1 systemd System and Service Manager
5 * A concept of system extension images is introduced. Such images may
6 be used to extend the /usr/ and /opt/ directory hierarchies at
7 runtime with additional files (even if the file system is read-only).
8 When a system extension image is activated, its /usr/ and /opt/
9 hierarchies and os-release information are combined via overlayfs
10 with the file system hierarchy of the host OS.
12 A new systemd-sysext tool can be used to merge, unmerge, list, and
13 refresh system extension hierarchies. See
14 https://www.freedesktop.org/software/systemd/man/systemd-sysext.html.
16 The systemd-sysext.service automatically merges installed system
17 extensions during boot (before basic.target, but not in very early
18 boot, since various file systems have to be mounted first).
20 The SYSEXT_LEVEL= field in os-release(5) may be used to specify the
21 supported system extension level.
23 * A new configuration file /etc/veritytab may be used to configure
24 integrity protection for block devices. Each line is in the format
25 "volume-name data-device hash-device roothash options".
27 * A new kernel command-line option systemd.verity.root-options= may be
28 used to configure dm-verity behaviour for the root device.
30 * The key file specified in /etc/crypttab (the third field) may now
31 refer to an AF_UNIX/SOCK_STREAM socket in the file system. The key is
32 acquired by connecting to that socket and reading from it. This
33 allows the implementation of a service to provide key information
34 dynamically, at the moment when it is needed.
36 * Support has been added to systemd-cryptsetup for extracting the
37 PKCS#11 token URI and encrypted key from the LUKS2 JSON embedded
38 metadata header. This allows the information how to open the
39 encrypted device to be embedded directly in the device and obviates
40 the need for configuration in an external file.
42 * systemd-cryptsetup gained support for unlocking LUKS2 volumes using
43 TPM2 hardware, as well as FIDO2 security tokens.
45 * systemd-repart may lock partitions using TPM2 hardware. This may be
46 useful for example to create an encrypted /var partition bound to the
47 machine on first boot.
49 * A new systemd-cryptenroll tool has been added to enroll TPM2, FIDO2
50 and PKCS#11 security tokens to LUKS volumes, list and destroy
53 http://0pointer.net/blog/unlocking-luks2-volumes-with-tpm2-fido2-pkcs11-security-hardware-on-systemd-248.html
55 It also supports enrolling "recovery keys" and regular passphrases.
57 * The libfido2 dependency is now based on dlopen(), so that the library
58 is used at runtime when installed, but not if not.
60 * systemd-cryptsetup gained support for two new options in
61 /etc/crypttab: no-write-workqueue and no-read-workqueue which request
62 synchronous processing of encryption/decryption IO.
64 * The manager may be configured at compile time to use fexecve instead
65 of execve when spawning children. Using fexecve closes a window
66 between checking the security context of an executable and spawning
67 it, but unfortunately the kernel displays stale information in the
68 comm field, which impacts ps output and such.
70 * The configuration option -Dcompat-gateway-hostname has been dropped.
71 "_gateway" is now the only supported name.
73 * The ConditionSecurity=tpm2 unit file setting may be used to check if
74 the system has at least one TPM2 (tpmrm class) device.
76 * The tables of system calls in seccomps filters are now automatically
77 generated from kernel lists exported on
78 https://fedora.juszkiewicz.com.pl/syscalls.html.
80 The following architectures should now have complete lists:
81 alpha, arc, arm64, arm, i386, ia64, m68k, mips64n32, mips64, mipso32,
82 powerpc, powerpc64, s390, s390x, tilegx, sparc, x86_64, x32.
84 * The MountAPIVFS= service file setting now additionally mounts a tmpfs
85 on /run/ if it is not already a mount point. A writable /run/ has
86 always been a requirement for a functioning system, but this was not
87 guaranteed when using a read-only image.
89 Users can always specify BindPaths= or InaccessiblePaths= as
90 overrides, and they will take precedence. If the host's root mount
91 point is used, there is no change in behaviour.
93 * New bind mounts and file system image mounts may be injected into the
94 mount namespace of a service (without restarting it). This is exposed
95 respectively as 'systemctl bind <unit> <path>…' and
96 'systemctl mount-image <unit> <image>…'.
98 * The StandardOuput= and StandardError= settings can now specify files
99 to be truncated for output (as "truncate:<path>").
101 * The ExecPaths= and NoExecPaths= settings may be used to specify
102 noexec for parts of the file system.
104 * sd-bus has a new function sd_bus_open_user_machine() to open a
105 connection to the session bus of a specific user in a local container
106 or on the local host. This is exposed in the -M switch to systemctl
109 systemctl --user -M lennart@foobar start foo
111 This will connect to the user bus of a user "lennart" in container
112 "foobar". If no container name is specified, the specified user on
113 the host itself is connected to
115 systemctl --user -M lennart@ start quux
117 * sd-bus also gained a convenience function sd_bus_reply() to call
118 sd_bus_send() with an existing reply message.
120 * sd-event allows rate limits to be set on event sources, for dealing
121 with high-priority event sources that might starve out others. See
122 the new man page sd_event_source_set_ratelimit(3) for details.
124 * systemd.link files gained a [Link] Promiscuous= switch, which allows
125 the device to be raised in promiscuous mode.
127 New [Link] TransmitQueues= and ReceiveQueues= settings allow the
128 number of TX and RX queues to be configured.
130 New [Link] TransmitQueueLength= setting allows the size of the TX
131 queue to be configured.
133 New [Link] GenericSegmentOffloadMaxBytes= and
134 GenericSegmentOffloadMaxSegments= allow capping the packet size and
135 the number of segments accepted in Generic Segment Offload.
137 * systemd.network files gained a [Network] RouteTable= configuration
138 switch to select the routing policy table.
140 systemd.network files gained a [RoutingPolicyRule] Type=
141 configuration switch (one of "blackhole, "unreachable", "prohibit").
143 systemd.network files gained a [IPv6AcceptRA] RouteDenyList= and
144 RouteAllowList= settings to ignore/accept route advertisements from
145 routers matching specified prefixes. The DenyList= setting has been
146 renamed to PrefixDenyList= and a new PrefixAllowList= option has been
149 systemd.network files gained a [DHCPv6] UseAddress= setting to
150 optionally ignore the address provided in the lease.
152 systemd.network files gained a [DHCPv6PrefixDelegation]
153 ManageTemporaryAddress= switch.
155 .network files gained a new ActivationPolicy= setting which allows
156 configuration how the UP state of an interface shall be managed,
157 i.e. whether the interface is always upped, always downed, or may be
158 upped/downed by the user using "ip dev".
160 * systemd.netdev files gained a [VLAN] Protocol=, IngressQOSMaps=,
161 EgressQOSMaps=, and [MACVLAN] BroadcastMulticastQueueLength=
162 configuration options for VLAN packet handling.
164 * udev rules may now set log_level= option. This allows debug logs to
165 be enabled for select events, e.g. just for a specific subsystem or
166 even a single device.
168 * udev now exports the VOLUME_ID, LOGICAL_VOLUME_ID, VOLUME_SET_ID, and
169 DATA_PREPARED_ID attributes for block devices (when available).
171 * udev now exports decoded DMI information about used memory slots as
172 device properties under the /sys/class/dmi/id/ pseudo device.
174 * /dev/ is not mounted noexec any more. This didn't provide any
175 significant security benefits and would conflicts with the executable
176 mappings used with /dev/sgx device nodes.
178 * Permissions for /dev/vsock are now set to 0o666, and /dev/vhost-vsock
179 and /dev/vhost-net are owned by the kvm group.
181 * The hardware database has been extended with a list of fingerprint
182 readers that correctly support autosuspend using data from libfprint.
184 * systemd-resolved can now answer DNSSEC questions through the stub
185 resolver interface in a way that allows local clients to do DNSSEC
186 validation themselves. For a question with DO+CD set, it'll proxy the
187 DNS query and respond with a mostly unmodified packet received from
190 * systemd-resolved learnt a new boolean option CacheFromLocalhost= in
191 resolved.conf. If true the service will provide caching even for DNS
192 lookups made to an upstream DNS server on the 127.0.0.1/::1
193 addresses. By default (and when the option is false) systemd-resolved
194 will not cache such lookups, in order to avoid duplicate local
195 caching, under the assumption the local upstream server caches
198 * systemd-nspawn gained a new --ambient-capability= setting
199 (AmbientCapability= in .nspawn files) to configure ambient
200 capabilities passed to the container payload.
202 * systemd-nspawn gained the ability to configure the firewall using the
203 nftables subsystem (in addition to the existing iptables
204 support). Similar, systemd-networkd's IPMasquerade= option now
205 supports nftables as backend, too. In both cases NAT on IPv6 is now
206 supported too, in addition to IPv4 (the iptables backend still is
209 * systemd-importd will now download .verity and .roothash.p7s files
210 along with the machine image (as exposed via machinectl pull-raw).
212 * systemd-oomd now gained a new DefaultMemoryPressureDurationSec=
213 setting to configure the time a unit's cgroup needs to exceed memory
214 pressure limits before action will be taken.
216 systemd-oomd is now considered fully supported (the usual
217 backwards-compatiblity promises apply). Swap is not required for
218 operation, but it is still recommended.
220 * systemd-timesyncd gained a new ConnectionRetrySec= setting which
221 configures the retry delay when trying to contact servers.
223 * systemd-stdio-bridge gained --system/--user options to connect to the
224 system bus (previous default) or the user session bus.
226 * When the hostname is set to "localhost", systemd-hostnamed will
227 accept this. Previously such a setting would be mostly silently
228 ignored. The goal is to honour configuration as specified by the
231 * systemd-hostnamed now exports the fallback hostname and the source of
232 the configured hostname ("static", "transient", or "fallback") as
235 * systemd-hostnamed now exports the "HardwareVendor" and
236 "HardwareModel" D-Bus properties, which are supposed to contain a
237 pair of cleaned up, human readable strings describing the system
238 vendor and model. It's typically source from the firmware's DMI
239 tables, but may be augmented from a new hwdb database. hostnamectl
240 shows this in the status output.
242 * systemd-localed may now call locale-gen to generate missing locales
243 on-demand (UTF-8-only). This improves integration with Debian-based
244 distributions (Debian/Ubuntu/PureOS/Tanglu/...) and Arch Linux.
246 * systemctl --check-inhibitors=true may now be used to obey inhibitors
247 even when invoked non-interactively. The old --ignore-inhibitors
248 switch is now deprecated and replaced by --check-inhibitors=false.
250 * systemctl import-environment will now emit a warning when called
251 without any arguments (i.e. to import the full environment block of
252 the called program). This command will usually be invoked from a
253 shell, which means that it'll inherit a bunch of variables which are
254 specific to that shell, and usually to the TTY the shell is connected
255 to, and don't have any meaning in the global context of the system or
256 user service manager. Instead, only specific variables should be
257 imported into the manager environment block.
259 Similarly, programs which update the manager environment block by
260 directly calling the D-Bus API of the manager, should also push
261 specific variables, and not the full inherited environment.
263 * systemctl's status output now shows unit state with a more careful
264 selection of Unicode characters: units in maintenance show a "○"
265 symbol instead of the usual "●", failed units show "×", and services
268 * coredumpctl gained a --debugger-arguments= switch to pass arguments
271 * networkctl now shows the link activation policy in status.
273 * Various tools gained --pager/--no-pager/--json switches to
274 enable/disable the pager and provide JSON output.
276 * Various tools now accept SYSTEMD_COLORS=16|256 to configure how
277 many terminal colours are used in output.
279 * less 568 or newer is now required. Hyperlink ANSI sequences in
280 terminal output are now used even if a pager is used, and older
281 versions of less are not able to display these sequences
282 correctly. SYSTEMD_URLIFY=0 may be used to disable it.
284 * Builds with support for separate / and /usr hierarchies (split-usr
285 builds, non-merged-usr builds) are now officially deprecated. A
286 warning is emitted during build. Support is slated to be removed in
287 about a year (when the Debian Bookworm release development starts).
289 * The main development branch has been renamed to 'main'.
291 * mmcblk[0-9]boot[0-9] devices will no longer be probed automatically
292 for partitions, as in the vast majority of cases they contain none
293 and are used internally by the bootloader (eg: uboot).
295 * systemd will now set the $SYSTEMD_EXEC_PID environment variable for
296 spawned processes to the PID of the process itself. This may be used
297 by programs for detecting whether they were forked off by the service
298 manager itself or are a process forked off further down the tree.
302 * KERNEL API INCOMPATIBILITY: Linux 4.14 introduced two new uevents
303 "bind" and "unbind" to the Linux device model. When this kernel
304 change was made, systemd-udevd was only minimally updated to handle
305 and propagate these new event types. The introduction of these new
306 uevents (which are typically generated for USB devices and devices
307 needing a firmware upload before being functional) resulted in a
308 number of issues which we so far didn't address. We hoped the kernel
309 maintainers would themselves address these issues in some form, but
310 that did not happen. To handle them properly, many (if not most) udev
311 rules files shipped in various packages need updating, and so do many
312 programs that monitor or enumerate devices with libudev or sd-device,
313 or otherwise process uevents. Please note that this incompatibility
314 is not fault of systemd or udev, but caused by an incompatible kernel
315 change that happened back in Linux 4.14, but is becoming more and
316 more visible as the new uevents are generated by more kernel drivers.
318 To minimize issues resulting from this kernel change (but not avoid
319 them entirely) starting with systemd-udevd 247 the udev "tags"
320 concept (which is a concept for marking and filtering devices during
321 enumeration and monitoring) has been reworked: udev tags are now
322 "sticky", meaning that once a tag is assigned to a device it will not
323 be removed from the device again until the device itself is removed
324 (i.e. unplugged). This makes sure that any application monitoring
325 devices that match a specific tag is guaranteed to both see uevents
326 where the device starts being relevant, and those where it stops
327 being relevant (the latter now regularly happening due to the new
328 "unbind" uevent type). The udev tags concept is hence now a concept
329 tied to a *device* instead of a device *event* — unlike for example
330 udev properties whose lifecycle (as before) is generally tied to a
331 device event, meaning that the previously determined properties are
332 forgotten whenever a new uevent is processed.
334 With the newly redefined udev tags concept, sometimes it's necessary
335 to determine which tags are the ones applied by the most recent
336 uevent/database update, in order to discern them from those
337 originating from earlier uevents/database updates of the same
338 device. To accommodate for this a new automatic property CURRENT_TAGS
339 has been added that works similar to the existing TAGS property but
340 only lists tags set by the most recent uevent/database
341 update. Similarly, the libudev/sd-device API has been updated with
342 new functions to enumerate these 'current' tags, in addition to the
343 existing APIs that now enumerate the 'sticky' ones.
345 To properly handle "bind"/"unbind" on Linux 4.14 and newer it is
346 essential that all udev rules files and applications are updated to
347 handle the new events. Specifically:
349 • All rule files that currently use a header guard similar to
350 ACTION!="add|change",GOTO="xyz_end" should be updated to use
351 ACTION=="remove",GOTO="xyz_end" instead, so that the
352 properties/tags they add are also applied whenever "bind" (or
353 "unbind") is seen. (This is most important for all physical device
354 types — those for which "bind" and "unbind" are currently
355 generated, for all other device types this change is still
356 recommended but not as important — but certainly prepares for
357 future kernel uevent type additions).
359 • Similarly, all code monitoring devices that contains an 'if' branch
360 discerning the "add" + "change" uevent actions from all other
361 uevents actions (i.e. considering devices only relevant after "add"
362 or "change", and irrelevant on all other events) should be reworked
363 to instead negatively check for "remove" only (i.e. considering
364 devices relevant after all event types, except for "remove", which
365 invalidates the device). Note that this also means that devices
366 should be considered relevant on "unbind", even though conceptually
367 this — in some form — invalidates the device. Since the precise
368 effect of "unbind" is not generically defined, devices should be
369 considered relevant even after "unbind", however I/O errors
370 accessing the device should then be handled gracefully.
372 • Any code that uses device tags for deciding whether a device is
373 relevant or not most likely needs to be updated to use the new
374 udev_device_has_current_tag() API (or sd_device_has_current_tag()
375 in case sd-device is used), to check whether the tag is set at the
376 moment an uevent is seen (as opposed to the existing
377 udev_device_has_tag() API which checks if the tag ever existed on
378 the device, following the API concept redefinition explained
381 We are very sorry for this breakage and the requirement to update
382 packages using these interfaces. We'd again like to underline that
383 this is not caused by systemd/udev changes, but result of a kernel
386 * UPCOMING INCOMPATIBILITY: So far most downstream distribution
387 packages have not retriggered devices once the udev package (or any
388 auxiliary package installing additional udev rules) is updated. We
389 intend to work with major distributions to change this, so that
390 "udevadm trigger -a change" is issued on such upgrades, ensuring that
391 the updated ruleset is applied to the devices already discovered, so
392 that (asynchronously) after the upgrade completed the udev database
393 is consistent with the updated rule set. This means udev rules must
394 be ready to be retriggered with a "change" action any time, and
395 result in correct and complete udev database entries. While the
396 majority of udev rule files known to us currently get this right,
397 some don't. Specifically, there are udev rules files included in
398 various packages that only set udev properties on the "add" action,
399 but do not handle the "change" action. If a device matching those
400 rules is retriggered with the "change" action (as is intended here)
401 it would suddenly lose the relevant properties. This always has been
402 problematic, but as soon as all udev devices are triggered on relevant
403 package upgrades this will become particularly so. It is strongly
404 recommended to fix offending rules so that they can handle a "change"
405 action at any time, and acquire all necessary udev properties even
406 then. Or in other words: the header guard mentioned above
407 (ACTION=="remove",GOTO="xyz_end") is the correct approach to handle
408 this, as it makes sure rules are rerun on "change" correctly, and
409 accumulate the correct and complete set of udev properties. udev rule
410 definitions that cannot handle "change" events being triggered at
411 arbitrary times should be considered buggy.
413 * The MountAPIVFS= service file setting now defaults to on if
414 RootImage= and RootDirectory= are used, which means that with those
415 two settings /proc/, /sys/ and /dev/ are automatically properly set
416 up for services. Previous behaviour may be restored by explicitly
417 setting MountAPIVFS=off.
419 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
420 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
421 latter it takes precedence over the former, similar to how most of
422 systemd's own configuration is handled. Given that PAM stack
423 definitions are primarily put together by OS vendors/distributions
424 (though possibly overridden by users), this systemd release moves its
425 own PAM stack configuration for the "systemd-user" PAM service (i.e.
426 for the PAM session invoked by the per-user user@.service instance)
427 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
428 packages' vendor versions of their PAM stack definitions from
429 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
430 desired the location to which systemd installs its PAM stack
431 configuration may be changed via the -Dpamconfdir Meson option.
433 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
434 libpwquality and libcryptsetup have been changed to be based on
435 dlopen(): instead of regular dynamic library dependencies declared in
436 the binary ELF headers, these libraries are now loaded on demand
437 only, if they are available. If the libraries cannot be found the
438 relevant operations will fail gracefully, or a suitable fallback
439 logic is chosen. This is supposed to be useful for general purpose
440 distributions, as it allows minimizing the list of dependencies the
441 systemd packages pull in, permitting building of more minimal OS
442 images, while still making use of these "weak" dependencies should
443 they be installed. Since many package managers automatically
444 synthesize package dependencies from ELF shared library dependencies,
445 some additional manual packaging work has to be done now to replace
446 those (slightly downgraded from "required" to "recommended" or
447 whatever is conceptually suitable for the package manager). Note that
448 this change does not alter build-time behaviour: as before the
449 build-time dependencies have to be installed during build, even if
450 they now are optional during runtime.
452 * sd-event.h gained a new call sd_event_add_time_relative() for
453 installing timers relative to the current time. This is mostly a
454 convenience wrapper around the pre-existing sd_event_add_time() call
455 which installs absolute timers.
457 * sd-event event sources may now be placed in a new "exit-on-failure"
458 mode, which may be controlled via the new
459 sd_event_source_get_exit_on_failure() and
460 sd_event_source_set_exit_on_failure() functions. If enabled, any
461 failure returned by the event source handler functions will result in
462 exiting the event loop (unlike the default behaviour of just
463 disabling the event source but continuing with the event loop). This
464 feature is useful to set for all event sources that define "primary"
465 program behaviour (where failure should be fatal) in contrast to
466 "auxiliary" behaviour (where failure should remain local).
468 * Most event source types sd-event supports now accept a NULL handler
469 function, in which case the event loop is exited once the event
470 source is to be dispatched, using the userdata pointer — converted to
471 a signed integer — as exit code of the event loop. Previously this
472 was supported for IO and signal event sources already. Exit event
473 sources still do not support this (simply because it makes little
474 sense there, as the event loop is already exiting when they are
477 * A new per-unit setting RootImageOptions= has been added which allows
478 tweaking the mount options for any file system mounted as effect of
479 the RootImage= setting.
481 * Another new per-unit setting MountImages= has been added, that allows
482 mounting additional disk images into the file system tree accessible
485 * Timer units gained a new FixedRandomDelay= boolean setting. If
486 enabled, the random delay configured with RandomizedDelaySec= is
487 selected in a way that is stable on a given system (though still
488 different for different units).
490 * Socket units gained a new setting Timestamping= that takes "us", "ns"
491 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
494 * systemd-repart now generates JSON output when requested with the new
497 * systemd-machined's OpenMachineShell() bus call will now pass
498 additional policy metadata data fields to the PolicyKit
499 authentication request.
501 * systemd-tmpfiles gained a new -E switch, which is equivalent to
502 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
503 --exclude=/sys. It's particularly useful in combination with --root=,
504 when operating on OS trees that do not have any of these four runtime
505 directories mounted, as this means no files below these subtrees are
506 created or modified, since those mount points should probably remain
509 * systemd-tmpfiles gained a new --image= switch which is like --root=,
510 but takes a disk image instead of a directory as argument. The
511 specified disk image is mounted inside a temporary mount namespace
512 and the tmpfiles.d/ drop-ins stored in the image are executed and
513 applied to the image. systemd-sysusers similarly gained a new
514 --image= switch, that allows the sysusers.d/ drop-ins stored in the
515 image to be applied onto the image.
517 * Similarly, the journalctl command also gained an --image= switch,
518 which is a quick one-step solution to look at the log data included
521 * journalctl's --output=cat option (which outputs the log content
522 without any metadata, just the pure text messages) will now make use
523 of terminal colors when run on a suitable terminal, similarly to the
526 * JSON group records now support a "description" string that may be
527 used to add a human-readable textual description to such groups. This
528 is supposed to match the user's GECOS field which traditionally
529 didn't have a counterpart for group records.
531 * The "systemd-dissect" tool that may be used to inspect OS disk images
532 and that was previously installed to /usr/lib/systemd/ has now been
533 moved to /usr/bin/, reflecting its updated status of an officially
534 supported tool with a stable interface. It gained support for a new
535 --mkdir switch which when combined with --mount has the effect of
536 creating the directory to mount the image to if it is missing
537 first. It also gained two new commands --copy-from and --copy-to for
538 copying files and directories in and out of an OS image without the
539 need to manually mount it. It also acquired support for a new option
540 --json= to generate JSON output when inspecting an OS image.
542 * The cgroup2 file system is now mounted with the
543 "memory_recursiveprot" mount option, supported since kernel 5.7. This
544 means that the MemoryLow= and MemoryMin= unit file settings now apply
545 recursively to whole subtrees.
547 * systemd-homed now defaults to using the btrfs file system — if
548 available — when creating home directories in LUKS volumes. This may
549 be changed with the DefaultFileSystemType= setting in homed.conf.
550 It's now the default file system in various major distributions and
551 has the major benefit for homed that it can be grown and shrunk while
552 mounted, unlike the other contenders ext4 and xfs, which can both be
553 grown online, but not shrunk (in fact xfs is the technically most
554 limited option here, as it cannot be shrunk at all).
556 * JSON user records managed by systemd-homed gained support for
557 "recovery keys". These are basically secondary passphrases that can
558 unlock user accounts/home directories. They are computer-generated
559 rather than user-chosen, and typically have greater entropy.
560 homectl's --recovery-key= option may be used to add a recovery key to
561 a user account. The generated recovery key is displayed as a QR code,
562 so that it can be scanned to be kept in a safe place. This feature is
563 particularly useful in combination with systemd-homed's support for
564 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
565 security tokens are lost. Recovery keys may be entered wherever the
566 system asks for a password.
568 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
569 home directory which indicates that a home directory has not been
570 deactivated cleanly when offline. This flag is useful to identify
571 home directories for which the offline discard logic did not run when
572 offlining, and where it would be a good idea to log in again to catch
575 * systemctl gained a new parameter --timestamp= which may be used to
576 change the style in which timestamps are output, i.e. whether to show
577 them in local timezone or UTC, or whether to show µs granularity.
579 * Alibaba's "pouch" container manager is now detected by
580 systemd-detect-virt, ConditionVirtualization= and similar
581 constructs. Similar, they now also recognize IBM PowerVM machine
584 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
585 place to use for propagating external mounts into the
586 container. Similarly /run/host/notify is now used as the socket path
587 for container payloads to communicate with the container manager
588 using sd_notify(). The container manager now uses the
589 /run/host/inaccessible/ directory to place "inaccessible" file nodes
590 of all relevant types which may be used by the container payload as
591 bind mount source to over-mount inodes to make them inaccessible.
592 /run/host/container-manager will now be initialized with the same
593 string as the $container environment variable passed to the
594 container's PID 1. /run/host/container-uuid will be initialized with
595 the same string as $container_uuid. This means the /run/host/
596 hierarchy is now the primary way to make host resources available to
597 the container. The Container Interface documents these new files and
600 https://systemd.io/CONTAINER_INTERFACE
602 * Support for the "ConditionNull=" unit file condition has been
603 deprecated and undocumented for 6 years. systemd started to warn
604 about its use 1.5 years ago. It has now been removed entirely.
606 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
607 a sd_bus_error struct and a list of error names, and checks if the
608 error matches one of these names. It's a convenience wrapper that is
609 useful in cases where multiple errors shall be handled the same way.
611 * A new system call filter list "@known" has been added, that contains
612 all system calls known at the time systemd was built.
614 * Behaviour of system call filter allow lists has changed slightly:
615 system calls that are contained in @known will result in a EPERM by
616 default, while those not contained in it result in ENOSYS. This
617 should improve compatibility because known system calls will thus be
618 communicated as prohibited, while unknown (and thus newer ones) will
619 be communicated as not implemented, which hopefully has the greatest
620 chance of triggering the right fallback code paths in client
623 * "systemd-analyze syscall-filter" will now show two separate sections
624 at the bottom of the output: system calls known during systemd build
625 time but not included in any of the filter groups shown above, and
626 system calls defined on the local kernel but known during systemd
629 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
630 systemd-nspawn all system call filter violations will be logged by
631 the kernel (audit). This is useful for tracking down system calls
632 invoked by container payloads that are prohibited by the container's
633 system call filter policy.
635 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
636 systemd-nspawn (and other programs that use seccomp) all seccomp
637 filtering is turned off.
639 * Two new unit file settings ProtectProc= and ProcSubset= have been
640 added that expose the hidepid= and subset= mount options of procfs.
641 All processes of the unit will only see processes in /proc that are
642 are owned by the unit's user. This is an important new sandboxing
643 option that is recommended to be set on all system services. All
644 long-running system services that are included in systemd itself set
645 this option now. This option is only supported on kernel 5.8 and
646 above, since the hidepid= option supported on older kernels was not a
647 per-mount option but actually applied to the whole PID namespace.
649 * Socket units gained a new boolean setting FlushPending=. If enabled
650 all pending socket data/connections are flushed whenever the socket
651 unit enters the "listening" state, i.e. after the associated service
654 * The unit file setting NUMAMask= gained a new "all" value: when used,
655 all existing NUMA nodes are added to the NUMA mask.
657 * A new "credentials" logic has been added to system services. This is
658 a simple mechanism to pass privileged data to services in a safe and
659 secure way. It's supposed to be used to pass per-service secret data
660 such as passwords or cryptographic keys but also associated less
661 private information such as user names, certificates, and similar to
662 system services. Each credential is identified by a short user-chosen
663 name and may contain arbitrary binary data. Two new unit file
664 settings have been added: SetCredential= and LoadCredential=. The
665 former allows setting a credential to a literal string, the latter
666 sets a credential to the contents of a file (or data read from a
667 user-chosen AF_UNIX stream socket). Credentials are passed to the
668 service via a special credentials directory, one file for each
669 credential. The path to the credentials directory is passed in a new
670 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
671 are passed in the file system they may be easily referenced in
672 ExecStart= command lines too, thus no explicit support for the
673 credentials logic in daemons is required (though ideally daemons
674 would look for the bits they need in $CREDENTIALS_DIRECTORY
675 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
676 backed by unswappable memory if privileges allow it, immutable if
677 privileges allow it, is accessible only to the service's UID, and is
678 automatically destroyed when the service stops.
680 * systemd-nspawn supports the same credentials logic. It can both
681 consume credentials passed to it via the aforementioned
682 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
683 to its payload. The service manager/PID 1 has been updated to match
684 this: it can also accept credentials from the container manager that
685 invokes it (in fact: any process that invokes it), and passes them on
686 to its services. Thus, credentials can be propagated recursively down
687 the tree: from a system's service manager to a systemd-nspawn
688 service, to the service manager that runs as container payload and to
689 the service it runs below. Credentials may also be added on the
690 systemd-nspawn command line, using new --set-credential= and
691 --load-credential= command line switches that match the
692 aforementioned service settings.
694 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
695 the partition drop-ins which may be used to format/LUKS
696 encrypt/populate any created partitions. The partitions are
697 encrypted/formatted/populated before they are registered in the
698 partition table, so that they appear atomically: either the
699 partitions do not exist yet or they exist fully encrypted, formatted,
700 and populated — there is no time window where they are
701 "half-initialized". Thus the system is robust to abrupt shutdown: if
702 the tool is terminated half-way during its operations on next boot it
703 will start from the beginning.
705 * systemd-repart's --size= operation gained a new "auto" value. If
706 specified, and operating on a loopback file it is automatically sized
707 to the minimal size the size constraints permit. This is useful to
708 use "systemd-repart" as an image builder for minimally sized images.
710 * systemd-resolved now gained a third IPC interface for requesting name
711 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
712 interface is now supported. The nss-resolve NSS module has been
713 modified to use this new interface instead of D-Bus. Using Varlink
714 has a major benefit over D-Bus: it works without a broker service,
715 and thus already during earliest boot, before the dbus daemon has
716 been started. This means name resolution via systemd-resolved now
717 works at the same time systemd-networkd operates: from earliest boot
718 on, including in the initrd.
720 * systemd-resolved gained support for a new DNSStubListenerExtra=
721 configuration file setting which may be used to specify additional IP
722 addresses the built-in DNS stub shall listen on, in addition to the
723 main one on 127.0.0.53:53.
725 * Name lookups issued via systemd-resolved's D-Bus and Varlink
726 interfaces (and thus also via glibc NSS if nss-resolve is used) will
727 now honour a trailing dot in the hostname: if specified the search
728 path logic is turned off. Thus "resolvectl query foo." is now
729 equivalent to "resolvectl query --search=off foo.".
731 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
732 exposes how /etc/resolv.conf is currently managed: by resolved (and
733 in which mode if so) or another subsystem. "resolvctl" will display
734 this property in its status output.
736 * The resolv.conf snippets systemd-resolved provides will now set "."
737 as the search domain if no other search domain is known. This turns
738 off the derivation of an implicit search domain by nss-dns for the
739 hostname, when the hostname is set to an FQDN. This change is done to
740 make nss-dns using resolv.conf provided by systemd-resolved behave
741 more similarly to nss-resolve.
743 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
744 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
745 "birth" time (btime) of a file in addition to the atime, mtime, and
748 * systemd-analyze gained a new verb "capability" that lists all known
749 capabilities by the systemd build and by the kernel.
751 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
752 advance the system clock to it at boot if it is noticed to be before
753 that time. Previously, PID 1 would only advance the time to an epoch
754 time that is set during build-time. With this new file OS builders
755 can change this epoch timestamp on individual OS images without
756 having to rebuild systemd.
758 * systemd-logind will now listen to the KEY_RESTART key from the Linux
759 input layer and reboot the system if it is pressed, similarly to how
760 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
761 was originally defined in the Multimedia context (to restart playback
762 of a song or film), but is now primarily used in various embedded
763 devices for "Reboot" buttons. Accordingly, systemd-logind will now
764 honour it as such. This may configured in more detail via the new
765 HandleRebootKey= and RebootKeyIgnoreInhibited=.
767 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
768 copying OS trees, for example in "systemd-nspawn --ephemeral",
769 "systemd-nspawn --template=", "machinectl clone" and similar. This is
770 useful when operating with OSTree images, which use hardlinks heavily
771 throughout, and where such copies previously resulting in "exploding"
774 * systemd-nspawn's --console= setting gained support for a new
775 "autopipe" value, which is identical to "interactive" when invoked on
776 a TTY, and "pipe" otherwise.
778 * systemd-networkd's .network files gained support for explicitly
779 configuring the multicast membership entries of bridge devices in the
780 [BridgeMDB] section. It also gained support for the PIE queuing
781 discipline in the [FlowQueuePIE] sections.
783 * systemd-networkd's .netdev files may now be used to create "BareUDP"
784 tunnels, configured in the new [BareUDP] setting.
786 * systemd-networkd's Gateway= setting in .network files now accepts the
787 special values "_dhcp4" and "_ipv6ra" to configure additional,
788 locally defined, explicit routes to the gateway acquired via DHCP or
789 IPv6 Router Advertisements. The old setting "_dhcp" is deprecated,
790 but still accepted for backwards compatibility.
792 * systemd-networkd's [IPv6PrefixDelegation] section and
793 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
794 IPv6SendRA= (the old names are still accepted for backwards
797 * systemd-networkd's .network files gained the DHCPv6PrefixDelegation=
798 boolean setting in [Network] section. If enabled, the delegated prefix
799 gained by another link will be configured, and an address within the
800 prefix will be assigned.
802 * systemd-networkd's .network files gained the Announce= boolean setting
803 in [DHCPv6PrefixDelegation] section. When enabled, the delegated
804 prefix will be announced through IPv6 router advertisement (IPv6 RA).
805 The setting is enabled by default.
807 * VXLAN tunnels may now be marked as independent of any underlying
808 network interface via the new Independent= boolean setting.
810 * systemctl gained support for two new verbs: "service-log-level" and
811 "service-log-target" may be used on services that implement the
812 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
813 adjust the log level and target. All of systemd's long-running
814 services support this now, but ideally all system services would
815 implement this interface to make the system more uniformly
818 * The SystemCallErrorNumber= unit file setting now accepts the new
819 "kill" and "log" actions, in addition to arbitrary error number
820 specifications as before. If "kill" the processes are killed on the
821 event, if "log" the offending system call is audit logged.
823 * A new SystemCallLog= unit file setting has been added that accepts a
824 list of system calls that shall be logged about (audit).
826 * The OS image dissection logic (as used by RootImage= in unit files or
827 systemd-nspawn's --image= switch) has gained support for identifying
828 and mounting explicit /usr/ partitions, which are now defined in the
829 discoverable partition specification. This should be useful for
830 environments where the root file system is
831 generated/formatted/populated dynamically on first boot and combined
832 with an immutable /usr/ tree that is supplied by the vendor.
834 * In the final phase of shutdown, within the systemd-shutdown binary
835 we'll now try to detach MD devices (i.e software RAID) in addition to
836 loopback block devices and DM devices as before. This is supposed to
837 be a safety net only, in order to increase robustness if things go
838 wrong. Storage subsystems are expected to properly detach their
839 storage volumes during regular shutdown already (or in case of
840 storage backing the root file system: in the initrd hook we return to
843 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
844 will now log the thread ID in their log output. This is useful when
845 working with heavily threaded programs.
847 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
848 not use the RDRAND CPU instruction. This is useful in environments
849 such as replay debuggers where non-deterministic behaviour is not
852 * The autopaging logic in systemd's various tools (such as systemctl)
853 has been updated to turn on "secure" mode in "less"
854 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
855 detected. This disables invoking external programs from the pager,
856 via the pipe logic. This behaviour may be overridden via the new
857 $SYSTEMD_PAGERSECURE environment variable.
859 * Units which have resource limits (.service, .mount, .swap, .slice,
860 .socket, and .slice) gained new configuration settings
861 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
862 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
863 limits and optional action taken by systemd-oomd.
865 * A new service systemd-oomd has been added. It monitors resource
866 contention for selected parts of the unit hierarchy using the PSI
867 information reported by the kernel, and kills processes when memory
868 or swap pressure is above configured limits. This service is only
869 enabled by default in developer mode (see below) and should be
870 considered a preview in this release. Behaviour details and option
871 names are subject to change without the usual backwards-compatibility
874 * A new helper oomctl has been added to introspect systemd-oomd state.
875 It is only enabled by default in developer mode and should be
876 considered a preview without the usual backwards-compatibility
879 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
880 enabled, systemd reads the system UID boundaries from /etc/login.defs
881 at runtime, instead of using the built-in values selected during
882 build. This is an option to improve compatibility for upgrades from
883 old systems. It's strongly recommended not to make use of this
884 functionality on new systems (or even enable it during build), as it
885 makes something runtime-configurable that is mostly an implementation
886 detail of the OS, and permits avoidable differences in deployments
887 that create all kinds of problems in the long run.
889 * New meson option '-Dmode=developer|release' has been added. When
890 'developer', additional checks and features are enabled that are
891 relevant during upstream development, e.g. verification that
892 semi-automatically-generated documentation has been properly updated
893 following API changes. Those checks are considered hints for
894 developers and are not actionable in downstream builds. In addition,
895 extra features that are not ready for general consumption may be
896 enabled in developer mode. It is thus recommended to set
897 '-Dmode=release' in end-user and distro builds.
899 * systemd-cryptsetup gained support for processing detached LUKS
900 headers specified on the kernel command line via the header=
901 parameter of the luks.options= kernel command line option. The same
902 device/path syntax as for key files is supported for header files
905 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
906 slot index data for devices that are connected through a PCI bridge
907 where the _SUN index is associated with the bridge instead of the
908 network device itself. Previously this would create ambiguous device
909 naming if multiple network interfaces were connected to the same PCI
910 bridge. Since this is a naming scheme incompatibility on systems that
911 possess hardware like this it has been introduced as new naming
912 scheme "v247". The previous scheme can be selected via the
913 "net.naming-scheme=v245" kernel command line parameter.
915 * ConditionFirstBoot= semantics have been modified to be safe towards
916 abnormal system power-off during first boot. Specifically, the
917 "systemd-machine-id-commit.service" service now acts as boot
918 milestone indicating when the first boot process is sufficiently
919 complete in order to not consider the next following boot also a
920 first boot. If the system is reset before this unit is reached the
921 first time, the next boot will still be considered a first boot; once
922 it has been reached, no further boots will be considered a first
923 boot. The "first-boot-complete.target" unit now acts as official hook
924 point to order against this. If a service shall be run on every boot
925 until the first boot fully succeeds it may thus be ordered before
926 this target unit (and pull it in) and carry ConditionFirstBoot=
929 * bootctl's set-default and set-oneshot commands now accept the three
930 special strings "@default", "@oneshot", "@current" in place of a boot
931 entry id. These strings are resolved to the current default and
932 oneshot boot loader entry, as well as the currently booted one. Thus
933 a command "bootctl set-default @current" may be used to make the
934 currently boot menu item the new default for all subsequent boots.
936 * "systemctl edit" has been updated to show the original effective unit
937 contents in commented form in the text editor.
939 * Units in user mode are now segregated into three new slices:
940 session.slice (units that form the core of graphical session),
941 app.slice ("normal" user applications), and background.slice
942 (low-priority tasks). Unless otherwise configured, user units are
943 placed in app.slice. The plan is to add resource limits and
944 protections for the different slices in the future.
946 * New GPT partition types for RISCV32/64 for the root and /usr
947 partitions, and their associated Verity partitions have been defined,
948 and are now understood by systemd-gpt-auto-generator, and the OS
949 image dissection logic.
951 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
952 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
953 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
954 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
955 Dey, Charles Lee, Chris Down, Christian Göttsche, Christof Efkemann,
956 Christoph Ruegge, Clemens Gruber, Daan De Meyer, Daniele Medri, Daniel
957 Mack, Daniel Rusek, Dan Streetman, David Tardon, Dimitri John Ledkov,
958 Dmitry Borodaenko, Elias Probst, Elisei Roca, ErrantSpore, Etienne
959 Doms, Fabrice Fontaine, fangxiuning, Felix Riemann, Florian Klink,
960 Franck Bui, Frantisek Sumsal, fwSmit, George Rawlinson, germanztz,
961 Gibeom Gwon, Glen Whitney, Gogo Gogsi, Göran Uddeborg, Grant Mathews,
962 Hans de Goede, Hans Ulrich Niedermann, Haochen Tong, Harald Seiler,
963 huangyong, Hubert Kario, igo95862, Ikey Doherty, Insun Pyo, Jan Chren,
964 Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann, Jonathan
965 Lebon, Jörg Thalheim, Josh Brobst, Juergen Hoetzel, Julien Humbert,
966 Kai-Chuan Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan
967 Gjoshev, Kyle Huey, Kyle Russell, Lee Whalen, Lennart Poettering,
968 lichangze, Luca Boccassi, Lucas Werkmeister, Luca Weiss, Marc
969 Kleine-Budde, Marco Wang, Martin Wilck, Marti Raudsepp, masmullin2000,
970 Máté Pozsgay, Matt Fenwick, Michael Biebl, Michael Scherer, Michal
971 Koutný, Michal Sekletár, Michal Suchanek, Mikael Szreder, Milo
972 Casagrande, mirabilos, Mitsuha_QuQ, mog422, Muhammet Kara, Nazar
973 Vinnichuk, Nicholas Narsing, Nicolas Fella, Njibhu, nl6720, Oğuz Ersen,
974 Olivier Le Moal, Ondrej Kozina, onlybugreports, Pass Automated Testing
975 Suite, Pat Coulthard, Pavel Sapezhko, Pedro Ruiz, perry_yuan, Peter
976 Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan C,
977 Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert Marko,
978 Roman Beranek, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer,
979 Samanta Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd,
980 Susant Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
981 Miettinen, vanou, Vito Caputo, Weblate, Wen Yang, Whired Planck,
982 williamvds, Yu, Li-Yu, Yuri Chornoivan, Yu Watanabe, Zbigniew
983 Jędrzejewski-Szmek, Zmicer Turok, Дамјан Георгиевски
989 * The service manager gained basic support for cgroup v2 freezer. Units
990 can now be suspended or resumed either using new systemctl verbs,
991 freeze and thaw respectively, or via D-Bus.
993 * PID 1 may now automatically load pre-compiled AppArmor policies from
994 /etc/apparmor/earlypolicy during early boot.
996 * The CPUAffinity= setting in service unit files now supports a new
997 special value "numa" that causes the CPU affinity masked to be set
998 based on the NUMA mask.
1000 * systemd will now log about all left-over processes remaining in a
1001 unit when the unit is stopped. It will now warn about services using
1002 KillMode=none, as this is generally an unsafe thing to make use of.
1004 * Two new unit file settings
1005 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
1006 added. They may be used to check whether a specific file system path
1007 resides on a block device that is encrypted on the block level
1008 (i.e. using dm-crypt/LUKS).
1010 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
1011 has been added that may be used for simple environment checks. This
1012 is particularly useful when passing in environment variables from a
1013 container manager (or from PAM in case of the systemd --user
1016 * .service unit files now accept a new setting CoredumpFilter= which
1017 allows configuration of the memory sections coredumps of the
1018 service's processes shall include.
1020 * .mount units gained a new ReadWriteOnly= boolean option. If set
1021 it will not be attempted to mount a file system read-only if mounting
1022 in read-write mode doesn't succeed. An option x-systemd.rw-only is
1023 available in /etc/fstab to control the same.
1025 * .socket units gained a new boolean setting PassPacketInfo=. If
1026 enabled, the kernel will attach additional per-packet metadata to all
1027 packets read from the socket, as an ancillary message. This controls
1028 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
1029 depending on socket type.
1031 * .service units gained a new setting RootHash= which may be used to
1032 specify the root hash for verity enabled disk images which are
1033 specified in RootImage=. RootVerity= may be used to specify a path to
1034 the Verity data matching a RootImage= file system. (The latter is
1035 only useful for images that do not contain the Verity data embedded
1036 into the same image that carries a GPT partition table following the
1037 Discoverable Partition Specification). Similarly, systemd-nspawn
1038 gained a new switch --verity-data= that takes a path to a file with
1039 the verity data of the disk image supplied in --image=, if the image
1040 doesn't contain the verity data itself.
1042 * .service units gained a new setting RootHashSignature= which takes
1043 either a base64 encoded PKCS#7 signature of the root hash specified
1044 with RootHash=, or a path to a file to read the signature from. This
1045 allows validation of the root hash against public keys available in
1046 the kernel keyring, and is only supported on recent kernels
1047 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
1048 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
1049 this mechanism has also been added to systemd-veritysetup.
1051 * .service unit files gained two new options
1052 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
1053 tune behaviour if a start or stop timeout is hit, i.e. whether to
1054 terminate the service with SIGTERM, SIGABRT or SIGKILL.
1056 * Most options in systemd that accept hexadecimal values prefixed with
1057 0x in additional to the usual decimal notation now also support octal
1058 notation when the 0o prefix is used and binary notation if the 0b
1061 * Various command line parameters and configuration file settings that
1062 configure key or certificate files now optionally take paths to
1063 AF_UNIX sockets in the file system. If configured that way a stream
1064 connection is made to the socket and the required data read from
1065 it. This is a simple and natural extension to the existing regular
1066 file logic, and permits other software to provide keys or
1067 certificates via simple IPC services, for example when unencrypted
1068 storage on disk is not desired. Specifically, systemd-networkd's
1069 Wireguard and MACSEC key file settings as well as
1070 systemd-journal-gatewayd's and systemd-journal-remote's PEM
1071 key/certificate parameters support this now.
1073 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
1074 configuration files that support specifier expansion learnt six new
1075 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
1076 resolve to the various ID fields from /etc/os-release, %l resolves to
1077 the "short" hostname of the system, i.e. the hostname configured in
1078 the kernel truncated at the first dot.
1080 * Support for the .include syntax in unit files has been removed. The
1081 concept has been obsolete for 6 years and we started warning about
1082 its pending removal 2 years ago (also see NEWS file below). It's
1085 * StandardError= and StandardOutput= in unit files no longer support
1086 the "syslog" and "syslog-console" switches. They were long removed
1087 from the documentation, but will now result in warnings when used,
1088 and be converted to "journal" and "journal+console" automatically.
1090 * If the service setting User= is set to the "nobody" user, a warning
1091 message is now written to the logs (but the value is nonetheless
1092 accepted). Setting User=nobody is unsafe, since the primary purpose
1093 of the "nobody" user is to own all files whose owner cannot be mapped
1094 locally. It's in particular used by the NFS subsystem and in user
1095 namespacing. By running a service under this user's UID it might get
1096 read and even write access to all these otherwise unmappable files,
1097 which is quite likely a major security problem.
1099 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
1100 and others) now have a size and inode limits applied (50% of RAM for
1101 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
1102 that the implicit kernel default is 50% too, so there is no change
1103 in the size limit for /tmp and /dev/shm.
1105 * nss-mymachines lost support for resolution of users and groups, and
1106 now only does resolution of hostnames. This functionality is now
1107 provided by nss-systemd. Thus, the 'mymachines' entry should be
1108 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
1109 (and 'systemd' added if it is not already there).
1111 * A new kernel command line option systemd.hostname= has been added
1112 that allows controlling the hostname that is initialized early during
1115 * A kernel command line option "udev.blockdev_read_only" has been
1116 added. If specified all hardware block devices that show up are
1117 immediately marked as read-only by udev. This option is useful for
1118 making sure that a specific boot under no circumstances modifies data
1119 on disk. Use "blockdev --setrw" to undo the effect of this, per
1122 * A new boolean kernel command line option systemd.swap= has been
1123 added, which may be used to turn off automatic activation of swap
1124 devices listed in /etc/fstab.
1126 * New kernel command line options systemd.condition-needs-update= and
1127 systemd.condition-first-boot= have been added, which override the
1128 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
1131 * A new kernel command line option systemd.clock-usec= has been added
1132 that allows setting the system clock to the specified time in µs
1133 since Jan 1st, 1970 early during boot. This is in particular useful
1134 in order to make test cases more reliable.
1136 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
1137 systemd-coredump to save core files for suid processes. When saving
1138 the core file, systemd-coredump will use the effective uid and gid of
1139 the process that faulted.
1141 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
1142 now automatically set to "Y" at boot, in order to enable pstore
1143 generation for collection with systemd-pstore.
1145 * We provide a set of udev rules to enable auto-suspend on PCI and USB
1146 devices that were tested to correctly support it. Previously, this
1147 was distributed as a set of udev rules, but has now been replaced by
1148 by a set of hwdb entries (and a much shorter udev rule to take action
1149 if the device modalias matches one of the new hwdb entries).
1151 As before, entries are periodically imported from the database
1152 maintained by the ChromiumOS project. If you have a device that
1153 supports auto-suspend correctly and where it should be enabled by
1154 default, please submit a patch that adds it to the database (see
1155 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
1157 * systemd-udevd gained the new configuration option timeout_signal= as well
1158 as a corresponding kernel command line option udev.timeout_signal=.
1159 The option can be used to configure the UNIX signal that the main
1160 daemon sends to the worker processes on timeout. Setting the signal
1161 to SIGABRT is useful for debugging.
1163 * .link files managed by systemd-udevd gained options RxFlowControl=,
1164 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
1165 order to configure various flow control parameters. They also gained
1166 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
1167 frame ring buffer sizes.
1169 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
1170 enabled systemd-networkd manages all routes configured by other tools.
1172 * .network files managed by systemd-networkd gained a new section
1173 [SR-IOV], in order to configure SR-IOV capable network devices.
1175 * systemd-networkd's [IPv6Prefix] section in .network files gained a
1176 new boolean setting Assign=. If enabled an address from the prefix is
1177 automatically assigned to the interface.
1179 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
1180 controls delegated prefixes assigned by DHCPv6 client. The section
1181 has three settings: SubnetID=, Assign=, and Token=. The setting
1182 SubnetID= allows explicit configuration of the preferred subnet that
1183 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
1184 Assign= is enabled (which is the default) an address from any acquired
1185 delegated prefix is automatically chosen and assigned to the
1186 interface. The setting Token= specifies an optional address generation
1189 * systemd-networkd's [Network] section gained a new setting
1190 IPv4AcceptLocal=. If enabled the interface accepts packets with local
1193 * systemd-networkd gained support for configuring the HTB queuing
1194 discipline in the [HierarchyTokenBucket] and
1195 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
1196 be configured in the [PFIFO] section, "GRED" in
1197 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
1198 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
1199 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
1200 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
1201 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
1202 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
1204 * systemd-networkd gained support for a new Termination= setting in the
1205 [CAN] section for configuring the termination resistor. It also
1206 gained a new ListenOnly= setting for controlling whether to only
1207 listen on CAN interfaces, without interfering with traffic otherwise
1208 (which is useful for debugging/monitoring CAN network
1209 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
1210 been added to configure various CAN-FD aspects.
1212 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
1213 When enabled, DHCPv6 will be attempted right-away without requiring an
1214 Router Advertisement packet suggesting it first (i.e. without the 'M'
1215 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
1216 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
1217 the RA packets suggest it.
1219 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
1220 which may be used to turn off use of the gateway information provided
1221 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
1222 used to configure how to process leases that lack a lifetime option.
1224 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
1225 setting SendVendorOption= allowing configuration of additional vendor
1226 options to send in the DHCP requests/responses. The [DHCPv6] section
1227 gained a new SendOption= setting for sending arbitrary DHCP
1228 options. RequestOptions= has been added to request arbitrary options
1229 from the server. UserClass= has been added to set the DHCP user class
1232 * systemd-networkd's [DHCPServer] section gained a new set of options
1233 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
1234 information about these three protocols in the DHCP lease. It also
1235 gained support for including "MUD" URLs ("Manufacturer Usage
1236 Description"). Support for "MUD" URLs was also added to the LLDP
1237 stack, configurable in the [LLDP] section in .network files.
1239 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
1240 mode. Also, the sections now support a new setting SourceMACAddress=.
1242 * systemd-networkd's .netdev files now support a new setting
1243 VLANProtocol= in the [Bridge] section that allows configuration of
1244 the VLAN protocol to use.
1246 * systemd-networkd supports a new Group= setting in the [Link] section
1247 of the .network files, to control the link group.
1249 * systemd-networkd's [Network] section gained a new
1250 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
1251 link local address is generated.
1253 * A new default .network file is now shipped that matches TUN/TAP
1254 devices that begin with "vt-" in their name. Such interfaces will
1255 have IP routing onto the host links set up automatically. This is
1256 supposed to be used by VM managers to trivially acquire a network
1257 interface which is fully set up for host communication, simply by
1258 carefully picking an interface name to use.
1260 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
1261 which sets the route priority for routes specified by the DHCP server.
1263 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
1264 which configures the vendor class information sent to DHCP server.
1266 * The BlackList= settings in .network files' [DHCPv4] and
1267 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
1268 are still understood to provide compatibility.
1270 * networkctl gained the new "forcerenew" command for forcing all DHCP
1271 server clients to renew their lease. The interface "status" output
1272 will now show numerous additional fields of information about an
1273 interface. There are new "up" and "down" commands to bring specific
1274 interfaces up or down.
1276 * systemd-resolved's DNS= configuration option now optionally accepts a
1277 port number (after ":") and a host name (after "#"). When the host
1278 name is specified, the DNS-over-TLS certificate is validated to match
1279 the specified hostname. Additionally, in case of IPv6 addresses, an
1280 interface may be specified (after "%").
1282 * systemd-resolved may be configured to forward single-label DNS names.
1283 This is not standard-conformant, but may make sense in setups where
1284 public DNS servers are not used.
1286 * systemd-resolved's DNS-over-TLS support gained SNI validation.
1288 * systemd-nspawn's --resolv-conf= switch gained a number of new
1289 supported values. Specifically, options starting with "replace-" are
1290 like those prefixed "copy-" but replace any existing resolv.conf
1291 file. And options ending in "-uplink" and "-stub" can now be used to
1292 propagate other flavours of resolv.conf into the container (as
1293 defined by systemd-resolved).
1295 * The various programs included in systemd can now optionally output
1296 their log messages on stderr prefixed with a timestamp, controlled by
1297 the $SYSTEMD_LOG_TIME environment variable.
1299 * systemctl gained a new "-P" switch that is a shortcut for "--value
1302 * "systemctl list-units" and "systemctl list-machines" no longer hide
1303 their first output column with --no-legend. To hide the first column,
1306 * "systemctl reboot" takes the option "--reboot-argument=".
1307 The optional positional argument to "systemctl reboot" is now
1308 being deprecated in favor of this option.
1310 * systemd-run gained a new switch --slice-inherit. If specified the
1311 unit it generates is placed in the same slice as the systemd-run
1314 * systemd-journald gained support for zstd compression of large fields
1315 in journal files. The hash tables in journal files have been hardened
1316 against hash collisions. This is an incompatible change and means
1317 that journal files created with new systemd versions are not readable
1318 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
1319 environment variable for systemd-journald.service is set to 0 this
1320 new hardening functionality may be turned off, so that generated
1321 journal files remain compatible with older journalctl
1324 * journalctl will now include a clickable link in the default output for
1325 each log message for which an URL with further documentation is
1326 known. This is only supported on terminal emulators that support
1327 clickable hyperlinks, and is turned off if a pager is used (since
1328 "less" still doesn't support hyperlinks,
1329 unfortunately). Documentation URLs may be included in log messages
1330 either by including a DOCUMENTATION= journal field in it, or by
1331 associating a journal message catalog entry with the log message's
1332 MESSAGE_ID, which then carries a "Documentation:" tag.
1334 * journald.conf gained a new boolean setting Audit= that may be used to
1335 control whether systemd-journald will enable audit during
1338 * when systemd-journald's log stream is broken up into multiple lines
1339 because the PID of the sender changed this is indicated in the
1340 generated log records via the _LINE_BREAK=pid-change field.
1342 * journalctl's "-o cat" output mode will now show one or more journal
1343 fields specified with --output-fields= instead of unconditionally
1344 MESSAGE=. This is useful to retrieve a very specific set of fields
1345 without any decoration.
1347 * The sd-journal.h API gained two new functions:
1348 sd_journal_enumerate_available_unique() and
1349 sd_journal_enumerate_available_data() that operate like their
1350 counterparts that lack the _available_ in the name, but skip items
1351 that cannot be read and processed by the local implementation
1352 (i.e. are compressed in an unsupported format or such),
1354 * coredumpctl gained a new --file= switch, matching the same one in
1355 journalctl: a specific journal file may be specified to read the
1358 * coredumps collected by systemd-coredump may now be compressed using
1361 * systemd-binfmt gained a new switch --unregister for unregistering all
1362 registered entries at once. This is now invoked automatically at
1363 shutdown, so that binary formats registered with the "F" flag will
1364 not block clean file system unmounting.
1366 * systemd-notify's --pid= switch gained new values: "parent", "self",
1367 "auto" for controlling which PID to send to the service manager: the
1368 systemd-notify process' PID, or the one of the process invoking it.
1370 * systemd-logind's Session bus object learnt a new method call
1371 SetType() for temporarily updating the session type of an already
1372 allocated session. This is useful for upgrading tty sessions to
1373 graphical ones once a compositor is invoked.
1375 * systemd-socket-proxy gained a new switch --exit-idle-time= for
1376 configuring an exit-on-idle time.
1378 * systemd-repart's --empty= setting gained a new value "create". If
1379 specified a new empty regular disk image file is created under the
1380 specified name. Its size may be specified with the new --size=
1381 option. The latter is also supported without the "create" mode, in
1382 order to grow existing disk image files to the specified size. These
1383 two new options are useful when creating or manipulating disk images
1384 instead of operating on actual block devices.
1386 * systemd-repart drop-ins now support a new UUID= setting to control
1387 the UUID to assign to a newly created partition.
1389 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
1392 * systemd-repart's Label= setting now support the usual, simple
1393 specifier expansion.
1395 * systemd-homed's LUKS backend gained the ability to discard empty file
1396 system blocks automatically when the user logs out. This is enabled
1397 by default to ensure that home directories take minimal space when
1398 logged out but get full size guarantees when logged in. This may be
1399 controlled with the new --luks-offline-discard= switch to homectl.
1401 * If systemd-homed detects that /home/ is encrypted as a whole it will
1402 now default to the directory or subvolume backends instead of the
1403 LUKS backend, in order to avoid double encryption. The default
1404 storage and file system may now be configured explicitly, too, via
1405 the new /etc/systemd/homed.conf configuration file.
1407 * systemd-homed now supports unlocking home directories with FIDO2
1408 security tokens that support the 'hmac-secret' extension, in addition
1409 to the existing support for PKCS#11 security token unlocking
1410 support. Note that many recent hardware security tokens support both
1411 interfaces. The FIDO2 support is accessible via homectl's
1412 --fido2-device= option.
1414 * homectl's --pkcs11-uri= setting now accepts two special parameters:
1415 if "auto" is specified and only one suitable PKCS#11 security token
1416 is plugged in, its URL is automatically determined and enrolled for
1417 unlocking the home directory. If "list" is specified a brief table of
1418 suitable PKCS#11 security tokens is shown. Similar, the new
1419 --fido2-device= option also supports these two special values, for
1420 automatically selecting and listing suitable FIDO2 devices.
1422 * The /etc/crypttab tmp option now optionally takes an argument
1423 selecting the file system to use. Moreover, the default is now
1424 changed from ext2 to ext4.
1426 * There's a new /etc/crypttab option "keyfile-erase". If specified the
1427 key file listed in the same line is removed after use, regardless if
1428 volume activation was successful or not. This is useful if the key
1429 file is only acquired transiently at runtime and shall be erased
1430 before the system continues to boot.
1432 * There's also a new /etc/crypttab option "try-empty-password". If
1433 specified, before asking the user for a password it is attempted to
1434 unlock the volume with an empty password. This is useful for
1435 installing encrypted images whose password shall be set on first boot
1436 instead of at installation time.
1438 * systemd-cryptsetup will now attempt to load the keys to unlock
1439 volumes with automatically from files in
1440 /etc/cryptsetup-keys.d/<volume>.key and
1441 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
1443 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
1444 /etc/crypttab, during boot.
1446 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
1447 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
1450 * A new generator systemd-xdg-autostart-generator has been added. It
1451 generates systemd unit files from XDG autostart .desktop files, and
1452 may be used to let the systemd user instance manage services that are
1453 started automatically as part of the desktop session.
1455 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
1456 to query and change the firmware's 'reboot into firmware' setup flag.
1458 * systemd-firstboot gained a new switch --kernel-command-line= that may
1459 be used to initialize the /etc/kernel/cmdline file of the image. It
1460 also gained a new switch --root-password-hashed= which is like
1461 --root-password= but accepts a pre-hashed UNIX password as
1462 argument. The new option --delete-root-password may be used to unset
1463 any password for the root user (dangerous!). The --root-shell= switch
1464 may be used to control the shell to use for the root account. A new
1465 --force option may be used to override any already set settings with
1466 the parameters specified on the command line (by default, the tool
1467 will not override what has already been set before, i.e. is purely
1470 * systemd-firstboot gained support for a new --image= switch, which is
1471 similar to --root= but accepts the path to a disk image file, on
1472 which it then operates.
1474 * A new sd-path.h API has been added to libsystemd. It provides a
1475 simple API for retrieving various search paths and primary
1476 directories for various resources.
1478 * A new call sd_notify_barrier() has been added to the sd-daemon.h
1479 API. The call will block until all previously sent sd_notify()
1480 messages have been processed by the service manager. This is useful
1481 to remove races caused by a process already having disappeared at the
1482 time a notification message is processed by the service manager,
1483 making correct attribution impossible. The systemd-notify tool will
1484 now make use of this call implicitly, but this can be turned off again
1485 via the new --no-block switch.
1487 * When sending a file descriptor (fd) to the service manager to keep
1488 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
1489 may be specified. If passed the service manager will refrain from
1490 poll()ing on the file descriptor. Traditionally (and when the
1491 parameter is not specified), the service manager will poll it for
1492 POLLHUP or POLLERR events, and immediately close the fds in that
1495 * The service manager (PID1) gained a new D-Bus method call
1496 SetShowStatus() which may be used to control whether it shall show
1497 boot-time status output on the console. This method has a similar
1498 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
1500 * The sd-bus API gained a number of convenience functions that take
1501 va_list arguments rather than "...". For example, there's now
1502 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
1503 it easier to build wrappers that accept variadic arguments and want
1504 to pass a ready va_list structure to sd-bus.
1506 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
1507 flag which alters how the userdata pointer to pass to the callbacks
1508 is determined. When the flag is set, the offset field is converted
1509 as-is into a pointer, without adding it to the object pointer the
1510 vtable is associated with.
1512 * sd-bus now exposes four new functions:
1513 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
1514 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
1515 validate strings to check if they qualify as various D-Bus concepts.
1517 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
1518 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
1519 that simplify adding argument names to D-Bus methods and signals.
1521 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
1523 * Various D-Bus APIs of systemd daemons now have man pages that
1524 document the methods, signals and properties.
1526 * The expectations on user/group name syntax are now documented in
1527 detail; documentation on how classic home directories may be
1528 converted into home directories managed by homed has been added;
1529 documentation regarding integration of homed/userdb functionality in
1530 desktops has been added:
1532 https://systemd.io/USER_NAMES
1533 https://systemd.io/CONVERTING_TO_HOMED
1534 https://systemd.io/USERDB_AND_DESKTOPS
1536 * Documentation for the on-disk Journal file format has been updated
1537 and has now moved to:
1539 https://systemd.io/JOURNAL_FILE_FORMAT
1541 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
1542 has been extended by a set of environment variables that expose
1543 select fields from the host's os-release file to the container
1544 payload. Similarly, host's os-release files can be mounted into the
1545 container underneath /run/host. Together, those mechanisms provide a
1546 standardized way to expose information about the host to the
1547 container payload. Both interfaces are implemented in systemd-nspawn.
1549 * All D-Bus services shipped in systemd now implement the generic
1550 LogControl1 D-Bus API which allows clients to change log level +
1551 target of the service during runtime.
1553 * Only relevant for developers: the mkosi.default symlink has been
1554 dropped from version control. Please create a symlink to one of the
1555 distribution-specific defaults in .mkosi/ based on your preference.
1557 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
1558 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
1559 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
1560 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
1561 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
1562 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
1563 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
1564 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
1565 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
1566 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
1567 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
1568 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
1569 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
1570 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
1571 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
1572 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
1573 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
1574 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
1575 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
1576 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
1577 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
1578 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
1579 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
1580 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
1581 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
1582 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
1583 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
1584 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
1585 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
1586 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
1587 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
1588 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
1589 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
1590 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
1591 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
1592 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
1593 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
1594 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
1596 – Warsaw, 2020-07-30
1600 * A new tool "systemd-repart" has been added, that operates as an
1601 idempotent declarative repartitioner for GPT partition tables.
1602 Specifically, a set of partitions that must or may exist can be
1603 configured via drop-in files, and during every boot the partition
1604 table on disk is compared with these files, creating missing
1605 partitions or growing existing ones based on configurable relative
1606 and absolute size constraints. The tool is strictly incremental,
1607 i.e. does not delete, shrink or move partitions, but only adds and
1608 grows them. The primary use-case is OS images that ship in minimized
1609 form, that on first boot are grown to the size of the underlying
1610 block device or augmented with additional partitions. For example,
1611 the root partition could be extended to cover the whole disk, or a
1612 swap or /home partitions could be added on first boot. It can also be
1613 used for systems that use an A/B update scheme but ship images with
1614 just the A partition, with B added on first boot. The tool is
1615 primarily intended to be run in the initrd, shortly before
1616 transitioning into the host OS, but can also be run after the
1617 transition took place. It automatically discovers the disk backing
1618 the root file system, and should hence not require any additional
1619 configuration besides the partition definition drop-ins. If no
1620 configuration drop-ins are present, no action is taken.
1622 * A new component "userdb" has been added, along with a small daemon
1623 "systemd-userdbd.service" and a client tool "userdbctl". The framework
1624 allows defining rich user and group records in a JSON format,
1625 extending on the classic "struct passwd" and "struct group"
1626 structures. Various components in systemd have been updated to
1627 process records in this format, including systemd-logind and
1628 pam-systemd. The user records are intended to be extensible, and
1629 allow setting various resource management, security and runtime
1630 parameters that shall be applied to processes and sessions of the
1631 user as they log in. This facility is intended to allow associating
1632 such metadata directly with user/group records so that they can be
1633 produced, extended and consumed in unified form. We hope that
1634 eventually frameworks such as sssd will generate records this way, so
1635 that for the first time resource management and various other
1636 per-user settings can be configured in LDAP directories and then
1637 provided to systemd (specifically to systemd-logind and pam-system)
1638 to apply on login. For further details see:
1640 https://systemd.io/USER_RECORD
1641 https://systemd.io/GROUP_RECORD
1642 https://systemd.io/USER_GROUP_API
1644 * A small new service systemd-homed.service has been added, that may be
1645 used to securely manage home directories with built-in encryption.
1646 The complete user record data is unified with the home directory,
1647 thus making home directories naturally migratable. Its primary
1648 back-end is based on LUKS volumes, but fscrypt, plain directories,
1649 and other storage schemes are also supported. This solves a couple of
1650 problems we saw with traditional ways to manage home directories, in
1651 particular when it comes to encryption. For further discussion of
1652 this, see the video of Lennart's talk at AllSystemsGo! 2019:
1654 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
1656 For further details about the format and expectations on home
1657 directories this new daemon makes, see:
1659 https://systemd.io/HOME_DIRECTORY
1661 * systemd-journald is now multi-instantiable. In addition to the main
1662 instance systemd-journald.service there's now a template unit
1663 systemd-journald@.service, with each instance defining a new named
1664 log 'namespace' (whose name is specified via the instance part of the
1665 unit name). A new unit file setting LogNamespace= has been added,
1666 taking such a namespace name, that assigns services to the specified
1667 log namespaces. As each log namespace is serviced by its own
1668 independent journal daemon, this functionality may be used to improve
1669 performance and increase isolation of applications, at the price of
1670 losing global message ordering. Each instance of journald has a
1671 separate set of configuration files, with possibly different disk
1672 usage limitations and other settings.
1674 journalctl now takes a new option --namespace= to show logs from a
1675 specific log namespace. The sd-journal.h API gained
1676 sd_journal_open_namespace() for opening the log stream of a specific
1677 log namespace. systemd-journald also gained the ability to exit on
1678 idle, which is useful in the context of log namespaces, as this means
1679 log daemons for log namespaces can be activated automatically on
1680 demand and will stop automatically when no longer used, minimizing
1683 * When systemd-tmpfiles copies a file tree using the 'C' line type it
1684 will now label every copied file according to the SELinux database.
1686 * When systemd/PID 1 detects it is used in the initrd it will now boot
1687 into initrd.target rather than default.target by default. This should
1688 make it simpler to build initrds with systemd as for many cases the
1689 only difference between a host OS image and an initrd image now is
1690 the presence of the /etc/initrd-release file.
1692 * A new kernel command line option systemd.cpu_affinity= is now
1693 understood. It's equivalent to the CPUAffinity= option in
1694 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
1695 itself and the default for all other processes.
1697 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
1698 equivalent), the SELinux database is now reloaded, ensuring that
1699 sockets and other file system objects are generated taking the new
1700 database into account.
1702 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
1703 "quiet" has been changed to imply that instead of
1704 "systemd.show-status=auto". In this mode, only messages about errors
1705 and significant delays in boot are shown on the console.
1707 * The sd-event.h API gained native support for the new Linux "pidfd"
1708 concept. This permits watching processes using file descriptors
1709 instead of PID numbers, which fixes a number of races and makes
1710 process supervision more robust and efficient. All of systemd's
1711 components will now use pidfds if the kernel supports it for process
1712 watching, with the exception of PID 1 itself, unfortunately. We hope
1713 to move PID 1 to exclusively using pidfds too eventually, but this
1714 requires some more kernel work first. (Background: PID 1 watches
1715 processes using waitid() with the P_ALL flag, and that does not play
1716 together nicely with pidfds yet.)
1718 * Closely related to this, the sd-event.h API gained two new calls
1719 sd_event_source_send_child_signal() (for sending a signal to a
1720 watched process) and sd_event_source_get_child_process_own() (for
1721 marking a process so that it is killed automatically whenever the
1722 event source watching it is freed).
1724 * systemd-networkd gained support for configuring Token Bucket Filter
1725 (TBF) parameters in its qdisc configuration support. Similarly,
1726 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
1727 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
1729 * systemd-networkd gained support for Intermediate Functional Block
1730 (IFB) network devices.
1732 * systemd-networkd gained support for configuring multi-path IP routes,
1733 using the new MultiPathRoute= setting in the [Route] section.
1735 * systemd-networkd's DHCPv4 client has been updated to support a new
1736 SendDecline= option. If enabled, duplicate address detection is done
1737 after a DHCP offer is received from the server. If a conflict is
1738 detected, the address is declined. The DHCPv4 client also gained
1739 support for a new RouteMTUBytes= setting that allows to configure the
1740 MTU size to be used for routes generated from DHCPv4 leases.
1742 * The PrefixRoute= setting in systemd-networkd's [Address] section of
1743 .network files has been deprecated, and replaced by AddPrefixRoute=,
1744 with its sense inverted.
1746 * The Gateway= setting of [Route] sections of .network files gained
1747 support for a special new value "_dhcp". If set, the configured
1748 static route uses the gateway host configured via DHCP.
1750 * New User= and SuppressPrefixLength= settings have been implemented
1751 for the [RoutingPolicyRule] section of .network files to configure
1752 source routing based on UID ranges and prefix length, respectively.
1754 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
1755 D-Bus message object as "sensitive". Those objects are erased from
1756 memory when they are freed. This concept is intended to be used for
1757 messages that contain security sensitive data. A new flag
1758 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
1759 in sd-bus vtables, causing any incoming and outgoing messages of
1760 those methods to be implicitly marked as "sensitive".
1762 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
1763 contents of a message (or parts thereof) to standard output for
1766 * systemd-sysusers gained support for creating users with the primary
1767 group named differently than the user.
1769 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
1770 gained support for growing XFS partitions. Previously it supported
1771 only ext4 and btrfs partitions.
1773 * The support for /etc/crypttab gained a new x-initrd.attach option. If
1774 set, the specified encrypted volume is unlocked already in the
1775 initrd. This concept corresponds to the x-initrd.mount option in
1778 * systemd-cryptsetup gained native support for unlocking encrypted
1779 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
1780 encryption of volumes to YubiKeys. This is exposed in the new
1781 pkcs11-uri= option in /etc/crypttab.
1783 * The /etc/fstab support in systemd now supports two new mount options
1784 x-systemd.{required,wanted}-by=, for explicitly configuring the units
1785 that the specified mount shall be pulled in by, in place of
1786 the usual local-fs.target/remote-fs.target.
1788 * The https://systemd.io/ web site has been relaunched, directly
1789 populated with most of the documentation included in the systemd
1790 repository. systemd also acquired a new logo, thanks to Tobias
1793 * systemd-udevd gained support for managing "alternative" network
1794 interface names, as supported by new Linux kernels. For the first
1795 time this permits assigning multiple (and longer!) names to a network
1796 interface. systemd-udevd will now by default assign the names
1797 generated via all supported naming schemes to each interface. This
1798 may be further tweaked with .link files and the AlternativeName= and
1799 AlternativeNamesPolicy= settings. Other components of systemd have
1800 been updated to support the new alternative names wherever
1801 appropriate. For example, systemd-nspawn will now generate
1802 alternative interface names for the host-facing side of container
1803 veth links based on the full container name without truncation.
1805 * systemd-nspawn interface naming logic has been updated in another way
1806 too: if the main interface name (i.e. as opposed to new-style
1807 "alternative" names) based on the container name is truncated, a
1808 simple hashing scheme is used to give different interface names to
1809 multiple containers whose names all begin with the same prefix. Since
1810 this changes the primary interface names pointing to containers if
1811 truncation happens, the old scheme may still be requested by
1812 selecting an older naming scheme, via the net.naming-scheme= kernel
1813 command line option.
1815 * PrivateUsers= in service files now works in services run by the
1816 systemd --user per-user instance of the service manager.
1818 * A new per-service sandboxing option ProtectClock= has been added that
1819 locks down write access to the system clock. It takes away device
1820 node access to /dev/rtc as well as the system calls that set the
1821 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
1822 Note that this option does not affect access to auxiliary services
1823 that allow changing the clock, for example access to
1826 * The systemd-id128 tool gained a new "show" verb for listing or
1827 resolving a number of well-known UUIDs/128bit IDs, currently mostly
1828 GPT partition table types.
1830 * The Discoverable Partitions Specification has been updated to support
1831 /var and /var/tmp partition discovery. Support for this has been
1832 added to systemd-gpt-auto-generator. For details see:
1834 https://systemd.io/DISCOVERABLE_PARTITIONS
1836 * "systemctl list-unit-files" has been updated to show a new column
1837 with the suggested enablement state based on the vendor preset files
1838 for the respective units.
1840 * "systemctl" gained a new option "--with-dependencies". If specified
1841 commands such as "systemctl status" or "systemctl cat" will now show
1842 all specified units along with all units they depend on.
1844 * networkctl gained support for showing per-interface logs in its
1847 * systemd-networkd-wait-online gained support for specifying the maximum
1848 operational state to wait for, and to wait for interfaces to
1851 * The [Match] section of .link and .network files now supports a new
1852 option PermanentMACAddress= which may be used to check against the
1853 permanent MAC address of a network device even if a randomized MAC
1856 * The [TrafficControlQueueingDiscipline] section in .network files has
1857 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
1858 dropped from the individual setting names.
1860 * Any .link and .network files that have an empty [Match] section (this
1861 also includes empty and commented-out files) will now be
1862 rejected. systemd-udev and systemd-networkd started warning about
1863 such files in version 243.
1865 * systemd-logind will now validate access to the operation of changing
1866 the virtual terminal via a polkit action. By default, only users
1867 with at least one session on a local VT are granted permission.
1869 * When systemd sets up PAM sessions that invoked service processes
1870 shall run in, the pam_setcred() API is now invoked, thus permitting
1871 PAM modules to set additional credentials for the processes.
1873 * portablectl attach/detach verbs now accept --now and --enable options
1874 to combine attachment with enablement and invocation, or detachment
1875 with stopping and disablement.
1877 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
1878 fixed, which in turn exposed bugs in unit configuration of services
1879 which have Type=oneshot and should only run once, but do not have
1880 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
1881 service may be started again after exiting successfully, for example
1882 as a dependency in another transaction. Affected services included
1883 some internal systemd services (most notably
1884 systemd-vconsole-setup.service, which was updated to have
1885 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
1886 plymouth has been suitably updated or patched before upgrading to
1887 this systemd release. See
1888 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
1889 additional discussion.
1891 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
1892 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
1893 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
1894 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
1895 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
1896 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
1897 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
1898 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
1899 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
1900 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
1901 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
1902 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
1903 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
1904 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
1905 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
1906 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
1907 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
1908 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
1909 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
1910 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
1911 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
1912 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
1913 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
1914 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
1917 – Warsaw, 2020-03-06
1921 * Support for the cpuset cgroups v2 controller has been added.
1922 Processes may be restricted to specific CPUs using the new
1923 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
1924 AllowedMemoryNodes= setting.
1926 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
1927 now be configured using a new RestartKillSignal= setting. This
1928 allows units which signals to request termination to implement
1929 different behaviour when stopping in preparation for a restart.
1931 * "systemctl clean" may now be used also for socket, mount, and swap
1934 * systemd will also read configuration options from the EFI variable
1935 SystemdOptions. This may be used to configure systemd behaviour when
1936 modifying the kernel command line is inconvenient, but configuration
1937 on disk is read too late, for example for the options related to
1938 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
1939 set the EFI variable.
1941 * systemd will now disable printk ratelimits in early boot. This should
1942 allow us to capture more logs from the early boot phase where normal
1943 storage is not available and the kernel ring buffer is used for
1944 logging. Configuration on the kernel command line has higher priority
1945 and overrides the systemd setting.
1947 systemd programs which log to /dev/kmsg directly use internal
1948 ratelimits to prevent runaway logging. (Normally this is only used
1949 during early boot, so in practice this change has very little
1952 * Unit files now support top level dropin directories of the form
1953 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
1954 that affects all corresponding unit files.
1956 * systemctl gained support for 'stop --job-mode=triggering' which will
1957 stop the specified unit and any units which could trigger it.
1959 * Unit status display now includes units triggering and triggered by
1960 the unit being shown.
1962 * The RuntimeMaxSec= setting is now supported by scopes, not just
1963 .service units. This is particularly useful for PAM sessions which
1964 create a scope unit for the user login. systemd.runtime_max_sec=
1965 setting may used with the pam_systemd module to limit the duration
1966 of the PAM session, for example for time-limited logins.
1968 * A new @pkey system call group is now defined to make it easier to
1969 allow-list memory protection syscalls for containers and services
1970 which need to use them.
1972 * systemd-udevd: removed the 30s timeout for killing stale workers on
1973 exit. systemd-udevd now waits for workers to finish. The hard-coded
1974 exit timeout of 30s was too short for some large installations, where
1975 driver initialization could be prematurely interrupted during initrd
1976 processing if the root file system had been mounted and init was
1977 preparing to switch root. If udevd is run without systemd and workers
1978 are hanging while udevd receives an exit signal, udevd will now exit
1979 when udev.event_timeout is reached for the last hanging worker. With
1980 systemd, the exit timeout can additionally be configured using
1981 TimeoutStopSec= in systemd-udevd.service.
1983 * udev now provides a program (fido_id) that identifies FIDO CTAP1
1984 ("U2F")/CTAP2 security tokens based on the usage declared in their
1985 report and descriptor and outputs suitable environment variables.
1986 This replaces the externally maintained allow lists of all known
1987 security tokens that were used previously.
1989 * Automatically generated autosuspend udev rules for allow-listed
1990 devices have been imported from the Chromium OS project. This should
1991 improve power saving with many more devices.
1993 * udev gained a new "CONST{key}=value" setting that allows matching
1994 against system-wide constants without forking a helper binary.
1995 Currently "arch" and "virt" keys are supported.
1997 * udev now opens CDROMs in non-exclusive mode when querying their
1998 capabilities. This should fix issues where other programs trying to
1999 use the CDROM cannot gain access to it, but carries a risk of
2000 interfering with programs writing to the disk, if they did not open
2001 the device in exclusive mode as they should.
2003 * systemd-networkd does not create a default route for IPv4 link local
2004 addressing anymore. The creation of the route was unexpected and was
2005 breaking routing in various cases, but people who rely on it being
2006 created implicitly will need to adjust. Such a route may be requested
2007 with DefaultRouteOnDevice=yes.
2009 Similarly, systemd-networkd will not assign a link-local IPv6 address
2010 when IPv6 link-local routing is not enabled.
2012 * Receive and transmit buffers may now be configured on links with
2013 the new RxBufferSize= and TxBufferSize= settings.
2015 * systemd-networkd may now advertise additional IPv6 routes. A new
2016 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
2019 * systemd-networkd may now configure "next hop" routes using the
2020 [NextHop] section and Gateway= and Id= settings.
2022 * systemd-networkd will now retain DHCP config on restarts by default
2023 (but this may be overridden using the KeepConfiguration= setting).
2024 The default for SendRelease= has been changed to true.
2026 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
2027 received from the server.
2029 The client will use the received SIP server list if UseSIP=yes is
2032 The client may be configured to request specific options from the
2033 server using a new RequestOptions= setting.
2035 The client may be configured to send arbitrary options to the server
2036 using a new SendOption= setting.
2038 A new IPServiceType= setting has been added to configure the "IP
2039 service type" value used by the client.
2041 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
2042 request prefix hints in the DHCPv6 solicitation.
2044 * The DHCPv4 server may be configured to send arbitrary options using
2045 a new SendOption= setting.
2047 * The DHCPv4 server may now be configured to emit SIP server list using
2048 the new EmitSIP= and SIP= settings.
2050 * systemd-networkd and networkctl may now renew DHCP leases on demand.
2051 networkctl has a new 'networkctl renew' verb.
2053 * systemd-networkd may now reconfigure links on demand. networkctl
2054 gained two new verbs: "reload" will reload the configuration, and
2055 "reconfigure DEVICE…" will reconfigure one or more devices.
2057 * .network files may now match on SSID and BSSID of a wireless network,
2058 i.e. the access point name and hardware address using the new SSID=
2059 and BSSID= options. networkctl will display the current SSID and
2060 BSSID for wireless links.
2062 .network files may also match on the wireless network type using the
2063 new WLANInterfaceType= option.
2065 * systemd-networkd now includes default configuration that enables
2066 link-local addressing when connected to an ad-hoc wireless network.
2068 * systemd-networkd may configure the Traffic Control queueing
2069 disciplines in the kernel using the new
2070 [TrafficControlQueueingDiscipline] section and Parent=,
2071 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
2072 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
2073 NetworkEmulatorDuplicateRate= settings.
2075 * systemd-tmpfiles gained a new w+ setting to append to files.
2077 * systemd-analyze dump will now report when the memory configuration in
2078 the kernel does not match what systemd has configured (usually,
2079 because some external program has modified the kernel configuration
2082 * systemd-analyze gained a new --base-time= switch instructs the
2083 'calendar' verb to resolve times relative to that timestamp instead
2084 of the present time.
2086 * journalctl --update-catalog now produces deterministic output (making
2087 reproducible image builds easier).
2089 * A new devicetree-overlay setting is now documented in the Boot Loader
2092 * The default value of the WatchdogSec= setting used in systemd
2093 services (the ones bundled with the project itself) may be set at
2094 configuration time using the -Dservice-watchdog= setting. If set to
2095 empty, the watchdogs will be disabled.
2097 * systemd-resolved validates IP addresses in certificates now when GnuTLS
2100 * libcryptsetup >= 2.0.1 is now required.
2102 * A configuration option -Duser-path= may be used to override the $PATH
2103 used by the user service manager. The default is again to use the same
2104 path as the system manager.
2106 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
2107 outputting the 128bit IDs in UUID format (i.e. in the "canonical
2110 * Service units gained a new sandboxing option ProtectKernelLogs= which
2111 makes sure the program cannot get direct access to the kernel log
2112 buffer anymore, i.e. the syslog() system call (not to be confused
2113 with the API of the same name in libc, which is not affected), the
2114 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
2115 inaccessible to the service. It's recommended to enable this setting
2116 for all services that should not be able to read from or write to the
2117 kernel log buffer, which are probably almost all.
2119 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
2120 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
2121 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
2122 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
2123 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
2124 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
2125 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
2126 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
2127 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
2128 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
2129 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
2130 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
2131 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
2132 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
2133 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
2134 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
2135 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
2136 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
2137 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
2138 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
2139 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
2140 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
2141 Zach Smith, Zbigniew Jędrzejewski-Szmek
2143 – Warsaw, 2019-11-29
2147 * This release enables unprivileged programs (i.e. requiring neither
2148 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
2149 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
2150 kernel for the whole UNIX group range, i.e. all processes. This
2151 change should be reasonably safe, as the kernel support for it was
2152 specifically implemented to allow safe access to ICMP Echo for
2153 processes lacking any privileges. If this is not desirable, it can be
2154 disabled again by setting the parameter to "1 0".
2156 * Previously, filters defined with SystemCallFilter= would have the
2157 effect that any calling of an offending system call would terminate
2158 the calling thread. This behaviour never made much sense, since
2159 killing individual threads of unsuspecting processes is likely to
2160 create more problems than it solves. With this release the default
2161 action changed from killing the thread to killing the whole
2162 process. For this to work correctly both a kernel version (>= 4.14)
2163 and a libseccomp version (>= 2.4.0) supporting this new seccomp
2164 action is required. If an older kernel or libseccomp is used the old
2165 behaviour continues to be used. This change does not affect any
2166 services that have no system call filters defined, or that use
2167 SystemCallErrorNumber= (and thus see EPERM or another error instead
2168 of being killed when calling an offending system call). Note that
2169 systemd documentation always claimed that the whole process is
2170 killed. With this change behaviour is thus adjusted to match the
2173 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
2174 4194304 by default, i.e. the full 22bit range the kernel allows, up
2175 from the old 16bit range. This should improve security and
2176 robustness, as PID collisions are made less likely (though certainly
2177 still possible). There are rumours this might create compatibility
2178 problems, though at this moment no practical ones are known to
2179 us. Downstream distributions are hence advised to undo this change in
2180 their builds if they are concerned about maximum compatibility, but
2181 for everybody else we recommend leaving the value bumped. Besides
2182 improving security and robustness this should also simplify things as
2183 the maximum number of allowed concurrent tasks was previously bounded
2184 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
2185 only a single knob is left ("kernel.threads-max"). There have been
2186 concerns that usability is affected by this change because larger PID
2187 numbers are harder to type, but we believe the change from 5 digits
2188 to 7 digits doesn't hamper usability.
2190 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
2191 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
2192 hierarchically set default memory protection values for a particular
2193 subtree of the unit hierarchy.
2195 * Memory protection directives can now take a value of zero, allowing
2196 explicit opting out of a default value propagated by an ancestor.
2198 * systemd now defaults to the "unified" cgroup hierarchy setup during
2199 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
2200 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
2201 change reflects the fact that cgroupsv2 support has matured
2202 substantially in both systemd and in the kernel, and is clearly the
2203 way forward. Downstream production distributions might want to
2204 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
2205 their builds as unfortunately the popular container managers have not
2206 caught up with the kernel API changes.
2208 * Man pages are not built by default anymore (html pages were already
2209 disabled by default), to make development builds quicker. When
2210 building systemd for a full installation with documentation, meson
2211 should be called with -Dman=true and/or -Dhtml=true as appropriate.
2212 The default was changed based on the assumption that quick one-off or
2213 repeated development builds are much more common than full optimized
2214 builds for installation, and people need to pass various other
2215 options to when doing "proper" builds anyway, so the gain from making
2216 development builds quicker is bigger than the one time disruption for
2219 Two scripts are created in the *build* directory to generate and
2220 preview man and html pages on demand, e.g.:
2222 build/man/man systemctl
2223 build/man/html systemd.index
2225 * libidn2 is used by default if both libidn2 and libidn are installed.
2226 Please use -Dlibidn=true if libidn is preferred.
2228 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
2229 big-endian machines. Before, bytes were written and read in native
2230 machine order as exposed by the native libc __cpu_mask interface.
2231 Now, little-endian order is always used (CPUs 0–7 are described by
2232 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
2233 This change fixes D-Bus calls that cross endianness boundary.
2235 The presentation format used for CPUAffinity= by "systemctl show" and
2236 "systemd-analyze dump" is changed to present CPU indices instead of
2237 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
2238 shown as CPUAffinity=03000000000000000000000000000… (on
2239 little-endian) or CPUAffinity=00000000000000300000000000000… (on
2240 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
2241 input format. The maximum integer that will be printed in the new
2242 format is 8191 (four digits), while the old format always used a very
2243 long number (with the length varying by architecture), so they can be
2244 unambiguously distinguished.
2246 * /usr/sbin/halt.local is no longer supported. Implementation in
2247 distributions was inconsistent and it seems this functionality was
2250 To replace this functionality, users should:
2251 - either define a new unit and make it a dependency of final.target
2252 (systemctl add-wants final.target my-halt-local.service)
2253 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
2254 and ensure that it accepts "halt", "poweroff", "reboot", and
2255 "kexec" as an argument, see the description in systemd-shutdown(8).
2257 * When a [Match] section in .link or .network file is empty (contains
2258 no match patterns), a warning will be emitted. Please add any "match
2259 all" pattern instead, e.g. OriginalName=* or Name=* in case all
2260 interfaces should really be matched.
2262 * A new setting NUMAPolicy= may be used to set process memory
2263 allocation policy. This setting can be specified in
2264 /etc/systemd/system.conf and hence will set the default policy for
2265 PID1. The default policy can be overridden on a per-service
2266 basis. The related setting NUMAMask= is used to specify NUMA node
2267 mask that should be associated with the selected policy.
2269 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
2270 generates when processes it manages are reaching their memory limits,
2271 and will place their units in a special state, and optionally kill or
2272 stop the whole unit.
2274 * The service manager will now expose bus properties for the IO
2275 resources used by units. This information is also shown in "systemctl
2276 status" now (for services that have IOAccounting=yes set). Moreover,
2277 the IO accounting data is included in the resource log message
2278 generated whenever a unit stops.
2280 * Units may now configure an explicit timeout to wait for when killed
2281 with SIGABRT, for example when a service watchdog is hit. Previously,
2282 the regular TimeoutStopSec= timeout was applied in this case too —
2283 now a separate timeout may be set using TimeoutAbortSec=.
2285 * Services may now send a special WATCHDOG=trigger message with
2286 sd_notify() to trigger an immediate "watchdog missed" event, and thus
2287 trigger service termination. This is useful both for testing watchdog
2288 handling, but also for defining error paths in services, that shall
2289 be handled the same way as watchdog events.
2291 * There are two new per-unit settings IPIngressFilterPath= and
2292 IPEgressFilterPath= which allow configuration of a BPF program
2293 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
2294 to apply to the IP packet ingress/egress path of all processes of a
2295 unit. This is useful to allow running systemd services with BPF
2296 programs set up externally.
2298 * systemctl gained a new "clean" verb for removing the state, cache,
2299 runtime or logs directories of a service while it is terminated. The
2300 new verb may also be used to remove the state maintained on disk for
2301 timer units that have Persistent= configured.
2303 * During the last phase of shutdown systemd will now automatically
2304 increase the log level configured in the "kernel.printk" sysctl so
2305 that any relevant loggable events happening during late shutdown are
2306 made visible. Previously, loggable events happening so late during
2307 shutdown were generally lost if the "kernel.printk" sysctl was set to
2308 high thresholds, as regular logging daemons are terminated at that
2309 time and thus nothing is written to disk.
2311 * If processes terminated during the last phase of shutdown do not exit
2312 quickly systemd will now show their names after a short time, to make
2313 debugging easier. After a longer timeout they are forcibly killed,
2316 * journalctl (and the other tools that display logs) will now highlight
2317 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
2318 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
2319 are now shown in blue color, to separate them visually from regular
2320 logs. References to configuration files are now turned into clickable
2321 links on terminals that support that.
2323 * systemd-journald will now stop logging to /var/log/journal during
2324 shutdown when /var/ is on a separate mount, so that it can be
2325 unmounted safely during shutdown.
2327 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
2329 * systemd-resolved "Cache=" configuration option in resolved.conf has
2330 been extended to also accept the 'no-negative' value. Previously,
2331 only a boolean option was allowed (yes/no), having yes as the
2332 default. If this option is set to 'no-negative', negative answers are
2333 not cached while the old cache heuristics are used positive answers.
2334 The default remains unchanged.
2336 * The predictable naming scheme for network devices now supports
2337 generating predictable names for "netdevsim" devices.
2339 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
2342 Those two changes form a new net.naming-policy-scheme= entry.
2343 Distributions which want to preserve naming stability may want to set
2344 the -Ddefault-net-naming-scheme= configuration option.
2346 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
2347 interfaces natively.
2349 * systemd-networkd's bridge FDB support now allows configuration of a
2350 destination address for each entry (Destination=), as well as the
2351 VXLAN VNI (VNI=), as well as an option to declare what an entry is
2352 associated with (AssociatedWith=).
2354 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
2355 option for configuring the maximum number of DHCP lease requests. It
2356 also learnt a new BlackList= option for deny-listing DHCP servers (a
2357 similar setting has also been added to the IPv6 RA client), as well
2358 as a SendRelease= option for configuring whether to send a DHCP
2359 RELEASE message when terminating.
2361 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
2362 separately in the [DHCPv4] and [DHCPv6] sections.
2364 * systemd-networkd's DHCP support will now optionally create an
2365 implicit host route to the DNS server specified in the DHCP lease, in
2366 addition to the routes listed explicitly in the lease. This should
2367 ensure that in multi-homed systems DNS traffic leaves the systems on
2368 the interface that acquired the DNS server information even if other
2369 routes such as default routes exist. This behaviour may be turned on
2370 with the new RoutesToDNS= option.
2372 * systemd-networkd's VXLAN support gained a new option
2373 GenericProtocolExtension= for enabling VXLAN Generic Protocol
2374 Extension support, as well as IPDoNotFragment= for setting the IP
2375 "Don't fragment" bit on outgoing packets. A similar option has been
2376 added to the GENEVE support.
2378 * In systemd-networkd's [Route] section you may now configure
2379 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
2380 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
2381 propagation. The Type= setting now supports local, broadcast,
2382 anycast, multicast, any, xresolve routes, too.
2384 * systemd-networkd's [Network] section learnt a new option
2385 DefaultRouteOnDevice= for automatically configuring a default route
2386 onto the network device.
2388 * systemd-networkd's bridging support gained two new options ProxyARP=
2389 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
2390 MulticastRouter= for configuring multicast routing behaviour. A new
2391 option MulticastIGMPVersion= may be used to change bridge's multicast
2392 Internet Group Management Protocol (IGMP) version.
2394 * systemd-networkd's FooOverUDP support gained the ability to configure
2395 local and peer IP addresses via Local= and Peer=. A new option
2396 PeerPort= may be used to configure the peer's IP port.
2398 * systemd-networkd's TUN support gained a new setting VnetHeader= for
2399 tweaking Generic Segment Offload support.
2401 * The address family for policy rules may be specified using the new
2402 Family= option in the [RoutingPolicyRule] section.
2404 * networkctl gained a new "delete" command for removing virtual network
2405 devices, as well as a new "--stats" switch for showing device
2408 * networkd.conf gained a new setting SpeedMeter= and
2409 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
2410 measured speed may be shown by 'networkctl status'.
2412 * "networkctl status" now displays MTU and queue lengths, and more
2413 detailed information about VXLAN and bridge devices.
2415 * systemd-networkd's .network and .link files gained a new Property=
2416 setting in the [Match] section, to match against devices with
2417 specific udev properties.
2419 * systemd-networkd's tunnel support gained a new option
2420 AssignToLoopback= for selecting whether to use the loopback device
2421 "lo" as underlying device.
2423 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
2424 been renamed to LinkLayerAddress=, and it now allows configuration of
2427 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
2428 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
2429 IPv6 configuration (static or DHCPv6) was found for a given
2430 interface. It will not touch the sysctl otherwise.
2432 * The order of entries is $PATH used by the user manager instance was
2433 changed to put bin/ entries before the corresponding sbin/ entries.
2434 It is recommended to not rely on this order, and only ever have one
2435 binary with a given name in the system paths under /usr.
2437 * A new tool systemd-network-generator has been added that may generate
2438 .network, .netdev and .link files from IP configuration specified on
2439 the kernel command line in the format used by Dracut.
2441 * The CriticalConnection= setting in .network files is now deprecated,
2442 and replaced by a new KeepConfiguration= setting which allows more
2443 detailed configuration of the IP configuration to keep in place.
2445 * systemd-analyze gained a few new verbs:
2447 - "systemd-analyze timestamp" parses and converts timestamps. This is
2448 similar to the existing "systemd-analyze calendar" command which
2449 does the same for recurring calendar events.
2451 - "systemd-analyze timespan" parses and converts timespans (i.e.
2452 durations as opposed to points in time).
2454 - "systemd-analyze condition" will parse and test ConditionXYZ=
2457 - "systemd-analyze exit-status" will parse and convert exit status
2458 codes to their names and back.
2460 - "systemd-analyze unit-files" will print a list of all unit
2461 file paths and unit aliases.
2463 * SuccessExitStatus=, RestartPreventExitStatus=, and
2464 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
2465 is equivalent to "65"). Those exit status name mappings may be
2466 displayed with the systemd-analyze exit-status verb describe above.
2468 * systemd-logind now exposes a per-session SetBrightness() bus call,
2469 which may be used to securely change the brightness of a kernel
2470 brightness device, if it belongs to the session's seat. By using this
2471 call unprivileged clients can make changes to "backlight" and "leds"
2472 devices securely with strict requirements on session membership.
2473 Desktop environments may use this to generically make brightness
2474 changes to such devices without shipping private SUID binaries or
2475 udev rules for that purpose.
2477 * "udevadm info" gained a --wait-for-initialization switch to wait for
2478 a device to be initialized.
2480 * systemd-hibernate-resume-generator will now look for resumeflags= on
2481 the kernel command line, which is similar to rootflags= and may be
2482 used to configure device timeout for the hibernation device.
2484 * sd-event learnt a new API call sd_event_source_disable_unref() for
2485 disabling and unref'ing an event source in a single function. A
2486 related call sd_event_source_disable_unrefp() has been added for use
2487 with gcc's cleanup extension.
2489 * The sd-id128.h public API gained a new definition
2490 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
2493 * "busctl introspect" gained a new switch --xml-interface for dumping
2494 XML introspection data unmodified.
2496 * PID 1 may now show the unit name instead of the unit description
2497 string in its status output during boot. This may be configured in
2498 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
2499 kernel command line option systemd.status_unit_format=.
2501 * PID 1 now understands a new option KExecWatchdogSec= in
2502 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
2503 Previously watchdog functionality was only available for regular
2504 reboots. The new setting defaults to off, because we don't know in
2505 the general case if the watchdog will be reset after kexec (some
2506 drivers do reset it, but not all), and the new userspace might not be
2507 configured to handle the watchdog.
2509 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
2510 RebootWatchdogSec= to more clearly communicate what it is about. The
2511 old name is still accepted for compatibility.
2513 * The systemd.debug_shell kernel command line option now optionally
2514 takes a tty name to spawn the debug shell on, which allows a
2515 different tty to be selected than the built-in default.
2517 * Service units gained a new ExecCondition= setting which will run
2518 before ExecStartPre= and either continue execution of the unit (for
2519 clean exit codes), stop execution without marking the unit failed
2520 (for exit codes 1 through 254), or stop execution and fail the unit
2521 (for exit code 255 or abnormal termination).
2523 * A new service systemd-pstore.service has been added that pulls data
2524 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
2527 * timedatectl gained new verbs for configuring per-interface NTP
2528 service configuration for systemd-timesyncd.
2530 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
2531 2019. (You can set non-UTF-8 locales though, if you know their name.)
2533 * If variable assignments in sysctl.d/ files are prefixed with "-" any
2534 failures to apply them are now ignored.
2536 * systemd-random-seed.service now optionally credits entropy when
2537 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
2538 true for the service to enable this behaviour, but please consult the
2539 documentation first, since this comes with a couple of caveats.
2541 * systemd-random-seed.service is now a synchronization point for full
2542 initialization of the kernel's entropy pool. Services that require
2543 /dev/urandom to be correctly initialized should be ordered after this
2546 * The systemd-boot boot loader has been updated to optionally maintain
2547 a random seed file in the EFI System Partition (ESP). During the boot
2548 phase, this random seed is read and updated with a new seed
2549 cryptographically derived from it. Another derived seed is passed to
2550 the OS. The latter seed is then credited to the kernel's entropy pool
2551 very early during userspace initialization (from PID 1). This allows
2552 systems to boot up with a fully initialized kernel entropy pool from
2553 earliest boot on, and thus entirely removes all entropy pool
2554 initialization delays from systems using systemd-boot. Special care
2555 is taken to ensure different seeds are derived on system images
2556 replicated to multiple systems. "bootctl status" will show whether
2557 a seed was received from the boot loader.
2559 * bootctl gained two new verbs:
2561 - "bootctl random-seed" will generate the file in ESP and an EFI
2562 variable to allow a random seed to be passed to the OS as described
2565 - "bootctl is-installed" checks whether systemd-boot is currently
2568 * bootctl will warn if it detects that boot entries are misconfigured
2569 (for example if the kernel image was removed without purging the
2572 * A new document has been added describing systemd's use and support
2573 for the kernel's entropy pool subsystem:
2575 https://systemd.io/RANDOM_SEEDS
2577 * When the system is hibernated the swap device to write the
2578 hibernation image to is now automatically picked from all available
2579 swap devices, preferring the swap device with the highest configured
2580 priority over all others, and picking the device with the most free
2581 space if there are multiple devices with the highest priority.
2583 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
2584 option that permits selecting the timeout how long to wait for a
2585 device with an encryption key before asking for the password.
2587 * IOWeight= has learnt to properly set the IO weight when using the
2588 BFQ scheduler officially found in kernels 5.0+.
2590 * A new mailing list has been created for reporting of security issues:
2591 systemd-security@redhat.com. For mode details, see
2592 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
2594 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
2595 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
2596 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
2597 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
2598 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
2599 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
2600 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
2601 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
2602 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
2603 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
2604 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
2605 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
2606 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
2607 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
2608 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
2609 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
2610 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
2611 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
2612 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
2613 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
2614 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
2615 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
2616 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
2617 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
2618 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
2619 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
2620 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
2621 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
2622 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
2623 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
2625 – Camerino, 2019-09-03
2629 * In .link files, MACAddressPolicy=persistent (the default) is changed
2630 to cover more devices. For devices like bridges, tun, tap, bond, and
2631 similar interfaces that do not have other identifying information,
2632 the interface name is used as the basis for persistent seed for MAC
2633 and IPv4LL addresses. The way that devices that were handled
2634 previously is not changed, and this change is about covering more
2635 devices then previously by the "persistent" policy.
2637 MACAddressPolicy=random may be used to force randomized MACs and
2638 IPv4LL addresses for a device if desired.
2640 Hint: the log output from udev (at debug level) was enhanced to
2641 clarify what policy is followed and which attributes are used.
2642 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
2643 may be used to view this.
2645 Hint: if a bridge interface is created without any slaves, and gains
2646 a slave later, then now the bridge does not inherit slave's MAC.
2647 To inherit slave's MAC, for example, create the following file:
2649 # /etc/systemd/network/98-bridge-inherit-mac.link
2654 MACAddressPolicy=none
2657 * The .device units generated by systemd-fstab-generator and other
2658 generators do not automatically pull in the corresponding .mount unit
2659 as a Wants= dependency. This means that simply plugging in the device
2660 will not cause the mount unit to be started automatically. But please
2661 note that the mount unit may be started for other reasons, in
2662 particular if it is part of local-fs.target, and any unit which
2663 (transitively) depends on local-fs.target is started.
2665 * networkctl list/status/lldp now accept globbing wildcards for network
2666 interface names to match against all existing interfaces.
2668 * The $PIDFILE environment variable is set to point the absolute path
2669 configured with PIDFile= for processes of that service.
2671 * The fallback DNS server list was augmented with Cloudflare public DNS
2672 servers. Use `-Ddns-servers=` to set a different fallback.
2674 * A new special target usb-gadget.target will be started automatically
2675 when a USB Device Controller is detected (which means that the system
2676 is a USB peripheral).
2678 * A new unit setting CPUQuotaPeriodSec= assigns the time period
2679 relatively to which the CPU time quota specified by CPUQuota= is
2682 * A new unit setting ProtectHostname= may be used to prevent services
2683 from modifying hostname information (even if they otherwise would
2684 have privileges to do so).
2686 * A new unit setting NetworkNamespacePath= may be used to specify a
2687 namespace for service or socket units through a path referring to a
2688 Linux network namespace pseudo-file.
2690 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
2691 have an effect on .socket units: when used the listening socket is
2692 created within the configured network namespace instead of the host
2695 * ExecStart= command lines in unit files may now be prefixed with ':'
2696 in which case environment variable substitution is
2697 disabled. (Supported for the other ExecXYZ= settings, too.)
2699 * .timer units gained two new boolean settings OnClockChange= and
2700 OnTimezoneChange= which may be used to also trigger a unit when the
2701 system clock is changed or the local timezone is
2702 modified. systemd-run has been updated to make these options easily
2703 accessible from the command line for transient timers.
2705 * Two new conditions for units have been added: ConditionMemory= may be
2706 used to conditionalize a unit based on installed system
2707 RAM. ConditionCPUs= may be used to conditionalize a unit based on
2708 installed CPU cores.
2710 * The @default system call filter group understood by SystemCallFilter=
2711 has been updated to include the new rseq() system call introduced in
2714 * A new time-set.target has been added that indicates that the system
2715 time has been set from a local source (possibly imprecise). The
2716 existing time-sync.target is stronger and indicates that the time has
2717 been synchronized with a precise external source. Services where
2718 approximate time is sufficient should use the new target.
2720 * "systemctl start" (and related commands) learnt a new
2721 --show-transaction option. If specified brief information about all
2722 jobs queued because of the requested operation is shown.
2724 * systemd-networkd recognizes a new operation state 'enslaved', used
2725 (instead of 'degraded' or 'carrier') for interfaces which form a
2726 bridge, bond, or similar, and an new 'degraded-carrier' operational
2727 state used for the bond or bridge master interface when one of the
2728 enslaved devices is not operational.
2730 * .network files learnt the new IgnoreCarrierLoss= option for leaving
2731 networks configured even if the carrier is lost.
2733 * The RequiredForOnline= setting in .network files may now specify a
2734 minimum operational state required for the interface to be considered
2735 "online" by systemd-networkd-wait-online. Related to this
2736 systemd-networkd-wait-online gained a new option --operational-state=
2737 to configure the same, and its --interface= option was updated to
2738 optionally also take an operational state specific for an interface.
2740 * systemd-networkd-wait-online gained a new setting --any for waiting
2741 for only one of the requested interfaces instead of all of them.
2743 * systemd-networkd now implements L2TP tunnels.
2745 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
2746 may be used to cause autonomous and onlink prefixes received in IPv6
2747 Router Advertisements to be ignored.
2749 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
2750 file settings may be used to tweak bridge behaviour.
2752 * The new TripleSampling= option in .network files may be used to
2753 configure CAN triple sampling.
2755 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
2756 used to point to private or preshared key for a WireGuard interface.
2758 * /etc/crypttab now supports the same-cpu-crypt and
2759 submit-from-crypt-cpus options to tweak encryption work scheduling
2762 * systemd-tmpfiles will now take a BSD file lock before operating on a
2763 contents of directory. This may be used to temporarily exclude
2764 directories from aging by taking the same lock (useful for example
2765 when extracting a tarball into /tmp or /var/tmp as a privileged user,
2766 which might create files with really old timestamps, which
2767 nevertheless should not be deleted). For further details, see:
2769 https://systemd.io/TEMPORARY_DIRECTORIES
2771 * systemd-tmpfiles' h line type gained support for the
2772 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
2773 controlling project quota inheritance.
2775 * sd-boot and bootctl now implement support for an Extended Boot Loader
2776 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
2777 addition to the ESP partition mounted to /efi or /boot/efi.
2778 Configuration file fragments, kernels, initrds and other EFI images
2779 to boot will be loaded from both the ESP and XBOOTLDR partitions.
2780 The XBOOTLDR partition was previously described by the Boot Loader
2781 Specification, but implementation was missing in sd-boot. Support for
2782 this concept allows using the sd-boot boot loader in more
2783 conservative scenarios where the boot loader itself is placed in the
2784 ESP but the kernels to boot (and their metadata) in a separate
2787 * A system may now be booted with systemd.volatile=overlay on the
2788 kernel command line, which causes the root file system to be set up
2789 an overlayfs mount combining the root-only root directory with a
2790 writable tmpfs. In this setup, the underlying root device is not
2791 modified, and any changes are lost at reboot.
2793 * Similar, systemd-nspawn can now boot containers with a volatile
2794 overlayfs root with the new --volatile=overlay switch.
2796 * systemd-nspawn can now consume OCI runtime bundles using a new
2797 --oci-bundle= option. This implementation is fully usable, with most
2798 features in the specification implemented, but since this a lot of
2799 new code and functionality, this feature should most likely not
2800 be used in production yet.
2802 * systemd-nspawn now supports various options described by the OCI
2803 runtime specification on the command-line and in .nspawn files:
2804 --inaccessible=/Inaccessible= may be used to mask parts of the file
2805 system tree, --console=/--pipe may be used to configure how standard
2806 input, output, and error are set up.
2808 * busctl learned the `emit` verb to generate D-Bus signals.
2810 * systemd-analyze cat-config may be used to gather and display
2811 configuration spread over multiple files, for example system and user
2812 presets, tmpfiles.d, sysusers.d, udev rules, etc.
2814 * systemd-analyze calendar now takes an optional new parameter
2815 --iterations= which may be used to show a maximum number of iterations
2816 the specified expression will elapse next.
2818 * The sd-bus C API gained support for naming method parameters in the
2821 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
2822 the reboot() system call expects.
2824 * journalctl learnt a new --cursor-file= option that points to a file
2825 from which a cursor should be loaded in the beginning and to which
2826 the updated cursor should be stored at the end.
2828 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
2829 detected by systemd-detect-virt (and may also be used in
2830 ConditionVirtualization=).
2832 * The behaviour of systemd-logind may now be modified with environment
2833 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
2834 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
2835 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
2836 skip the relevant operation completely (when set to false), or to
2837 create a flag file in /run/systemd (when set to true), instead of
2838 actually commencing the real operation when requested. The presence
2839 of /run/systemd/reboot-to-firmware-setup,
2840 /run/systemd/reboot-to-boot-loader-menu, and
2841 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
2842 boot loader implementations to replace some steps logind performs
2843 during reboot with their own operations.
2845 * systemctl can be used to request a reboot into the boot loader menu
2846 or a specific boot loader entry with the new --boot-load-menu= and
2847 --boot-loader-entry= options to a reboot command. (This requires a
2848 boot loader that supports this, for example sd-boot.)
2850 * kernel-install will no longer unconditionally create the output
2851 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
2852 snippets, but will do only if the machine-specific parent directory
2853 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
2854 to create this parent directory during sd-boot installation.
2856 This makes it easier to use kernel-install with plugins which support
2857 a different layout of the bootloader partitions (for example grub2).
2859 * During package installation (with `ninja install`), we would create
2860 symlinks for getty@tty1.service, systemd-networkd.service,
2861 systemd-networkd.socket, systemd-resolved.service,
2862 remote-cryptsetup.target, remote-fs.target,
2863 systemd-networkd-wait-online.service, and systemd-timesyncd.service
2864 in /etc, as if `systemctl enable` was called for those units, to make
2865 the system usable immediately after installation. Now this is not
2866 done anymore, and instead calling `systemctl preset-all` is
2867 recommended after the first installation of systemd.
2869 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
2870 is built on seccomp. When turned on creation of SUID/SGID files is
2873 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
2874 implied if DynamicUser= is turned on for a service. This hardens
2875 these services, so that they neither can benefit from nor create
2876 SUID/SGID executables. This is a minor compatibility breakage, given
2877 that when DynamicUser= was first introduced SUID/SGID behaviour was
2878 unaffected. However, the security benefit of these two options is
2879 substantial, and the setting is still relatively new, hence we opted
2880 to make it mandatory for services with dynamic users.
2882 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
2883 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
2884 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
2885 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
2886 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
2887 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
2888 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
2889 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
2890 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
2891 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
2892 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
2893 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
2894 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
2895 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
2896 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
2897 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
2898 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
2899 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2901 — Warsaw, 2019-04-11
2905 * The default locale can now be configured at compile time. Otherwise,
2906 a suitable default will be selected automatically (one of C.UTF-8,
2907 en_US.UTF-8, and C).
2909 * The version string shown by systemd and other tools now includes the
2910 git commit hash when built from git. An override may be specified
2911 during compilation, which is intended to be used by distributions to
2912 include the package release information.
2914 * systemd-cat can now filter standard input and standard error streams
2915 for different syslog priorities using the new --stderr-priority=
2918 * systemd-journald and systemd-journal-remote reject entries which
2919 contain too many fields (CVE-2018-16865) and set limits on the
2920 process' command line length (CVE-2018-16864).
2922 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
2925 * A new network device NamePolicy "keep" is implemented for link files,
2926 and used by default in 99-default.link (the fallback configuration
2927 provided by systemd). With this policy, if the network device name
2928 was already set by userspace, the device will not be renamed again.
2929 This matches the naming scheme that was implemented before
2930 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
2931 is also enabled by default, even if not specified. Effectively, this
2932 means that if naming-scheme >= 240 is specified, network devices will
2933 be renamed according to the configuration, even if they have been
2934 renamed already, if "keep" is not specified as the naming policy in
2935 the .link file. The 99-default.link file provided by systemd includes
2936 "keep" for backwards compatibility, but it is recommended for user
2937 installed .link files to *not* include it.
2939 The "kernel" policy, which keeps kernel names declared to be
2940 "persistent", now works again as documented.
2942 * kernel-install script now optionally takes the paths to one or more
2943 initrd files, and passes them to all plugins.
2945 * The mincore() system call has been dropped from the @system-service
2946 system call filter group, as it is pretty exotic and may potentially
2947 used for side-channel attacks.
2949 * -fPIE is dropped from compiler and linker options. Please specify
2950 -Db_pie=true option to meson to build position-independent
2951 executables. Note that the meson option is supported since meson-0.49.
2953 * The fs.protected_regular and fs.protected_fifos sysctls, which were
2954 added in Linux 4.19 to make some data spoofing attacks harder, are
2955 now enabled by default. While this will hopefully improve the
2956 security of most installations, it is technically a backwards
2957 incompatible change; to disable these sysctls again, place the
2958 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
2960 fs.protected_regular = 0
2961 fs.protected_fifos = 0
2963 Note that the similar hardlink and symlink protection has been
2964 enabled since v199, and may be disabled likewise.
2966 * The files read from the EnvironmentFile= setting in unit files now
2967 parse backslashes inside quotes literally, matching the behaviour of
2970 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
2971 now automatically become NOPs when run in a chroot() environment.
2973 * The tmpfiles.d/ "C" line type will now copy directory trees not only
2974 when the destination is so far missing, but also if it already exists
2975 as a directory and is empty. This is useful to cater for systems
2976 where directory trees are put together from multiple separate mount
2977 points but otherwise empty.
2979 * A new function sd_bus_close_unref() (and the associated
2980 sd_bus_close_unrefp()) has been added to libsystemd, that combines
2981 sd_bus_close() and sd_bus_unref() in one.
2983 * udevadm control learnt a new option for --ping for testing whether a
2984 systemd-udevd instance is running and reacting.
2986 * udevadm trigger learnt a new option for --wait-daemon for waiting
2987 systemd-udevd daemon to be initialized.
2989 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
2990 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
2991 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
2992 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
2993 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
2994 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
2995 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
2996 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
2997 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
2998 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
2999 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
3000 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
3001 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
3002 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
3003 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
3004 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
3005 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
3007 — Berlin, 2019-02-14
3011 * NoNewPrivileges=yes has been set for all long-running services
3012 implemented by systemd. Previously, this was problematic due to
3013 SELinux (as this would also prohibit the transition from PID1's label
3014 to the service's label). This restriction has since been lifted, but
3015 an SELinux policy update is required.
3016 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
3018 * DynamicUser=yes is dropped from systemd-networkd.service,
3019 systemd-resolved.service and systemd-timesyncd.service, which was
3020 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
3021 and since v236 for systemd-timesyncd.service. The users and groups
3022 systemd-network, systemd-resolve and systemd-timesync are created
3023 by systemd-sysusers again. Distributors or system administrators
3024 may need to create these users and groups if they not exist (or need
3025 to re-enable DynamicUser= for those units) while upgrading systemd.
3026 Also, the clock file for systemd-timesyncd may need to move from
3027 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
3029 * When unit files are loaded from disk, previously systemd would
3030 sometimes (depending on the unit loading order) load units from the
3031 target path of symlinks in .wants/ or .requires/ directories of other
3032 units. This meant that unit could be loaded from different paths
3033 depending on whether the unit was requested explicitly or as a
3034 dependency of another unit, not honouring the priority of directories
3035 in search path. It also meant that it was possible to successfully
3036 load and start units which are not found in the unit search path, as
3037 long as they were requested as a dependency and linked to from
3038 .wants/ or .requires/. The target paths of those symlinks are not
3039 used for loading units anymore and the unit file must be found in
3042 * A new service type has been added: Type=exec. It's very similar to
3043 Type=simple but ensures the service manager will wait for both fork()
3044 and execve() of the main service binary to complete before proceeding
3045 with follow-up units. This is primarily useful so that the manager
3046 propagates any errors in the preparation phase of service execution
3047 back to the job that requested the unit to be started. For example,
3048 consider a service that has ExecStart= set to a file system binary
3049 that doesn't exist. With Type=simple starting the unit would be
3050 considered instantly successful, as only fork() has to complete
3051 successfully and the manager does not wait for execve(), and hence
3052 its failure is seen "too late". With the new Type=exec service type
3053 starting the unit will fail, as the manager will wait for the
3054 execve() and notice its failure, which is then propagated back to the
3057 NOTE: with the next release 241 of systemd we intend to change the
3058 systemd-run tool to default to Type=exec for transient services
3059 started by it. This should be mostly safe, but in specific corner
3060 cases might result in problems, as the systemd-run tool will then
3061 block on NSS calls (such as user name look-ups due to User=) done
3062 between the fork() and execve(), which under specific circumstances
3063 might cause problems. It is recommended to specify "-p Type=simple"
3064 explicitly in the few cases where this applies. For regular,
3065 non-transient services (i.e. those defined with unit files on disk)
3066 we will continue to default to Type=simple.
3068 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
3069 userspace processes is set to 1024 (soft) and 4096
3070 (hard). Previously, systemd passed this on unmodified to all
3071 processes it forked off. With this systemd release the hard limit
3072 systemd passes on is increased to 512K, overriding the kernel's
3073 defaults and substantially increasing the number of simultaneous file
3074 descriptors unprivileged userspace processes can allocate. Note that
3075 the soft limit remains at 1024 for compatibility reasons: the
3076 traditional UNIX select() call cannot deal with file descriptors >=
3077 1024 and increasing the soft limit globally might thus result in
3078 programs unexpectedly allocating a high file descriptor and thus
3079 failing abnormally when attempting to use it with select() (of
3080 course, programs shouldn't use select() anymore, and prefer
3081 poll()/epoll, but the call unfortunately remains undeservedly popular
3082 at this time). This change reflects the fact that file descriptor
3083 handling in the Linux kernel has been optimized in more recent
3084 kernels and allocating large numbers of them should be much cheaper
3085 both in memory and in performance than it used to be. Programs that
3086 want to take benefit of the increased limit have to "opt-in" into
3087 high file descriptors explicitly by raising their soft limit. Of
3088 course, when they do that they must acknowledge that they cannot use
3089 select() anymore (and neither can any shared library they use — or
3090 any shared library used by any shared library they use and so on).
3091 Which default hard limit is most appropriate is of course hard to
3092 decide. However, given reports that ~300K file descriptors are used
3093 in real-life applications we believe 512K is sufficiently high as new
3094 default for now. Note that there are also reports that using very
3095 high hard limits (e.g. 1G) is problematic: some software allocates
3096 large arrays with one element for each potential file descriptor
3097 (Java, …) — a high hard limit thus triggers excessively large memory
3098 allocations in these applications. Hopefully, the new default of 512K
3099 is a good middle ground: higher than what real-life applications
3100 currently need, and low enough for avoid triggering excessively large
3101 allocations in problematic software. (And yes, somebody should fix
3104 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
3105 to the highest possible values, as separate accounting of file
3106 descriptors is no longer necessary, as memcg tracks them correctly as
3107 part of the memory accounting anyway. Thus, from the four limits on
3108 file descriptors currently enforced (fs.file-max, fs.nr_open,
3109 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
3110 and keep only the latter two. A set of build-time options
3111 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
3112 has been added to revert this change in behaviour, which might be
3113 an option for systems that turn off memcg in the kernel.
3115 * When no /etc/locale.conf file exists (and hence no locale settings
3116 are in place), systemd will now use the "C.UTF-8" locale by default,
3117 and set LANG= to it. This locale is supported by various
3118 distributions including Fedora, with clear indications that upstream
3119 glibc is going to make it available too. This locale enables UTF-8
3120 mode by default, which appears appropriate for 2018.
3122 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
3123 default. This effectively switches the RFC3704 Reverse Path filtering
3124 from Strict mode to Loose mode. This is more appropriate for hosts
3125 that have multiple links with routes to the same networks (e.g.
3126 a client with a Wi-Fi and Ethernet both connected to the internet).
3128 Consult the kernel documentation for details on this sysctl:
3129 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
3131 * CPUAccounting=yes no longer enables the CPU controller when using
3132 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
3133 statistics are now provided independently from the CPU controller.
3135 * Support for disabling a particular cgroup controller within a sub-tree
3136 has been added through the DisableControllers= directive.
3138 * cgroup_no_v1=all on the kernel command line now also implies
3139 using the unified cgroup hierarchy, unless one explicitly passes
3140 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
3142 * The new "MemoryMin=" unit file property may now be used to set the
3143 memory usage protection limit of processes invoked by the unit. This
3144 controls the cgroup v2 memory.min attribute. Similarly, the new
3145 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
3146 cgroup v2 io.latency cgroup property for configuring per-service I/O
3149 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
3150 to the cgroup v1 "devices" cgroup controller.
3152 * systemd-escape now is able to combine --unescape with --template. It
3153 also learnt a new option --instance for extracting and unescaping the
3154 instance part of a unit name.
3156 * sd-bus now provides the sd_bus_message_readv() which is similar to
3157 sd_bus_message_read() but takes a va_list object. The pair
3158 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
3159 has been added for configuring the default method call timeout to
3160 use. sd_bus_error_move() may be used to efficiently move the contents
3161 from one sd_bus_error structure to another, invalidating the
3162 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
3163 be used to control whether a bus connection object is automatically
3164 flushed when an sd-event loop is exited.
3166 * When processing classic BSD syslog log messages, journald will now
3167 save the original time-stamp string supplied in the new
3168 SYSLOG_TIMESTAMP= journal field. This permits consumers to
3169 reconstruct the original BSD syslog message more correctly.
3171 * StandardOutput=/StandardError= in service files gained support for
3172 new "append:…" parameters, for connecting STDOUT/STDERR of a service
3173 to a file, and appending to it.
3175 * The signal to use as last step of killing of unit processes is now
3176 configurable. Previously it was hard-coded to SIGKILL, which may now
3177 be overridden with the new KillSignal= setting. Note that this is the
3178 signal used when regular termination (i.e. SIGTERM) does not suffice.
3179 Similarly, the signal used when aborting a program in case of a
3180 watchdog timeout may now be configured too (WatchdogSignal=).
3182 * The XDG_SESSION_DESKTOP environment variable may now be configured in
3183 the pam_systemd argument line, using the new desktop= switch. This is
3184 useful to initialize it properly from a display manager without
3185 having to touch C code.
3187 * Most configuration options that previously accepted percentage values
3188 now also accept permille values with the '‰' suffix (instead of '%').
3190 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
3193 * systemd-resolved's configuration file resolved.conf gained a new
3194 option ReadEtcHosts= which may be used to turn off processing and
3195 honoring /etc/hosts entries.
3197 * The "--wait" switch may now be passed to "systemctl
3198 is-system-running", in which case the tool will synchronously wait
3199 until the system finished start-up.
3201 * hostnamed gained a new bus call to determine the DMI product UUID.
3203 * On x86-64 systemd will now prefer using the RDRAND processor
3204 instruction over /dev/urandom whenever it requires randomness that
3205 neither has to be crypto-grade nor should be reproducible. This
3206 should substantially reduce the amount of entropy systemd requests
3207 from the kernel during initialization on such systems, though not
3208 reduce it to zero. (Why not zero? systemd still needs to allocate
3209 UUIDs and such uniquely, which require high-quality randomness.)
3211 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
3212 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
3213 for forcing the "Other Information" bit in IPv6 RA messages. The
3214 bonding logic gained four new options AdActorSystemPriority=,
3215 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
3216 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
3217 shuffling of flows. The tunnel logic gained a new
3218 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
3219 Deployment. The policy rule logic gained four new options IPProtocol=,
3220 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
3221 support for the MulticastToUnicast= option. networkd also gained
3222 support for configuring static IPv4 ARP or IPv6 neighbor entries.
3224 * .preset files (as read by 'systemctl preset') may now be used to
3225 instantiate services.
3227 * /etc/crypttab now understands the sector-size= option to configure
3228 the sector size for an encrypted partition.
3230 * Key material for encrypted disks may now be placed on a formatted
3231 medium, and referenced from /etc/crypttab by the UUID of the file
3232 system, followed by "=" suffixed by the path to the key file.
3234 * The "collect" udev component has been removed without replacement, as
3235 it is neither used nor maintained.
3237 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
3238 LogsDirectory=, ConfigurationDirectory= settings are used in a
3239 service the executed processes will now receive a set of environment
3240 variables containing the full paths of these directories.
3241 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
3242 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
3243 are used. Note that these options may be used multiple times per
3244 service in which case the resulting paths will be concatenated and
3245 separated by colons.
3247 * Predictable interface naming has been extended to cover InfiniBand
3248 NICs. They will be exposed with an "ib" prefix.
3250 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
3251 which case the respective line failing is ignored.
3253 * .link files may now be used to configure the equivalent to the
3254 "ethtool advertise" commands.
3256 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
3257 alternative to libudev.h. Previously, the latter was just an internal
3258 wrapper around the former, but now these two APIs are exposed
3261 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
3262 which calculates an app-specific boot ID similar to how
3263 sd_id128_get_machine_app_specific() generates an app-specific machine
3266 * A new tool systemd-id128 has been added that can be used to determine
3267 and generate various 128bit IDs.
3269 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
3272 * systemd-hibernate-resume-generator will now honor the "noresume"
3273 kernel command line option, in which case it will bypass resuming
3274 from any hibernated image.
3276 * The systemd-sleep.conf configuration file gained new options
3277 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
3278 AllowHybridSleep= for prohibiting specific sleep modes even if the
3279 kernel exports them.
3281 * portablectl is now officially supported and has thus moved to
3284 * bootctl learnt the two new commands "set-default" and "set-oneshot"
3285 for setting the default boot loader item to boot to (either
3286 persistently or only for the next boot). This is currently only
3287 compatible with sd-boot, but may be implemented on other boot loaders
3288 too, that follow the boot loader interface. The updated interface is
3289 now documented here:
3291 https://systemd.io/BOOT_LOADER_INTERFACE
3293 * A new kernel command line option systemd.early_core_pattern= is now
3294 understood which may be used to influence the core_pattern PID 1
3295 installs during early boot.
3297 * busctl learnt two new options -j and --json= for outputting method
3298 call replies, properties and monitoring output in JSON.
3300 * journalctl's JSON output now supports simple ANSI coloring as well as
3301 a new "json-seq" mode for generating RFC7464 output.
3303 * Unit files now support the %g/%G specifiers that resolve to the UNIX
3304 group/GID of the service manager runs as, similar to the existing
3305 %u/%U specifiers that resolve to the UNIX user/UID.
3307 * systemd-logind learnt a new global configuration option
3308 UserStopDelaySec= that may be set in logind.conf. It specifies how
3309 long the systemd --user instance shall remain started after a user
3310 logs out. This is useful to speed up repetitive re-connections of the
3311 same user, as it means the user's service manager doesn't have to be
3312 stopped/restarted on each iteration, but can be reused between
3313 subsequent options. This setting defaults to 10s. systemd-logind also
3314 exports two new properties on its Manager D-Bus objects indicating
3315 whether the system's lid is currently closed, and whether the system
3318 * systemd gained support for a generic boot counting logic, which
3319 generically permits automatic reverting to older boot loader entries
3320 if newer updated ones don't work. The boot loader side is implemented
3321 in sd-boot, but is kept open for other boot loaders too. For details
3324 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
3326 * The SuccessAction=/FailureAction= unit file settings now learnt two
3327 new parameters: "exit" and "exit-force", which result in immediate
3328 exiting of the service manager, and are only useful in systemd --user
3329 and container environments.
3331 * Unit files gained support for a pair of options
3332 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
3333 exit status to use as service manager exit status when
3334 SuccessAction=/FailureAction= is set to exit or exit-force.
3336 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
3337 options may now be used to configure the log rate limiting applied by
3338 journald per-service.
3340 * systemd-analyze gained a new verb "timespan" for parsing and
3341 normalizing time span values (i.e. strings like "5min 7s 8us").
3343 * systemd-analyze also gained a new verb "security" for analyzing the
3344 security and sand-boxing settings of services in order to determine an
3345 "exposure level" for them, indicating whether a service would benefit
3346 from more sand-boxing options turned on for them.
3348 * "systemd-analyze syscall-filter" will now also show system calls
3349 supported by the local kernel but not included in any of the defined
3352 * .nspawn files now understand the Ephemeral= setting, matching the
3353 --ephemeral command line switch.
3355 * sd-event gained the new APIs sd_event_source_get_floating() and
3356 sd_event_source_set_floating() for controlling whether a specific
3357 event source is "floating", i.e. destroyed along with the even loop
3360 * Unit objects on D-Bus gained a new "Refs" property that lists all
3361 clients that currently have a reference on the unit (to ensure it is
3364 * The JoinControllers= option in system.conf is no longer supported, as
3365 it didn't work correctly, is hard to support properly, is legacy (as
3366 the concept only exists on cgroup v1) and apparently wasn't used.
3368 * Journal messages that are generated whenever a unit enters the failed
3369 state are now tagged with a unique MESSAGE_ID. Similarly, messages
3370 generated whenever a service process exits are now made recognizable,
3371 too. A tagged message is also emitted whenever a unit enters the
3372 "dead" state on success.
3374 * systemd-run gained a new switch --working-directory= for configuring
3375 the working directory of the service to start. A shortcut -d is
3376 equivalent, setting the working directory of the service to the
3377 current working directory of the invoking program. The new --shell
3378 (or just -S) option has been added for invoking the $SHELL of the
3379 caller as a service, and implies --pty --same-dir --wait --collect
3380 --service-type=exec. Or in other words, "systemd-run -S" is now the
3381 quickest way to quickly get an interactive in a fully clean and
3382 well-defined system service context.
3384 * machinectl gained a new verb "import-fs" for importing an OS tree
3385 from a directory. Moreover, when a directory or tarball is imported
3386 and single top-level directory found with the OS itself below the OS
3387 tree is automatically mangled and moved one level up.
3389 * systemd-importd will no longer set up an implicit btrfs loop-back
3390 file system on /var/lib/machines. If one is already set up, it will
3391 continue to be used.
3393 * A new generator "systemd-run-generator" has been added. It will
3394 synthesize a unit from one or more program command lines included in
3395 the kernel command line. This is very useful in container managers
3398 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
3400 This will run "systemd-nspawn" on an image, invoke the specified
3401 command line and immediately shut down the container again, returning
3402 the command line's exit code.
3404 * The block device locking logic is now documented:
3406 https://systemd.io/BLOCK_DEVICE_LOCKING
3408 * loginctl and machinectl now optionally output the various tables in
3409 JSON using the --output= switch. It is our intention to add similar
3410 support to systemctl and all other commands.
3412 * udevadm's query and trigger verb now optionally take a .device unit
3415 * systemd-udevd's network naming logic now understands a new
3416 net.naming-scheme= kernel command line switch, which may be used to
3417 pick a specific version of the naming scheme. This helps stabilizing
3418 interface names even as systemd/udev are updated and the naming logic
3421 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
3422 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
3423 initialize one to all 0xFF.
3425 * After loading the SELinux policy systemd will now recursively relabel
3426 all files and directories listed in
3427 /run/systemd/relabel-extra.d/*.relabel (which should be simple
3428 newline separated lists of paths) in addition to the ones it already
3429 implicitly relabels in /run, /dev and /sys. After the relabelling is
3430 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
3431 removed. This is useful to permit initrds (i.e. code running before
3432 the SELinux policy is in effect) to generate files in the host
3433 filesystem safely and ensure that the correct label is applied during
3434 the transition to the host OS.
3436 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
3437 mknod() handling in user namespaces. Previously mknod() would always
3438 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
3439 but device nodes generated that way cannot be opened, and attempts to
3440 open them result in EPERM. This breaks the "graceful fallback" logic
3441 in systemd's PrivateDevices= sand-boxing option. This option is
3442 implemented defensively, so that when systemd detects it runs in a
3443 restricted environment (such as a user namespace, or an environment
3444 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
3445 where device nodes cannot be created the effect of PrivateDevices= is
3446 bypassed (following the logic that 2nd-level sand-boxing is not
3447 essential if the system systemd runs in is itself already sand-boxed
3448 as a whole). This logic breaks with 4.18 in container managers where
3449 user namespacing is used: suddenly PrivateDevices= succeeds setting
3450 up a private /dev/ file system containing devices nodes — but when
3451 these are opened they don't work.
3453 At this point it is recommended that container managers utilizing
3454 user namespaces that intend to run systemd in the payload explicitly
3455 block mknod() with seccomp or similar, so that the graceful fallback
3458 We are very sorry for the breakage and the requirement to change
3459 container configurations for newer kernels. It's purely caused by an
3460 incompatible kernel change. The relevant kernel developers have been
3461 notified about this userspace breakage quickly, but they chose to
3464 * PermissionsStartOnly= setting is deprecated (but is still supported
3465 for backwards compatibility). The same functionality is provided by
3466 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
3469 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
3470 pam_systemd anymore.
3472 * The naming scheme for network devices was changed to always rename
3473 devices, even if they were already renamed by userspace. The "kernel"
3474 policy was changed to only apply as a fallback, if no other naming
3477 * The requirements to build systemd is bumped to meson-0.46 and
3480 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
3481 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
3482 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
3483 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
3484 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
3485 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
3486 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
3487 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
3488 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
3489 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
3490 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
3491 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
3492 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
3493 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
3494 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
3495 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
3496 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
3497 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
3498 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
3499 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
3500 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
3501 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
3502 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
3503 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
3504 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
3505 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
3506 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
3507 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
3508 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
3509 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
3510 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
3511 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
3512 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
3513 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
3514 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
3515 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
3516 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
3517 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
3518 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
3519 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
3520 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
3521 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
3522 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
3523 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
3524 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
3526 — Warsaw, 2018-12-21
3530 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
3531 builtin will name network interfaces differently than in previous
3532 versions for virtual network interfaces created with SR-IOV and NPAR
3533 and for devices where the PCI network controller device does not have
3534 a slot number associated.
3536 SR-IOV virtual devices are now named based on the name of the parent
3537 interface, with a suffix of "v<N>", where <N> is the virtual device
3538 number. Previously those virtual devices were named as if completely
3541 The ninth and later NPAR virtual devices will be named following the
3542 scheme used for the first eight NPAR partitions. Previously those
3543 devices were not renamed and the kernel default (eth<n>) was used.
3545 "net_id" will also generate names for PCI devices where the PCI
3546 network controller device does not have an associated slot number
3547 itself, but one of its parents does. Previously those devices were
3548 not renamed and the kernel default (eth<n>) was used.
3550 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
3551 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
3552 the unit. So, it is expected that the default behavior of
3553 systemd-logind is not changed. However, if distribution packagers or
3554 administrators disabled or modified IPAddressDeny= setting by a
3555 drop-in config file, then it may be necessary to update the file to
3556 re-enable AF_INET and AF_INET6 to support network user name services,
3559 * When the RestrictNamespaces= unit property is specified multiple
3560 times, then the specified types are merged now. Previously, only the
3561 last assignment was used. So, if distribution packagers or
3562 administrators modified the setting by a drop-in config file, then it
3563 may be necessary to update the file.
3565 * When OnFailure= is used in combination with Restart= on a service
3566 unit, then the specified units will no longer be triggered on
3567 failures that result in restarting. Previously, the specified units
3568 would be activated each time the unit failed, even when the unit was
3569 going to be restarted automatically. This behaviour contradicted the
3570 documentation. With this release the code is adjusted to match the
3573 * systemd-tmpfiles will now print a notice whenever it encounters
3574 tmpfiles.d/ lines referencing the /var/run/ directory. It will
3575 recommend reworking them to use the /run/ directory instead (for
3576 which /var/run/ is simply a symlinked compatibility alias). This way
3577 systemd-tmpfiles can properly detect line conflicts and merge lines
3578 referencing the same file by two paths, without having to access
3581 * systemctl disable/unmask/preset/preset-all cannot be used with
3582 --runtime. Previously this was allowed, but resulted in unintuitive
3583 behaviour that wasn't useful. systemctl disable/unmask will now undo
3584 both runtime and persistent enablement/masking, i.e. it will remove
3585 any relevant symlinks both in /run and /etc.
3587 * Note that all long-running system services shipped with systemd will
3588 now default to a system call allow list (rather than a deny list, as
3589 before). In particular, systemd-udevd will now enforce one too. For
3590 most cases this should be safe, however downstream distributions
3591 which disabled sandboxing of systemd-udevd (specifically the
3592 MountFlags= setting), might want to disable this security feature
3593 too, as the default allow-listing will prohibit all mount, swap,
3594 reboot and clock changing operations from udev rules.
3596 * sd-boot acquired new loader configuration settings to optionally turn
3597 off Windows and MacOS boot partition discovery as well as
3598 reboot-into-firmware menu items. It is also able to pick a better
3599 screen resolution for HiDPI systems, and now provides loader
3600 configuration settings to change the resolution explicitly.
3602 * systemd-resolved now supports DNS-over-TLS. It's still
3603 turned off by default, use DNSOverTLS=opportunistic to turn it on in
3604 resolved.conf. We intend to make this the default as soon as couple
3605 of additional techniques for optimizing the initial latency caused by
3606 establishing a TLS/TCP connection are implemented.
3608 * systemd-resolved.service and systemd-networkd.service now set
3609 DynamicUser=yes. The users systemd-resolve and systemd-network are
3610 not created by systemd-sysusers anymore.
3612 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
3613 that embed a network facing module into any process using getpwuid()
3614 or related call: the dynamic allocation of the user ID for
3615 systemd-resolved.service means the service manager has to check NSS
3616 if the user name is already taken when forking off the service. Since
3617 the user in the common case won't be defined in /etc/passwd the
3618 lookup is likely to trigger nss-ldap which in turn might use NSS to
3619 ask systemd-resolved for hostname lookups. This will hence result in
3620 a deadlock: a user name lookup in order to start
3621 systemd-resolved.service will result in a hostname lookup for which
3622 systemd-resolved.service needs to be started already. There are
3623 multiple ways to work around this problem: pre-allocate the
3624 "systemd-resolve" user on such systems, so that nss-ldap won't be
3625 triggered; or use a different NSS package that doesn't do networking
3626 in-process but provides a local asynchronous name cache; or configure
3627 the NSS package to avoid lookups for UIDs in the range `pkg-config
3628 systemd --variable=dynamicuidmin` … `pkg-config systemd
3629 --variable=dynamicuidmax`, so that it does not consider itself
3630 authoritative for the same UID range systemd allocates dynamic users
3633 * The systemd-resolve tool has been renamed to resolvectl (it also
3634 remains available under the old name, for compatibility), and its
3635 interface is now verb-based, similar in style to the other <xyz>ctl
3636 tools, such as systemctl or loginctl.
3638 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
3639 compatibility. It may be symlinked under the 'resolvconf' name, in
3640 which case it will take arguments and input compatible with the
3641 Debian and FreeBSD resolvconf tool.
3643 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
3644 where the system initially suspends, and after a timeout resumes and
3647 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
3648 set the client will only send a DUID as client identifier.
3650 * The nss-systemd glibc NSS module will now enumerate dynamic users and
3651 groups in effect. Previously, it could resolve UIDs/GIDs to user
3652 names/groups and vice versa, but did not support enumeration.
3654 * journald's Compress= configuration setting now optionally accepts a
3655 byte threshold value. All journal objects larger than this threshold
3656 will be compressed, smaller ones will not. Previously this threshold
3657 was not configurable and set to 512.
3659 * A new system.conf setting NoNewPrivileges= is now available which may
3660 be used to turn off acquisition of new privileges system-wide
3661 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
3662 for all its children). Note that turning this option on means setuid
3663 binaries and file system capabilities lose their special powers.
3664 While turning on this option is a big step towards a more secure
3665 system, doing so is likely to break numerous pre-existing UNIX tools,
3666 in particular su and sudo.
3668 * A new service systemd-time-sync-wait.service has been added. If
3669 enabled it will delay the time-sync.target unit at boot until time
3670 synchronization has been received from the network. This
3671 functionality is useful on systems lacking a local RTC or where it is
3672 acceptable that the boot process shall be delayed by external network
3675 * When hibernating, systemd will now inform the kernel of the image
3676 write offset, on kernels new enough to support this. This means swap
3677 files should work for hibernation now.
3679 * When loading unit files, systemd will now look for drop-in unit files
3680 extensions in additional places. Previously, for a unit file name
3681 "foo-bar-baz.service" it would look for dropin files in
3682 "foo-bar-baz.service.d/*.conf". Now, it will also look in
3683 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
3684 service name truncated after all inner dashes. This scheme allows
3685 writing drop-ins easily that apply to a whole set of unit files at
3686 once. It's particularly useful for mount and slice units (as their
3687 naming is prefix based), but is also useful for service and other
3688 units, for packages that install multiple unit files at once,
3689 following a strict naming regime of beginning the unit file name with
3690 the package's name. Two new specifiers are now supported in unit
3691 files to match this: %j and %J are replaced by the part of the unit
3692 name following the last dash.
3694 * Unit files and other configuration files that support specifier
3695 expansion now understand another three new specifiers: %T and %V will
3696 resolve to /tmp and /var/tmp respectively, or whatever temporary
3697 directory has been set for the calling user. %E will expand to either
3698 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
3700 * The ExecStart= lines of unit files are no longer required to
3701 reference absolute paths. If non-absolute paths are specified the
3702 specified binary name is searched within the service manager's
3703 built-in $PATH, which may be queried with 'systemd-path
3704 search-binaries-default'. It's generally recommended to continue to
3705 use absolute paths for all binaries specified in unit files.
3707 * Units gained a new load state "bad-setting", which is used when a
3708 unit file was loaded, but contained fatal errors which prevent it
3709 from being started (for example, a service unit has been defined
3710 lacking both ExecStart= and ExecStop= lines).
3712 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
3713 support alternative debuggers, for example lldb. The old name
3714 continues to be available however, for compatibility reasons. Use the
3715 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
3716 to pick an alternative debugger instead of the default gdb.
3718 * systemctl and the other tools will now output escape sequences that
3719 generate proper clickable hyperlinks in various terminal emulators
3720 where useful (for example, in the "systemctl status" output you can
3721 now click on the unit file name to quickly open it in the
3722 editor/viewer of your choice). Note that not all terminal emulators
3723 support this functionality yet, but many do. Unfortunately, the
3724 "less" pager doesn't support this yet, hence this functionality is
3725 currently automatically turned off when a pager is started (which
3726 happens quite often due to auto-paging). We hope to remove this
3727 limitation as soon as "less" learns these escape sequences. This new
3728 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
3729 environment variable. For details on these escape sequences see:
3730 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
3732 * networkd's .network files now support a new IPv6MTUBytes= option for
3733 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
3734 option in the [Route] section to configure the MTU to use for
3735 specific routes. It also gained support for configuration of the DHCP
3736 "UserClass" option through the new UserClass= setting. It gained
3737 three new options in the new [CAN] section for configuring CAN
3738 networks. The MULTICAST and ALLMULTI interface flags may now be
3739 controlled explicitly with the new Multicast= and AllMulticast=
3742 * networkd will now automatically make use of the kernel's route
3743 expiration feature, if it is available.
3745 * udevd's .link files now support setting the number of receive and
3746 transmit channels, using the RxChannels=, TxChannels=,
3747 OtherChannels=, CombinedChannels= settings.
3749 * Support for UDPSegmentationOffload= has been removed, given its
3750 limited support in hardware, and waning software support.
3752 * networkd's .netdev files now support creating "netdevsim" interfaces.
3754 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
3755 to query the unit belonging to a specific kernel control group.
3757 * systemd-analyze gained a new verb "cat-config", which may be used to
3758 dump the contents of any configuration file, with all its matching
3759 drop-in files added in, and honouring the usual search and masking
3760 logic applied to systemd configuration files. For example use
3761 "systemd-analyze cat-config systemd/system.conf" to get the complete
3762 system configuration file of systemd how it would be loaded by PID 1
3763 itself. Similar to this, various tools such as systemd-tmpfiles or
3764 systemd-sysusers, gained a new option "--cat-config", which does the
3765 corresponding operation for their own configuration settings. For
3766 example, "systemd-tmpfiles --cat-config" will now output the full
3767 list of tmpfiles.d/ lines in place.
3769 * timedatectl gained three new verbs: "show" shows bus properties of
3770 systemd-timedated, "timesync-status" shows the current NTP
3771 synchronization state of systemd-timesyncd, and "show-timesync"
3772 shows bus properties of systemd-timesyncd.
3774 * systemd-timesyncd gained a bus interface on which it exposes details
3777 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
3778 understood by systemd-timedated. It takes a colon-separated list of
3779 unit names of NTP client services. The list is used by
3780 "timedatectl set-ntp".
3782 * systemd-nspawn gained a new --rlimit= switch for setting initial
3783 resource limits for the container payload. There's a new switch
3784 --hostname= to explicitly override the container's hostname. A new
3785 --no-new-privileges= switch may be used to control the
3786 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
3787 --oom-score-adjust= switch controls the OOM scoring adjustment value
3788 for the payload. The new --cpu-affinity= switch controls the CPU
3789 affinity of the container payload. The new --resolv-conf= switch
3790 allows more detailed control of /etc/resolv.conf handling of the
3791 container. Similarly, the new --timezone= switch allows more detailed
3792 control of /etc/localtime handling of the container.
3794 * systemd-detect-virt gained a new --list switch, which will print a
3795 list of all currently known VM and container environments.
3797 * Support for "Portable Services" has been added, see
3798 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
3799 experimental, but this is expected to change soon. Reflecting this
3800 experimental state, the "portablectl" binary is not installed into
3801 /usr/bin yet. The binary has to be called with the full path
3802 /usr/lib/systemd/portablectl instead.
3804 * journalctl's and systemctl's -o switch now knows a new log output
3805 mode "with-unit". The output it generates is very similar to the
3806 regular "short" mode, but displays the unit name instead of the
3807 syslog tag for each log line. Also, the date is shown with timezone
3808 information. This mode is probably more useful than the classic
3809 "short" output mode for most purposes, except where pixel-perfect
3810 compatibility with classic /var/log/messages formatting is required.
3812 * A new --dump-bus-properties switch has been added to the systemd
3813 binary, which may be used to dump all supported D-Bus properties.
3814 (Options which are still supported, but are deprecated, are *not*
3817 * sd-bus gained a set of new calls:
3818 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
3819 enable/disable the "floating" state of a bus slot object,
3820 i.e. whether the slot object pins the bus it is allocated for into
3821 memory or if the bus slot object gets disconnected when the bus goes
3822 away. sd_bus_open_with_description(),
3823 sd_bus_open_user_with_description(),
3824 sd_bus_open_system_with_description() may be used to allocate bus
3825 objects and set their description string already during allocation.
3827 * sd-event gained support for watching inotify events from the event
3828 loop, in an efficient way, sharing inotify handles between multiple
3829 users. For this a new function sd_event_add_inotify() has been added.
3831 * sd-event and sd-bus gained support for calling special user-supplied
3832 destructor functions for userdata pointers associated with
3833 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
3834 functions sd_bus_slot_set_destroy_callback,
3835 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
3836 sd_bus_track_get_destroy_callback,
3837 sd_event_source_set_destroy_callback,
3838 sd_event_source_get_destroy_callback have been added.
3840 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
3842 * PID 1 will now automatically reschedule .timer units whenever the
3843 local timezone changes. (They previously got rescheduled
3844 automatically when the system clock changed.)
3846 * New documentation has been added to document cgroups delegation,
3847 portable services and the various code quality tools we have set up:
3849 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
3850 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
3851 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
3853 * The Boot Loader Specification has been added to the source tree.
3855 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
3857 While moving it into our source tree we have updated it and further
3858 changes are now accepted through the usual github PR workflow.
3860 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
3861 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
3862 earlier PAM modules. The data in these fields is used to initialize
3863 the session scope's resource properties. Thus external PAM modules
3864 may now configure per-session limits, for example sourced from
3865 external user databases.
3867 * socket units with Accept=yes will now maintain a "refused" counter in
3868 addition to the existing "accepted" counter, counting connections
3869 refused due to the enforced limits.
3871 * The "systemd-path search-binaries-default" command may now be use to
3872 query the default, built-in $PATH PID 1 will pass to the services it
3875 * A new unit file setting PrivateMounts= has been added. It's a boolean
3876 option. If enabled the unit's processes are invoked in their own file
3877 system namespace. Note that this behaviour is also implied if any
3878 other file system namespacing options (such as PrivateTmp=,
3879 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
3880 primarily useful for services that do not use any of the other file
3881 system namespacing options. One such service is systemd-udevd.service
3882 where this is now used by default.
3884 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
3885 when the system is booted in UEFI "secure mode".
3887 * A new unit "system-update-pre.target" is added, which defines an
3888 optional synchronization point for offline system updates, as
3889 implemented by the pre-existing "system-update.target" unit. It
3890 allows ordering services before the service that executes the actual
3891 update process in a generic way.
3893 * Systemd now emits warnings whenever .include syntax is used.
3895 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
3896 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
3897 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
3898 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
3899 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
3900 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
3901 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
3902 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
3903 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
3904 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
3905 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
3906 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
3907 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
3908 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
3909 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
3910 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
3911 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
3912 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
3913 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
3914 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
3915 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
3916 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
3917 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
3918 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
3919 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
3920 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
3921 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
3922 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
3923 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
3925 — Berlin, 2018-06-22
3929 * The MemoryAccounting= unit property now defaults to on. After
3930 discussions with the upstream control group maintainers we learnt
3931 that the negative impact of cgroup memory accounting on current
3932 kernels is finally relatively minimal, so that it should be safe to
3933 enable this by default without affecting system performance. Besides
3934 memory accounting only task accounting is turned on by default, all
3935 other forms of resource accounting (CPU, IO, IP) remain off for now,
3936 because it's not clear yet that their impact is small enough to move
3937 from opt-in to opt-out. We recommend downstreams to leave memory
3938 accounting on by default if kernel 4.14 or higher is primarily
3939 used. On very resource constrained systems or when support for old
3940 kernels is a necessity, -Dmemory-accounting-default=false can be used
3941 to revert this change.
3943 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
3944 %udev_rules_update) and the journal catalog (%journal_catalog_update)
3945 from the upgrade scriptlets of individual packages now do nothing.
3946 Transfiletriggers have been added which will perform those updates
3947 once at the end of the transaction.
3949 Similar transfiletriggers have been added to execute any sysctl.d
3950 and binfmt.d rules. Thus, it should be unnecessary to provide any
3951 scriptlets to execute this configuration from package installation
3954 * systemd-sysusers gained a mode where the configuration to execute is
3955 specified on the command line, but this configuration is not executed
3956 directly, but instead it is merged with the configuration on disk,
3957 and the result is executed. This is useful for package installation
3958 scripts which want to create the user before installing any files on
3959 disk (in case some of those files are owned by that user), while
3960 still allowing local admin overrides.
3962 This functionality is exposed to rpm scriptlets through a new
3963 %sysusers_create_package macro. Old %sysusers_create and
3964 %sysusers_create_inline macros are deprecated.
3966 A transfiletrigger for sysusers.d configuration is now installed,
3967 which means that it should be unnecessary to call systemd-sysusers from
3968 package installation scripts, unless the package installs any files
3969 owned by those newly-created users, in which case
3970 %sysusers_create_package should be used.
3972 * Analogous change has been done for systemd-tmpfiles: it gained a mode
3973 where the command-line configuration is merged with the configuration
3974 on disk. This is exposed as the new %tmpfiles_create_package macro,
3975 and %tmpfiles_create is deprecated. A transfiletrigger is installed
3976 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
3977 from package installation scripts.
3979 * sysusers.d configuration for a user may now also specify the group
3980 number, in addition to the user number ("u username 123:456"), or
3981 without the user number ("u username -:456").
3983 * Configution items for systemd-sysusers can now be specified as
3984 positional arguments when the new --inline switch is used.
3986 * The login shell of users created through sysusers.d may now be
3987 specified (previously, it was always /bin/sh for root and
3988 /sbin/nologin for other users).
3990 * systemd-analyze gained a new --global switch to look at global user
3991 configuration. It also gained a unit-paths verb to list the unit load
3992 paths that are compiled into systemd (which can be used with
3993 --systemd, --user, or --global).
3995 * udevadm trigger gained a new --settle/-w option to wait for any
3996 triggered events to finish (but just those, and not any other events
3997 which are triggered meanwhile).
3999 * The action that systemd-logind takes when the lid is closed and the
4000 machine is connected to external power can now be configured using
4001 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
4002 was determined by HandleLidSwitch=, and, for backwards compatibility,
4003 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
4005 * journalctl will periodically call sd_journal_process() to make it
4006 resilient against inotify queue overruns when journal files are
4007 rotated very quickly.
4009 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
4010 sd_bus_get_n_queued_write — may be used to check the number of
4011 pending bus messages.
4013 * systemd gained a new
4014 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
4015 which can be used to migrate foreign processes to scope and service
4016 units. The primary user for this new API is systemd itself: the
4017 systemd --user instance uses this call of the systemd --system
4018 instance to migrate processes if it itself gets the request to
4019 migrate processes and the kernel refuses this due to access
4020 restrictions. Thanks to this "systemd-run --scope --user …" works
4021 again in pure cgroup v2 environments when invoked from the user
4024 * A new TemporaryFileSystem= setting can be used to mask out part of
4025 the real file system tree with tmpfs mounts. This may be combined
4026 with BindPaths= and BindReadOnlyPaths= to hide files or directories
4027 not relevant to the unit, while still allowing some paths lower in
4028 the tree to be accessed.
4030 ProtectHome=tmpfs may now be used to hide user home and runtime
4031 directories from units, in a way that is mostly equivalent to
4032 "TemporaryFileSystem=/home /run/user /root".
4034 * Non-service units are now started with KeyringMode=shared by default.
4035 This means that mount and swapon and other mount tools have access
4036 to keys in the main keyring.
4038 * /sys/fs/bpf is now mounted automatically.
4040 * QNX virtualization is now detected by systemd-detect-virt and may
4041 be used in ConditionVirtualization=.
4043 * IPAccounting= may now be enabled also for slice units.
4045 * A new -Dsplit-bin= build configuration switch may be used to specify
4046 whether bin and sbin directories are merged, or if they should be
4047 included separately in $PATH and various listings of executable
4048 directories. The build configuration scripts will try to autodetect
4049 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
4050 system, but distributions are encouraged to configure this
4053 * A new -Dok-color= build configuration switch may be used to change
4054 the colour of "OK" status messages.
4056 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
4057 PrivateNetwork=yes was buggy in previous versions of systemd. This
4058 means that after the upgrade and daemon-reexec, any such units must
4061 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
4062 will not exclude read-only files owned by root from cleanup.
4064 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
4065 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
4066 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
4067 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
4068 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
4069 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
4070 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
4071 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
4072 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
4073 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
4074 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
4075 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
4076 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
4077 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
4078 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
4079 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
4081 — Warsaw, 2018-03-05
4085 * Some keyboards come with a zoom see-saw or rocker which until now got
4086 mapped to the Linux "zoomin/out" keys in hwdb. However, these
4087 keycodes are not recognized by any major desktop. They now produce
4088 Up/Down key events so that they can be used for scrolling.
4090 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
4091 slightly: previously, if an argument was specified for lines of this
4092 type (i.e. the right-most column was set) this string was appended to
4093 existing files each time systemd-tmpfiles was run. This behaviour was
4094 different from what the documentation said, and not particularly
4095 useful, as repeated systemd-tmpfiles invocations would not be
4096 idempotent and grow such files without bounds. With this release
4097 behaviour has been altered to match what the documentation says:
4098 lines of this type only have an effect if the indicated files don't
4099 exist yet, and only then the argument string is written to the file.
4101 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
4102 systemd-tmpfiles behaviour: previously, read-only files owned by root
4103 were always excluded from the file "aging" algorithm (i.e. the
4104 automatic clean-up of directories like /tmp based on
4105 atime/mtime/ctime). We intend to drop this restriction, and age files
4106 by default even when owned by root and read-only. This behaviour was
4107 inherited from older tools, but there have been requests to remove
4108 it, and it's not obvious why this restriction was made in the first
4109 place. Please speak up now, if you are aware of software that requires
4110 this behaviour, otherwise we'll remove the restriction in v238.
4112 * A new environment variable $SYSTEMD_OFFLINE is now understood by
4113 systemctl. It takes a boolean argument. If on, systemctl assumes it
4114 operates on an "offline" OS tree, and will not attempt to talk to the
4115 service manager. Previously, this mode was implicitly enabled if a
4116 chroot() environment was detected, and this new environment variable
4117 now provides explicit control.
4119 * .path and .socket units may now be created transiently, too.
4120 Previously only service, mount, automount and timer units were
4121 supported as transient units. The systemd-run tool has been updated
4122 to expose this new functionality, you may hence use it now to bind
4123 arbitrary commands to path or socket activation on-the-fly from the
4124 command line. Moreover, almost all properties are now exposed for the
4125 unit types that already supported transient operation.
4127 * The systemd-mount command gained support for a new --owner= parameter
4128 which takes a user name, which is then resolved and included in uid=
4129 and gid= mount options string of the file system to mount.
4131 * A new unit condition ConditionControlGroupController= has been added
4132 that checks whether a specific cgroup controller is available.
4134 * Unit files, udev's .link files, and systemd-networkd's .netdev and
4135 .network files all gained support for a new condition
4136 ConditionKernelVersion= for checking against specific kernel
4139 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
4140 support for configuring device flags in the Flags= setting. In the
4141 same files, the [Tunnel] section gained support for configuring
4142 AllowLocalRemote=. The [Route] section in .network files gained
4143 support for configuring InitialCongestionWindow=,
4144 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
4145 understands RapidCommit=.
4147 * systemd-networkd's DHCPv6 support gained support for Prefix
4150 * sd-bus gained support for a new "watch-bind" feature. When this
4151 feature is enabled, an sd_bus connection may be set up to connect to
4152 an AF_UNIX socket in the file system as soon as it is created. This
4153 functionality is useful for writing early-boot services that
4154 automatically connect to the system bus as soon as it is started,
4155 without ugly time-based polling. systemd-networkd and
4156 systemd-resolved have been updated to make use of this
4157 functionality. busctl exposes this functionality in a new
4158 --watch-bind= command line switch.
4160 * sd-bus will now optionally synthesize a local "Connected" signal as
4161 soon as a D-Bus connection is set up fully. This message mirrors the
4162 already existing "Disconnected" signal which is synthesized when the
4163 connection is terminated. This signal is generally useful but
4164 particularly handy in combination with the "watch-bind" feature
4165 described above. Synthesizing of this message has to be requested
4166 explicitly through the new API call sd_bus_set_connected_signal(). In
4167 addition a new call sd_bus_is_ready() has been added that checks
4168 whether a connection is fully set up (i.e. between the "Connected" and
4169 "Disconnected" signals).
4171 * sd-bus gained two new calls sd_bus_request_name_async() and
4172 sd_bus_release_name_async() for asynchronously registering bus
4173 names. Similar, there is now sd_bus_add_match_async() for installing
4174 a signal match asynchronously. All of systemd's own services have
4175 been updated to make use of these calls. Doing these operations
4176 asynchronously has two benefits: it reduces the risk of deadlocks in
4177 case of cyclic dependencies between bus services, and it speeds up
4178 service initialization since synchronization points for bus
4179 round-trips are removed.
4181 * sd-bus gained two new calls sd_bus_match_signal() and
4182 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
4183 and sd_bus_add_match_async() but instead of taking a D-Bus match
4184 string take match fields as normal function parameters.
4186 * sd-bus gained two new calls sd_bus_set_sender() and
4187 sd_bus_message_set_sender() for setting the sender name of outgoing
4188 messages (either for all outgoing messages or for just one specific
4189 one). These calls are only useful in direct connections as on
4190 brokered connections the broker fills in the sender anyway,
4191 overwriting whatever the client filled in.
4193 * sd-event gained a new pseudo-handle that may be specified on all API
4194 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
4195 used this refers to the default event loop object of the calling
4196 thread. Note however that this does not implicitly allocate one —
4197 which has to be done prior by using sd_event_default(). Similarly
4198 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
4199 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
4200 to the default bus of the specified type of the calling thread. Here
4201 too this does not implicitly allocate bus connection objects, this
4202 has to be done prior with sd_bus_default() and friends.
4204 * sd-event gained a new call pair
4205 sd_event_source_{get|set}_io_fd_own(). This may be used to request
4206 automatic closure of the file descriptor an IO event source watches
4207 when the event source is destroyed.
4209 * systemd-networkd gained support for natively configuring WireGuard
4212 * In previous versions systemd synthesized user records both for the
4213 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
4214 internally. In order to simplify distribution-wide renames of the
4215 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
4216 new transitional flag file has been added: if
4217 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
4218 user and group record within the systemd codebase is disabled.
4220 * systemd-notify gained a new --uid= option for selecting the source
4221 user/UID to use for notification messages sent to the service
4224 * journalctl gained a new --grep= option to list only entries in which
4225 the message matches a certain pattern. By default matching is case
4226 insensitive if the pattern is lowercase, and case sensitive
4227 otherwise. Option --case-sensitive=yes|no can be used to override
4228 this an specify case sensitivity or case insensitivity.
4230 * There's now a "systemd-analyze service-watchdogs" command for printing
4231 the current state of the service runtime watchdog, and optionally
4232 enabling or disabling the per-service watchdogs system-wide if given a
4233 boolean argument (i.e. the concept you configure in WatchdogSec=), for
4234 debugging purposes. There's also a kernel command line option
4235 systemd.service_watchdogs= for controlling the same.
4237 * Two new "log-level" and "log-target" options for systemd-analyze were
4238 added that merge the now deprecated get-log-level, set-log-level and
4239 get-log-target, set-log-target pairs. The deprecated options are still
4240 understood for backwards compatibility. The two new options print the
4241 current value when no arguments are given, and set them when a
4242 level/target is given as an argument.
4244 * sysusers.d's "u" lines now optionally accept both a UID and a GID
4245 specification, separated by a ":" character, in order to create users
4246 where UID and GID do not match.
4248 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
4249 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
4250 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
4251 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
4252 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
4253 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
4254 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
4255 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
4256 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
4257 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
4258 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
4259 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
4260 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
4261 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
4262 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
4263 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
4264 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
4265 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
4266 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
4267 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
4274 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
4275 in v235 has been extended to also set the dummy.ko module option
4276 numdummies=0, preventing the kernel from automatically creating
4277 dummy0. All dummy interfaces must now be explicitly created.
4279 * Unknown '%' specifiers in configuration files are now rejected. This
4280 applies to units and tmpfiles.d configuration. Any percent characters
4281 that are followed by a letter or digit that are not supposed to be
4282 interpreted as the beginning of a specifier should be escaped by
4283 doubling ("%%"). (So "size=5%" is still accepted, as well as
4284 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
4285 valid specifiers today.)
4287 * systemd-resolved now maintains a new dynamic
4288 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
4289 recommended to make /etc/resolv.conf a symlink to it. This file
4290 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
4291 includes dynamically acquired search domains, achieving more correct
4292 DNS resolution by software that bypasses local DNS APIs such as NSS.
4294 * The "uaccess" udev tag has been dropped from /dev/kvm and
4295 /dev/dri/renderD*. These devices now have the 0666 permissions by
4296 default (but this may be changed at build-time). /dev/dri/renderD*
4297 will now be owned by the "render" group along with /dev/kfd.
4299 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
4300 systemd-journal-gatewayd.service and
4301 systemd-journal-upload.service. This means "nss-systemd" must be
4302 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
4303 services are resolved properly.
4305 * In /etc/fstab two new mount options are now understood:
4306 x-systemd.makefs and x-systemd.growfs. The former has the effect that
4307 the configured file system is formatted before it is mounted, the
4308 latter that the file system is resized to the full block device size
4309 after it is mounted (i.e. if the file system is smaller than the
4310 partition it resides on, it's grown). This is similar to the fsck
4311 logic in /etc/fstab, and pulls in systemd-makefs@.service and
4312 systemd-growfs@.service as necessary, similar to
4313 systemd-fsck@.service. Resizing is currently only supported on ext4
4316 * In systemd-networkd, the IPv6 RA logic now optionally may announce
4317 DNS server and domain information.
4319 * Support for the LUKS2 on-disk format for encrypted partitions has
4320 been added. This requires libcryptsetup2 during compilation and
4323 * The systemd --user instance will now signal "readiness" when its
4324 basic.target unit has been reached, instead of when the run queue ran
4325 empty for the first time.
4327 * Tmpfiles.d with user configuration are now also supported.
4328 systemd-tmpfiles gained a new --user switch, and snippets placed in
4329 ~/.config/user-tmpfiles.d/ and corresponding directories will be
4330 executed by systemd-tmpfiles --user running in the new
4331 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
4332 running in the user session.
4334 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
4335 %S resolves to the top-level state directory (/var/lib for the system
4336 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
4337 top-level cache directory (/var/cache for the system instance,
4338 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
4339 logs directory (/var/log for the system instance,
4340 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
4341 existing %t specifier, that resolves to the top-level runtime
4342 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
4345 * journalctl learnt a new parameter --output-fields= for limiting the
4346 set of journal fields to output in verbose and JSON output modes.
4348 * systemd-timesyncd's configuration file gained a new option
4349 RootDistanceMaxSec= for setting the maximum root distance of servers
4350 it'll use, as well as the new options PollIntervalMinSec= and
4351 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
4353 * bootctl gained a new command "list" for listing all available boot
4354 menu items on systems that follow the boot loader specification.
4356 * systemctl gained a new --dry-run switch that shows what would be done
4357 instead of doing it, and is currently supported by the shutdown and
4360 * ConditionSecurity= can now detect the TOMOYO security module.
4362 * Unit file [Install] sections are now also respected in unit drop-in
4363 files. This is intended to be used by drop-ins under /usr/lib/.
4365 * systemd-firstboot may now also set the initial keyboard mapping.
4367 * Udev "changed" events for devices which are exposed as systemd
4368 .device units are now propagated to units specified in
4369 ReloadPropagatedFrom= as reload requests.
4371 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
4372 unit template name (i.e. a name in the form of 'foobar@.service',
4373 without the instance component between the '@' and - the '.'), then
4374 the escaped sysfs path of the device is automatically used as the
4377 * SystemCallFilter= in unit files has been extended so that an "errno"
4378 can be specified individually for each system call. Example:
4379 SystemCallFilter=~uname:EILSEQ.
4381 * The cgroup delegation logic has been substantially updated. Delegate=
4382 now optionally takes a list of controllers (instead of a boolean, as
4383 before), which lists the controllers to delegate at least.
4385 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
4387 * A new LogLevelMax= setting configures the maximum log level any
4388 process of the service may log at (i.e. anything with a lesser
4389 priority than what is specified is automatically dropped). A new
4390 LogExtraFields= setting allows configuration of additional journal
4391 fields to attach to all log records generated by any of the unit's
4394 * New StandardInputData= and StandardInputText= settings along with the
4395 new option StandardInput=data may be used to configure textual or
4396 binary data that shall be passed to the executed service process via
4397 standard input, encoded in-line in the unit file.
4399 * StandardInput=, StandardOutput= and StandardError= may now be used to
4400 connect stdin/stdout/stderr of executed processes directly with a
4401 file or AF_UNIX socket in the file system, using the new "file:" option.
4403 * A new unit file option CollectMode= has been added, that allows
4404 tweaking the garbage collection logic for units. It may be used to
4405 tell systemd to garbage collect units that have failed automatically
4406 (normally it only GCs units that exited successfully). systemd-run
4407 and systemd-mount expose this new functionality with a new -G option.
4409 * "machinectl bind" may now be used to bind mount non-directories
4410 (i.e. regularfiles, devices, fifos, sockets).
4412 * systemd-analyze gained a new verb "calendar" for validating and
4413 testing calendar time specifications to use for OnCalendar= in timer
4414 units. Besides validating the expression it will calculate the next
4415 time the specified expression would elapse.
4417 * In addition to the pre-existing FailureAction= unit file setting
4418 there's now SuccessAction=, for configuring a shutdown action to
4419 execute when a unit completes successfully. This is useful in
4420 particular inside containers that shall terminate after some workload
4421 has been completed. Also, both options are now supported for all unit
4422 types, not just services.
4424 * networkds's IP rule support gained two new options
4425 IncomingInterface= and OutgoingInterface= for configuring the incoming
4426 and outgoing interfaces of configured rules. systemd-networkd also
4427 gained support for "vxcan" network devices.
4429 * networkd gained a new setting RequiredForOnline=, taking a
4430 boolean. If set, systemd-wait-online will take it into consideration
4431 when determining that the system is up, otherwise it will ignore the
4432 interface for this purpose.
4434 * The sd_notify() protocol gained support for a new operation: with
4435 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
4436 store again, ahead of POLLHUP or POLLERR when they are removed
4439 * A new document doc/UIDS-GIDS.md has been added to the source tree,
4440 that documents the UID/GID range and assignment assumptions and
4441 requirements of systemd.
4443 * The watchdog device PID 1 will ping may now be configured through the
4444 WatchdogDevice= configuration file setting, or by setting the
4445 systemd.watchdog_service= kernel commandline option.
4447 * systemd-resolved's gained support for registering DNS-SD services on
4448 the local network using MulticastDNS. Services may either be
4449 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
4450 the same dir below /run, /usr/lib), or through its D-Bus API.
4452 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
4453 extend the effective start, runtime, and stop time. The service must
4454 continue to send EXTEND_TIMEOUT_USEC within the period specified to
4455 prevent the service manager from making the service as timedout.
4457 * systemd-resolved's DNSSEC support gained support for RFC 8080
4458 (Ed25519 keys and signatures).
4460 * The systemd-resolve command line tool gained a new set of options
4461 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
4462 --set-nta= and --revert to configure per-interface DNS configuration
4463 dynamically during runtime. It's useful for pushing DNS information
4464 into systemd-resolved from DNS hook scripts that various interface
4465 managing software supports (such as pppd).
4467 * systemd-nspawn gained a new --network-namespace-path= command line
4468 option, which may be used to make a container join an existing
4469 network namespace, by specifying a path to a "netns" file.
4471 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
4472 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
4473 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
4474 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
4475 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
4476 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
4477 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
4478 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
4479 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
4480 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
4481 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
4482 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
4483 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
4484 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
4485 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
4486 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
4487 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
4488 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
4489 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
4490 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
4491 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
4492 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
4493 Jędrzejewski-Szmek, Zeal Jagannatha
4495 — Berlin, 2017-12-14
4499 * INCOMPATIBILITY: systemd-logind.service and other long-running
4500 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
4501 communication with the outside. This generally improves security of
4502 the system, and is in almost all cases a safe and good choice, as
4503 these services do not and should not provide any network-facing
4504 functionality. However, systemd-logind uses the glibc NSS API to
4505 query the user database. This creates problems on systems where NSS
4506 is set up to directly consult network services for user database
4507 lookups. In particular, this creates incompatibilities with the
4508 "nss-nis" module, which attempts to directly contact the NIS/YP
4509 network servers it is configured for, and will now consistently
4510 fail. In such cases, it is possible to turn off IP sandboxing for
4511 systemd-logind.service (set IPAddressDeny= in its [Service] section
4512 to the empty string, via a .d/ unit file drop-in). Downstream
4513 distributions might want to update their nss-nis packaging to include
4514 such a drop-in snippet, accordingly, to hide this incompatibility
4515 from the user. Another option is to make use of glibc's nscd service
4516 to proxy such network requests through a privilege-separated, minimal
4517 local caching daemon, or to switch to more modern technologies such
4518 sssd, whose NSS hook-ups generally do not involve direct network
4519 access. In general, we think it's definitely time to question the
4520 implementation choices of nss-nis, i.e. whether it's a good idea
4521 today to embed a network-facing loadable module into all local
4522 processes that need to query the user database, including the most
4523 trivial and benign ones, such as "ls". For more details about
4524 IPAddressDeny= see below.
4526 * A new modprobe.d drop-in is now shipped by default that sets the
4527 bonding module option max_bonds=0. This overrides the kernel default,
4528 to avoid conflicts and ambiguity as to whether or not bond0 should be
4529 managed by systemd-networkd or not. This resolves multiple issues
4530 with bond0 properties not being applied, when bond0 is configured
4531 with systemd-networkd. Distributors may choose to not package this,
4532 however in that case users will be prevented from correctly managing
4533 bond0 interface using systemd-networkd.
4535 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
4536 which print the logging level and target of the system manager. They
4537 complement the existing "set-log-level" and "set-log-target" verbs
4538 used to change those values.
4540 * journald.conf gained a new boolean setting ReadKMsg= which defaults
4541 to on. If turned off kernel log messages will not be read by
4542 systemd-journald or included in the logs. It also gained a new
4543 setting LineMax= for configuring the maximum line length in
4544 STDOUT/STDERR log streams. The new default for this value is 48K, up
4545 from the previous hardcoded 2048.
4547 * A new unit setting RuntimeDirectoryPreserve= has been added, which
4548 allows more detailed control of what to do with a runtime directory
4549 configured with RuntimeDirectory= (i.e. a directory below /run or
4550 $XDG_RUNTIME_DIR) after a unit is stopped.
4552 * The RuntimeDirectory= setting for units gained support for creating
4553 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
4554 one top-level directory.
4556 * Units gained new options StateDirectory=, CacheDirectory=,
4557 LogsDirectory= and ConfigurationDirectory= which are closely related
4558 to RuntimeDirectory= but manage per-service directories below
4559 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
4560 possible to write unit files which when activated automatically gain
4561 properly owned service specific directories in these locations, thus
4562 making unit files self-contained and increasing compatibility with
4563 stateless systems and factory reset where /etc or /var are
4564 unpopulated at boot. Matching these new settings there's also
4565 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
4566 ConfigurationDirectoryMode= for configuring the access mode of these
4567 directories. These settings are particularly useful in combination
4568 with DynamicUser=yes as they provide secure, properly-owned,
4569 writable, and stateful locations for storage, excluded from the
4570 sandbox that such services live in otherwise.
4572 * Automake support has been removed from this release. systemd is now
4575 * systemd-journald will now aggressively cache client metadata during
4576 runtime, speeding up log write performance under pressure. This comes
4577 at a small price though: as much of the metadata is read
4578 asynchronously from /proc/ (and isn't implicitly attached to log
4579 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
4580 metadata stored alongside a log entry might be slightly
4581 out-of-date. Previously it could only be slightly newer than the log
4582 message. The time window is small however, and given that the kernel
4583 is unlikely to be improved anytime soon in this regard, this appears
4586 * nss-myhostname/systemd-resolved will now by default synthesize an
4587 A/AAAA resource record for the "_gateway" hostname, pointing to the
4588 current default IP gateway. Previously it did that for the "gateway"
4589 name, hampering adoption, as some distributions wanted to leave that
4590 hostname open for local use. The old behaviour may still be
4591 requested at build time.
4593 * systemd-networkd's [Address] section in .network files gained a new
4594 Scope= setting for configuring the IP address scope. The [Network]
4595 section gained a new boolean setting ConfigureWithoutCarrier= that
4596 tells systemd-networkd to ignore link sensing when configuring the
4597 device. The [DHCP] section gained a new Anonymize= boolean option for
4598 turning on a number of options suggested in RFC 7844. A new
4599 [RoutingPolicyRule] section has been added for configuring the IP
4600 routing policy. The [Route] section has gained support for a new
4601 Type= setting which permits configuring
4602 blackhole/unreachable/prohibit routes.
4604 * The [VRF] section in .netdev files gained a new Table= setting for
4605 configuring the routing table to use. The [Tunnel] section gained a
4606 new Independent= boolean field for configuring tunnels independent of
4607 an underlying network interface. The [Bridge] section gained a new
4608 GroupForwardMask= option for configuration of propagation of link
4609 local frames between bridge ports.
4611 * The WakeOnLan= setting in .link files gained support for a number of
4612 new modes. A new TCP6SegmentationOffload= setting has been added for
4613 configuring TCP/IPv6 hardware segmentation offload.
4615 * The IPv6 RA sender implementation may now optionally send out RDNSS
4616 and RDNSSL records to supply DNS configuration to peers.
4618 * systemd-nspawn gained support for a new --system-call-filter= command
4619 line option for adding and removing entries in the default system
4620 call filter it applies. Moreover systemd-nspawn has been changed to
4621 implement a system call allow list instead of a deny list.
4623 * systemd-run gained support for a new --pipe command line option. If
4624 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
4625 are directly passed on to the activated transient service
4626 executable. This allows invoking arbitrary processes as systemd
4627 services (for example to take benefit of dependency management,
4628 accounting management, resource management or log management that is
4629 done automatically for services) — while still allowing them to be
4630 integrated in a classic UNIX shell pipeline.
4632 * When a service sends RELOAD=1 via sd_notify() and reload propagation
4633 using ReloadPropagationTo= is configured, a reload is now propagated
4634 to configured units. (Previously this was only done on explicitly
4635 requested reloads, using "systemctl reload" or an equivalent
4638 * For each service unit a restart counter is now kept: it is increased
4639 each time the service is restarted due to Restart=, and may be
4640 queried using "systemctl show -p NRestarts …".
4642 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
4643 @signal and @timer have been added, for usage with SystemCallFilter=
4644 in unit files and the new --system-call-filter= command line option
4645 of systemd-nspawn (see above).
4647 * ExecStart= lines in unit files gained two new modifiers: when a
4648 command line is prefixed with "!" the command will be executed as
4649 configured, except for the credentials applied by
4650 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
4651 "+", but does still apply namespacing options unlike "+". There's
4652 also "!!" now, which is mostly identical, but becomes a NOP on
4653 systems that support ambient capabilities. This is useful to write
4654 unit files that work with ambient capabilities where possible but
4655 automatically fall back to traditional privilege dropping mechanisms
4656 on systems where this is not supported.
4658 * ListenNetlink= settings in socket units now support RDMA netlink
4661 * A new unit file setting LockPersonality= has been added which permits
4662 locking down the chosen execution domain ("personality") of a service
4665 * A new special target "getty-pre.target" has been added, which is
4666 ordered before all text logins, and may be used to order services
4667 before textual logins acquire access to the console.
4669 * systemd will now attempt to load the virtio-rng.ko kernel module very
4670 early on if a VM environment supporting this is detected. This should
4671 improve entropy during early boot in virtualized environments.
4673 * A _netdev option is now supported in /etc/crypttab that operates in a
4674 similar way as the same option in /etc/fstab: it permits configuring
4675 encrypted devices that need to be ordered after the network is up.
4676 Following this logic, two new special targets
4677 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
4678 added that are to cryptsetup.target what remote-fs.target and
4679 remote-fs-pre.target are to local-fs.target.
4681 * Service units gained a new UnsetEnvironment= setting which permits
4682 unsetting specific environment variables for services that are
4683 normally passed to it (for example in order to mask out locale
4684 settings for specific services that can't deal with it).
4686 * Units acquired a new boolean option IPAccounting=. When turned on, IP
4687 traffic accounting (packet count as well as byte count) is done for
4688 the service, and shown as part of "systemctl status" or "systemd-run
4691 * Service units acquired two new options IPAddressAllow= and
4692 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
4693 for configuring a simple IP access control list for all sockets of
4694 the unit. These options are available also on .slice and .socket
4695 units, permitting flexible access list configuration for individual
4696 services as well as groups of services (as defined by a slice unit),
4697 including system-wide. Note that IP ACLs configured this way are
4698 enforced on every single IPv4 and IPv6 socket created by any process
4699 of the service unit, and apply to ingress as well as egress traffic.
4701 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
4702 structured log message is generated each time the unit is stopped,
4703 containing information about the consumed resources of this
4706 * A new setting KeyringMode= has been added to unit files, which may be
4707 used to control how the kernel keyring is set up for executed
4710 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
4711 "systemctl kexec" and "systemctl exit" are now always asynchronous in
4712 behaviour (that is: these commands return immediately after the
4713 operation was enqueued instead of waiting for the operation to
4714 complete). Previously, "systemctl poweroff" and "systemctl reboot"
4715 were asynchronous on systems using systemd-logind (i.e. almost
4716 always, and like they were on sysvinit), and the other three commands
4717 were unconditionally synchronous. With this release this is cleaned
4718 up, and callers will see the same asynchronous behaviour on all
4719 systems for all five operations.
4721 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
4724 * .timer units now accept calendar specifications in other timezones
4725 than UTC or the local timezone.
4727 * The tmpfiles snippet var.conf has been changed to create
4728 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
4729 the "utmp" group already, and it appears to be generally understood
4730 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
4731 databases. Previously this was implemented correctly for all these
4732 databases excepts btmp, which has been opened up like this now
4733 too. Note that while the other databases are world-readable
4734 (i.e. 0644), btmp is not and remains more restrictive.
4736 * The systemd-resolve tool gained a new --reset-server-features
4737 switch. When invoked like this systemd-resolved will forget
4738 everything it learnt about the features supported by the configured
4739 upstream DNS servers, and restarts the feature probing logic on the
4740 next resolver look-up for them at the highest feature level
4743 * The status dump systemd-resolved sends to the logs upon receiving
4744 SIGUSR1 now also includes information about all DNS servers it is
4745 configured to use, and the features levels it probed for them.
4747 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
4748 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
4749 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
4750 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
4751 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
4752 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
4753 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
4754 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
4755 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
4756 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
4757 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
4758 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
4759 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
4760 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
4761 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
4762 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
4763 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
4764 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
4765 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
4766 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
4768 — Berlin, 2017-10-06
4772 * Meson is now supported as build system in addition to Automake. It is
4773 our plan to remove Automake in one of our next releases, so that
4774 Meson becomes our exclusive build system. Hence, please start using
4775 the Meson build system in your downstream packaging. There's plenty
4776 of documentation around how to use Meson, the extremely brief
4779 ./autogen.sh && ./configure && make && sudo make install
4783 meson build && ninja -C build && sudo ninja -C build install
4785 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
4786 which permits configuring a timeout on the time a job is
4787 running. This is particularly useful for setting timeouts on jobs for
4790 * Unit files gained two new options ConditionUser= and ConditionGroup=
4791 for conditionalizing units based on the identity of the user/group
4792 running a systemd user instance.
4794 * systemd-networkd now understands a new FlowLabel= setting in the
4795 [VXLAN] section of .network files, as well as a Priority= in
4796 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
4797 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
4798 gained support for configuration of GENEVE links, and IPv6 address
4799 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
4801 * .link files now understand a new Port= setting.
4803 * systemd-networkd's DHCP support gained support for DHCP option 119
4804 (domain search list).
4806 * systemd-networkd gained support for serving IPv6 address ranges using
4807 the Router Advertisement protocol. The new .network configuration
4808 section [IPv6Prefix] may be used to configure the ranges to
4809 serve. This is implemented based on a new, minimal, native server
4810 implementation of RA.
4812 * journalctl's --output= switch gained support for a new parameter
4813 "short-iso-precise" for a mode where timestamps are shown as precise
4816 * systemd-udevd's "net_id" builtin may now generate stable network
4817 interface names from IBM PowerVM VIO devices as well as ACPI platform
4820 * MulticastDNS support in systemd-resolved may now be explicitly
4821 enabled/disabled using the new MulticastDNS= configuration file
4824 * systemd-resolved may now optionally use libidn2 instead of the libidn
4825 for processing internationalized domain names. Support for libidn2
4826 should be considered experimental and should not be enabled by
4829 * "machinectl pull-tar" and related call may now do verification of
4830 downloaded images using SUSE-style .sha256 checksum files in addition
4831 to the already existing support for validating using Ubuntu-style
4834 * sd-bus gained support for a new sd_bus_message_appendv() call which
4835 is va_list equivalent of sd_bus_message_append().
4837 * sd-boot gained support for validating images using SHIM/MOK.
4839 * The SMACK code learnt support for "onlycap".
4841 * systemd-mount --umount is now much smarter in figuring out how to
4842 properly unmount a device given its mount or device path.
4844 * The code to call libnss_dns as a fallback from libnss_resolve when
4845 the communication with systemd-resolved fails was removed. This
4846 fallback was redundant and interfered with the [!UNAVAIL=return]
4847 suffix. See nss-resolve(8) for the recommended configuration.
4849 * systemd-logind may now be restarted without losing state. It stores
4850 the file descriptors for devices it manages in the system manager
4851 using the FDSTORE= mechanism. Please note that further changes in
4852 other components may be required to make use of this (for example
4853 Xorg has code to listen for stops of systemd-logind and terminate
4854 itself when logind is stopped or restarted, in order to avoid using
4855 stale file descriptors for graphical devices, which is now
4856 counterproductive and must be reverted in order for restarts of
4857 systemd-logind to be safe. See
4858 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
4860 * All kernel-install plugins are called with the environment variable
4861 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
4862 /etc/machine-id. If the machine ID could not be determined,
4863 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
4864 anything in the entry directory (passed as the second argument) if
4865 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
4866 temporary directory is passed as the entry directory and removed
4867 after all the plugins exit.
4869 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
4870 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
4871 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
4872 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
4873 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
4874 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
4875 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
4876 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
4877 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
4878 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
4879 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
4880 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
4881 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
4882 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
4883 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
4884 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
4885 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
4886 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
4887 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
4888 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
4889 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
4890 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
4891 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
4892 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
4893 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
4894 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
4895 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
4896 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
4897 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
4900 — Berlin, 2017-07-12
4904 * The "hybrid" control group mode has been modified to improve
4905 compatibility with "legacy" cgroups-v1 setups. Specifically, the
4906 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
4907 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
4908 cgroups-v1 hierarchy), the only externally visible change being that
4909 the cgroups-v2 hierarchy is also mounted, to
4910 /sys/fs/cgroup/unified. This should provide a large degree of
4911 compatibility with "legacy" cgroups-v1, while taking benefit of the
4912 better management capabilities of cgroups-v2.
4914 * The default control group setup mode may be selected both a boot-time
4915 via a set of kernel command line parameters (specifically:
4916 systemd.unified_cgroup_hierarchy= and
4917 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
4918 default selected on the configure command line
4919 (--with-default-hierarchy=). The upstream default is "hybrid"
4920 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
4921 this will change in a future systemd version to be "unified" (pure
4922 cgroups-v2 mode). The third option for the compile time option is
4923 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
4924 distributions to default to "hybrid" mode for release distributions,
4925 starting with v233. We recommend "unified" for development
4926 distributions (specifically: distributions such as Fedora's rawhide)
4927 as that's where things are headed in the long run. Use "legacy" for
4928 greatest stability and compatibility only.
4930 * Note one current limitation of "unified" and "hybrid" control group
4931 setup modes: the kernel currently does not permit the systemd --user
4932 instance (i.e. unprivileged code) to migrate processes between two
4933 disconnected cgroup subtrees, even if both are managed and owned by
4934 the user. This effectively means "systemd-run --user --scope" doesn't
4935 work when invoked from outside of any "systemd --user" service or
4936 scope. Specifically, it is not supported from session scopes. We are
4937 working on fixing this in a future systemd version. (See #3388 for
4938 further details about this.)
4940 * DBus policy files are now installed into /usr rather than /etc. Make
4941 sure your system has dbus >= 1.9.18 running before upgrading to this
4942 version, or override the install path with --with-dbuspolicydir= .
4944 * All python scripts shipped with systemd (specifically: the various
4945 tests written in Python) now require Python 3.
4947 * systemd unit tests can now run standalone (without the source or
4948 build directories), and can be installed into /usr/lib/systemd/tests/
4949 with 'make install-tests'.
4951 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
4952 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
4955 * Support for the %c, %r, %R specifiers in unit files has been
4956 removed. Specifiers are not supposed to be dependent on configuration
4957 in the unit file itself (so that they resolve the same regardless
4958 where used in the unit files), but these specifiers were influenced
4959 by the Slice= option.
4961 * The shell invoked by debug-shell.service now defaults to /bin/sh in
4962 all cases. If distributions want to use a different shell for this
4963 purpose (for example Fedora's /sbin/sushell) they need to specify
4964 this explicitly at configure time using --with-debug-shell=.
4966 * The confirmation spawn prompt has been reworked to offer the
4969 (c)ontinue, proceed without asking anymore
4970 (D)ump, show the state of the unit
4971 (f)ail, don't execute the command and pretend it failed
4973 (i)nfo, show a short summary of the unit
4974 (j)obs, show jobs that are in progress
4975 (s)kip, don't execute the command and pretend it succeeded
4976 (y)es, execute the command
4978 The 'n' choice for the confirmation spawn prompt has been removed,
4979 because its meaning was confusing.
4981 The prompt may now also be redirected to an alternative console by
4982 specifying the console as parameter to systemd.confirm_spawn=.
4984 * Services of Type=notify require a READY=1 notification to be sent
4985 during startup. If no such message is sent, the service now fails,
4986 even if the main process exited with a successful exit code.
4988 * Services that fail to start up correctly now always have their
4989 ExecStopPost= commands executed. Previously, they'd enter "failed"
4990 state directly, without executing these commands.
4992 * The option MulticastDNS= of network configuration files has acquired
4993 an actual implementation. With MulticastDNS=yes a host can resolve
4994 names of remote hosts and reply to mDNS A and AAAA requests.
4996 * When units are about to be started an additional check is now done to
4997 ensure that all dependencies of type BindsTo= (when used in
4998 combination with After=) have been started.
5000 * systemd-analyze gained a new verb "syscall-filter" which shows which
5001 system call groups are defined for the SystemCallFilter= unit file
5002 setting, and which system calls they contain.
5004 * A new system call filter group "@filesystem" has been added,
5005 consisting of various file system related system calls. Group
5006 "@reboot" has been added, covering reboot, kexec and shutdown related
5007 calls. Finally, group "@swap" has been added covering swap
5008 configuration related calls.
5010 * A new unit file option RestrictNamespaces= has been added that may be
5011 used to restrict access to the various process namespace types the
5012 Linux kernel provides. Specifically, it may be used to take away the
5013 right for a service unit to create additional file system, network,
5014 user, and other namespaces. This sandboxing option is particularly
5015 relevant due to the high amount of recently discovered namespacing
5016 related vulnerabilities in the kernel.
5018 * systemd-udev's .link files gained support for a new AutoNegotiation=
5019 setting for configuring Ethernet auto-negotiation.
5021 * systemd-networkd's .network files gained support for a new
5022 ListenPort= setting in the [DHCP] section to explicitly configure the
5023 UDP client port the DHCP client shall listen on.
5025 * .network files gained a new Unmanaged= boolean setting for explicitly
5026 excluding one or more interfaces from management by systemd-networkd.
5028 * The systemd-networkd ProxyARP= option has been renamed to
5029 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
5030 renamed to ReduceARPProxy=. The old names continue to be available
5033 * systemd-networkd gained support for configuring IPv6 Proxy NDP
5034 addresses via the new IPv6ProxyNDPAddress= .network file setting.
5036 * systemd-networkd's bonding device support gained support for two new
5037 configuration options ActiveSlave= and PrimarySlave=.
5039 * The various options in the [Match] section of .network files gained
5040 support for negative matching.
5042 * New systemd-specific mount options are now understood in /etc/fstab:
5044 x-systemd.mount-timeout= may be used to configure the maximum
5045 permitted runtime of the mount command.
5047 x-systemd.device-bound may be set to bind a mount point to its
5048 backing device unit, in order to automatically remove a mount point
5049 if its backing device is unplugged. This option may also be
5050 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
5051 on the block device, which is now automatically set for all CDROM
5052 drives, so that mounted CDs are automatically unmounted when they are
5053 removed from the drive.
5055 x-systemd.after= and x-systemd.before= may be used to explicitly
5056 order a mount after or before another unit or mount point.
5058 * Enqueued start jobs for device units are now automatically garbage
5059 collected if there are no jobs waiting for them anymore.
5061 * systemctl list-jobs gained two new switches: with --after, for every
5062 queued job the jobs it's waiting for are shown; with --before the
5063 jobs which it's blocking are shown.
5065 * systemd-nspawn gained support for ephemeral boots from disk images
5066 (or in other words: --ephemeral and --image= may now be
5067 combined). Moreover, ephemeral boots are now supported for normal
5068 directories, even if the backing file system is not btrfs. Of course,
5069 if the file system does not support file system snapshots or
5070 reflinks, the initial copy operation will be relatively expensive, but
5071 this should still be suitable for many use cases.
5073 * Calendar time specifications in .timer units now support
5074 specifications relative to the end of a month by using "~" instead of
5075 "-" as separator between month and day. For example, "*-02~03" means
5076 "the third last day in February". In addition a new syntax for
5077 repeated events has been added using the "/" character. For example,
5078 "9..17/2:00" means "every two hours from 9am to 5pm".
5080 * systemd-socket-proxyd gained a new parameter --connections-max= for
5081 configuring the maximum number of concurrent connections.
5083 * sd-id128 gained a new API for generating unique IDs for the host in a
5084 way that does not leak the machine ID. Specifically,
5085 sd_id128_get_machine_app_specific() derives an ID based on the
5086 machine ID a in well-defined, non-reversible, stable way. This is
5087 useful whenever an identifier for the host is needed but where the
5088 identifier shall not be useful to identify the system beyond the
5089 scope of the application itself. (Internally this uses HMAC-SHA256 as
5090 keyed hash function using the machine ID as input.)
5092 * NotifyAccess= gained a new supported value "exec". When set
5093 notifications are accepted from all processes systemd itself invoked,
5094 including all control processes.
5096 * .nspawn files gained support for defining overlay mounts using the
5097 Overlay= and OverlayReadOnly= options. Previously this functionality
5098 was only available on the systemd-nspawn command line.
5100 * systemd-nspawn's --bind= and --overlay= options gained support for
5101 bind/overlay mounts whose source lies within the container tree by
5102 prefixing the source path with "+".
5104 * systemd-nspawn's --bind= and --overlay= options gained support for
5105 automatically allocating a temporary source directory in /var/tmp
5106 that is removed when the container dies. Specifically, if the source
5107 directory is specified as empty string this mechanism is selected. An
5108 example usage is --overlay=+/var::/var, which creates an overlay
5109 mount based on the original /var contained in the image, overlaid
5110 with a temporary directory in the host's /var/tmp. This way changes
5111 to /var are automatically flushed when the container shuts down.
5113 * systemd-nspawn --image= option does now permit raw file system block
5114 devices (in addition to images containing partition tables, as
5117 * The disk image dissection logic in systemd-nspawn gained support for
5118 automatically setting up LUKS encrypted as well as Verity protected
5119 partitions. When a container is booted from an encrypted image the
5120 passphrase is queried at start-up time. When a container with Verity
5121 data is started, the root hash is search in a ".roothash" file
5122 accompanying the disk image (alternatively, pass the root hash via
5123 the new --root-hash= command line option).
5125 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
5126 be used to dissect disk images the same way as systemd-nspawn does
5127 it, following the Bootable Partition Specification. It may even be
5128 used to mount disk images with complex partition setups (including
5129 LUKS and Verity partitions) to a local host directory, in order to
5130 inspect them. This tool is not considered public API (yet), and is
5131 thus not installed into /usr/bin. Please do not rely on its
5132 existence, since it might go away or be changed in later systemd
5135 * A new generator "systemd-verity-generator" has been added, similar in
5136 style to "systemd-cryptsetup-generator", permitting automatic setup of
5137 Verity root partitions when systemd boots up. In order to make use of
5138 this your partition setup should follow the Discoverable Partitions
5139 Specification, and the GPT partition ID of the root file system
5140 partition should be identical to the upper 128bit of the Verity root
5141 hash. The GPT partition ID of the Verity partition protecting it
5142 should be the lower 128bit of the Verity root hash. If the partition
5143 image follows this model it is sufficient to specify a single
5144 "roothash=" kernel command line argument to both configure which root
5145 image and verity partition to use as well as the root hash for
5146 it. Note that systemd-nspawn's Verity support follows the same
5147 semantics, meaning that disk images with proper Verity data in place
5148 may be booted in containers with systemd-nspawn as well as on
5149 physical systems via the verity generator. Also note that the "mkosi"
5150 tool available at https://github.com/systemd/mkosi has been updated
5151 to generate Verity protected disk images following this scheme. In
5152 fact, it has been updated to generate disk images that optionally
5153 implement a complete UEFI SecureBoot trust chain, involving a signed
5154 kernel and initrd image that incorporates such a root hash as well as
5155 a Verity-enabled root partition.
5157 * The hardware database (hwdb) udev supports has been updated to carry
5158 accelerometer quirks.
5160 * All system services are now run with a fresh kernel keyring set up
5161 for them. The invocation ID is stored by default in it, thus
5162 providing a safe, non-overridable way to determine the invocation
5165 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
5166 options for bind mounting arbitrary paths in a service-specific
5167 way. When these options are used, arbitrary host or service files and
5168 directories may be mounted to arbitrary locations in the service's
5171 * Documentation has been added that lists all of systemd's low-level
5172 environment variables:
5174 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
5176 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
5177 whether a specific socket file descriptor matches a specified socket
5180 * systemd-firstboot has been updated to check for the
5181 systemd.firstboot= kernel command line option. It accepts a boolean
5182 and when set to false the first boot questions are skipped.
5184 * systemd-fstab-generator has been updated to check for the
5185 systemd.volatile= kernel command line option, which either takes an
5186 optional boolean parameter or the special value "state". If used the
5187 system may be booted in a "volatile" boot mode. Specifically,
5188 "systemd.volatile" is used, the root directory will be mounted as
5189 tmpfs, and only /usr is mounted from the actual root file system. If
5190 "systemd.volatile=state" is used, the root directory will be mounted
5191 as usual, but /var is mounted as tmpfs. This concept provides similar
5192 functionality as systemd-nspawn's --volatile= option, but provides it
5193 on physical boots. Use this option for implementing stateless
5194 systems, or testing systems with all state and/or configuration reset
5195 to the defaults. (Note though that many distributions are not
5196 prepared to boot up without a populated /etc or /var, though.)
5198 * systemd-gpt-auto-generator gained support for LUKS encrypted root
5199 partitions. Previously it only supported LUKS encrypted partitions
5200 for all other uses, except for the root partition itself.
5202 * Socket units gained support for listening on AF_VSOCK sockets for
5203 communication in virtualized QEMU environments.
5205 * The "configure" script gained a new option --with-fallback-hostname=
5206 for specifying the fallback hostname to use if none is configured in
5207 /etc/hostname. For example, by specifying
5208 --with-fallback-hostname=fedora it is possible to default to a
5209 hostname of "fedora" on pristine installations.
5211 * systemd-cgls gained support for a new --unit= switch for listing only
5212 the control groups of a specific unit. Similar --user-unit= has been
5213 added for listing only the control groups of a specific user unit.
5215 * systemd-mount gained a new --umount switch for unmounting a mount or
5216 automount point (and all mount/automount points below it).
5218 * systemd will now refuse full configuration reloads (via systemctl
5219 daemon-reload and related calls) unless at least 16MiB of free space
5220 are available in /run. This is a safety precaution in order to ensure
5221 that generators can safely operate after the reload completed.
5223 * A new unit file option RootImage= has been added, which has a similar
5224 effect as RootDirectory= but mounts the service's root directory from
5225 a disk image instead of plain directory. This logic reuses the same
5226 image dissection and mount logic that systemd-nspawn already uses,
5227 and hence supports any disk images systemd-nspawn supports, including
5228 those following the Discoverable Partition Specification, as well as
5229 Verity enabled images. This option enables systemd to run system
5230 services directly off disk images acting as resource bundles,
5231 possibly even including full integrity data.
5233 * A new MountAPIVFS= unit file option has been added, taking a boolean
5234 argument. If enabled /proc, /sys and /dev (collectively called the
5235 "API VFS") will be mounted for the service. This is only relevant if
5236 RootDirectory= or RootImage= is used for the service, as these mounts
5237 are of course in place in the host mount namespace anyway.
5239 * systemd-nspawn gained support for a new --pivot-root= switch. If
5240 specified the root directory within the container image is pivoted to
5241 the specified mount point, while the original root disk is moved to a
5242 different place. This option enables booting of ostree images
5243 directly with systemd-nspawn.
5245 * The systemd build scripts will no longer complain if the NTP server
5246 addresses are not changed from the defaults. Google now supports
5247 these NTP servers officially. We still recommend downstreams to
5248 properly register an NTP pool with the NTP pool project though.
5250 * coredumpctl gained a new "--reverse" option for printing the list
5251 of coredumps in reverse order.
5253 * coredumpctl will now show additional information about truncated and
5254 inaccessible coredumps, as well as coredumps that are still being
5255 processed. It also gained a new --quiet switch for suppressing
5256 additional informational message in its output.
5258 * coredumpctl gained support for only showing coredumps newer and/or
5259 older than specific timestamps, using the new --since= and --until=
5260 options, reminiscent of journalctl's options by the same name.
5262 * The systemd-coredump logic has been improved so that it may be reused
5263 to collect backtraces in non-compiled languages, for example in
5264 scripting languages such as Python.
5266 * machinectl will now show the UID shift of local containers, if user
5267 namespacing is enabled for them.
5269 * systemd will now optionally run "environment generator" binaries at
5270 configuration load time. They may be used to add environment
5271 variables to the environment block passed to services invoked. One
5272 user environment generator is shipped by default that sets up
5273 environment variables based on files dropped into /etc/environment.d
5274 and ~/.config/environment.d/.
5276 * systemd-resolved now includes the new, recently published 2017 DNSSEC
5279 * hostnamed has been updated to report a new chassis type of
5280 "convertible" to cover "foldable" laptops that can both act as a
5281 tablet and as a laptop, such as various Lenovo Yoga devices.
5283 Contributions from: Adrián López, Alexander Galanin, Alexander
5284 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
5285 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
5286 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
5287 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
5288 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
5289 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
5290 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
5291 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
5292 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
5293 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
5294 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
5295 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
5296 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
5297 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
5298 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
5299 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
5300 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
5301 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
5302 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
5303 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
5304 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
5305 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
5306 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
5307 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
5308 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
5309 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
5310 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
5313 — Berlin, 2017-03-01
5317 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
5318 RestrictAddressFamilies= enabled. These sandboxing options should
5319 generally be compatible with the various external udev call-out
5320 binaries we are aware of, however there may be exceptions, in
5321 particular when exotic languages for these call-outs are used. In
5322 this case, consider turning off these settings locally.
5324 * The new RemoveIPC= option can be used to remove IPC objects owned by
5325 the user or group of a service when that service exits.
5327 * The new ProtectKernelModules= option can be used to disable explicit
5328 load and unload operations of kernel modules by a service. In
5329 addition access to /usr/lib/modules is removed if this option is set.
5331 * ProtectSystem= option gained a new value "strict", which causes the
5332 whole file system tree with the exception of /dev, /proc, and /sys,
5333 to be remounted read-only for a service.
5335 * The new ProtectKernelTunables= option can be used to disable
5336 modification of configuration files in /sys and /proc by a service.
5337 Various directories and files are remounted read-only, so access is
5338 restricted even if the file permissions would allow it.
5340 * The new ProtectControlGroups= option can be used to disable write
5341 access by a service to /sys/fs/cgroup.
5343 * Various systemd services have been hardened with
5344 ProtectKernelTunables=yes, ProtectControlGroups=yes,
5345 RestrictAddressFamilies=.
5347 * Support for dynamically creating users for the lifetime of a service
5348 has been added. If DynamicUser=yes is specified, user and group IDs
5349 will be allocated from the range 61184…65519 for the lifetime of the
5350 service. They can be resolved using the new nss-systemd.so NSS
5351 module. The module must be enabled in /etc/nsswitch.conf. Services
5352 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
5353 any resources allocated by the service will be cleaned up when the
5354 service exits. They also have ProtectHome=read-only and
5355 ProtectSystem=strict enabled, so they are not able to make any
5356 permanent modifications to the system.
5358 * The nss-systemd module also always resolves root and nobody, making
5359 it possible to have no /etc/passwd or /etc/group files in minimal
5360 container or chroot environments.
5362 * Services may be started with their own user namespace using the new
5363 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
5364 under which the service is running are mapped. All other users are
5367 * Support for the cgroup namespace has been added to systemd-nspawn. If
5368 supported by kernel, the container system started by systemd-nspawn
5369 will have its own view of the cgroup hierarchy. This new behaviour
5370 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
5372 * The new MemorySwapMax= option can be used to limit the maximum swap
5373 usage under the unified cgroup hierarchy.
5375 * Support for the CPU controller in the unified cgroup hierarchy has
5376 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
5377 options. This controller requires out-of-tree patches for the kernel
5378 and the support is provisional.
5380 * Mount and automount units may now be created transiently
5381 (i.e. dynamically at runtime via the bus API, instead of requiring
5382 unit files in the file system).
5384 * systemd-mount is a new tool which may mount file systems – much like
5385 mount(8), optionally pulling in additional dependencies through
5386 transient .mount and .automount units. For example, this tool
5387 automatically runs fsck on a backing block device before mounting,
5388 and allows the automount logic to be used dynamically from the
5389 command line for establishing mount points. This tool is particularly
5390 useful when dealing with removable media, as it will ensure fsck is
5391 run – if necessary – before the first access and that the file system
5392 is quickly unmounted after each access by utilizing the automount
5393 logic. This maximizes the chance that the file system on the
5394 removable media stays in a clean state, and if it isn't in a clean
5395 state is fixed automatically.
5397 * LazyUnmount=yes option for mount units has been added to expose the
5398 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
5401 * /efi will be used as the mount point of the EFI boot partition, if
5402 the directory is present, and the mount point was not configured
5403 through other means (e.g. fstab). If /efi directory does not exist,
5404 /boot will be used as before. This makes it easier to automatically
5405 mount the EFI partition on systems where /boot is used for something
5408 * When operating on GPT disk images for containers, systemd-nspawn will
5409 now mount the ESP to /boot or /efi according to the same rules as PID
5410 1 running on a host. This allows tools like "bootctl" to operate
5411 correctly within such containers, in order to make container images
5412 bootable on physical systems.
5414 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
5416 * Two new user session targets have been added to support running
5417 graphical sessions under the systemd --user instance:
5418 graphical-session.target and graphical-session-pre.target. See
5419 systemd.special(7) for a description of how those targets should be
5422 * The vconsole initialization code has been significantly reworked to
5423 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
5424 support unicode keymaps. Font and keymap configuration will now be
5425 copied to all allocated virtual consoles.
5427 * FreeBSD's bhyve virtualization is now detected.
5429 * Information recorded in the journal for core dumps now includes the
5430 contents of /proc/mountinfo and the command line of the process at
5431 the top of the process hierarchy (which is usually the init process
5434 * systemd-journal-gatewayd learned the --directory= option to serve
5435 files from the specified location.
5437 * journalctl --root=… can be used to peruse the journal in the
5438 /var/log/ directories inside of a container tree. This is similar to
5439 the existing --machine= option, but does not require the container to
5442 * The hardware database has been extended to support
5443 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
5446 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
5447 specify the click rate for mice which include a horizontal wheel with
5448 a click rate that is different than the one for the vertical wheel.
5450 * systemd-run gained a new --wait option that makes service execution
5451 synchronous. (Specifically, the command will not return until the
5452 specified service binary exited.)
5454 * systemctl gained a new --wait option that causes the start command to
5455 wait until the units being started have terminated again.
5457 * A new journal output mode "short-full" has been added which displays
5458 timestamps with abbreviated English day names and adds a timezone
5459 suffix. Those timestamps include more information than the default
5460 "short" output mode, and can be passed directly to journalctl's
5461 --since= and --until= options.
5463 * /etc/resolv.conf will be bind-mounted into containers started by
5464 systemd-nspawn, if possible, so any changes to resolv.conf contents
5465 are automatically propagated to the container.
5467 * The number of instances for socket-activated services originating
5468 from a single IP address can be limited with
5469 MaxConnectionsPerSource=, extending the existing setting of
5472 * systemd-networkd gained support for vcan ("Virtual CAN") interface
5475 * .netdev and .network configuration can now be extended through
5478 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
5479 Segmentation Offload, Generic Receive Offload, Large Receive Offload
5480 can be enabled and disabled using the new UDPSegmentationOffload=,
5481 TCPSegmentationOffload=, GenericSegmentationOffload=,
5482 GenericReceiveOffload=, LargeReceiveOffload= options in the
5483 [Link] section of .link files.
5485 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
5486 Port VLAN ID can be configured for bridge devices using the new STP=,
5487 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
5488 section of .netdev files.
5490 * The route table to which routes received over DHCP or RA should be
5491 added can be configured with the new RouteTable= option in the [DHCP]
5492 and [IPv6AcceptRA] sections of .network files.
5494 * The Address Resolution Protocol can be disabled on links managed by
5495 systemd-networkd using the ARP=no setting in the [Link] section of
5498 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
5499 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
5500 encode information about the result and exit codes of the current
5501 service runtime cycle.
5503 * systemd-sysctl will now configure kernel parameters in the order
5504 they occur in the configuration files. This matches what sysctl
5505 has been traditionally doing.
5507 * kernel-install "plugins" that are executed to perform various
5508 tasks after a new kernel is added and before an old one is removed
5509 can now return a special value to terminate the procedure and
5510 prevent any later plugins from running.
5512 * Journald's SplitMode=login setting has been deprecated. It has been
5513 removed from documentation, and its use is discouraged. In a future
5514 release it will be completely removed, and made equivalent to current
5515 default of SplitMode=uid.
5517 * Storage=both option setting in /etc/systemd/coredump.conf has been
5518 removed. With fast LZ4 compression storing the core dump twice is not
5521 * The --share-system systemd-nspawn option has been replaced with an
5522 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
5523 this functionality is discouraged. In addition the variables
5524 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
5525 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
5526 individual namespaces.
5528 * "machinectl list" now shows the IP address of running containers in
5529 the output, as well as OS release information.
5531 * "loginctl list" now shows the TTY of each session in the output.
5533 * sd-bus gained new API calls sd_bus_track_set_recursive(),
5534 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
5535 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
5536 tracking objects in a "recursive" mode, where a single client can be
5537 counted multiple times, if it takes multiple references.
5539 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
5540 sd_bus_get_exit_on_disconnect(). They may be used to make a
5541 process using sd-bus automatically exit if the bus connection is
5544 * Bus clients of the service manager may now "pin" loaded units into
5545 memory, by taking an explicit reference on them. This is useful to
5546 ensure the client can retrieve runtime data about the service even
5547 after the service completed execution. Taking such a reference is
5548 available only for privileged clients and should be helpful to watch
5549 running services in a race-free manner, and in particular collect
5550 information about exit statuses and results.
5552 * The nss-resolve module has been changed to strictly return UNAVAIL
5553 when communication via D-Bus with resolved failed, and NOTFOUND when
5554 a lookup completed but was negative. This means it is now possible to
5555 neatly configure fallbacks using nsswitch.conf result checking
5556 expressions. Taking benefit of this, the new recommended
5557 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
5559 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
5561 * A new setting CtrlAltDelBurstAction= has been added to
5562 /etc/systemd/system.conf which may be used to configure the precise
5563 behaviour if the user on the console presses Ctrl-Alt-Del more often
5564 than 7 times in 2s. Previously this would unconditionally result in
5565 an expedited, immediate reboot. With this new setting the precise
5566 operation may be configured in more detail, and also turned off
5569 * In .netdev files two new settings RemoteChecksumTx= and
5570 RemoteChecksumRx= are now understood that permit configuring the
5571 remote checksumming logic for VXLAN networks.
5573 * The service manager learnt a new "invocation ID" concept for invoked
5574 services. Each runtime cycle of a service will get a new invocation
5575 ID (a 128bit random UUID) assigned that identifies the current
5576 run of the service uniquely and globally. A new invocation ID
5577 is generated each time a service starts up. The journal will store
5578 the invocation ID of a service along with any logged messages, thus
5579 making the invocation ID useful for matching the online runtime of a
5580 service with the offline log data it generated in a safe way without
5581 relying on synchronized timestamps. In many ways this new service
5582 invocation ID concept is similar to the kernel's boot ID concept that
5583 uniquely and globally identifies the runtime of each boot. The
5584 invocation ID of a service is passed to the service itself via an
5585 environment variable ($INVOCATION_ID). A new bus call
5586 GetUnitByInvocationID() has been added that is similar to GetUnit()
5587 but instead of retrieving the bus path for a unit by its name
5588 retrieves it by its invocation ID. The returned path is valid only as
5589 long as the passed invocation ID is current.
5591 * systemd-resolved gained a new "DNSStubListener" setting in
5592 resolved.conf. It either takes a boolean value or the special values
5593 "udp" and "tcp", and configures whether to enable the stub DNS
5594 listener on 127.0.0.53:53.
5596 * IP addresses configured via networkd may now carry additional
5597 configuration settings supported by the kernel. New options include:
5598 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
5599 PrefixRoute=, AutoJoin=.
5601 * The PAM configuration fragment file for "user@.service" shipped with
5602 systemd (i.e. the --user instance of systemd) has been stripped to
5603 the minimum necessary to make the system boot. Previously, it
5604 contained Fedora-specific stanzas that did not apply to other
5605 distributions. It is expected that downstream distributions add
5606 additional configuration lines, matching their needs to this file,
5607 using it only as rough template of what systemd itself needs. Note
5608 that this reduced fragment does not even include an invocation of
5609 pam_limits which most distributions probably want to add, even though
5610 systemd itself does not need it. (There's also the new build time
5611 option --with-pamconfdir=no to disable installation of the PAM
5614 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
5615 capability is now also dropped from its set (in addition to
5616 CAP_SYS_MKNOD as before).
5618 * In service unit files it is now possible to connect a specific named
5619 file descriptor with stdin/stdout/stdout of an executed service. The
5620 name may be specified in matching .socket units using the
5621 FileDescriptorName= setting.
5623 * A number of journal settings may now be configured on the kernel
5624 command line. Specifically, the following options are now understood:
5625 systemd.journald.max_level_console=,
5626 systemd.journald.max_level_store=,
5627 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
5628 systemd.journald.max_level_wall=.
5630 * "systemctl is-enabled --full" will now show by which symlinks a unit
5631 file is enabled in the unit dependency tree.
5633 * Support for VeraCrypt encrypted partitions has been added to the
5634 "cryptsetup" logic and /etc/crypttab.
5636 * systemd-detect-virt gained support for a new --private-users switch
5637 that checks whether the invoking processes are running inside a user
5638 namespace. Similar, a new special value "private-users" for the
5639 existing ConditionVirtualization= setting has been added, permitting
5640 skipping of specific units in user namespace environments.
5642 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
5643 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
5644 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
5645 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
5646 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
5647 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
5648 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
5649 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
5650 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
5651 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
5652 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
5653 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
5654 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
5655 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
5656 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
5657 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
5658 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
5659 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
5660 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
5661 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
5662 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
5663 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
5664 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
5665 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
5666 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
5667 Jędrzejewski-Szmek, Zeal Jagannatha
5669 — Santa Fe, 2016-11-03
5673 * In service units the various ExecXYZ= settings have been extended
5674 with an additional special character as first argument of the
5675 assigned value: if the character '+' is used the specified command
5676 line it will be run with full privileges, regardless of User=,
5677 Group=, CapabilityBoundingSet= and similar options. The effect is
5678 similar to the existing PermissionsStartOnly= option, but allows
5679 configuration of this concept for each executed command line
5682 * Services may now alter the service watchdog timeout at runtime by
5683 sending a WATCHDOG_USEC= message via sd_notify().
5685 * MemoryLimit= and related unit settings now optionally take percentage
5686 specifications. The percentage is taken relative to the amount of
5687 physical memory in the system (or in case of containers, the assigned
5688 amount of memory). This allows scaling service resources neatly with
5689 the amount of RAM available on the system. Similarly, systemd-logind's
5690 RuntimeDirectorySize= option now also optionally takes percentage
5693 * In similar fashion TasksMax= takes percentage values now, too. The
5694 value is taken relative to the configured maximum number of processes
5695 on the system. The per-service task maximum has been changed to 15%
5696 using this functionality. (Effectively this is an increase of 512 →
5697 4915 for service units, given the kernel's default pid_max setting.)
5699 * Calendar time specifications in .timer units now understand a ".."
5700 syntax for time ranges. Example: "4..7:10" may now be used for
5701 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
5704 * The InaccessableDirectories=, ReadOnlyDirectories= and
5705 ReadWriteDirectories= unit file settings have been renamed to
5706 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
5707 applied to all kinds of file nodes, and not just directories, with
5708 the exception of symlinks. Specifically these settings may now be
5709 used on block and character device nodes, UNIX sockets and FIFOS as
5710 well as regular files. The old names of these settings remain
5711 available for compatibility.
5713 * systemd will now log about all service processes it kills forcibly
5714 (using SIGKILL) because they remained after the clean shutdown phase
5715 of the service completed. This should help identifying services that
5716 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
5717 systemd-logind's configuration a similar log message is generated for
5718 processes killed at the end of each session due to this setting.
5720 * systemd will now set the $JOURNAL_STREAM environment variable for all
5721 services whose stdout/stderr are connected to the Journal (which
5722 effectively means by default: all services). The variable contains
5723 the device and inode number of the file descriptor used for
5724 stdout/stderr. This may be used by invoked programs to detect whether
5725 their stdout/stderr is connected to the Journal, in which case they
5726 can switch over to direct Journal communication, thus being able to
5727 pass extended, structured metadata along with their log messages. As
5728 one example, this is now used by glib's logging primitives.
5730 * When using systemd's default tmp.mount unit for /tmp, the mount point
5731 will now be established with the "nosuid" and "nodev" options. This
5732 avoids privilege escalation attacks that put traps and exploits into
5733 /tmp. However, this might cause problems if you e. g. put container
5734 images or overlays into /tmp; if you need this, override tmp.mount's
5735 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
5738 * systemd now supports the "memory" cgroup controller also on
5741 * The systemd-cgtop tool now optionally takes a control group path as
5742 command line argument. If specified, the control group list shown is
5743 limited to subgroups of that group.
5745 * The SystemCallFilter= unit file setting gained support for
5746 pre-defined, named system call filter sets. For example
5747 SystemCallFilter=@clock is now an effective way to make all clock
5748 changing-related system calls unavailable to a service. A number of
5749 similar pre-defined groups are defined. Writing system call filters
5750 for system services is simplified substantially with this new
5751 concept. Accordingly, all of systemd's own, long-running services now
5752 enable system call filtering based on this, by default.
5754 * A new service setting MemoryDenyWriteExecute= has been added, taking
5755 a boolean value. If turned on, a service may no longer create memory
5756 mappings that are writable and executable at the same time. This
5757 enhances security for services where this is enabled as it becomes
5758 harder to dynamically write and then execute memory in exploited
5759 service processes. This option has been enabled for all of systemd's
5760 own long-running services.
5762 * A new RestrictRealtime= service setting has been added, taking a
5763 boolean argument. If set the service's processes may no longer
5764 acquire realtime scheduling. This improves security as realtime
5765 scheduling may otherwise be used to easily freeze the system.
5767 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
5768 value. This may be used for requesting that the system manager inside
5769 of the container reports start-up completion to nspawn which then
5770 propagates this notification further to the service manager
5771 supervising nspawn itself. A related option NotifyReady= in .nspawn
5772 files has been added too. This functionality allows ordering of the
5773 start-up of multiple containers using the usual systemd ordering
5776 * machinectl gained a new command "stop" that is an alias for
5779 * systemd-resolved gained support for contacting DNS servers on
5780 link-local IPv6 addresses.
5782 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
5783 its caches. A method call for requesting the same operation has been
5784 added to the bus API too, and is made available via "systemd-resolve
5787 * systemd-resolve gained a new --status switch. If passed a brief
5788 summary of the used DNS configuration with per-interface information
5791 * resolved.conf gained a new Cache= boolean option, defaulting to
5792 on. If turned off local DNS caching is disabled. This comes with a
5793 performance penalty in particular when DNSSEC is enabled. Note that
5794 resolved disables its internal caching implicitly anyway, when the
5795 configured DNS server is on a host-local IP address such as ::1 or
5796 127.0.0.1, thus automatically avoiding double local caching.
5798 * systemd-resolved now listens on the local IP address 127.0.0.53:53
5799 for DNS requests. This improves compatibility with local programs
5800 that do not use the libc NSS or systemd-resolved's bus APIs for name
5801 resolution. This minimal DNS service is only available to local
5802 programs and does not implement the full DNS protocol, but enough to
5803 cover local DNS clients. A new, static resolv.conf file, listing just
5804 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
5805 now recommended to make /etc/resolv.conf a symlink to this file in
5806 order to route all DNS lookups to systemd-resolved, regardless if
5807 done via NSS, the bus API or raw DNS packets. Note that this local
5808 DNS service is not as fully featured as the libc NSS or
5809 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
5810 used to deliver link-local address information (as this implies
5811 sending a local interface index along), LLMNR/mDNS support via this
5812 interface is severely restricted. It is thus strongly recommended for
5813 all applications to use the libc NSS API or native systemd-resolved
5816 * systemd-networkd's bridge support learned a new setting
5817 VLANFiltering= for controlling VLAN filtering. Moreover a new section
5818 in .network files has been added for configuring VLAN bridging in
5819 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
5821 * systemd-networkd's IPv6 Router Advertisement code now makes use of
5822 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
5823 now be acquired without relying on DHCPv6. Two new options
5824 UseDomains= and UseDNS= have been added to configure this behaviour.
5826 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
5827 renamed IPv6AcceptRA=, without altering its behaviour. The old
5828 setting name remains available for compatibility reasons.
5830 * The systemd-networkd VTI/VTI6 tunneling support gained new options
5831 Key=, InputKey= and OutputKey=.
5833 * systemd-networkd gained support for VRF ("Virtual Routing Function")
5834 interface configuration.
5836 * "systemctl edit" may now be used to create new unit files by
5837 specifying the --force switch.
5839 * sd-event gained a new function sd_event_get_iteration() for
5840 requesting the current iteration counter of the event loop. It starts
5841 at zero and is increased by one with each event loop iteration.
5843 * A new rpm macro %systemd_ordering is provided by the macros.systemd
5844 file. It can be used in lieu of %systemd_requires in packages which
5845 don't use any systemd functionality and are intended to be installed
5846 in minimal containers without systemd present. This macro provides
5847 ordering dependencies to ensure that if the package is installed in
5848 the same rpm transaction as systemd, systemd will be installed before
5849 the scriptlets for the package are executed, allowing unit presets
5852 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
5853 been added to simplify packaging of generators.
5855 * The os-release file gained VERSION_CODENAME field for the
5856 distribution nickname (e.g. VERSION_CODENAME=woody).
5858 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
5859 can be set to disable parsing of metadata and the creation
5860 of persistent symlinks for that device.
5862 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
5863 to make them available to logged-in users has been reverted.
5865 * Much of the common code of the various systemd components is now
5866 built into an internal shared library libsystemd-shared-231.so
5867 (incorporating the systemd version number in the name, to be updated
5868 with future releases) that the components link to. This should
5869 decrease systemd footprint both in memory during runtime and on
5870 disk. Note that the shared library is not for public use, and is
5871 neither API nor ABI stable, but is likely to change with every new
5872 released update. Packagers need to make sure that binaries
5873 linking to libsystemd-shared.so are updated in step with the
5876 * Configuration for "mkosi" is now part of the systemd
5877 repository. mkosi is a tool to easily build legacy-free OS images,
5878 and is available on github: https://github.com/systemd/mkosi. If
5879 "mkosi" is invoked in the build tree a new raw OS image is generated
5880 incorporating the systemd sources currently being worked on and a
5881 clean, fresh distribution installation. The generated OS image may be
5882 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
5883 UEFI PC. This functionality is particularly useful to easily test
5884 local changes made to systemd in a pristine, defined environment. See
5885 doc/HACKING for details.
5887 * configure learned the --with-support-url= option to specify the
5888 distribution's bugtracker.
5890 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
5891 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
5892 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
5893 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
5894 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
5895 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
5896 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
5897 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
5898 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
5899 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
5900 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
5901 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
5902 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
5903 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
5904 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
5905 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
5906 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
5907 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
5908 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
5910 — Berlin, 2016-07-25
5914 * DNSSEC is now turned on by default in systemd-resolved (in
5915 "allow-downgrade" mode), but may be turned off during compile time by
5916 passing "--with-default-dnssec=no" to "configure" (and of course,
5917 during runtime with DNSSEC= in resolved.conf). We recommend
5918 downstreams to leave this on at least during development cycles and
5919 report any issues with the DNSSEC logic upstream. We are very
5920 interested in collecting feedback about the DNSSEC validator and its
5921 limitations in the wild. Note however, that DNSSEC support is
5922 probably nothing downstreams should turn on in stable distros just
5923 yet, as it might create incompatibilities with a few DNS servers and
5924 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
5925 automatically whenever we detect such incompatible setups, but there
5926 might be systems we do not cover yet. Hence: please help us testing
5927 the DNSSEC code, leave this on where you can, report back, but then
5928 again don't consider turning this on in your stable, LTS or
5929 production release just yet. (Note that you have to enable
5930 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
5931 and its DNSSEC mode for hostname resolution from local
5934 * systemd-resolve conveniently resolves DANE records with the --tlsa
5935 option and OPENPGPKEY records with the --openpgp option. It also
5936 supports dumping raw DNS record data via the new --raw= switch.
5938 * systemd-logind will now by default terminate user processes that are
5939 part of the user session scope unit (session-XX.scope) when the user
5940 logs out. This behavior is controlled by the KillUserProcesses=
5941 setting in logind.conf, and the previous default of "no" is now
5942 changed to "yes". This means that user sessions will be properly
5943 cleaned up after, but additional steps are necessary to allow
5944 intentionally long-running processes to survive logout.
5946 While the user is logged in at least once, user@.service is running,
5947 and any service that should survive the end of any individual login
5948 session can be started at a user service or scope using systemd-run.
5949 systemd-run(1) man page has been extended with an example which shows
5950 how to run screen in a scope unit underneath user@.service. The same
5951 command works for tmux.
5953 After the user logs out of all sessions, user@.service will be
5954 terminated too, by default, unless the user has "lingering" enabled.
5955 To effectively allow users to run long-term tasks even if they are
5956 logged out, lingering must be enabled for them. See loginctl(1) for
5957 details. The default polkit policy was modified to allow users to
5958 set lingering for themselves without authentication.
5960 Previous defaults can be restored at compile time by the
5961 --without-kill-user-processes option to "configure".
5963 * systemd-logind gained new configuration settings SessionsMax= and
5964 InhibitorsMax=, both with a default of 8192. It will not register new
5965 user sessions or inhibitors above this limit.
5967 * systemd-logind will now reload configuration on SIGHUP.
5969 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
5970 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
5971 enable. Also, support for the "io" cgroup controller in the unified
5972 hierarchy has been added, so that the "memory", "pids" and "io" are
5973 now the controllers that are supported on the unified hierarchy.
5975 WARNING: it is not possible to use previous systemd versions with
5976 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
5977 is necessary to also update systemd in the initramfs if using the
5978 unified hierarchy. An updated SELinux policy is also required.
5980 * LLDP support has been extended, and both passive (receive-only) and
5981 active (sender) modes are supported. Passive mode ("routers-only") is
5982 enabled by default in systemd-networkd. Active LLDP mode is enabled
5983 by default for containers on the internal network. The "networkctl
5984 lldp" command may be used to list information gathered. "networkctl
5985 status" will also show basic LLDP information on connected peers now.
5987 * The IAID and DUID unique identifier sent in DHCP requests may now be
5988 configured for the system and each .network file managed by
5989 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
5991 * systemd-networkd gained support for configuring proxy ARP support for
5992 each interface, via the ProxyArp= setting in .network files. It also
5993 gained support for configuring the multicast querier feature of
5994 bridge devices, via the new MulticastQuerier= setting in .netdev
5995 files. Similarly, snooping on the IGMP traffic can be controlled
5996 via the new setting MulticastSnooping=.
5998 A new setting PreferredLifetime= has been added for addresses
5999 configured in .network file to configure the lifetime intended for an
6002 The systemd-networkd DHCP server gained the option EmitRouter=, which
6003 defaults to yes, to configure whether the DHCP Option 3 (Router)
6006 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
6007 systemd-socket-activate and installed into /usr/bin. It is now fully
6010 * systemd-journald now uses separate threads to flush changes to disk
6011 when closing journal files, thus reducing impact of slow disk I/O on
6012 logging performance.
6014 * The sd-journal API gained two new calls
6015 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
6016 can be used to open journal files using file descriptors instead of
6017 file or directory paths. sd_journal_open_container() has been
6018 deprecated, sd_journal_open_directory_fd() should be used instead
6019 with the flag SD_JOURNAL_OS_ROOT.
6021 * journalctl learned a new output mode "-o short-unix" that outputs log
6022 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
6023 UTC). It also gained support for a new --no-hostname setting to
6024 suppress the hostname column in the family of "short" output modes.
6026 * systemd-ask-password now optionally skips printing of the password to
6027 stdout with --no-output which can be useful in scripts.
6029 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
6030 (devices tagged with ID_MAKER_TOOL) are now tagged with
6031 "uaccess" and are available to logged in users.
6033 * The DeviceAllow= unit setting now supports specifiers (with "%").
6035 * "systemctl show" gained a new --value switch, which allows print a
6036 only the contents of a specific unit property, without also printing
6037 the property's name. Similar support was added to "show*" verbs
6038 of loginctl and machinectl that output "key=value" lists.
6040 * A new unit type "generated" was added for files dynamically generated
6041 by generator tools. Similarly, a new unit type "transient" is used
6042 for unit files created using the runtime API. "systemctl enable" will
6043 refuse to operate on such files.
6045 * A new command "systemctl revert" has been added that may be used to
6046 revert to the vendor version of a unit file, in case local changes
6047 have been made by adding drop-ins or overriding the unit file.
6049 * "machinectl clean" gained a new verb to automatically remove all or
6050 just hidden container images.
6052 * systemd-tmpfiles gained support for a new line type "e" for emptying
6053 directories, if they exist, without creating them if they don't.
6055 * systemd-nspawn gained support for automatically patching the UID/GIDs
6056 of the owners and the ACLs of all files and directories in a
6057 container tree to match the UID/GID user namespacing range selected
6058 for the container invocation. This mode is enabled via the new
6059 --private-users-chown switch. It also gained support for
6060 automatically choosing a free, previously unused UID/GID range when
6061 starting a container, via the new --private-users=pick setting (which
6062 implies --private-users-chown). Together, these options for the first
6063 time make user namespacing for nspawn containers fully automatic and
6064 thus deployable. The systemd-nspawn@.service template unit file has
6065 been changed to use this functionality by default.
6067 * systemd-nspawn gained a new --network-zone= switch, that allows
6068 creating ad-hoc virtual Ethernet links between multiple containers,
6069 that only exist as long as at least one container referencing them is
6070 running. This allows easy connecting of multiple containers with a
6071 common link that implements an Ethernet broadcast domain. Each of
6072 these network "zones" may be named relatively freely by the user, and
6073 may be referenced by any number of containers, but each container may
6074 only reference one of these "zones". On the lower level, this is
6075 implemented by an automatically managed bridge network interface for
6076 each zone, that is created when the first container referencing its
6077 zone is created and removed when the last one referencing its zone
6080 * The default start timeout may now be configured on the kernel command
6081 line via systemd.default_timeout_start_sec=. It was already
6082 configurable via the DefaultTimeoutStartSec= option in
6083 /etc/systemd/system.conf.
6085 * Socket units gained a new TriggerLimitIntervalSec= and
6086 TriggerLimitBurst= setting to configure a limit on the activation
6087 rate of the socket unit.
6089 * The LimitNICE= setting now optionally takes normal UNIX nice values
6090 in addition to the raw integer limit value. If the specified
6091 parameter is prefixed with "+" or "-" and is in the range -20…19 the
6092 value is understood as UNIX nice value. If not prefixed like this it
6093 is understood as raw RLIMIT_NICE limit.
6095 * Note that the effect of the PrivateDevices= unit file setting changed
6096 slightly with this release: the per-device /dev file system will be
6097 mounted read-only from this version on, and will have "noexec"
6098 set. This (minor) change of behavior might cause some (exceptional)
6099 legacy software to break, when PrivateDevices=yes is set for its
6100 service. Please leave PrivateDevices= off if you run into problems
6103 * systemd-bootchart has been split out to a separate repository:
6104 https://github.com/systemd/systemd-bootchart
6106 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
6107 merged into the kernel in its current form.
6109 * The compatibility libraries libsystemd-daemon.so,
6110 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
6111 which have been deprecated since systemd-209 have been removed along
6112 with the corresponding pkg-config files. All symbols provided by
6113 those libraries are provided by libsystemd.so.
6115 * The Capabilities= unit file setting has been removed (it is ignored
6116 for backwards compatibility). AmbientCapabilities= and
6117 CapabilityBoundingSet= should be used instead.
6119 * A new special target has been added, initrd-root-device.target,
6120 which creates a synchronization point for dependencies of the root
6121 device in early userspace. Initramfs builders must ensure that this
6122 target is now included in early userspace.
6124 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
6125 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
6126 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
6127 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
6128 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
6129 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
6130 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
6131 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
6132 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
6133 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
6134 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
6135 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
6136 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
6137 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
6138 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
6139 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
6140 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
6141 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
6142 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
6143 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
6144 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
6145 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
6146 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
6147 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
6148 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
6151 — Fairfax, 2016-05-21
6155 * The systemd-resolved DNS resolver service has gained a substantial
6156 set of new features, most prominently it may now act as a DNSSEC
6157 validating stub resolver. DNSSEC mode is currently turned off by
6158 default, but is expected to be turned on by default in one of the
6159 next releases. For now, we invite everybody to test the DNSSEC logic
6160 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
6161 service also gained a full set of D-Bus interfaces, including calls
6162 to configure DNS and DNSSEC settings per link (for use by external
6163 network management software). systemd-resolved and systemd-networkd
6164 now distinguish between "search" and "routing" domains. The former
6165 are used to qualify single-label names, the latter are used purely
6166 for routing lookups within certain domains to specific links.
6167 resolved now also synthesizes RRs for all entries from /etc/hosts.
6169 * The systemd-resolve tool (which is a client utility for
6170 systemd-resolved) has been improved considerably and is now fully
6171 supported and documented. Hence it has moved from /usr/lib/systemd to
6174 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
6177 * The coredump collection logic has been reworked: when a coredump is
6178 collected it is now written to disk, compressed and processed
6179 (including stacktrace extraction) from a new instantiated service
6180 systemd-coredump@.service, instead of directly from the
6181 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
6182 processing large coredumps can take up a substantial amount of
6183 resources and time, and this previously happened entirely outside of
6184 systemd's service supervision. With the new logic the core_pattern
6185 hook only does minimal metadata collection before passing off control
6186 to the new instantiated service, which is configured with a time
6187 limit, a nice level and other settings to minimize negative impact on
6188 the rest of the system. Also note that the new logic will honour the
6189 RLIMIT_CORE setting of the crashed process, which now allows users
6190 and processes to turn off coredumping for their processes by setting
6193 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
6194 and all forked processes by default. Previously, PID 1 would leave
6195 the setting at "0" for all processes, as set by the kernel. Note that
6196 the resource limit traditionally has no effect on the generated
6197 coredumps on the system if the /proc/sys/kernel/core_pattern hook
6198 logic is used. Since the limit is now honoured (see above) its
6199 default has been changed so that the coredumping logic is enabled by
6200 default for all processes, while allowing specific opt-out.
6202 * When the stacktrace is extracted from processes of system users, this
6203 is now done as "systemd-coredump" user, in order to sandbox this
6204 potentially security sensitive parsing operation. (Note that when
6205 processing coredumps of normal users this is done under the user ID
6206 of process that crashed, as before.) Packagers should take notice
6207 that it is now necessary to create the "systemd-coredump" system user
6208 and group at package installation time.
6210 * The systemd-activate socket activation testing tool gained support
6211 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
6212 and --seqpacket switches. It also has been extended to support both
6213 new-style and inetd-style file descriptor passing. Use the new
6214 --inetd switch to request inetd-style file descriptor passing.
6216 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
6217 variable, which takes a boolean value. If set to false, ANSI color
6218 output is disabled in the tools even when run on a terminal that
6221 * The VXLAN support in networkd now supports two new settings
6222 DestinationPort= and PortRange=.
6224 * A new systemd.machine_id= kernel command line switch has been added,
6225 that may be used to set the machine ID in /etc/machine-id if it is
6226 not initialized yet. This command line option has no effect if the
6227 file is already initialized.
6229 * systemd-nspawn gained a new --as-pid2 switch that invokes any
6230 specified command line as PID 2 rather than PID 1 in the
6231 container. In this mode PID 1 is a minimal stub init process that
6232 implements the special POSIX and Linux semantics of PID 1 regarding
6233 signal and child process management. Note that this stub init process
6234 is implemented in nspawn itself and requires no support from the
6235 container image. This new logic is useful to support running
6236 arbitrary commands in the container, as normal processes are
6237 generally not prepared to run as PID 1.
6239 * systemd-nspawn gained a new --chdir= switch for setting the current
6240 working directory for the process started in the container.
6242 * "journalctl /dev/sda" will now output all kernel log messages for
6243 specified device from the current boot, in addition to all devices
6244 that are parents of it. This should make log output about devices
6245 pretty useful, as long as kernel drivers attach enough metadata to
6246 the log messages. (The usual SATA drivers do.)
6248 * The sd-journal API gained two new calls
6249 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
6250 that report whether log data from /run or /var has been found.
6252 * journalctl gained a new switch "--fields" that prints all journal
6253 record field names currently in use in the journal. This is backed
6254 by two new sd-journal API calls sd_journal_enumerate_fields() and
6255 sd_journal_restart_fields().
6257 * Most configurable timeouts in systemd now expect an argument of
6258 "infinity" to turn them off, instead of "0" as before. The semantics
6259 from now on is that a timeout of "0" means "now", and "infinity"
6260 means "never". To maintain backwards compatibility, "0" continues to
6261 turn off previously existing timeout settings.
6263 * "systemctl reload-or-try-restart" has been renamed to "systemctl
6264 try-reload-or-restart" to clarify what it actually does: the "try"
6265 logic applies to both reloading and restarting, not just restarting.
6266 The old name continues to be accepted for compatibility.
6268 * On boot-up, when PID 1 detects that the system clock is behind the
6269 release date of the systemd version in use, the clock is now set
6270 to the latter. Previously, this was already done in timesyncd, in order
6271 to avoid running with clocks set to the various clock epochs such as
6272 1902, 1938 or 1970. With this change the logic is now done in PID 1
6273 in addition to timesyncd during early boot-up, so that it is enforced
6274 before the first process is spawned by systemd. Note that the logic
6275 in timesyncd remains, as it is more comprehensive and ensures
6276 clock monotonicity by maintaining a persistent timestamp file in
6277 /var. Since /var is generally not available in earliest boot or the
6278 initrd, this part of the logic remains in timesyncd, and is not done
6281 * Support for tweaking details in net_cls.class_id through the
6282 NetClass= configuration directive has been removed, as the kernel
6283 people have decided to deprecate that controller in cgroup v2.
6284 Userspace tools such as nftables are moving over to setting rules
6285 that are specific to the full cgroup path of a task, which obsoletes
6286 these controllers anyway. The NetClass= directive is kept around for
6287 legacy compatibility reasons. For a more in-depth description of the
6288 kernel change, please refer to the respective upstream commit:
6290 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
6292 * A new service setting RuntimeMaxSec= has been added that may be used
6293 to specify a maximum runtime for a service. If the timeout is hit, the
6294 service is terminated and put into a failure state.
6296 * A new service setting AmbientCapabilities= has been added. It allows
6297 configuration of additional Linux process capabilities that are
6298 passed to the activated processes. This is only available on very
6301 * The process resource limit settings in service units may now be used
6302 to configure hard and soft limits individually.
6304 * The various libsystemd APIs such as sd-bus or sd-event now publicly
6305 expose support for gcc's __attribute__((cleanup())) C extension.
6306 Specifically, for many object destructor functions alternative
6307 versions have been added that have names suffixed with "p" and take a
6308 pointer to a pointer to the object to destroy, instead of just a
6309 pointer to the object itself. This is useful because these destructor
6310 functions may be used directly as parameters to the cleanup
6311 construct. Internally, systemd has been a heavy user of this GCC
6312 extension for a long time, and with this change similar support is
6313 now available to consumers of the library outside of systemd. Note
6314 that by using this extension in your sources compatibility with old
6315 and strictly ANSI compatible C compilers is lost. However, all gcc or
6316 LLVM versions of recent years support this extension.
6318 * Timer units gained support for a new setting RandomizedDelaySec= that
6319 allows configuring some additional randomized delay to the configured
6320 time. This is useful to spread out timer events to avoid load peaks in
6321 clusters or larger setups.
6323 * Calendar time specifications now support sub-second accuracy.
6325 * Socket units now support listening on SCTP and UDP-lite protocol
6328 * The sd-event API now comes with a full set of man pages.
6330 * Older versions of systemd contained experimental support for
6331 compressing journal files and coredumps with the LZ4 compressor that
6332 was not compatible with the lz4 binary (due to API limitations of the
6333 lz4 library). This support has been removed; only support for files
6334 compatible with the lz4 binary remains. This LZ4 logic is now
6335 officially supported and no longer considered experimental.
6337 * The dkr image import logic has been removed again from importd. dkr's
6338 micro-services focus doesn't fit into the machine image focus of
6339 importd, and quickly got out of date with the upstream dkr API.
6341 * Creation of the /run/lock/lockdev/ directory was dropped from
6342 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
6343 been available for many years. If you still need this, you need to
6344 create your own tmpfiles.d config file with:
6346 d /run/lock/lockdev 0775 root lock -
6348 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
6349 and RebootArgument= have been moved from the [Service] section of
6350 unit files to [Unit], and they are now supported on all unit types,
6351 not just service units. Of course, systemd will continue to
6352 understand these settings also at the old location, in order to
6353 maintain compatibility.
6355 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
6356 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
6357 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
6358 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
6359 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
6360 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
6361 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
6362 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
6363 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
6364 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
6365 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
6366 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
6367 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
6368 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
6369 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
6370 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
6371 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
6372 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
6373 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
6375 — Berlin, 2016-02-11
6379 * A number of properties previously only settable in unit
6380 files are now also available as properties to set when
6381 creating transient units programmatically via the bus, as it
6382 is exposed with systemd-run's --property=
6383 setting. Specifically, these are: SyslogIdentifier=,
6384 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
6385 EnvironmentFile=, ReadWriteDirectories=,
6386 ReadOnlyDirectories=, InaccessibleDirectories=,
6387 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
6389 * When creating transient services via the bus API it is now
6390 possible to pass in a set of file descriptors to use as
6391 STDIN/STDOUT/STDERR for the invoked process.
6393 * Slice units may now be created transiently via the bus APIs,
6394 similar to the way service and scope units may already be
6395 created transiently.
6397 * Wherever systemd expects a calendar timestamp specification
6398 (like in journalctl's --since= and --until= switches) UTC
6399 timestamps are now supported. Timestamps suffixed with "UTC"
6400 are now considered to be in Universal Time Coordinated
6401 instead of the local timezone. Also, timestamps may now
6402 optionally be specified with sub-second accuracy. Both of
6403 these additions also apply to recurring calendar event
6404 specification, such as OnCalendar= in timer units.
6406 * journalctl gained a new "--sync" switch that asks the
6407 journal daemon to write all so far unwritten log messages to
6408 disk and sync the files, before returning.
6410 * systemd-tmpfiles learned two new line types "q" and "Q" that
6411 operate like "v", but also set up a basic btrfs quota
6412 hierarchy when used on a btrfs file system with quota
6415 * tmpfiles' "v", "q" and "Q" will now create a plain directory
6416 instead of a subvolume (even on a btrfs file system) if the
6417 root directory is a plain directory, and not a
6418 subvolume. This should simplify things with certain chroot()
6419 environments which are not aware of the concept of btrfs
6422 * systemd-detect-virt gained a new --chroot switch to detect
6423 whether execution takes place in a chroot() environment.
6425 * CPUAffinity= now takes CPU index ranges in addition to
6428 * The various memory-related resource limit settings (such as
6429 LimitAS=) now understand the usual K, M, G, … suffixes to
6430 the base of 1024 (IEC). Similar, the time-related resource
6431 limit settings understand the usual min, h, day, … suffixes
6434 * There's a new system.conf setting DefaultTasksMax= to
6435 control the default TasksMax= setting for services and
6436 scopes running on the system. (TasksMax= is the primary
6437 setting that exposes the "pids" cgroup controller on systemd
6438 and was introduced in the previous systemd release.) The
6439 setting now defaults to 512, which means services that are
6440 not explicitly configured otherwise will only be able to
6441 create 512 processes or threads at maximum, from this
6442 version on. Note that this means that thread- or
6443 process-heavy services might need to be reconfigured to set
6444 TasksMax= to a higher value. It is sufficient to set
6445 TasksMax= in these specific unit files to a higher value, or
6446 even "infinity". Similar, there's now a logind.conf setting
6447 UserTasksMax= that defaults to 4096 and limits the total
6448 number of processes or tasks each user may own
6449 concurrently. nspawn containers also have the TasksMax=
6450 value set by default now, to 8192. Note that all of this
6451 only has an effect if the "pids" cgroup controller is
6452 enabled in the kernel. The general benefit of these changes
6453 should be a more robust and safer system, that provides a
6454 certain amount of per-service fork() bomb protection.
6456 * systemd-nspawn gained the new --network-veth-extra= switch
6457 to define additional and arbitrarily-named virtual Ethernet
6458 links between the host and the container.
6460 * A new service execution setting PassEnvironment= has been
6461 added that allows importing select environment variables
6462 from PID1's environment block into the environment block of
6465 * Timer units gained support for a new RemainAfterElapse=
6466 setting which takes a boolean argument. It defaults to on,
6467 exposing behaviour unchanged to previous releases. If set to
6468 off, timer units are unloaded after they elapsed if they
6469 cannot elapse again. This is particularly useful for
6470 transient timer units, which shall not stay around longer
6471 than until they first elapse.
6473 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
6474 default now (the kernel default is 16). This is beneficial
6475 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
6476 allows substantially larger numbers of queued
6477 datagrams. This should increase the capability of systemd to
6478 parallelize boot-up, as logging and sd_notify() are unlikely
6479 to stall execution anymore. If you need to change the value
6480 from the new defaults, use the usual sysctl.d/ snippets.
6482 * The compression framing format used by the journal or
6483 coredump processing has changed to be in line with what the
6484 official LZ4 tools generate. LZ4 compression support in
6485 systemd was considered unsupported previously, as the format
6486 was not compatible with the normal tools. With this release
6487 this has changed now, and it is hence safe for downstream
6488 distributions to turn it on. While not compressing as well
6489 as the XZ, LZ4 is substantially faster, which makes
6490 it a good default choice for the compression logic in the
6491 journal and in coredump handling.
6493 * Any reference to /etc/mtab has been dropped from
6494 systemd. The file has been obsolete since a while, but
6495 systemd refused to work on systems where it was incorrectly
6496 set up (it should be a symlink or non-existent). Please make
6497 sure to update to util-linux 2.27.1 or newer in conjunction
6498 with this systemd release, which also drops any reference to
6499 /etc/mtab. If you maintain a distribution make sure that no
6500 software you package still references it, as this is a
6501 likely source of bugs. There's also a glibc bug pending,
6502 asking for removal of any reference to this obsolete file:
6504 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
6506 Note that only util-linux versions built with
6507 --enable-libmount-force-mountinfo are supported.
6509 * Support for the ".snapshot" unit type has been removed. This
6510 feature turned out to be little useful and little used, and
6511 has now been removed from the core and from systemctl.
6513 * The dependency types RequiresOverridable= and
6514 RequisiteOverridable= have been removed from systemd. They
6515 have been used only very sparingly to our knowledge and
6516 other options that provide a similar effect (such as
6517 systemctl --mode=ignore-dependencies) are much more useful
6518 and commonly used. Moreover, they were only half-way
6519 implemented as the option to control behaviour regarding
6520 these dependencies was never added to systemctl. By removing
6521 these dependency types the execution engine becomes a bit
6522 simpler. Unit files that use these dependencies should be
6523 changed to use the non-Overridable dependency types
6524 instead. In fact, when parsing unit files with these
6525 options, that's what systemd will automatically convert them
6526 too, but it will also warn, asking users to fix the unit
6527 files accordingly. Removal of these dependency types should
6528 only affect a negligible number of unit files in the wild.
6530 * Behaviour of networkd's IPForward= option changed
6531 (again). It will no longer maintain a per-interface setting,
6532 but propagate one way from interfaces where this is enabled
6533 to the global kernel setting. The global setting will be
6534 enabled when requested by a network that is set up, but
6535 never be disabled again. This change was made to make sure
6536 IPv4 and IPv6 behaviour regarding packet forwarding is
6537 similar (as the Linux IPv6 stack does not support
6538 per-interface control of this setting) and to minimize
6541 * In unit files the behaviour of %u, %U, %h, %s has
6542 changed. These specifiers will now unconditionally resolve
6543 to the various user database fields of the user that the
6544 systemd instance is running as, instead of the user
6545 configured in the specific unit via User=. Note that this
6546 effectively doesn't change much, as resolving of these
6547 specifiers was already turned off in the --system instance
6548 of systemd, as we cannot do NSS lookups from PID 1. In the
6549 --user instance of systemd these specifiers where correctly
6550 resolved, but hardly made any sense, since the user instance
6551 lacks privileges to do user switches anyway, and User= is
6552 hence useless. Moreover, even in the --user instance of
6553 systemd behaviour was awkward as it would only take settings
6554 from User= assignment placed before the specifier into
6555 account. In order to unify and simplify the logic around
6556 this the specifiers will now always resolve to the
6557 credentials of the user invoking the manager (which in case
6558 of PID 1 is the root user).
6560 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
6561 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
6562 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
6563 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
6564 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
6565 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
6566 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
6567 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
6568 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
6569 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
6570 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
6571 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
6572 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
6573 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
6576 — Berlin, 2015-11-18
6580 * systemd now depends on util-linux v2.27. More specifically,
6581 the newly added mount monitor feature in libmount now
6582 replaces systemd's former own implementation.
6584 * libmount mandates /etc/mtab not to be regular file, and
6585 systemd now enforces this condition at early boot.
6586 /etc/mtab has been deprecated and warned about for a very
6587 long time, so systems running systemd should already have
6588 stopped having this file around as anything else than a
6589 symlink to /proc/self/mounts.
6591 * Support for the "pids" cgroup controller has been added. It
6592 allows accounting the number of tasks in a cgroup and
6593 enforcing limits on it. This adds two new setting
6594 TasksAccounting= and TasksMax= to each unit, as well as a
6595 global option DefaultTasksAccounting=.
6597 * Support for the "net_cls" cgroup controller has been added.
6598 It allows assigning a net class ID to each task in the
6599 cgroup, which can then be used in firewall rules and traffic
6600 shaping configurations. Note that the kernel netfilter net
6601 class code does not currently work reliably for ingress
6602 packets on unestablished sockets.
6604 This adds a new config directive called NetClass= to CGroup
6605 enabled units. Allowed values are positive numbers for fixed
6606 assignments and "auto" for picking a free value
6609 * 'systemctl is-system-running' now returns 'offline' if the
6610 system is not booted with systemd. This command can now be
6611 used as a substitute for 'systemd-notify --booted'.
6613 * Watchdog timeouts have been increased to 3 minutes for all
6614 in-tree service files. Apparently, disk IO issues are more
6615 frequent than we hoped, and user reported >1 minute waiting
6618 * 'machine-id-commit' functionality has been merged into
6619 'machine-id-setup --commit'. The separate binary has been
6622 * The WorkingDirectory= directive in unit files may now be set
6623 to the special value '~'. In this case, the working
6624 directory is set to the home directory of the user
6625 configured in User=.
6627 * "machinectl shell" will now open the shell in the home
6628 directory of the selected user by default.
6630 * The CrashChVT= configuration file setting is renamed to
6631 CrashChangeVT=, following our usual logic of not
6632 abbreviating unnecessarily. The old directive is still
6633 supported for compat reasons. Also, this directive now takes
6634 an integer value between 1 and 63, or a boolean value. The
6635 formerly supported '-1' value for disabling stays around for
6638 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
6639 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
6640 RootDirectory= properties can now be set for transient
6643 * The systemd-analyze tool gained a new "set-log-target" verb
6644 to change the logging target the system manager logs to
6645 dynamically during runtime. This is similar to how
6646 "systemd-analyze set-log-level" already changes the log
6649 * In nspawn /sys is now mounted as tmpfs, with only a selected
6650 set of subdirectories mounted in from the real sysfs. This
6651 enhances security slightly, and is useful for ensuring user
6652 namespaces work correctly.
6654 * Support for USB FunctionFS activation has been added. This
6655 allows implementation of USB gadget services that are
6656 activated as soon as they are requested, so that they don't
6657 have to run continuously, similar to classic socket
6660 * The "systemctl exit" command now optionally takes an
6661 additional parameter that sets the exit code to return from
6662 the systemd manager when exiting. This is only relevant when
6663 running the systemd user instance, or when running the
6664 system instance in a container.
6666 * sd-bus gained the new API calls sd_bus_path_encode_many()
6667 and sd_bus_path_decode_many() that allow easy encoding and
6668 decoding of multiple identifier strings inside a D-Bus
6669 object path. Another new call sd_bus_default_flush_close()
6670 has been added to flush and close per-thread default
6673 * systemd-cgtop gained support for a -M/--machine= switch to
6674 show the control groups within a certain container only.
6676 * "systemctl kill" gained support for an optional --fail
6677 switch. If specified the requested operation will fail of no
6678 processes have been killed, because the unit had no
6679 processes attached, or similar.
6681 * A new systemd.crash_reboot=1 kernel command line option has
6682 been added that triggers a reboot after crashing. This can
6683 also be set through CrashReboot= in systemd.conf.
6685 * The RuntimeDirectory= setting now understands unit
6686 specifiers like %i or %f.
6688 * A new (still internal) library API sd-ipv4acd has been added,
6689 that implements address conflict detection for IPv4. It's
6690 based on code from sd-ipv4ll, and will be useful for
6691 detecting DHCP address conflicts.
6693 * File descriptors passed during socket activation may now be
6694 named. A new API sd_listen_fds_with_names() is added to
6695 access the names. The default names may be overridden,
6696 either in the .socket file using the FileDescriptorName=
6697 parameter, or by passing FDNAME= when storing the file
6698 descriptors using sd_notify().
6700 * systemd-networkd gained support for:
6702 - Setting the IPv6 Router Advertisement settings via
6703 IPv6AcceptRouterAdvertisements= in .network files.
6705 - Configuring the HelloTimeSec=, MaxAgeSec= and
6706 ForwardDelaySec= bridge parameters in .netdev files.
6708 - Configuring PreferredSource= for static routes in
6711 * The "ask-password" framework used to query for LUKS harddisk
6712 passwords or SSL passwords during boot gained support for
6713 caching passwords in the kernel keyring, if it is
6714 available. This makes sure that the user only has to type in
6715 a passphrase once if there are multiple objects to unlock
6716 with the same one. Previously, such password caching was
6717 available only when Plymouth was used; this moves the
6718 caching logic into the systemd codebase itself. The
6719 "systemd-ask-password" utility gained a new --keyname=
6720 switch to control which kernel keyring key to use for
6721 caching a password in. This functionality is also useful for
6722 enabling display managers such as gdm to automatically
6723 unlock the user's GNOME keyring if its passphrase, the
6724 user's password and the harddisk password are the same, if
6725 gdm-autologin is used.
6727 * When downloading tar or raw images using "machinectl
6728 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
6729 file is now also downloaded, if it is available and stored
6730 next to the image file.
6732 * Units of type ".socket" gained a new boolean setting
6733 Writable= which is only useful in conjunction with
6734 ListenSpecial=. If true, enables opening the specified
6735 special file in O_RDWR mode rather than O_RDONLY mode.
6737 * systemd-rfkill has been reworked to become a singleton
6738 service that is activated through /dev/rfkill on each rfkill
6739 state change and saves the settings to disk. This way,
6740 systemd-rfkill is now compatible with devices that exist
6741 only intermittendly, and even restores state if the previous
6742 system shutdown was abrupt rather than clean.
6744 * The journal daemon gained support for vacuuming old journal
6745 files controlled by the number of files that shall remain,
6746 in addition to the already existing control by size and by
6747 date. This is useful as journal interleaving performance
6748 degrades with too many separate journal files, and allows
6749 putting an effective limit on them. The new setting defaults
6750 to 100, but this may be changed by setting SystemMaxFiles=
6751 and RuntimeMaxFiles= in journald.conf. Also, the
6752 "journalctl" tool gained the new --vacuum-files= switch to
6753 manually vacuum journal files to leave only the specified
6754 number of files in place.
6756 * udev will now create /dev/disk/by-path links for ATA devices
6757 on kernels where that is supported.
6759 * Galician, Serbian, Turkish and Korean translations were added.
6761 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
6762 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
6763 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
6764 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
6765 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
6766 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
6767 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
6768 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
6769 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
6770 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
6771 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
6772 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
6773 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
6774 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
6775 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
6776 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
6777 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
6778 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
6780 — Berlin, 2015-10-07
6784 * The DHCP implementation of systemd-networkd gained a set of
6787 - The DHCP server now supports emitting DNS and NTP
6788 information. It may be enabled and configured via
6789 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
6790 and NTP information is enabled, but no servers are
6791 configured, the corresponding uplink information (if there
6792 is any) is propagated.
6794 - Server and client now support transmission and reception
6795 of timezone information. It can be configured via the
6796 newly introduced network options UseTimezone=,
6797 EmitTimezone=, and Timezone=. Transmission of timezone
6798 information is enabled between host and containers by
6799 default now: the container will change its local timezone
6800 to what the host has set.
6802 - Lease timeouts can now be configured via
6803 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
6805 - The DHCP server improved on the stability of
6806 leases. Clients are more likely to get the same lease
6807 information back, even if the server loses state.
6809 - The DHCP server supports two new configuration options to
6810 control the lease address pool metrics, PoolOffset= and
6813 * The encapsulation limit of tunnels in systemd-networkd may
6814 now be configured via 'EncapsulationLimit='. It allows
6815 modifying the maximum additional levels of encapsulation
6816 that are permitted to be prepended to a packet.
6818 * systemd now supports the concept of user buses replacing
6819 session buses, if used with dbus-1.10 (and enabled via dbus
6820 --enable-user-session). It previously only supported this on
6821 kdbus-enabled systems, and this release expands this to
6822 'dbus-daemon' systems.
6824 * systemd-networkd now supports predictable interface names
6827 * systemd now optionally supports the new Linux kernel
6828 "unified" control group hierarchy. If enabled via the kernel
6829 command-line option 'systemd.unified_cgroup_hierarchy=1',
6830 systemd will try to mount the unified cgroup hierarchy
6831 directly on /sys/fs/cgroup. If not enabled, or not
6832 available, systemd will fall back to the legacy cgroup
6833 hierarchy setup, as before. Host system and containers can
6834 mix and match legacy and unified hierarchies as they
6835 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
6836 environment variable to individually select the hierarchy to
6837 use for executed containers. By default, nspawn will use the
6838 unified hierarchy for the containers if the host uses the
6839 unified hierarchy, and the legacy hierarchy otherwise.
6840 Please note that at this point the unified hierarchy is an
6841 experimental kernel feature and is likely to change in one
6842 of the next kernel releases. Therefore, it should not be
6843 enabled by default in downstream distributions yet. The
6844 minimum required kernel version for the unified hierarchy to
6845 work is 4.2. Note that when the unified hierarchy is used
6846 for the first time delegated access to controllers is
6847 safe. Because of this systemd-nspawn containers will get
6848 access to controllers now, as will systemd user
6849 sessions. This means containers and user sessions may now
6850 manage their own resources, partitioning up what the system
6853 * A new special scope unit "init.scope" has been introduced
6854 that encapsulates PID 1 of the system. It may be used to
6855 determine resource usage and enforce resource limits on PID
6856 1 itself. PID 1 hence moved out of the root of the control
6859 * The cgtop tool gained support for filtering out kernel
6860 threads when counting tasks in a control group. Also, the
6861 count of processes is now recursively summed up by
6862 default. Two options -k and --recursive= have been added to
6863 revert to old behaviour. The tool has also been updated to
6864 work correctly in containers now.
6866 * systemd-nspawn's --bind= and --bind-ro= options have been
6867 extended to allow creation of non-recursive bind mounts.
6869 * libsystemd gained two new calls sd_pid_get_cgroup() and
6870 sd_peer_get_cgroup() which return the control group path of
6871 a process or peer of a connected AF_UNIX socket. This
6872 function call is particularly useful when implementing
6873 delegated subtrees support in the control group hierarchy.
6875 * The "sd-event" event loop API of libsystemd now supports
6876 correct dequeuing of real-time signals, without losing
6879 * When systemd requests a polkit decision when managing units it
6880 will now add additional fields to the request, including unit
6881 name and desired operation. This enables more powerful polkit
6882 policies, that make decisions depending on these parameters.
6884 * nspawn learnt support for .nspawn settings files, that may
6885 accompany the image files or directories of containers, and
6886 may contain additional settings for the container. This is
6887 an alternative to configuring container parameters via the
6888 nspawn command line.
6890 Contributions from: Cristian Rodríguez, Daniel Mack, David
6891 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
6892 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
6893 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
6894 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
6895 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
6896 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
6897 Andersen, Tom Gundersen, Torstein Husebø
6899 — Berlin, 2015-09-08
6903 * machinectl gained a new verb 'shell' which opens a fresh
6904 shell on the target container or the host. It is similar to
6905 the existing 'login' command of machinectl, but spawns the
6906 shell directly without prompting for username or
6907 password. The pseudo machine '.host' now refers to the local
6908 host and is used by default. Hence, 'machinectl shell' can
6909 be used as replacement for 'su -' which spawns a session as
6910 a fresh systemd unit in a way that is fully isolated from
6911 the originating session.
6913 * systemd-networkd learned to cope with private-zone DHCP
6914 options and allows other programs to query the values.
6916 * SELinux access control when enabling/disabling units is no
6917 longer enforced with this release. The previous implementation
6918 was incorrect, and a new corrected implementation is not yet
6919 available. As unit file operations are still protected via
6920 polkit and D-Bus policy this is not a security problem. Yet,
6921 distributions which care about optimal SELinux support should
6922 probably not stabilize on this release.
6924 * sd-bus gained support for matches of type "arg0has=", that
6925 test for membership of strings in string arrays sent in bus
6928 * systemd-resolved now dumps the contents of its DNS and LLMNR
6929 caches to the logs on reception of the SIGUSR1 signal. This
6930 is useful to debug DNS behaviour.
6932 * The coredumpctl tool gained a new --directory= option to
6933 operate on journal files in a specific directory.
6935 * "systemctl reboot" and related commands gained a new
6936 "--message=" option which may be used to set a free-text
6937 wall message when shutting down or rebooting the
6938 system. This message is also logged, which is useful for
6939 figuring out the reason for a reboot or shutdown a
6942 * The "systemd-resolve-host" tool's -i switch now takes
6943 network interface numbers as alternative to interface names.
6945 * A new unit file setting for services has been introduced:
6946 UtmpMode= allows configuration of how precisely systemd
6947 handles utmp and wtmp entries for the service if this is
6948 enabled. This allows writing services that appear similar to
6949 user sessions in the output of the "w", "who", "last" and
6952 * systemd-resolved will now locally synthesize DNS resource
6953 records for the "localhost" and "gateway" domains as well as
6954 the local hostname. This should ensure that clients querying
6955 RRs via resolved will get similar results as those going via
6956 NSS, if nss-myhostname is enabled.
6958 Contributions from: Alastair Hughes, Alex Crawford, Daniel
6959 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
6960 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
6961 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
6962 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
6963 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
6964 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
6965 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
6966 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
6967 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
6968 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
6969 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
6971 — Berlin, 2015-08-27
6975 * The systemd-efi-boot-generator functionality was merged into
6976 systemd-gpt-auto-generator.
6978 * systemd-networkd now supports Group Policy for vxlan
6979 devices. It can be enabled via the new boolean configuration
6980 option called 'GroupPolicyExtension='.
6982 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
6983 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
6984 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
6986 — Berlin, 2015-07-31
6990 * The python-systemd code has been removed from the systemd repository.
6991 A new repository has been created which accommodates the code from
6992 now on, and we kindly ask distributions to create a separate package
6993 for this: https://github.com/systemd/python-systemd
6995 * The systemd daemon will now reload its main configuration
6996 (/etc/systemd/system.conf) on daemon-reload.
6998 * sd-dhcp now exposes vendor specific extensions via
6999 sd_dhcp_lease_get_vendor_specific().
7001 * systemd-networkd gained a number of new configuration options.
7003 - A new boolean configuration option for TAP devices called
7004 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
7005 device, thus allowing to send and receive GSO packets.
7007 - A new tunnel configuration option called 'CopyDSCP='.
7008 If enabled, the DSCP field of ip6 tunnels is copied into the
7009 decapsulated packet.
7011 - A set of boolean bridge configuration options were added.
7012 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
7013 and 'UnicastFlood=' are now parsed by networkd and applied to the
7014 respective bridge link device via the respective IFLA_BRPORT_*
7017 - A new string configuration option to override the hostname sent
7018 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
7019 is true, networkd will use the configured hostname instead of the
7020 system hostname when sending DHCP requests.
7022 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
7023 networkd will configure the IPv6 flow-label of the tunnel device
7024 according to RFC2460.
7026 - The 'macvtap' virtual network devices are now supported, similar to
7027 the already supported 'macvlan' devices.
7029 * systemd-resolved now implements RFC5452 to improve resilience against
7030 cache poisoning. Additionally, source port randomization is enabled
7031 by default to further protect against DNS spoofing attacks.
7033 * nss-mymachines now supports translating UIDs and GIDs of running
7034 containers with user-namespaces enabled. If a container 'foo'
7035 translates a host uid 'UID' to the container uid 'TUID', then
7036 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
7037 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
7038 mapped as 'vg-foo-TGID'.
7040 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
7041 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
7042 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
7043 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
7044 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
7045 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
7046 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
7047 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
7048 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
7049 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
7051 — Berlin, 2015-07-29
7055 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
7056 There are no known issues with current sysfs, and udev does not need
7057 or should be used to work around such bugs.
7059 * udev does no longer enable USB HID power management. Several reports
7060 indicate, that some devices cannot handle that setting.
7062 * The udev accelerometer helper was removed. The functionality
7063 is now fully included in iio-sensor-proxy. But this means,
7064 older iio-sensor-proxy versions will no longer provide
7065 accelerometer/orientation data with this systemd version.
7066 Please upgrade iio-sensor-proxy to version 1.0.
7068 * networkd gained a new configuration option IPv6PrivacyExtensions=
7069 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
7070 for Stateless Address") on selected networks.
7072 * For the sake of fewer build-time dependencies and less code in the
7073 main repository, the python bindings are about to be removed in the
7074 next release. A new repository has been created which accommodates
7075 the code from now on, and we kindly ask distributions to create a
7076 separate package for this. The removal will take place in v223.
7078 https://github.com/systemd/python-systemd
7080 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
7081 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
7082 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
7083 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
7084 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
7085 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
7086 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
7087 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
7088 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
7089 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
7091 — Berlin, 2015-07-07
7095 * The sd-bus.h and sd-event.h APIs have now been declared
7096 stable and have been added to the official interface of
7097 libsystemd.so. sd-bus implements an alternative D-Bus client
7098 library, that is relatively easy to use, very efficient and
7099 supports both classic D-Bus as well as kdbus as transport
7100 backend. sd-event is a generic event loop abstraction that
7101 is built around Linux epoll, but adds features such as event
7102 prioritization or efficient timer handling. Both APIs are good
7103 choices for C programs looking for a bus and/or event loop
7104 implementation that is minimal and does not have to be
7105 portable to other kernels.
7107 * kdbus support is no longer compile-time optional. It is now
7108 always built-in. However, it can still be disabled at
7109 runtime using the kdbus=0 kernel command line setting, and
7110 that setting may be changed to default to off, by specifying
7111 --disable-kdbus at build-time. Note though that the kernel
7112 command line setting has no effect if the kdbus.ko kernel
7113 module is not installed, in which case kdbus is (obviously)
7114 also disabled. We encourage all downstream distributions to
7115 begin testing kdbus by adding it to the kernel images in the
7116 development distributions, and leaving kdbus support in
7119 * The minimal required util-linux version has been bumped to
7122 * Support for chkconfig (--enable-chkconfig) was removed in
7123 favor of calling an abstraction tool
7124 /lib/systemd/systemd-sysv-install. This needs to be
7125 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
7126 in README for details.
7128 * If there's a systemd unit and a SysV init script for the
7129 same service name, and the user executes "systemctl enable"
7130 for it (or a related call), then this will now enable both
7131 (or execute the related operation on both), not just the
7134 * The libudev API documentation has been converted from gtkdoc
7137 * gudev has been removed from the systemd tree, it is now an
7140 * The systemd-cgtop tool learnt a new --raw switch to generate
7141 "raw" (machine parsable) output.
7143 * networkd's IPForwarding= .network file setting learnt the
7144 new setting "kernel", which ensures that networkd does not
7145 change the IP forwarding sysctl from the default kernel
7148 * The systemd-logind bus API now exposes a new boolean
7149 property "Docked" that reports whether logind considers the
7150 system "docked", i.e. connected to a docking station or not.
7152 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
7153 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
7154 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
7155 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
7156 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
7157 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
7158 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
7159 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
7160 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
7161 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
7162 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
7163 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
7164 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
7165 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
7166 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
7167 Fink, Zbigniew Jędrzejewski-Szmek
7169 — Berlin, 2015-06-19
7173 * The gudev library has been extracted into a separate repository
7174 available at: https://git.gnome.org/browse/libgudev/
7175 It is now managed as part of the Gnome project. Distributions
7176 are recommended to pass --disable-gudev to systemd and use
7177 gudev from the Gnome project instead. gudev is still included
7178 in systemd, for now. It will be removed soon, though. Please
7179 also see the announcement-thread on systemd-devel:
7180 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
7182 * systemd now exposes a CPUUsageNSec= property for each
7183 service unit on the bus, that contains the overall consumed
7184 CPU time of a service (the sum of what each process of the
7185 service consumed). This value is only available if
7186 CPUAccounting= is turned on for a service, and is then shown
7187 in the "systemctl status" output.
7189 * Support for configuring alternative mappings of the old SysV
7190 runlevels to systemd targets has been removed. They are now
7191 hardcoded in a way that runlevels 2, 3, 4 all map to
7192 multi-user.target and 5 to graphical.target (which
7193 previously was already the default behaviour).
7195 * The auto-mounter logic gained support for mount point
7196 expiry, using a new TimeoutIdleSec= setting in .automount
7197 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
7199 * The EFI System Partition (ESP) as mounted to /boot by
7200 systemd-efi-boot-generator will now be unmounted
7201 automatically after 2 minutes of not being used. This should
7202 minimize the risk of ESP corruptions.
7204 * New /etc/fstab options x-systemd.requires= and
7205 x-systemd.requires-mounts-for= are now supported to express
7206 additional dependencies for mounts. This is useful for
7207 journaling file systems that support external journal
7208 devices or overlay file systems that require underlying file
7209 systems to be mounted.
7211 * systemd does not support direct live-upgrades (via systemctl
7212 daemon-reexec) from versions older than v44 anymore. As no
7213 distribution we are aware of shipped such old versions in a
7214 stable release this should not be problematic.
7216 * When systemd forks off a new per-connection service instance
7217 it will now set the $REMOTE_ADDR environment variable to the
7218 remote IP address, and $REMOTE_PORT environment variable to
7219 the remote IP port. This behaviour is similar to the
7220 corresponding environment variables defined by CGI.
7222 * systemd-networkd gained support for uplink failure
7223 detection. The BindCarrier= option allows binding interface
7224 configuration dynamically to the link sense of other
7225 interfaces. This is useful to achieve behaviour like in
7228 * systemd-networkd gained support for configuring the DHCP
7229 client identifier to use when requesting leases.
7231 * systemd-networkd now has a per-network UseNTP= option to
7232 configure whether NTP server information acquired via DHCP
7233 is passed on to services like systemd-timesyncd.
7235 * systemd-networkd gained support for vti6 tunnels.
7237 * Note that systemd-networkd manages the sysctl variable
7238 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
7239 it is configured for since v219. The variable controls IP
7240 forwarding, and is a per-interface alternative to the global
7241 /proc/sys/net/ipv[46]/ip_forward. This setting is
7242 configurable in the IPForward= option, which defaults to
7243 "no". This means if networkd is used for an interface it is
7244 no longer sufficient to set the global sysctl option to turn
7245 on IP forwarding! Instead, the .network file option
7246 IPForward= needs to be turned on! Note that the
7247 implementation of this behaviour was broken in v219 and has
7250 * Many bonding and vxlan options are now configurable in
7253 * systemd-nspawn gained a new --property= setting to set unit
7254 properties for the container scope. This is useful for
7255 setting resource parameters (e.g. "CPUShares=500") on
7256 containers started from the command line.
7258 * systemd-nspawn gained a new --private-users= switch to make
7259 use of user namespacing available on recent Linux kernels.
7261 * systemd-nspawn may now be called as part of a shell pipeline
7262 in which case the pipes used for stdin and stdout are passed
7263 directly to the process invoked in the container, without
7264 indirection via a pseudo tty.
7266 * systemd-nspawn gained a new switch to control the UNIX
7267 signal to use when killing the init process of the container
7270 * systemd-nspawn gained a new --overlay= switch for mounting
7271 overlay file systems into the container using the new kernel
7274 * When a container image is imported via systemd-importd and
7275 the host file system is not btrfs, a loopback block device
7276 file is created in /var/lib/machines.raw with a btrfs file
7277 system inside. It is then mounted to /var/lib/machines to
7278 enable btrfs features for container management. The loopback
7279 file and btrfs file system is grown as needed when container
7280 images are imported via systemd-importd.
7282 * systemd-machined/systemd-importd gained support for btrfs
7283 quota, to enforce container disk space limits on disk. This
7284 is exposed in "machinectl set-limit".
7286 * systemd-importd now can import containers from local .tar,
7287 .raw and .qcow2 images, and export them to .tar and .raw. It
7288 can also import dkr v2 images now from the network (on top
7291 * systemd-importd gained support for verifying downloaded
7292 images with gpg2 (previously only gpg1 was supported).
7294 * systemd-machined, systemd-logind, systemd: most bus calls are
7295 now accessible to unprivileged processes via polkit. Also,
7296 systemd-logind will now allow users to kill their own sessions
7297 without further privileges or authorization.
7299 * systemd-shutdownd has been removed. This service was
7300 previously responsible for implementing scheduled shutdowns
7301 as exposed in /usr/bin/shutdown's time parameter. This
7302 functionality has now been moved into systemd-logind and is
7303 accessible via a bus interface.
7305 * "systemctl reboot" gained a new switch --firmware-setup that
7306 can be used to reboot into the EFI firmware setup, if that
7307 is available. systemd-logind now exposes an API on the bus
7308 to trigger such reboots, in case graphical desktop UIs want
7309 to cover this functionality.
7311 * "systemctl enable", "systemctl disable" and "systemctl mask"
7312 now support a new "--now" switch. If specified the units
7313 that are enabled will also be started, and the ones
7314 disabled/masked also stopped.
7316 * The Gummiboot EFI boot loader tool has been merged into
7317 systemd, and renamed to "systemd-boot". The bootctl tool has been
7318 updated to support systemd-boot.
7320 * An EFI kernel stub has been added that may be used to create
7321 kernel EFI binaries that contain not only the actual kernel,
7322 but also an initrd, boot splash, command line and OS release
7323 information. This combined binary can then be signed as a
7324 single image, so that the firmware can verify it all in one
7325 step. systemd-boot has special support for EFI binaries created
7326 like this and can extract OS release information from them
7327 and show them in the boot menu. This functionality is useful
7328 to implement cryptographically verified boot schemes.
7330 * Optional support has been added to systemd-fsck to pass
7331 fsck's progress report to an AF_UNIX socket in the file
7334 * udev will no longer create device symlinks for all block devices by
7335 default. A deny list for excluding special block devices from this
7336 logic has been turned into a allow list that requires picking block
7337 devices explicitly that require device symlinks.
7339 * A new (currently still internal) API sd-device.h has been
7340 added to libsystemd. This modernized API is supposed to
7341 replace libudev eventually. In fact, already much of libudev
7342 is now just a wrapper around sd-device.h.
7344 * A new hwdb database for storing metadata about pointing
7345 stick devices has been added.
7347 * systemd-tmpfiles gained support for setting file attributes
7348 similar to the "chattr" tool with new 'h' and 'H' lines.
7350 * systemd-journald will no longer unconditionally set the
7351 btrfs NOCOW flag on new journal files. This is instead done
7352 with tmpfiles snippet using the new 'h' line type. This
7353 allows easy disabling of this logic, by masking the
7354 journal-nocow.conf tmpfiles file.
7356 * systemd-journald will now translate audit message types to
7357 human readable identifiers when writing them to the
7358 journal. This should improve readability of audit messages.
7360 * The LUKS logic gained support for the offset= and skip=
7361 options in /etc/crypttab, as previously implemented by
7364 * /usr/lib/os-release gained a new optional field VARIANT= for
7365 distributions that support multiple variants (such as a
7366 desktop edition, a server edition, …)
7368 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
7369 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
7370 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
7371 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
7372 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
7373 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
7374 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
7375 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
7376 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
7377 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
7378 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
7379 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
7380 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
7381 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
7382 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
7383 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
7384 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
7385 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
7386 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
7387 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
7388 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
7389 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
7390 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
7391 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
7392 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
7393 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
7394 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
7396 — Berlin, 2015-05-22
7400 * Introduce a new API "sd-hwdb.h" for querying the hardware
7401 metadata database. With this minimal interface one can query
7402 and enumerate the udev hwdb, decoupled from the old libudev
7403 library. libudev's interface for this is now only a wrapper
7404 around sd-hwdb. A new tool systemd-hwdb has been added to
7405 interface with and update the database.
7407 * When any of systemd's tools copies files (for example due to
7408 tmpfiles' C lines) a btrfs reflink will attempted first,
7409 before bytewise copying is done.
7411 * systemd-nspawn gained a new --ephemeral switch. When
7412 specified a btrfs snapshot is taken of the container's root
7413 directory, and immediately removed when the container
7414 terminates again. Thus, a container can be started whose
7415 changes never alter the container's root directory, and are
7416 lost on container termination. This switch can also be used
7417 for starting a container off the root file system of the
7418 host without affecting the host OS. This switch is only
7419 available on btrfs file systems.
7421 * systemd-nspawn gained a new --template= switch. It takes the
7422 path to a container tree to use as template for the tree
7423 specified via --directory=, should that directory be
7424 missing. This allows instantiating containers dynamically,
7425 on first run. This switch is only available on btrfs file
7428 * When a .mount unit refers to a mount point on which multiple
7429 mounts are stacked, and the .mount unit is stopped all of
7430 the stacked mount points will now be unmounted until no
7431 mount point remains.
7433 * systemd now has an explicit notion of supported and
7434 unsupported unit types. Jobs enqueued for unsupported unit
7435 types will now fail with an "unsupported" error code. More
7436 specifically .swap, .automount and .device units are not
7437 supported in containers, .busname units are not supported on
7438 non-kdbus systems. .swap and .automount are also not
7439 supported if their respective kernel compile time options
7442 * machinectl gained support for two new "copy-from" and
7443 "copy-to" commands for copying files from a running
7444 container to the host or vice versa.
7446 * machinectl gained support for a new "bind" command to bind
7447 mount host directories into local containers. This is
7448 currently only supported for nspawn containers.
7450 * networkd gained support for configuring bridge forwarding
7451 database entries (fdb) from .network files.
7453 * A new tiny daemon "systemd-importd" has been added that can
7454 download container images in tar, raw, qcow2 or dkr formats,
7455 and make them available locally in /var/lib/machines, so
7456 that they can run as nspawn containers. The daemon can GPG
7457 verify the downloads (not supported for dkr, since it has no
7458 provisions for verifying downloads). It will transparently
7459 decompress bz2, xz, gzip compressed downloads if necessary,
7460 and restore sparse files on disk. The daemon uses privilege
7461 separation to ensure the actual download logic runs with
7462 fewer privileges than the daemon itself. machinectl has
7463 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
7464 make the functionality of importd available to the
7465 user. With this in place the Fedora and Ubuntu "Cloud"
7466 images can be downloaded and booted as containers unmodified
7467 (the Fedora images lack the appropriate GPG signature files
7468 currently, so they cannot be verified, but this will change
7469 soon, hopefully). Note that downloading images is currently
7470 only fully supported on btrfs.
7472 * machinectl is now able to list container images found in
7473 /var/lib/machines, along with some metadata about sizes of
7474 disk and similar. If the directory is located on btrfs and
7475 quota is enabled, this includes quota display. A new command
7476 "image-status" has been added that shows additional
7477 information about images.
7479 * machinectl is now able to clone container images
7480 efficiently, if the underlying file system (btrfs) supports
7481 it, with the new "machinectl clone" command. It also
7482 gained commands for renaming and removing images, as well as
7483 marking them read-only or read-write (supported also on
7484 legacy file systems).
7486 * networkd gained support for collecting LLDP network
7487 announcements, from hardware that supports this. This is
7488 shown in networkctl output.
7490 * systemd-run gained support for a new -t (--pty) switch for
7491 invoking a binary on a pty whose input and output is
7492 connected to the invoking terminal. This allows executing
7493 processes as system services while interactively
7494 communicating with them via the terminal. Most interestingly
7495 this is supported across container boundaries. Invoking
7496 "systemd-run -t /bin/bash" is an alternative to running a
7497 full login session, the difference being that the former
7498 will not register a session, nor go through the PAM session
7501 * tmpfiles gained support for a new "v" line type for creating
7502 btrfs subvolumes. If the underlying file system is a legacy
7503 file system, this automatically degrades to creating a
7504 normal directory. Among others /var/lib/machines is now
7505 created like this at boot, should it be missing.
7507 * The directory /var/lib/containers/ has been deprecated and
7508 been replaced by /var/lib/machines. The term "machines" has
7509 been used in the systemd context as generic term for both
7510 VMs and containers, and hence appears more appropriate for
7511 this, as the directory can also contain raw images bootable
7514 * systemd-nspawn when invoked with -M but without --directory=
7515 or --image= is now capable of searching for the container
7516 root directory, subvolume or disk image automatically, in
7517 /var/lib/machines. systemd-nspawn@.service has been updated
7518 to make use of this, thus allowing it to be used for raw
7521 * A new machines.target unit has been introduced that is
7522 supposed to group all containers/VMs invoked as services on
7523 the system. systemd-nspawn@.service has been updated to
7524 integrate with that.
7526 * machinectl gained a new "start" command, for invoking a
7527 container as a service. "machinectl start foo" is mostly
7528 equivalent to "systemctl start systemd-nspawn@foo.service",
7529 but handles escaping in a nicer way.
7531 * systemd-nspawn will now mount most of the cgroupfs tree
7532 read-only into each container, with the exception of the
7533 container's own subtree in the name=systemd hierarchy.
7535 * journald now sets the special FS_NOCOW file flag for its
7536 journal files. This should improve performance on btrfs, by
7537 avoiding heavy fragmentation when journald's write-pattern
7538 is used on COW file systems. It degrades btrfs' data
7539 integrity guarantees for the files to the same levels as for
7540 ext3/ext4 however. This should be OK though as journald does
7541 its own data integrity checks and all its objects are
7542 checksummed on disk. Also, journald should handle btrfs disk
7543 full events a lot more gracefully now, by processing SIGBUS
7544 errors, and not relying on fallocate() anymore.
7546 * When journald detects that journal files it is writing to
7547 have been deleted it will immediately start new journal
7550 * systemd now provides a way to store file descriptors
7551 per-service in PID 1. This is useful for daemons to ensure
7552 that fds they require are not lost during a daemon
7553 restart. The fds are passed to the daemon on the next
7554 invocation in the same way socket activation fds are
7555 passed. This is now used by journald to ensure that the
7556 various sockets connected to all the system's stdout/stderr
7557 are not lost when journald is restarted. File descriptors
7558 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
7559 an extension to sd_notify(). Note that a limit is enforced
7560 on the number of fds a service can store in PID 1, and it
7561 defaults to 0, so that no fds may be stored, unless this is
7562 explicitly turned on.
7564 * The default TERM variable to use for units connected to a
7565 terminal, when no other value is explicitly is set is now
7566 vt220 rather than vt102. This should be fairly safe still,
7567 but allows PgUp/PgDn work.
7569 * The /etc/crypttab option header= as known from Debian is now
7572 * "loginctl user-status" and "loginctl session-status" will
7573 now show the last 10 lines of log messages of the
7574 user/session following the status output. Similar,
7575 "machinectl status" will show the last 10 log lines
7576 associated with a virtual machine or container
7577 service. (Note that this is usually not the log messages
7578 done in the VM/container itself, but simply what the
7579 container manager logs. For nspawn this includes all console
7582 * "loginctl session-status" without further argument will now
7583 show the status of the session of the caller. Similar,
7584 "lock-session", "unlock-session", "activate",
7585 "enable-linger", "disable-linger" may now be called without
7586 session/user parameter in which case they apply to the
7587 caller's session/user.
7589 * An X11 session scriptlet is now shipped that uploads
7590 $DISPLAY and $XAUTHORITY into the environment of the systemd
7591 --user daemon if a session begins. This should improve
7592 compatibility with X11 enabled applications run as systemd
7595 * Generators are now subject to masking via /etc and /run, the
7596 same way as unit files.
7598 * networkd .network files gained support for configuring
7599 per-link IPv4/IPv6 packet forwarding as well as IPv4
7600 masquerading. This is by default turned on for veth links to
7601 containers, as registered by systemd-nspawn. This means that
7602 nspawn containers run with --network-veth will now get
7603 automatic routed access to the host's networks without any
7604 further configuration or setup, as long as networkd runs on
7607 * systemd-nspawn gained the --port= (-p) switch to expose TCP
7608 or UDP posts of a container on the host. With this in place
7609 it is possible to run containers with private veth links
7610 (--network-veth), and have their functionality exposed on
7611 the host as if their services were running directly on the
7614 * systemd-nspawn's --network-veth switch now gained a short
7615 version "-n", since with the changes above it is now truly
7616 useful out-of-the-box. The systemd-nspawn@.service has been
7617 updated to make use of it too by default.
7619 * systemd-nspawn will now maintain a per-image R/W lock, to
7620 ensure that the same image is not started more than once
7621 writable. (It's OK to run an image multiple times
7622 simultaneously in read-only mode.)
7624 * systemd-nspawn's --image= option is now capable of
7625 dissecting and booting MBR and GPT disk images that contain
7626 only a single active Linux partition. Previously it
7627 supported only GPT disk images with proper GPT type
7628 IDs. This allows running cloud images from major
7629 distributions directly with systemd-nspawn, without
7632 * In addition to collecting mouse dpi data in the udev
7633 hardware database, there's now support for collecting angle
7634 information for mouse scroll wheels. The database is
7635 supposed to guarantee similar scrolling behavior on mice
7636 that it knows about. There's also support for collecting
7637 information about Touchpad types.
7639 * udev's input_id built-in will now also collect touch screen
7640 dimension data and attach it to probed devices.
7642 * /etc/os-release gained support for a Distribution Privacy
7645 * networkd gained support for creating "ipvlan", "gretap",
7646 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
7648 * systemd-tmpfiles gained support for "a" lines for setting
7651 * systemd-nspawn will now mount /tmp in the container to
7652 tmpfs, automatically.
7654 * systemd now exposes the memory.usage_in_bytes cgroup
7655 attribute and shows it for each service in the "systemctl
7656 status" output, if available.
7658 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
7659 immediate reboot is triggered. This useful if shutdown is
7660 hung and is unable to complete, to expedite the
7661 operation. Note that this kind of reboot will still unmount
7662 all file systems, and hence should not result in fsck being
7665 * A .device unit for an optical block device will now be
7666 considered active only when a medium is in the drive. Also,
7667 mount units are now bound to their backing devices thus
7668 triggering automatic unmounting when devices become
7669 unavailable. With this in place systemd will now
7670 automatically unmount left-over mounts when a CD-ROM is
7671 ejected or an USB stick is yanked from the system.
7673 * networkd-wait-online now has support for waiting for
7674 specific interfaces only (with globbing), and for giving up
7675 after a configurable timeout.
7677 * networkd now exits when idle. It will be automatically
7678 restarted as soon as interfaces show up, are removed or
7679 change state. networkd will stay around as long as there is
7680 at least one DHCP state machine or similar around, that keep
7683 * networkd may now configure IPv6 link-local addressing in
7684 addition to IPv4 link-local addressing.
7686 * The IPv6 "token" for use in SLAAC may now be configured for
7687 each .network interface in networkd.
7689 * Routes configured with networkd may now be assigned a scope
7692 * networkd's [Match] sections now support globbing and lists
7693 of multiple space-separated matches per item.
7695 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
7696 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
7697 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
7698 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
7699 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
7700 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
7701 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
7702 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
7703 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
7704 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
7705 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
7706 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
7707 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
7708 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
7709 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
7710 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
7711 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
7712 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
7713 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
7714 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
7715 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
7716 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
7717 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
7718 Hoffmann, Zbigniew Jędrzejewski-Szmek
7720 — Berlin, 2015-02-16
7724 * When querying unit file enablement status (for example via
7725 "systemctl is-enabled"), a new state "indirect" is now known
7726 which indicates that a unit might not be enabled itself, but
7727 another unit listed in its Also= setting might be.
7729 * Similar to the various existing ConditionXYZ= settings for
7730 units, there are now matching AssertXYZ= settings. While
7731 failing conditions cause a unit to be skipped, but its job
7732 to succeed, failing assertions declared like this will cause
7733 a unit start operation and its job to fail.
7735 * hostnamed now knows a new chassis type "embedded".
7737 * systemctl gained a new "edit" command. When used on a unit
7738 file, this allows extending unit files with .d/ drop-in
7739 configuration snippets or editing the full file (after
7740 copying it from /usr/lib to /etc). This will invoke the
7741 user's editor (as configured with $EDITOR), and reload the
7742 modified configuration after editing.
7744 * "systemctl status" now shows the suggested enablement state
7745 for a unit, as declared in the (usually vendor-supplied)
7746 system preset files.
7748 * nss-myhostname will now resolve the single-label hostname
7749 "gateway" to the locally configured default IP routing
7750 gateways, ordered by their metrics. This assigns a stable
7751 name to the used gateways, regardless which ones are
7752 currently configured. Note that the name will only be
7753 resolved after all other name sources (if nss-myhostname is
7754 configured properly) and should hence not negatively impact
7755 systems that use the single-label hostname "gateway" in
7758 * systemd-inhibit now allows filtering by mode when listing
7761 * Scope and service units gained a new "Delegate" boolean
7762 property, which, when set, allows processes running inside the
7763 unit to further partition resources. This is primarily
7764 useful for systemd user instances as well as container
7767 * journald will now pick up audit messages directly from
7768 the kernel, and log them like any other log message. The
7769 audit fields are split up and fully indexed. This means that
7770 journalctl in many ways is now a (nicer!) alternative to
7771 ausearch, the traditional audit client. Note that this
7772 implements only a minimal audit client. If you want the
7773 special audit modes like reboot-on-log-overflow, please use
7774 the traditional auditd instead, which can be used in
7775 parallel to journald.
7777 * The ConditionSecurity= unit file option now understands the
7778 special string "audit" to check whether auditing is
7781 * journalctl gained two new commands --vacuum-size= and
7782 --vacuum-time= to delete old journal files until the
7783 remaining ones take up no more than the specified size on disk,
7784 or are not older than the specified time.
7786 * A new, native PPPoE library has been added to sd-network,
7787 systemd's library of light-weight networking protocols. This
7788 library will be used in a future version of networkd to
7789 enable PPPoE communication without an external pppd daemon.
7791 * The busctl tool now understands a new "capture" verb that
7792 works similar to "monitor", but writes a packet capture
7793 trace to STDOUT that can be redirected to a file which is
7794 compatible with libcap's capture file format. This can then
7795 be loaded in Wireshark and similar tools to inspect bus
7798 * The busctl tool now understands a new "tree" verb that shows
7799 the object trees of a specific service on the bus, or of all
7802 * The busctl tool now understands a new "introspect" verb that
7803 shows all interfaces and members of objects on the bus,
7804 including their signature and values. This is particularly
7805 useful to get more information about bus objects shown by
7806 the new "busctl tree" command.
7808 * The busctl tool now understands new verbs "call",
7809 "set-property" and "get-property" for invoking bus method
7810 calls, setting and getting bus object properties in a
7813 * busctl gained a new --augment-creds= argument that controls
7814 whether the tool shall augment credential information it
7815 gets from the bus with data from /proc, in a possibly
7818 * nspawn's --link-journal= switch gained two new values
7819 "try-guest" and "try-host" that work like "guest" and
7820 "host", but do not fail if the host has no persistent
7821 journaling enabled. -j is now equivalent to
7822 --link-journal=try-guest.
7824 * macvlan network devices created by nspawn will now have
7825 stable MAC addresses.
7827 * A new SmackProcessLabel= unit setting has been added, which
7828 controls the SMACK security label processes forked off by
7829 the respective unit shall use.
7831 * If compiled with --enable-xkbcommon, systemd-localed will
7832 verify x11 keymap settings by compiling the given keymap. It
7833 will spew out warnings if the compilation fails. This
7834 requires libxkbcommon to be installed.
7836 * When a coredump is collected, a larger number of metadata
7837 fields is now collected and included in the journal records
7838 created for it. More specifically, control group membership,
7839 environment variables, memory maps, working directory,
7840 chroot directory, /proc/$PID/status, and a list of open file
7841 descriptors is now stored in the log entry.
7843 * The udev hwdb now contains DPI information for mice. For
7846 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
7848 * All systemd programs that read standalone configuration
7849 files in /etc now also support a corresponding series of
7850 .conf.d configuration directories in /etc/, /run/,
7851 /usr/local/lib/, /usr/lib/, and (if configured with
7852 --enable-split-usr) /lib/. In particular, the following
7853 configuration files now have corresponding configuration
7854 directories: system.conf user.conf, logind.conf,
7855 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
7856 resolved.conf, timesyncd.conf, journal-remote.conf, and
7857 journal-upload.conf. Note that distributions should use the
7858 configuration directories in /usr/lib/; the directories in
7859 /etc/ are reserved for the system administrator.
7861 * systemd-rfkill will no longer take the rfkill device name
7862 into account when storing rfkill state on disk, as the name
7863 might be dynamically assigned and not stable. Instead, the
7864 ID_PATH udev variable combined with the rfkill type (wlan,
7865 bluetooth, …) is used.
7867 * A new service systemd-machine-id-commit.service has been
7868 added. When used on systems where /etc is read-only during
7869 boot, and /etc/machine-id is not initialized (but an empty
7870 file), this service will copy the temporary machine ID
7871 created as replacement into /etc after the system is fully
7872 booted up. This is useful for systems that are freshly
7873 installed with a non-initialized machine ID, but should get
7874 a fixed machine ID for subsequent boots.
7876 * networkd's .netdev files now provide a large set of
7877 configuration parameters for VXLAN devices. Similarly, the
7878 bridge port cost parameter is now configurable in .network
7879 files. There's also new support for configuring IP source
7880 routing. networkd .link files gained support for a new
7881 OriginalName= match that is useful to match against the
7882 original interface name the kernel assigned. .network files
7883 may include MTU= and MACAddress= fields for altering the MTU
7884 and MAC address while being connected to a specific network
7887 * The LUKS logic gained supported for configuring
7888 UUID-specific key files. There's also new support for naming
7889 LUKS device from the kernel command line, using the new
7890 luks.name= argument.
7892 * Timer units may now be transiently created via the bus API
7893 (this was previously already available for scope and service
7894 units). In addition it is now possible to create multiple
7895 transient units at the same time with a single bus call. The
7896 "systemd-run" tool has been updated to make use of this for
7897 running commands on a specified time, in at(1)-style.
7899 * tmpfiles gained support for "t" lines, for assigning
7900 extended attributes to files. Among other uses this may be
7901 used to assign SMACK labels to files.
7903 Contributions from: Alin Rauta, Alison Chaiken, Andrej
7904 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
7905 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
7906 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
7907 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
7908 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
7909 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
7910 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
7911 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
7912 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
7913 Michael Biebl, Michael Chapman, Michael Marineau, Michal
7914 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
7915 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
7916 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
7917 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
7918 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
7919 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
7920 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
7922 — Berlin, 2014-12-10
7926 * journalctl gained the new options -t/--identifier= to match
7927 on the syslog identifier (aka "tag"), as well as --utc to
7928 show log timestamps in the UTC timezone. journalctl now also
7929 accepts -n/--lines=all to disable line capping in a pager.
7931 * journalctl gained a new switch, --flush, that synchronously
7932 flushes logs from /run/log/journal to /var/log/journal if
7933 persistent storage is enabled. systemd-journal-flush.service
7934 now waits until the operation is complete.
7936 * Services can notify the manager before they start a reload
7937 (by sending RELOADING=1) or shutdown (by sending
7938 STOPPING=1). This allows the manager to track and show the
7939 internal state of daemons and closes a race condition when
7940 the process is still running but has closed its D-Bus
7943 * Services with Type=oneshot do not have to have any ExecStart
7946 * User units are now loaded also from
7947 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
7948 /run/systemd/user directory that was already previously
7949 supported, but is under the control of the user.
7951 * Job timeouts (i.e. timeouts on the time a job that is
7952 queued stays in the run queue) can now optionally result in
7953 immediate reboot or power-off actions (JobTimeoutAction= and
7954 JobTimeoutRebootArgument=). This is useful on ".target"
7955 units, to limit the maximum time a target remains
7956 undispatched in the run queue, and to trigger an emergency
7957 operation in such a case. This is now used by default to
7958 turn off the system if boot-up (as defined by everything in
7959 basic.target) hangs and does not complete for at least
7960 15min. Also, if power-off or reboot hang for at least 30min
7961 an immediate power-off/reboot operation is triggered. This
7962 functionality is particularly useful to increase reliability
7963 on embedded devices, but also on laptops which might
7964 accidentally get powered on when carried in a backpack and
7965 whose boot stays stuck in a hard disk encryption passphrase
7968 * systemd-logind can be configured to also handle lid switch
7969 events even when the machine is docked or multiple displays
7970 are attached (HandleLidSwitchDocked= option).
7972 * A helper binary and a service have been added which can be
7973 used to resume from hibernation in the initramfs. A
7974 generator will parse the resume= option on the kernel
7975 command line to trigger resume.
7977 * A user console daemon systemd-consoled has been
7978 added. Currently, it is a preview, and will so far open a
7979 single terminal on each session of the user marked as
7980 Desktop=systemd-console.
7982 * Route metrics can be specified for DHCP routes added by
7985 * The SELinux context of socket-activated services can be set
7986 from the information provided by the networking stack
7987 (SELinuxContextFromNet= option).
7989 * Userspace firmware loading support has been removed and
7990 the minimum supported kernel version is thus bumped to 3.7.
7992 * Timeout for udev workers has been increased from 1 to 3
7993 minutes, but a warning will be printed after 1 minute to
7994 help diagnose kernel modules that take a long time to load.
7996 * Udev rules can now remove tags on devices with TAG-="foobar".
7998 * systemd's readahead implementation has been removed. In many
7999 circumstances it didn't give expected benefits even for
8000 rotational disk drives and was becoming less relevant in the
8001 age of SSDs. As none of the developers has been using
8002 rotating media anymore, and nobody stepped up to actively
8003 maintain this component of systemd it has now been removed.
8005 * Swap units can use Options= to specify discard options.
8006 Discard options specified for swaps in /etc/fstab are now
8009 * Docker containers are now detected as a separate type of
8012 * The Password Agent protocol gained support for queries where
8013 the user input is shown, useful e.g. for user names.
8014 systemd-ask-password gained a new --echo option to turn that
8017 * The default sysctl.d/ snippets will now set:
8019 net.core.default_qdisc = fq_codel
8021 This selects Fair Queuing Controlled Delay as the default
8022 queuing discipline for network interfaces. fq_codel helps
8023 fight the network bufferbloat problem. It is believed to be
8024 a good default with no tuning required for most workloads.
8025 Downstream distributions may override this choice. On 10Gbit
8026 servers that do not do forwarding, "fq" may perform better.
8027 Systems without a good clocksource should use "pfifo_fast".
8029 * If kdbus is enabled during build a new option BusPolicy= is
8030 available for service units, that allows locking all service
8031 processes into a stricter bus policy, in order to limit
8032 access to various bus services, or even hide most of them
8033 from the service's view entirely.
8035 * networkctl will now show the .network and .link file
8036 networkd has applied to a specific interface.
8038 * sd-login gained a new API call sd_session_get_desktop() to
8039 query which desktop environment has been selected for a
8042 * UNIX utmp support is now compile-time optional to support
8043 legacy-free systems.
8045 * systemctl gained two new commands "add-wants" and
8046 "add-requires" for pulling in units from specific targets
8049 * If the word "rescue" is specified on the kernel command line
8050 the system will now boot into rescue mode (aka
8051 rescue.target), which was previously available only by
8052 specifying "1" or "systemd.unit=rescue.target" on the kernel
8053 command line. This new kernel command line option nicely
8054 mirrors the already existing "emergency" kernel command line
8057 * New kernel command line options mount.usr=, mount.usrflags=,
8058 mount.usrfstype= have been added that match root=, rootflags=,
8059 rootfstype= but allow mounting a specific file system to
8062 * The $NOTIFY_SOCKET is now also passed to control processes of
8063 services, not only the main process.
8065 * This version reenables support for fsck's -l switch. This
8066 means at least version v2.25 of util-linux is required for
8067 operation, otherwise dead-locks on device nodes may
8068 occur. Again: you need to update util-linux to at least
8069 v2.25 when updating systemd to v217.
8071 * The "multi-seat-x" tool has been removed from systemd, as
8072 its functionality has been integrated into X servers 1.16,
8073 and the tool is hence redundant. It is recommended to update
8074 display managers invoking this tool to simply invoke X
8075 directly from now on, again.
8077 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
8078 message flag has been added for all of systemd's polkit
8079 authenticated method calls has been added. In particular this
8080 now allows optional interactive authorization via polkit for
8081 many of PID1's privileged operations such as unit file
8082 enabling and disabling.
8084 * "udevadm hwdb --update" learnt a new switch "--usr" for
8085 placing the rebuilt hardware database in /usr instead of
8086 /etc. When used only hardware database entries stored in
8087 /usr will be used, and any user database entries in /etc are
8088 ignored. This functionality is useful for vendors to ship a
8089 pre-built database on systems where local configuration is
8090 unnecessary or unlikely.
8092 * Calendar time specifications in .timer units now also
8093 understand the strings "semi-annually", "quarterly" and
8094 "minutely" as shortcuts (in addition to the preexisting
8095 "annually", "hourly", …).
8097 * systemd-tmpfiles will now correctly create files in /dev
8098 at boot which are marked for creation only at boot. It is
8099 recommended to always create static device nodes with 'c!'
8100 and 'b!', so that they are created only at boot and not
8101 overwritten at runtime.
8103 * When the watchdog logic is used for a service (WatchdogSec=)
8104 and the watchdog timeout is hit the service will now be
8105 terminated with SIGABRT (instead of just SIGTERM), in order
8106 to make sure a proper coredump and backtrace is
8107 generated. This ensures that hanging services will result in
8108 similar coredump/backtrace behaviour as services that hit a
8111 Contributions from: Andreas Henriksson, Andrei Borzenkov,
8112 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
8113 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
8114 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
8115 Herrmann, David Sommerseth, David Strauss, Emil Renner
8116 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
8117 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
8118 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
8119 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
8120 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
8121 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
8122 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
8123 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
8124 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
8125 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
8126 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
8127 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
8128 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
8129 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
8130 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
8131 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
8134 — Berlin, 2014-10-28
8138 * timedated no longer reads NTP implementation unit names from
8139 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
8140 implementations should add a
8142 Conflicts=systemd-timesyncd.service
8144 to their unit files to take over and replace systemd's NTP
8145 default functionality.
8147 * systemd-sysusers gained a new line type "r" for configuring
8148 which UID/GID ranges to allocate system users/groups
8149 from. Lines of type "u" may now add an additional column
8150 that specifies the home directory for the system user to be
8151 created. Also, systemd-sysusers may now optionally read user
8152 information from STDIN instead of a file. This is useful for
8153 invoking it from RPM preinst scriptlets that need to create
8154 users before the first RPM file is installed since these
8155 files might need to be owned by them. A new
8156 %sysusers_create_inline RPM macro has been introduced to do
8157 just that. systemd-sysusers now updates the shadow files as
8158 well as the user/group databases, which should enhance
8159 compatibility with certain tools like grpck.
8161 * A number of bus APIs of PID 1 now optionally consult polkit to
8162 permit access for otherwise unprivileged clients under certain
8163 conditions. Note that this currently doesn't support
8164 interactive authentication yet, but this is expected to be
8165 added eventually, too.
8167 * /etc/machine-info now has new fields for configuring the
8168 deployment environment of the machine, as well as the
8169 location of the machine. hostnamectl has been updated with
8170 new command to update these fields.
8172 * systemd-timesyncd has been updated to automatically acquire
8173 NTP server information from systemd-networkd, which might
8174 have been discovered via DHCP.
8176 * systemd-resolved now includes a caching DNS stub resolver
8177 and a complete LLMNR name resolution implementation. A new
8178 NSS module "nss-resolve" has been added which can be used
8179 instead of glibc's own "nss-dns" to resolve hostnames via
8180 systemd-resolved. Hostnames, addresses and arbitrary RRs may
8181 be resolved via systemd-resolved D-Bus APIs. In contrast to
8182 the glibc internal resolver systemd-resolved is aware of
8183 multi-homed system, and keeps DNS server and caches separate
8184 and per-interface. Queries are sent simultaneously on all
8185 interfaces that have DNS servers configured, in order to
8186 properly handle VPNs and local LANs which might resolve
8187 separate sets of domain names. systemd-resolved may acquire
8188 DNS server information from systemd-networkd automatically,
8189 which in turn might have discovered them via DHCP. A tool
8190 "systemd-resolve-host" has been added that may be used to
8191 query the DNS logic in resolved. systemd-resolved implements
8192 IDNA and automatically uses IDNA or UTF-8 encoding depending
8193 on whether classic DNS or LLMNR is used as transport. In the
8194 next releases we intend to add a DNSSEC and mDNS/DNS-SD
8195 implementation to systemd-resolved.
8197 * A new NSS module nss-mymachines has been added, that
8198 automatically resolves the names of all local registered
8199 containers to their respective IP addresses.
8201 * A new client tool "networkctl" for systemd-networkd has been
8202 added. It currently is entirely passive and will query
8203 networking configuration from udev, rtnetlink and networkd,
8204 and present it to the user in a very friendly
8205 way. Eventually, we hope to extend it to become a full
8206 control utility for networkd.
8208 * .socket units gained a new DeferAcceptSec= setting that
8209 controls the kernels' TCP_DEFER_ACCEPT sockopt for
8210 TCP. Similarly, support for controlling TCP keep-alive
8211 settings has been added (KeepAliveTimeSec=,
8212 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
8213 turning off Nagle's algorithm on TCP has been added
8216 * logind learned a new session type "web", for use in projects
8217 like Cockpit which register web clients as PAM sessions.
8219 * timer units with at least one OnCalendar= setting will now
8220 be started only after time-sync.target has been
8221 reached. This way they will not elapse before the system
8222 clock has been corrected by a local NTP client or
8223 similar. This is particular useful on RTC-less embedded
8224 machines, that come up with an invalid system clock.
8226 * systemd-nspawn's --network-veth= switch should now result in
8227 stable MAC addresses for both the outer and the inner side
8230 * systemd-nspawn gained a new --volatile= switch for running
8231 container instances with /etc or /var unpopulated.
8233 * The kdbus client code has been updated to use the new Linux
8234 3.17 memfd subsystem instead of the old kdbus-specific one.
8236 * systemd-networkd's DHCP client and server now support
8237 FORCERENEW. There are also new configuration options to
8238 configure the vendor client identifier and broadcast mode
8241 * systemd will no longer inform the kernel about the current
8242 timezone, as this is necessarily incorrect and racy as the
8243 kernel has no understanding of DST and similar
8244 concepts. This hence means FAT timestamps will be always
8245 considered UTC, similar to what Android is already
8246 doing. Also, when the RTC is configured to the local time
8247 (rather than UTC) systemd will never synchronize back to it,
8248 as this might confuse Windows at a later boot.
8250 * systemd-analyze gained a new command "verify" for offline
8251 validation of unit files.
8253 * systemd-networkd gained support for a couple of additional
8254 settings for bonding networking setups. Also, the metric for
8255 statically configured routes may now be configured. For
8256 network interfaces where this is appropriate the peer IP
8257 address may now be configured.
8259 * systemd-networkd's DHCP client will no longer request
8260 broadcasting by default, as this tripped up some networks.
8261 For hardware where broadcast is required the feature should
8262 be switched back on using RequestBroadcast=yes.
8264 * systemd-networkd will now set up IPv4LL addresses (when
8265 enabled) even if DHCP is configured successfully.
8267 * udev will now default to respect network device names given
8268 by the kernel when the kernel indicates that these are
8269 predictable. This behavior can be tweaked by changing
8270 NamePolicy= in the relevant .link file.
8272 * A new library systemd-terminal has been added that
8273 implements full TTY stream parsing and rendering. This
8274 library is supposed to be used later on for implementing a
8275 full userspace VT subsystem, replacing the current kernel
8278 * A new tool systemd-journal-upload has been added to push
8279 journal data to a remote system running
8280 systemd-journal-remote.
8282 * journald will no longer forward all local data to another
8283 running syslog daemon. This change has been made because
8284 rsyslog (which appears to be the most commonly used syslog
8285 implementation these days) no longer makes use of this, and
8286 instead pulls the data out of the journal on its own. Since
8287 forwarding the messages to a non-existent syslog server is
8288 more expensive than we assumed we have now turned this
8289 off. If you run a syslog server that is not a recent rsyslog
8290 version, you have to turn this option on again
8291 (ForwardToSyslog= in journald.conf).
8293 * journald now optionally supports the LZ4 compressor for
8294 larger journal fields. This compressor should perform much
8295 better than XZ which was the previous default.
8297 * machinectl now shows the IP addresses of local containers,
8298 if it knows them, plus the interface name of the container.
8300 * A new tool "systemd-escape" has been added that makes it
8301 easy to escape strings to build unit names and similar.
8303 * sd_notify() messages may now include a new ERRNO= field
8304 which is parsed and collected by systemd and shown among the
8305 "systemctl status" output for a service.
8307 * A new component "systemd-firstboot" has been added that
8308 queries the most basic systemd information (timezone,
8309 hostname, root password) interactively on first
8310 boot. Alternatively it may also be used to provision these
8311 things offline on OS images installed into directories.
8313 * The default sysctl.d/ snippets will now set
8315 net.ipv4.conf.default.promote_secondaries=1
8317 This has the benefit of no flushing secondary IP addresses
8318 when primary addresses are removed.
8320 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
8321 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
8322 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
8323 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
8324 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
8325 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
8326 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
8327 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
8328 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
8329 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
8330 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
8331 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
8332 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
8333 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
8334 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
8336 — Berlin, 2014-08-19
8340 * A new tool systemd-sysusers has been added. This tool
8341 creates system users and groups in /etc/passwd and
8342 /etc/group, based on static declarative system user/group
8343 definitions in /usr/lib/sysusers.d/. This is useful to
8344 enable factory resets and volatile systems that boot up with
8345 an empty /etc directory, and thus need system users and
8346 groups created during early boot. systemd now also ships
8347 with two default sysusers.d/ files for the most basic
8348 users and groups systemd and the core operating system
8351 * A new tmpfiles snippet has been added that rebuilds the
8352 essential files in /etc on boot, should they be missing.
8354 * A directive for ensuring automatic clean-up of
8355 /var/cache/man/ has been removed from the default
8356 configuration. This line should now be shipped by the man
8357 implementation. The necessary change has been made to the
8358 man-db implementation. Note that you need to update your man
8359 implementation to one that ships this line, otherwise no
8360 automatic clean-up of /var/cache/man will take place.
8362 * A new condition ConditionNeedsUpdate= has been added that
8363 may conditionalize services to only run when /etc or /var
8364 are "older" than the vendor operating system resources in
8365 /usr. This is useful for reconstructing or updating /etc
8366 after an offline update of /usr or a factory reset, on the
8367 next reboot. Services that want to run once after such an
8368 update or reset should use this condition and order
8369 themselves before the new systemd-update-done.service, which
8370 will mark the two directories as fully updated. A number of
8371 service files have been added making use of this, to rebuild
8372 the udev hardware database, the journald message catalog and
8373 dynamic loader cache (ldconfig). The systemd-sysusers tool
8374 described above also makes use of this now. With this in
8375 place it is now possible to start up a minimal operating
8376 system with /etc empty cleanly. For more information on the
8377 concepts involved see this recent blog story:
8379 http://0pointer.de/blog/projects/stateless.html
8381 * A new system group "input" has been introduced, and all
8382 input device nodes get this group assigned. This is useful
8383 for system-level software to get access to input devices. It
8384 complements what is already done for "audio" and "video".
8386 * systemd-networkd learnt minimal DHCPv4 server support in
8387 addition to the existing DHCPv4 client support. It also
8388 learnt DHCPv6 client and IPv6 Router Solicitation client
8389 support. The DHCPv4 client gained support for static routes
8390 passed in from the server. Note that the [DHCPv4] section
8391 known in older systemd-networkd versions has been renamed to
8392 [DHCP] and is now also used by the DHCPv6 client. Existing
8393 .network files using settings of this section should be
8394 updated, though compatibility is maintained. Optionally, the
8395 client hostname may now be sent to the DHCP server.
8397 * networkd gained support for vxlan virtual networks as well
8398 as tun/tap and dummy devices.
8400 * networkd gained support for automatic allocation of address
8401 ranges for interfaces from a system-wide pool of
8402 addresses. This is useful for dynamically managing a large
8403 number of interfaces with a single network configuration
8404 file. In particular this is useful to easily assign
8405 appropriate IP addresses to the veth links of a large number
8406 of nspawn instances.
8408 * RPM macros for processing sysusers, sysctl and binfmt
8409 drop-in snippets at package installation time have been
8412 * The /etc/os-release file should now be placed in
8413 /usr/lib/os-release. The old location is automatically
8414 created as symlink. /usr/lib is the more appropriate
8415 location of this file, since it shall actually describe the
8416 vendor operating system shipped in /usr, and not the
8417 configuration stored in /etc.
8419 * .mount units gained a new boolean SloppyOptions= setting
8420 that maps to mount(8)'s -s option which enables permissive
8421 parsing of unknown mount options.
8423 * tmpfiles learnt a new "L+" directive which creates a symlink
8424 but (unlike "L") deletes a pre-existing file first, should
8425 it already exist and not already be the correct
8426 symlink. Similarly, "b+", "c+" and "p+" directives have been
8427 added as well, which create block and character devices, as
8428 well as fifos in the filesystem, possibly removing any
8429 pre-existing files of different types.
8431 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
8432 'argument' field (which so far specified the source to
8433 symlink/copy the files from) is now optional. If omitted the
8434 same file os copied from /usr/share/factory/ suffixed by the
8435 full destination path. This is useful for populating /etc
8436 with essential files, by copying them from vendor defaults
8437 shipped in /usr/share/factory/etc.
8439 * A new command "systemctl preset-all" has been added that
8440 applies the service preset settings to all installed unit
8441 files. A new switch --preset-mode= has been added that
8442 controls whether only enable or only disable operations
8445 * A new command "systemctl is-system-running" has been added
8446 that allows checking the overall state of the system, for
8447 example whether it is fully up and running.
8449 * When the system boots up with an empty /etc, the equivalent
8450 to "systemctl preset-all" is executed during early boot, to
8451 make sure all default services are enabled after a factory
8454 * systemd now contains a minimal preset file that enables the
8455 most basic services systemd ships by default.
8457 * Unit files' [Install] section gained a new DefaultInstance=
8458 field for defining the default instance to create if a
8459 template unit is enabled with no instance specified.
8461 * A new passive target cryptsetup-pre.target has been added
8462 that may be used by services that need to make they run and
8463 finish before the first LUKS cryptographic device is set up.
8465 * The /dev/loop-control and /dev/btrfs-control device nodes
8466 are now owned by the "disk" group by default, opening up
8467 access to this group.
8469 * systemd-coredump will now automatically generate a
8470 stack trace of all core dumps taking place on the system,
8471 based on elfutils' libdw library. This stack trace is logged
8474 * systemd-coredump may now optionally store coredumps directly
8475 on disk (in /var/lib/systemd/coredump, possibly compressed),
8476 instead of storing them unconditionally in the journal. This
8477 mode is the new default. A new configuration file
8478 /etc/systemd/coredump.conf has been added to configure this
8479 and other parameters of systemd-coredump.
8481 * coredumpctl gained a new "info" verb to show details about a
8482 specific coredump. A new switch "-1" has also been added
8483 that makes sure to only show information about the most
8484 recent entry instead of all entries. Also, as the tool is
8485 generally useful now the "systemd-" prefix of the binary
8486 name has been removed. Distributions that want to maintain
8487 compatibility with the old name should add a symlink from
8488 the old name to the new name.
8490 * journald's SplitMode= now defaults to "uid". This makes sure
8491 that unprivileged users can access their own coredumps with
8492 coredumpctl without restrictions.
8494 * New kernel command line options "systemd.wants=" (for
8495 pulling an additional unit during boot), "systemd.mask="
8496 (for masking a specific unit for the boot), and
8497 "systemd.debug-shell" (for enabling the debug shell on tty9)
8498 have been added. This is implemented in the new generator
8499 "systemd-debug-generator".
8501 * systemd-nspawn will now by default filter a couple of
8502 syscalls for containers, among them those required for
8503 kernel module loading, direct x86 IO port access, swap
8504 management, and kexec. Most importantly though
8505 open_by_handle_at() is now prohibited for containers,
8506 closing a hole similar to a recently discussed vulnerability
8507 in docker regarding access to files on file hierarchies the
8508 container should normally not have access to. Note that, for
8509 nspawn, we generally make no security claims anyway (and
8510 this is explicitly documented in the man page), so this is
8511 just a fix for one of the most obvious problems.
8513 * A new man page file-hierarchy(7) has been added that
8514 contains a minimized, modernized version of the file system
8515 layout systemd expects, similar in style to the FHS
8516 specification or hier(5). A new tool systemd-path(1) has
8517 been added to query many of these paths for the local
8520 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
8521 longer done. Since the directory now has a per-user size
8522 limit, and is cleaned on logout this appears unnecessary,
8523 in particular since this now brings the lifecycle of this
8524 directory closer in line with how IPC objects are handled.
8526 * systemd.pc now exports a number of additional directories,
8527 including $libdir (which is useful to identify the library
8528 path for the primary architecture of the system), and a
8529 couple of drop-in directories.
8531 * udev's predictable network interface names now use the dev_port
8532 sysfs attribute, introduced in linux 3.15 instead of dev_id to
8533 distinguish between ports of the same PCI function. dev_id should
8534 only be used for ports using the same HW address, hence the need
8537 * machined has been updated to export the OS version of a
8538 container (read from /etc/os-release and
8539 /usr/lib/os-release) on the bus. This is now shown in
8540 "machinectl status" for a machine.
8542 * A new service setting RestartForceExitStatus= has been
8543 added. If configured to a set of exit signals or process
8544 return values, the service will be restarted when the main
8545 daemon process exits with any of them, regardless of the
8548 * systemctl's -H switch for connecting to remote systemd
8549 machines has been extended so that it may be used to
8550 directly connect to a specific container on the
8551 host. "systemctl -H root@foobar:waldi" will now connect as
8552 user "root" to host "foobar", and then proceed directly to
8553 the container named "waldi". Note that currently you have to
8554 authenticate as user "root" for this to work, as entering
8555 containers is a privileged operation.
8557 Contributions from: Andreas Henriksson, Benjamin Steinwender,
8558 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
8559 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
8560 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
8561 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
8562 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
8563 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
8564 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
8565 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
8566 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
8567 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
8568 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
8570 — Berlin, 2014-07-03
8574 * As an experimental feature, udev now tries to lock the
8575 disk device node (flock(LOCK_SH|LOCK_NB)) while it
8576 executes events for the disk or any of its partitions.
8577 Applications like partitioning programs can lock the
8578 disk device node (flock(LOCK_EX)) and claim temporary
8579 device ownership that way; udev will entirely skip all event
8580 handling for this disk and its partitions. If the disk
8581 was opened for writing, the close will trigger a partition
8582 table rescan in udev's "watch" facility, and if needed
8583 synthesize "change" events for the disk and all its partitions.
8584 This is now unconditionally enabled, and if it turns out to
8585 cause major problems, we might turn it on only for specific
8586 devices, or might need to disable it entirely. Device Mapper
8587 devices are excluded from this logic.
8589 * We temporarily dropped the "-l" switch for fsck invocations,
8590 since they collide with the flock() logic above. util-linux
8591 upstream has been changed already to avoid this conflict,
8592 and we will re-add "-l" as soon as util-linux with this
8593 change has been released.
8595 * The dependency on libattr has been removed. Since a long
8596 time, the extended attribute calls have moved to glibc, and
8597 libattr is thus unnecessary.
8599 * Virtualization detection works without privileges now. This
8600 means the systemd-detect-virt binary no longer requires
8601 CAP_SYS_PTRACE file capabilities, and our daemons can run
8602 with fewer privileges.
8604 * systemd-networkd now runs under its own "systemd-network"
8605 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
8606 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
8607 loses the ability to write to files owned by root this way.
8609 * Similarly, systemd-resolved now runs under its own
8610 "systemd-resolve" user with no capabilities remaining.
8612 * Similarly, systemd-bus-proxyd now runs under its own
8613 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
8615 * systemd-networkd gained support for setting up "veth"
8616 virtual Ethernet devices for container connectivity, as well
8617 as GRE and VTI tunnels.
8619 * systemd-networkd will no longer automatically attempt to
8620 manually load kernel modules necessary for certain tunnel
8621 transports. Instead, it is assumed the kernel loads them
8622 automatically when required. This only works correctly on
8623 very new kernels. On older kernels, please consider adding
8624 the kernel modules to /etc/modules-load.d/ as a work-around.
8626 * The resolv.conf file systemd-resolved generates has been
8627 moved to /run/systemd/resolve/. If you have a symlink from
8628 /etc/resolv.conf, it might be necessary to correct it.
8630 * Two new service settings, ProtectHome= and ProtectSystem=,
8631 have been added. When enabled, they will make the user data
8632 (such as /home) inaccessible or read-only and the system
8633 (such as /usr) read-only, for specific services. This allows
8634 very light-weight per-service sandboxing to avoid
8635 modifications of user data or system files from
8636 services. These two new switches have been enabled for all
8637 of systemd's long-running services, where appropriate.
8639 * Socket units gained new SocketUser= and SocketGroup=
8640 settings to set the owner user and group of AF_UNIX sockets
8641 and FIFOs in the file system.
8643 * Socket units gained a new RemoveOnStop= setting. If enabled,
8644 all FIFOS and sockets in the file system will be removed
8645 when the specific socket unit is stopped.
8647 * Socket units gained a new Symlinks= setting. It takes a list
8648 of symlinks to create to file system sockets or FIFOs
8649 created by the specific Unix sockets. This is useful to
8650 manage symlinks to socket nodes with the same lifecycle as
8653 * The /dev/log socket and /dev/initctl FIFO have been moved to
8654 /run, and have been replaced by symlinks. This allows
8655 connecting to these facilities even if PrivateDevices=yes is
8656 used for a service (which makes /dev/log itself unavailable,
8657 but /run is left). This also has the benefit of ensuring
8658 that /dev only contains device nodes, directories and
8659 symlinks, and nothing else.
8661 * sd-daemon gained two new calls sd_pid_notify() and
8662 sd_pid_notifyf(). They are similar to sd_notify() and
8663 sd_notifyf(), but allow overriding of the source PID of
8664 notification messages if permissions permit this. This is
8665 useful to send notify messages on behalf of a different
8666 process (for example, the parent process). The
8667 systemd-notify tool has been updated to make use of this
8668 when sending messages (so that notification messages now
8669 originate from the shell script invoking systemd-notify and
8670 not the systemd-notify process itself. This should minimize
8671 a race where systemd fails to associate notification
8672 messages to services when the originating process already
8675 * A new "on-abnormal" setting for Restart= has been added. If
8676 set, it will result in automatic restarts on all "abnormal"
8677 reasons for a process to exit, which includes unclean
8678 signals, core dumps, timeouts and watchdog timeouts, but
8679 does not include clean and unclean exit codes or clean
8680 signals. Restart=on-abnormal is an alternative for
8681 Restart=on-failure for services that shall be able to
8682 terminate and avoid restarts on certain errors, by
8683 indicating so with an unclean exit code. Restart=on-failure
8684 or Restart=on-abnormal is now the recommended setting for
8685 all long-running services.
8687 * If the InaccessibleDirectories= service setting points to a
8688 mount point (or if there are any submounts contained within
8689 it), it is now attempted to completely unmount it, to make
8690 the file systems truly unavailable for the respective
8693 * The ReadOnlyDirectories= service setting and
8694 systemd-nspawn's --read-only parameter are now recursively
8695 applied to all submounts, too.
8697 * Mount units may now be created transiently via the bus APIs.
8699 * The support for SysV and LSB init scripts has been removed
8700 from the systemd daemon itself. Instead, it is now
8701 implemented as a generator that creates native systemd units
8702 from these scripts when needed. This enables us to remove a
8703 substantial amount of legacy code from PID 1, following the
8704 fact that many distributions only ship a very small number
8705 of LSB/SysV init scripts nowadays.
8707 * Privileged Xen (dom0) domains are not considered
8708 virtualization anymore by the virtualization detection
8709 logic. After all, they generally have unrestricted access to
8710 the hardware and usually are used to manage the unprivileged
8713 * systemd-tmpfiles gained a new "C" line type, for copying
8714 files or entire directories.
8716 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
8717 lines. So far, they have been non-globbing versions of the
8718 latter, and have thus been redundant. In future, it is
8719 recommended to only use "z". "m" has hence been removed
8720 from the documentation, even though it stays supported.
8722 * A tmpfiles snippet to recreate the most basic structure in
8723 /var has been added. This is enough to create the /var/run →
8724 /run symlink and create a couple of structural
8725 directories. This allows systems to boot up with an empty or
8726 volatile /var. Of course, while with this change, the core OS
8727 now is capable with dealing with a volatile /var, not all
8728 user services are ready for it. However, we hope that sooner
8729 or later, many service daemons will be changed upstream so
8730 that they are able to automatically create their necessary
8731 directories in /var at boot, should they be missing. This is
8732 the first step to allow state-less systems that only require
8733 the vendor image for /usr to boot.
8735 * systemd-nspawn has gained a new --tmpfs= switch to mount an
8736 empty tmpfs instance to a specific directory. This is
8737 particularly useful for making use of the automatic
8738 reconstruction of /var (see above), by passing --tmpfs=/var.
8740 * Access modes specified in tmpfiles snippets may now be
8741 prefixed with "~", which indicates that they shall be masked
8742 by whether the existing file or directory is currently
8743 writable, readable or executable at all. Also, if specified,
8744 the sgid/suid/sticky bits will be masked for all
8747 * A new passive target unit "network-pre.target" has been
8748 added which is useful for services that shall run before any
8749 network is configured, for example firewall scripts.
8751 * The "floppy" group that previously owned the /dev/fd*
8752 devices is no longer used. The "disk" group is now used
8753 instead. Distributions should probably deprecate usage of
8756 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
8757 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
8758 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
8759 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
8760 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
8761 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
8764 — Berlin, 2014-06-11
8768 * A new "systemd-timesyncd" daemon has been added for
8769 synchronizing the system clock across the network. It
8770 implements an SNTP client. In contrast to NTP
8771 implementations such as chrony or the NTP reference server,
8772 this only implements a client side, and does not bother with
8773 the full NTP complexity, focusing only on querying time from
8774 one remote server and synchronizing the local clock to
8775 it. Unless you intend to serve NTP to networked clients or
8776 want to connect to local hardware clocks, this simple NTP
8777 client should be more than appropriate for most
8778 installations. The daemon runs with minimal privileges, and
8779 has been hooked up with networkd to only operate when
8780 network connectivity is available. The daemon saves the
8781 current clock to disk every time a new NTP sync has been
8782 acquired, and uses this to possibly correct the system clock
8783 early at bootup, in order to accommodate for systems that
8784 lack an RTC such as the Raspberry Pi and embedded devices,
8785 and to make sure that time monotonically progresses on these
8786 systems, even if it is not always correct. To make use of
8787 this daemon, a new system user and group "systemd-timesync"
8788 needs to be created on installation of systemd.
8790 * The queue "seqnum" interface of libudev has been disabled, as
8791 it was generally incompatible with device namespacing as
8792 sequence numbers of devices go "missing" if the devices are
8793 part of a different namespace.
8795 * "systemctl list-timers" and "systemctl list-sockets" gained
8796 a --recursive switch for showing units of these types also
8797 for all local containers, similar in style to the already
8798 supported --recursive switch for "systemctl list-units".
8800 * A new RebootArgument= setting has been added for service
8801 units, which may be used to specify a kernel reboot argument
8802 to use when triggering reboots with StartLimitAction=.
8804 * A new FailureAction= setting has been added for service
8805 units which may be used to specify an operation to trigger
8806 when a service fails. This works similarly to
8807 StartLimitAction=, but unlike it, controls what is done
8808 immediately rather than only after several attempts to
8809 restart the service in question.
8811 * hostnamed got updated to also expose the kernel name,
8812 release, and version on the bus. This is useful for
8813 executing commands like hostnamectl with the -H switch.
8814 systemd-analyze makes use of this to properly display
8815 details when running non-locally.
8817 * The bootchart tool can now show cgroup information in the
8818 graphs it generates.
8820 * The CFS CPU quota cgroup attribute is now exposed for
8821 services. The new CPUQuota= switch has been added for this
8822 which takes a percentage value. Setting this will have the
8823 result that a service may never get more CPU time than the
8824 specified percentage, even if the machine is otherwise idle.
8826 * systemd-networkd learned IPIP and SIT tunnel support.
8828 * LSB init scripts exposing a dependency on $network will now
8829 get a dependency on network-online.target rather than simply
8830 network.target. This should bring LSB handling closer to
8831 what it was on SysV systems.
8833 * A new fsck.repair= kernel option has been added to control
8834 how fsck shall deal with unclean file systems at boot.
8836 * The (.ini) configuration file parser will now silently ignore
8837 sections whose names begin with "X-". This may be used to maintain
8838 application-specific extension sections in unit files.
8840 * machined gained a new API to query the IP addresses of
8841 registered containers. "machinectl status" has been updated
8842 to show these addresses in its output.
8844 * A new call sd_uid_get_display() has been added to the
8845 sd-login APIs for querying the "primary" session of a
8846 user. The "primary" session of the user is elected from the
8847 user's sessions and generally a graphical session is
8848 preferred over a text one.
8850 * A minimal systemd-resolved daemon has been added. It
8851 currently simply acts as a companion to systemd-networkd and
8852 manages resolv.conf based on per-interface DNS
8853 configuration, possibly supplied via DHCP. In the long run
8854 we hope to extend this into a local DNSSEC enabled DNS and
8857 * The systemd-networkd-wait-online tool is now enabled by
8858 default. It will delay network-online.target until a network
8859 connection has been configured. The tool primarily integrates
8860 with networkd, but will also make a best effort to make sense
8861 of network configuration performed in some other way.
8863 * Two new service options StartupCPUShares= and
8864 StartupBlockIOWeight= have been added that work similarly to
8865 CPUShares= and BlockIOWeight= however only apply during
8866 system startup. This is useful to prioritize certain services
8867 differently during bootup than during normal runtime.
8869 * hostnamed has been changed to prefer the statically
8870 configured hostname in /etc/hostname (unless set to
8871 'localhost' or empty) over any dynamic one supplied by
8872 dhcp. With this change, the rules for picking the hostname
8873 match more closely the rules of other configuration settings
8874 where the local administrator's configuration in /etc always
8875 overrides any other settings.
8877 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
8878 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
8879 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
8880 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
8881 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
8882 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
8883 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
8884 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
8885 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
8886 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
8887 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
8888 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
8889 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
8890 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
8891 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
8892 Lindskog, WaLyong Cho, Will Woods, Zbigniew
8895 — Beijing, 2014-05-28
8899 * When restoring the screen brightness at boot, stay away from
8900 the darkest setting or from the lowest 5% of the available
8901 range, depending on which is the larger value of both. This
8902 should effectively protect the user from rebooting into a
8903 black screen, should the brightness have been set to minimum
8906 * sd-login gained a new sd_machine_get_class() call to
8907 determine the class ("vm" or "container") of a machine
8908 registered with machined.
8910 * sd-login gained new calls
8911 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
8912 to query the identity of the peer of a local AF_UNIX
8913 connection. They operate similarly to their sd_pid_get_xyz()
8916 * PID 1 will now maintain a system-wide system state engine
8917 with the states "starting", "running", "degraded",
8918 "maintenance", "stopping". These states are bound to system
8919 startup, normal runtime, runtime with at least one failed
8920 service, rescue/emergency mode and system shutdown. This
8921 state is shown in the "systemctl status" output when no unit
8922 name is passed. It is useful to determine system state, in
8923 particularly when doing so for many systems or containers at
8926 * A new command "list-machines" has been added to "systemctl"
8927 that lists all local OS containers and shows their system
8928 state (see above), if systemd runs inside of them.
8930 * systemctl gained a new "-r" switch to recursively enumerate
8931 units on all local containers, when used with the
8932 "list-unit" command (which is the default one that is
8933 executed when no parameters are specified).
8935 * The GPT automatic partition discovery logic will now honour
8936 two GPT partition flags: one may be set on a partition to
8937 cause it to be mounted read-only, and the other may be set
8938 on a partition to ignore it during automatic discovery.
8940 * Two new GPT type UUIDs have been added for automatic root
8941 partition discovery, for 32-bit and 64-bit ARM. This is not
8942 particularly useful for discovering the root directory on
8943 these architectures during bare-metal boots (since UEFI is
8944 not common there), but still very useful to allow booting of
8945 ARM disk images in nspawn with the -i option.
8947 * MAC addresses of interfaces created with nspawn's
8948 --network-interface= switch will now be generated from the
8949 machine name, and thus be stable between multiple invocations
8952 * logind will now automatically remove all IPC objects owned
8953 by a user if she or he fully logs out. This makes sure that
8954 users who are logged out cannot continue to consume IPC
8955 resources. This covers SysV memory, semaphores and message
8956 queues as well as POSIX shared memory and message
8957 queues. Traditionally, SysV and POSIX IPC had no lifecycle
8958 limits. With this functionality, that is corrected. This may
8959 be turned off by using the RemoveIPC= switch of logind.conf.
8961 * The systemd-machine-id-setup and tmpfiles tools gained a
8962 --root= switch to operate on a specific root directory,
8965 * journald can now forward logged messages to the TTYs of all
8966 logged in users ("wall"). This is the default for all
8967 emergency messages now.
8969 * A new tool systemd-journal-remote has been added to stream
8970 journal log messages across the network.
8972 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
8973 controller trees are mounted into it. Note that the
8974 directories mounted beneath it are not read-only. This is a
8975 security measure and is particularly useful because glibc
8976 actually includes a search logic to pick any tmpfs it can
8977 find to implement shm_open() if /dev/shm is not available
8978 (which it might very well be in namespaced setups).
8980 * machinectl gained a new "poweroff" command to cleanly power
8981 down a local OS container.
8983 * The PrivateDevices= unit file setting will now also drop the
8984 CAP_MKNOD capability from the capability bound set, and
8985 imply DevicePolicy=closed.
8987 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
8988 comprehensively on all long-running systemd services where
8989 this is appropriate.
8991 * systemd-udevd will now run in a disassociated mount
8992 namespace. To mount directories from udev rules, make sure to
8993 pull in mount units via SYSTEMD_WANTS properties.
8995 * The kdbus support gained support for uploading policy into
8996 the kernel. sd-bus gained support for creating "monitoring"
8997 connections that can eavesdrop into all bus communication
8998 for debugging purposes.
9000 * Timestamps may now be specified in seconds since the UNIX
9001 epoch Jan 1st, 1970 by specifying "@" followed by the value
9004 * Native tcpwrap support in systemd has been removed. tcpwrap
9005 is old code, not really maintained anymore and has serious
9006 shortcomings, and better options such as firewalls
9007 exist. For setups that require tcpwrap usage, please
9008 consider invoking your socket-activated service via tcpd,
9009 like on traditional inetd.
9011 * A new system.conf configuration option
9012 DefaultTimerAccuracySec= has been added that controls the
9013 default AccuracySec= setting of .timer units.
9015 * Timer units gained a new WakeSystem= switch. If enabled,
9016 timers configured this way will cause the system to resume
9017 from system suspend (if the system supports that, which most
9020 * Timer units gained a new Persistent= switch. If enabled,
9021 timers configured this way will save to disk when they have
9022 been last triggered. This information is then used on next
9023 reboot to possible execute overdue timer events, that
9024 could not take place because the system was powered off.
9025 This enables simple anacron-like behaviour for timer units.
9027 * systemctl's "list-timers" will now also list the time a
9028 timer unit was last triggered in addition to the next time
9029 it will be triggered.
9031 * systemd-networkd will now assign predictable IPv4LL
9032 addresses to its local interfaces.
9034 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
9035 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
9036 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
9037 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
9038 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
9039 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
9040 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
9041 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
9044 — Berlin, 2014-03-25
9048 * A new unit file setting RestrictAddressFamilies= has been
9049 added to restrict which socket address families unit
9050 processes gain access to. This takes address family names
9051 like "AF_INET" or "AF_UNIX", and is useful to minimize the
9052 attack surface of services via exotic protocol stacks. This
9053 is built on seccomp system call filters.
9055 * Two new unit file settings RuntimeDirectory= and
9056 RuntimeDirectoryMode= have been added that may be used to
9057 manage a per-daemon runtime directories below /run. This is
9058 an alternative for setting up directory permissions with
9059 tmpfiles snippets, and has the advantage that the runtime
9060 directory's lifetime is bound to the daemon runtime and that
9061 the daemon starts up with an empty directory each time. This
9062 is particularly useful when writing services that drop
9063 privileges using the User= or Group= setting.
9065 * The DeviceAllow= unit setting now supports globbing for
9066 matching against device group names.
9068 * The systemd configuration file system.conf gained new
9069 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
9070 DefaultMemoryAccounting= to globally turn on/off accounting
9071 for specific resources (cgroups) for all units. These
9072 settings may still be overridden individually in each unit
9075 * systemd-gpt-auto-generator is now able to discover /srv and
9076 root partitions in addition to /home and swap partitions. It
9077 also supports LUKS-encrypted partitions now. With this in
9078 place, automatic discovery of partitions to mount following
9079 the Discoverable Partitions Specification
9080 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
9081 is now a lot more complete. This allows booting without
9082 /etc/fstab and without root= on the kernel command line on
9083 systems prepared appropriately.
9085 * systemd-nspawn gained a new --image= switch which allows
9086 booting up disk images and Linux installations on any block
9087 device that follow the Discoverable Partitions Specification
9088 (see above). This means that installations made with
9089 appropriately updated installers may now be started and
9090 deployed using container managers, completely
9091 unmodified. (We hope that libvirt-lxc will add support for
9092 this feature soon, too.)
9094 * systemd-nspawn gained a new --network-macvlan= setting to
9095 set up a private macvlan interface for the
9096 container. Similarly, systemd-networkd gained a new
9097 Kind=macvlan setting in .netdev files.
9099 * systemd-networkd now supports configuring local addresses
9102 * A new tool systemd-network-wait-online has been added to
9103 synchronously wait for network connectivity using
9106 * The sd-bus.h bus API gained a new sd_bus_track object for
9107 tracking the lifecycle of bus peers. Note that sd-bus.h is
9108 still not a public API though (unless you specify
9109 --enable-kdbus on the configure command line, which however
9110 voids your warranty and you get no API stability guarantee).
9112 * The $XDG_RUNTIME_DIR runtime directories for each user are
9113 now individual tmpfs instances, which has the benefit of
9114 introducing separate pools for each user, with individual
9115 size limits, and thus making sure that unprivileged clients
9116 can no longer negatively impact the system or other users by
9117 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
9118 RuntimeDirectorySize= has been introduced that allows
9119 controlling the default size limit for all users. It
9120 defaults to 10% of the available physical memory. This is no
9121 replacement for quotas on tmpfs though (which the kernel
9122 still does not support), as /dev/shm and /tmp are still
9123 shared resources used by both the system and unprivileged
9126 * logind will now automatically turn off automatic suspending
9127 on laptop lid close when more than one display is
9128 connected. This was previously expected to be implemented
9129 individually in desktop environments (such as GNOME),
9130 however has been added to logind now, in order to fix a
9131 boot-time race where a desktop environment might not have
9132 been started yet and thus not been able to take an inhibitor
9133 lock at the time where logind already suspends the system
9134 due to a closed lid.
9136 * logind will now wait at least 30s after each system
9137 suspend/resume cycle, and 3min after system boot before
9138 suspending the system due to a closed laptop lid. This
9139 should give USB docking stations and similar enough time to
9140 be probed and configured after system resume and boot in
9141 order to then act as suspend blocker.
9143 * systemd-run gained a new --property= setting which allows
9144 initialization of resource control properties (and others)
9145 for the created scope or service unit. Example: "systemd-run
9146 --property=BlockIOWeight=10 updatedb" may be used to run
9147 updatedb at a low block IO scheduling weight.
9149 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
9150 now also work in --scope mode.
9152 * When systemd is compiled with kdbus support, basic support
9153 for enforced policies is now in place. (Note that enabling
9154 kdbus still voids your warranty and no API compatibility
9157 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
9158 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
9159 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
9160 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
9161 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
9162 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
9163 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
9164 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
9165 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
9166 Zbigniew Jędrzejewski-Szmek
9168 — Berlin, 2014-03-12
9172 * systemd will now relabel /dev after loading the SMACK policy
9173 according to SMACK rules.
9175 * A new unit file option AppArmorProfile= has been added to
9176 set the AppArmor profile for the processes of a unit.
9178 * A new condition check ConditionArchitecture= has been added
9179 to conditionalize units based on the system architecture, as
9180 reported by uname()'s "machine" field.
9182 * systemd-networkd now supports matching on the system
9183 virtualization, architecture, kernel command line, hostname
9186 * logind is now a lot more aggressive when suspending the
9187 machine due to a closed laptop lid. Instead of acting only
9188 on the lid close action, it will continuously watch the lid
9189 status and act on it. This is useful for laptops where the
9190 power button is on the outside of the chassis so that it can
9191 be reached without opening the lid (such as the Lenovo
9192 Yoga). On those machines, logind will now immediately
9193 re-suspend the machine if the power button has been
9194 accidentally pressed while the laptop was suspended and in a
9195 backpack or similar.
9197 * logind will now watch SW_DOCK switches and inhibit reaction
9198 to the lid switch if it is pressed. This means that logind
9199 will not suspend the machine anymore if the lid is closed
9200 and the system is docked, if the laptop supports SW_DOCK
9201 notifications via the input layer. Note that ACPI docking
9202 stations do not generate this currently. Also note that this
9203 logic is usually not fully sufficient and Desktop
9204 Environments should take a lid switch inhibitor lock when an
9205 external display is connected, as systemd will not watch
9208 * nspawn will now make use of the devices cgroup controller by
9209 default, and only permit creation of and access to the usual
9210 API device nodes like /dev/null or /dev/random, as well as
9211 access to (but not creation of) the pty devices.
9213 * We will now ship a default .network file for
9214 systemd-networkd that automatically configures DHCP for
9215 network interfaces created by nspawn's --network-veth or
9216 --network-bridge= switches.
9218 * systemd will now understand the usual M, K, G, T suffixes
9219 according to SI conventions (i.e. to the base 1000) when
9220 referring to throughput and hardware metrics. It will stay
9221 with IEC conventions (i.e. to the base 1024) for software
9222 metrics, according to what is customary according to
9223 Wikipedia. We explicitly document which base applies for
9224 each configuration option.
9226 * The DeviceAllow= setting in unit files now supports a syntax to
9227 allow-list an entire group of devices node majors at once, based on
9228 the /proc/devices listing. For example, with the string "char-pts",
9229 it is now possible to allow-list all current and future pseudo-TTYs
9232 * sd-event learned a new "post" event source. Event sources of
9233 this type are triggered by the dispatching of any event
9234 source of a type that is not "post". This is useful for
9235 implementing clean-up and check event sources that are
9236 triggered by other work being done in the program.
9238 * systemd-networkd is no longer statically enabled, but uses
9239 the usual [Install] sections so that it can be
9240 enabled/disabled using systemctl. It still is enabled by
9243 * When creating a veth interface pair with systemd-nspawn, the
9244 host side will now be prefixed with "vb-" if
9245 --network-bridge= is used, and with "ve-" if --network-veth
9246 is used. This way, it is easy to distinguish these cases on
9247 the host, for example to apply different configuration to
9248 them with systemd-networkd.
9250 * The compatibility libraries for libsystemd-journal.so,
9251 libsystem-id128.so, libsystemd-login.so and
9252 libsystemd-daemon.so do not make use of IFUNC
9253 anymore. Instead, we now build libsystemd.so multiple times
9254 under these alternative names. This means that the footprint
9255 is drastically increased, but given that these are
9256 transitional compatibility libraries, this should not matter
9257 much. This change has been made necessary to support the ARM
9258 platform for these compatibility libraries, as the ARM
9259 toolchain is not really at the same level as the toolchain
9260 for other architectures like x86 and does not support
9261 IFUNC. Please make sure to use --enable-compat-libs only
9262 during a transitional period!
9264 * The .include syntax has been deprecated and is not documented
9265 anymore. Drop-in files in .d directories should be used instead.
9267 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
9268 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
9269 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
9270 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
9271 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
9272 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
9273 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
9274 Zbigniew Jędrzejewski-Szmek
9276 — Berlin, 2014-02-24
9280 * A new component "systemd-networkd" has been added that can
9281 be used to configure local network interfaces statically or
9282 via DHCP. It is capable of bringing up bridges, VLANs, and
9283 bonding. Currently, no hook-ups for interactive network
9284 configuration are provided. Use this for your initrd,
9285 container, embedded, or server setup if you need a simple,
9286 yet powerful, network configuration solution. This
9287 configuration subsystem is quite nifty, as it allows wildcard
9288 hotplug matching in interfaces. For example, with a single
9289 configuration snippet, you can configure that all Ethernet
9290 interfaces showing up are automatically added to a bridge,
9291 or similar. It supports link-sensing and more.
9293 * A new tool "systemd-socket-proxyd" has been added which can
9294 act as a bidirectional proxy for TCP sockets. This is
9295 useful for adding socket activation support to services that
9296 do not actually support socket activation, including virtual
9297 machines and the like.
9299 * Add a new tool to save/restore rfkill state on
9302 * Save/restore state of keyboard backlights in addition to
9303 display backlights on shutdown/boot.
9305 * udev learned a new SECLABEL{} construct to label device
9306 nodes with a specific security label when they appear. For
9307 now, only SECLABEL{selinux} is supported, but the syntax is
9308 prepared for additional security frameworks.
9310 * udev gained a new scheme to configure link-level attributes
9311 from files in /etc/systemd/network/*.link. These files can
9312 match against MAC address, device path, driver name and type,
9313 and will apply attributes like the naming policy, link speed,
9314 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
9315 address assignment policy (randomized, …).
9317 * The configuration of network interface naming rules for
9318 "permanent interface names" has changed: a new NamePolicy=
9319 setting in the [Link] section of .link files determines the
9320 priority of possible naming schemes (onboard, slot, MAC,
9321 path). The default value of this setting is determined by
9322 /usr/lib/net/links/99-default.link. Old
9323 80-net-name-slot.rules udev configuration file has been
9324 removed, so local configuration overriding this file should
9325 be adapted to override 99-default.link instead.
9327 * When the User= switch is used in a unit file, also
9328 initialize $SHELL= based on the user database entry.
9330 * systemd no longer depends on libdbus. All communication is
9331 now done with sd-bus, systemd's low-level bus library
9334 * kdbus support has been added to PID 1 itself. When kdbus is
9335 enabled, this causes PID 1 to set up the system bus and
9336 enable support for a new ".busname" unit type that
9337 encapsulates bus name activation on kdbus. It works a little
9338 bit like ".socket" units, except for bus names. A new
9339 generator has been added that converts classic dbus1 service
9340 activation files automatically into native systemd .busname
9343 * sd-bus: add a light-weight vtable implementation that allows
9344 defining objects on the bus with a simple static const
9345 vtable array of its methods, signals and properties.
9347 * systemd will not generate or install static dbus
9348 introspection data anymore to /usr/share/dbus-1/interfaces,
9349 as the precise format of these files is unclear, and
9350 nothing makes use of it.
9352 * A proxy daemon is now provided to proxy clients connecting
9353 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
9354 compatibility with classic D-Bus.
9356 * A bus driver implementation has been added that supports the
9357 classic D-Bus bus driver calls on kdbus, also for
9358 compatibility purposes.
9360 * A new API "sd-event.h" has been added that implements a
9361 minimal event loop API built around epoll. It provides a
9362 couple of features that direct epoll usage is lacking:
9363 prioritization of events, scales to large numbers of timer
9364 events, per-event timer slack (accuracy), system-wide
9365 coalescing of timer events, exit handlers, watchdog
9366 supervision support using systemd's sd_notify() API, child
9369 * A new API "sd-rntl.h" has been added that provides an API
9370 around the route netlink interface of the kernel, similar in
9371 style to "sd-bus.h".
9373 * A new API "sd-dhcp-client.h" has been added that provides a
9374 small DHCPv4 client-side implementation. This is used by
9377 * There is a new kernel command line option
9378 "systemd.restore_state=0|1". When set to "0", none of the
9379 systemd tools will restore saved runtime state to hardware
9380 devices. More specifically, the rfkill and backlight states
9383 * The FsckPassNo= compatibility option in mount/service units
9384 has been removed. The fstab generator will now add the
9385 necessary dependencies automatically, and does not require
9386 PID1's support for that anymore.
9388 * journalctl gained a new switch, --list-boots, that lists
9389 recent boots with their times and boot IDs.
9391 * The various tools like systemctl, loginctl, timedatectl,
9392 busctl, systemd-run, … have gained a new switch "-M" to
9393 connect to a specific, local OS container (as direct
9394 connection, without requiring SSH). This works on any
9395 container that is registered with machined, such as those
9396 created by libvirt-lxc or nspawn.
9398 * systemd-run and systemd-analyze also gained support for "-H"
9399 to connect to remote hosts via SSH. This is particularly
9400 useful for systemd-run because it enables queuing of jobs
9401 onto remote systems.
9403 * machinectl gained a new command "login" to open a getty
9404 login in any local container. This works with any container
9405 that is registered with machined (such as those created by
9406 libvirt-lxc or nspawn), and which runs systemd inside.
9408 * machinectl gained a new "reboot" command that may be used to
9409 trigger a reboot on a specific container that is registered
9410 with machined. This works on any container that runs an init
9411 system of some kind.
9413 * systemctl gained a new "list-timers" command to print a nice
9414 listing of installed timer units with the times they elapse
9417 * Alternative reboot() parameters may now be specified on the
9418 "systemctl reboot" command line and are passed to the
9419 reboot() system call.
9421 * systemctl gained a new --job-mode= switch to configure the
9422 mode to queue a job with. This is a more generic version of
9423 --fail, --irreversible, and --ignore-dependencies, which are
9424 still available but not advertised anymore.
9426 * /etc/systemd/system.conf gained new settings to configure
9427 various default timeouts of units, as well as the default
9428 start limit interval and burst. These may still be overridden
9431 * PID1 will now export on the bus profile data of the security
9432 policy upload process (such as the SELinux policy upload to
9435 * journald: when forwarding logs to the console, include
9436 timestamps (following the setting in
9437 /sys/module/printk/parameters/time).
9439 * OnCalendar= in timer units now understands the special
9440 strings "yearly" and "annually". (Both are equivalent)
9442 * The accuracy of timer units is now configurable with the new
9443 AccuracySec= setting. It defaults to 1min.
9445 * A new dependency type JoinsNamespaceOf= has been added that
9446 allows running two services within the same /tmp and network
9447 namespace, if PrivateNetwork= or PrivateTmp= are used.
9449 * A new command "cat" has been added to systemctl. It outputs
9450 the original unit file of a unit, and concatenates the
9451 contents of additional "drop-in" unit file snippets, so that
9452 the full configuration is shown.
9454 * systemctl now supports globbing on the various "list-xyz"
9455 commands, like "list-units" or "list-sockets", as well as on
9456 those commands which take multiple unit names.
9458 * journalctl's --unit= switch gained support for globbing.
9460 * All systemd daemons now make use of the watchdog logic so
9461 that systemd automatically notices when they hang.
9463 * If the $container_ttys environment variable is set,
9464 getty-generator will automatically spawn a getty for each
9465 listed tty. This is useful for container managers to request
9466 login gettys to be spawned on as many ttys as needed.
9468 * %h, %s, %U specifier support is not available anymore when
9469 used in unit files for PID 1. This is because NSS calls are
9470 not safe from PID 1. They stay available for --user
9471 instances of systemd, and as special case for the root user.
9473 * loginctl gained a new "--no-legend" switch to turn off output
9476 * The "sd-login.h" API gained three new calls:
9477 sd_session_is_remote(), sd_session_get_remote_user(),
9478 sd_session_get_remote_host() to query information about
9481 * The udev hardware database now also carries vendor/product
9482 information of SDIO devices.
9484 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
9485 determine whether watchdog notifications are requested by
9488 * Socket-activated per-connection services now include a
9489 short description of the connection parameters in the
9492 * tmpfiles gained a new "--boot" option. When this is not used,
9493 only lines where the command character is not suffixed with
9494 "!" are executed. When this option is specified, those
9495 options are executed too. This partitions tmpfiles
9496 directives into those that can be safely executed at any
9497 time, and those which should be run only at boot (for
9498 example, a line that creates /run/nologin).
9500 * A new API "sd-resolve.h" has been added which provides a simple
9501 asynchronous wrapper around glibc NSS hostname resolution
9502 calls, such as getaddrinfo(). In contrast to glibc's
9503 getaddrinfo_a(), it does not use signals. In contrast to most
9504 other asynchronous name resolution libraries, this one does
9505 not reimplement DNS, but reuses NSS, so that alternate
9506 hostname resolution systems continue to work, such as mDNS,
9507 LDAP, etc. This API is based on libasyncns, but it has been
9508 cleaned up for inclusion in systemd.
9510 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
9511 "sd-daemon.h" are no longer found in individual libraries
9512 libsystemd-journal.so, libsystemd-login.so,
9513 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
9514 merged them into a single library, libsystemd.so, which
9515 provides all symbols. The reason for this is cyclic
9516 dependencies, as these libraries tend to use each other's
9517 symbols. So far, we have managed to workaround that by linking
9518 a copy of a good part of our code into each of these
9519 libraries again and again, which, however, makes certain
9520 things hard to do, like sharing static variables. Also, it
9521 substantially increases footprint. With this change, there
9522 is only one library for the basic APIs systemd
9523 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
9524 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
9525 library as well, however are subject to the --enable-kdbus
9526 switch (see below). Note that "sd-dhcp-client.h" is not part
9527 of this library (this is because it only consumes, never
9528 provides, services of/to other APIs). To make the transition
9529 easy from the separate libraries to the unified one, we
9530 provide the --enable-compat-libs compile-time switch which
9531 will generate stub libraries that are compatible with the
9532 old ones but redirect all calls to the new one.
9534 * All of the kdbus logic and the new APIs "sd-bus.h",
9535 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
9536 and "sd-utf8.h" are compile-time optional via the
9537 "--enable-kdbus" switch, and they are not compiled in by
9538 default. To make use of kdbus, you have to explicitly enable
9539 the switch. Note however, that neither the kernel nor the
9540 userspace API for all of this is considered stable yet. We
9541 want to maintain the freedom to still change the APIs for
9542 now. By specifying this build-time switch, you acknowledge
9543 that you are aware of the instability of the current
9546 * Also, note that while kdbus is pretty much complete,
9547 it lacks one thing: proper policy support. This means you
9548 can build a fully working system with all features; however,
9549 it will be highly insecure. Policy support will be added in
9550 one of the next releases, at the same time that we will
9551 declare the APIs stable.
9553 * When the kernel command line argument "kdbus" is specified,
9554 systemd will automatically load the kdbus.ko kernel module. At
9555 this stage of development, it is only useful for testing kdbus
9556 and should not be used in production. Note: if "--enable-kdbus"
9557 is specified, and the kdbus.ko kernel module is available, and
9558 "kdbus" is added to the kernel command line, the entire system
9559 runs with kdbus instead of dbus-daemon, with the above mentioned
9560 problem of missing the system policy enforcement. Also a future
9561 version of kdbus.ko or a newer systemd will not be compatible with
9562 each other, and will unlikely be able to boot the machine if only
9563 one of them is updated.
9565 * systemctl gained a new "import-environment" command which
9566 uploads the caller's environment (or parts thereof) into the
9567 service manager so that it is inherited by services started
9568 by the manager. This is useful to upload variables like
9569 $DISPLAY into the user service manager.
9571 * A new PrivateDevices= switch has been added to service units
9572 which allows running a service with a namespaced /dev
9573 directory that does not contain any device nodes for
9574 physical devices. More specifically, it only includes devices
9575 such as /dev/null, /dev/urandom, and /dev/zero which are API
9578 * logind has been extended to support behaviour like VT
9579 switching on seats that do not support a VT. This makes
9580 multi-session available on seats that are not the first seat
9581 (seat0), and on systems where kernel support for VTs has
9582 been disabled at compile-time.
9584 * If a process holds a delay lock for system sleep or shutdown
9585 and fails to release it in time, we will now log its
9586 identity. This makes it easier to identify processes that
9587 cause slow suspends or power-offs.
9589 * When parsing /etc/crypttab, support for a new key-slot=
9590 option as supported by Debian is added. It allows indicating
9591 which LUKS slot to use on disk, speeding up key loading.
9593 * The sd_journal_sendv() API call has been checked and
9594 officially declared to be async-signal-safe so that it may
9595 be invoked from signal handlers for logging purposes.
9597 * Boot-time status output is now enabled automatically after a
9598 short timeout if boot does not progress, in order to give
9599 the user an indication what she or he is waiting for.
9601 * The boot-time output has been improved to show how much time
9602 remains until jobs expire.
9604 * The KillMode= switch in service units gained a new possible
9605 value "mixed". If set, and the unit is shut down, then the
9606 initial SIGTERM signal is sent only to the main daemon
9607 process, while the following SIGKILL signal is sent to
9608 all remaining processes of the service.
9610 * When a scope unit is registered, a new property "Controller"
9611 may be set. If set to a valid bus name, systemd will send a
9612 RequestStop() signal to this name when it would like to shut
9613 down the scope. This may be used to hook manager logic into
9614 the shutdown logic of scope units. Also, scope units may now
9615 be put in a special "abandoned" state, in which case the
9616 manager process which created them takes no further
9617 responsibilities for it.
9619 * When reading unit files, systemd will now verify
9620 the access mode of these files, and warn about certain
9621 suspicious combinations. This has been added to make it
9622 easier to track down packaging bugs where unit files are
9623 marked executable or world-writable.
9625 * systemd-nspawn gained a new "--setenv=" switch to set
9626 container-wide environment variables. The similar option in
9627 systemd-activate was renamed from "--environment=" to
9628 "--setenv=" for consistency.
9630 * systemd-nspawn has been updated to create a new kdbus domain
9631 for each container that is invoked, thus allowing each
9632 container to have its own set of system and user buses,
9633 independent of the host.
9635 * systemd-nspawn gained a new --drop-capability= switch to run
9636 the container with less capabilities than the default. Both
9637 --drop-capability= and --capability= now take the special
9638 string "all" for dropping or keeping all capabilities.
9640 * systemd-nspawn gained new switches for executing containers
9641 with specific SELinux labels set.
9643 * systemd-nspawn gained a new --quiet switch to not generate
9644 any additional output but the container's own console
9647 * systemd-nspawn gained a new --share-system switch to run a
9648 container without PID namespacing enabled.
9650 * systemd-nspawn gained a new --register= switch to control
9651 whether the container is registered with systemd-machined or
9652 not. This is useful for containers that do not run full
9653 OS images, but only specific apps.
9655 * systemd-nspawn gained a new --keep-unit which may be used
9656 when invoked as the only program from a service unit, and
9657 results in registration of the unit service itself in
9658 systemd-machined, instead of a newly opened scope unit.
9660 * systemd-nspawn gained a new --network-interface= switch for
9661 moving arbitrary interfaces to the container. The new
9662 --network-veth switch creates a virtual Ethernet connection
9663 between host and container. The new --network-bridge=
9664 switch then allows assigning the host side of this virtual
9665 Ethernet connection to a bridge device.
9667 * systemd-nspawn gained a new --personality= switch for
9668 setting the kernel personality for the container. This is
9669 useful when running a 32-bit container on a 64-bit host. A
9670 similar option Personality= is now also available for service
9673 * logind will now also track a "Desktop" identifier for each
9674 session which encodes the desktop environment of it. This is
9675 useful for desktop environments that want to identify
9676 multiple running sessions of itself easily.
9678 * A new SELinuxContext= setting for service units has been
9679 added that allows setting a specific SELinux execution
9680 context for a service.
9682 * Most systemd client tools will now honour $SYSTEMD_LESS for
9683 settings of the "less" pager. By default, these tools will
9684 override $LESS to allow certain operations to work, such as
9685 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
9686 influence this logic.
9688 * systemd's "seccomp" hook-up has been changed to make use of
9689 the libseccomp library instead of using its own
9690 implementation. This has benefits for portability among
9693 * For usage together with SystemCallFilter=, a new
9694 SystemCallErrorNumber= setting has been introduced that
9695 allows configuration of a system error number to be returned
9696 on filtered system calls, instead of immediately killing the
9697 process. Also, SystemCallArchitectures= has been added to
9698 limit access to system calls of a particular architecture
9699 (in order to turn off support for unused secondary
9700 architectures). There is also a global
9701 SystemCallArchitectures= setting in system.conf now to turn
9702 off support for non-native system calls system-wide.
9704 * systemd requires a kernel with a working name_to_handle_at(),
9705 please see the kernel config requirements in the README file.
9707 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
9708 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
9709 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
9710 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
9711 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
9712 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
9713 Elia Pinto, Florian Weimer, George McCollister, Goffredo
9714 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
9715 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
9716 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
9717 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
9718 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
9719 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
9720 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
9721 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
9722 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
9723 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
9724 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
9725 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
9726 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
9727 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
9728 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
9729 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
9730 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
9732 — Berlin, 2014-02-20
9736 * logind has gained support for facilitating privileged input
9737 and drm device access for unprivileged clients. This work is
9738 useful to allow Wayland display servers (and similar
9739 programs, such as kmscon) to run under the user's ID and
9740 access input and drm devices which are normally
9741 protected. When this is used (and the kernel is new enough)
9742 logind will "mute" IO on the file descriptors passed to
9743 Wayland as long as it is in the background and "unmute" it
9744 if it returns into the foreground. This allows secure
9745 session switching without allowing background sessions to
9746 eavesdrop on input and display data. This also introduces
9747 session switching support if VT support is turned off in the
9748 kernel, and on seats that are not seat0.
9750 * A new kernel command line option luks.options= is understood
9751 now which allows specifying LUKS options for usage for LUKS
9752 encrypted partitions specified with luks.uuid=.
9754 * tmpfiles.d(5) snippets may now use specifier expansion in
9755 path names. More specifically %m, %b, %H, %v, are now
9756 replaced by the local machine id, boot id, hostname, and
9757 kernel version number.
9759 * A new tmpfiles.d(5) command "m" has been introduced which
9760 may be used to change the owner/group/access mode of a file
9761 or directory if it exists, but do nothing if it does not.
9763 * This release removes high-level support for the
9764 MemorySoftLimit= cgroup setting. The underlying kernel
9765 cgroup attribute memory.soft_limit= is currently badly
9766 designed and likely to be removed from the kernel API in its
9767 current form, hence we should not expose it for now.
9769 * The memory.use_hierarchy cgroup attribute is now enabled for
9770 all cgroups systemd creates in the memory cgroup
9771 hierarchy. This option is likely to be come the built-in
9772 default in the kernel anyway, and the non-hierarchical mode
9773 never made much sense in the intrinsically hierarchical
9776 * A new field _SYSTEMD_SLICE= is logged along with all journal
9777 messages containing the slice a message was generated
9778 from. This is useful to allow easy per-customer filtering of
9779 logs among other things.
9781 * systemd-journald will no longer adjust the group of journal
9782 files it creates to the "systemd-journal" group. Instead we
9783 rely on the journal directory to be owned by the
9784 "systemd-journal" group, and its setgid bit set, so that the
9785 kernel file system layer will automatically enforce that
9786 journal files inherit this group assignment. The reason for
9787 this change is that we cannot allow NSS look-ups from
9788 journald which would be necessary to resolve
9789 "systemd-journal" to a numeric GID, because this might
9790 create deadlocks if NSS involves synchronous queries to
9791 other daemons (such as nscd, or sssd) which in turn are
9792 logging clients of journald and might block on it, which
9793 would then dead lock. A tmpfiles.d(5) snippet included in
9794 systemd will make sure the setgid bit and group are
9795 properly set on the journal directory if it exists on every
9796 boot. However, we recommend adjusting it manually after
9797 upgrades too (or from RPM scriptlets), so that the change is
9798 not delayed until next reboot.
9800 * Backlight and random seed files in /var/lib/ have moved into
9801 the /var/lib/systemd/ directory, in order to centralize all
9802 systemd generated files in one directory.
9804 * Boot time performance measurements (as displayed by
9805 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
9806 performance information if that's available to determine how
9807 much time BIOS and boot loader initialization required. With
9808 a sufficiently new BIOS you hence no longer need to boot
9809 with Gummiboot to get access to such information.
9811 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
9812 Cristian Rodríguez, Dave Reisner, David Herrmann, David
9813 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
9814 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
9815 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
9816 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
9817 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9819 — Berlin, 2013-10-02
9823 * The Restart= option for services now understands a new
9824 on-watchdog setting, which will restart the service
9825 automatically if the service stops sending out watchdog keep
9826 alive messages (as configured with WatchdogSec=).
9828 * The getty generator (which is responsible for bringing up a
9829 getty on configured serial consoles) will no longer only
9830 start a getty on the primary kernel console but on all
9831 others, too. This makes the order in which console= is
9832 specified on the kernel command line less important.
9834 * libsystemd-logind gained a new sd_session_get_vt() call to
9835 retrieve the VT number of a session.
9837 * If the option "tries=0" is set for an entry of /etc/crypttab
9838 its passphrase is queried indefinitely instead of any
9839 maximum number of tries.
9841 * If a service with a configure PID file terminates its PID
9842 file will now be removed automatically if it still exists
9843 afterwards. This should put an end to stale PID files.
9845 * systemd-run will now also take relative binary path names
9846 for execution and no longer insists on absolute paths.
9848 * InaccessibleDirectories= and ReadOnlyDirectories= now take
9849 paths that are optionally prefixed with "-" to indicate that
9850 it should not be considered a failure if they do not exist.
9852 * journalctl -o (and similar commands) now understands a new
9853 output mode "short-precise", it is similar to "short" but
9854 shows timestamps with usec accuracy.
9856 * The option "discard" (as known from Debian) is now
9857 synonymous to "allow-discards" in /etc/crypttab. In fact,
9858 "discard" is preferred now (since it is easier to remember
9861 * Some licensing clean-ups were made, so that more code is now
9862 LGPL-2.1 licensed than before.
9864 * A minimal tool to save/restore the display backlight
9865 brightness across reboots has been added. It will store the
9866 backlight setting as late as possible at shutdown, and
9867 restore it as early as possible during reboot.
9869 * A logic to automatically discover and enable home and swap
9870 partitions on GPT disks has been added. With this in place
9871 /etc/fstab becomes optional for many setups as systemd can
9872 discover certain partitions located on the root disk
9873 automatically. Home partitions are recognized under their
9874 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
9875 partitions are recognized under their GPT type ID
9876 0657fd6da4ab43c484e50933c84b4f4f.
9878 * systemd will no longer pass any environment from the kernel
9879 or initrd to system services. If you want to set an
9880 environment for all services, do so via the kernel command
9881 line systemd.setenv= assignment.
9883 * The systemd-sysctl tool no longer natively reads the file
9884 /etc/sysctl.conf. If desired, the file should be symlinked
9885 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
9886 legacy support by a symlink rather than built-in code, it
9887 also makes the otherwise hidden order of application of the
9888 different files visible. (Note that this partly reverts to a
9889 pre-198 application order of sysctl knobs!)
9891 * The "systemctl set-log-level" and "systemctl dump" commands
9892 have been moved to systemd-analyze.
9894 * systemd-run learned the new --remain-after-exit switch,
9895 which causes the scope unit not to be cleaned up
9896 automatically after the process terminated.
9898 * tmpfiles learned a new --exclude-prefix= switch to exclude
9899 certain paths from operation.
9901 * journald will now automatically flush all messages to disk
9902 as soon as a message at the log level CRIT, ALERT or EMERG
9905 Contributions from: Andrew Cook, Brandon Philips, Christian
9906 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
9907 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
9908 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
9909 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
9910 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
9911 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
9912 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
9913 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
9914 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
9915 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
9916 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
9917 William Giokas, Zbigniew Jędrzejewski-Szmek
9919 — Berlin, 2013-09-13
9923 * The documentation has been updated to cover the various new
9924 concepts introduced with 205.
9926 * Unit files now understand the new %v specifier which
9927 resolves to the kernel version string as returned by "uname
9930 * systemctl now supports filtering the unit list output by
9931 load state, active state and sub state, using the new
9934 * "systemctl status" will now show the results of the
9935 condition checks (like ConditionPathExists= and similar) of
9936 the last start attempts of the unit. They are also logged to
9939 * "journalctl -b" may now be used to look for boot output of a
9940 specific boot. Try "journalctl -b -1" for the previous boot,
9941 but the syntax is substantially more powerful.
9943 * "journalctl --show-cursor" has been added which prints the
9944 cursor string the last shown log line. This may then be used
9945 with the new "journalctl --after-cursor=" switch to continue
9946 browsing logs from that point on.
9948 * "journalctl --force" may now be used to force regeneration
9951 * Creation of "dead" device nodes has been moved from udev
9952 into kmod and tmpfiles. Previously, udev would read the kmod
9953 databases to pre-generate dead device nodes based on meta
9954 information contained in kernel modules, so that these would
9955 be auto-loaded on access rather then at boot. As this
9956 does not really have much to do with the exposing actual
9957 kernel devices to userspace this has always been slightly
9958 alien in the udev codebase. Following the new scheme kmod
9959 will now generate a runtime snippet for tmpfiles from the
9960 module meta information and it now is tmpfiles' job to the
9961 create the nodes. This also allows overriding access and
9962 other parameters for the nodes using the usual tmpfiles
9963 facilities. As side effect this allows us to remove the
9964 CAP_SYS_MKNOD capability bit from udevd entirely.
9966 * logind's device ACLs may now be applied to these "dead"
9967 devices nodes too, thus finally allowing managed access to
9968 devices such as /dev/snd/sequencer without loading the
9969 backing module right-away.
9971 * A new RPM macro has been added that may be used to apply
9972 tmpfiles configuration during package installation.
9974 * systemd-detect-virt and ConditionVirtualization= now can
9975 detect User-Mode-Linux machines (UML).
9977 * journald will now implicitly log the effective capabilities
9978 set of processes in the message metadata.
9980 * systemd-cryptsetup has gained support for TrueCrypt volumes.
9982 * The initrd interface has been simplified (more specifically,
9983 support for passing performance data via environment
9984 variables and fsck results via files in /run has been
9985 removed). These features were non-essential, and are
9986 nowadays available in a much nicer way by having systemd in
9987 the initrd serialize its state and have the hosts systemd
9988 deserialize it again.
9990 * The udev "keymap" data files and tools to apply keyboard
9991 specific mappings of scan to key codes, and force-release
9992 scan code lists have been entirely replaced by a udev
9993 "keyboard" builtin and a hwdb data file.
9995 * systemd will now honour the kernel's "quiet" command line
9996 argument also during late shutdown, resulting in a
9997 completely silent shutdown when used.
9999 * There's now an option to control the SO_REUSEPORT socket
10000 option in .socket units.
10002 * Instance units will now automatically get a per-template
10003 subslice of system.slice unless something else is explicitly
10004 configured. For example, instances of sshd@.service will now
10005 implicitly be placed in system-sshd.slice rather than
10006 system.slice as before.
10008 * Test coverage support may now be enabled at build time.
10010 Contributions from: Dave Reisner, Frederic Crozat, Harald
10011 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
10012 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
10013 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
10014 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
10015 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
10016 Giokas, Zbigniew Jędrzejewski-Szmek
10018 — Berlin, 2013-07-23
10022 * Two new unit types have been introduced:
10024 Scope units are very similar to service units, however, are
10025 created out of pre-existing processes — instead of PID 1
10026 forking off the processes. By using scope units it is
10027 possible for system services and applications to group their
10028 own child processes (worker processes) in a powerful way
10029 which then maybe used to organize them, or kill them
10030 together, or apply resource limits on them.
10032 Slice units may be used to partition system resources in an
10033 hierarchical fashion and then assign other units to them. By
10034 default there are now three slices: system.slice (for all
10035 system services), user.slice (for all user sessions),
10036 machine.slice (for VMs and containers).
10038 Slices and scopes have been introduced primarily in
10039 context of the work to move cgroup handling to a
10040 single-writer scheme, where only PID 1
10041 creates/removes/manages cgroups.
10043 * There's a new concept of "transient" units. In contrast to
10044 normal units these units are created via an API at runtime,
10045 not from configuration from disk. More specifically this
10046 means it is now possible to run arbitrary programs as
10047 independent services, with all execution parameters passed
10048 in via bus APIs rather than read from disk. Transient units
10049 make systemd substantially more dynamic then it ever was,
10050 and useful as a general batch manager.
10052 * logind has been updated to make use of scope and slice units
10053 for managing user sessions. As a user logs in he will get
10054 his own private slice unit, to which all sessions are added
10055 as scope units. We also added support for automatically
10056 adding an instance of user@.service for the user into the
10057 slice. Effectively logind will no longer create cgroup
10058 hierarchies on its own now, it will defer entirely to PID 1
10059 for this by means of scope, service and slice units. Since
10060 user sessions this way become entities managed by PID 1
10061 the output of "systemctl" is now a lot more comprehensive.
10063 * A new mini-daemon "systemd-machined" has been added which
10064 may be used by virtualization managers to register local
10065 VMs/containers. nspawn has been updated accordingly, and
10066 libvirt will be updated shortly. machined will collect a bit
10067 of meta information about the VMs/containers, and assign
10068 them their own scope unit (see above). The collected
10069 meta-data is then made available via the "machinectl" tool,
10070 and exposed in "ps" and similar tools. machined/machinectl
10071 is compile-time optional.
10073 * As discussed earlier, the low-level cgroup configuration
10074 options ControlGroup=, ControlGroupModify=,
10075 ControlGroupPersistent=, ControlGroupAttribute= have been
10076 removed. Please use high-level attribute settings instead as
10077 well as slice units.
10079 * A new bus call SetUnitProperties() has been added to alter
10080 various runtime parameters of a unit. This is primarily
10081 useful to alter cgroup parameters dynamically in a nice way,
10082 but will be extended later on to make more properties
10083 modifiable at runtime. systemctl gained a new set-properties
10084 command that wraps this call.
10086 * A new tool "systemd-run" has been added which can be used to
10087 run arbitrary command lines as transient services or scopes,
10088 while configuring a number of settings via the command
10089 line. This tool is currently very basic, however already
10090 very useful. We plan to extend this tool to even allow
10091 queuing of execution jobs with time triggers from the
10092 command line, similar in fashion to "at".
10094 * nspawn will now inform the user explicitly that kernels with
10095 audit enabled break containers, and suggest the user to turn
10098 * Support for detecting the IMA and AppArmor security
10099 frameworks with ConditionSecurity= has been added.
10101 * journalctl gained a new "-k" switch for showing only kernel
10102 messages, mimicking dmesg output; in addition to "--user"
10103 and "--system" switches for showing only user's own logs
10106 * systemd-delta can now show information about drop-in
10107 snippets extending unit files.
10109 * libsystemd-bus has been substantially updated but is still
10110 not available as public API.
10112 * systemd will now look for the "debug" argument on the kernel
10113 command line and enable debug logging, similar to what
10114 "systemd.log_level=debug" already did before.
10116 * "systemctl set-default", "systemctl get-default" has been
10117 added to configure the default.target symlink, which
10118 controls what to boot into by default.
10120 * "systemctl set-log-level" has been added as a convenient
10121 way to raise and lower systemd logging threshold.
10123 * "systemd-analyze plot" will now show the time the various
10124 generators needed for execution, as well as information
10125 about the unit file loading.
10127 * libsystemd-journal gained a new sd_journal_open_files() call
10128 for opening specific journal files. journactl also gained a
10129 new switch to expose this new functionality. Previously we
10130 only supported opening all files from a directory, or all
10131 files from the system, as opening individual files only is
10132 racy due to journal file rotation.
10134 * systemd gained the new DefaultEnvironment= setting in
10135 /etc/systemd/system.conf to set environment variables for
10138 * If a privileged process logs a journal message with the
10139 OBJECT_PID= field set, then journald will automatically
10140 augment this with additional OBJECT_UID=, OBJECT_GID=,
10141 OBJECT_COMM=, OBJECT_EXE=, … fields. This is useful if
10142 system services want to log events about specific client
10143 processes. journactl/systemctl has been updated to make use
10144 of this information if all log messages regarding a specific
10147 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
10148 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
10149 Reisner, David Coppa, David King, David Strauss, Eelco
10150 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
10151 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
10152 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
10153 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
10154 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
10155 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
10156 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
10157 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
10158 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
10159 Łukasz Stelmach, 장동준
10163 * The Python bindings gained some minimal support for the APIs
10164 exposed by libsystemd-logind.
10166 * ConditionSecurity= gained support for detecting SMACK. Since
10167 this condition already supports SELinux and AppArmor we only
10168 miss IMA for this. Patches welcome!
10170 Contributions from: Karol Lewandowski, Lennart Poettering,
10171 Zbigniew Jędrzejewski-Szmek
10175 * systemd-nspawn will now create /etc/resolv.conf if
10176 necessary, before bind-mounting the host's file onto it.
10178 * systemd-nspawn will now store meta information about a
10179 container on the container's cgroup as extended attribute
10180 fields, including the root directory.
10182 * The cgroup hierarchy has been reworked in many ways. All
10183 objects any of the components systemd creates in the cgroup
10184 tree are now suffixed. More specifically, user sessions are
10185 now placed in cgroups suffixed with ".session", users in
10186 cgroups suffixed with ".user", and nspawn containers in
10187 cgroups suffixed with ".nspawn". Furthermore, all cgroup
10188 names are now escaped in a simple scheme to avoid collision
10189 of userspace object names with kernel filenames. This work
10190 is preparation for making these objects relocatable in the
10191 cgroup tree, in order to allow easy resource partitioning of
10192 these objects without causing naming conflicts.
10194 * systemctl list-dependencies gained the new switches
10195 --plain, --reverse, --after and --before.
10197 * systemd-inhibit now shows the process name of processes that
10198 have taken an inhibitor lock.
10200 * nss-myhostname will now also resolve "localhost"
10201 implicitly. This makes /etc/hosts an optional file and
10202 nicely handles that on IPv6 ::1 maps to both "localhost" and
10203 the local hostname.
10205 * libsystemd-logind.so gained a new call
10206 sd_get_machine_names() to enumerate running containers and
10207 VMs (currently only supported by very new libvirt and
10208 nspawn). sd_login_monitor can now be used to watch
10209 VMs/containers coming and going.
10211 * .include is not allowed recursively anymore, and only in
10212 unit files. Usually it is better to use drop-in snippets in
10213 .d/*.conf anyway, as introduced with systemd 198.
10215 * systemd-analyze gained a new "critical-chain" command that
10216 determines the slowest chain of units run during system
10217 boot-up. It is very useful for tracking down where
10218 optimizing boot time is the most beneficial.
10220 * systemd will no longer allow manipulating service paths in
10221 the name=systemd:/system cgroup tree using ControlGroup= in
10222 units. (But is still fine with it in all other dirs.)
10224 * There's a new systemd-nspawn@.service service file that may
10225 be used to easily run nspawn containers as system
10226 services. With the container's root directory in
10227 /var/lib/container/foobar it is now sufficient to run
10228 "systemctl start systemd-nspawn@foobar.service" to boot it.
10230 * systemd-cgls gained a new parameter "--machine" to list only
10231 the processes within a certain container.
10233 * ConditionSecurity= now can check for "apparmor". We still
10234 are lacking checks for SMACK and IMA for this condition
10235 check though. Patches welcome!
10237 * A new configuration file /etc/systemd/sleep.conf has been
10238 added that may be used to configure which kernel operation
10239 systemd is supposed to execute when "suspend", "hibernate"
10240 or "hybrid-sleep" is requested. This makes the new kernel
10241 "freeze" state accessible to the user.
10243 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
10244 the passed argument if applicable.
10246 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
10247 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
10248 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
10249 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
10250 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
10251 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
10252 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
10257 * The output of 'systemctl list-jobs' got some polishing. The
10258 '--type=' argument may now be passed more than once. A new
10259 command 'systemctl list-sockets' has been added which shows
10260 a list of kernel sockets systemd is listening on with the
10261 socket units they belong to, plus the units these socket
10264 * The experimental libsystemd-bus library got substantial
10265 updates to work in conjunction with the (also experimental)
10266 kdbus kernel project. It works well enough to exchange
10267 messages with some sophistication. Note that kdbus is not
10268 ready yet, and the library is mostly an elaborate test case
10269 for now, and not installable.
10271 * systemd gained a new unit 'systemd-static-nodes.service'
10272 that generates static device nodes earlier during boot, and
10273 can run in conjunction with udev.
10275 * libsystemd-login gained a new call sd_pid_get_user_unit()
10276 to retrieve the user systemd unit a process is running
10277 in. This is useful for systems where systemd is used as
10280 * systemd-nspawn now places all containers in the new /machine
10281 top-level cgroup directory in the name=systemd
10282 hierarchy. libvirt will soon do the same, so that we get a
10283 uniform separation of /system, /user and /machine for system
10284 services, user processes and containers/virtual
10285 machines. This new cgroup hierarchy is also useful to stick
10286 stable names to specific container instances, which can be
10287 recognized later this way (this name may be controlled
10288 via systemd-nspawn's new -M switch). libsystemd-login also
10289 gained a new call sd_pid_get_machine_name() to retrieve the
10290 name of the container/VM a specific process belongs to.
10292 * bootchart can now store its data in the journal.
10294 * libsystemd-journal gained a new call
10295 sd_journal_add_conjunction() for AND expressions to the
10296 matching logic. This can be used to express more complex
10297 logical expressions.
10299 * journactl can now take multiple --unit= and --user-unit=
10302 * The cryptsetup logic now understands the "luks.key=" kernel
10303 command line switch for specifying a file to read the
10304 decryption key from. Also, if a configured key file is not
10305 found the tool will now automatically fall back to prompting
10308 * Python systemd.journal module was updated to wrap recently
10309 added functions from libsystemd-journal. The interface was
10310 changed to bring the low level interface in s.j._Reader
10311 closer to the C API, and the high level interface in
10312 s.j.Reader was updated to wrap and convert all data about
10315 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
10316 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
10317 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
10318 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
10319 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
10320 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
10324 * journalctl --update-catalog now understands a new --root=
10325 option to operate on catalogs found in a different root
10328 * During shutdown after systemd has terminated all running
10329 services a final killing loop kills all remaining left-over
10330 processes. We will now print the name of these processes
10331 when we send SIGKILL to them, since this usually indicates a
10334 * If /etc/crypttab refers to password files stored on
10335 configured mount points automatic dependencies will now be
10336 generated to ensure the specific mount is established first
10337 before the key file is attempted to be read.
10339 * 'systemctl status' will now show information about the
10340 network sockets a socket unit is listening on.
10342 * 'systemctl status' will also shown information about any
10343 drop-in configuration file for units. (Drop-In configuration
10344 files in this context are files such as
10345 /etc/systemd/systemd/foobar.service.d/*.conf)
10347 * systemd-cgtop now optionally shows summed up CPU times of
10348 cgroups. Press '%' while running cgtop to switch between
10349 percentage and absolute mode. This is useful to determine
10350 which cgroups use up the most CPU time over the entire
10351 runtime of the system. systemd-cgtop has also been updated
10352 to be 'pipeable' for processing with further shell tools.
10354 * 'hostnamectl set-hostname' will now allow setting of FQDN
10357 * The formatting and parsing of time span values has been
10358 changed. The parser now understands fractional expressions
10359 such as "5.5h". The formatter will now output fractional
10360 expressions for all time spans under 1min, i.e. "5.123456s"
10361 rather than "5s 123ms 456us". For time spans under 1s
10362 millisecond values are shown, for those under 1ms
10363 microsecond values are shown. This should greatly improve
10364 all time-related output of systemd.
10366 * libsystemd-login and libsystemd-journal gained new
10367 functions for querying the poll() events mask and poll()
10368 timeout value for integration into arbitrary event
10371 * localectl gained the ability to list available X11 keymaps
10372 (models, layouts, variants, options).
10374 * 'systemd-analyze dot' gained the ability to filter for
10375 specific units via shell-style globs, to create smaller,
10376 more useful graphs. I.e. it is now possible to create simple
10377 graphs of all the dependencies between only target units, or
10378 of all units that Avahi has dependencies with.
10380 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
10381 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
10382 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
10383 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
10384 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
10385 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
10386 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
10390 * The boot-time readahead implementation for rotating media
10391 will now read the read-ahead data in multiple passes which
10392 consist of all read requests made in equidistant time
10393 intervals. This means instead of strictly reading read-ahead
10394 data in its physical order on disk we now try to find a
10395 middle ground between physical and access time order.
10397 * /etc/os-release files gained a new BUILD_ID= field for usage
10398 on operating systems that provide continuous builds of OS
10401 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
10402 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
10403 William Douglas, Zbigniew Jędrzejewski-Szmek
10407 * systemd-python gained an API exposing libsystemd-daemon.
10409 * The SMACK setup logic gained support for uploading CIPSO
10412 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
10413 ReadOnlyDirectories= and InaccessibleDirectories= has
10414 changed. The private /tmp and /var/tmp directories are now
10415 shared by all processes of a service (which means
10416 ExecStartPre= may now leave data in /tmp that ExecStart= of
10417 the same service can still access). When a service is
10418 stopped its temporary directories are immediately deleted
10419 (normal clean-up with tmpfiles is still done in addition to
10422 * By default, systemd will now set a couple of sysctl
10423 variables in the kernel: the safe sysrq options are turned
10424 on, IP route verification is turned on, and source routing
10425 disabled. The recently added hardlink and softlink
10426 protection of the kernel is turned on. These settings should
10427 be reasonably safe, and good defaults for all new systems.
10429 * The predictable network naming logic may now be turned off
10430 with a new kernel command line switch: net.ifnames=0.
10432 * A new libsystemd-bus module has been added that implements a
10433 pretty complete D-Bus client library. For details see:
10435 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
10437 * journald will now explicitly flush the journal files to disk
10438 at the latest 5min after each write. The file will then also
10439 be marked offline until the next write. This should increase
10440 reliability in case of a crash. The synchronization delay
10441 can be configured via SyncIntervalSec= in journald.conf.
10443 * There's a new remote-fs-setup.target unit that can be used
10444 to pull in specific services when at least one remote file
10445 system is to be mounted.
10447 * There are new targets timers.target and paths.target as
10448 canonical targets to pull user timer and path units in
10449 from. This complements sockets.target with a similar
10450 purpose for socket units.
10452 * libudev gained a new call udev_device_set_attribute_value()
10453 to set sysfs attributes of a device.
10455 * The udev daemon now sets the default number of worker
10456 processes executed in parallel based on the number of available
10457 CPUs instead of the amount of available RAM. This is supposed
10458 to provide a more reliable default and limit a too aggressive
10459 parallelism for setups with 1000s of devices connected.
10461 Contributions from: Auke Kok, Colin Walters, Cristian
10462 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
10463 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
10464 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
10465 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
10466 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
10467 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
10468 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
10469 Zbigniew Jędrzejewski-Szmek
10473 * Configuration of unit files may now be extended via drop-in
10474 files without having to edit/override the unit files
10475 themselves. More specifically, if the administrator wants to
10476 change one value for a service file foobar.service he can
10477 now do so by dropping in a configuration snippet into
10478 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
10479 will load all these snippets and apply them on top of the
10480 main unit configuration file, possibly extending or
10481 overriding its settings. Using these drop-in snippets is
10482 generally nicer than the two earlier options for changing
10483 unit files locally: copying the files from
10484 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
10485 them there; or creating a new file in /etc/systemd/system/
10486 that incorporates the original one via ".include". Drop-in
10487 snippets into these .d/ directories can be placed in any
10488 directory systemd looks for units in, and the usual
10489 overriding semantics between /usr/lib, /etc and /run apply
10492 * Most unit file settings which take lists of items can now be
10493 reset by assigning the empty string to them. For example,
10494 normally, settings such as Environment=FOO=BAR append a new
10495 environment variable assignment to the environment block,
10496 each time they are used. By assigning Environment= the empty
10497 string the environment block can be reset to empty. This is
10498 particularly useful with the .d/*.conf drop-in snippets
10499 mentioned above, since this adds the ability to reset list
10500 settings from vendor unit files via these drop-ins.
10502 * systemctl gained a new "list-dependencies" command for
10503 listing the dependencies of a unit recursively.
10505 * Inhibitors are now honored and listed by "systemctl
10506 suspend", "systemctl poweroff" (and similar) too, not only
10507 GNOME. These commands will also list active sessions by
10510 * Resource limits (as exposed by the various control group
10511 controllers) can now be controlled dynamically at runtime
10512 for all units. More specifically, you can now use a command
10513 like "systemctl set-cgroup-attr foobar.service cpu.shares
10514 2000" to alter the CPU shares a specific service gets. These
10515 settings are stored persistently on disk, and thus allow the
10516 administrator to easily adjust the resource usage of
10517 services with a few simple commands. This dynamic resource
10518 management logic is also available to other programs via the
10519 bus. Almost any kernel cgroup attribute and controller is
10522 * systemd-vconsole-setup will now copy all font settings to
10523 all allocated VTs, where it previously applied them only to
10526 * libsystemd-login gained the new sd_session_get_tty() API
10529 * This release drops support for a few legacy or
10530 distribution-specific LSB facility names when parsing init
10531 scripts: $x-display-manager, $mail-transfer-agent,
10532 $mail-transport-agent, $mail-transfer-agent, $smtp,
10533 $null. Also, the mail-transfer-agent.target unit backing
10534 this has been removed. Distributions which want to retain
10535 compatibility with this should carry the burden for
10536 supporting this themselves and patch support for these back
10537 in, if they really need to. Also, the facilities $syslog and
10538 $local_fs are now ignored, since systemd does not support
10539 early-boot LSB init scripts anymore, and these facilities
10540 are implied anyway for normal services. syslog.target has
10543 * There are new bus calls on PID1's Manager object for
10544 cancelling jobs, and removing snapshot units. Previously,
10545 both calls were only available on the Job and Snapshot
10546 objects themselves.
10548 * systemd-journal-gatewayd gained SSL support.
10550 * The various "environment" files, such as /etc/locale.conf
10551 now support continuation lines with a backslash ("\") as
10552 last character in the line, similarly in style (but different)
10553 to how this is supported in shells.
10555 * For normal user processes the _SYSTEMD_USER_UNIT= field is
10556 now implicitly appended to every log entry logged. systemctl
10557 has been updated to filter by this field when operating on a
10558 user systemd instance.
10560 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
10561 CAP_AUDIT_CONTROL capabilities to the capabilities set for
10562 the container. This makes it easier to boot unmodified
10563 Fedora systems in a container, which however still requires
10564 audit=0 to be passed on the kernel command line. Auditing in
10565 kernel and userspace is unfortunately still too broken in
10566 context of containers, hence we recommend compiling it out
10567 of the kernel or using audit=0. Hopefully this will be fixed
10568 one day for good in the kernel.
10570 * nspawn gained the new --bind= and --bind-ro= parameters to
10571 bind mount specific directories from the host into the
10574 * nspawn will now mount its own devpts file system instance
10575 into the container, in order not to leak pty devices from
10576 the host into the container.
10578 * systemd will now read the firmware boot time performance
10579 information from the EFI variables, if the used boot loader
10580 supports this, and takes it into account for boot performance
10581 analysis via "systemd-analyze". This is currently supported
10582 only in conjunction with Gummiboot, but could be supported
10583 by other boot loaders too. For details see:
10585 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
10587 * A new generator has been added that automatically mounts the
10588 EFI System Partition (ESP) to /boot, if that directory
10589 exists, is empty, and no other file system has been
10590 configured to be mounted there.
10592 * logind will now send out PrepareForSleep(false) out
10593 unconditionally, after coming back from suspend. This may be
10594 used by applications as asynchronous notification for
10595 system resume events.
10597 * "systemctl unlock-sessions" has been added, that allows
10598 unlocking the screens of all user sessions at once, similar
10599 to how "systemctl lock-sessions" already locked all users
10600 sessions. This is backed by a new D-Bus call UnlockSessions().
10602 * "loginctl seat-status" will now show the master device of a
10603 seat. (i.e. the device of a seat that needs to be around for
10604 the seat to be considered available, usually the graphics
10607 * tmpfiles gained a new "X" line type, that allows
10608 configuration of files and directories (with wildcards) that
10609 shall be excluded from automatic cleanup ("aging").
10611 * udev default rules set the device node permissions now only
10612 at "add" events, and do not change them any longer with a
10613 later "change" event.
10615 * The log messages for lid events and power/sleep keypresses
10616 now carry a message ID.
10618 * We now have a substantially larger unit test suite, but this
10619 continues to be work in progress.
10621 * udevadm hwdb gained a new --root= parameter to change the
10622 root directory to operate relative to.
10624 * logind will now issue a background sync() request to the kernel
10625 early at shutdown, so that dirty buffers are flushed to disk early
10626 instead of at the last moment, in order to optimize shutdown
10629 * A new bootctl tool has been added that is an interface for
10630 certain boot loader operations. This is currently a preview
10631 and is likely to be extended into a small mechanism daemon
10632 like timedated, localed, hostnamed, and can be used by
10633 graphical UIs to enumerate available boot options, and
10634 request boot into firmware operations.
10636 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
10637 the rest of the package. It also has been updated to work
10638 correctly in initrds.
10640 * polkit previously has been runtime optional, and is now also
10641 compile time optional via a configure switch.
10643 * systemd-analyze has been reimplemented in C. Also "systemctl
10644 dot" has moved into systemd-analyze.
10646 * "systemctl status" with no further parameters will now print
10647 the status of all active or failed units.
10649 * Operations such as "systemctl start" can now be executed
10650 with a new mode "--irreversible" which may be used to queue
10651 operations that cannot accidentally be reversed by a later
10652 job queuing. This is by default used to make shutdown
10653 requests more robust.
10655 * The Python API of systemd now gained a new module for
10656 reading journal files.
10658 * A new tool kernel-install has been added that can install
10659 kernel images according to the Boot Loader Specification:
10661 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
10663 * Boot time console output has been improved to provide
10664 animated boot time output for hanging jobs.
10666 * A new tool systemd-activate has been added which can be used
10667 to test socket activation with, directly from the command
10668 line. This should make it much easier to test and debug
10669 socket activation in daemons.
10671 * journalctl gained a new "--reverse" (or -r) option to show
10672 journal output in reverse order (i.e. newest line first).
10674 * journalctl gained a new "--pager-end" (or -e) option to jump
10675 to immediately jump to the end of the journal in the
10676 pager. This is only supported in conjunction with "less".
10678 * journalctl gained a new "--user-unit=" option, that works
10679 similarly to "--unit=" but filters for user units rather than
10682 * A number of unit files to ease adoption of systemd in
10683 initrds has been added. This moves some minimal logic from
10684 the various initrd implementations into systemd proper.
10686 * The journal files are now owned by a new group
10687 "systemd-journal", which exists specifically to allow access
10688 to the journal, and nothing else. Previously, we used the
10689 "adm" group for that, which however possibly covers more
10690 than just journal/log file access. This new group is now
10691 already used by systemd-journal-gatewayd to ensure this
10692 daemon gets access to the journal files and as little else
10693 as possible. Note that "make install" will also set FS ACLs
10694 up for /var/log/journal to give "adm" and "wheel" read
10695 access to it, in addition to "systemd-journal" which owns
10696 the journal files. We recommend that packaging scripts also
10697 add read access to "adm" + "wheel" to /var/log/journal, and
10698 all existing/future journal files. To normal users and
10699 administrators little changes, however packagers need to
10700 ensure to create the "systemd-journal" system group at
10701 package installation time.
10703 * The systemd-journal-gatewayd now runs as unprivileged user
10704 systemd-journal-gateway:systemd-journal-gateway. Packaging
10705 scripts need to create these system user/group at
10708 * timedated now exposes a new boolean property CanNTP that
10709 indicates whether a local NTP service is available or not.
10711 * systemd-detect-virt will now also detect xen PVs
10713 * The pstore file system is now mounted by default, if it is
10716 * In addition to the SELinux and IMA policies we will now also
10717 load SMACK policies at early boot.
10719 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
10720 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
10721 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
10722 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
10723 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
10724 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
10725 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
10726 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
10727 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
10728 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
10729 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
10730 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
10731 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
10732 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
10736 * Timer units now support calendar time events in addition to
10737 monotonic time events. That means you can now trigger a unit
10738 based on a calendar time specification such as "Thu,Fri
10739 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
10740 or fifth day of any month of the year 2013, given that it is
10741 a thursday or friday. This brings timer event support
10742 considerably closer to cron's capabilities. For details on
10743 the supported calendar time specification language see
10746 * udev now supports a number of different naming policies for
10747 network interfaces for predictable names, and a combination
10748 of these policies is now the default. Please see this wiki
10749 document for details:
10751 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
10753 * Auke Kok's bootchart implementation has been added to the
10754 systemd tree. It is an optional component that can graph the
10755 boot in quite some detail. It is one of the best bootchart
10756 implementations around and minimal in its code and
10759 * nss-myhostname has been integrated into the systemd source
10760 tree. nss-myhostname guarantees that the local hostname
10761 always stays resolvable via NSS. It has been a weak
10762 requirement of systemd-hostnamed since a long time, and
10763 since its code is actually trivial we decided to just
10764 include it in systemd's source tree. It can be turned off
10765 with a configure switch.
10767 * The read-ahead logic is now capable of properly detecting
10768 whether a btrfs file system is on SSD or rotating media, in
10769 order to optimize the read-ahead scheme. Previously, it was
10770 only capable of detecting this on traditional file systems
10773 * In udev, additional device properties are now read from the
10774 IAB in addition to the OUI database. Also, Bluetooth company
10775 identities are attached to the devices as well.
10777 * In service files %U may be used as specifier that is
10778 replaced by the configured user name of the service.
10780 * nspawn may now be invoked without a controlling TTY. This
10781 makes it suitable for invocation as its own service. This
10782 may be used to set up a simple containerized server system
10783 using only core OS tools.
10785 * systemd and nspawn can now accept socket file descriptors
10786 when they are started for socket activation. This enables
10787 implementation of socket activated nspawn
10788 containers. i.e. think about autospawning an entire OS image
10789 when the first SSH or HTTP connection is received. We expect
10790 that similar functionality will also be added to libvirt-lxc
10793 * journalctl will now suppress ANSI color codes when
10794 presenting log data.
10796 * systemctl will no longer show control group information for
10797 a unit if the control group is empty anyway.
10799 * logind can now automatically suspend/hibernate/shutdown the
10802 * /etc/machine-info and hostnamed now also expose the chassis
10803 type of the system. This can be used to determine whether
10804 the local system is a laptop, desktop, handset or
10805 tablet. This information may either be configured by the
10806 user/vendor or is automatically determined from ACPI and DMI
10807 information if possible.
10809 * A number of polkit actions are now bound together with "imply"
10810 rules. This should simplify creating UIs because many actions
10811 will now authenticate similar ones as well.
10813 * Unit files learnt a new condition ConditionACPower= which
10814 may be used to conditionalize a unit depending on whether an
10815 AC power source is connected or not, of whether the system
10816 is running on battery power.
10818 * systemctl gained a new "is-failed" verb that may be used in
10819 shell scripts and suchlike to check whether a specific unit
10820 is in the "failed" state.
10822 * The EnvironmentFile= setting in unit files now supports file
10823 globbing, and can hence be used to easily read a number of
10824 environment files at once.
10826 * systemd will no longer detect and recognize specific
10827 distributions. All distribution-specific #ifdeffery has been
10828 removed, systemd is now fully generic and
10829 distribution-agnostic. Effectively, not too much is lost as
10830 a lot of the code is still accessible via explicit configure
10831 switches. However, support for some distribution specific
10832 legacy configuration file formats has been dropped. We
10833 recommend distributions to simply adopt the configuration
10834 files everybody else uses now and convert the old
10835 configuration from packaging scripts. Most distributions
10836 already did that. If that's not possible or desirable,
10837 distributions are welcome to forward port the specific
10838 pieces of code locally from the git history.
10840 * When logging a message about a unit systemd will now always
10841 log the unit name in the message meta data.
10843 * localectl will now also discover system locale data that is
10844 not stored in locale archives, but directly unpacked.
10846 * logind will no longer unconditionally use framebuffer
10847 devices as seat masters, i.e. as devices that are required
10848 to be existing before a seat is considered preset. Instead,
10849 it will now look for all devices that are tagged as
10850 "seat-master" in udev. By default, framebuffer devices will
10851 be marked as such, but depending on local systems, other
10852 devices might be marked as well. This may be used to
10853 integrate graphics cards using closed source drivers (such
10854 as NVidia ones) more nicely into logind. Note however, that
10855 we recommend using the open source NVidia drivers instead,
10856 and no udev rules for the closed-source drivers will be
10857 shipped from us upstream.
10859 Contributions from: Adam Williamson, Alessandro Crismani, Auke
10860 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
10861 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
10862 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
10863 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
10864 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
10865 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
10866 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
10867 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
10868 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
10869 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
10870 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
10875 * udev gained support for loading additional device properties
10876 from an indexed database that is keyed by vendor/product IDs
10877 and similar device identifiers. For the beginning this
10878 "hwdb" is populated with data from the well-known PCI and
10879 USB database, but also includes PNP, ACPI and OID data. In
10880 the longer run this indexed database shall grow into
10881 becoming the one central database for non-essential
10882 userspace device metadata. Previously, data from the PCI/USB
10883 database was only attached to select devices, since the
10884 lookup was a relatively expensive operation due to O(n) time
10885 complexity (with n being the number of entries in the
10886 database). Since this is now O(1), we decided to add in this
10887 data for all devices where this is available, by
10888 default. Note that the indexed database needs to be rebuilt
10889 when new data files are installed. To achieve this you need
10890 to update your packaging scripts to invoke "udevadm hwdb
10891 --update" after installation of hwdb data files. For
10892 RPM-based distributions we introduced the new
10893 %udev_hwdb_update macro for this purpose.
10895 * The Journal gained support for the "Message Catalog", an
10896 indexed database to link up additional information with
10897 journal entries. For further details please check:
10899 https://www.freedesktop.org/wiki/Software/systemd/catalog
10901 The indexed message catalog database also needs to be
10902 rebuilt after installation of message catalog files. Use
10903 "journalctl --update-catalog" for this. For RPM-based
10904 distributions we introduced the %journal_catalog_update
10905 macro for this purpose.
10907 * The Python Journal bindings gained support for the standard
10908 Python logging framework.
10910 * The Journal API gained new functions for checking whether
10911 the underlying file system of a journal file is capable of
10912 properly reporting file change notifications, or whether
10913 applications that want to reflect journal changes "live"
10914 need to recheck journal files continuously in appropriate
10917 * It is now possible to set the "age" field for tmpfiles
10918 entries to 0, indicating that files matching this entry
10919 shall always be removed when the directories are cleaned up.
10921 * coredumpctl gained a new "gdb" verb which invokes gdb
10922 right-away on the selected coredump.
10924 * There's now support for "hybrid sleep" on kernels that
10925 support this, in addition to "suspend" and "hibernate". Use
10926 "systemctl hybrid-sleep" to make use of this.
10928 * logind's HandleSuspendKey= setting (and related settings)
10929 now gained support for a new "lock" setting to simply
10930 request the screen lock on all local sessions, instead of
10931 actually executing a suspend or hibernation.
10933 * systemd will now mount the EFI variables file system by
10936 * Socket units now gained support for configuration of the
10937 SMACK security label.
10939 * timedatectl will now output the time of the last and next
10940 daylight saving change.
10942 * We dropped support for various legacy and distro-specific
10943 concepts, such as insserv, early-boot SysV services
10944 (i.e. those for non-standard runlevels such as 'b' or 'S')
10945 or ArchLinux /etc/rc.conf support. We recommend the
10946 distributions who still need support this to either continue
10947 to maintain the necessary patches downstream, or find a
10948 different solution. (Talk to us if you have questions!)
10950 * Various systemd components will now bypass polkit checks for
10951 root and otherwise handle properly if polkit is not found to
10952 be around. This should fix most issues for polkit-less
10953 systems. Quite frankly this should have been this way since
10954 day one. It is absolutely our intention to make systemd work
10955 fine on polkit-less systems, and we consider it a bug if
10956 something does not work as it should if polkit is not around.
10958 * For embedded systems it is now possible to build udev and
10959 systemd without blkid and/or kmod support.
10961 * "systemctl switch-root" is now capable of switching root
10962 more than once. I.e. in addition to transitions from the
10963 initrd to the host OS it is now possible to transition to
10964 further OS images from the host. This is useful to implement
10965 offline updating tools.
10967 * Various other additions have been made to the RPM macros
10968 shipped with systemd. Use %udev_rules_update() after
10969 installing new udev rules files. %_udevhwdbdir,
10970 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
10971 %_sysctldir are now available which resolve to the right
10972 directories for packages to place various data files in.
10974 * journalctl gained the new --full switch (in addition to
10975 --all, to disable ellipsation for long messages.
10977 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
10978 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
10979 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
10980 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
10981 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
10982 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
10983 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
10984 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
10985 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
10989 * journalctl gained new --since= and --until= switches to
10990 filter by time. It also now supports nice filtering for
10991 units via --unit=/-u.
10993 * Type=oneshot services may use ExecReload= and do the
10996 * The journal daemon now supports time-based rotation and
10997 vacuuming, in addition to the usual disk-space based
11000 * The journal will now index the available field values for
11001 each field name. This enables clients to show pretty drop
11002 downs of available match values when filtering. The bash
11003 completion of journalctl has been updated
11004 accordingly. journalctl gained a new switch -F to list all
11005 values a certain field takes in the journal database.
11007 * More service events are now written as structured messages
11008 to the journal, and made recognizable via message IDs.
11010 * The timedated, localed and hostnamed mini-services which
11011 previously only provided support for changing time, locale
11012 and hostname settings from graphical DEs such as GNOME now
11013 also have a minimal (but very useful) text-based client
11014 utility each. This is probably the nicest way to changing
11015 these settings from the command line now, especially since
11016 it lists available options and is fully integrated with bash
11019 * There's now a new tool "systemd-coredumpctl" to list and
11020 extract coredumps from the journal.
11022 * We now install a README each in /var/log/ and
11023 /etc/rc.d/init.d explaining where the system logs and init
11024 scripts went. This hopefully should help folks who go to
11025 that dirs and look into the otherwise now empty void and
11026 scratch their heads.
11028 * When user-services are invoked (by systemd --user) the
11029 $MANAGERPID env var is set to the PID of systemd.
11031 * SIGRTMIN+24 when sent to a --user instance will now result
11032 in immediate termination of systemd.
11034 * gatewayd received numerous feature additions such as a
11035 "follow" mode, for live syncing and filtering.
11037 * browse.html now allows filtering and showing detailed
11038 information on specific entries. Keyboard navigation and
11039 mouse screen support has been added.
11041 * gatewayd/journalctl now supports HTML5/JSON
11042 Server-Sent-Events as output.
11044 * The SysV init script compatibility logic will now
11045 heuristically determine whether a script supports the
11046 "reload" verb, and only then make this available as
11047 "systemctl reload".
11049 * "systemctl status --follow" has been removed, use "journalctl
11052 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
11053 have been removed since they are hardly useful to be
11056 * And I'd like to take the opportunity to specifically mention
11057 Zbigniew for his great contributions. Zbigniew, you rock!
11059 Contributions from: Andrew Eikum, Christian Hesse, Colin
11060 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
11061 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
11062 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
11063 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
11064 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
11065 Jędrzejewski-Szmek, Сковорода Никита Андреевич
11069 * If /etc/vconsole.conf is non-existent or empty we will no
11070 longer load any console font or key map at boot by
11071 default. Instead the kernel defaults will be left
11072 intact. This is definitely the right thing to do, as no
11073 configuration should mean no configuration, and hard-coding
11074 font names that are different on all archs is probably a bad
11075 idea. Also, the kernel default key map and font should be
11076 good enough for most cases anyway, and mostly identical to
11077 the userspace fonts/key maps we previously overloaded them
11078 with. If distributions want to continue to default to a
11079 non-kernel font or key map they should ship a default
11080 /etc/vconsole.conf with the appropriate contents.
11082 Contributions from: Colin Walters, Daniel J Walsh, Dave
11083 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
11084 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
11088 * journalctl gained a new --cursor= switch to show entries
11089 starting from the specified location in the journal.
11091 * We now enforce a size limit on journal entry fields exported
11092 with "-o json" in journalctl. Fields larger than 4K will be
11093 assigned null. This can be turned off with --all.
11095 * An (optional) journal gateway daemon is now available as
11096 "systemd-journal-gatewayd.service". This service provides
11097 access to the journal via HTTP and JSON. This functionality
11098 will be used to implement live log synchronization in both
11099 pull and push modes, but has various other users too, such
11100 as easy log access for debugging of embedded devices. Right
11101 now it is already useful to retrieve the journal via HTTP:
11103 # systemctl start systemd-journal-gatewayd.service
11104 # wget http://localhost:19531/entries
11106 This will download the journal contents in a
11107 /var/log/messages compatible format. The same as JSON:
11109 # curl -H"Accept: application/json" http://localhost:19531/entries
11111 This service is also accessible via a web browser where a
11112 single static HTML5 app is served that uses the JSON logic
11113 to enable the user to do some basic browsing of the
11114 journal. This will be extended later on. Here's an example
11115 screenshot of this app in its current state:
11117 http://0pointer.de/public/journal-gatewayd
11119 Contributions from: Kay Sievers, Lennart Poettering, Robert
11120 Milasan, Tom Gundersen
11124 * The bash completion logic is now available for journalctl
11127 * We do not mount the "cpuset" controller anymore together with
11128 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
11129 started if no parameters are assigned to it. "cpuset" hence
11130 broke code that assumed it could create "cpu" groups and
11133 * journalctl -f will now subscribe to terminal size changes,
11134 and line break accordingly.
11136 Contributions from: Dave Reisner, Kay Sievers, Lennart
11137 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
11141 * nspawn will now create a symlink /etc/localtime in the
11142 container environment, copying the host's timezone
11143 setting. Previously this has been done via a bind mount, but
11144 since symlinks cannot be bind mounted this has now been
11145 changed to create/update the appropriate symlink.
11147 * journalctl -n's line number argument is now optional, and
11148 will default to 10 if omitted.
11150 * journald will now log the maximum size the journal files may
11151 take up on disk. This is particularly useful if the default
11152 built-in logic of determining this parameter from the file
11153 system size is used. Use "systemctl status
11154 systemd-journald.service" to see this information.
11156 * The multi-seat X wrapper tool has been stripped down. As X
11157 is now capable of enumerating graphics devices via udev in a
11158 seat-aware way the wrapper is not strictly necessary
11159 anymore. A stripped down temporary stop-gap is still shipped
11160 until the upstream display managers have been updated to
11161 fully support the new X logic. Expect this wrapper to be
11162 removed entirely in one of the next releases.
11164 * HandleSleepKey= in logind.conf has been split up into
11165 HandleSuspendKey= and HandleHibernateKey=. The old setting
11166 is not available anymore. X11 and the kernel are
11167 distinguishing between these keys and we should too. This
11168 also means the inhibition lock for these keys has been split
11171 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
11172 Poettering, Lukas Nykryn, Václav Pavlín
11176 * Whenever a unit changes state we will now log this to the
11177 journal and show along the unit's own log output in
11178 "systemctl status".
11180 * ConditionPathIsMountPoint= can now properly detect bind
11181 mount points too. (Previously, a bind mount of one file
11182 system to another place in the same file system could not be
11183 detected as mount, since they shared struct stat's st_dev
11186 * We will now mount the cgroup controllers cpu, cpuacct,
11187 cpuset and the controllers net_cls, net_prio together by
11190 * nspawn containers will now have a virtualized boot
11191 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
11192 over with a randomized ID at container initialization). This
11193 has the effect of making "journalctl -b" do the right thing
11196 * The JSON output journal serialization has been updated not
11197 to generate "endless" list objects anymore, but rather one
11198 JSON object per line. This is more in line how most JSON
11199 parsers expect JSON objects. The new output mode
11200 "json-pretty" has been added to provide similar output, but
11201 neatly aligned for readability by humans.
11203 * We dropped all explicit sync() invocations in the shutdown
11204 code. The kernel does this implicitly anyway in the kernel
11205 reboot() syscall. halt(8)'s -n option is now a compatibility
11208 * We now support virtualized reboot() in containers, as
11209 supported by newer kernels. We will fall back to exit() if
11210 CAP_SYS_REBOOT is not available to the container. Also,
11211 nspawn makes use of this now and will actually reboot the
11212 container if the containerized OS asks for that.
11214 * journalctl will only show local log output by default
11215 now. Use --merge (-m) to show remote log output, too.
11217 * libsystemd-journal gained the new sd_journal_get_usage()
11218 call to determine the current disk usage of all journal
11219 files. This is exposed in the new "journalctl --disk-usage"
11222 * journald gained a new configuration setting SplitMode= in
11223 journald.conf which may be used to control how user journals
11224 are split off. See journald.conf(5) for details.
11226 * A new condition type ConditionFileNotEmpty= has been added.
11228 * tmpfiles' "w" lines now support file globbing, to write
11229 multiple files at once.
11231 * We added Python bindings for the journal submission
11232 APIs. More Python APIs for a number of selected APIs will
11233 likely follow. Note that we intend to add native bindings
11234 only for the Python language, as we consider it common
11235 enough to deserve bindings shipped within systemd. There are
11236 various projects outside of systemd that provide bindings
11237 for languages such as PHP or Lua.
11239 * Many conditions will now resolve specifiers such as %i. In
11240 addition, PathChanged= and related directives of .path units
11241 now support specifiers as well.
11243 * There's now a new RPM macro definition for the system preset
11246 * journald will now warn if it ca not forward a message to the
11247 syslog daemon because its socket is full.
11249 * timedated will no longer write or process /etc/timezone,
11250 except on Debian. As we do not support late mounted /usr
11251 anymore /etc/localtime always being a symlink is now safe,
11252 and hence the information in /etc/timezone is not necessary
11255 * logind will now always reserve one VT for a text getty (VT6
11256 by default). Previously if more than 6 X sessions where
11257 started they took up all the VTs with auto-spawned gettys,
11258 so that no text gettys were available anymore.
11260 * udev will now automatically inform the btrfs kernel logic
11261 about btrfs RAID components showing up. This should make
11262 simple hotplug based btrfs RAID assembly work.
11264 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
11265 (but not for its children which will stay at the kernel
11266 default). This should allow setups with a lot more listening
11269 * systemd will now always pass the configured timezone to the
11270 kernel at boot. timedated will do the same when the timezone
11273 * logind's inhibition logic has been updated. By default,
11274 logind will now handle the lid switch, the power and sleep
11275 keys all the time, even in graphical sessions. If DEs want
11276 to handle these events on their own they should take the new
11277 handle-power-key, handle-sleep-key and handle-lid-switch
11278 inhibitors during their runtime. A simple way to achieve
11279 that is to invoke the DE wrapped in an invocation of:
11281 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch …
11283 * Access to unit operations is now checked via SELinux taking
11284 the unit file label and client process label into account.
11286 * systemd will now notify the administrator in the journal
11287 when he over-mounts a non-empty directory.
11289 * There are new specifiers that are resolved in unit files,
11290 for the hostname (%H), the machine ID (%m) and the boot ID
11293 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
11294 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
11295 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
11296 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
11297 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
11298 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
11299 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
11303 * Support for reading structured kernel messages from
11304 /dev/kmsg has now been added and is enabled by default.
11306 * Support for reading kernel messages from /proc/kmsg has now
11307 been removed. If you want kernel messages in the journal
11308 make sure to run a recent kernel (>= 3.5) that supports
11309 reading structured messages from /dev/kmsg (see
11310 above). /proc/kmsg is now exclusive property of classic
11311 syslog daemons again.
11313 * The libudev API gained the new
11314 udev_device_new_from_device_id() call.
11316 * The logic for file system namespace (ReadOnlyDirectory=,
11317 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
11318 require pivot_root() anymore. This means fewer temporary
11319 directories are created below /tmp for this feature.
11321 * nspawn containers will now see and receive all submounts
11322 made on the host OS below the root file system of the
11325 * Forward Secure Sealing is now supported for Journal files,
11326 which provide cryptographical sealing of journal files so
11327 that attackers cannot alter log history anymore without this
11328 being detectable. Lennart will soon post a blog story about
11329 this explaining it in more detail.
11331 * There are two new service settings RestartPreventExitStatus=
11332 and SuccessExitStatus= which allow configuration of exit
11333 status (exit code or signal) which will be excepted from the
11334 restart logic, resp. consider successful.
11336 * journalctl gained the new --verify switch that can be used
11337 to check the integrity of the structure of journal files and
11338 (if Forward Secure Sealing is enabled) the contents of
11341 * nspawn containers will now be run with /dev/stdin, /dev/fd/
11342 and similar symlinks pre-created. This makes running shells
11343 as container init process a lot more fun.
11345 * The fstab support can now handle PARTUUID= and PARTLABEL=
11348 * A new ConditionHost= condition has been added to match
11349 against the hostname (with globs) and machine ID. This is
11350 useful for clusters where a single OS image is used to
11351 provision a large number of hosts which shall run slightly
11352 different sets of services.
11354 * Services which hit the restart limit will now be placed in a
11357 Contributions from: Bertram Poettering, Dave Reisner, Huang
11358 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
11359 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
11363 * When running in --user mode systemd will now become a
11364 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
11365 tree a lot more organized.
11367 * A new PartOf= unit dependency type has been introduced that
11368 may be used to group services in a natural way.
11370 * "systemctl enable" may now be used to enable instances of
11373 * journalctl now prints error log levels in red, and
11374 warning/notice log levels in bright white. It also supports
11375 filtering by log level now.
11377 * cgtop gained a new -n switch (similar to top), to configure
11378 the maximum number of iterations to run for. It also gained
11379 -b, to run in batch mode (accepting no input).
11381 * The suffix ".service" may now be omitted on most systemctl
11382 command lines involving service unit names.
11384 * There's a new bus call in logind to lock all sessions, as
11385 well as a loginctl verb for it "lock-sessions".
11387 * libsystemd-logind.so gained a new call sd_journal_perror()
11388 that works similar to libc perror() but logs to the journal
11389 and encodes structured information about the error number.
11391 * /etc/crypttab entries now understand the new keyfile-size=
11394 * shutdown(8) now can send a (configurable) wall message when
11395 a shutdown is cancelled.
11397 * The mount propagation mode for the root file system will now
11398 default to "shared", which is useful to make containers work
11399 nicely out-of-the-box so that they receive new mounts from
11400 the host. This can be undone locally by running "mount
11401 --make-rprivate /" if needed.
11403 * The prefdm.service file has been removed. Distributions
11404 should maintain this unit downstream if they intend to keep
11405 it around. However, we recommend writing normal unit files
11406 for display managers instead.
11408 * Since systemd is a crucial part of the OS we will now
11409 default to a number of compiler switches that improve
11410 security (hardening) such as read-only relocations, stack
11411 protection, and suchlike.
11413 * The TimeoutSec= setting for services is now split into
11414 TimeoutStartSec= and TimeoutStopSec= to allow configuration
11415 of individual time outs for the start and the stop phase of
11418 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
11419 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
11420 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
11421 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
11422 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
11423 Gundersen, Zbigniew Jędrzejewski-Szmek
11427 * The journal and id128 C APIs are now fully documented as man
11430 * Extra safety checks have been added when transitioning from
11431 the initial RAM disk to the main system to avoid accidental
11434 * /etc/crypttab entries now understand the new keyfile-offset=
11437 * systemctl -t can now be used to filter by unit load state.
11439 * The journal C API gained the new sd_journal_wait() call to
11440 make writing synchronous journal clients easier.
11442 * journalctl gained the new -D switch to show journals from a
11443 specific directory.
11445 * journalctl now displays a special marker between log
11446 messages of two different boots.
11448 * The journal is now explicitly flushed to /var via a service
11449 systemd-journal-flush.service, rather than implicitly simply
11450 by seeing /var/log/journal to be writable.
11452 * journalctl (and the journal C APIs) can now match for much
11453 more complex expressions, with alternatives and
11456 * When transitioning from the initial RAM disk to the main
11457 system we will now kill all processes in a killing spree to
11458 ensure no processes stay around by accident.
11460 * Three new specifiers may be used in unit files: %u, %h, %s
11461 resolve to the user name, user home directory resp. user
11462 shell. This is useful for running systemd user instances.
11464 * We now automatically rotate journal files if their data
11465 object hash table gets a fill level > 75%. We also size the
11466 hash table based on the configured maximum file size. This
11467 together should lower hash collisions drastically and thus
11468 speed things up a bit.
11470 * journalctl gained the new "--header" switch to introspect
11471 header data of journal files.
11473 * A new setting SystemCallFilters= has been added to services which may
11474 be used to apply deny lists or allow lists to system calls. This is
11475 based on SECCOMP Mode 2 of Linux 3.5.
11477 * nspawn gained a new --link-journal= switch (and quicker: -j)
11478 to link the container journal with the host. This makes it
11479 very easy to centralize log viewing on the host for all
11480 guests while still keeping the journal files separated.
11482 * Many bugfixes and optimizations
11484 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
11485 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
11486 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
11491 * Several tools now understand kernel command line arguments,
11492 which are only read when run in an initial RAM disk. They
11493 usually follow closely their normal counterparts, but are
11496 * There's a new tool to analyze the readahead files that are
11497 automatically generated at boot. Use:
11499 /usr/lib/systemd/systemd-readahead analyze /.readahead
11501 * We now provide an early debug shell on tty9 if this enabled. Use:
11503 systemctl enable debug-shell.service
11505 * All plymouth related units have been moved into the Plymouth
11506 package. Please make sure to upgrade your Plymouth version
11509 * systemd-tmpfiles now supports getting passed the basename of
11510 a configuration file only, in which case it will look for it
11511 in all appropriate directories automatically.
11513 * udevadm info now takes a /dev or /sys path as argument, and
11514 does the right thing. Example:
11516 udevadm info /dev/sda
11517 udevadm info /sys/class/block/sda
11519 * systemctl now prints a warning if a unit is stopped but a
11520 unit that might trigger it continues to run. Example: a
11521 service is stopped but the socket that activates it is left
11524 * "systemctl status" will now mention if the log output was
11525 shortened due to rotation since a service has been started.
11527 * The journal API now exposes functions to determine the
11528 "cutoff" times due to rotation.
11530 * journald now understands SIGUSR1 and SIGUSR2 for triggering
11531 immediately flushing of runtime logs to /var if possible,
11532 resp. for triggering immediate rotation of the journal
11535 * It is now considered an error if a service is attempted to
11536 be stopped that is not loaded.
11538 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
11540 * systemd-analyze now supports Python 3
11542 * tmpfiles now supports cleaning up directories via aging
11543 where the first level dirs are always kept around but
11544 directories beneath it automatically aged. This is enabled
11545 by prefixing the age field with '~'.
11547 * Seat objects now expose CanGraphical, CanTTY properties
11548 which is required to deal with very fast bootups where the
11549 display manager might be running before the graphics drivers
11550 completed initialization.
11552 * Seat objects now expose a State property.
11554 * We now include RPM macros for service enabling/disabling
11555 based on the preset logic. We recommend RPM based
11556 distributions to make use of these macros if possible. This
11557 makes it simpler to reuse RPM spec files across
11560 * We now make sure that the collected systemd unit name is
11561 always valid when services log to the journal via
11564 * There's a new man page kernel-command-line(7) detailing all
11565 command line options we understand.
11567 * The fstab generator may now be disabled at boot by passing
11568 fstab=0 on the kernel command line.
11570 * A new kernel command line option modules-load= is now understood
11571 to load a specific kernel module statically, early at boot.
11573 * Unit names specified on the systemctl command line are now
11574 automatically escaped as needed. Also, if file system or
11575 device paths are specified they are automatically turned
11576 into the appropriate mount or device unit names. Example:
11578 systemctl status /home
11579 systemctl status /dev/sda
11581 * The SysVConsole= configuration option has been removed from
11582 system.conf parsing.
11584 * The SysV search path is no longer exported on the D-Bus
11587 * The Names= option has been removed from unit file parsing.
11589 * There's a new man page bootup(7) detailing the boot process.
11591 * Every unit and every generator we ship with systemd now
11592 comes with full documentation. The self-explanatory boot is
11595 * A couple of services gained "systemd-" prefixes in their
11596 name if they wrap systemd code, rather than only external
11597 code. Among them fsck@.service which is now
11598 systemd-fsck@.service.
11600 * The HaveWatchdog property has been removed from the D-Bus
11603 * systemd.confirm_spawn= on the kernel command line should now
11606 * There's a new man page crypttab(5) which details all options
11607 we actually understand.
11609 * systemd-nspawn gained a new --capability= switch to pass
11610 additional capabilities to the container.
11612 * timedated will now read known NTP implementation unit names
11613 from /usr/lib/systemd/ntp-units.d/*.list,
11614 systemd-timedated-ntp.target has been removed.
11616 * journalctl gained a new switch "-b" that lists log data of
11617 the current boot only.
11619 * The notify socket is in the abstract namespace again, in
11620 order to support daemons which chroot() at start-up.
11622 * There is a new Storage= configuration option for journald
11623 which allows configuration of where log data should go. This
11624 also provides a way to disable journal logging entirely, so
11625 that data collected is only forwarded to the console, the
11626 kernel log buffer or another syslog implementation.
11628 * Many bugfixes and optimizations
11630 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
11631 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
11632 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
11633 Shawn Landden, Tom Gundersen
11637 * "systemctl help <unit>" now shows the man page if one is
11640 * Several new man pages have been added.
11642 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
11643 MaxLevelConsole= can now be specified in
11644 journald.conf. These options allow reducing the amount of
11645 data stored on disk or forwarded by the log level.
11647 * TimerSlackNSec= can now be specified in system.conf for
11648 PID1. This allows system-wide power savings.
11650 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
11651 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
11656 * logind is now capable of (optionally) handling power and
11657 sleep keys as well as the lid switch.
11659 * journalctl now understands the syntax "journalctl
11660 /usr/bin/avahi-daemon" to get all log output of a specific
11663 * CapabilityBoundingSet= in system.conf now also influences
11664 the capability bound set of usermode helpers of the kernel.
11666 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
11667 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
11668 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
11669 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
11673 * Note that we skipped 139 releases here in order to set the
11674 new version to something that is greater than both udev's
11675 and systemd's most recent version number.
11677 * udev: all udev sources are merged into the systemd source tree now.
11678 All future udev development will happen in the systemd tree. It
11679 is still fully supported to use the udev daemon and tools without
11680 systemd running, like in initramfs or other init systems. Building
11681 udev though, will require the *build* of the systemd tree, but
11682 udev can be properly *run* without systemd.
11684 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
11685 should be used to create dead device nodes as workarounds for broken
11688 * udev: RUN+="socket:…" and udev_monitor_new_from_socket() is
11689 no longer supported. udev_monitor_new_from_netlink() needs to be
11690 used to subscribe to events.
11692 * udev: when udevd is started by systemd, processes which are left
11693 behind by forking them off of udev rules, are unconditionally cleaned
11694 up and killed now after the event handling has finished. Services or
11695 daemons must be started as systemd services. Services can be
11696 pulled-in by udev to get started, but they can no longer be directly
11697 forked by udev rules.
11699 * udev: the daemon binary is called systemd-udevd now and installed
11700 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
11701 to adapt to that, create symlink, or rename the binary after building
11704 * libudev no longer provides these symbols:
11705 udev_monitor_from_socket()
11706 udev_queue_get_failed_list_entry()
11707 udev_get_{dev,sys,run}_path()
11708 The versions number was bumped and symbol versioning introduced.
11710 * systemd-loginctl and systemd-journalctl have been renamed
11711 to loginctl and journalctl to match systemctl.
11713 * The config files: /etc/systemd/systemd-logind.conf and
11714 /etc/systemd/systemd-journald.conf have been renamed to
11715 logind.conf and journald.conf. Package updates should rename
11716 the files to the new names on upgrade.
11718 * For almost all files the license is now LGPL2.1+, changed
11719 from the previous GPL2.0+. Exceptions are some minor stuff
11720 of udev (which will be changed to LGPL2.1 eventually, too),
11721 and the MIT licensed sd-daemon.[ch] library that is suitable
11722 to be used as drop-in files.
11724 * systemd and logind now handle system sleep states, in
11725 particular suspending and hibernating.
11727 * logind now implements a sleep/shutdown/idle inhibiting logic
11728 suitable for a variety of uses. Soonishly Lennart will blog
11729 about this in more detail.
11731 * var-run.mount and var-lock.mount are no longer provided
11732 (which previously bind mounted these directories to their new
11733 places). Distributions which have not converted these
11734 directories to symlinks should consider stealing these files
11735 from git history and add them downstream.
11737 * We introduced the Documentation= field for units and added
11738 this to all our shipped units. This is useful to make it
11739 easier to explore the boot and the purpose of the various
11742 * All smaller setup units (such as
11743 systemd-vconsole-setup.service) now detect properly if they
11744 are run in a container and are skipped when
11745 appropriate. This guarantees an entirely noise-free boot in
11746 Linux container environments such as systemd-nspawn.
11748 * A framework for implementing offline system updates is now
11749 integrated, for details see:
11750 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
11752 * A new service type Type=idle is available now which helps us
11753 avoiding ugly interleaving of getty output and boot status
11756 * There's now a system-wide CapabilityBoundingSet= option to
11757 globally reduce the set of capabilities for the
11758 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
11759 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
11760 even CAP_NET_ADMIN system-wide for secure systems.
11762 * There are now system-wide DefaultLimitXXX= options to
11763 globally change the defaults of the various resource limits
11764 for all units started by PID 1.
11766 * Harald Hoyer's systemd test suite has been integrated into
11767 systemd which allows easy testing of systemd builds in qemu
11768 and nspawn. (This is really awesome! Ask us for details!)
11770 * The fstab parser is now implemented as generator, not inside
11773 * systemctl will now warn you if .mount units generated from
11774 /etc/fstab are out of date due to changes in fstab that
11775 have not been read by systemd yet.
11777 * systemd is now suitable for usage in initrds. Dracut has
11778 already been updated to make use of this. With this in place
11779 initrds get a slight bit faster but primarily are much
11780 easier to introspect and debug since "systemctl status" in
11781 the host system can be used to introspect initrd services,
11782 and the journal from the initrd is kept around too.
11784 * systemd-delta has been added, a tool to explore differences
11785 between user/admin configuration and vendor defaults.
11787 * PrivateTmp= now affects both /tmp and /var/tmp.
11789 * Boot time status messages are now much prettier and feature
11790 proper english language. Booting up systemd has never been
11793 * Read-ahead pack files now include the inode number of all
11794 files to pre-cache. When the inode changes the pre-caching
11795 is not attempted. This should be nicer to deal with updated
11796 packages which might result in changes of read-ahead
11799 * We now temporaritly lower the kernel's read_ahead_kb variable
11800 when collecting read-ahead data to ensure the kernel's
11801 built-in read-ahead does not add noise to our measurements
11802 of necessary blocks to pre-cache.
11804 * There's now RequiresMountsFor= to add automatic dependencies
11805 for all mounts necessary for a specific file system path.
11807 * MountAuto= and SwapAuto= have been removed from
11808 system.conf. Mounting file systems at boot has to take place
11811 * nspawn now learned a new switch --uuid= to set the machine
11812 ID on the command line.
11814 * nspawn now learned the -b switch to automatically search
11815 for an init system.
11817 * vt102 is now the default TERM for serial TTYs, upgraded from
11820 * systemd-logind now works on VT-less systems.
11822 * The build tree has been reorganized. The individual
11823 components now have directories of their own.
11825 * A new condition type ConditionPathIsReadWrite= is now available.
11827 * nspawn learned the new -C switch to create cgroups for the
11828 container in other hierarchies.
11830 * We now have support for hardware watchdogs, configurable in
11833 * The scheduled shutdown logic now has a public API.
11835 * We now mount /tmp as tmpfs by default, but this can be
11836 masked and /etc/fstab can override it.
11838 * Since udisks does not make use of /media anymore we are not
11839 mounting a tmpfs on it anymore.
11841 * journalctl gained a new --local switch to only interleave
11842 locally generated journal files.
11844 * We can now load the IMA policy at boot automatically.
11846 * The GTK tools have been split off into a systemd-ui.
11848 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
11849 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
11850 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
11851 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
11852 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
11853 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
11854 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
11855 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
11856 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
11861 * This is mostly a bugfix release
11863 * Support optional initialization of the machine ID from the
11864 KVM or container configured UUID.
11866 * Support immediate reboots with "systemctl reboot -ff"
11868 * Show /etc/os-release data in systemd-analyze output
11870 * Many bugfixes for the journal, including endianness fixes and
11871 ensuring that disk space enforcement works
11873 * sd-login.h is C++ compatible again
11875 * Extend the /etc/os-release format on request of the Debian
11878 * We now refuse non-UTF8 strings used in various configuration
11879 and unit files. This is done to ensure we do not pass invalid
11880 data over D-Bus or expose it elsewhere.
11882 * Register Mimo USB Screens as suitable for automatic seat
11885 * Read SELinux client context from journal clients in a race
11888 * Reorder configuration file lookup order. /etc now always
11889 overrides /run in order to allow the administrator to always
11890 and unconditionally override vendor-supplied or
11891 automatically generated data.
11893 * The various user visible bits of the journal now have man
11894 pages. We still lack man pages for the journal API calls
11897 * We now ship all man pages in HTML format again in the
11900 Contributions from: Dave Reisner, Dirk Eibach, Frederic
11901 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
11902 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
11907 * This is mostly a bugfix release
11909 * systems lacking /etc/os-release are no longer supported.
11911 * Various functionality updates to libsystemd-login.so
11913 * Track class of PAM logins to distinguish greeters from
11914 normal user logins.
11916 Contributions from: Kay Sievers, Lennart Poettering, Michael
11921 * This is an important bugfix release for v41.
11923 * Building man pages is now optional which should be useful
11924 for those building systemd from git but unwilling to install
11927 * Watchdog support for supervising services is now usable. In
11928 a future release support for hardware watchdogs
11929 (i.e. /dev/watchdog) will be added building on this.
11931 * Service start rate limiting is now configurable and can be
11932 turned off per service. When a start rate limit is hit a
11933 reboot can automatically be triggered.
11935 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
11937 Contributions from: Benjamin Franzke, Bill Nottingham,
11938 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
11939 Schmidt, Michał Górny, Piotr Drąg
11943 * The systemd binary is installed /usr/lib/systemd/systemd now;
11944 An existing /sbin/init symlink needs to be adapted with the
11947 * The code that loads kernel modules has been ported to invoke
11948 libkmod directly, instead of modprobe. This means we do not
11949 support systems with module-init-tools anymore.
11951 * Watchdog support is now already useful, but still not
11954 * A new kernel command line option systemd.setenv= is
11955 understood to set system wide environment variables
11956 dynamically at boot.
11958 * We now limit the set of capabilities of systemd-journald.
11960 * We now set SIGPIPE to ignore by default, since it only is
11961 useful in shell pipelines, and has little use in general
11962 code. This can be disabled with IgnoreSIPIPE=no in unit
11965 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
11966 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
11971 * This is mostly a bugfix release
11973 * We now expose the reason why a service failed in the
11974 "Result" D-Bus property.
11976 * Rudimentary service watchdog support (will be completed over
11977 the next few releases.)
11979 * When systemd forks off in order execute some service we will
11980 now immediately changes its argv[0] to reflect which process
11981 it will execute. This is useful to minimize the time window
11982 with a generic argv[0], which makes bootcharts more useful
11984 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
11985 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
11986 Mike Kazantsev, Ray Strode
11990 * This is mostly a test release, but incorporates many
11993 * New systemd-cgtop tool to show control groups by their
11996 * Linking against libacl for ACLs is optional again. If
11997 disabled, support tracking device access for active logins
11998 goes becomes unavailable, and so does access to the user
11999 journals by the respective users.
12001 * If a group "adm" exists, journal files are automatically
12002 owned by them, thus allow members of this group full access
12003 to the system journal as well as all user journals.
12005 * The journal now stores the SELinux context of the logging
12006 client for all entries.
12008 * Add C++ inclusion guards to all public headers
12010 * New output mode "cat" in the journal to print only text
12011 messages, without any meta data like date or time.
12013 * Include tiny X server wrapper as a temporary stop-gap to
12014 teach XOrg udev display enumeration. This is used by display
12015 managers such as gdm, and will go away as soon as XOrg
12016 learned native udev hotplugging for display devices.
12018 * Add new systemd-cat tool for executing arbitrary programs
12019 with STDERR/STDOUT connected to the journal. Can also act as
12020 BSD logger replacement, and does so by default.
12022 * Optionally store all locally generated coredumps in the
12023 journal along with meta data.
12025 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
12026 writing short strings to files (for usage for /sys), and for
12027 creating symlinks, character and block device nodes.
12029 * New unit file option ControlGroupPersistent= to make cgroups
12030 persistent, following the mechanisms outlined in
12031 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
12033 * Support multiple local RTCs in a sane way
12035 * No longer monopolize IO when replaying readahead data on
12036 rotating disks, since we might starve non-file-system IO to
12037 death, since fanotify() will not see accesses done by blkid,
12040 * Do not show kernel threads in systemd-cgls anymore, unless
12041 requested with new -k switch.
12043 Contributions from: Dan Horák, Kay Sievers, Lennart
12044 Poettering, Michal Schmidt
12048 * This is mostly a test release, but incorporates many
12051 * The git repository moved to:
12052 git://anongit.freedesktop.org/systemd/systemd
12053 ssh://git.freedesktop.org/git/systemd/systemd
12055 * First release with the journal
12056 http://0pointer.de/blog/projects/the-journal.html
12058 * The journal replaces both systemd-kmsg-syslogd and
12059 systemd-stdout-bridge.
12061 * New sd_pid_get_unit() API call in libsystemd-logind
12063 * Many systemadm clean-ups
12065 * Introduce remote-fs-pre.target which is ordered before all
12066 remote mounts and may be used to start services before all
12069 * Added Mageia support
12071 * Add bash completion for systemd-loginctl
12073 * Actively monitor PID file creation for daemons which exit in
12074 the parent process before having finished writing the PID
12075 file in the daemon process. Daemons which do this need to be
12076 fixed (i.e. PID file creation must have finished before the
12077 parent exits), but we now react a bit more gracefully to them.
12079 * Add colourful boot output, mimicking the well-known output
12080 of existing distributions.
12082 * New option PassCredentials= for socket units, for
12083 compatibility with a recent kernel ABI breakage.
12085 * /etc/rc.local is now hooked in via a generator binary, and
12086 thus will no longer act as synchronization point during
12089 * systemctl list-unit-files now supports --root=.
12091 * systemd-tmpfiles now understands two new commands: z, Z for
12092 relabelling files according to the SELinux database. This is
12093 useful to apply SELinux labels to specific files in /sys,
12094 among other things.
12096 * Output of SysV services is now forwarded to both the console
12097 and the journal by default, not only just the console.
12099 * New man pages for all APIs from libsystemd-login.
12101 * The build tree got reorganized and the build system is a
12102 lot more modular allowing embedded setups to specifically
12103 select the components of systemd they are interested in.
12105 * Support for Linux systems lacking the kernel VT subsystem is
12108 * configure's --with-rootdir= got renamed to
12109 --with-rootprefix= to follow the naming used by udev and
12112 * Unless specified otherwise we will now install to /usr instead
12113 of /usr/local by default.
12115 * Processes with '@' in argv[0][0] are now excluded from the
12116 final shut-down killing spree, following the logic explained
12118 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
12120 * All processes remaining in a service cgroup when we enter
12121 the START or START_PRE states are now killed with
12122 SIGKILL. That means it is no longer possible to spawn
12123 background processes from ExecStart= lines (which was never
12124 supported anyway, and bad style).
12126 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
12127 reloading of units together.
12129 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
12130 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
12131 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
12132 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
12133 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek