560 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
9561 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
9562 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
9563 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
9564 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
9565 William Giokas, Zbigniew Jędrzejewski-Szmek
9567 — Berlin, 2013-09-13
9571 * The documentation has been updated to cover the various new
9572 concepts introduced with 205.
9574 * Unit files now understand the new %v specifier which
9575 resolves to the kernel version string as returned by "uname
9578 * systemctl now supports filtering the unit list output by
9579 load state, active state and sub state, using the new
9582 * "systemctl status" will now show the results of the
9583 condition checks (like ConditionPathExists= and similar) of
9584 the last start attempts of the unit. They are also logged to
9587 * "journalctl -b" may now be used to look for boot output of a
9588 specific boot. Try "journalctl -b -1" for the previous boot,
9589 but the syntax is substantially more powerful.
9591 * "journalctl --show-cursor" has been added which prints the
9592 cursor string the last shown log line. This may then be used
9593 with the new "journalctl --after-cursor=" switch to continue
9594 browsing logs from that point on.
9596 * "journalctl --force" may now be used to force regeneration
9599 * Creation of "dead" device nodes has been moved from udev
9600 into kmod and tmpfiles. Previously, udev would read the kmod
9601 databases to pre-generate dead device nodes based on meta
9602 information contained in kernel modules, so that these would
9603 be auto-loaded on access rather then at boot. As this
9604 does not really have much to do with the exposing actual
9605 kernel devices to userspace this has always been slightly
9606 alien in the udev codebase. Following the new scheme kmod
9607 will now generate a runtime snippet for tmpfiles from the
9608 module meta information and it now is tmpfiles' job to the
9609 create the nodes. This also allows overriding access and
9610 other parameters for the nodes using the usual tmpfiles
9611 facilities. As side effect this allows us to remove the
9612 CAP_SYS_MKNOD capability bit from udevd entirely.
9614 * logind's device ACLs may now be applied to these "dead"
9615 devices nodes too, thus finally allowing managed access to
9616 devices such as /dev/snd/sequencer without loading the
9617 backing module right-away.
9619 * A new RPM macro has been added that may be used to apply
9620 tmpfiles configuration during package installation.
9622 * systemd-detect-virt and ConditionVirtualization= now can
9623 detect User-Mode-Linux machines (UML).
9625 * journald will now implicitly log the effective capabilities
9626 set of processes in the message metadata.
9628 * systemd-cryptsetup has gained support for TrueCrypt volumes.
9630 * The initrd interface has been simplified (more specifically,
9631 support for passing performance data via environment
9632 variables and fsck results via files in /run has been
9633 removed). These features were non-essential, and are
9634 nowadays available in a much nicer way by having systemd in
9635 the initrd serialize its state and have the hosts systemd
9636 deserialize it again.
9638 * The udev "keymap" data files and tools to apply keyboard
9639 specific mappings of scan to key codes, and force-release
9640 scan code lists have been entirely replaced by a udev
9641 "keyboard" builtin and a hwdb data file.
9643 * systemd will now honour the kernel's "quiet" command line
9644 argument also during late shutdown, resulting in a
9645 completely silent shutdown when used.
9647 * There's now an option to control the SO_REUSEPORT socket
9648 option in .socket units.
9650 * Instance units will now automatically get a per-template
9651 subslice of system.slice unless something else is explicitly
9652 configured. For example, instances of sshd@.service will now
9653 implicitly be placed in system-sshd.slice rather than
9654 system.slice as before.
9656 * Test coverage support may now be enabled at build time.
9658 Contributions from: Dave Reisner, Frederic Crozat, Harald
9659 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
9660 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
9661 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
9662 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
9663 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
9664 Giokas, Zbigniew Jędrzejewski-Szmek
9666 — Berlin, 2013-07-23
9670 * Two new unit types have been introduced:
9672 Scope units are very similar to service units, however, are
9673 created out of pre-existing processes — instead of PID 1
9674 forking off the processes. By using scope units it is
9675 possible for system services and applications to group their
9676 own child processes (worker processes) in a powerful way
9677 which then maybe used to organize them, or kill them
9678 together, or apply resource limits on them.
9680 Slice units may be used to partition system resources in an
9681 hierarchical fashion and then assign other units to them. By
9682 default there are now three slices: system.slice (for all
9683 system services), user.slice (for all user sessions),
9684 machine.slice (for VMs and containers).
9686 Slices and scopes have been introduced primarily in
9687 context of the work to move cgroup handling to a
9688 single-writer scheme, where only PID 1
9689 creates/removes/manages cgroups.
9691 * There's a new concept of "transient" units. In contrast to
9692 normal units these units are created via an API at runtime,
9693 not from configuration from disk. More specifically this
9694 means it is now possible to run arbitrary programs as
9695 independent services, with all execution parameters passed
9696 in via bus APIs rather than read from disk. Transient units
9697 make systemd substantially more dynamic then it ever was,
9698 and useful as a general batch manager.
9700 * logind has been updated to make use of scope and slice units
9701 for managing user sessions. As a user logs in he will get
9702 his own private slice unit, to which all sessions are added
9703 as scope units. We also added support for automatically
9704 adding an instance of user@.service for the user into the
9705 slice. Effectively logind will no longer create cgroup
9706 hierarchies on its own now, it will defer entirely to PID 1
9707 for this by means of scope, service and slice units. Since
9708 user sessions this way become entities managed by PID 1
9709 the output of "systemctl" is now a lot more comprehensive.
9711 * A new mini-daemon "systemd-machined" has been added which
9712 may be used by virtualization managers to register local
9713 VMs/containers. nspawn has been updated accordingly, and
9714 libvirt will be updated shortly. machined will collect a bit
9715 of meta information about the VMs/containers, and assign
9716 them their own scope unit (see above). The collected
9717 meta-data is then made available via the "machinectl" tool,
9718 and exposed in "ps" and similar tools. machined/machinectl
9719 is compile-time optional.
9721 * As discussed earlier, the low-level cgroup configuration
9722 options ControlGroup=, ControlGroupModify=,
9723 ControlGroupPersistent=, ControlGroupAttribute= have been
9724 removed. Please use high-level attribute settings instead as
9725 well as slice units.
9727 * A new bus call SetUnitProperties() has been added to alter
9728 various runtime parameters of a unit. This is primarily
9729 useful to alter cgroup parameters dynamically in a nice way,
9730 but will be extended later on to make more properties
9731 modifiable at runtime. systemctl gained a new set-properties
9732 command that wraps this call.
9734 * A new tool "systemd-run" has been added which can be used to
9735 run arbitrary command lines as transient services or scopes,
9736 while configuring a number of settings via the command
9737 line. This tool is currently very basic, however already
9738 very useful. We plan to extend this tool to even allow
9739 queuing of execution jobs with time triggers from the
9740 command line, similar in fashion to "at".
9742 * nspawn will now inform the user explicitly that kernels with
9743 audit enabled break containers, and suggest the user to turn
9746 * Support for detecting the IMA and AppArmor security
9747 frameworks with ConditionSecurity= has been added.
9749 * journalctl gained a new "-k" switch for showing only kernel
9750 messages, mimicking dmesg output; in addition to "--user"
9751 and "--system" switches for showing only user's own logs
9754 * systemd-delta can now show information about drop-in
9755 snippets extending unit files.
9757 * libsystemd-bus has been substantially updated but is still
9758 not available as public API.
9760 * systemd will now look for the "debug" argument on the kernel
9761 command line and enable debug logging, similar to what
9762 "systemd.log_level=debug" already did before.
9764 * "systemctl set-default", "systemctl get-default" has been
9765 added to configure the default.target symlink, which
9766 controls what to boot into by default.
9768 * "systemctl set-log-level" has been added as a convenient
9769 way to raise and lower systemd logging threshold.
9771 * "systemd-analyze plot" will now show the time the various
9772 generators needed for execution, as well as information
9773 about the unit file loading.
9775 * libsystemd-journal gained a new sd_journal_open_files() call
9776 for opening specific journal files. journactl also gained a
9777 new switch to expose this new functionality. Previously we
9778 only supported opening all files from a directory, or all
9779 files from the system, as opening individual files only is
9780 racy due to journal file rotation.
9782 * systemd gained the new DefaultEnvironment= setting in
9783 /etc/systemd/system.conf to set environment variables for
9786 * If a privileged process logs a journal message with the
9787 OBJECT_PID= field set, then journald will automatically
9788 augment this with additional OBJECT_UID=, OBJECT_GID=,
9789 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
9790 system services want to log events about specific client
9791 processes. journactl/systemctl has been updated to make use
9792 of this information if all log messages regarding a specific
9795 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
9796 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
9797 Reisner, David Coppa, David King, David Strauss, Eelco
9798 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
9799 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
9800 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
9801 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
9802 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
9803 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
9804 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
9805 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
9806 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
9807 Łukasz Stelmach, 장동준
9811 * The Python bindings gained some minimal support for the APIs
9812 exposed by libsystemd-logind.
9814 * ConditionSecurity= gained support for detecting SMACK. Since
9815 this condition already supports SELinux and AppArmor we only
9816 miss IMA for this. Patches welcome!
9818 Contributions from: Karol Lewandowski, Lennart Poettering,
9819 Zbigniew Jędrzejewski-Szmek
9823 * systemd-nspawn will now create /etc/resolv.conf if
9824 necessary, before bind-mounting the host's file onto it.
9826 * systemd-nspawn will now store meta information about a
9827 container on the container's cgroup as extended attribute
9828 fields, including the root directory.
9830 * The cgroup hierarchy has been reworked in many ways. All
9831 objects any of the components systemd creates in the cgroup
9832 tree are now suffixed. More specifically, user sessions are
9833 now placed in cgroups suffixed with ".session", users in
9834 cgroups suffixed with ".user", and nspawn containers in
9835 cgroups suffixed with ".nspawn". Furthermore, all cgroup
9836 names are now escaped in a simple scheme to avoid collision
9837 of userspace object names with kernel filenames. This work
9838 is preparation for making these objects relocatable in the
9839 cgroup tree, in order to allow easy resource partitioning of
9840 these objects without causing naming conflicts.
9842 * systemctl list-dependencies gained the new switches
9843 --plain, --reverse, --after and --before.
9845 * systemd-inhibit now shows the process name of processes that
9846 have taken an inhibitor lock.
9848 * nss-myhostname will now also resolve "localhost"
9849 implicitly. This makes /etc/hosts an optional file and
9850 nicely handles that on IPv6 ::1 maps to both "localhost" and
9853 * libsystemd-logind.so gained a new call
9854 sd_get_machine_names() to enumerate running containers and
9855 VMs (currently only supported by very new libvirt and
9856 nspawn). sd_login_monitor can now be used to watch
9857 VMs/containers coming and going.
9859 * .include is not allowed recursively anymore, and only in
9860 unit files. Usually it is better to use drop-in snippets in
9861 .d/*.conf anyway, as introduced with systemd 198.
9863 * systemd-analyze gained a new "critical-chain" command that
9864 determines the slowest chain of units run during system
9865 boot-up. It is very useful for tracking down where
9866 optimizing boot time is the most beneficial.
9868 * systemd will no longer allow manipulating service paths in
9869 the name=systemd:/system cgroup tree using ControlGroup= in
9870 units. (But is still fine with it in all other dirs.)
9872 * There's a new systemd-nspawn@.service service file that may
9873 be used to easily run nspawn containers as system
9874 services. With the container's root directory in
9875 /var/lib/container/foobar it is now sufficient to run
9876 "systemctl start systemd-nspawn@foobar.service" to boot it.
9878 * systemd-cgls gained a new parameter "--machine" to list only
9879 the processes within a certain container.
9881 * ConditionSecurity= now can check for "apparmor". We still
9882 are lacking checks for SMACK and IMA for this condition
9883 check though. Patches welcome!
9885 * A new configuration file /etc/systemd/sleep.conf has been
9886 added that may be used to configure which kernel operation
9887 systemd is supposed to execute when "suspend", "hibernate"
9888 or "hybrid-sleep" is requested. This makes the new kernel
9889 "freeze" state accessible to the user.
9891 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
9892 the passed argument if applicable.
9894 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
9895 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
9896 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
9897 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
9898 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
9899 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
9900 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
9905 * The output of 'systemctl list-jobs' got some polishing. The
9906 '--type=' argument may now be passed more than once. A new
9907 command 'systemctl list-sockets' has been added which shows
9908 a list of kernel sockets systemd is listening on with the
9909 socket units they belong to, plus the units these socket
9912 * The experimental libsystemd-bus library got substantial
9913 updates to work in conjunction with the (also experimental)
9914 kdbus kernel project. It works well enough to exchange
9915 messages with some sophistication. Note that kdbus is not
9916 ready yet, and the library is mostly an elaborate test case
9917 for now, and not installable.
9919 * systemd gained a new unit 'systemd-static-nodes.service'
9920 that generates static device nodes earlier during boot, and
9921 can run in conjunction with udev.
9923 * libsystemd-login gained a new call sd_pid_get_user_unit()
9924 to retrieve the user systemd unit a process is running
9925 in. This is useful for systems where systemd is used as
9928 * systemd-nspawn now places all containers in the new /machine
9929 top-level cgroup directory in the name=systemd
9930 hierarchy. libvirt will soon do the same, so that we get a
9931 uniform separation of /system, /user and /machine for system
9932 services, user processes and containers/virtual
9933 machines. This new cgroup hierarchy is also useful to stick
9934 stable names to specific container instances, which can be
9935 recognized later this way (this name may be controlled
9936 via systemd-nspawn's new -M switch). libsystemd-login also
9937 gained a new call sd_pid_get_machine_name() to retrieve the
9938 name of the container/VM a specific process belongs to.
9940 * bootchart can now store its data in the journal.
9942 * libsystemd-journal gained a new call
9943 sd_journal_add_conjunction() for AND expressions to the
9944 matching logic. This can be used to express more complex
9945 logical expressions.
9947 * journactl can now take multiple --unit= and --user-unit=
9950 * The cryptsetup logic now understands the "luks.key=" kernel
9951 command line switch for specifying a file to read the
9952 decryption key from. Also, if a configured key file is not
9953 found the tool will now automatically fall back to prompting
9956 * Python systemd.journal module was updated to wrap recently
9957 added functions from libsystemd-journal. The interface was
9958 changed to bring the low level interface in s.j._Reader
9959 closer to the C API, and the high level interface in
9960 s.j.Reader was updated to wrap and convert all data about
9963 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
9964 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
9965 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
9966 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
9967 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
9968 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9972 * journalctl --update-catalog now understands a new --root=
9973 option to operate on catalogs found in a different root
9976 * During shutdown after systemd has terminated all running
9977 services a final killing loop kills all remaining left-over
9978 processes. We will now print the name of these processes
9979 when we send SIGKILL to them, since this usually indicates a
9982 * If /etc/crypttab refers to password files stored on
9983 configured mount points automatic dependencies will now be
9984 generated to ensure the specific mount is established first
9985 before the key file is attempted to be read.
9987 * 'systemctl status' will now show information about the
9988 network sockets a socket unit is listening on.
9990 * 'systemctl status' will also shown information about any
9991 drop-in configuration file for units. (Drop-In configuration
9992 files in this context are files such as
9993 /etc/systemd/systemd/foobar.service.d/*.conf)
9995 * systemd-cgtop now optionally shows summed up CPU times of
9996 cgroups. Press '%' while running cgtop to switch between
9997 percentage and absolute mode. This is useful to determine
9998 which cgroups use up the most CPU time over the entire
9999 runtime of the system. systemd-cgtop has also been updated
10000 to be 'pipeable' for processing with further shell tools.
10002 * 'hostnamectl set-hostname' will now allow setting of FQDN
10005 * The formatting and parsing of time span values has been
10006 changed. The parser now understands fractional expressions
10007 such as "5.5h". The formatter will now output fractional
10008 expressions for all time spans under 1min, i.e. "5.123456s"
10009 rather than "5s 123ms 456us". For time spans under 1s
10010 millisecond values are shown, for those under 1ms
10011 microsecond values are shown. This should greatly improve
10012 all time-related output of systemd.
10014 * libsystemd-login and libsystemd-journal gained new
10015 functions for querying the poll() events mask and poll()
10016 timeout value for integration into arbitrary event
10019 * localectl gained the ability to list available X11 keymaps
10020 (models, layouts, variants, options).
10022 * 'systemd-analyze dot' gained the ability to filter for
10023 specific units via shell-style globs, to create smaller,
10024 more useful graphs. I.e. it is now possible to create simple
10025 graphs of all the dependencies between only target units, or
10026 of all units that Avahi has dependencies with.
10028 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
10029 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
10030 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
10031 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
10032 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
10033 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
10034 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
10038 * The boot-time readahead implementation for rotating media
10039 will now read the read-ahead data in multiple passes which
10040 consist of all read requests made in equidistant time
10041 intervals. This means instead of strictly reading read-ahead
10042 data in its physical order on disk we now try to find a
10043 middle ground between physical and access time order.
10045 * /etc/os-release files gained a new BUILD_ID= field for usage
10046 on operating systems that provide continuous builds of OS
10049 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
10050 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
10051 William Douglas, Zbigniew Jędrzejewski-Szmek
10055 * systemd-python gained an API exposing libsystemd-daemon.
10057 * The SMACK setup logic gained support for uploading CIPSO
10060 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
10061 ReadOnlyDirectories= and InaccessibleDirectories= has
10062 changed. The private /tmp and /var/tmp directories are now
10063 shared by all processes of a service (which means
10064 ExecStartPre= may now leave data in /tmp that ExecStart= of
10065 the same service can still access). When a service is
10066 stopped its temporary directories are immediately deleted
10067 (normal clean-up with tmpfiles is still done in addition to
10070 * By default, systemd will now set a couple of sysctl
10071 variables in the kernel: the safe sysrq options are turned
10072 on, IP route verification is turned on, and source routing
10073 disabled. The recently added hardlink and softlink
10074 protection of the kernel is turned on. These settings should
10075 be reasonably safe, and good defaults for all new systems.
10077 * The predictable network naming logic may now be turned off
10078 with a new kernel command line switch: net.ifnames=0.
10080 * A new libsystemd-bus module has been added that implements a
10081 pretty complete D-Bus client library. For details see:
10083 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
10085 * journald will now explicitly flush the journal files to disk
10086 at the latest 5min after each write. The file will then also
10087 be marked offline until the next write. This should increase
10088 reliability in case of a crash. The synchronization delay
10089 can be configured via SyncIntervalSec= in journald.conf.
10091 * There's a new remote-fs-setup.target unit that can be used
10092 to pull in specific services when at least one remote file
10093 system is to be mounted.
10095 * There are new targets timers.target and paths.target as
10096 canonical targets to pull user timer and path units in
10097 from. This complements sockets.target with a similar
10098 purpose for socket units.
10100 * libudev gained a new call udev_device_set_attribute_value()
10101 to set sysfs attributes of a device.
10103 * The udev daemon now sets the default number of worker
10104 processes executed in parallel based on the number of available
10105 CPUs instead of the amount of available RAM. This is supposed
10106 to provide a more reliable default and limit a too aggressive
10107 parallelism for setups with 1000s of devices connected.
10109 Contributions from: Auke Kok, Colin Walters, Cristian
10110 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
10111 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
10112 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
10113 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
10114 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
10115 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
10116 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
10117 Zbigniew Jędrzejewski-Szmek
10121 * Configuration of unit files may now be extended via drop-in
10122 files without having to edit/override the unit files
10123 themselves. More specifically, if the administrator wants to
10124 change one value for a service file foobar.service he can
10125 now do so by dropping in a configuration snippet into
10126 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
10127 will load all these snippets and apply them on top of the
10128 main unit configuration file, possibly extending or
10129 overriding its settings. Using these drop-in snippets is
10130 generally nicer than the two earlier options for changing
10131 unit files locally: copying the files from
10132 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
10133 them there; or creating a new file in /etc/systemd/system/
10134 that incorporates the original one via ".include". Drop-in
10135 snippets into these .d/ directories can be placed in any
10136 directory systemd looks for units in, and the usual
10137 overriding semantics between /usr/lib, /etc and /run apply
10140 * Most unit file settings which take lists of items can now be
10141 reset by assigning the empty string to them. For example,
10142 normally, settings such as Environment=FOO=BAR append a new
10143 environment variable assignment to the environment block,
10144 each time they are used. By assigning Environment= the empty
10145 string the environment block can be reset to empty. This is
10146 particularly useful with the .d/*.conf drop-in snippets
10147 mentioned above, since this adds the ability to reset list
10148 settings from vendor unit files via these drop-ins.
10150 * systemctl gained a new "list-dependencies" command for
10151 listing the dependencies of a unit recursively.
10153 * Inhibitors are now honored and listed by "systemctl
10154 suspend", "systemctl poweroff" (and similar) too, not only
10155 GNOME. These commands will also list active sessions by
10158 * Resource limits (as exposed by the various control group
10159 controllers) can now be controlled dynamically at runtime
10160 for all units. More specifically, you can now use a command
10161 like "systemctl set-cgroup-attr foobar.service cpu.shares
10162 2000" to alter the CPU shares a specific service gets. These
10163 settings are stored persistently on disk, and thus allow the
10164 administrator to easily adjust the resource usage of
10165 services with a few simple commands. This dynamic resource
10166 management logic is also available to other programs via the
10167 bus. Almost any kernel cgroup attribute and controller is
10170 * systemd-vconsole-setup will now copy all font settings to
10171 all allocated VTs, where it previously applied them only to
10174 * libsystemd-login gained the new sd_session_get_tty() API
10177 * This release drops support for a few legacy or
10178 distribution-specific LSB facility names when parsing init
10179 scripts: $x-display-manager, $mail-transfer-agent,
10180 $mail-transport-agent, $mail-transfer-agent, $smtp,
10181 $null. Also, the mail-transfer-agent.target unit backing
10182 this has been removed. Distributions which want to retain
10183 compatibility with this should carry the burden for
10184 supporting this themselves and patch support for these back
10185 in, if they really need to. Also, the facilities $syslog and
10186 $local_fs are now ignored, since systemd does not support
10187 early-boot LSB init scripts anymore, and these facilities
10188 are implied anyway for normal services. syslog.target has
10191 * There are new bus calls on PID1's Manager object for
10192 cancelling jobs, and removing snapshot units. Previously,
10193 both calls were only available on the Job and Snapshot
10194 objects themselves.
10196 * systemd-journal-gatewayd gained SSL support.
10198 * The various "environment" files, such as /etc/locale.conf
10199 now support continuation lines with a backslash ("\") as
10200 last character in the line, similarly in style (but different)
10201 to how this is supported in shells.
10203 * For normal user processes the _SYSTEMD_USER_UNIT= field is
10204 now implicitly appended to every log entry logged. systemctl
10205 has been updated to filter by this field when operating on a
10206 user systemd instance.
10208 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
10209 CAP_AUDIT_CONTROL capabilities to the capabilities set for
10210 the container. This makes it easier to boot unmodified
10211 Fedora systems in a container, which however still requires
10212 audit=0 to be passed on the kernel command line. Auditing in
10213 kernel and userspace is unfortunately still too broken in
10214 context of containers, hence we recommend compiling it out
10215 of the kernel or using audit=0. Hopefully this will be fixed
10216 one day for good in the kernel.
10218 * nspawn gained the new --bind= and --bind-ro= parameters to
10219 bind mount specific directories from the host into the
10222 * nspawn will now mount its own devpts file system instance
10223 into the container, in order not to leak pty devices from
10224 the host into the container.
10226 * systemd will now read the firmware boot time performance
10227 information from the EFI variables, if the used boot loader
10228 supports this, and takes it into account for boot performance
10229 analysis via "systemd-analyze". This is currently supported
10230 only in conjunction with Gummiboot, but could be supported
10231 by other boot loaders too. For details see:
10233 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
10235 * A new generator has been added that automatically mounts the
10236 EFI System Partition (ESP) to /boot, if that directory
10237 exists, is empty, and no other file system has been
10238 configured to be mounted there.
10240 * logind will now send out PrepareForSleep(false) out
10241 unconditionally, after coming back from suspend. This may be
10242 used by applications as asynchronous notification for
10243 system resume events.
10245 * "systemctl unlock-sessions" has been added, that allows
10246 unlocking the screens of all user sessions at once, similar
10247 to how "systemctl lock-sessions" already locked all users
10248 sessions. This is backed by a new D-Bus call UnlockSessions().
10250 * "loginctl seat-status" will now show the master device of a
10251 seat. (i.e. the device of a seat that needs to be around for
10252 the seat to be considered available, usually the graphics
10255 * tmpfiles gained a new "X" line type, that allows
10256 configuration of files and directories (with wildcards) that
10257 shall be excluded from automatic cleanup ("aging").
10259 * udev default rules set the device node permissions now only
10260 at "add" events, and do not change them any longer with a
10261 later "change" event.
10263 * The log messages for lid events and power/sleep keypresses
10264 now carry a message ID.
10266 * We now have a substantially larger unit test suite, but this
10267 continues to be work in progress.
10269 * udevadm hwdb gained a new --root= parameter to change the
10270 root directory to operate relative to.
10272 * logind will now issue a background sync() request to the kernel
10273 early at shutdown, so that dirty buffers are flushed to disk early
10274 instead of at the last moment, in order to optimize shutdown
10277 * A new bootctl tool has been added that is an interface for
10278 certain boot loader operations. This is currently a preview
10279 and is likely to be extended into a small mechanism daemon
10280 like timedated, localed, hostnamed, and can be used by
10281 graphical UIs to enumerate available boot options, and
10282 request boot into firmware operations.
10284 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
10285 the rest of the package. It also has been updated to work
10286 correctly in initrds.
10288 * polkit previously has been runtime optional, and is now also
10289 compile time optional via a configure switch.
10291 * systemd-analyze has been reimplemented in C. Also "systemctl
10292 dot" has moved into systemd-analyze.
10294 * "systemctl status" with no further parameters will now print
10295 the status of all active or failed units.
10297 * Operations such as "systemctl start" can now be executed
10298 with a new mode "--irreversible" which may be used to queue
10299 operations that cannot accidentally be reversed by a later
10300 job queuing. This is by default used to make shutdown
10301 requests more robust.
10303 * The Python API of systemd now gained a new module for
10304 reading journal files.
10306 * A new tool kernel-install has been added that can install
10307 kernel images according to the Boot Loader Specification:
10309 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
10311 * Boot time console output has been improved to provide
10312 animated boot time output for hanging jobs.
10314 * A new tool systemd-activate has been added which can be used
10315 to test socket activation with, directly from the command
10316 line. This should make it much easier to test and debug
10317 socket activation in daemons.
10319 * journalctl gained a new "--reverse" (or -r) option to show
10320 journal output in reverse order (i.e. newest line first).
10322 * journalctl gained a new "--pager-end" (or -e) option to jump
10323 to immediately jump to the end of the journal in the
10324 pager. This is only supported in conjunction with "less".
10326 * journalctl gained a new "--user-unit=" option, that works
10327 similarly to "--unit=" but filters for user units rather than
10330 * A number of unit files to ease adoption of systemd in
10331 initrds has been added. This moves some minimal logic from
10332 the various initrd implementations into systemd proper.
10334 * The journal files are now owned by a new group
10335 "systemd-journal", which exists specifically to allow access
10336 to the journal, and nothing else. Previously, we used the
10337 "adm" group for that, which however possibly covers more
10338 than just journal/log file access. This new group is now
10339 already used by systemd-journal-gatewayd to ensure this
10340 daemon gets access to the journal files and as little else
10341 as possible. Note that "make install" will also set FS ACLs
10342 up for /var/log/journal to give "adm" and "wheel" read
10343 access to it, in addition to "systemd-journal" which owns
10344 the journal files. We recommend that packaging scripts also
10345 add read access to "adm" + "wheel" to /var/log/journal, and
10346 all existing/future journal files. To normal users and
10347 administrators little changes, however packagers need to
10348 ensure to create the "systemd-journal" system group at
10349 package installation time.
10351 * The systemd-journal-gatewayd now runs as unprivileged user
10352 systemd-journal-gateway:systemd-journal-gateway. Packaging
10353 scripts need to create these system user/group at
10356 * timedated now exposes a new boolean property CanNTP that
10357 indicates whether a local NTP service is available or not.
10359 * systemd-detect-virt will now also detect xen PVs
10361 * The pstore file system is now mounted by default, if it is
10364 * In addition to the SELinux and IMA policies we will now also
10365 load SMACK policies at early boot.
10367 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
10368 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
10369 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
10370 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
10371 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
10372 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
10373 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
10374 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
10375 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
10376 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
10377 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
10378 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
10379 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
10380 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
10384 * Timer units now support calendar time events in addition to
10385 monotonic time events. That means you can now trigger a unit
10386 based on a calendar time specification such as "Thu,Fri
10387 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
10388 or fifth day of any month of the year 2013, given that it is
10389 a thursday or friday. This brings timer event support
10390 considerably closer to cron's capabilities. For details on
10391 the supported calendar time specification language see
10394 * udev now supports a number of different naming policies for
10395 network interfaces for predictable names, and a combination
10396 of these policies is now the default. Please see this wiki
10397 document for details:
10399 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
10401 * Auke Kok's bootchart implementation has been added to the
10402 systemd tree. It is an optional component that can graph the
10403 boot in quite some detail. It is one of the best bootchart
10404 implementations around and minimal in its code and
10407 * nss-myhostname has been integrated into the systemd source
10408 tree. nss-myhostname guarantees that the local hostname
10409 always stays resolvable via NSS. It has been a weak
10410 requirement of systemd-hostnamed since a long time, and
10411 since its code is actually trivial we decided to just
10412 include it in systemd's source tree. It can be turned off
10413 with a configure switch.
10415 * The read-ahead logic is now capable of properly detecting
10416 whether a btrfs file system is on SSD or rotating media, in
10417 order to optimize the read-ahead scheme. Previously, it was
10418 only capable of detecting this on traditional file systems
10421 * In udev, additional device properties are now read from the
10422 IAB in addition to the OUI database. Also, Bluetooth company
10423 identities are attached to the devices as well.
10425 * In service files %U may be used as specifier that is
10426 replaced by the configured user name of the service.
10428 * nspawn may now be invoked without a controlling TTY. This
10429 makes it suitable for invocation as its own service. This
10430 may be used to set up a simple containerized server system
10431 using only core OS tools.
10433 * systemd and nspawn can now accept socket file descriptors
10434 when they are started for socket activation. This enables
10435 implementation of socket activated nspawn
10436 containers. i.e. think about autospawning an entire OS image
10437 when the first SSH or HTTP connection is received. We expect
10438 that similar functionality will also be added to libvirt-lxc
10441 * journalctl will now suppress ANSI color codes when
10442 presenting log data.
10444 * systemctl will no longer show control group information for
10445 a unit if the control group is empty anyway.
10447 * logind can now automatically suspend/hibernate/shutdown the
10450 * /etc/machine-info and hostnamed now also expose the chassis
10451 type of the system. This can be used to determine whether
10452 the local system is a laptop, desktop, handset or
10453 tablet. This information may either be configured by the
10454 user/vendor or is automatically determined from ACPI and DMI
10455 information if possible.
10457 * A number of polkit actions are now bound together with "imply"
10458 rules. This should simplify creating UIs because many actions
10459 will now authenticate similar ones as well.
10461 * Unit files learnt a new condition ConditionACPower= which
10462 may be used to conditionalize a unit depending on whether an
10463 AC power source is connected or not, of whether the system
10464 is running on battery power.
10466 * systemctl gained a new "is-failed" verb that may be used in
10467 shell scripts and suchlike to check whether a specific unit
10468 is in the "failed" state.
10470 * The EnvironmentFile= setting in unit files now supports file
10471 globbing, and can hence be used to easily read a number of
10472 environment files at once.
10474 * systemd will no longer detect and recognize specific
10475 distributions. All distribution-specific #ifdeffery has been
10476 removed, systemd is now fully generic and
10477 distribution-agnostic. Effectively, not too much is lost as
10478 a lot of the code is still accessible via explicit configure
10479 switches. However, support for some distribution specific
10480 legacy configuration file formats has been dropped. We
10481 recommend distributions to simply adopt the configuration
10482 files everybody else uses now and convert the old
10483 configuration from packaging scripts. Most distributions
10484 already did that. If that's not possible or desirable,
10485 distributions are welcome to forward port the specific
10486 pieces of code locally from the git history.
10488 * When logging a message about a unit systemd will now always
10489 log the unit name in the message meta data.
10491 * localectl will now also discover system locale data that is
10492 not stored in locale archives, but directly unpacked.
10494 * logind will no longer unconditionally use framebuffer
10495 devices as seat masters, i.e. as devices that are required
10496 to be existing before a seat is considered preset. Instead,
10497 it will now look for all devices that are tagged as
10498 "seat-master" in udev. By default, framebuffer devices will
10499 be marked as such, but depending on local systems, other
10500 devices might be marked as well. This may be used to
10501 integrate graphics cards using closed source drivers (such
10502 as NVidia ones) more nicely into logind. Note however, that
10503 we recommend using the open source NVidia drivers instead,
10504 and no udev rules for the closed-source drivers will be
10505 shipped from us upstream.
10507 Contributions from: Adam Williamson, Alessandro Crismani, Auke
10508 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
10509 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
10510 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
10511 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
10512 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
10513 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
10514 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
10515 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
10516 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
10517 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
10518 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
10523 * udev gained support for loading additional device properties
10524 from an indexed database that is keyed by vendor/product IDs
10525 and similar device identifiers. For the beginning this
10526 "hwdb" is populated with data from the well-known PCI and
10527 USB database, but also includes PNP, ACPI and OID data. In
10528 the longer run this indexed database shall grow into
10529 becoming the one central database for non-essential
10530 userspace device metadata. Previously, data from the PCI/USB
10531 database was only attached to select devices, since the
10532 lookup was a relatively expensive operation due to O(n) time
10533 complexity (with n being the number of entries in the
10534 database). Since this is now O(1), we decided to add in this
10535 data for all devices where this is available, by
10536 default. Note that the indexed database needs to be rebuilt
10537 when new data files are installed. To achieve this you need
10538 to update your packaging scripts to invoke "udevadm hwdb
10539 --update" after installation of hwdb data files. For
10540 RPM-based distributions we introduced the new
10541 %udev_hwdb_update macro for this purpose.
10543 * The Journal gained support for the "Message Catalog", an
10544 indexed database to link up additional information with
10545 journal entries. For further details please check:
10547 https://www.freedesktop.org/wiki/Software/systemd/catalog
10549 The indexed message catalog database also needs to be
10550 rebuilt after installation of message catalog files. Use
10551 "journalctl --update-catalog" for this. For RPM-based
10552 distributions we introduced the %journal_catalog_update
10553 macro for this purpose.
10555 * The Python Journal bindings gained support for the standard
10556 Python logging framework.
10558 * The Journal API gained new functions for checking whether
10559 the underlying file system of a journal file is capable of
10560 properly reporting file change notifications, or whether
10561 applications that want to reflect journal changes "live"
10562 need to recheck journal files continuously in appropriate
10565 * It is now possible to set the "age" field for tmpfiles
10566 entries to 0, indicating that files matching this entry
10567 shall always be removed when the directories are cleaned up.
10569 * coredumpctl gained a new "gdb" verb which invokes gdb
10570 right-away on the selected coredump.
10572 * There's now support for "hybrid sleep" on kernels that
10573 support this, in addition to "suspend" and "hibernate". Use
10574 "systemctl hybrid-sleep" to make use of this.
10576 * logind's HandleSuspendKey= setting (and related settings)
10577 now gained support for a new "lock" setting to simply
10578 request the screen lock on all local sessions, instead of
10579 actually executing a suspend or hibernation.
10581 * systemd will now mount the EFI variables file system by
10584 * Socket units now gained support for configuration of the
10585 SMACK security label.
10587 * timedatectl will now output the time of the last and next
10588 daylight saving change.
10590 * We dropped support for various legacy and distro-specific
10591 concepts, such as insserv, early-boot SysV services
10592 (i.e. those for non-standard runlevels such as 'b' or 'S')
10593 or ArchLinux /etc/rc.conf support. We recommend the
10594 distributions who still need support this to either continue
10595 to maintain the necessary patches downstream, or find a
10596 different solution. (Talk to us if you have questions!)
10598 * Various systemd components will now bypass polkit checks for
10599 root and otherwise handle properly if polkit is not found to
10600 be around. This should fix most issues for polkit-less
10601 systems. Quite frankly this should have been this way since
10602 day one. It is absolutely our intention to make systemd work
10603 fine on polkit-less systems, and we consider it a bug if
10604 something does not work as it should if polkit is not around.
10606 * For embedded systems it is now possible to build udev and
10607 systemd without blkid and/or kmod support.
10609 * "systemctl switch-root" is now capable of switching root
10610 more than once. I.e. in addition to transitions from the
10611 initrd to the host OS it is now possible to transition to
10612 further OS images from the host. This is useful to implement
10613 offline updating tools.
10615 * Various other additions have been made to the RPM macros
10616 shipped with systemd. Use %udev_rules_update() after
10617 installing new udev rules files. %_udevhwdbdir,
10618 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
10619 %_sysctldir are now available which resolve to the right
10620 directories for packages to place various data files in.
10622 * journalctl gained the new --full switch (in addition to
10623 --all, to disable ellipsation for long messages.
10625 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
10626 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
10627 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
10628 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
10629 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
10630 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
10631 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
10632 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
10633 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
10637 * journalctl gained new --since= and --until= switches to
10638 filter by time. It also now supports nice filtering for
10639 units via --unit=/-u.
10641 * Type=oneshot services may use ExecReload= and do the
10644 * The journal daemon now supports time-based rotation and
10645 vacuuming, in addition to the usual disk-space based
10648 * The journal will now index the available field values for
10649 each field name. This enables clients to show pretty drop
10650 downs of available match values when filtering. The bash
10651 completion of journalctl has been updated
10652 accordingly. journalctl gained a new switch -F to list all
10653 values a certain field takes in the journal database.
10655 * More service events are now written as structured messages
10656 to the journal, and made recognizable via message IDs.
10658 * The timedated, localed and hostnamed mini-services which
10659 previously only provided support for changing time, locale
10660 and hostname settings from graphical DEs such as GNOME now
10661 also have a minimal (but very useful) text-based client
10662 utility each. This is probably the nicest way to changing
10663 these settings from the command line now, especially since
10664 it lists available options and is fully integrated with bash
10667 * There's now a new tool "systemd-coredumpctl" to list and
10668 extract coredumps from the journal.
10670 * We now install a README each in /var/log/ and
10671 /etc/rc.d/init.d explaining where the system logs and init
10672 scripts went. This hopefully should help folks who go to
10673 that dirs and look into the otherwise now empty void and
10674 scratch their heads.
10676 * When user-services are invoked (by systemd --user) the
10677 $MANAGERPID env var is set to the PID of systemd.
10679 * SIGRTMIN+24 when sent to a --user instance will now result
10680 in immediate termination of systemd.
10682 * gatewayd received numerous feature additions such as a
10683 "follow" mode, for live syncing and filtering.
10685 * browse.html now allows filtering and showing detailed
10686 information on specific entries. Keyboard navigation and
10687 mouse screen support has been added.
10689 * gatewayd/journalctl now supports HTML5/JSON
10690 Server-Sent-Events as output.
10692 * The SysV init script compatibility logic will now
10693 heuristically determine whether a script supports the
10694 "reload" verb, and only then make this available as
10695 "systemctl reload".
10697 * "systemctl status --follow" has been removed, use "journalctl
10700 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
10701 have been removed since they are hardly useful to be
10704 * And I'd like to take the opportunity to specifically mention
10705 Zbigniew for his great contributions. Zbigniew, you rock!
10707 Contributions from: Andrew Eikum, Christian Hesse, Colin
10708 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
10709 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
10710 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
10711 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
10712 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
10713 Jędrzejewski-Szmek, Сковорода Никита Андреевич
10717 * If /etc/vconsole.conf is non-existent or empty we will no
10718 longer load any console font or key map at boot by
10719 default. Instead the kernel defaults will be left
10720 intact. This is definitely the right thing to do, as no
10721 configuration should mean no configuration, and hard-coding
10722 font names that are different on all archs is probably a bad
10723 idea. Also, the kernel default key map and font should be
10724 good enough for most cases anyway, and mostly identical to
10725 the userspace fonts/key maps we previously overloaded them
10726 with. If distributions want to continue to default to a
10727 non-kernel font or key map they should ship a default
10728 /etc/vconsole.conf with the appropriate contents.
10730 Contributions from: Colin Walters, Daniel J Walsh, Dave
10731 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
10732 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
10736 * journalctl gained a new --cursor= switch to show entries
10737 starting from the specified location in the journal.
10739 * We now enforce a size limit on journal entry fields exported
10740 with "-o json" in journalctl. Fields larger than 4K will be
10741 assigned null. This can be turned off with --all.
10743 * An (optional) journal gateway daemon is now available as
10744 "systemd-journal-gatewayd.service". This service provides
10745 access to the journal via HTTP and JSON. This functionality
10746 will be used to implement live log synchronization in both
10747 pull and push modes, but has various other users too, such
10748 as easy log access for debugging of embedded devices. Right
10749 now it is already useful to retrieve the journal via HTTP:
10751 # systemctl start systemd-journal-gatewayd.service
10752 # wget http://localhost:19531/entries
10754 This will download the journal contents in a
10755 /var/log/messages compatible format. The same as JSON:
10757 # curl -H"Accept: application/json" http://localhost:19531/entries
10759 This service is also accessible via a web browser where a
10760 single static HTML5 app is served that uses the JSON logic
10761 to enable the user to do some basic browsing of the
10762 journal. This will be extended later on. Here's an example
10763 screenshot of this app in its current state:
10765 http://0pointer.de/public/journal-gatewayd
10767 Contributions from: Kay Sievers, Lennart Poettering, Robert
10768 Milasan, Tom Gundersen
10772 * The bash completion logic is now available for journalctl
10775 * We do not mount the "cpuset" controller anymore together with
10776 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
10777 started if no parameters are assigned to it. "cpuset" hence
10778 broke code that assumed it could create "cpu" groups and
10781 * journalctl -f will now subscribe to terminal size changes,
10782 and line break accordingly.
10784 Contributions from: Dave Reisner, Kay Sievers, Lennart
10785 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
10789 * nspawn will now create a symlink /etc/localtime in the
10790 container environment, copying the host's timezone
10791 setting. Previously this has been done via a bind mount, but
10792 since symlinks cannot be bind mounted this has now been
10793 changed to create/update the appropriate symlink.
10795 * journalctl -n's line number argument is now optional, and
10796 will default to 10 if omitted.
10798 * journald will now log the maximum size the journal files may
10799 take up on disk. This is particularly useful if the default
10800 built-in logic of determining this parameter from the file
10801 system size is used. Use "systemctl status
10802 systemd-journald.service" to see this information.
10804 * The multi-seat X wrapper tool has been stripped down. As X
10805 is now capable of enumerating graphics devices via udev in a
10806 seat-aware way the wrapper is not strictly necessary
10807 anymore. A stripped down temporary stop-gap is still shipped
10808 until the upstream display managers have been updated to
10809 fully support the new X logic. Expect this wrapper to be
10810 removed entirely in one of the next releases.
10812 * HandleSleepKey= in logind.conf has been split up into
10813 HandleSuspendKey= and HandleHibernateKey=. The old setting
10814 is not available anymore. X11 and the kernel are
10815 distinguishing between these keys and we should too. This
10816 also means the inhibition lock for these keys has been split
10819 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
10820 Poettering, Lukas Nykryn, Václav Pavlín
10824 * Whenever a unit changes state we will now log this to the
10825 journal and show along the unit's own log output in
10826 "systemctl status".
10828 * ConditionPathIsMountPoint= can now properly detect bind
10829 mount points too. (Previously, a bind mount of one file
10830 system to another place in the same file system could not be
10831 detected as mount, since they shared struct stat's st_dev
10834 * We will now mount the cgroup controllers cpu, cpuacct,
10835 cpuset and the controllers net_cls, net_prio together by
10838 * nspawn containers will now have a virtualized boot
10839 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
10840 over with a randomized ID at container initialization). This
10841 has the effect of making "journalctl -b" do the right thing
10844 * The JSON output journal serialization has been updated not
10845 to generate "endless" list objects anymore, but rather one
10846 JSON object per line. This is more in line how most JSON
10847 parsers expect JSON objects. The new output mode
10848 "json-pretty" has been added to provide similar output, but
10849 neatly aligned for readability by humans.
10851 * We dropped all explicit sync() invocations in the shutdown
10852 code. The kernel does this implicitly anyway in the kernel
10853 reboot() syscall. halt(8)'s -n option is now a compatibility
10856 * We now support virtualized reboot() in containers, as
10857 supported by newer kernels. We will fall back to exit() if
10858 CAP_SYS_REBOOT is not available to the container. Also,
10859 nspawn makes use of this now and will actually reboot the
10860 container if the containerized OS asks for that.
10862 * journalctl will only show local log output by default
10863 now. Use --merge (-m) to show remote log output, too.
10865 * libsystemd-journal gained the new sd_journal_get_usage()
10866 call to determine the current disk usage of all journal
10867 files. This is exposed in the new "journalctl --disk-usage"
10870 * journald gained a new configuration setting SplitMode= in
10871 journald.conf which may be used to control how user journals
10872 are split off. See journald.conf(5) for details.
10874 * A new condition type ConditionFileNotEmpty= has been added.
10876 * tmpfiles' "w" lines now support file globbing, to write
10877 multiple files at once.
10879 * We added Python bindings for the journal submission
10880 APIs. More Python APIs for a number of selected APIs will
10881 likely follow. Note that we intend to add native bindings
10882 only for the Python language, as we consider it common
10883 enough to deserve bindings shipped within systemd. There are
10884 various projects outside of systemd that provide bindings
10885 for languages such as PHP or Lua.
10887 * Many conditions will now resolve specifiers such as %i. In
10888 addition, PathChanged= and related directives of .path units
10889 now support specifiers as well.
10891 * There's now a new RPM macro definition for the system preset
10894 * journald will now warn if it ca not forward a message to the
10895 syslog daemon because its socket is full.
10897 * timedated will no longer write or process /etc/timezone,
10898 except on Debian. As we do not support late mounted /usr
10899 anymore /etc/localtime always being a symlink is now safe,
10900 and hence the information in /etc/timezone is not necessary
10903 * logind will now always reserve one VT for a text getty (VT6
10904 by default). Previously if more than 6 X sessions where
10905 started they took up all the VTs with auto-spawned gettys,
10906 so that no text gettys were available anymore.
10908 * udev will now automatically inform the btrfs kernel logic
10909 about btrfs RAID components showing up. This should make
10910 simple hotplug based btrfs RAID assembly work.
10912 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
10913 (but not for its children which will stay at the kernel
10914 default). This should allow setups with a lot more listening
10917 * systemd will now always pass the configured timezone to the
10918 kernel at boot. timedated will do the same when the timezone
10921 * logind's inhibition logic has been updated. By default,
10922 logind will now handle the lid switch, the power and sleep
10923 keys all the time, even in graphical sessions. If DEs want
10924 to handle these events on their own they should take the new
10925 handle-power-key, handle-sleep-key and handle-lid-switch
10926 inhibitors during their runtime. A simple way to achieve
10927 that is to invoke the DE wrapped in an invocation of:
10929 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
10931 * Access to unit operations is now checked via SELinux taking
10932 the unit file label and client process label into account.
10934 * systemd will now notify the administrator in the journal
10935 when he over-mounts a non-empty directory.
10937 * There are new specifiers that are resolved in unit files,
10938 for the hostname (%H), the machine ID (%m) and the boot ID
10941 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
10942 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
10943 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
10944 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
10945 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
10946 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
10947 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
10951 * Support for reading structured kernel messages from
10952 /dev/kmsg has now been added and is enabled by default.
10954 * Support for reading kernel messages from /proc/kmsg has now
10955 been removed. If you want kernel messages in the journal
10956 make sure to run a recent kernel (>= 3.5) that supports
10957 reading structured messages from /dev/kmsg (see
10958 above). /proc/kmsg is now exclusive property of classic
10959 syslog daemons again.
10961 * The libudev API gained the new
10962 udev_device_new_from_device_id() call.
10964 * The logic for file system namespace (ReadOnlyDirectory=,
10965 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
10966 require pivot_root() anymore. This means fewer temporary
10967 directories are created below /tmp for this feature.
10969 * nspawn containers will now see and receive all submounts
10970 made on the host OS below the root file system of the
10973 * Forward Secure Sealing is now supported for Journal files,
10974 which provide cryptographical sealing of journal files so
10975 that attackers cannot alter log history anymore without this
10976 being detectable. Lennart will soon post a blog story about
10977 this explaining it in more detail.
10979 * There are two new service settings RestartPreventExitStatus=
10980 and SuccessExitStatus= which allow configuration of exit
10981 status (exit code or signal) which will be excepted from the
10982 restart logic, resp. consider successful.
10984 * journalctl gained the new --verify switch that can be used
10985 to check the integrity of the structure of journal files and
10986 (if Forward Secure Sealing is enabled) the contents of
10989 * nspawn containers will now be run with /dev/stdin, /dev/fd/
10990 and similar symlinks pre-created. This makes running shells
10991 as container init process a lot more fun.
10993 * The fstab support can now handle PARTUUID= and PARTLABEL=
10996 * A new ConditionHost= condition has been added to match
10997 against the hostname (with globs) and machine ID. This is
10998 useful for clusters where a single OS image is used to
10999 provision a large number of hosts which shall run slightly
11000 different sets of services.
11002 * Services which hit the restart limit will now be placed in a
11005 Contributions from: Bertram Poettering, Dave Reisner, Huang
11006 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
11007 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
11011 * When running in --user mode systemd will now become a
11012 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
11013 tree a lot more organized.
11015 * A new PartOf= unit dependency type has been introduced that
11016 may be used to group services in a natural way.
11018 * "systemctl enable" may now be used to enable instances of
11021 * journalctl now prints error log levels in red, and
11022 warning/notice log levels in bright white. It also supports
11023 filtering by log level now.
11025 * cgtop gained a new -n switch (similar to top), to configure
11026 the maximum number of iterations to run for. It also gained
11027 -b, to run in batch mode (accepting no input).
11029 * The suffix ".service" may now be omitted on most systemctl
11030 command lines involving service unit names.
11032 * There's a new bus call in logind to lock all sessions, as
11033 well as a loginctl verb for it "lock-sessions".
11035 * libsystemd-logind.so gained a new call sd_journal_perror()
11036 that works similar to libc perror() but logs to the journal
11037 and encodes structured information about the error number.
11039 * /etc/crypttab entries now understand the new keyfile-size=
11042 * shutdown(8) now can send a (configurable) wall message when
11043 a shutdown is cancelled.
11045 * The mount propagation mode for the root file system will now
11046 default to "shared", which is useful to make containers work
11047 nicely out-of-the-box so that they receive new mounts from
11048 the host. This can be undone locally by running "mount
11049 --make-rprivate /" if needed.
11051 * The prefdm.service file has been removed. Distributions
11052 should maintain this unit downstream if they intend to keep
11053 it around. However, we recommend writing normal unit files
11054 for display managers instead.
11056 * Since systemd is a crucial part of the OS we will now
11057 default to a number of compiler switches that improve
11058 security (hardening) such as read-only relocations, stack
11059 protection, and suchlike.
11061 * The TimeoutSec= setting for services is now split into
11062 TimeoutStartSec= and TimeoutStopSec= to allow configuration
11063 of individual time outs for the start and the stop phase of
11066 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
11067 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
11068 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
11069 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
11070 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
11071 Gundersen, Zbigniew Jędrzejewski-Szmek
11075 * The journal and id128 C APIs are now fully documented as man
11078 * Extra safety checks have been added when transitioning from
11079 the initial RAM disk to the main system to avoid accidental
11082 * /etc/crypttab entries now understand the new keyfile-offset=
11085 * systemctl -t can now be used to filter by unit load state.
11087 * The journal C API gained the new sd_journal_wait() call to
11088 make writing synchronous journal clients easier.
11090 * journalctl gained the new -D switch to show journals from a
11091 specific directory.
11093 * journalctl now displays a special marker between log
11094 messages of two different boots.
11096 * The journal is now explicitly flushed to /var via a service
11097 systemd-journal-flush.service, rather than implicitly simply
11098 by seeing /var/log/journal to be writable.
11100 * journalctl (and the journal C APIs) can now match for much
11101 more complex expressions, with alternatives and
11104 * When transitioning from the initial RAM disk to the main
11105 system we will now kill all processes in a killing spree to
11106 ensure no processes stay around by accident.
11108 * Three new specifiers may be used in unit files: %u, %h, %s
11109 resolve to the user name, user home directory resp. user
11110 shell. This is useful for running systemd user instances.
11112 * We now automatically rotate journal files if their data
11113 object hash table gets a fill level > 75%. We also size the
11114 hash table based on the configured maximum file size. This
11115 together should lower hash collisions drastically and thus
11116 speed things up a bit.
11118 * journalctl gained the new "--header" switch to introspect
11119 header data of journal files.
11121 * A new setting SystemCallFilters= has been added to services which may
11122 be used to apply deny lists or allow lists to system calls. This is
11123 based on SECCOMP Mode 2 of Linux 3.5.
11125 * nspawn gained a new --link-journal= switch (and quicker: -j)
11126 to link the container journal with the host. This makes it
11127 very easy to centralize log viewing on the host for all
11128 guests while still keeping the journal files separated.
11130 * Many bugfixes and optimizations
11132 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
11133 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
11134 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
11139 * Several tools now understand kernel command line arguments,
11140 which are only read when run in an initial RAM disk. They
11141 usually follow closely their normal counterparts, but are
11144 * There's a new tool to analyze the readahead files that are
11145 automatically generated at boot. Use:
11147 /usr/lib/systemd/systemd-readahead analyze /.readahead
11149 * We now provide an early debug shell on tty9 if this enabled. Use:
11151 systemctl enable debug-shell.service
11153 * All plymouth related units have been moved into the Plymouth
11154 package. Please make sure to upgrade your Plymouth version
11157 * systemd-tmpfiles now supports getting passed the basename of
11158 a configuration file only, in which case it will look for it
11159 in all appropriate directories automatically.
11161 * udevadm info now takes a /dev or /sys path as argument, and
11162 does the right thing. Example:
11164 udevadm info /dev/sda
11165 udevadm info /sys/class/block/sda
11167 * systemctl now prints a warning if a unit is stopped but a
11168 unit that might trigger it continues to run. Example: a
11169 service is stopped but the socket that activates it is left
11172 * "systemctl status" will now mention if the log output was
11173 shortened due to rotation since a service has been started.
11175 * The journal API now exposes functions to determine the
11176 "cutoff" times due to rotation.
11178 * journald now understands SIGUSR1 and SIGUSR2 for triggering
11179 immediately flushing of runtime logs to /var if possible,
11180 resp. for triggering immediate rotation of the journal
11183 * It is now considered an error if a service is attempted to
11184 be stopped that is not loaded.
11186 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
11188 * systemd-analyze now supports Python 3
11190 * tmpfiles now supports cleaning up directories via aging
11191 where the first level dirs are always kept around but
11192 directories beneath it automatically aged. This is enabled
11193 by prefixing the age field with '~'.
11195 * Seat objects now expose CanGraphical, CanTTY properties
11196 which is required to deal with very fast bootups where the
11197 display manager might be running before the graphics drivers
11198 completed initialization.
11200 * Seat objects now expose a State property.
11202 * We now include RPM macros for service enabling/disabling
11203 based on the preset logic. We recommend RPM based
11204 distributions to make use of these macros if possible. This
11205 makes it simpler to reuse RPM spec files across
11208 * We now make sure that the collected systemd unit name is
11209 always valid when services log to the journal via
11212 * There's a new man page kernel-command-line(7) detailing all
11213 command line options we understand.
11215 * The fstab generator may now be disabled at boot by passing
11216 fstab=0 on the kernel command line.
11218 * A new kernel command line option modules-load= is now understood
11219 to load a specific kernel module statically, early at boot.
11221 * Unit names specified on the systemctl command line are now
11222 automatically escaped as needed. Also, if file system or
11223 device paths are specified they are automatically turned
11224 into the appropriate mount or device unit names. Example:
11226 systemctl status /home
11227 systemctl status /dev/sda
11229 * The SysVConsole= configuration option has been removed from
11230 system.conf parsing.
11232 * The SysV search path is no longer exported on the D-Bus
11235 * The Names= option has been removed from unit file parsing.
11237 * There's a new man page bootup(7) detailing the boot process.
11239 * Every unit and every generator we ship with systemd now
11240 comes with full documentation. The self-explanatory boot is
11243 * A couple of services gained "systemd-" prefixes in their
11244 name if they wrap systemd code, rather than only external
11245 code. Among them fsck@.service which is now
11246 systemd-fsck@.service.
11248 * The HaveWatchdog property has been removed from the D-Bus
11251 * systemd.confirm_spawn= on the kernel command line should now
11254 * There's a new man page crypttab(5) which details all options
11255 we actually understand.
11257 * systemd-nspawn gained a new --capability= switch to pass
11258 additional capabilities to the container.
11260 * timedated will now read known NTP implementation unit names
11261 from /usr/lib/systemd/ntp-units.d/*.list,
11262 systemd-timedated-ntp.target has been removed.
11264 * journalctl gained a new switch "-b" that lists log data of
11265 the current boot only.
11267 * The notify socket is in the abstract namespace again, in
11268 order to support daemons which chroot() at start-up.
11270 * There is a new Storage= configuration option for journald
11271 which allows configuration of where log data should go. This
11272 also provides a way to disable journal logging entirely, so
11273 that data collected is only forwarded to the console, the
11274 kernel log buffer or another syslog implementation.
11276 * Many bugfixes and optimizations
11278 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
11279 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
11280 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
11281 Shawn Landden, Tom Gundersen
11285 * "systemctl help <unit>" now shows the man page if one is
11288 * Several new man pages have been added.
11290 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
11291 MaxLevelConsole= can now be specified in
11292 journald.conf. These options allow reducing the amount of
11293 data stored on disk or forwarded by the log level.
11295 * TimerSlackNSec= can now be specified in system.conf for
11296 PID1. This allows system-wide power savings.
11298 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
11299 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
11304 * logind is now capable of (optionally) handling power and
11305 sleep keys as well as the lid switch.
11307 * journalctl now understands the syntax "journalctl
11308 /usr/bin/avahi-daemon" to get all log output of a specific
11311 * CapabilityBoundingSet= in system.conf now also influences
11312 the capability bound set of usermode helpers of the kernel.
11314 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
11315 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
11316 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
11317 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
11321 * Note that we skipped 139 releases here in order to set the
11322 new version to something that is greater than both udev's
11323 and systemd's most recent version number.
11325 * udev: all udev sources are merged into the systemd source tree now.
11326 All future udev development will happen in the systemd tree. It
11327 is still fully supported to use the udev daemon and tools without
11328 systemd running, like in initramfs or other init systems. Building
11329 udev though, will require the *build* of the systemd tree, but
11330 udev can be properly *run* without systemd.
11332 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
11333 should be used to create dead device nodes as workarounds for broken
11336 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
11337 no longer supported. udev_monitor_new_from_netlink() needs to be
11338 used to subscribe to events.
11340 * udev: when udevd is started by systemd, processes which are left
11341 behind by forking them off of udev rules, are unconditionally cleaned
11342 up and killed now after the event handling has finished. Services or
11343 daemons must be started as systemd services. Services can be
11344 pulled-in by udev to get started, but they can no longer be directly
11345 forked by udev rules.
11347 * udev: the daemon binary is called systemd-udevd now and installed
11348 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
11349 to adapt to that, create symlink, or rename the binary after building
11352 * libudev no longer provides these symbols:
11353 udev_monitor_from_socket()
11354 udev_queue_get_failed_list_entry()
11355 udev_get_{dev,sys,run}_path()
11356 The versions number was bumped and symbol versioning introduced.
11358 * systemd-loginctl and systemd-journalctl have been renamed
11359 to loginctl and journalctl to match systemctl.
11361 * The config files: /etc/systemd/systemd-logind.conf and
11362 /etc/systemd/systemd-journald.conf have been renamed to
11363 logind.conf and journald.conf. Package updates should rename
11364 the files to the new names on upgrade.
11366 * For almost all files the license is now LGPL2.1+, changed
11367 from the previous GPL2.0+. Exceptions are some minor stuff
11368 of udev (which will be changed to LGPL2.1 eventually, too),
11369 and the MIT licensed sd-daemon.[ch] library that is suitable
11370 to be used as drop-in files.
11372 * systemd and logind now handle system sleep states, in
11373 particular suspending and hibernating.
11375 * logind now implements a sleep/shutdown/idle inhibiting logic
11376 suitable for a variety of uses. Soonishly Lennart will blog
11377 about this in more detail.
11379 * var-run.mount and var-lock.mount are no longer provided
11380 (which previously bind mounted these directories to their new
11381 places). Distributions which have not converted these
11382 directories to symlinks should consider stealing these files
11383 from git history and add them downstream.
11385 * We introduced the Documentation= field for units and added
11386 this to all our shipped units. This is useful to make it
11387 easier to explore the boot and the purpose of the various
11390 * All smaller setup units (such as
11391 systemd-vconsole-setup.service) now detect properly if they
11392 are run in a container and are skipped when
11393 appropriate. This guarantees an entirely noise-free boot in
11394 Linux container environments such as systemd-nspawn.
11396 * A framework for implementing offline system updates is now
11397 integrated, for details see:
11398 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
11400 * A new service type Type=idle is available now which helps us
11401 avoiding ugly interleaving of getty output and boot status
11404 * There's now a system-wide CapabilityBoundingSet= option to
11405 globally reduce the set of capabilities for the
11406 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
11407 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
11408 even CAP_NET_ADMIN system-wide for secure systems.
11410 * There are now system-wide DefaultLimitXXX= options to
11411 globally change the defaults of the various resource limits
11412 for all units started by PID 1.
11414 * Harald Hoyer's systemd test suite has been integrated into
11415 systemd which allows easy testing of systemd builds in qemu
11416 and nspawn. (This is really awesome! Ask us for details!)
11418 * The fstab parser is now implemented as generator, not inside
11421 * systemctl will now warn you if .mount units generated from
11422 /etc/fstab are out of date due to changes in fstab that
11423 have not been read by systemd yet.
11425 * systemd is now suitable for usage in initrds. Dracut has
11426 already been updated to make use of this. With this in place
11427 initrds get a slight bit faster but primarily are much
11428 easier to introspect and debug since "systemctl status" in
11429 the host system can be used to introspect initrd services,
11430 and the journal from the initrd is kept around too.
11432 * systemd-delta has been added, a tool to explore differences
11433 between user/admin configuration and vendor defaults.
11435 * PrivateTmp= now affects both /tmp and /var/tmp.
11437 * Boot time status messages are now much prettier and feature
11438 proper english language. Booting up systemd has never been
11441 * Read-ahead pack files now include the inode number of all
11442 files to pre-cache. When the inode changes the pre-caching
11443 is not attempted. This should be nicer to deal with updated
11444 packages which might result in changes of read-ahead
11447 * We now temporaritly lower the kernel's read_ahead_kb variable
11448 when collecting read-ahead data to ensure the kernel's
11449 built-in read-ahead does not add noise to our measurements
11450 of necessary blocks to pre-cache.
11452 * There's now RequiresMountsFor= to add automatic dependencies
11453 for all mounts necessary for a specific file system path.
11455 * MountAuto= and SwapAuto= have been removed from
11456 system.conf. Mounting file systems at boot has to take place
11459 * nspawn now learned a new switch --uuid= to set the machine
11460 ID on the command line.
11462 * nspawn now learned the -b switch to automatically search
11463 for an init system.
11465 * vt102 is now the default TERM for serial TTYs, upgraded from
11468 * systemd-logind now works on VT-less systems.
11470 * The build tree has been reorganized. The individual
11471 components now have directories of their own.
11473 * A new condition type ConditionPathIsReadWrite= is now available.
11475 * nspawn learned the new -C switch to create cgroups for the
11476 container in other hierarchies.
11478 * We now have support for hardware watchdogs, configurable in
11481 * The scheduled shutdown logic now has a public API.
11483 * We now mount /tmp as tmpfs by default, but this can be
11484 masked and /etc/fstab can override it.
11486 * Since udisks does not make use of /media anymore we are not
11487 mounting a tmpfs on it anymore.
11489 * journalctl gained a new --local switch to only interleave
11490 locally generated journal files.
11492 * We can now load the IMA policy at boot automatically.
11494 * The GTK tools have been split off into a systemd-ui.
11496 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
11497 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
11498 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
11499 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
11500 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
11501 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
11502 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
11503 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
11504 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
11509 * This is mostly a bugfix release
11511 * Support optional initialization of the machine ID from the
11512 KVM or container configured UUID.
11514 * Support immediate reboots with "systemctl reboot -ff"
11516 * Show /etc/os-release data in systemd-analyze output
11518 * Many bugfixes for the journal, including endianness fixes and
11519 ensuring that disk space enforcement works
11521 * sd-login.h is C++ compatible again
11523 * Extend the /etc/os-release format on request of the Debian
11526 * We now refuse non-UTF8 strings used in various configuration
11527 and unit files. This is done to ensure we do not pass invalid
11528 data over D-Bus or expose it elsewhere.
11530 * Register Mimo USB Screens as suitable for automatic seat
11533 * Read SELinux client context from journal clients in a race
11536 * Reorder configuration file lookup order. /etc now always
11537 overrides /run in order to allow the administrator to always
11538 and unconditionally override vendor-supplied or
11539 automatically generated data.
11541 * The various user visible bits of the journal now have man
11542 pages. We still lack man pages for the journal API calls
11545 * We now ship all man pages in HTML format again in the
11548 Contributions from: Dave Reisner, Dirk Eibach, Frederic
11549 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
11550 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
11555 * This is mostly a bugfix release
11557 * systems lacking /etc/os-release are no longer supported.
11559 * Various functionality updates to libsystemd-login.so
11561 * Track class of PAM logins to distinguish greeters from
11562 normal user logins.
11564 Contributions from: Kay Sievers, Lennart Poettering, Michael
11569 * This is an important bugfix release for v41.
11571 * Building man pages is now optional which should be useful
11572 for those building systemd from git but unwilling to install
11575 * Watchdog support for supervising services is now usable. In
11576 a future release support for hardware watchdogs
11577 (i.e. /dev/watchdog) will be added building on this.
11579 * Service start rate limiting is now configurable and can be
11580 turned off per service. When a start rate limit is hit a
11581 reboot can automatically be triggered.
11583 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
11585 Contributions from: Benjamin Franzke, Bill Nottingham,
11586 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
11587 Schmidt, Michał Górny, Piotr Drąg
11591 * The systemd binary is installed /usr/lib/systemd/systemd now;
11592 An existing /sbin/init symlink needs to be adapted with the
11595 * The code that loads kernel modules has been ported to invoke
11596 libkmod directly, instead of modprobe. This means we do not
11597 support systems with module-init-tools anymore.
11599 * Watchdog support is now already useful, but still not
11602 * A new kernel command line option systemd.setenv= is
11603 understood to set system wide environment variables
11604 dynamically at boot.
11606 * We now limit the set of capabilities of systemd-journald.
11608 * We now set SIGPIPE to ignore by default, since it only is
11609 useful in shell pipelines, and has little use in general
11610 code. This can be disabled with IgnoreSIPIPE=no in unit
11613 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
11614 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
11619 * This is mostly a bugfix release
11621 * We now expose the reason why a service failed in the
11622 "Result" D-Bus property.
11624 * Rudimentary service watchdog support (will be completed over
11625 the next few releases.)
11627 * When systemd forks off in order execute some service we will
11628 now immediately changes its argv[0] to reflect which process
11629 it will execute. This is useful to minimize the time window
11630 with a generic argv[0], which makes bootcharts more useful
11632 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
11633 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
11634 Mike Kazantsev, Ray Strode
11638 * This is mostly a test release, but incorporates many
11641 * New systemd-cgtop tool to show control groups by their
11644 * Linking against libacl for ACLs is optional again. If
11645 disabled, support tracking device access for active logins
11646 goes becomes unavailable, and so does access to the user
11647 journals by the respective users.
11649 * If a group "adm" exists, journal files are automatically
11650 owned by them, thus allow members of this group full access
11651 to the system journal as well as all user journals.
11653 * The journal now stores the SELinux context of the logging
11654 client for all entries.
11656 * Add C++ inclusion guards to all public headers
11658 * New output mode "cat" in the journal to print only text
11659 messages, without any meta data like date or time.
11661 * Include tiny X server wrapper as a temporary stop-gap to
11662 teach XOrg udev display enumeration. This is used by display
11663 managers such as gdm, and will go away as soon as XOrg
11664 learned native udev hotplugging for display devices.
11666 * Add new systemd-cat tool for executing arbitrary programs
11667 with STDERR/STDOUT connected to the journal. Can also act as
11668 BSD logger replacement, and does so by default.
11670 * Optionally store all locally generated coredumps in the
11671 journal along with meta data.
11673 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
11674 writing short strings to files (for usage for /sys), and for
11675 creating symlinks, character and block device nodes.
11677 * New unit file option ControlGroupPersistent= to make cgroups
11678 persistent, following the mechanisms outlined in
11679 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
11681 * Support multiple local RTCs in a sane way
11683 * No longer monopolize IO when replaying readahead data on
11684 rotating disks, since we might starve non-file-system IO to
11685 death, since fanotify() will not see accesses done by blkid,
11688 * Do not show kernel threads in systemd-cgls anymore, unless
11689 requested with new -k switch.
11691 Contributions from: Dan Horák, Kay Sievers, Lennart
11692 Poettering, Michal Schmidt
11696 * This is mostly a test release, but incorporates many
11699 * The git repository moved to:
11700 git://anongit.freedesktop.org/systemd/systemd
11701 ssh://git.freedesktop.org/git/systemd/systemd
11703 * First release with the journal
11704 http://0pointer.de/blog/projects/the-journal.html
11706 * The journal replaces both systemd-kmsg-syslogd and
11707 systemd-stdout-bridge.
11709 * New sd_pid_get_unit() API call in libsystemd-logind
11711 * Many systemadm clean-ups
11713 * Introduce remote-fs-pre.target which is ordered before all
11714 remote mounts and may be used to start services before all
11717 * Added Mageia support
11719 * Add bash completion for systemd-loginctl
11721 * Actively monitor PID file creation for daemons which exit in
11722 the parent process before having finished writing the PID
11723 file in the daemon process. Daemons which do this need to be
11724 fixed (i.e. PID file creation must have finished before the
11725 parent exits), but we now react a bit more gracefully to them.
11727 * Add colourful boot output, mimicking the well-known output
11728 of existing distributions.
11730 * New option PassCredentials= for socket units, for
11731 compatibility with a recent kernel ABI breakage.
11733 * /etc/rc.local is now hooked in via a generator binary, and
11734 thus will no longer act as synchronization point during
11737 * systemctl list-unit-files now supports --root=.
11739 * systemd-tmpfiles now understands two new commands: z, Z for
11740 relabelling files according to the SELinux database. This is
11741 useful to apply SELinux labels to specific files in /sys,
11742 among other things.
11744 * Output of SysV services is now forwarded to both the console
11745 and the journal by default, not only just the console.
11747 * New man pages for all APIs from libsystemd-login.
11749 * The build tree got reorganized and the build system is a
11750 lot more modular allowing embedded setups to specifically
11751 select the components of systemd they are interested in.
11753 * Support for Linux systems lacking the kernel VT subsystem is
11756 * configure's --with-rootdir= got renamed to
11757 --with-rootprefix= to follow the naming used by udev and
11760 * Unless specified otherwise we will now install to /usr instead
11761 of /usr/local by default.
11763 * Processes with '@' in argv[0][0] are now excluded from the
11764 final shut-down killing spree, following the logic explained
11766 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
11768 * All processes remaining in a service cgroup when we enter
11769 the START or START_PRE states are now killed with
11770 SIGKILL. That means it is no longer possible to spawn
11771 background processes from ExecStart= lines (which was never
11772 supported anyway, and bad style).
11774 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
11775 reloading of units together.
11777 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
11778 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
11779 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
11780 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
11781 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek