1 systemd System and Service Manager
5 * When operating on disk images via the --image= switch of various
6 tools (such as systemd-nspawn or systemd-dissect), or when udev finds
7 no 'root=' parameter on the kernel command line, and multiple
8 suitable root or /usr/ partitions exist in the image, then a simple
9 comparison inspired by strverscmp() is done on the GPT partition
10 label, and the newest partition is picked. This permits a simple and
11 generic whole-file-system A/B update logic where new operating system
12 versions are dropped into partitions whose label is then updated with
13 a matching version identifier.
15 * systemd-sysusers now supports querying the passwords to set for the
16 users it creates via the "credentials" logic introduced in v247: the
17 passwd.hashed-password.<user> and passwd.plaintext-password.<user>
18 credentials are consulted for the password to use (either in UNIX
19 hashed form, or literally). By default these credentials are inherited
20 down from PID1 (which in turn imports it from a container manager if
21 there is one). This permits easy configuration of user passwords
22 during first boot. Example:
24 # systemd-nspawn -i foo.raw --volatile=yes --set-credential=passwd.plaintext-password.root:foo
26 Note that systemd-sysusers operates in purely additive mode: it
27 executes no operation if the declared users already exist, and hence
28 doesn't set any passwords as effect of the command line above if the
29 specified root user exists already in the image. (Note that
30 --volatile=yes ensures it doesn't, though.)
32 * systemd-firstboot now also supports querying various system
33 parameters via the credential subsystems. Thus, as above this may be
34 used to initialize important system parameters on first boot of
35 previously unprovisioned images (i.e. images with a mostly empty
38 * PID 1 may now show both the unit name and the unit description
39 strings in its status output during boot. This may be configured with
40 StatusUnitFormat=combined in system.conf or
41 systemd.status-unit-format=combined on the kernel command line.
43 * The systemd-machine-id-setup tool now supports a --image= switch for
44 provisioning a machine ID file into an OS disk image, similar to how
45 --root= operates on an OS file tree. This matches the existing switch
46 of the same name for systemd-tmpfiles, systemd-firstboot, and
47 systemd-sysusers tools.
49 * Similarly, systemd-repart gained support for the --image= switch too.
50 In combination with the existing --size= option, this makes the tool
51 particularly useful for easily growing disk images in a single
52 invocation, following the declarative rules included in the image
55 * systemd-repart's partition configuration files gained support for a
56 new switch MakeDirectories= which may be used to create arbitrary
57 directories inside file systems that are created, before registering
58 them in the partition table. This is useful in particular for root
59 partitions to create mount point directories for other partitions
60 included in the image. For example, a disk image that contains a
61 root, /home/, and /var/ partitions, may set MakeDirectories=yes to
62 create /home/ and /var/ as empty directories in the root file system
63 on its creation, so that the resulting image can be mounted
64 immediately, even in read-only mode.
66 * systemd-repart's CopyBlocks= setting gained support for the special
67 value "auto". If used, a suitable matching partition on the booted OS
68 is found as source to copy blocks from. This is useful when
69 implementing replicating installers, that are booted from one medium
70 and then stream their own root partition onto the target medium.
72 * systemd-repart's partition configuration files gained support for a
73 Flags=, a ReadOnly= and a NoAuto= setting, allowing control of these
74 GPT partition flags for the created partitions: this is useful for
75 marking newly created partitions as read-only, or as not being
76 subject for automatic mounting from creation on.
78 * The /etc/os-release file has been extended with two new (optional)
79 variables IMAGE_VERSION= and IMAGE_ID=, carrying identity and version
80 information for OS images that are updated comprehensively and
81 atomically as one image. Two new specifiers %M, %A now resolve to
82 these two fields in the various configuration options that resolve
85 * portablectl gained a new switch --extension= for enabling portable
86 service images with extensions that follow the extension image
87 concept introduced with v248, and thus allows layering multiple
88 images when setting up the root filesystem of the service.
90 * systemd-coredump will now extract ELF build-id information from
91 processes dumping core and include it in the coredump report.
92 Moreover, it will look for ELF .note.package sections with
93 distribution packaging meta-information about the crashing process.
94 This is useful to directly embed the rpm or deb (or any other)
95 package name and version in ELF files, making it easy to match
96 coredump reports with the specific package for which the software was
97 compiled. This is particularly useful on environments with ELF files
98 from multiple vendors, different distributions and versions, as is
99 common today in our containerized and sand-boxed world. For further
102 https://systemd.io/COREDUMP_PACKAGE_METADATA
104 * A new udev hardware database has been added for FireWire devices
107 * The "net_id" built-in of udev has been updated with three
108 backwards-incompatible changes:
110 - PCI hotplug slot names on s390 systems are now parsed as
111 hexadecimal numbers. They were incorrectly parsed as decimal
112 previously, or ignored if the name was not a valid decimal
115 - PCI onboard indices up to 65535 are allowed. Previously, numbers
116 above 16383 were rejected. This primarily impacts s390 systems,
117 where values up to 65535 are used.
119 - Invalid characters in interface names are replaced with "_".
121 The new version of the net naming scheme is "v249". The previous
122 scheme can be selected via the "net.naming-scheme=v247" kernel
123 command line parameter.
125 * sd-bus' sd_bus_is_ready() and sd_bus_is_open() calls now accept a
126 NULL bus object, for which they will return false. Or in other words,
127 an unallocated bus connection is neither ready nor open.
129 * The sd-device API acquired a new API function
130 sd_device_get_usec_initialized() that returns the monotonic time when
131 the udev device first appeared in the database.
133 * sd-device gained a new APIs sd_device_trigger_with_uuid() and
134 sd_device_get_trigger_uuid(). The former is similar to
135 sd_device_trigger() but returns a randomly generated UUID that is
136 associated with the synthetic uevent generated by the call. This UUID
137 may be read from the sd_device object a monitor eventually receives,
138 via the sd_device_get_trigger_uuid(). This interface requires kernel
139 4.13 or above to work, and allows tracking a synthetic uevent through
140 the entire device management stack. The "udevadm trigger --settle"
141 logic has been updated to make use of this concept if available to
142 wait precisely for the uevents it generates. "udevadm trigger" also
143 gained a new parameter --uuid that prints the UUID for each generated
146 * sd-device also gained new APIs sd_device_new_from_ifname() and
147 sd_device_new_from_ifindex() for allocating an sd-device object for
148 the specified network interface. The former accepts an interface name
149 (either a primary or an alternative name), the latter an interface
152 * The native Journal protocol has been documented. Clients may talk
153 this as alternative to the classic BSD syslog protocol for locally
154 delivering log records to the Journal. The protocol has been stable
155 for a long time and in fact been implemented already in a variety
156 of alternative client libraries. This documentation makes the support
159 https://systemd.io/JOURNAL_NATIVE_PROTOCOL
161 * A new BPFProgram= setting has been added to service files. It may be
162 set to a path to a loaded kernel BPF program, i.e. a path to a bpffs
163 file, or a bind mount or symlink to one. This may be used to upload
164 and manage BPF programs externally and then hook arbitrary systemd
167 * The "home.arpa" domain that has been officially declared as the
168 choice for domain for local home networks per RFC 8375 has been added
169 to the default NTA list of resolved, since DNSSEC is generally not
170 available on private domains.
172 * The CPUAffinity= setting of unit files now resolves "%" specifiers.
174 * A new ManageForeignRoutingPolicyRules= setting has been added to
175 .network files which may be used to exclude foreign-created routing
176 policy rules from systemd-networkd management.
178 * systemd-network-wait-online gained two new switches -4 and -6 that
179 may be used to tweak whether to wait for only IPv4 or only IPv6
182 * .network files gained a new RequiredFamilyForOnline= setting to
183 fine-tune whether to require an IPv4 or IPv6 address in order to
184 consider an interface "online".
186 * networkctl will now show an over-all "online" state in the per-link
189 * In .network files a new OutgoingInterface= setting has been added to
190 specify the output interface in bridge FDB setups.
192 * In .network files the Multipath group ID may now be configured for
193 [NextHop] entries, via the new Group= setting.
195 * The DHCP server logic configured in .network files gained a new
196 setting RelayTarget= that turns the server into a DHCP server relay.
197 The RelayAgentCircuitId= and RelayAgentRemoteId= settings may be used
198 to further tweak the DHCP relay behaviour.
200 * The DHCP server logic also gained a new ServerAddress= setting in
201 .network files that explicitly specifies the server IP address to
202 use. If not specified, the address is determined automatically, as
205 * The DHCP server logic in systemd-networkd gained support for static
206 DHCP leases, configurable via the [DHCPServerStaticLease]
207 section. This allows explicitly mapping specific MAC addresses to
208 fixed IP addresses and vice versa.
210 * The RestrictAddressFamilies= setting in service files now supports a
211 new special value "none". If specified sockets of all address
212 families will be made unavailable to services configured that way.
214 * systemd-fstab-generator and systemd-repart have been updated to
215 support booting from disks that carry only a /usr/ partition but no
216 root partition yet, and where systemd-repart can add it in on the
217 first boot. This is useful for implementing systems that ship with a
218 single /usr/ file system, and whose root file system shall be set up
219 and formatted on a LUKS-encrypted volume whose key is generated
220 locally (and possibly enrolled in the TPM) during the first boot.
222 * The [Address] section of .network files now accepts a new
223 RouteMetric= setting that configures the routing metric to use for
224 the prefix route created as effect of the address configuration.
225 Similarly, the [DHCPv6PrefixDelegation] and [IPv6Prefix] sections
226 gained matching settings for their prefix routes. (The option of the
227 same name in the [DHCPv6] section is moved to [IPv6AcceptRA], since
228 it conceptually belongs there; the old option is still understood for
231 * The DHCPv6 IAID and DUID are now explicitly configurable in .network
234 * A new udev property ID_NET_DHCP_BROADCAST on network interface
235 devices is now honoured by systemd-networkd, controlling whether to
236 issue DHCP offers via broadcasting. This is used to ensure that s390
237 layer 3 network interfaces work out-of-the-box with systemd-networkd.
239 * nss-myhostname and systemd-resolved will now synthesize address
240 records for a new special hostname "_outbound". The name will always
241 resolve to the local IP addresses most likely used for outbound
242 connections towards the default routes. On multi-homed hosts this is
243 useful to have a stable handle referring to "the" local IP address
244 that matters most, to the point where this is defined.
246 * The Discoverable Partition Specification has been updated with a new
247 GPT partition flag "grow-file-system" defined for its partition
248 types. Whenever partitions with this flag set are automatically
249 mounted (i.e. via systemd-gpt-auto-generator or the --image= switch
250 of systemd-nspawn or other tools; and as opposed to explicit mounting
251 via /etc/fstab), the file system within the partition is
252 automatically grown to the full size of the partition. If the file
253 system size already matches the partition size this flag has no
254 effect. Previously, this functionality has been available via the
255 explicit x-systemd.growfs mount option, and this new flag extends
256 this to automatically discovered mounts. A new GrowFileSystem=
257 setting has been added to systemd-repart drop-in files that allows
258 configuring this partition flag. This new flag defaults to on for
259 partitions automatically created by systemd-repart, except if they
260 are marked read-only. See the specification for further details:
262 https://systemd.io/DISCOVERABLE_PARTITIONS
264 * .network files gained a new setting RoutesToNTP= in the [DHCPv4]
265 section. If enabled (which is the default), and an NTP server address
266 is acquired through a DHCP lease on this interface an explicit route
267 to this address is created on this interface to ensure that NTP
268 traffic to the NTP server acquired on an interface is also routed
269 through that interface. The pre-existing RoutesToDNS= setting that
270 implements the same for DNS servers is now enabled by default.
272 * A pair of service settings SocketBindAllow= + SocketBindDeny= have
273 been added that may be used to restrict the network interfaces
274 sockets created by the service may be bound to. This is implemented
277 * A new ConditionFirmware= setting has been added to unit files to
278 conditionalize on certain firmware features. At the moment it may
279 check whether running on an UEFI system, a device.tree system, or if
280 the system is compatible with some specified device-tree feature.
282 * A new ConditionOSRelease= setting has been added to unit files to
283 check os-release(5) fields. The "=", "!=", "<", "<=", ">=", ">"
284 operators may be used to check if some field has some specific value
285 or do an alphanumerical comparison. Equality comparisons are useful
286 for fields like ID, but relative comparisons for fields like
287 VERSION_ID or IMAGE_VERSION.
289 * hostnamed gained a new Describe() D-Bus method that returns a JSON
290 serialization of the host data it exposes. This is exposed via
291 "hostnamectl --json=" to acquire a host identity description in JSON.
292 It's our intention to add a similar features to most services and
293 objects systemd manages, in order to simplify integration with
294 program code that can consume JSON.
296 * Similarly, networkd gained a Describe() method on its Manager and
297 Link bus objects. This is exposed via "networkctl --json=".
299 * hostnamectl's various "get-xyz"/"set-xyz" verb pairs
300 (e.g. "hostnamectl get-hostname", "hostnamectl "set-hostname") have
301 been replaced by a single "xyz" verb (e.g. "hostnamectl hostname")
302 that is used both to get the value (when no argument is given), and
303 to set the value (when an argument is specified). The old names
304 continue to be supported for compatibility.
306 * systemd-detect-virt and ConditionVirtualization= are now able to
307 correctly identify Amazon EC2 environments.
309 * The LogLevelMax= setting of unit files now applies not only to log
310 messages generated *by* the service, but also to log messages
311 generated *about* the service by PID 1. To suppress logs concerning a
312 specific service comprehensively, set this option to a high log
315 * bootctl gained support for a new --make-machine-id-directory= switch
316 that allows precise control on whether to create the top-level
317 per-machine directory in the boot partition that typically contains
318 Type 1 boot loader entries.
320 * During build SBAT data to include in the systemd-boot EFI PE binaries
321 may be specified now.
323 * /etc/crypttab learnt a new option "headless". If specified any
324 requests to query the user interactively for passwords or PINs will
325 be skipped. This is useful on systems that are headless, i.e. where
326 an interactive user is generally not present.
328 * /etc/crypttab also learnt a new option "password-echo=" that allows
329 configuring whether the encryption password prompt shall echo the
330 typed password and if so, do so literally or via asterisks. (The
331 default is the same behaviour as before: provide echo feedback via
334 * FIDO2 support in systemd-cryptenroll/systemd-cryptsetup and
335 systemd-homed has been updated to allow explicit configuration of the
336 "user presence" and "user verification" checks, as well as whether a
337 PIN is required for authentication, via the new switches
338 --fido2-with-user-presence=, --fido2-with-user-verification=,
339 --fido2-with-client-pin= to systemd-cryptenroll and homectl. Which
340 features are available, and may be enabled or disabled depends on the
343 * systemd-nspawn's --private-user= switch now accepts the special value
344 "identity" which configures a user namespacing environment with an
345 identity mapping of 65535 UIDs. This means the container UID 0 is
346 mapped to the host UID 0, and the UID 1 to host UID 1. On first look
347 this doesn't appear to be useful, however it does reduce the attack
348 surface a bit, since the resulting container will possess process
349 capabilities only within its namespace and not on the host.
351 * systemd-nspawn's --private-user-chown switch has been replaced by a
352 more generic --private-user-ownership= switch that accepts one of
353 three values: "chown" is equivalent to the old --private-user-chown,
354 and "off" is equivalent to the absence of the old switch. The value
355 "map" uses the new UID mapping mounts of Linux 5.12 to map ownership
356 of files and directories of the underlying image to the chosen UID
357 range for the container. "auto" is equivalent to "map" if UID mapping
358 mount are supported, otherwise it is equivalent to "chown". The short
359 -U switch systemd-nspawn now implies --private-user-ownership=auto
360 instead of the old --private-user-chown. Effectively this means: if
361 the backing file system supports UID mapping mounts the feature is
362 now used by default if -U is used. Generally, it's a good idea to use
363 UID mapping mounts instead of recursive chown()ing, since it allows
364 running containers off immutable images (since no modifications of
365 the images need to take place), and share images between multiple
366 instances. Moreover, the recursive chown()ing operation is slow and
367 can be avoided. Conceptually it's also a good thing if transient UID
368 range uses do not leak into persistent file ownership anymore. TLDR:
369 finally, the last major drawback of user namespacing has been
370 removed, and -U should always be used (unless you use btrfs, where
371 UID mapped mounts do not exist; or your container actually needs
372 privileges on the host).
374 * nss-systemd now synthesizes user and group shadow records in addition
375 to the main user and group records. Thus, hashed passwords managed by
376 systemd-homed are now accessible via the shadow database.
378 * The userdb logic (and thus nss-systemd, and so on) now read
379 additional user/group definitions in JSON format from the drop-in
380 directories /etc/userdb/, /run/userdb/, /run/host/userdb/ and
381 /usr/lib/userdb/. This is a simple and powerful mechanism for making
382 additional users available to the system, with full integration into
383 NSS including the shadow databases. Since the full JSON user/group
384 record format is supported this may also be used to define users with
385 resource management settings and other runtime settings that
386 pam_systemd and systemd-logind enforce at login.
388 * The userdbctl tool gained two new switches --with-dropin= and
389 --with-varlink= which can be used to fine-tune the sources used for
390 user database lookups.
392 * systemd-nspawn gained a new switch --bind-user= for binding a host
393 user account into the container. This does three things: the user's
394 home directory is bind mounted from the host into the container,
395 below the /run/userdb/home/ hierarchy. A free UID is picked in the
396 container, and a user namespacing UID mapping to the host user's UID
397 installed. And finally, a minimal JSON user and group record (along
398 with its hashed password) is dropped into /run/host/userdb/. These
399 records are picked up automatically by the userdb drop-in logic
400 describe above, and allow the user to login with the same password as
401 on the host. Effectively this means: if host and container run new
402 enough systemd versions making a host user available to the container
405 * systemd-journal-gatewayd now supports the switches --user, --system,
406 --merge, --file= that are equivalent to the same switches of
407 journalctl, and permit exposing only the specified subset of the
410 * The OnFailure= dependency between units is now augmented with a
411 implicit reverse dependency OnFailureOf= (this new dependency cannot
412 be configured directly it's only created as effect of an OnFailure=
413 dependency in the reverse order — it's visible in "systemctl show"
414 however). Similar, Slice= now has an reverse dependency SliceOf=,
415 that is also not configurable directly, but useful to determine all
416 units that are members of a slice.
418 * A pair of new dependency types between units PropagatesStopTo= +
419 StopPropagatedFrom= has been added, that allows propagation of unit
420 stop events between two units. It operates similar to the existing
421 PropagatesReloadTo= + ReloadPropagatedFrom= dependencies.
423 * A new dependency type OnSuccess= has been added (plus the reverse
424 dependency OnSuccessOf=, which cannot be configured directly, but
425 exists only as effect of the reverse OnSuccess=). It is similar to
426 OnFailure=, but triggers in the opposite case: when a service exits
427 cleanly. This allows "chaining up" of services where one or more
428 services are started once another service has successfully completed.
430 * A new dependency type Upholds= has been added (plus the reverse
431 dependency UpheldBy=, which cannot be configured directly, but exists
432 only as effect of Upholds=). This dependency type is a stronger form
433 of Wants=: if a unit has an UpHolds= dependency on some other unit
434 and the former is active then the latter is started whenever it is
435 found inactive (and no job is queued for it). This is an alternative
436 to Restart= inside service units, but less configurable, and the
437 request to uphold a unit is not encoded in the unit itself but in
438 another unit that intends to uphold it.
440 * The systemd-ask-password tool now also supports reading passwords
441 from the credentials subsystem, via the new --credential= switch.
443 * The systemd-ask-password tool learnt a new switch --emoji= which may
444 be used to explicit control whether the lock and key emoji (🔐) is
445 shown in the password prompt on suitable TTYs.
447 * The --echo switch of systemd-ask-password now optionally takes a
448 parameter that controls character echo. It may either show asterisks
449 (default, as before), turn echo off entirely, or echo the typed
450 characters literally.
452 * The systemd-ask-password tool also gained a new -n switch for
453 suppressing output of a trailing newline character when writing the
454 acquired password to standard output, similar to /bin/echo's -n
457 * New documentation has been added that describes the organization of
458 the systemd source code tree:
460 https://systemd.io/ARCHITECTURE
462 * Units using ConditionNeedsUpdate= will no longer be activated in
465 * It is now possible to list a template unit in the WantedBy= or
466 RequiredBy= settings of the [Install] section of another template
467 unit, which will be instantiated using the same instance name.
469 * A new MemoryAvailable property is available for units. If the unit,
470 or the slice(s) it is part of, have a memory limit set via MemoryMax=/
471 MemoryHigh=, MemoryAvailable will indicate how much more memory the
472 unit can claim before hitting the limit(s).
474 * systemd-coredump will now try to stay below the cgroup memory limit
475 placed on itself or one of the slices it runs under, if the storage
476 area for core files (/var/lib/systemd/coredump/) is placed on a tmpfs,
477 since files written on such filesystems count toward the cgroup memory
478 limit. If there is not enough available memory in such cases to store
479 the core file uncompressed, systemd-coredump will skip to compressed
480 storage directly (if enabled) and it will avoid analyzing the core file
481 to print backtrace and metadata in the journal.
483 * tmpfiles.d/ drop-ins gained a new '=' modifier to check if the type
484 of a path matches the configured expectations, and remove it if not.
486 * tmpfiles.d/'s 'Age' now accepts an 'age-by' argument, which allows to
487 specify which of the several available filesystem timestamps (access
488 time, birth time, change time, modification time) to look at when
489 deciding whether a path has aged enough to be cleaned.
491 * A new IPv6StableSecretAddress= setting has been added to .network
492 files, which takes an IPv6 address to use as secret for IPv6 address
495 * The [DHCPServer] logic in .network files gained support for a new
496 UplinkInterface= setting that permits configuration of the uplink
497 interface name to propagate DHCP lease information from.
499 * The WakeOnLan= setting in .link files now accepts a list of flags
500 instead of a single one, to configure multiple wake-on-LAN policies.
502 * User-space defined tracepoints (USDT) have been added to udev at
503 strategic locations. This is useful for tracing udev behaviour and
504 performance with bpftrace and similar tools.
506 * systemd-journald-upload gained a new NetworkTimeoutSec= option for
507 setting a network timeout time.
509 * If a system service is running in a new mount namespace (RootDirectory=
510 and friends), all file systems will be mounted with MS_NOSUID by
511 default, unless the system is running with SELinux enabled.
513 * When enumerating time zones the timedatectl tool will now consult the
514 'tzdata.zi' file shipped by the IANA time zone database package, in
515 addition to 'zone1970.tab', as before. This makes sure time zone
516 aliases are now correctly supported. Some distributions so far did
517 not install this additional file, most do however. If you
518 distribution does not install it yet, it might make sense to change
521 Contributions from: Aakash Singh, adrian5, Albert Brox,
522 Alexander Sverdlin, Alexander Tsoy, Alexey Rubtsov, alexlzhu,
523 Allen Webb, Alvin Šipraga, Alyssa Ross, Anders Wenhaug,
524 Andrea Pappacoda, Anita Zhang, asavah, Balint Reczey, Bertrand Jacquin,
525 borna-blazevic, caoxia2008cxx, Carlo Teubner, Christian Göttsche,
526 Christian Hesse, Daniel Schaefer, Dan Streetman,
527 David Santamaría Rogado, David Tardon, Deepak Rawat, dgcampea,
528 Dimitri John Ledkov, ei-ke, Emilio Herrera, Emil Renner Berthing,
529 Eric Cook, Flos Lonicerae, Franck Bui, Francois Gervais,
530 Frantisek Sumsal, Gibeom Gwon, gitm0, Hamish Moffatt, Hans de Goede,
531 Harsh Barsaiyan, Henri Chain, Hristo Venev, Icenowy Zheng, Igor Zhbanov,
532 imayoda, Jakub Warczarek, James Buren, Jan Janssen, Jan Macku,
533 Jan Synacek, Jason Francis, Jayanth Ananthapadmanaban, Jeremy Szu,
534 Jérôme Carretero, Jesse Stricker, jiangchuangang, Joerg Behrmann,
535 Jóhann B. Guðmundsson, Jörg Deckert, Jörg Thalheim, Juergen Hoetzel,
536 Julia Kartseva, Kai-Heng Feng, Khem Raj, KoyamaSohei, laineantti,
537 Lennart Poettering, LetzteInstanz, Luca Adrian L, Luca Boccassi,
538 Lucas Magasweran, Mantas Mikulėnas, Marco Antonio Mauro, Mark Wielaard,
539 Masahiro Matsuya, Matt Johnston, Michael Catanzaro, Michal Koutný,
540 Michal Sekletár, Mike Crowe, Mike Kazantsev, Milan, milaq,
541 Miroslav Suchý, Morten Linderud, nerdopolis, nl6720, Noah Meyerhans,
542 Oleg Popov, Olle Lundberg, Ondrej Kozina, Paweł Marciniak, Perry.Yuan,
543 Peter Hutterer, Peter Kjellerstedt, Peter Morrow, Phaedrus Leeds,
544 plattrap, qhill, Raul Tambre, Roman Beranek, Roshan Shariff,
545 Ryan Hendrickson, Samuel BF, scootergrisen, Sebastian Blunt,
546 Seong-ho Cho, Sergey Bugaev, Sevan Janiyan, Sibo Dong, simmon,
547 Simon Watts, Srinidhi Kaushik, Štěpán Němec, Steve Bonds, Susant Sahani,
548 sverdlin, syyhao1994, Takashi Sakamoto, Topi Miettinen, tramsay,
549 Trent Piepho, Uwe Kleine-König, Viktor Mihajlovski, Vincent Dechenaux,
550 Vito Caputo, William A. Kennington III, Yangyang Shen, Yegor Alexeyev,
551 Yi Gao, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, zsien, наб
553 — Edinburgh, 2021-07-07
557 * A concept of system extension images is introduced. Such images may
558 be used to extend the /usr/ and /opt/ directory hierarchies at
559 runtime with additional files (even if the file system is read-only).
560 When a system extension image is activated, its /usr/ and /opt/
561 hierarchies and os-release information are combined via overlayfs
562 with the file system hierarchy of the host OS.
564 A new systemd-sysext tool can be used to merge, unmerge, list, and
565 refresh system extension hierarchies. See
566 https://www.freedesktop.org/software/systemd/man/systemd-sysext.html.
568 The systemd-sysext.service automatically merges installed system
569 extensions during boot (before basic.target, but not in very early
570 boot, since various file systems have to be mounted first).
572 The SYSEXT_LEVEL= field in os-release(5) may be used to specify the
573 supported system extension level.
575 * A new ExtensionImages= unit setting can be used to apply the same
576 system extension image concept from systemd-sysext to the namespaced
577 file hierarchy of specific services, following the same rules and
580 * Support for a new special "root=tmpfs" kernel command-line option has
581 been added. When specified, a tmpfs is mounted on /, and mount.usr=
582 should be used to point to the operating system implementation.
584 * A new configuration file /etc/veritytab may be used to configure
585 dm-verity integrity protection for block devices. Each line is in the
586 format "volume-name data-device hash-device roothash options",
587 similar to /etc/crypttab.
589 * A new kernel command-line option systemd.verity.root_options= may be
590 used to configure dm-verity behaviour for the root device.
592 * The key file specified in /etc/crypttab (the third field) may now
593 refer to an AF_UNIX/SOCK_STREAM socket in the file system. The key is
594 acquired by connecting to that socket and reading from it. This
595 allows the implementation of a service to provide key information
596 dynamically, at the moment when it is needed.
598 * When the hostname is set explicitly to "localhost", systemd-hostnamed
599 will respect this. Previously such a setting would be mostly silently
600 ignored. The goal is to honour configuration as specified by the
603 * The fallback hostname that will be used by the system manager and
604 systemd-hostnamed can now be configured in two new ways: by setting
605 DEFAULT_HOSTNAME= in os-release(5), or by setting
606 $SYSTEMD_DEFAULT_HOSTNAME in the environment block. As before, it can
607 also be configured during compilation. The environment variable is
608 intended for testing and local overrides, the os-release(5) field is
609 intended to allow customization by different variants of a
610 distribution that share the same compiled packages.
612 * The environment block of the manager itself may be configured through
613 a new ManagerEnvironment= setting in system.conf or user.conf. This
614 complements existing ways to set the environment block (the kernel
615 command line for the system manager, the inherited environment and
616 user@.service unit file settings for the user manager).
618 * systemd-hostnamed now exports the default hostname and the source of
619 the configured hostname ("static", "transient", or "default") as
622 * systemd-hostnamed now exports the "HardwareVendor" and
623 "HardwareModel" D-Bus properties, which are supposed to contain a
624 pair of cleaned up, human readable strings describing the system's
625 vendor and model. It's typically sourced from the firmware's DMI
626 tables, but may be augmented from a new hwdb database. hostnamectl
627 shows this in the status output.
629 * Support has been added to systemd-cryptsetup for extracting the
630 PKCS#11 token URI and encrypted key from the LUKS2 JSON embedded
631 metadata header. This allows the information how to open the
632 encrypted device to be embedded directly in the device and obviates
633 the need for configuration in an external file.
635 * systemd-cryptsetup gained support for unlocking LUKS2 volumes using
636 TPM2 hardware, as well as FIDO2 security tokens (in addition to the
637 pre-existing support for PKCS#11 security tokens).
639 * systemd-repart may enroll encrypted partitions using TPM2
640 hardware. This may be useful for example to create an encrypted /var
641 partition bound to the machine on first boot.
643 * A new systemd-cryptenroll tool has been added to enroll TPM2, FIDO2
644 and PKCS#11 security tokens to LUKS volumes, list and destroy
647 http://0pointer.net/blog/unlocking-luks2-volumes-with-tpm2-fido2-pkcs11-security-hardware-on-systemd-248.html
649 It also supports enrolling "recovery keys" and regular passphrases.
651 * The libfido2 dependency is now based on dlopen(), so that the library
652 is used at runtime when installed, but is not a hard runtime
655 * systemd-cryptsetup gained support for two new options in
656 /etc/crypttab: "no-write-workqueue" and "no-read-workqueue" which
657 request synchronous processing of encryption/decryption IO.
659 * The manager may be configured at compile time to use the fexecve()
660 instead of the execve() system call when spawning processes. Using
661 fexecve() closes a window between checking the security context of an
662 executable and spawning it, but unfortunately the kernel displays
663 stale information in the process' "comm" field, which impacts ps
666 * The configuration option -Dcompat-gateway-hostname has been dropped.
667 "_gateway" is now the only supported name.
669 * The ConditionSecurity=tpm2 unit file setting may be used to check if
670 the system has at least one TPM2 (tpmrm class) device.
672 * A new ConditionCPUFeature= has been added that may be used to
673 conditionalize units based on CPU features. For example,
674 ConditionCPUFeature=rdrand will condition a unit so that it is only
675 run when the system CPU supports the RDRAND opcode.
677 * The existing ConditionControlGroupController= setting has been
678 extended with two new values "v1" and "v2". "v2" means that the
679 unified v2 cgroup hierarchy is used, and "v1" means that legacy v1
680 hierarchy or the hybrid hierarchy are used.
682 * A new PrivateIPC= setting on a unit file allows executed processes to
683 be moved into a private IPC namespace, with separate System V IPC
684 identifiers and POSIX message queues.
686 A new IPCNamespacePath= allows the unit to be joined to an existing
689 * The tables of system calls in seccomp filters are now automatically
690 generated from kernel lists exported on
691 https://fedora.juszkiewicz.com.pl/syscalls.html.
693 The following architectures should now have complete lists:
694 alpha, arc, arm64, arm, i386, ia64, m68k, mips64n32, mips64, mipso32,
695 powerpc, powerpc64, s390, s390x, tilegx, sparc, x86_64, x32.
697 * The MountAPIVFS= service file setting now additionally mounts a tmpfs
698 on /run/ if it is not already a mount point. A writable /run/ has
699 always been a requirement for a functioning system, but this was not
700 guaranteed when using a read-only image.
702 Users can always specify BindPaths= or InaccessiblePaths= as
703 overrides, and they will take precedence. If the host's root mount
704 point is used, there is no change in behaviour.
706 * New bind mounts and file system image mounts may be injected into the
707 mount namespace of a service (without restarting it). This is exposed
708 respectively as 'systemctl bind <unit> <path>…' and
709 'systemctl mount-image <unit> <image>…'.
711 * The StandardOutput= and StandardError= settings can now specify files
712 to be truncated for output (as "truncate:<path>").
714 * The ExecPaths= and NoExecPaths= settings may be used to specify
715 noexec for parts of the file system.
717 * sd-bus has a new function sd_bus_open_user_machine() to open a
718 connection to the session bus of a specific user in a local container
719 or on the local host. This is exposed in the existing -M switch to
720 systemctl and similar tools:
722 systemctl --user -M lennart@foobar start foo
724 This will connect to the user bus of a user "lennart" in container
725 "foobar". If no container name is specified, the specified user on
726 the host itself is connected to
728 systemctl --user -M lennart@ start quux
730 * sd-bus also gained a convenience function sd_bus_message_send() to
731 simplify invocations of sd_bus_send(), taking only a single
732 parameter: the message to send.
734 * sd-event allows rate limits to be set on event sources, for dealing
735 with high-priority event sources that might starve out others. See
736 the new man page sd_event_source_set_ratelimit(3) for details.
738 * systemd.link files gained a [Link] Promiscuous= switch, which allows
739 the device to be raised in promiscuous mode.
741 New [Link] TransmitQueues= and ReceiveQueues= settings allow the
742 number of TX and RX queues to be configured.
744 New [Link] TransmitQueueLength= setting allows the size of the TX
745 queue to be configured.
747 New [Link] GenericSegmentOffloadMaxBytes= and
748 GenericSegmentOffloadMaxSegments= allow capping the packet size and
749 the number of segments accepted in Generic Segment Offload.
751 * systemd-networkd gained support for the "B.A.T.M.A.N. advanced"
752 wireless routing protocol that operates on ISO/OSI Layer 2 only and
753 uses ethernet frames to route/bridge packets. This encompasses a new
754 "batadv" netdev Type=, a new [BatmanAdvanced] section with a bunch of
755 new settings in .netdev files, and a new BatmanAdvanced= setting in
758 * systemd.network files gained a [Network] RouteTable= configuration
759 switch to select the routing policy table.
761 systemd.network files gained a [RoutingPolicyRule] Type=
762 configuration switch (one of "blackhole, "unreachable", "prohibit").
764 systemd.network files gained a [IPv6AcceptRA] RouteDenyList= and
765 RouteAllowList= settings to ignore/accept route advertisements from
766 routers matching specified prefixes. The DenyList= setting has been
767 renamed to PrefixDenyList= and a new PrefixAllowList= option has been
770 systemd.network files gained a [DHCPv6] UseAddress= setting to
771 optionally ignore the address provided in the lease.
773 systemd.network files gained a [DHCPv6PrefixDelegation]
774 ManageTemporaryAddress= switch.
776 systemd.network files gained a new ActivationPolicy= setting which
777 allows configuring how the UP state of an interface shall be managed,
778 i.e. whether the interface is always upped, always downed, or may be
779 upped/downed by the user using "ip link set dev".
781 * The default for the Broadcast= setting in .network files has slightly
782 changed: the broadcast address will not be configured for wireguard
785 * systemd.netdev files gained a [VLAN] Protocol=, IngressQOSMaps=,
786 EgressQOSMaps=, and [MACVLAN] BroadcastMulticastQueueLength=
787 configuration options for VLAN packet handling.
789 * udev rules may now set log_level= option. This allows debug logs to
790 be enabled for select events, e.g. just for a specific subsystem or
791 even a single device.
793 * udev now exports the VOLUME_ID, LOGICAL_VOLUME_ID, VOLUME_SET_ID, and
794 DATA_PREPARED_ID properties for block devices with ISO9660 file
797 * udev now exports decoded DMI information about installed memory slots
798 as device properties under the /sys/class/dmi/id/ pseudo device.
800 * /dev/ is not mounted noexec anymore. This didn't provide any
801 significant security benefits and would conflict with the executable
802 mappings used with /dev/sgx device nodes. The previous behaviour can
803 be restored for individual services with NoExecPaths=/dev (or by allow-
804 listing and excluding /dev from ExecPaths=).
806 * Permissions for /dev/vsock are now set to 0o666, and /dev/vhost-vsock
807 and /dev/vhost-net are owned by the kvm group.
809 * The hardware database has been extended with a list of fingerprint
810 readers that correctly support USB auto-suspend using data from
813 * systemd-resolved can now answer DNSSEC questions through the stub
814 resolver interface in a way that allows local clients to do DNSSEC
815 validation themselves. For a question with DO+CD set, it'll proxy the
816 DNS query and respond with a mostly unmodified packet received from
819 * systemd-resolved learnt a new boolean option CacheFromLocalhost= in
820 resolved.conf. If true the service will provide caching even for DNS
821 lookups made to an upstream DNS server on the 127.0.0.1/::1
822 addresses. By default (and when the option is false) systemd-resolved
823 will not cache such lookups, in order to avoid duplicate local
824 caching, under the assumption the local upstream server caches
827 * systemd-resolved now implements RFC5001 NSID in its local DNS
828 stub. This may be used by local clients to determine whether they are
829 talking to the DNS resolver stub or a different DNS server.
831 * When resolving host names and other records resolvectl will now
832 report where the data was acquired from (i.e. the local cache, the
833 network, locally synthesized, …) and whether the network traffic it
834 effected was encrypted or not. Moreover the tool acquired a number of
835 new options --cache=, --synthesize=, --network=, --zone=,
836 --trust-anchor=, --validate= that take booleans and may be used to
837 tweak a lookup, i.e. whether it may be answered from cached
838 information, locally synthesized information, information acquired
839 through the network, the local mDNS/LLMNR zone, the DNSSEC trust
840 anchor, and whether DNSSEC validation shall be executed for the
843 * systemd-nspawn gained a new --ambient-capability= setting
844 (AmbientCapability= in .nspawn files) to configure ambient
845 capabilities passed to the container payload.
847 * systemd-nspawn gained the ability to configure the firewall using the
848 nftables subsystem (in addition to the existing iptables
849 support). Similarly, systemd-networkd's IPMasquerade= option now
850 supports nftables as back-end, too. In both cases NAT on IPv6 is now
851 supported too, in addition to IPv4 (the iptables back-end still is
854 "IPMasquerade=yes", which was the same as "IPMasquerade=ipv4" before,
855 retains its meaning, but has been deprecated. Please switch to either
856 "ivp4" or "both" (if covering IPv6 is desired).
858 * systemd-importd will now download .verity and .roothash.p7s files
859 along with the machine image (as exposed via machinectl pull-raw).
861 * systemd-oomd now gained a new DefaultMemoryPressureDurationSec=
862 setting to configure the time a unit's cgroup needs to exceed memory
863 pressure limits before action will be taken, and a new
864 ManagedOOMPreference=none|avoid|omit setting to avoid killing certain
867 systemd-oomd is now considered fully supported (the usual
868 backwards-compatiblity promises apply). Swap is not required for
869 operation, but it is still recommended.
871 * systemd-timesyncd gained a new ConnectionRetrySec= setting which
872 configures the retry delay when trying to contact servers.
874 * systemd-stdio-bridge gained --system/--user options to connect to the
875 system bus (previous default) or the user session bus.
877 * systemd-localed may now call locale-gen to generate missing locales
878 on-demand (UTF-8-only). This improves integration with Debian-based
879 distributions (Debian/Ubuntu/PureOS/Tanglu/...) and Arch Linux.
881 * systemctl --check-inhibitors=true may now be used to obey inhibitors
882 even when invoked non-interactively. The old --ignore-inhibitors
883 switch is now deprecated and replaced by --check-inhibitors=false.
885 * systemctl import-environment will now emit a warning when called
886 without any arguments (i.e. to import the full environment block of
887 the called program). This command will usually be invoked from a
888 shell, which means that it'll inherit a bunch of variables which are
889 specific to that shell, and usually to the TTY the shell is connected
890 to, and don't have any meaning in the global context of the system or
891 user service manager. Instead, only specific variables should be
892 imported into the manager environment block.
894 Similarly, programs which update the manager environment block by
895 directly calling the D-Bus API of the manager, should also push
896 specific variables, and not the full inherited environment.
898 * systemctl's status output now shows unit state with a more careful
899 choice of Unicode characters: units in maintenance show a "○" symbol
900 instead of the usual "●", failed units show "×", and services being
903 * coredumpctl gained a --debugger-arguments= switch to pass arguments
904 to the debugger. It also gained support for showing coredump info in
905 a simple JSON format.
907 * systemctl/loginctl/machinectl's --signal= option now accept a special
908 value "list", which may be used to show a brief table with known
909 process signals and their numbers.
911 * networkctl now shows the link activation policy in status.
913 * Various tools gained --pager/--no-pager/--json= switches to
914 enable/disable the pager and provide JSON output.
916 * Various tools now accept two new values for the SYSTEMD_COLORS
917 environment variable: "16" and "256", to configure how many terminal
918 colors are used in output.
920 * less 568 or newer is now required for the auto-paging logic of the
921 various tools. Hyperlink ANSI sequences in terminal output are now
922 used even if a pager is used, and older versions of less are not able
923 to display these sequences correctly. SYSTEMD_URLIFY=0 may be used to
924 disable this output again.
926 * Builds with support for separate / and /usr/ hierarchies ("split-usr"
927 builds, non-merged-usr builds) are now officially deprecated. A
928 warning is emitted during build. Support is slated to be removed in
929 about a year (when the Debian Bookworm release development starts).
931 * Systems with the legacy cgroup v1 hierarchy are now marked as
932 "tainted", to make it clearer that using the legacy hierarchy is not
935 * systemd-localed will now refuse to configure a keymap which is not
936 installed in the file system. This is intended as a bug fix, but
937 could break cases where systemd-localed was used to configure the
938 keymap in advanced of it being installed. It is necessary to install
939 the keymap file first.
941 * The main git development branch has been renamed to 'main'.
943 * mmcblk[0-9]boot[0-9] devices will no longer be probed automatically
944 for partitions, as in the vast majority of cases they contain none
945 and are used internally by the bootloader (eg: uboot).
947 * systemd will now set the $SYSTEMD_EXEC_PID environment variable for
948 spawned processes to the PID of the process itself. This may be used
949 by programs for detecting whether they were forked off by the service
950 manager itself or are a process forked off further down the tree.
952 * The sd-device API gained four new calls: sd_device_get_action() to
953 determine the uevent add/remove/change/… action the device object has
954 been seen for, sd_device_get_seqno() to determine the uevent sequence
955 number, sd_device_new_from_stat_rdev() to allocate a new sd_device
956 object from stat(2) data of a device node, and sd_device_trigger() to
957 write to the 'uevent' attribute of a device.
959 * For most tools the --no-legend= switch has been replaced by
960 --legend=no and --legend=yes, to force whether tables are shown with
963 * Units acquired a new property "Markers" that takes a list of zero,
964 one or two of the following strings: "needs-reload" and
965 "needs-restart". These markers may be set via "systemctl
966 set-property". Once a marker is set, "systemctl reload-or-restart
967 --marked" may be invoked to execute the operation the units are
968 marked for. This is useful for package managers that want to mark
969 units for restart/reload while updating, but effect the actual
970 operations at a later step at once.
972 * The sd_bus_message_read_strv() API call of sd-bus may now also be
973 used to parse arrays of D-Bus signatures and D-Bus paths, in addition
976 * bootctl will now report whether the UEFI firmware used a TPM2 device
977 and measured the boot process into it.
979 * systemd-tmpfiles learnt support for a new environment variable
980 $SYSTEMD_TMPFILES_FORCE_SUBVOL which takes a boolean value. If true
981 the v/q/Q lines in tmpfiles.d/ snippets will create btrfs subvolumes
982 even if the root fs of the system is not itself a btrfs volume.
984 * systemd-detect-virt/ConditionVirtualization= will now explicitly
985 detect Docker/Podman environments where possible. Moreover, they
986 should be able to generically detect any container manager as long as
987 it assigns the container a cgroup.
989 * portablectl gained a new "reattach" verb for detaching/reattaching a
990 portable service image, useful for updating images on-the-fly.
992 * Intel SGX enclave device nodes (which expose a security feature of
993 newer Intel CPUs) will now be owned by a new system group "sgx".
995 Contributions from: Adam Nielsen, Adrian Vovk, AJ Jordan, Alan Perry,
996 Alastair Pharo, Alexander Batischev, Ali Abdallah, Andrew Balmos,
997 Anita Zhang, Annika Wickert, Ansgar Burchardt, Antonio Terceiro,
998 Antonius Frie, Ardy, Arian van Putten, Ariel Fermani, Arnaud T,
999 A S Alam, Bastien Nocera, Benjamin Berg, Benjamin Robin, Björn Daase,
1000 caoxia, Carlo Wood, Charles Lee, ChopperRob, chri2, Christian Ehrhardt,
1001 Christian Hesse, Christopher Obbard, clayton craft, corvusnix, cprn,
1002 Daan De Meyer, Daniele Medri, Daniel Rusek, Dan Sanders, Dan Streetman,
1003 Darren Ng, David Edmundson, David Tardon, Deepak Rawat, Devon Pringle,
1004 Dmitry Borodaenko, dropsignal, Einsler Lee, Endre Szabo,
1005 Evgeny Vereshchagin, Fabian Affolter, Fangrui Song, Felipe Borges,
1006 feliperodriguesfr, Felix Stupp, Florian Hülsmann, Florian Klink,
1007 Florian Westphal, Franck Bui, Frantisek Sumsal, Gablegritule,
1008 Gaël PORTAY, Gaurav, Giedrius Statkevičius, Greg Depoire-Ferrer,
1009 Gustavo Costa, Hans de Goede, Hela Basa, heretoenhance, hide,
1010 Iago López Galeiras, igo95862, Ilya Dmitrichenko, Jameer Pathan,
1011 Jan Tojnar, Jiehong, Jinyuan Si, Joerg Behrmann, John Slade,
1012 Jonathan G. Underwood, Jonathan McDowell, Josh Triplett, Joshua Watt,
1013 Julia Cartwright, Julien Humbert, Kairui Song, Karel Zak,
1014 Kevin Backhouse, Kevin P. Fleming, Khem Raj, Konomi, krissgjeng,
1015 l4gfcm, Lajos Veres, Lennart Poettering, Lincoln Ramsay, Luca Boccassi,
1016 Luca BRUNO, Lucas Werkmeister, Luka Kudra, Luna Jernberg,
1017 Marc-André Lureau, Martin Wilck, Matthias Klumpp, Matt Turner,
1018 Michael Gisbers, Michael Marley, Michael Trapp, Michal Fabik,
1019 Michał Kopeć, Michal Koutný, Michal Sekletár, Michele Guerini Rocco,
1020 Mike Gilbert, milovlad, moson-mo, Nick, nihilix-melix, Oğuz Ersen,
1021 Ondrej Mosnacek, pali, Pavel Hrdina, Pavel Sapezhko, Perry Yuan,
1022 Peter Hutterer, Pierre Dubouilh, Piotr Drąg, Pjotr Vertaalt,
1023 Richard Laager, RussianNeuroMancer, Sam Lunt, Sebastiaan van Stijn,
1024 Sergey Bugaev, shenyangyang4, simmon, Simonas Kazlauskas,
1025 Slimane Selyan Amiri, Stefan Agner, Steve Ramage, Susant Sahani,
1026 Sven Mueller, Tad Fisher, Takashi Iwai, Thomas Haller, Tom Shield,
1027 Topi Miettinen, Torsten Hilbrich, tpgxyz, Tyler Hicks, ulf-f,
1028 Ulrich Ölmann, Vincent Pelletier, Vinnie Magro, Vito Caputo, Vlad,
1029 walbit-de, Whired Planck, wouter bolsterlee, Xℹ Ruoyao, Yangyang Shen,
1030 Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek,
1031 Zmicer Turok, Дамјан Георгиевски
1033 — Berlin, 2021-03-30
1037 * KERNEL API INCOMPATIBILITY: Linux 4.14 introduced two new uevents
1038 "bind" and "unbind" to the Linux device model. When this kernel
1039 change was made, systemd-udevd was only minimally updated to handle
1040 and propagate these new event types. The introduction of these new
1041 uevents (which are typically generated for USB devices and devices
1042 needing a firmware upload before being functional) resulted in a
1043 number of issues which we so far didn't address. We hoped the kernel
1044 maintainers would themselves address these issues in some form, but
1045 that did not happen. To handle them properly, many (if not most) udev
1046 rules files shipped in various packages need updating, and so do many
1047 programs that monitor or enumerate devices with libudev or sd-device,
1048 or otherwise process uevents. Please note that this incompatibility
1049 is not fault of systemd or udev, but caused by an incompatible kernel
1050 change that happened back in Linux 4.14, but is becoming more and
1051 more visible as the new uevents are generated by more kernel drivers.
1053 To minimize issues resulting from this kernel change (but not avoid
1054 them entirely) starting with systemd-udevd 247 the udev "tags"
1055 concept (which is a concept for marking and filtering devices during
1056 enumeration and monitoring) has been reworked: udev tags are now
1057 "sticky", meaning that once a tag is assigned to a device it will not
1058 be removed from the device again until the device itself is removed
1059 (i.e. unplugged). This makes sure that any application monitoring
1060 devices that match a specific tag is guaranteed to both see uevents
1061 where the device starts being relevant, and those where it stops
1062 being relevant (the latter now regularly happening due to the new
1063 "unbind" uevent type). The udev tags concept is hence now a concept
1064 tied to a *device* instead of a device *event* — unlike for example
1065 udev properties whose lifecycle (as before) is generally tied to a
1066 device event, meaning that the previously determined properties are
1067 forgotten whenever a new uevent is processed.
1069 With the newly redefined udev tags concept, sometimes it's necessary
1070 to determine which tags are the ones applied by the most recent
1071 uevent/database update, in order to discern them from those
1072 originating from earlier uevents/database updates of the same
1073 device. To accommodate for this a new automatic property CURRENT_TAGS
1074 has been added that works similar to the existing TAGS property but
1075 only lists tags set by the most recent uevent/database
1076 update. Similarly, the libudev/sd-device API has been updated with
1077 new functions to enumerate these 'current' tags, in addition to the
1078 existing APIs that now enumerate the 'sticky' ones.
1080 To properly handle "bind"/"unbind" on Linux 4.14 and newer it is
1081 essential that all udev rules files and applications are updated to
1082 handle the new events. Specifically:
1084 • All rule files that currently use a header guard similar to
1085 ACTION!="add|change",GOTO="xyz_end" should be updated to use
1086 ACTION=="remove",GOTO="xyz_end" instead, so that the
1087 properties/tags they add are also applied whenever "bind" (or
1088 "unbind") is seen. (This is most important for all physical device
1089 types — those for which "bind" and "unbind" are currently
1090 generated, for all other device types this change is still
1091 recommended but not as important — but certainly prepares for
1092 future kernel uevent type additions).
1094 • Similarly, all code monitoring devices that contains an 'if' branch
1095 discerning the "add" + "change" uevent actions from all other
1096 uevents actions (i.e. considering devices only relevant after "add"
1097 or "change", and irrelevant on all other events) should be reworked
1098 to instead negatively check for "remove" only (i.e. considering
1099 devices relevant after all event types, except for "remove", which
1100 invalidates the device). Note that this also means that devices
1101 should be considered relevant on "unbind", even though conceptually
1102 this — in some form — invalidates the device. Since the precise
1103 effect of "unbind" is not generically defined, devices should be
1104 considered relevant even after "unbind", however I/O errors
1105 accessing the device should then be handled gracefully.
1107 • Any code that uses device tags for deciding whether a device is
1108 relevant or not most likely needs to be updated to use the new
1109 udev_device_has_current_tag() API (or sd_device_has_current_tag()
1110 in case sd-device is used), to check whether the tag is set at the
1111 moment an uevent is seen (as opposed to the existing
1112 udev_device_has_tag() API which checks if the tag ever existed on
1113 the device, following the API concept redefinition explained
1116 We are very sorry for this breakage and the requirement to update
1117 packages using these interfaces. We'd again like to underline that
1118 this is not caused by systemd/udev changes, but result of a kernel
1121 * UPCOMING INCOMPATIBILITY: So far most downstream distribution
1122 packages have not retriggered devices once the udev package (or any
1123 auxiliary package installing additional udev rules) is updated. We
1124 intend to work with major distributions to change this, so that
1125 "udevadm trigger -a change" is issued on such upgrades, ensuring that
1126 the updated ruleset is applied to the devices already discovered, so
1127 that (asynchronously) after the upgrade completed the udev database
1128 is consistent with the updated rule set. This means udev rules must
1129 be ready to be retriggered with a "change" action any time, and
1130 result in correct and complete udev database entries. While the
1131 majority of udev rule files known to us currently get this right,
1132 some don't. Specifically, there are udev rules files included in
1133 various packages that only set udev properties on the "add" action,
1134 but do not handle the "change" action. If a device matching those
1135 rules is retriggered with the "change" action (as is intended here)
1136 it would suddenly lose the relevant properties. This always has been
1137 problematic, but as soon as all udev devices are triggered on relevant
1138 package upgrades this will become particularly so. It is strongly
1139 recommended to fix offending rules so that they can handle a "change"
1140 action at any time, and acquire all necessary udev properties even
1141 then. Or in other words: the header guard mentioned above
1142 (ACTION=="remove",GOTO="xyz_end") is the correct approach to handle
1143 this, as it makes sure rules are rerun on "change" correctly, and
1144 accumulate the correct and complete set of udev properties. udev rule
1145 definitions that cannot handle "change" events being triggered at
1146 arbitrary times should be considered buggy.
1148 * The MountAPIVFS= service file setting now defaults to on if
1149 RootImage= and RootDirectory= are used, which means that with those
1150 two settings /proc/, /sys/ and /dev/ are automatically properly set
1151 up for services. Previous behaviour may be restored by explicitly
1152 setting MountAPIVFS=off.
1154 * Since PAM 1.2.0 (2015) configuration snippets may be placed in
1155 /usr/lib/pam.d/ in addition to /etc/pam.d/. If a file exists in the
1156 latter it takes precedence over the former, similar to how most of
1157 systemd's own configuration is handled. Given that PAM stack
1158 definitions are primarily put together by OS vendors/distributions
1159 (though possibly overridden by users), this systemd release moves its
1160 own PAM stack configuration for the "systemd-user" PAM service (i.e.
1161 for the PAM session invoked by the per-user user@.service instance)
1162 from /etc/pam.d/ to /usr/lib/pam.d/. We recommend moving all
1163 packages' vendor versions of their PAM stack definitions from
1164 /etc/pam.d/ to /usr/lib/pam.d/, but if such OS-wide migration is not
1165 desired the location to which systemd installs its PAM stack
1166 configuration may be changed via the -Dpamconfdir Meson option.
1168 * The runtime dependencies on libqrencode, libpcre2, libidn/libidn2,
1169 libpwquality and libcryptsetup have been changed to be based on
1170 dlopen(): instead of regular dynamic library dependencies declared in
1171 the binary ELF headers, these libraries are now loaded on demand
1172 only, if they are available. If the libraries cannot be found the
1173 relevant operations will fail gracefully, or a suitable fallback
1174 logic is chosen. This is supposed to be useful for general purpose
1175 distributions, as it allows minimizing the list of dependencies the
1176 systemd packages pull in, permitting building of more minimal OS
1177 images, while still making use of these "weak" dependencies should
1178 they be installed. Since many package managers automatically
1179 synthesize package dependencies from ELF shared library dependencies,
1180 some additional manual packaging work has to be done now to replace
1181 those (slightly downgraded from "required" to "recommended" or
1182 whatever is conceptually suitable for the package manager). Note that
1183 this change does not alter build-time behaviour: as before the
1184 build-time dependencies have to be installed during build, even if
1185 they now are optional during runtime.
1187 * sd-event.h gained a new call sd_event_add_time_relative() for
1188 installing timers relative to the current time. This is mostly a
1189 convenience wrapper around the pre-existing sd_event_add_time() call
1190 which installs absolute timers.
1192 * sd-event event sources may now be placed in a new "exit-on-failure"
1193 mode, which may be controlled via the new
1194 sd_event_source_get_exit_on_failure() and
1195 sd_event_source_set_exit_on_failure() functions. If enabled, any
1196 failure returned by the event source handler functions will result in
1197 exiting the event loop (unlike the default behaviour of just
1198 disabling the event source but continuing with the event loop). This
1199 feature is useful to set for all event sources that define "primary"
1200 program behaviour (where failure should be fatal) in contrast to
1201 "auxiliary" behaviour (where failure should remain local).
1203 * Most event source types sd-event supports now accept a NULL handler
1204 function, in which case the event loop is exited once the event
1205 source is to be dispatched, using the userdata pointer — converted to
1206 a signed integer — as exit code of the event loop. Previously this
1207 was supported for IO and signal event sources already. Exit event
1208 sources still do not support this (simply because it makes little
1209 sense there, as the event loop is already exiting when they are
1212 * A new per-unit setting RootImageOptions= has been added which allows
1213 tweaking the mount options for any file system mounted as effect of
1214 the RootImage= setting.
1216 * Another new per-unit setting MountImages= has been added, that allows
1217 mounting additional disk images into the file system tree accessible
1220 * Timer units gained a new FixedRandomDelay= boolean setting. If
1221 enabled, the random delay configured with RandomizedDelaySec= is
1222 selected in a way that is stable on a given system (though still
1223 different for different units).
1225 * Socket units gained a new setting Timestamping= that takes "us", "ns"
1226 or "off". This controls the SO_TIMESTAMP/SO_TIMESTAMPNS socket
1229 * systemd-repart now generates JSON output when requested with the new
1232 * systemd-machined's OpenMachineShell() bus call will now pass
1233 additional policy metadata data fields to the PolicyKit
1234 authentication request.
1236 * systemd-tmpfiles gained a new -E switch, which is equivalent to
1237 --exclude-prefix=/dev --exclude-prefix=/proc --exclude=/run
1238 --exclude=/sys. It's particularly useful in combination with --root=,
1239 when operating on OS trees that do not have any of these four runtime
1240 directories mounted, as this means no files below these subtrees are
1241 created or modified, since those mount points should probably remain
1244 * systemd-tmpfiles gained a new --image= switch which is like --root=,
1245 but takes a disk image instead of a directory as argument. The
1246 specified disk image is mounted inside a temporary mount namespace
1247 and the tmpfiles.d/ drop-ins stored in the image are executed and
1248 applied to the image. systemd-sysusers similarly gained a new
1249 --image= switch, that allows the sysusers.d/ drop-ins stored in the
1250 image to be applied onto the image.
1252 * Similarly, the journalctl command also gained an --image= switch,
1253 which is a quick one-step solution to look at the log data included
1256 * journalctl's --output=cat option (which outputs the log content
1257 without any metadata, just the pure text messages) will now make use
1258 of terminal colors when run on a suitable terminal, similarly to the
1261 * JSON group records now support a "description" string that may be
1262 used to add a human-readable textual description to such groups. This
1263 is supposed to match the user's GECOS field which traditionally
1264 didn't have a counterpart for group records.
1266 * The "systemd-dissect" tool that may be used to inspect OS disk images
1267 and that was previously installed to /usr/lib/systemd/ has now been
1268 moved to /usr/bin/, reflecting its updated status of an officially
1269 supported tool with a stable interface. It gained support for a new
1270 --mkdir switch which when combined with --mount has the effect of
1271 creating the directory to mount the image to if it is missing
1272 first. It also gained two new commands --copy-from and --copy-to for
1273 copying files and directories in and out of an OS image without the
1274 need to manually mount it. It also acquired support for a new option
1275 --json= to generate JSON output when inspecting an OS image.
1277 * The cgroup2 file system is now mounted with the
1278 "memory_recursiveprot" mount option, supported since kernel 5.7. This
1279 means that the MemoryLow= and MemoryMin= unit file settings now apply
1280 recursively to whole subtrees.
1282 * systemd-homed now defaults to using the btrfs file system — if
1283 available — when creating home directories in LUKS volumes. This may
1284 be changed with the DefaultFileSystemType= setting in homed.conf.
1285 It's now the default file system in various major distributions and
1286 has the major benefit for homed that it can be grown and shrunk while
1287 mounted, unlike the other contenders ext4 and xfs, which can both be
1288 grown online, but not shrunk (in fact xfs is the technically most
1289 limited option here, as it cannot be shrunk at all).
1291 * JSON user records managed by systemd-homed gained support for
1292 "recovery keys". These are basically secondary passphrases that can
1293 unlock user accounts/home directories. They are computer-generated
1294 rather than user-chosen, and typically have greater entropy.
1295 homectl's --recovery-key= option may be used to add a recovery key to
1296 a user account. The generated recovery key is displayed as a QR code,
1297 so that it can be scanned to be kept in a safe place. This feature is
1298 particularly useful in combination with systemd-homed's support for
1299 FIDO2 or PKCS#11 authentication, as a secure fallback in case the
1300 security tokens are lost. Recovery keys may be entered wherever the
1301 system asks for a password.
1303 * systemd-homed now maintains a "dirty" flag for each LUKS encrypted
1304 home directory which indicates that a home directory has not been
1305 deactivated cleanly when offline. This flag is useful to identify
1306 home directories for which the offline discard logic did not run when
1307 offlining, and where it would be a good idea to log in again to catch
1310 * systemctl gained a new parameter --timestamp= which may be used to
1311 change the style in which timestamps are output, i.e. whether to show
1312 them in local timezone or UTC, or whether to show µs granularity.
1314 * Alibaba's "pouch" container manager is now detected by
1315 systemd-detect-virt, ConditionVirtualization= and similar
1316 constructs. Similar, they now also recognize IBM PowerVM machine
1319 * systemd-nspawn has been reworked to use the /run/host/incoming/ as
1320 place to use for propagating external mounts into the
1321 container. Similarly /run/host/notify is now used as the socket path
1322 for container payloads to communicate with the container manager
1323 using sd_notify(). The container manager now uses the
1324 /run/host/inaccessible/ directory to place "inaccessible" file nodes
1325 of all relevant types which may be used by the container payload as
1326 bind mount source to over-mount inodes to make them inaccessible.
1327 /run/host/container-manager will now be initialized with the same
1328 string as the $container environment variable passed to the
1329 container's PID 1. /run/host/container-uuid will be initialized with
1330 the same string as $container_uuid. This means the /run/host/
1331 hierarchy is now the primary way to make host resources available to
1332 the container. The Container Interface documents these new files and
1335 https://systemd.io/CONTAINER_INTERFACE
1337 * Support for the "ConditionNull=" unit file condition has been
1338 deprecated and undocumented for 6 years. systemd started to warn
1339 about its use 1.5 years ago. It has now been removed entirely.
1341 * sd-bus.h gained a new API call sd_bus_error_has_names(), which takes
1342 a sd_bus_error struct and a list of error names, and checks if the
1343 error matches one of these names. It's a convenience wrapper that is
1344 useful in cases where multiple errors shall be handled the same way.
1346 * A new system call filter list "@known" has been added, that contains
1347 all system calls known at the time systemd was built.
1349 * Behaviour of system call filter allow lists has changed slightly:
1350 system calls that are contained in @known will result in EPERM by
1351 default, while those not contained in it result in ENOSYS. This
1352 should improve compatibility because known system calls will thus be
1353 communicated as prohibited, while unknown (and thus newer ones) will
1354 be communicated as not implemented, which hopefully has the greatest
1355 chance of triggering the right fallback code paths in client
1358 * "systemd-analyze syscall-filter" will now show two separate sections
1359 at the bottom of the output: system calls known during systemd build
1360 time but not included in any of the filter groups shown above, and
1361 system calls defined on the local kernel but known during systemd
1364 * If the $SYSTEMD_LOG_SECCOMP=1 environment variable is set for
1365 systemd-nspawn all system call filter violations will be logged by
1366 the kernel (audit). This is useful for tracking down system calls
1367 invoked by container payloads that are prohibited by the container's
1368 system call filter policy.
1370 * If the $SYSTEMD_SECCOMP=0 environment variable is set for
1371 systemd-nspawn (and other programs that use seccomp) all seccomp
1372 filtering is turned off.
1374 * Two new unit file settings ProtectProc= and ProcSubset= have been
1375 added that expose the hidepid= and subset= mount options of procfs.
1376 All processes of the unit will only see processes in /proc that are
1377 are owned by the unit's user. This is an important new sandboxing
1378 option that is recommended to be set on all system services. All
1379 long-running system services that are included in systemd itself set
1380 this option now. This option is only supported on kernel 5.8 and
1381 above, since the hidepid= option supported on older kernels was not a
1382 per-mount option but actually applied to the whole PID namespace.
1384 * Socket units gained a new boolean setting FlushPending=. If enabled
1385 all pending socket data/connections are flushed whenever the socket
1386 unit enters the "listening" state, i.e. after the associated service
1389 * The unit file setting NUMAMask= gained a new "all" value: when used,
1390 all existing NUMA nodes are added to the NUMA mask.
1392 * A new "credentials" logic has been added to system services. This is
1393 a simple mechanism to pass privileged data to services in a safe and
1394 secure way. It's supposed to be used to pass per-service secret data
1395 such as passwords or cryptographic keys but also associated less
1396 private information such as user names, certificates, and similar to
1397 system services. Each credential is identified by a short user-chosen
1398 name and may contain arbitrary binary data. Two new unit file
1399 settings have been added: SetCredential= and LoadCredential=. The
1400 former allows setting a credential to a literal string, the latter
1401 sets a credential to the contents of a file (or data read from a
1402 user-chosen AF_UNIX stream socket). Credentials are passed to the
1403 service via a special credentials directory, one file for each
1404 credential. The path to the credentials directory is passed in a new
1405 $CREDENTIALS_DIRECTORY environment variable. Since the credentials
1406 are passed in the file system they may be easily referenced in
1407 ExecStart= command lines too, thus no explicit support for the
1408 credentials logic in daemons is required (though ideally daemons
1409 would look for the bits they need in $CREDENTIALS_DIRECTORY
1410 themselves automatically, if set). The $CREDENTIALS_DIRECTORY is
1411 backed by unswappable memory if privileges allow it, immutable if
1412 privileges allow it, is accessible only to the service's UID, and is
1413 automatically destroyed when the service stops.
1415 * systemd-nspawn supports the same credentials logic. It can both
1416 consume credentials passed to it via the aforementioned
1417 $CREDENTIALS_DIRECTORY protocol as well as pass these credentials on
1418 to its payload. The service manager/PID 1 has been updated to match
1419 this: it can also accept credentials from the container manager that
1420 invokes it (in fact: any process that invokes it), and passes them on
1421 to its services. Thus, credentials can be propagated recursively down
1422 the tree: from a system's service manager to a systemd-nspawn
1423 service, to the service manager that runs as container payload and to
1424 the service it runs below. Credentials may also be added on the
1425 systemd-nspawn command line, using new --set-credential= and
1426 --load-credential= command line switches that match the
1427 aforementioned service settings.
1429 * systemd-repart gained new settings Format=, Encrypt=, CopyFiles= in
1430 the partition drop-ins which may be used to format/LUKS
1431 encrypt/populate any created partitions. The partitions are
1432 encrypted/formatted/populated before they are registered in the
1433 partition table, so that they appear atomically: either the
1434 partitions do not exist yet or they exist fully encrypted, formatted,
1435 and populated — there is no time window where they are
1436 "half-initialized". Thus the system is robust to abrupt shutdown: if
1437 the tool is terminated half-way during its operations on next boot it
1438 will start from the beginning.
1440 * systemd-repart's --size= operation gained a new "auto" value. If
1441 specified, and operating on a loopback file it is automatically sized
1442 to the minimal size the size constraints permit. This is useful to
1443 use "systemd-repart" as an image builder for minimally sized images.
1445 * systemd-resolved now gained a third IPC interface for requesting name
1446 resolution: besides D-Bus and local DNS to 127.0.0.53 a Varlink
1447 interface is now supported. The nss-resolve NSS module has been
1448 modified to use this new interface instead of D-Bus. Using Varlink
1449 has a major benefit over D-Bus: it works without a broker service,
1450 and thus already during earliest boot, before the dbus daemon has
1451 been started. This means name resolution via systemd-resolved now
1452 works at the same time systemd-networkd operates: from earliest boot
1453 on, including in the initrd.
1455 * systemd-resolved gained support for a new DNSStubListenerExtra=
1456 configuration file setting which may be used to specify additional IP
1457 addresses the built-in DNS stub shall listen on, in addition to the
1458 main one on 127.0.0.53:53.
1460 * Name lookups issued via systemd-resolved's D-Bus and Varlink
1461 interfaces (and thus also via glibc NSS if nss-resolve is used) will
1462 now honour a trailing dot in the hostname: if specified the search
1463 path logic is turned off. Thus "resolvectl query foo." is now
1464 equivalent to "resolvectl query --search=off foo.".
1466 * systemd-resolved gained a new D-Bus property "ResolvConfMode" that
1467 exposes how /etc/resolv.conf is currently managed: by resolved (and
1468 in which mode if so) or another subsystem. "resolvctl" will display
1469 this property in its status output.
1471 * The resolv.conf snippets systemd-resolved provides will now set "."
1472 as the search domain if no other search domain is known. This turns
1473 off the derivation of an implicit search domain by nss-dns for the
1474 hostname, when the hostname is set to an FQDN. This change is done to
1475 make nss-dns using resolv.conf provided by systemd-resolved behave
1476 more similarly to nss-resolve.
1478 * systemd-tmpfiles' file "aging" logic (i.e. the automatic clean-up of
1479 /tmp/ and /var/tmp/ based on file timestamps) now looks at the
1480 "birth" time (btime) of a file in addition to the atime, mtime, and
1483 * systemd-analyze gained a new verb "capability" that lists all known
1484 capabilities by the systemd build and by the kernel.
1486 * If a file /usr/lib/clock-epoch exists, PID 1 will read its mtime and
1487 advance the system clock to it at boot if it is noticed to be before
1488 that time. Previously, PID 1 would only advance the time to an epoch
1489 time that is set during build-time. With this new file OS builders
1490 can change this epoch timestamp on individual OS images without
1491 having to rebuild systemd.
1493 * systemd-logind will now listen to the KEY_RESTART key from the Linux
1494 input layer and reboot the system if it is pressed, similarly to how
1495 it already handles KEY_POWER, KEY_SUSPEND or KEY_SLEEP. KEY_RESTART
1496 was originally defined in the Multimedia context (to restart playback
1497 of a song or film), but is now primarily used in various embedded
1498 devices for "Reboot" buttons. Accordingly, systemd-logind will now
1499 honour it as such. This may configured in more detail via the new
1500 HandleRebootKey= and RebootKeyIgnoreInhibited=.
1502 * systemd-nspawn/systemd-machined will now reconstruct hardlinks when
1503 copying OS trees, for example in "systemd-nspawn --ephemeral",
1504 "systemd-nspawn --template=", "machinectl clone" and similar. This is
1505 useful when operating with OSTree images, which use hardlinks heavily
1506 throughout, and where such copies previously resulting in "exploding"
1509 * systemd-nspawn's --console= setting gained support for a new
1510 "autopipe" value, which is identical to "interactive" when invoked on
1511 a TTY, and "pipe" otherwise.
1513 * systemd-networkd's .network files gained support for explicitly
1514 configuring the multicast membership entries of bridge devices in the
1515 [BridgeMDB] section. It also gained support for the PIE queuing
1516 discipline in the [FlowQueuePIE] sections.
1518 * systemd-networkd's .netdev files may now be used to create "BareUDP"
1519 tunnels, configured in the new [BareUDP] setting.
1521 * systemd-networkd's Gateway= setting in .network files now accepts the
1522 special values "_dhcp4" and "_ipv6ra" to configure additional,
1523 locally defined, explicit routes to the gateway acquired via DHCP or
1524 IPv6 Router Advertisements. The old setting "_dhcp" is deprecated,
1525 but still accepted for backwards compatibility.
1527 * systemd-networkd's [IPv6PrefixDelegation] section and
1528 IPv6PrefixDelegation= options have been renamed as [IPv6SendRA] and
1529 IPv6SendRA= (the old names are still accepted for backwards
1532 * systemd-networkd's .network files gained the DHCPv6PrefixDelegation=
1533 boolean setting in [Network] section. If enabled, the delegated prefix
1534 gained by another link will be configured, and an address within the
1535 prefix will be assigned.
1537 * systemd-networkd's .network files gained the Announce= boolean setting
1538 in [DHCPv6PrefixDelegation] section. When enabled, the delegated
1539 prefix will be announced through IPv6 router advertisement (IPv6 RA).
1540 The setting is enabled by default.
1542 * VXLAN tunnels may now be marked as independent of any underlying
1543 network interface via the new Independent= boolean setting.
1545 * systemctl gained support for two new verbs: "service-log-level" and
1546 "service-log-target" may be used on services that implement the
1547 generic org.freedesktop.LogControl1 D-Bus interface to dynamically
1548 adjust the log level and target. All of systemd's long-running
1549 services support this now, but ideally all system services would
1550 implement this interface to make the system more uniformly
1553 * The SystemCallErrorNumber= unit file setting now accepts the new
1554 "kill" and "log" actions, in addition to arbitrary error number
1555 specifications as before. If "kill" the processes are killed on the
1556 event, if "log" the offending system call is audit logged.
1558 * A new SystemCallLog= unit file setting has been added that accepts a
1559 list of system calls that shall be logged about (audit).
1561 * The OS image dissection logic (as used by RootImage= in unit files or
1562 systemd-nspawn's --image= switch) has gained support for identifying
1563 and mounting explicit /usr/ partitions, which are now defined in the
1564 discoverable partition specification. This should be useful for
1565 environments where the root file system is
1566 generated/formatted/populated dynamically on first boot and combined
1567 with an immutable /usr/ tree that is supplied by the vendor.
1569 * In the final phase of shutdown, within the systemd-shutdown binary
1570 we'll now try to detach MD devices (i.e software RAID) in addition to
1571 loopback block devices and DM devices as before. This is supposed to
1572 be a safety net only, in order to increase robustness if things go
1573 wrong. Storage subsystems are expected to properly detach their
1574 storage volumes during regular shutdown already (or in case of
1575 storage backing the root file system: in the initrd hook we return to
1578 * If the SYSTEMD_LOG_TID environment variable is set all systemd tools
1579 will now log the thread ID in their log output. This is useful when
1580 working with heavily threaded programs.
1582 * If the SYSTEMD_RDRAND environment variable is set to "0", systemd will
1583 not use the RDRAND CPU instruction. This is useful in environments
1584 such as replay debuggers where non-deterministic behaviour is not
1587 * The autopaging logic in systemd's various tools (such as systemctl)
1588 has been updated to turn on "secure" mode in "less"
1589 (i.e. $LESSECURE=1) if execution in a "sudo" environment is
1590 detected. This disables invoking external programs from the pager,
1591 via the pipe logic. This behaviour may be overridden via the new
1592 $SYSTEMD_PAGERSECURE environment variable.
1594 * Units which have resource limits (.service, .mount, .swap, .slice,
1595 .socket, and .slice) gained new configuration settings
1596 ManagedOOMSwap=, ManagedOOMMemoryPressure=, and
1597 ManagedOOMMemoryPressureLimitPercent= that specify resource pressure
1598 limits and optional action taken by systemd-oomd.
1600 * A new service systemd-oomd has been added. It monitors resource
1601 contention for selected parts of the unit hierarchy using the PSI
1602 information reported by the kernel, and kills processes when memory
1603 or swap pressure is above configured limits. This service is only
1604 enabled by default in developer mode (see below) and should be
1605 considered a preview in this release. Behaviour details and option
1606 names are subject to change without the usual backwards-compatibility
1609 * A new helper oomctl has been added to introspect systemd-oomd state.
1610 It is only enabled by default in developer mode and should be
1611 considered a preview without the usual backwards-compatibility
1614 * New meson option -Dcompat-mutable-uid-boundaries= has been added. If
1615 enabled, systemd reads the system UID boundaries from /etc/login.defs
1616 at runtime, instead of using the built-in values selected during
1617 build. This is an option to improve compatibility for upgrades from
1618 old systems. It's strongly recommended not to make use of this
1619 functionality on new systems (or even enable it during build), as it
1620 makes something runtime-configurable that is mostly an implementation
1621 detail of the OS, and permits avoidable differences in deployments
1622 that create all kinds of problems in the long run.
1624 * New meson option '-Dmode=developer|release' has been added. When
1625 'developer', additional checks and features are enabled that are
1626 relevant during upstream development, e.g. verification that
1627 semi-automatically-generated documentation has been properly updated
1628 following API changes. Those checks are considered hints for
1629 developers and are not actionable in downstream builds. In addition,
1630 extra features that are not ready for general consumption may be
1631 enabled in developer mode. It is thus recommended to set
1632 '-Dmode=release' in end-user and distro builds.
1634 * systemd-cryptsetup gained support for processing detached LUKS
1635 headers specified on the kernel command line via the header=
1636 parameter of the luks.options= kernel command line option. The same
1637 device/path syntax as for key files is supported for header files
1640 * The "net_id" built-in of udev has been updated to ignore ACPI _SUN
1641 slot index data for devices that are connected through a PCI bridge
1642 where the _SUN index is associated with the bridge instead of the
1643 network device itself. Previously this would create ambiguous device
1644 naming if multiple network interfaces were connected to the same PCI
1645 bridge. Since this is a naming scheme incompatibility on systems that
1646 possess hardware like this it has been introduced as new naming
1647 scheme "v247". The previous scheme can be selected via the
1648 "net.naming-scheme=v245" kernel command line parameter.
1650 * ConditionFirstBoot= semantics have been modified to be safe towards
1651 abnormal system power-off during first boot. Specifically, the
1652 "systemd-machine-id-commit.service" service now acts as boot
1653 milestone indicating when the first boot process is sufficiently
1654 complete in order to not consider the next following boot also a
1655 first boot. If the system is reset before this unit is reached the
1656 first time, the next boot will still be considered a first boot; once
1657 it has been reached, no further boots will be considered a first
1658 boot. The "first-boot-complete.target" unit now acts as official hook
1659 point to order against this. If a service shall be run on every boot
1660 until the first boot fully succeeds it may thus be ordered before
1661 this target unit (and pull it in) and carry ConditionFirstBoot=
1664 * bootctl's set-default and set-oneshot commands now accept the three
1665 special strings "@default", "@oneshot", "@current" in place of a boot
1666 entry id. These strings are resolved to the current default and
1667 oneshot boot loader entry, as well as the currently booted one. Thus
1668 a command "bootctl set-default @current" may be used to make the
1669 currently boot menu item the new default for all subsequent boots.
1671 * "systemctl edit" has been updated to show the original effective unit
1672 contents in commented form in the text editor.
1674 * Units in user mode are now segregated into three new slices:
1675 session.slice (units that form the core of graphical session),
1676 app.slice ("normal" user applications), and background.slice
1677 (low-priority tasks). Unless otherwise configured, user units are
1678 placed in app.slice. The plan is to add resource limits and
1679 protections for the different slices in the future.
1681 * New GPT partition types for RISCV32/64 for the root and /usr
1682 partitions, and their associated Verity partitions have been defined,
1683 and are now understood by systemd-gpt-auto-generator, and the OS
1684 image dissection logic.
1686 Contributions from: Adolfo Jayme Barrientos, afg, Alec Moskvin, Alyssa
1687 Ross, Amitanand Chikorde, Andrew Hangsleben, Anita Zhang, Ansgar
1688 Burchardt, Arian van Putten, Aurelien Jarno, Axel Rasmussen, bauen1,
1689 Beniamino Galvani, Benjamin Berg, Bjørn Mork, brainrom, Chandradeep
1690 Dey, Charles Lee, Chris Down, Christian Göttsche, Christof Efkemann,
1691 Christoph Ruegge, Clemens Gruber, Daan De Meyer, Daniele Medri, Daniel
1692 Mack, Daniel Rusek, Dan Streetman, David Tardon, Dimitri John Ledkov,
1693 Dmitry Borodaenko, Elias Probst, Elisei Roca, ErrantSpore, Etienne
1694 Doms, Fabrice Fontaine, fangxiuning, Felix Riemann, Florian Klink,
1695 Franck Bui, Frantisek Sumsal, fwSmit, George Rawlinson, germanztz,
1696 Gibeom Gwon, Glen Whitney, Gogo Gogsi, Göran Uddeborg, Grant Mathews,
1697 Hans de Goede, Hans Ulrich Niedermann, Haochen Tong, Harald Seiler,
1698 huangyong, Hubert Kario, igo95862, Ikey Doherty, Insun Pyo, Jan Chren,
1699 Jan Schlüter, Jérémy Nouhaud, Jian-Hong Pan, Joerg Behrmann, Jonathan
1700 Lebon, Jörg Thalheim, Josh Brobst, Juergen Hoetzel, Julien Humbert,
1701 Kai-Chuan Hsieh, Kairui Song, Kamil Dudka, Kir Kolyshkin, Kristijan
1702 Gjoshev, Kyle Huey, Kyle Russell, Lee Whalen, Lennart Poettering,
1703 lichangze, Luca Boccassi, Lucas Werkmeister, Luca Weiss, Marc
1704 Kleine-Budde, Marco Wang, Martin Wilck, Marti Raudsepp, masmullin2000,
1705 Máté Pozsgay, Matt Fenwick, Michael Biebl, Michael Scherer, Michal
1706 Koutný, Michal Sekletár, Michal Suchanek, Mikael Szreder, Milo
1707 Casagrande, mirabilos, Mitsuha_QuQ, mog422, Muhammet Kara, Nazar
1708 Vinnichuk, Nicholas Narsing, Nicolas Fella, Njibhu, nl6720, Oğuz Ersen,
1709 Olivier Le Moal, Ondrej Kozina, onlybugreports, Pass Automated Testing
1710 Suite, Pat Coulthard, Pavel Sapezhko, Pedro Ruiz, perry_yuan, Peter
1711 Hutterer, Phaedrus Leeds, PhoenixDiscord, Piotr Drąg, Plan C,
1712 Purushottam choudhary, Rasmus Villemoes, Renaud Métrich, Robert Marko,
1713 Roman Beranek, Ronan Pigott, Roy Chen (陳彥廷), RussianNeuroMancer,
1714 Samanta Navarro, Samuel BF, scootergrisen, Sorin Ionescu, Steve Dodd,
1715 Susant Sahani, Timo Rothenpieler, Tobias Hunger, Tobias Kaufmann, Topi
1716 Miettinen, vanou, Vito Caputo, Weblate, Wen Yang, Whired Planck,
1717 williamvds, Yu, Li-Yu, Yuri Chornoivan, Yu Watanabe, Zbigniew
1718 Jędrzejewski-Szmek, Zmicer Turok, Дамјан Георгиевски
1720 – Warsaw, 2020-11-26
1724 * The service manager gained basic support for cgroup v2 freezer. Units
1725 can now be suspended or resumed either using new systemctl verbs,
1726 freeze and thaw respectively, or via D-Bus.
1728 * PID 1 may now automatically load pre-compiled AppArmor policies from
1729 /etc/apparmor/earlypolicy during early boot.
1731 * The CPUAffinity= setting in service unit files now supports a new
1732 special value "numa" that causes the CPU affinity masked to be set
1733 based on the NUMA mask.
1735 * systemd will now log about all left-over processes remaining in a
1736 unit when the unit is stopped. It will now warn about services using
1737 KillMode=none, as this is generally an unsafe thing to make use of.
1739 * Two new unit file settings
1740 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
1741 added. They may be used to check whether a specific file system path
1742 resides on a block device that is encrypted on the block level
1743 (i.e. using dm-crypt/LUKS).
1745 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
1746 has been added that may be used for simple environment checks. This
1747 is particularly useful when passing in environment variables from a
1748 container manager (or from PAM in case of the systemd --user
1751 * .service unit files now accept a new setting CoredumpFilter= which
1752 allows configuration of the memory sections coredumps of the
1753 service's processes shall include.
1755 * .mount units gained a new ReadWriteOnly= boolean option. If set
1756 it will not be attempted to mount a file system read-only if mounting
1757 in read-write mode doesn't succeed. An option x-systemd.rw-only is
1758 available in /etc/fstab to control the same.
1760 * .socket units gained a new boolean setting PassPacketInfo=. If
1761 enabled, the kernel will attach additional per-packet metadata to all
1762 packets read from the socket, as an ancillary message. This controls
1763 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
1764 depending on socket type.
1766 * .service units gained a new setting RootHash= which may be used to
1767 specify the root hash for verity enabled disk images which are
1768 specified in RootImage=. RootVerity= may be used to specify a path to
1769 the Verity data matching a RootImage= file system. (The latter is
1770 only useful for images that do not contain the Verity data embedded
1771 into the same image that carries a GPT partition table following the
1772 Discoverable Partition Specification). Similarly, systemd-nspawn
1773 gained a new switch --verity-data= that takes a path to a file with
1774 the verity data of the disk image supplied in --image=, if the image
1775 doesn't contain the verity data itself.
1777 * .service units gained a new setting RootHashSignature= which takes
1778 either a base64 encoded PKCS#7 signature of the root hash specified
1779 with RootHash=, or a path to a file to read the signature from. This
1780 allows validation of the root hash against public keys available in
1781 the kernel keyring, and is only supported on recent kernels
1782 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
1783 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
1784 this mechanism has also been added to systemd-veritysetup.
1786 * .service unit files gained two new options
1787 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
1788 tune behaviour if a start or stop timeout is hit, i.e. whether to
1789 terminate the service with SIGTERM, SIGABRT or SIGKILL.
1791 * Most options in systemd that accept hexadecimal values prefixed with
1792 0x in additional to the usual decimal notation now also support octal
1793 notation when the 0o prefix is used and binary notation if the 0b
1796 * Various command line parameters and configuration file settings that
1797 configure key or certificate files now optionally take paths to
1798 AF_UNIX sockets in the file system. If configured that way a stream
1799 connection is made to the socket and the required data read from
1800 it. This is a simple and natural extension to the existing regular
1801 file logic, and permits other software to provide keys or
1802 certificates via simple IPC services, for example when unencrypted
1803 storage on disk is not desired. Specifically, systemd-networkd's
1804 Wireguard and MACSEC key file settings as well as
1805 systemd-journal-gatewayd's and systemd-journal-remote's PEM
1806 key/certificate parameters support this now.
1808 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
1809 configuration files that support specifier expansion learnt six new
1810 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
1811 resolve to the various ID fields from /etc/os-release, %l resolves to
1812 the "short" hostname of the system, i.e. the hostname configured in
1813 the kernel truncated at the first dot.
1815 * Support for the .include syntax in unit files has been removed. The
1816 concept has been obsolete for 6 years and we started warning about
1817 its pending removal 2 years ago (also see NEWS file below). It's
1820 * StandardError= and StandardOutput= in unit files no longer support
1821 the "syslog" and "syslog-console" switches. They were long removed
1822 from the documentation, but will now result in warnings when used,
1823 and be converted to "journal" and "journal+console" automatically.
1825 * If the service setting User= is set to the "nobody" user, a warning
1826 message is now written to the logs (but the value is nonetheless
1827 accepted). Setting User=nobody is unsafe, since the primary purpose
1828 of the "nobody" user is to own all files whose owner cannot be mapped
1829 locally. It's in particular used by the NFS subsystem and in user
1830 namespacing. By running a service under this user's UID it might get
1831 read and even write access to all these otherwise unmappable files,
1832 which is quite likely a major security problem.
1834 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
1835 and others) now have a size and inode limits applied (50% of RAM for
1836 /tmp and /dev/shm, 10% of RAM for other mounts, etc.). Please note
1837 that the implicit kernel default is 50% too, so there is no change
1838 in the size limit for /tmp and /dev/shm.
1840 * nss-mymachines lost support for resolution of users and groups, and
1841 now only does resolution of hostnames. This functionality is now
1842 provided by nss-systemd. Thus, the 'mymachines' entry should be
1843 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
1844 (and 'systemd' added if it is not already there).
1846 * A new kernel command line option systemd.hostname= has been added
1847 that allows controlling the hostname that is initialized early during
1850 * A kernel command line option "udev.blockdev_read_only" has been
1851 added. If specified all hardware block devices that show up are
1852 immediately marked as read-only by udev. This option is useful for
1853 making sure that a specific boot under no circumstances modifies data
1854 on disk. Use "blockdev --setrw" to undo the effect of this, per
1857 * A new boolean kernel command line option systemd.swap= has been
1858 added, which may be used to turn off automatic activation of swap
1859 devices listed in /etc/fstab.
1861 * New kernel command line options systemd.condition-needs-update= and
1862 systemd.condition-first-boot= have been added, which override the
1863 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
1866 * A new kernel command line option systemd.clock-usec= has been added
1867 that allows setting the system clock to the specified time in µs
1868 since Jan 1st, 1970 early during boot. This is in particular useful
1869 in order to make test cases more reliable.
1871 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
1872 systemd-coredump to save core files for suid processes. When saving
1873 the core file, systemd-coredump will use the effective uid and gid of
1874 the process that faulted.
1876 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
1877 now automatically set to "Y" at boot, in order to enable pstore
1878 generation for collection with systemd-pstore.
1880 * We provide a set of udev rules to enable auto-suspend on PCI and USB
1881 devices that were tested to correctly support it. Previously, this
1882 was distributed as a set of udev rules, but has now been replaced by
1883 by a set of hwdb entries (and a much shorter udev rule to take action
1884 if the device modalias matches one of the new hwdb entries).
1886 As before, entries are periodically imported from the database
1887 maintained by the ChromiumOS project. If you have a device that
1888 supports auto-suspend correctly and where it should be enabled by
1889 default, please submit a patch that adds it to the database (see
1890 /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
1892 * systemd-udevd gained the new configuration option timeout_signal= as well
1893 as a corresponding kernel command line option udev.timeout_signal=.
1894 The option can be used to configure the UNIX signal that the main
1895 daemon sends to the worker processes on timeout. Setting the signal
1896 to SIGABRT is useful for debugging.
1898 * .link files managed by systemd-udevd gained options RxFlowControl=,
1899 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
1900 order to configure various flow control parameters. They also gained
1901 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
1902 frame ring buffer sizes.
1904 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
1905 enabled systemd-networkd manages all routes configured by other tools.
1907 * .network files managed by systemd-networkd gained a new section
1908 [SR-IOV], in order to configure SR-IOV capable network devices.
1910 * systemd-networkd's [IPv6Prefix] section in .network files gained a
1911 new boolean setting Assign=. If enabled an address from the prefix is
1912 automatically assigned to the interface.
1914 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
1915 controls delegated prefixes assigned by DHCPv6 client. The section
1916 has three settings: SubnetID=, Assign=, and Token=. The setting
1917 SubnetID= allows explicit configuration of the preferred subnet that
1918 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
1919 Assign= is enabled (which is the default) an address from any acquired
1920 delegated prefix is automatically chosen and assigned to the
1921 interface. The setting Token= specifies an optional address generation
1924 * systemd-networkd's [Network] section gained a new setting
1925 IPv4AcceptLocal=. If enabled the interface accepts packets with local
1928 * systemd-networkd gained support for configuring the HTB queuing
1929 discipline in the [HierarchyTokenBucket] and
1930 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
1931 be configured in the [PFIFO] section, "GRED" in
1932 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
1933 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
1934 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
1935 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
1936 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
1937 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
1939 * systemd-networkd gained support for a new Termination= setting in the
1940 [CAN] section for configuring the termination resistor. It also
1941 gained a new ListenOnly= setting for controlling whether to only
1942 listen on CAN interfaces, without interfering with traffic otherwise
1943 (which is useful for debugging/monitoring CAN network
1944 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
1945 been added to configure various CAN-FD aspects.
1947 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
1948 When enabled, DHCPv6 will be attempted right-away without requiring an
1949 Router Advertisement packet suggesting it first (i.e. without the 'M'
1950 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
1951 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
1952 the RA packets suggest it.
1954 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
1955 which may be used to turn off use of the gateway information provided
1956 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
1957 used to configure how to process leases that lack a lifetime option.
1959 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
1960 setting SendVendorOption= allowing configuration of additional vendor
1961 options to send in the DHCP requests/responses. The [DHCPv6] section
1962 gained a new SendOption= setting for sending arbitrary DHCP
1963 options. RequestOptions= has been added to request arbitrary options
1964 from the server. UserClass= has been added to set the DHCP user class
1967 * systemd-networkd's [DHCPServer] section gained a new set of options
1968 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
1969 information about these three protocols in the DHCP lease. It also
1970 gained support for including "MUD" URLs ("Manufacturer Usage
1971 Description"). Support for "MUD" URLs was also added to the LLDP
1972 stack, configurable in the [LLDP] section in .network files.
1974 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
1975 mode. Also, the sections now support a new setting SourceMACAddress=.
1977 * systemd-networkd's .netdev files now support a new setting
1978 VLANProtocol= in the [Bridge] section that allows configuration of
1979 the VLAN protocol to use.
1981 * systemd-networkd supports a new Group= setting in the [Link] section
1982 of the .network files, to control the link group.
1984 * systemd-networkd's [Network] section gained a new
1985 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
1986 link local address is generated.
1988 * A new default .network file is now shipped that matches TUN/TAP
1989 devices that begin with "vt-" in their name. Such interfaces will
1990 have IP routing onto the host links set up automatically. This is
1991 supposed to be used by VM managers to trivially acquire a network
1992 interface which is fully set up for host communication, simply by
1993 carefully picking an interface name to use.
1995 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
1996 which sets the route priority for routes specified by the DHCP server.
1998 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
1999 which configures the vendor class information sent to DHCP server.
2001 * The BlackList= settings in .network files' [DHCPv4] and
2002 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
2003 are still understood to provide compatibility.
2005 * networkctl gained the new "forcerenew" command for forcing all DHCP
2006 server clients to renew their lease. The interface "status" output
2007 will now show numerous additional fields of information about an
2008 interface. There are new "up" and "down" commands to bring specific
2009 interfaces up or down.
2011 * systemd-resolved's DNS= configuration option now optionally accepts a
2012 port number (after ":") and a host name (after "#"). When the host
2013 name is specified, the DNS-over-TLS certificate is validated to match
2014 the specified hostname. Additionally, in case of IPv6 addresses, an
2015 interface may be specified (after "%").
2017 * systemd-resolved may be configured to forward single-label DNS names.
2018 This is not standard-conformant, but may make sense in setups where
2019 public DNS servers are not used.
2021 * systemd-resolved's DNS-over-TLS support gained SNI validation.
2023 * systemd-nspawn's --resolv-conf= switch gained a number of new
2024 supported values. Specifically, options starting with "replace-" are
2025 like those prefixed "copy-" but replace any existing resolv.conf
2026 file. And options ending in "-uplink" and "-stub" can now be used to
2027 propagate other flavours of resolv.conf into the container (as
2028 defined by systemd-resolved).
2030 * The various programs included in systemd can now optionally output
2031 their log messages on stderr prefixed with a timestamp, controlled by
2032 the $SYSTEMD_LOG_TIME environment variable.
2034 * systemctl gained a new "-P" switch that is a shortcut for "--value
2037 * "systemctl list-units" and "systemctl list-machines" no longer hide
2038 their first output column with --no-legend. To hide the first column,
2041 * "systemctl reboot" takes the option "--reboot-argument=".
2042 The optional positional argument to "systemctl reboot" is now
2043 being deprecated in favor of this option.
2045 * systemd-run gained a new switch --slice-inherit. If specified the
2046 unit it generates is placed in the same slice as the systemd-run
2049 * systemd-journald gained support for zstd compression of large fields
2050 in journal files. The hash tables in journal files have been hardened
2051 against hash collisions. This is an incompatible change and means
2052 that journal files created with new systemd versions are not readable
2053 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
2054 environment variable for systemd-journald.service is set to 0 this
2055 new hardening functionality may be turned off, so that generated
2056 journal files remain compatible with older journalctl
2059 * journalctl will now include a clickable link in the default output for
2060 each log message for which an URL with further documentation is
2061 known. This is only supported on terminal emulators that support
2062 clickable hyperlinks, and is turned off if a pager is used (since
2063 "less" still doesn't support hyperlinks,
2064 unfortunately). Documentation URLs may be included in log messages
2065 either by including a DOCUMENTATION= journal field in it, or by
2066 associating a journal message catalog entry with the log message's
2067 MESSAGE_ID, which then carries a "Documentation:" tag.
2069 * journald.conf gained a new boolean setting Audit= that may be used to
2070 control whether systemd-journald will enable audit during
2073 * when systemd-journald's log stream is broken up into multiple lines
2074 because the PID of the sender changed this is indicated in the
2075 generated log records via the _LINE_BREAK=pid-change field.
2077 * journalctl's "-o cat" output mode will now show one or more journal
2078 fields specified with --output-fields= instead of unconditionally
2079 MESSAGE=. This is useful to retrieve a very specific set of fields
2080 without any decoration.
2082 * The sd-journal.h API gained two new functions:
2083 sd_journal_enumerate_available_unique() and
2084 sd_journal_enumerate_available_data() that operate like their
2085 counterparts that lack the _available_ in the name, but skip items
2086 that cannot be read and processed by the local implementation
2087 (i.e. are compressed in an unsupported format or such),
2089 * coredumpctl gained a new --file= switch, matching the same one in
2090 journalctl: a specific journal file may be specified to read the
2093 * coredumps collected by systemd-coredump may now be compressed using
2096 * systemd-binfmt gained a new switch --unregister for unregistering all
2097 registered entries at once. This is now invoked automatically at
2098 shutdown, so that binary formats registered with the "F" flag will
2099 not block clean file system unmounting.
2101 * systemd-notify's --pid= switch gained new values: "parent", "self",
2102 "auto" for controlling which PID to send to the service manager: the
2103 systemd-notify process' PID, or the one of the process invoking it.
2105 * systemd-logind's Session bus object learnt a new method call
2106 SetType() for temporarily updating the session type of an already
2107 allocated session. This is useful for upgrading tty sessions to
2108 graphical ones once a compositor is invoked.
2110 * systemd-socket-proxy gained a new switch --exit-idle-time= for
2111 configuring an exit-on-idle time.
2113 * systemd-repart's --empty= setting gained a new value "create". If
2114 specified a new empty regular disk image file is created under the
2115 specified name. Its size may be specified with the new --size=
2116 option. The latter is also supported without the "create" mode, in
2117 order to grow existing disk image files to the specified size. These
2118 two new options are useful when creating or manipulating disk images
2119 instead of operating on actual block devices.
2121 * systemd-repart drop-ins now support a new UUID= setting to control
2122 the UUID to assign to a newly created partition.
2124 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
2127 * systemd-repart's Label= setting now support the usual, simple
2128 specifier expansion.
2130 * systemd-homed's LUKS backend gained the ability to discard empty file
2131 system blocks automatically when the user logs out. This is enabled
2132 by default to ensure that home directories take minimal space when
2133 logged out but get full size guarantees when logged in. This may be
2134 controlled with the new --luks-offline-discard= switch to homectl.
2136 * If systemd-homed detects that /home/ is encrypted as a whole it will
2137 now default to the directory or subvolume backends instead of the
2138 LUKS backend, in order to avoid double encryption. The default
2139 storage and file system may now be configured explicitly, too, via
2140 the new /etc/systemd/homed.conf configuration file.
2142 * systemd-homed now supports unlocking home directories with FIDO2
2143 security tokens that support the 'hmac-secret' extension, in addition
2144 to the existing support for PKCS#11 security token unlocking
2145 support. Note that many recent hardware security tokens support both
2146 interfaces. The FIDO2 support is accessible via homectl's
2147 --fido2-device= option.
2149 * homectl's --pkcs11-uri= setting now accepts two special parameters:
2150 if "auto" is specified and only one suitable PKCS#11 security token
2151 is plugged in, its URL is automatically determined and enrolled for
2152 unlocking the home directory. If "list" is specified a brief table of
2153 suitable PKCS#11 security tokens is shown. Similar, the new
2154 --fido2-device= option also supports these two special values, for
2155 automatically selecting and listing suitable FIDO2 devices.
2157 * The /etc/crypttab tmp option now optionally takes an argument
2158 selecting the file system to use. Moreover, the default is now
2159 changed from ext2 to ext4.
2161 * There's a new /etc/crypttab option "keyfile-erase". If specified the
2162 key file listed in the same line is removed after use, regardless if
2163 volume activation was successful or not. This is useful if the key
2164 file is only acquired transiently at runtime and shall be erased
2165 before the system continues to boot.
2167 * There's also a new /etc/crypttab option "try-empty-password". If
2168 specified, before asking the user for a password it is attempted to
2169 unlock the volume with an empty password. This is useful for
2170 installing encrypted images whose password shall be set on first boot
2171 instead of at installation time.
2173 * systemd-cryptsetup will now attempt to load the keys to unlock
2174 volumes with automatically from files in
2175 /etc/cryptsetup-keys.d/<volume>.key and
2176 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
2178 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
2179 /etc/crypttab, during boot.
2181 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
2182 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
2185 * A new generator systemd-xdg-autostart-generator has been added. It
2186 generates systemd unit files from XDG autostart .desktop files, and
2187 may be used to let the systemd user instance manage services that are
2188 started automatically as part of the desktop session.
2190 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
2191 to query and change the firmware's 'reboot into firmware' setup flag.
2193 * systemd-firstboot gained a new switch --kernel-command-line= that may
2194 be used to initialize the /etc/kernel/cmdline file of the image. It
2195 also gained a new switch --root-password-hashed= which is like
2196 --root-password= but accepts a pre-hashed UNIX password as
2197 argument. The new option --delete-root-password may be used to unset
2198 any password for the root user (dangerous!). The --root-shell= switch
2199 may be used to control the shell to use for the root account. A new
2200 --force option may be used to override any already set settings with
2201 the parameters specified on the command line (by default, the tool
2202 will not override what has already been set before, i.e. is purely
2205 * systemd-firstboot gained support for a new --image= switch, which is
2206 similar to --root= but accepts the path to a disk image file, on
2207 which it then operates.
2209 * A new sd-path.h API has been added to libsystemd. It provides a
2210 simple API for retrieving various search paths and primary
2211 directories for various resources.
2213 * A new call sd_notify_barrier() has been added to the sd-daemon.h
2214 API. The call will block until all previously sent sd_notify()
2215 messages have been processed by the service manager. This is useful
2216 to remove races caused by a process already having disappeared at the
2217 time a notification message is processed by the service manager,
2218 making correct attribution impossible. The systemd-notify tool will
2219 now make use of this call implicitly, but this can be turned off again
2220 via the new --no-block switch.
2222 * When sending a file descriptor (fd) to the service manager to keep
2223 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
2224 may be specified. If passed the service manager will refrain from
2225 poll()ing on the file descriptor. Traditionally (and when the
2226 parameter is not specified), the service manager will poll it for
2227 POLLHUP or POLLERR events, and immediately close the fds in that
2230 * The service manager (PID1) gained a new D-Bus method call
2231 SetShowStatus() which may be used to control whether it shall show
2232 boot-time status output on the console. This method has a similar
2233 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
2235 * The sd-bus API gained a number of convenience functions that take
2236 va_list arguments rather than "...". For example, there's now
2237 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
2238 it easier to build wrappers that accept variadic arguments and want
2239 to pass a ready va_list structure to sd-bus.
2241 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
2242 flag which alters how the userdata pointer to pass to the callbacks
2243 is determined. When the flag is set, the offset field is converted
2244 as-is into a pointer, without adding it to the object pointer the
2245 vtable is associated with.
2247 * sd-bus now exposes four new functions:
2248 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
2249 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
2250 validate strings to check if they qualify as various D-Bus concepts.
2252 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
2253 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
2254 that simplify adding argument names to D-Bus methods and signals.
2256 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
2258 * Various D-Bus APIs of systemd daemons now have man pages that
2259 document the methods, signals and properties.
2261 * The expectations on user/group name syntax are now documented in
2262 detail; documentation on how classic home directories may be
2263 converted into home directories managed by homed has been added;
2264 documentation regarding integration of homed/userdb functionality in
2265 desktops has been added:
2267 https://systemd.io/USER_NAMES
2268 https://systemd.io/CONVERTING_TO_HOMED
2269 https://systemd.io/USERDB_AND_DESKTOPS
2271 * Documentation for the on-disk Journal file format has been updated
2272 and has now moved to:
2274 https://systemd.io/JOURNAL_FILE_FORMAT
2276 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
2277 has been extended by a set of environment variables that expose
2278 select fields from the host's os-release file to the container
2279 payload. Similarly, host's os-release files can be mounted into the
2280 container underneath /run/host. Together, those mechanisms provide a
2281 standardized way to expose information about the host to the
2282 container payload. Both interfaces are implemented in systemd-nspawn.
2284 * All D-Bus services shipped in systemd now implement the generic
2285 LogControl1 D-Bus API which allows clients to change log level +
2286 target of the service during runtime.
2288 * Only relevant for developers: the mkosi.default symlink has been
2289 dropped from version control. Please create a symlink to one of the
2290 distribution-specific defaults in .mkosi/ based on your preference.
2292 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
2293 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
2294 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
2295 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
2296 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
2297 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
2298 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
2299 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
2300 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
2301 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
2302 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
2303 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
2304 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
2305 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
2306 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
2307 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
2308 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
2309 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
2310 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
2311 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
2312 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
2313 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
2314 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
2315 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
2316 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
2317 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
2318 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
2319 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
2320 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
2321 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
2322 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
2323 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
2324 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
2325 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
2326 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
2327 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
2328 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
2329 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
2331 – Warsaw, 2020-07-30
2335 * A new tool "systemd-repart" has been added, that operates as an
2336 idempotent declarative repartitioner for GPT partition tables.
2337 Specifically, a set of partitions that must or may exist can be
2338 configured via drop-in files, and during every boot the partition
2339 table on disk is compared with these files, creating missing
2340 partitions or growing existing ones based on configurable relative
2341 and absolute size constraints. The tool is strictly incremental,
2342 i.e. does not delete, shrink or move partitions, but only adds and
2343 grows them. The primary use-case is OS images that ship in minimized
2344 form, that on first boot are grown to the size of the underlying
2345 block device or augmented with additional partitions. For example,
2346 the root partition could be extended to cover the whole disk, or a
2347 swap or /home partitions could be added on first boot. It can also be
2348 used for systems that use an A/B update scheme but ship images with
2349 just the A partition, with B added on first boot. The tool is
2350 primarily intended to be run in the initrd, shortly before
2351 transitioning into the host OS, but can also be run after the
2352 transition took place. It automatically discovers the disk backing
2353 the root file system, and should hence not require any additional
2354 configuration besides the partition definition drop-ins. If no
2355 configuration drop-ins are present, no action is taken.
2357 * A new component "userdb" has been added, along with a small daemon
2358 "systemd-userdbd.service" and a client tool "userdbctl". The framework
2359 allows defining rich user and group records in a JSON format,
2360 extending on the classic "struct passwd" and "struct group"
2361 structures. Various components in systemd have been updated to
2362 process records in this format, including systemd-logind and
2363 pam-systemd. The user records are intended to be extensible, and
2364 allow setting various resource management, security and runtime
2365 parameters that shall be applied to processes and sessions of the
2366 user as they log in. This facility is intended to allow associating
2367 such metadata directly with user/group records so that they can be
2368 produced, extended and consumed in unified form. We hope that
2369 eventually frameworks such as sssd will generate records this way, so
2370 that for the first time resource management and various other
2371 per-user settings can be configured in LDAP directories and then
2372 provided to systemd (specifically to systemd-logind and pam-system)
2373 to apply on login. For further details see:
2375 https://systemd.io/USER_RECORD
2376 https://systemd.io/GROUP_RECORD
2377 https://systemd.io/USER_GROUP_API
2379 * A small new service systemd-homed.service has been added, that may be
2380 used to securely manage home directories with built-in encryption.
2381 The complete user record data is unified with the home directory,
2382 thus making home directories naturally migratable. Its primary
2383 back-end is based on LUKS volumes, but fscrypt, plain directories,
2384 and other storage schemes are also supported. This solves a couple of
2385 problems we saw with traditional ways to manage home directories, in
2386 particular when it comes to encryption. For further discussion of
2387 this, see the video of Lennart's talk at AllSystemsGo! 2019:
2389 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
2391 For further details about the format and expectations on home
2392 directories this new daemon makes, see:
2394 https://systemd.io/HOME_DIRECTORY
2396 * systemd-journald is now multi-instantiable. In addition to the main
2397 instance systemd-journald.service there's now a template unit
2398 systemd-journald@.service, with each instance defining a new named
2399 log 'namespace' (whose name is specified via the instance part of the
2400 unit name). A new unit file setting LogNamespace= has been added,
2401 taking such a namespace name, that assigns services to the specified
2402 log namespaces. As each log namespace is serviced by its own
2403 independent journal daemon, this functionality may be used to improve
2404 performance and increase isolation of applications, at the price of
2405 losing global message ordering. Each instance of journald has a
2406 separate set of configuration files, with possibly different disk
2407 usage limitations and other settings.
2409 journalctl now takes a new option --namespace= to show logs from a
2410 specific log namespace. The sd-journal.h API gained
2411 sd_journal_open_namespace() for opening the log stream of a specific
2412 log namespace. systemd-journald also gained the ability to exit on
2413 idle, which is useful in the context of log namespaces, as this means
2414 log daemons for log namespaces can be activated automatically on
2415 demand and will stop automatically when no longer used, minimizing
2418 * When systemd-tmpfiles copies a file tree using the 'C' line type it
2419 will now label every copied file according to the SELinux database.
2421 * When systemd/PID 1 detects it is used in the initrd it will now boot
2422 into initrd.target rather than default.target by default. This should
2423 make it simpler to build initrds with systemd as for many cases the
2424 only difference between a host OS image and an initrd image now is
2425 the presence of the /etc/initrd-release file.
2427 * A new kernel command line option systemd.cpu_affinity= is now
2428 understood. It's equivalent to the CPUAffinity= option in
2429 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
2430 itself and the default for all other processes.
2432 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
2433 equivalent), the SELinux database is now reloaded, ensuring that
2434 sockets and other file system objects are generated taking the new
2435 database into account.
2437 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
2438 "quiet" has been changed to imply that instead of
2439 "systemd.show-status=auto". In this mode, only messages about errors
2440 and significant delays in boot are shown on the console.
2442 * The sd-event.h API gained native support for the new Linux "pidfd"
2443 concept. This permits watching processes using file descriptors
2444 instead of PID numbers, which fixes a number of races and makes
2445 process supervision more robust and efficient. All of systemd's
2446 components will now use pidfds if the kernel supports it for process
2447 watching, with the exception of PID 1 itself, unfortunately. We hope
2448 to move PID 1 to exclusively using pidfds too eventually, but this
2449 requires some more kernel work first. (Background: PID 1 watches
2450 processes using waitid() with the P_ALL flag, and that does not play
2451 together nicely with pidfds yet.)
2453 * Closely related to this, the sd-event.h API gained two new calls
2454 sd_event_source_send_child_signal() (for sending a signal to a
2455 watched process) and sd_event_source_get_child_process_own() (for
2456 marking a process so that it is killed automatically whenever the
2457 event source watching it is freed).
2459 * systemd-networkd gained support for configuring Token Bucket Filter
2460 (TBF) parameters in its qdisc configuration support. Similarly,
2461 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
2462 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
2464 * systemd-networkd gained support for Intermediate Functional Block
2465 (IFB) network devices.
2467 * systemd-networkd gained support for configuring multi-path IP routes,
2468 using the new MultiPathRoute= setting in the [Route] section.
2470 * systemd-networkd's DHCPv4 client has been updated to support a new
2471 SendDecline= option. If enabled, duplicate address detection is done
2472 after a DHCP offer is received from the server. If a conflict is
2473 detected, the address is declined. The DHCPv4 client also gained
2474 support for a new RouteMTUBytes= setting that allows to configure the
2475 MTU size to be used for routes generated from DHCPv4 leases.
2477 * The PrefixRoute= setting in systemd-networkd's [Address] section of
2478 .network files has been deprecated, and replaced by AddPrefixRoute=,
2479 with its sense inverted.
2481 * The Gateway= setting of [Route] sections of .network files gained
2482 support for a special new value "_dhcp". If set, the configured
2483 static route uses the gateway host configured via DHCP.
2485 * New User= and SuppressPrefixLength= settings have been implemented
2486 for the [RoutingPolicyRule] section of .network files to configure
2487 source routing based on UID ranges and prefix length, respectively.
2489 * The Type= match property of .link files has been generalized to
2490 always match the device type shown by 'networkctl status', even for
2491 devices where udev does not set DEVTYPE=. This allows e.g. Type=ether
2494 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
2495 D-Bus message object as "sensitive". Those objects are erased from
2496 memory when they are freed. This concept is intended to be used for
2497 messages that contain security sensitive data. A new flag
2498 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
2499 in sd-bus vtables, causing any incoming and outgoing messages of
2500 those methods to be implicitly marked as "sensitive".
2502 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
2503 contents of a message (or parts thereof) to standard output for
2506 * systemd-sysusers gained support for creating users with the primary
2507 group named differently than the user.
2509 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
2510 gained support for growing XFS partitions. Previously it supported
2511 only ext4 and btrfs partitions.
2513 * The support for /etc/crypttab gained a new x-initrd.attach option. If
2514 set, the specified encrypted volume is unlocked already in the
2515 initrd. This concept corresponds to the x-initrd.mount option in
2518 * systemd-cryptsetup gained native support for unlocking encrypted
2519 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
2520 encryption of volumes to YubiKeys. This is exposed in the new
2521 pkcs11-uri= option in /etc/crypttab.
2523 * The /etc/fstab support in systemd now supports two new mount options
2524 x-systemd.{required,wanted}-by=, for explicitly configuring the units
2525 that the specified mount shall be pulled in by, in place of
2526 the usual local-fs.target/remote-fs.target.
2528 * The https://systemd.io/ web site has been relaunched, directly
2529 populated with most of the documentation included in the systemd
2530 repository. systemd also acquired a new logo, thanks to Tobias
2533 * systemd-udevd gained support for managing "alternative" network
2534 interface names, as supported by new Linux kernels. For the first
2535 time this permits assigning multiple (and longer!) names to a network
2536 interface. systemd-udevd will now by default assign the names
2537 generated via all supported naming schemes to each interface. This
2538 may be further tweaked with .link files and the AlternativeName= and
2539 AlternativeNamesPolicy= settings. Other components of systemd have
2540 been updated to support the new alternative names wherever
2541 appropriate. For example, systemd-nspawn will now generate
2542 alternative interface names for the host-facing side of container
2543 veth links based on the full container name without truncation.
2545 * systemd-nspawn interface naming logic has been updated in another way
2546 too: if the main interface name (i.e. as opposed to new-style
2547 "alternative" names) based on the container name is truncated, a
2548 simple hashing scheme is used to give different interface names to
2549 multiple containers whose names all begin with the same prefix. Since
2550 this changes the primary interface names pointing to containers if
2551 truncation happens, the old scheme may still be requested by
2552 selecting an older naming scheme, via the net.naming-scheme= kernel
2553 command line option.
2555 * PrivateUsers= in service files now works in services run by the
2556 systemd --user per-user instance of the service manager.
2558 * A new per-service sandboxing option ProtectClock= has been added that
2559 locks down write access to the system clock. It takes away device
2560 node access to /dev/rtc as well as the system calls that set the
2561 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
2562 Note that this option does not affect access to auxiliary services
2563 that allow changing the clock, for example access to
2566 * The systemd-id128 tool gained a new "show" verb for listing or
2567 resolving a number of well-known UUIDs/128bit IDs, currently mostly
2568 GPT partition table types.
2570 * The Discoverable Partitions Specification has been updated to support
2571 /var and /var/tmp partition discovery. Support for this has been
2572 added to systemd-gpt-auto-generator. For details see:
2574 https://systemd.io/DISCOVERABLE_PARTITIONS
2576 * "systemctl list-unit-files" has been updated to show a new column
2577 with the suggested enablement state based on the vendor preset files
2578 for the respective units.
2580 * "systemctl" gained a new option "--with-dependencies". If specified
2581 commands such as "systemctl status" or "systemctl cat" will now show
2582 all specified units along with all units they depend on.
2584 * networkctl gained support for showing per-interface logs in its
2587 * systemd-networkd-wait-online gained support for specifying the maximum
2588 operational state to wait for, and to wait for interfaces to
2591 * The [Match] section of .link and .network files now supports a new
2592 option PermanentMACAddress= which may be used to check against the
2593 permanent MAC address of a network device even if a randomized MAC
2596 * The [TrafficControlQueueingDiscipline] section in .network files has
2597 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
2598 dropped from the individual setting names.
2600 * Any .link and .network files that have an empty [Match] section (this
2601 also includes empty and commented-out files) will now be
2602 rejected. systemd-udev and systemd-networkd started warning about
2603 such files in version 243.
2605 * systemd-logind will now validate access to the operation of changing
2606 the virtual terminal via a polkit action. By default, only users
2607 with at least one session on a local VT are granted permission.
2609 * When systemd sets up PAM sessions that invoked service processes
2610 shall run in, the pam_setcred() API is now invoked, thus permitting
2611 PAM modules to set additional credentials for the processes.
2613 * portablectl attach/detach verbs now accept --now and --enable options
2614 to combine attachment with enablement and invocation, or detachment
2615 with stopping and disablement.
2617 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
2618 fixed, which in turn exposed bugs in unit configuration of services
2619 which have Type=oneshot and should only run once, but do not have
2620 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
2621 service may be started again after exiting successfully, for example
2622 as a dependency in another transaction. Affected services included
2623 some internal systemd services (most notably
2624 systemd-vconsole-setup.service, which was updated to have
2625 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
2626 plymouth has been suitably updated or patched before upgrading to
2627 this systemd release. See
2628 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
2629 additional discussion.
2631 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
2632 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
2633 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
2634 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
2635 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
2636 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
2637 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
2638 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
2639 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
2640 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
2641 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
2642 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
2643 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
2644 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
2645 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
2646 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
2647 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
2648 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
2649 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
2650 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
2651 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
2652 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
2653 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
2654 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
2657 – Warsaw, 2020-03-06
2661 * Support for the cpuset cgroups v2 controller has been added.
2662 Processes may be restricted to specific CPUs using the new
2663 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
2664 AllowedMemoryNodes= setting.
2666 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
2667 now be configured using a new RestartKillSignal= setting. This
2668 allows units which signals to request termination to implement
2669 different behaviour when stopping in preparation for a restart.
2671 * "systemctl clean" may now be used also for socket, mount, and swap
2674 * systemd will also read configuration options from the EFI variable
2675 SystemdOptions. This may be used to configure systemd behaviour when
2676 modifying the kernel command line is inconvenient, but configuration
2677 on disk is read too late, for example for the options related to
2678 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
2679 set the EFI variable.
2681 * systemd will now disable printk ratelimits in early boot. This should
2682 allow us to capture more logs from the early boot phase where normal
2683 storage is not available and the kernel ring buffer is used for
2684 logging. Configuration on the kernel command line has higher priority
2685 and overrides the systemd setting.
2687 systemd programs which log to /dev/kmsg directly use internal
2688 ratelimits to prevent runaway logging. (Normally this is only used
2689 during early boot, so in practice this change has very little
2692 * Unit files now support top level dropin directories of the form
2693 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
2694 that affects all corresponding unit files.
2696 * systemctl gained support for 'stop --job-mode=triggering' which will
2697 stop the specified unit and any units which could trigger it.
2699 * Unit status display now includes units triggering and triggered by
2700 the unit being shown.
2702 * The RuntimeMaxSec= setting is now supported by scopes, not just
2703 .service units. This is particularly useful for PAM sessions which
2704 create a scope unit for the user login. systemd.runtime_max_sec=
2705 setting may used with the pam_systemd module to limit the duration
2706 of the PAM session, for example for time-limited logins.
2708 * A new @pkey system call group is now defined to make it easier to
2709 allow-list memory protection syscalls for containers and services
2710 which need to use them.
2712 * systemd-udevd: removed the 30s timeout for killing stale workers on
2713 exit. systemd-udevd now waits for workers to finish. The hard-coded
2714 exit timeout of 30s was too short for some large installations, where
2715 driver initialization could be prematurely interrupted during initrd
2716 processing if the root file system had been mounted and init was
2717 preparing to switch root. If udevd is run without systemd and workers
2718 are hanging while udevd receives an exit signal, udevd will now exit
2719 when udev.event_timeout is reached for the last hanging worker. With
2720 systemd, the exit timeout can additionally be configured using
2721 TimeoutStopSec= in systemd-udevd.service.
2723 * udev now provides a program (fido_id) that identifies FIDO CTAP1
2724 ("U2F")/CTAP2 security tokens based on the usage declared in their
2725 report and descriptor and outputs suitable environment variables.
2726 This replaces the externally maintained allow lists of all known
2727 security tokens that were used previously.
2729 * Automatically generated autosuspend udev rules for allow-listed
2730 devices have been imported from the Chromium OS project. This should
2731 improve power saving with many more devices.
2733 * udev gained a new "CONST{key}=value" setting that allows matching
2734 against system-wide constants without forking a helper binary.
2735 Currently "arch" and "virt" keys are supported.
2737 * udev now opens CDROMs in non-exclusive mode when querying their
2738 capabilities. This should fix issues where other programs trying to
2739 use the CDROM cannot gain access to it, but carries a risk of
2740 interfering with programs writing to the disk, if they did not open
2741 the device in exclusive mode as they should.
2743 * systemd-networkd does not create a default route for IPv4 link local
2744 addressing anymore. The creation of the route was unexpected and was
2745 breaking routing in various cases, but people who rely on it being
2746 created implicitly will need to adjust. Such a route may be requested
2747 with DefaultRouteOnDevice=yes.
2749 Similarly, systemd-networkd will not assign a link-local IPv6 address
2750 when IPv6 link-local routing is not enabled.
2752 * Receive and transmit buffers may now be configured on links with
2753 the new RxBufferSize= and TxBufferSize= settings.
2755 * systemd-networkd may now advertise additional IPv6 routes. A new
2756 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
2759 * systemd-networkd may now configure "next hop" routes using the
2760 [NextHop] section and Gateway= and Id= settings.
2762 * systemd-networkd will now retain DHCP config on restarts by default
2763 (but this may be overridden using the KeepConfiguration= setting).
2764 The default for SendRelease= has been changed to true.
2766 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
2767 received from the server.
2769 The client will use the received SIP server list if UseSIP=yes is
2772 The client may be configured to request specific options from the
2773 server using a new RequestOptions= setting.
2775 The client may be configured to send arbitrary options to the server
2776 using a new SendOption= setting.
2778 A new IPServiceType= setting has been added to configure the "IP
2779 service type" value used by the client.
2781 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
2782 request prefix hints in the DHCPv6 solicitation.
2784 * The DHCPv4 server may be configured to send arbitrary options using
2785 a new SendOption= setting.
2787 * The DHCPv4 server may now be configured to emit SIP server list using
2788 the new EmitSIP= and SIP= settings.
2790 * systemd-networkd and networkctl may now renew DHCP leases on demand.
2791 networkctl has a new 'networkctl renew' verb.
2793 * systemd-networkd may now reconfigure links on demand. networkctl
2794 gained two new verbs: "reload" will reload the configuration, and
2795 "reconfigure DEVICE…" will reconfigure one or more devices.
2797 * .network files may now match on SSID and BSSID of a wireless network,
2798 i.e. the access point name and hardware address using the new SSID=
2799 and BSSID= options. networkctl will display the current SSID and
2800 BSSID for wireless links.
2802 .network files may also match on the wireless network type using the
2803 new WLANInterfaceType= option.
2805 * systemd-networkd now includes default configuration that enables
2806 link-local addressing when connected to an ad-hoc wireless network.
2808 * systemd-networkd may configure the Traffic Control queueing
2809 disciplines in the kernel using the new
2810 [TrafficControlQueueingDiscipline] section and Parent=,
2811 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
2812 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
2813 NetworkEmulatorDuplicateRate= settings.
2815 * systemd-tmpfiles gained a new w+ setting to append to files.
2817 * systemd-analyze dump will now report when the memory configuration in
2818 the kernel does not match what systemd has configured (usually,
2819 because some external program has modified the kernel configuration
2822 * systemd-analyze gained a new --base-time= switch instructs the
2823 'calendar' verb to resolve times relative to that timestamp instead
2824 of the present time.
2826 * journalctl --update-catalog now produces deterministic output (making
2827 reproducible image builds easier).
2829 * A new devicetree-overlay setting is now documented in the Boot Loader
2832 * The default value of the WatchdogSec= setting used in systemd
2833 services (the ones bundled with the project itself) may be set at
2834 configuration time using the -Dservice-watchdog= setting. If set to
2835 empty, the watchdogs will be disabled.
2837 * systemd-resolved validates IP addresses in certificates now when GnuTLS
2840 * libcryptsetup >= 2.0.1 is now required.
2842 * A configuration option -Duser-path= may be used to override the $PATH
2843 used by the user service manager. The default is again to use the same
2844 path as the system manager.
2846 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
2847 outputting the 128bit IDs in UUID format (i.e. in the "canonical
2850 * Service units gained a new sandboxing option ProtectKernelLogs= which
2851 makes sure the program cannot get direct access to the kernel log
2852 buffer anymore, i.e. the syslog() system call (not to be confused
2853 with the API of the same name in libc, which is not affected), the
2854 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
2855 inaccessible to the service. It's recommended to enable this setting
2856 for all services that should not be able to read from or write to the
2857 kernel log buffer, which are probably almost all.
2859 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
2860 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
2861 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
2862 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
2863 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
2864 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
2865 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
2866 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
2867 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
2868 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
2869 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
2870 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
2871 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
2872 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
2873 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
2874 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
2875 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
2876 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
2877 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
2878 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
2879 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
2880 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
2881 Zach Smith, Zbigniew Jędrzejewski-Szmek
2883 – Warsaw, 2019-11-29
2887 * This release enables unprivileged programs (i.e. requiring neither
2888 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
2889 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
2890 kernel for the whole UNIX group range, i.e. all processes. This
2891 change should be reasonably safe, as the kernel support for it was
2892 specifically implemented to allow safe access to ICMP Echo for
2893 processes lacking any privileges. If this is not desirable, it can be
2894 disabled again by setting the parameter to "1 0".
2896 * Previously, filters defined with SystemCallFilter= would have the
2897 effect that any calling of an offending system call would terminate
2898 the calling thread. This behaviour never made much sense, since
2899 killing individual threads of unsuspecting processes is likely to
2900 create more problems than it solves. With this release the default
2901 action changed from killing the thread to killing the whole
2902 process. For this to work correctly both a kernel version (>= 4.14)
2903 and a libseccomp version (>= 2.4.0) supporting this new seccomp
2904 action is required. If an older kernel or libseccomp is used the old
2905 behaviour continues to be used. This change does not affect any
2906 services that have no system call filters defined, or that use
2907 SystemCallErrorNumber= (and thus see EPERM or another error instead
2908 of being killed when calling an offending system call). Note that
2909 systemd documentation always claimed that the whole process is
2910 killed. With this change behaviour is thus adjusted to match the
2913 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
2914 4194304 by default, i.e. the full 22bit range the kernel allows, up
2915 from the old 16bit range. This should improve security and
2916 robustness, as PID collisions are made less likely (though certainly
2917 still possible). There are rumours this might create compatibility
2918 problems, though at this moment no practical ones are known to
2919 us. Downstream distributions are hence advised to undo this change in
2920 their builds if they are concerned about maximum compatibility, but
2921 for everybody else we recommend leaving the value bumped. Besides
2922 improving security and robustness this should also simplify things as
2923 the maximum number of allowed concurrent tasks was previously bounded
2924 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
2925 only a single knob is left ("kernel.threads-max"). There have been
2926 concerns that usability is affected by this change because larger PID
2927 numbers are harder to type, but we believe the change from 5 digits
2928 to 7 digits doesn't hamper usability.
2930 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
2931 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
2932 hierarchically set default memory protection values for a particular
2933 subtree of the unit hierarchy.
2935 * Memory protection directives can now take a value of zero, allowing
2936 explicit opting out of a default value propagated by an ancestor.
2938 * systemd now defaults to the "unified" cgroup hierarchy setup during
2939 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
2940 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
2941 change reflects the fact that cgroupsv2 support has matured
2942 substantially in both systemd and in the kernel, and is clearly the
2943 way forward. Downstream production distributions might want to
2944 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
2945 their builds as unfortunately the popular container managers have not
2946 caught up with the kernel API changes.
2948 * Man pages are not built by default anymore (html pages were already
2949 disabled by default), to make development builds quicker. When
2950 building systemd for a full installation with documentation, meson
2951 should be called with -Dman=true and/or -Dhtml=true as appropriate.
2952 The default was changed based on the assumption that quick one-off or
2953 repeated development builds are much more common than full optimized
2954 builds for installation, and people need to pass various other
2955 options to when doing "proper" builds anyway, so the gain from making
2956 development builds quicker is bigger than the one time disruption for
2959 Two scripts are created in the *build* directory to generate and
2960 preview man and html pages on demand, e.g.:
2962 build/man/man systemctl
2963 build/man/html systemd.index
2965 * libidn2 is used by default if both libidn2 and libidn are installed.
2966 Please use -Dlibidn=true if libidn is preferred.
2968 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
2969 big-endian machines. Before, bytes were written and read in native
2970 machine order as exposed by the native libc __cpu_mask interface.
2971 Now, little-endian order is always used (CPUs 0–7 are described by
2972 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
2973 This change fixes D-Bus calls that cross endianness boundary.
2975 The presentation format used for CPUAffinity= by "systemctl show" and
2976 "systemd-analyze dump" is changed to present CPU indices instead of
2977 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
2978 shown as CPUAffinity=03000000000000000000000000000… (on
2979 little-endian) or CPUAffinity=00000000000000300000000000000… (on
2980 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
2981 input format. The maximum integer that will be printed in the new
2982 format is 8191 (four digits), while the old format always used a very
2983 long number (with the length varying by architecture), so they can be
2984 unambiguously distinguished.
2986 * /usr/sbin/halt.local is no longer supported. Implementation in
2987 distributions was inconsistent and it seems this functionality was
2990 To replace this functionality, users should:
2991 - either define a new unit and make it a dependency of final.target
2992 (systemctl add-wants final.target my-halt-local.service)
2993 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
2994 and ensure that it accepts "halt", "poweroff", "reboot", and
2995 "kexec" as an argument, see the description in systemd-shutdown(8).
2997 * When a [Match] section in .link or .network file is empty (contains
2998 no match patterns), a warning will be emitted. Please add any "match
2999 all" pattern instead, e.g. OriginalName=* or Name=* in case all
3000 interfaces should really be matched.
3002 * A new setting NUMAPolicy= may be used to set process memory
3003 allocation policy. This setting can be specified in
3004 /etc/systemd/system.conf and hence will set the default policy for
3005 PID1. The default policy can be overridden on a per-service
3006 basis. The related setting NUMAMask= is used to specify NUMA node
3007 mask that should be associated with the selected policy.
3009 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
3010 generates when processes it manages are reaching their memory limits,
3011 and will place their units in a special state, and optionally kill or
3012 stop the whole unit.
3014 * The service manager will now expose bus properties for the IO
3015 resources used by units. This information is also shown in "systemctl
3016 status" now (for services that have IOAccounting=yes set). Moreover,
3017 the IO accounting data is included in the resource log message
3018 generated whenever a unit stops.
3020 * Units may now configure an explicit timeout to wait for when killed
3021 with SIGABRT, for example when a service watchdog is hit. Previously,
3022 the regular TimeoutStopSec= timeout was applied in this case too —
3023 now a separate timeout may be set using TimeoutAbortSec=.
3025 * Services may now send a special WATCHDOG=trigger message with
3026 sd_notify() to trigger an immediate "watchdog missed" event, and thus
3027 trigger service termination. This is useful both for testing watchdog
3028 handling, but also for defining error paths in services, that shall
3029 be handled the same way as watchdog events.
3031 * There are two new per-unit settings IPIngressFilterPath= and
3032 IPEgressFilterPath= which allow configuration of a BPF program
3033 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
3034 to apply to the IP packet ingress/egress path of all processes of a
3035 unit. This is useful to allow running systemd services with BPF
3036 programs set up externally.
3038 * systemctl gained a new "clean" verb for removing the state, cache,
3039 runtime or logs directories of a service while it is terminated. The
3040 new verb may also be used to remove the state maintained on disk for
3041 timer units that have Persistent= configured.
3043 * During the last phase of shutdown systemd will now automatically
3044 increase the log level configured in the "kernel.printk" sysctl so
3045 that any relevant loggable events happening during late shutdown are
3046 made visible. Previously, loggable events happening so late during
3047 shutdown were generally lost if the "kernel.printk" sysctl was set to
3048 high thresholds, as regular logging daemons are terminated at that
3049 time and thus nothing is written to disk.
3051 * If processes terminated during the last phase of shutdown do not exit
3052 quickly systemd will now show their names after a short time, to make
3053 debugging easier. After a longer timeout they are forcibly killed,
3056 * journalctl (and the other tools that display logs) will now highlight
3057 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
3058 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
3059 are now shown in blue color, to separate them visually from regular
3060 logs. References to configuration files are now turned into clickable
3061 links on terminals that support that.
3063 * systemd-journald will now stop logging to /var/log/journal during
3064 shutdown when /var/ is on a separate mount, so that it can be
3065 unmounted safely during shutdown.
3067 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
3069 * systemd-resolved "Cache=" configuration option in resolved.conf has
3070 been extended to also accept the 'no-negative' value. Previously,
3071 only a boolean option was allowed (yes/no), having yes as the
3072 default. If this option is set to 'no-negative', negative answers are
3073 not cached while the old cache heuristics are used positive answers.
3074 The default remains unchanged.
3076 * The predictable naming scheme for network devices now supports
3077 generating predictable names for "netdevsim" devices.
3079 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
3082 Those two changes form a new net.naming-policy-scheme= entry.
3083 Distributions which want to preserve naming stability may want to set
3084 the -Ddefault-net-naming-scheme= configuration option.
3086 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
3087 interfaces natively.
3089 * systemd-networkd's bridge FDB support now allows configuration of a
3090 destination address for each entry (Destination=), as well as the
3091 VXLAN VNI (VNI=), as well as an option to declare what an entry is
3092 associated with (AssociatedWith=).
3094 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
3095 option for configuring the maximum number of DHCP lease requests. It
3096 also learnt a new BlackList= option for deny-listing DHCP servers (a
3097 similar setting has also been added to the IPv6 RA client), as well
3098 as a SendRelease= option for configuring whether to send a DHCP
3099 RELEASE message when terminating.
3101 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
3102 separately in the [DHCPv4] and [DHCPv6] sections.
3104 * systemd-networkd's DHCP support will now optionally create an
3105 implicit host route to the DNS server specified in the DHCP lease, in
3106 addition to the routes listed explicitly in the lease. This should
3107 ensure that in multi-homed systems DNS traffic leaves the systems on
3108 the interface that acquired the DNS server information even if other
3109 routes such as default routes exist. This behaviour may be turned on
3110 with the new RoutesToDNS= option.
3112 * systemd-networkd's VXLAN support gained a new option
3113 GenericProtocolExtension= for enabling VXLAN Generic Protocol
3114 Extension support, as well as IPDoNotFragment= for setting the IP
3115 "Don't fragment" bit on outgoing packets. A similar option has been
3116 added to the GENEVE support.
3118 * In systemd-networkd's [Route] section you may now configure
3119 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
3120 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
3121 propagation. The Type= setting now supports local, broadcast,
3122 anycast, multicast, any, xresolve routes, too.
3124 * systemd-networkd's [Network] section learnt a new option
3125 DefaultRouteOnDevice= for automatically configuring a default route
3126 onto the network device.
3128 * systemd-networkd's bridging support gained two new options ProxyARP=
3129 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
3130 MulticastRouter= for configuring multicast routing behaviour. A new
3131 option MulticastIGMPVersion= may be used to change bridge's multicast
3132 Internet Group Management Protocol (IGMP) version.
3134 * systemd-networkd's FooOverUDP support gained the ability to configure
3135 local and peer IP addresses via Local= and Peer=. A new option
3136 PeerPort= may be used to configure the peer's IP port.
3138 * systemd-networkd's TUN support gained a new setting VnetHeader= for
3139 tweaking Generic Segment Offload support.
3141 * The address family for policy rules may be specified using the new
3142 Family= option in the [RoutingPolicyRule] section.
3144 * networkctl gained a new "delete" command for removing virtual network
3145 devices, as well as a new "--stats" switch for showing device
3148 * networkd.conf gained a new setting SpeedMeter= and
3149 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
3150 measured speed may be shown by 'networkctl status'.
3152 * "networkctl status" now displays MTU and queue lengths, and more
3153 detailed information about VXLAN and bridge devices.
3155 * systemd-networkd's .network and .link files gained a new Property=
3156 setting in the [Match] section, to match against devices with
3157 specific udev properties.
3159 * systemd-networkd's tunnel support gained a new option
3160 AssignToLoopback= for selecting whether to use the loopback device
3161 "lo" as underlying device.
3163 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
3164 been renamed to LinkLayerAddress=, and it now allows configuration of
3167 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
3168 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
3169 IPv6 configuration (static or DHCPv6) was found for a given
3170 interface. It will not touch the sysctl otherwise.
3172 * The order of entries is $PATH used by the user manager instance was
3173 changed to put bin/ entries before the corresponding sbin/ entries.
3174 It is recommended to not rely on this order, and only ever have one
3175 binary with a given name in the system paths under /usr.
3177 * A new tool systemd-network-generator has been added that may generate
3178 .network, .netdev and .link files from IP configuration specified on
3179 the kernel command line in the format used by Dracut.
3181 * The CriticalConnection= setting in .network files is now deprecated,
3182 and replaced by a new KeepConfiguration= setting which allows more
3183 detailed configuration of the IP configuration to keep in place.
3185 * systemd-analyze gained a few new verbs:
3187 - "systemd-analyze timestamp" parses and converts timestamps. This is
3188 similar to the existing "systemd-analyze calendar" command which
3189 does the same for recurring calendar events.
3191 - "systemd-analyze timespan" parses and converts timespans (i.e.
3192 durations as opposed to points in time).
3194 - "systemd-analyze condition" will parse and test ConditionXYZ=
3197 - "systemd-analyze exit-status" will parse and convert exit status
3198 codes to their names and back.
3200 - "systemd-analyze unit-files" will print a list of all unit
3201 file paths and unit aliases.
3203 * SuccessExitStatus=, RestartPreventExitStatus=, and
3204 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
3205 is equivalent to "65"). Those exit status name mappings may be
3206 displayed with the systemd-analyze exit-status verb describe above.
3208 * systemd-logind now exposes a per-session SetBrightness() bus call,
3209 which may be used to securely change the brightness of a kernel
3210 brightness device, if it belongs to the session's seat. By using this
3211 call unprivileged clients can make changes to "backlight" and "leds"
3212 devices securely with strict requirements on session membership.
3213 Desktop environments may use this to generically make brightness
3214 changes to such devices without shipping private SUID binaries or
3215 udev rules for that purpose.
3217 * "udevadm info" gained a --wait-for-initialization switch to wait for
3218 a device to be initialized.
3220 * systemd-hibernate-resume-generator will now look for resumeflags= on
3221 the kernel command line, which is similar to rootflags= and may be
3222 used to configure device timeout for the hibernation device.
3224 * sd-event learnt a new API call sd_event_source_disable_unref() for
3225 disabling and unref'ing an event source in a single function. A
3226 related call sd_event_source_disable_unrefp() has been added for use
3227 with gcc's cleanup extension.
3229 * The sd-id128.h public API gained a new definition
3230 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
3233 * "busctl introspect" gained a new switch --xml-interface for dumping
3234 XML introspection data unmodified.
3236 * PID 1 may now show the unit name instead of the unit description
3237 string in its status output during boot. This may be configured in
3238 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
3239 kernel command line option systemd.status_unit_format=.
3241 * PID 1 now understands a new option KExecWatchdogSec= in
3242 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
3243 Previously watchdog functionality was only available for regular
3244 reboots. The new setting defaults to off, because we don't know in
3245 the general case if the watchdog will be reset after kexec (some
3246 drivers do reset it, but not all), and the new userspace might not be
3247 configured to handle the watchdog.
3249 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
3250 RebootWatchdogSec= to more clearly communicate what it is about. The
3251 old name is still accepted for compatibility.
3253 * The systemd.debug_shell kernel command line option now optionally
3254 takes a tty name to spawn the debug shell on, which allows a
3255 different tty to be selected than the built-in default.
3257 * Service units gained a new ExecCondition= setting which will run
3258 before ExecStartPre= and either continue execution of the unit (for
3259 clean exit codes), stop execution without marking the unit failed
3260 (for exit codes 1 through 254), or stop execution and fail the unit
3261 (for exit code 255 or abnormal termination).
3263 * A new service systemd-pstore.service has been added that pulls data
3264 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
3267 * timedatectl gained new verbs for configuring per-interface NTP
3268 service configuration for systemd-timesyncd.
3270 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
3271 2019. (You can set non-UTF-8 locales though, if you know their name.)
3273 * If variable assignments in sysctl.d/ files are prefixed with "-" any
3274 failures to apply them are now ignored.
3276 * systemd-random-seed.service now optionally credits entropy when
3277 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
3278 true for the service to enable this behaviour, but please consult the
3279 documentation first, since this comes with a couple of caveats.
3281 * systemd-random-seed.service is now a synchronization point for full
3282 initialization of the kernel's entropy pool. Services that require
3283 /dev/urandom to be correctly initialized should be ordered after this
3286 * The systemd-boot boot loader has been updated to optionally maintain
3287 a random seed file in the EFI System Partition (ESP). During the boot
3288 phase, this random seed is read and updated with a new seed
3289 cryptographically derived from it. Another derived seed is passed to
3290 the OS. The latter seed is then credited to the kernel's entropy pool
3291 very early during userspace initialization (from PID 1). This allows
3292 systems to boot up with a fully initialized kernel entropy pool from
3293 earliest boot on, and thus entirely removes all entropy pool
3294 initialization delays from systems using systemd-boot. Special care
3295 is taken to ensure different seeds are derived on system images
3296 replicated to multiple systems. "bootctl status" will show whether
3297 a seed was received from the boot loader.
3299 * bootctl gained two new verbs:
3301 - "bootctl random-seed" will generate the file in ESP and an EFI
3302 variable to allow a random seed to be passed to the OS as described
3305 - "bootctl is-installed" checks whether systemd-boot is currently
3308 * bootctl will warn if it detects that boot entries are misconfigured
3309 (for example if the kernel image was removed without purging the
3312 * A new document has been added describing systemd's use and support
3313 for the kernel's entropy pool subsystem:
3315 https://systemd.io/RANDOM_SEEDS
3317 * When the system is hibernated the swap device to write the
3318 hibernation image to is now automatically picked from all available
3319 swap devices, preferring the swap device with the highest configured
3320 priority over all others, and picking the device with the most free
3321 space if there are multiple devices with the highest priority.
3323 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
3324 option that permits selecting the timeout how long to wait for a
3325 device with an encryption key before asking for the password.
3327 * IOWeight= has learnt to properly set the IO weight when using the
3328 BFQ scheduler officially found in kernels 5.0+.
3330 * A new mailing list has been created for reporting of security issues:
3331 systemd-security@redhat.com. For mode details, see
3332 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
3334 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
3335 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
3336 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
3337 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
3338 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
3339 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
3340 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
3341 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
3342 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
3343 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
3344 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
3345 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
3346 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
3347 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
3348 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
3349 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
3350 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
3351 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
3352 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
3353 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
3354 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
3355 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
3356 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
3357 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
3358 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
3359 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
3360 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
3361 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
3362 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
3363 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
3365 – Camerino, 2019-09-03
3369 * In .link files, MACAddressPolicy=persistent (the default) is changed
3370 to cover more devices. For devices like bridges, tun, tap, bond, and
3371 similar interfaces that do not have other identifying information,
3372 the interface name is used as the basis for persistent seed for MAC
3373 and IPv4LL addresses. The way that devices that were handled
3374 previously is not changed, and this change is about covering more
3375 devices then previously by the "persistent" policy.
3377 MACAddressPolicy=random may be used to force randomized MACs and
3378 IPv4LL addresses for a device if desired.
3380 Hint: the log output from udev (at debug level) was enhanced to
3381 clarify what policy is followed and which attributes are used.
3382 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
3383 may be used to view this.
3385 Hint: if a bridge interface is created without any slaves, and gains
3386 a slave later, then now the bridge does not inherit slave's MAC.
3387 To inherit slave's MAC, for example, create the following file:
3389 # /etc/systemd/network/98-bridge-inherit-mac.link
3394 MACAddressPolicy=none
3397 * The .device units generated by systemd-fstab-generator and other
3398 generators do not automatically pull in the corresponding .mount unit
3399 as a Wants= dependency. This means that simply plugging in the device
3400 will not cause the mount unit to be started automatically. But please
3401 note that the mount unit may be started for other reasons, in
3402 particular if it is part of local-fs.target, and any unit which
3403 (transitively) depends on local-fs.target is started.
3405 * networkctl list/status/lldp now accept globbing wildcards for network
3406 interface names to match against all existing interfaces.
3408 * The $PIDFILE environment variable is set to point the absolute path
3409 configured with PIDFile= for processes of that service.
3411 * The fallback DNS server list was augmented with Cloudflare public DNS
3412 servers. Use `-Ddns-servers=` to set a different fallback.
3414 * A new special target usb-gadget.target will be started automatically
3415 when a USB Device Controller is detected (which means that the system
3416 is a USB peripheral).
3418 * A new unit setting CPUQuotaPeriodSec= assigns the time period
3419 relatively to which the CPU time quota specified by CPUQuota= is
3422 * A new unit setting ProtectHostname= may be used to prevent services
3423 from modifying hostname information (even if they otherwise would
3424 have privileges to do so).
3426 * A new unit setting NetworkNamespacePath= may be used to specify a
3427 namespace for service or socket units through a path referring to a
3428 Linux network namespace pseudo-file.
3430 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
3431 have an effect on .socket units: when used the listening socket is
3432 created within the configured network namespace instead of the host
3435 * ExecStart= command lines in unit files may now be prefixed with ':'
3436 in which case environment variable substitution is
3437 disabled. (Supported for the other ExecXYZ= settings, too.)
3439 * .timer units gained two new boolean settings OnClockChange= and
3440 OnTimezoneChange= which may be used to also trigger a unit when the
3441 system clock is changed or the local timezone is
3442 modified. systemd-run has been updated to make these options easily
3443 accessible from the command line for transient timers.
3445 * Two new conditions for units have been added: ConditionMemory= may be
3446 used to conditionalize a unit based on installed system
3447 RAM. ConditionCPUs= may be used to conditionalize a unit based on
3448 installed CPU cores.
3450 * The @default system call filter group understood by SystemCallFilter=
3451 has been updated to include the new rseq() system call introduced in
3454 * A new time-set.target has been added that indicates that the system
3455 time has been set from a local source (possibly imprecise). The
3456 existing time-sync.target is stronger and indicates that the time has
3457 been synchronized with a precise external source. Services where
3458 approximate time is sufficient should use the new target.
3460 * "systemctl start" (and related commands) learnt a new
3461 --show-transaction option. If specified brief information about all
3462 jobs queued because of the requested operation is shown.
3464 * systemd-networkd recognizes a new operation state 'enslaved', used
3465 (instead of 'degraded' or 'carrier') for interfaces which form a
3466 bridge, bond, or similar, and an new 'degraded-carrier' operational
3467 state used for the bond or bridge master interface when one of the
3468 enslaved devices is not operational.
3470 * .network files learnt the new IgnoreCarrierLoss= option for leaving
3471 networks configured even if the carrier is lost.
3473 * The RequiredForOnline= setting in .network files may now specify a
3474 minimum operational state required for the interface to be considered
3475 "online" by systemd-networkd-wait-online. Related to this
3476 systemd-networkd-wait-online gained a new option --operational-state=
3477 to configure the same, and its --interface= option was updated to
3478 optionally also take an operational state specific for an interface.
3480 * systemd-networkd-wait-online gained a new setting --any for waiting
3481 for only one of the requested interfaces instead of all of them.
3483 * systemd-networkd now implements L2TP tunnels.
3485 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
3486 may be used to cause autonomous and onlink prefixes received in IPv6
3487 Router Advertisements to be ignored.
3489 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
3490 file settings may be used to tweak bridge behaviour.
3492 * The new TripleSampling= option in .network files may be used to
3493 configure CAN triple sampling.
3495 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
3496 used to point to private or preshared key for a WireGuard interface.
3498 * /etc/crypttab now supports the same-cpu-crypt and
3499 submit-from-crypt-cpus options to tweak encryption work scheduling
3502 * systemd-tmpfiles will now take a BSD file lock before operating on a
3503 contents of directory. This may be used to temporarily exclude
3504 directories from aging by taking the same lock (useful for example
3505 when extracting a tarball into /tmp or /var/tmp as a privileged user,
3506 which might create files with really old timestamps, which
3507 nevertheless should not be deleted). For further details, see:
3509 https://systemd.io/TEMPORARY_DIRECTORIES
3511 * systemd-tmpfiles' h line type gained support for the
3512 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
3513 controlling project quota inheritance.
3515 * sd-boot and bootctl now implement support for an Extended Boot Loader
3516 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
3517 addition to the ESP partition mounted to /efi or /boot/efi.
3518 Configuration file fragments, kernels, initrds and other EFI images
3519 to boot will be loaded from both the ESP and XBOOTLDR partitions.
3520 The XBOOTLDR partition was previously described by the Boot Loader
3521 Specification, but implementation was missing in sd-boot. Support for
3522 this concept allows using the sd-boot boot loader in more
3523 conservative scenarios where the boot loader itself is placed in the
3524 ESP but the kernels to boot (and their metadata) in a separate
3527 * A system may now be booted with systemd.volatile=overlay on the
3528 kernel command line, which causes the root file system to be set up
3529 an overlayfs mount combining the root-only root directory with a
3530 writable tmpfs. In this setup, the underlying root device is not
3531 modified, and any changes are lost at reboot.
3533 * Similar, systemd-nspawn can now boot containers with a volatile
3534 overlayfs root with the new --volatile=overlay switch.
3536 * systemd-nspawn can now consume OCI runtime bundles using a new
3537 --oci-bundle= option. This implementation is fully usable, with most
3538 features in the specification implemented, but since this a lot of
3539 new code and functionality, this feature should most likely not
3540 be used in production yet.
3542 * systemd-nspawn now supports various options described by the OCI
3543 runtime specification on the command-line and in .nspawn files:
3544 --inaccessible=/Inaccessible= may be used to mask parts of the file
3545 system tree, --console=/--pipe may be used to configure how standard
3546 input, output, and error are set up.
3548 * busctl learned the `emit` verb to generate D-Bus signals.
3550 * systemd-analyze cat-config may be used to gather and display
3551 configuration spread over multiple files, for example system and user
3552 presets, tmpfiles.d, sysusers.d, udev rules, etc.
3554 * systemd-analyze calendar now takes an optional new parameter
3555 --iterations= which may be used to show a maximum number of iterations
3556 the specified expression will elapse next.
3558 * The sd-bus C API gained support for naming method parameters in the
3561 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
3562 the reboot() system call expects.
3564 * journalctl learnt a new --cursor-file= option that points to a file
3565 from which a cursor should be loaded in the beginning and to which
3566 the updated cursor should be stored at the end.
3568 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
3569 detected by systemd-detect-virt (and may also be used in
3570 ConditionVirtualization=).
3572 * The behaviour of systemd-logind may now be modified with environment
3573 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
3574 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
3575 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
3576 skip the relevant operation completely (when set to false), or to
3577 create a flag file in /run/systemd (when set to true), instead of
3578 actually commencing the real operation when requested. The presence
3579 of /run/systemd/reboot-to-firmware-setup,
3580 /run/systemd/reboot-to-boot-loader-menu, and
3581 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
3582 boot loader implementations to replace some steps logind performs
3583 during reboot with their own operations.
3585 * systemctl can be used to request a reboot into the boot loader menu
3586 or a specific boot loader entry with the new --boot-load-menu= and
3587 --boot-loader-entry= options to a reboot command. (This requires a
3588 boot loader that supports this, for example sd-boot.)
3590 * kernel-install will no longer unconditionally create the output
3591 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
3592 snippets, but will do only if the machine-specific parent directory
3593 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
3594 to create this parent directory during sd-boot installation.
3596 This makes it easier to use kernel-install with plugins which support
3597 a different layout of the bootloader partitions (for example grub2).
3599 * During package installation (with `ninja install`), we would create
3600 symlinks for getty@tty1.service, systemd-networkd.service,
3601 systemd-networkd.socket, systemd-resolved.service,
3602 remote-cryptsetup.target, remote-fs.target,
3603 systemd-networkd-wait-online.service, and systemd-timesyncd.service
3604 in /etc, as if `systemctl enable` was called for those units, to make
3605 the system usable immediately after installation. Now this is not
3606 done anymore, and instead calling `systemctl preset-all` is
3607 recommended after the first installation of systemd.
3609 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
3610 is built on seccomp. When turned on creation of SUID/SGID files is
3613 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
3614 implied if DynamicUser= is turned on for a service. This hardens
3615 these services, so that they neither can benefit from nor create
3616 SUID/SGID executables. This is a minor compatibility breakage, given
3617 that when DynamicUser= was first introduced SUID/SGID behaviour was
3618 unaffected. However, the security benefit of these two options is
3619 substantial, and the setting is still relatively new, hence we opted
3620 to make it mandatory for services with dynamic users.
3622 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
3623 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
3624 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
3625 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
3626 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
3627 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
3628 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
3629 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
3630 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
3631 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
3632 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
3633 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
3634 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
3635 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
3636 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
3637 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
3638 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
3639 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
3641 — Warsaw, 2019-04-11
3645 * The default locale can now be configured at compile time. Otherwise,
3646 a suitable default will be selected automatically (one of C.UTF-8,
3647 en_US.UTF-8, and C).
3649 * The version string shown by systemd and other tools now includes the
3650 git commit hash when built from git. An override may be specified
3651 during compilation, which is intended to be used by distributions to
3652 include the package release information.
3654 * systemd-cat can now filter standard input and standard error streams
3655 for different syslog priorities using the new --stderr-priority=
3658 * systemd-journald and systemd-journal-remote reject entries which
3659 contain too many fields (CVE-2018-16865) and set limits on the
3660 process' command line length (CVE-2018-16864).
3662 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
3665 * A new network device NamePolicy "keep" is implemented for link files,
3666 and used by default in 99-default.link (the fallback configuration
3667 provided by systemd). With this policy, if the network device name
3668 was already set by userspace, the device will not be renamed again.
3669 This matches the naming scheme that was implemented before
3670 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
3671 is also enabled by default, even if not specified. Effectively, this
3672 means that if naming-scheme >= 240 is specified, network devices will
3673 be renamed according to the configuration, even if they have been
3674 renamed already, if "keep" is not specified as the naming policy in
3675 the .link file. The 99-default.link file provided by systemd includes
3676 "keep" for backwards compatibility, but it is recommended for user
3677 installed .link files to *not* include it.
3679 The "kernel" policy, which keeps kernel names declared to be
3680 "persistent", now works again as documented.
3682 * kernel-install script now optionally takes the paths to one or more
3683 initrd files, and passes them to all plugins.
3685 * The mincore() system call has been dropped from the @system-service
3686 system call filter group, as it is pretty exotic and may potentially
3687 used for side-channel attacks.
3689 * -fPIE is dropped from compiler and linker options. Please specify
3690 -Db_pie=true option to meson to build position-independent
3691 executables. Note that the meson option is supported since meson-0.49.
3693 * The fs.protected_regular and fs.protected_fifos sysctls, which were
3694 added in Linux 4.19 to make some data spoofing attacks harder, are
3695 now enabled by default. While this will hopefully improve the
3696 security of most installations, it is technically a backwards
3697 incompatible change; to disable these sysctls again, place the
3698 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
3700 fs.protected_regular = 0
3701 fs.protected_fifos = 0
3703 Note that the similar hardlink and symlink protection has been
3704 enabled since v199, and may be disabled likewise.
3706 * The files read from the EnvironmentFile= setting in unit files now
3707 parse backslashes inside quotes literally, matching the behaviour of
3710 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
3711 now automatically become NOPs when run in a chroot() environment.
3713 * The tmpfiles.d/ "C" line type will now copy directory trees not only
3714 when the destination is so far missing, but also if it already exists
3715 as a directory and is empty. This is useful to cater for systems
3716 where directory trees are put together from multiple separate mount
3717 points but otherwise empty.
3719 * A new function sd_bus_close_unref() (and the associated
3720 sd_bus_close_unrefp()) has been added to libsystemd, that combines
3721 sd_bus_close() and sd_bus_unref() in one.
3723 * udevadm control learnt a new option for --ping for testing whether a
3724 systemd-udevd instance is running and reacting.
3726 * udevadm trigger learnt a new option for --wait-daemon for waiting
3727 systemd-udevd daemon to be initialized.
3729 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
3730 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
3731 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
3732 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
3733 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
3734 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
3735 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
3736 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
3737 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
3738 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
3739 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
3740 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
3741 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
3742 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
3743 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
3744 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
3745 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
3747 — Berlin, 2019-02-14
3751 * NoNewPrivileges=yes has been set for all long-running services
3752 implemented by systemd. Previously, this was problematic due to
3753 SELinux (as this would also prohibit the transition from PID1's label
3754 to the service's label). This restriction has since been lifted, but
3755 an SELinux policy update is required.
3756 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
3758 * DynamicUser=yes is dropped from systemd-networkd.service,
3759 systemd-resolved.service and systemd-timesyncd.service, which was
3760 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
3761 and since v236 for systemd-timesyncd.service. The users and groups
3762 systemd-network, systemd-resolve and systemd-timesync are created
3763 by systemd-sysusers again. Distributors or system administrators
3764 may need to create these users and groups if they not exist (or need
3765 to re-enable DynamicUser= for those units) while upgrading systemd.
3766 Also, the clock file for systemd-timesyncd may need to move from
3767 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
3769 * When unit files are loaded from disk, previously systemd would
3770 sometimes (depending on the unit loading order) load units from the
3771 target path of symlinks in .wants/ or .requires/ directories of other
3772 units. This meant that unit could be loaded from different paths
3773 depending on whether the unit was requested explicitly or as a
3774 dependency of another unit, not honouring the priority of directories
3775 in search path. It also meant that it was possible to successfully
3776 load and start units which are not found in the unit search path, as
3777 long as they were requested as a dependency and linked to from
3778 .wants/ or .requires/. The target paths of those symlinks are not
3779 used for loading units anymore and the unit file must be found in
3782 * A new service type has been added: Type=exec. It's very similar to
3783 Type=simple but ensures the service manager will wait for both fork()
3784 and execve() of the main service binary to complete before proceeding
3785 with follow-up units. This is primarily useful so that the manager
3786 propagates any errors in the preparation phase of service execution
3787 back to the job that requested the unit to be started. For example,
3788 consider a service that has ExecStart= set to a file system binary
3789 that doesn't exist. With Type=simple starting the unit would be
3790 considered instantly successful, as only fork() has to complete
3791 successfully and the manager does not wait for execve(), and hence
3792 its failure is seen "too late". With the new Type=exec service type
3793 starting the unit will fail, as the manager will wait for the
3794 execve() and notice its failure, which is then propagated back to the
3797 NOTE: with the next release 241 of systemd we intend to change the
3798 systemd-run tool to default to Type=exec for transient services
3799 started by it. This should be mostly safe, but in specific corner
3800 cases might result in problems, as the systemd-run tool will then
3801 block on NSS calls (such as user name look-ups due to User=) done
3802 between the fork() and execve(), which under specific circumstances
3803 might cause problems. It is recommended to specify "-p Type=simple"
3804 explicitly in the few cases where this applies. For regular,
3805 non-transient services (i.e. those defined with unit files on disk)
3806 we will continue to default to Type=simple.
3808 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
3809 userspace processes is set to 1024 (soft) and 4096
3810 (hard). Previously, systemd passed this on unmodified to all
3811 processes it forked off. With this systemd release the hard limit
3812 systemd passes on is increased to 512K, overriding the kernel's
3813 defaults and substantially increasing the number of simultaneous file
3814 descriptors unprivileged userspace processes can allocate. Note that
3815 the soft limit remains at 1024 for compatibility reasons: the
3816 traditional UNIX select() call cannot deal with file descriptors >=
3817 1024 and increasing the soft limit globally might thus result in
3818 programs unexpectedly allocating a high file descriptor and thus
3819 failing abnormally when attempting to use it with select() (of
3820 course, programs shouldn't use select() anymore, and prefer
3821 poll()/epoll, but the call unfortunately remains undeservedly popular
3822 at this time). This change reflects the fact that file descriptor
3823 handling in the Linux kernel has been optimized in more recent
3824 kernels and allocating large numbers of them should be much cheaper
3825 both in memory and in performance than it used to be. Programs that
3826 want to take benefit of the increased limit have to "opt-in" into
3827 high file descriptors explicitly by raising their soft limit. Of
3828 course, when they do that they must acknowledge that they cannot use
3829 select() anymore (and neither can any shared library they use — or
3830 any shared library used by any shared library they use and so on).
3831 Which default hard limit is most appropriate is of course hard to
3832 decide. However, given reports that ~300K file descriptors are used
3833 in real-life applications we believe 512K is sufficiently high as new
3834 default for now. Note that there are also reports that using very
3835 high hard limits (e.g. 1G) is problematic: some software allocates
3836 large arrays with one element for each potential file descriptor
3837 (Java, …) — a high hard limit thus triggers excessively large memory
3838 allocations in these applications. Hopefully, the new default of 512K
3839 is a good middle ground: higher than what real-life applications
3840 currently need, and low enough for avoid triggering excessively large
3841 allocations in problematic software. (And yes, somebody should fix
3844 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
3845 to the highest possible values, as separate accounting of file
3846 descriptors is no longer necessary, as memcg tracks them correctly as
3847 part of the memory accounting anyway. Thus, from the four limits on
3848 file descriptors currently enforced (fs.file-max, fs.nr_open,
3849 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
3850 and keep only the latter two. A set of build-time options
3851 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
3852 has been added to revert this change in behaviour, which might be
3853 an option for systems that turn off memcg in the kernel.
3855 * When no /etc/locale.conf file exists (and hence no locale settings
3856 are in place), systemd will now use the "C.UTF-8" locale by default,
3857 and set LANG= to it. This locale is supported by various
3858 distributions including Fedora, with clear indications that upstream
3859 glibc is going to make it available too. This locale enables UTF-8
3860 mode by default, which appears appropriate for 2018.
3862 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
3863 default. This effectively switches the RFC3704 Reverse Path filtering
3864 from Strict mode to Loose mode. This is more appropriate for hosts
3865 that have multiple links with routes to the same networks (e.g.
3866 a client with a Wi-Fi and Ethernet both connected to the internet).
3868 Consult the kernel documentation for details on this sysctl:
3869 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
3871 * CPUAccounting=yes no longer enables the CPU controller when using
3872 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
3873 statistics are now provided independently from the CPU controller.
3875 * Support for disabling a particular cgroup controller within a sub-tree
3876 has been added through the DisableControllers= directive.
3878 * cgroup_no_v1=all on the kernel command line now also implies
3879 using the unified cgroup hierarchy, unless one explicitly passes
3880 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
3882 * The new "MemoryMin=" unit file property may now be used to set the
3883 memory usage protection limit of processes invoked by the unit. This
3884 controls the cgroup v2 memory.min attribute. Similarly, the new
3885 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
3886 cgroup v2 io.latency cgroup property for configuring per-service I/O
3889 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
3890 to the cgroup v1 "devices" cgroup controller.
3892 * systemd-escape now is able to combine --unescape with --template. It
3893 also learnt a new option --instance for extracting and unescaping the
3894 instance part of a unit name.
3896 * sd-bus now provides the sd_bus_message_readv() which is similar to
3897 sd_bus_message_read() but takes a va_list object. The pair
3898 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
3899 has been added for configuring the default method call timeout to
3900 use. sd_bus_error_move() may be used to efficiently move the contents
3901 from one sd_bus_error structure to another, invalidating the
3902 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
3903 be used to control whether a bus connection object is automatically
3904 flushed when an sd-event loop is exited.
3906 * When processing classic BSD syslog log messages, journald will now
3907 save the original time-stamp string supplied in the new
3908 SYSLOG_TIMESTAMP= journal field. This permits consumers to
3909 reconstruct the original BSD syslog message more correctly.
3911 * StandardOutput=/StandardError= in service files gained support for
3912 new "append:…" parameters, for connecting STDOUT/STDERR of a service
3913 to a file, and appending to it.
3915 * The signal to use as last step of killing of unit processes is now
3916 configurable. Previously it was hard-coded to SIGKILL, which may now
3917 be overridden with the new KillSignal= setting. Note that this is the
3918 signal used when regular termination (i.e. SIGTERM) does not suffice.
3919 Similarly, the signal used when aborting a program in case of a
3920 watchdog timeout may now be configured too (WatchdogSignal=).
3922 * The XDG_SESSION_DESKTOP environment variable may now be configured in
3923 the pam_systemd argument line, using the new desktop= switch. This is
3924 useful to initialize it properly from a display manager without
3925 having to touch C code.
3927 * Most configuration options that previously accepted percentage values
3928 now also accept permille values with the '‰' suffix (instead of '%').
3930 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
3933 * systemd-resolved's configuration file resolved.conf gained a new
3934 option ReadEtcHosts= which may be used to turn off processing and
3935 honoring /etc/hosts entries.
3937 * The "--wait" switch may now be passed to "systemctl
3938 is-system-running", in which case the tool will synchronously wait
3939 until the system finished start-up.
3941 * hostnamed gained a new bus call to determine the DMI product UUID.
3943 * On x86-64 systemd will now prefer using the RDRAND processor
3944 instruction over /dev/urandom whenever it requires randomness that
3945 neither has to be crypto-grade nor should be reproducible. This
3946 should substantially reduce the amount of entropy systemd requests
3947 from the kernel during initialization on such systems, though not
3948 reduce it to zero. (Why not zero? systemd still needs to allocate
3949 UUIDs and such uniquely, which require high-quality randomness.)
3951 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
3952 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
3953 for forcing the "Other Information" bit in IPv6 RA messages. The
3954 bonding logic gained four new options AdActorSystemPriority=,
3955 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
3956 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
3957 shuffling of flows. The tunnel logic gained a new
3958 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
3959 Deployment. The policy rule logic gained four new options IPProtocol=,
3960 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
3961 support for the MulticastToUnicast= option. networkd also gained
3962 support for configuring static IPv4 ARP or IPv6 neighbor entries.
3964 * .preset files (as read by 'systemctl preset') may now be used to
3965 instantiate services.
3967 * /etc/crypttab now understands the sector-size= option to configure
3968 the sector size for an encrypted partition.
3970 * Key material for encrypted disks may now be placed on a formatted
3971 medium, and referenced from /etc/crypttab by the UUID of the file
3972 system, followed by "=" suffixed by the path to the key file.
3974 * The "collect" udev component has been removed without replacement, as
3975 it is neither used nor maintained.
3977 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
3978 LogsDirectory=, ConfigurationDirectory= settings are used in a
3979 service the executed processes will now receive a set of environment
3980 variables containing the full paths of these directories.
3981 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
3982 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
3983 are used. Note that these options may be used multiple times per
3984 service in which case the resulting paths will be concatenated and
3985 separated by colons.
3987 * Predictable interface naming has been extended to cover InfiniBand
3988 NICs. They will be exposed with an "ib" prefix.
3990 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
3991 which case the respective line failing is ignored.
3993 * .link files may now be used to configure the equivalent to the
3994 "ethtool advertise" commands.
3996 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
3997 alternative to libudev.h. Previously, the latter was just an internal
3998 wrapper around the former, but now these two APIs are exposed
4001 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
4002 which calculates an app-specific boot ID similar to how
4003 sd_id128_get_machine_app_specific() generates an app-specific machine
4006 * A new tool systemd-id128 has been added that can be used to determine
4007 and generate various 128bit IDs.
4009 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
4012 * systemd-hibernate-resume-generator will now honor the "noresume"
4013 kernel command line option, in which case it will bypass resuming
4014 from any hibernated image.
4016 * The systemd-sleep.conf configuration file gained new options
4017 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
4018 AllowHybridSleep= for prohibiting specific sleep modes even if the
4019 kernel exports them.
4021 * portablectl is now officially supported and has thus moved to
4024 * bootctl learnt the two new commands "set-default" and "set-oneshot"
4025 for setting the default boot loader item to boot to (either
4026 persistently or only for the next boot). This is currently only
4027 compatible with sd-boot, but may be implemented on other boot loaders
4028 too, that follow the boot loader interface. The updated interface is
4029 now documented here:
4031 https://systemd.io/BOOT_LOADER_INTERFACE
4033 * A new kernel command line option systemd.early_core_pattern= is now
4034 understood which may be used to influence the core_pattern PID 1
4035 installs during early boot.
4037 * busctl learnt two new options -j and --json= for outputting method
4038 call replies, properties and monitoring output in JSON.
4040 * journalctl's JSON output now supports simple ANSI coloring as well as
4041 a new "json-seq" mode for generating RFC7464 output.
4043 * Unit files now support the %g/%G specifiers that resolve to the UNIX
4044 group/GID of the service manager runs as, similar to the existing
4045 %u/%U specifiers that resolve to the UNIX user/UID.
4047 * systemd-logind learnt a new global configuration option
4048 UserStopDelaySec= that may be set in logind.conf. It specifies how
4049 long the systemd --user instance shall remain started after a user
4050 logs out. This is useful to speed up repetitive re-connections of the
4051 same user, as it means the user's service manager doesn't have to be
4052 stopped/restarted on each iteration, but can be reused between
4053 subsequent options. This setting defaults to 10s. systemd-logind also
4054 exports two new properties on its Manager D-Bus objects indicating
4055 whether the system's lid is currently closed, and whether the system
4058 * systemd gained support for a generic boot counting logic, which
4059 generically permits automatic reverting to older boot loader entries
4060 if newer updated ones don't work. The boot loader side is implemented
4061 in sd-boot, but is kept open for other boot loaders too. For details
4064 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
4066 * The SuccessAction=/FailureAction= unit file settings now learnt two
4067 new parameters: "exit" and "exit-force", which result in immediate
4068 exiting of the service manager, and are only useful in systemd --user
4069 and container environments.
4071 * Unit files gained support for a pair of options
4072 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
4073 exit status to use as service manager exit status when
4074 SuccessAction=/FailureAction= is set to exit or exit-force.
4076 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
4077 options may now be used to configure the log rate limiting applied by
4078 journald per-service.
4080 * systemd-analyze gained a new verb "timespan" for parsing and
4081 normalizing time span values (i.e. strings like "5min 7s 8us").
4083 * systemd-analyze also gained a new verb "security" for analyzing the
4084 security and sand-boxing settings of services in order to determine an
4085 "exposure level" for them, indicating whether a service would benefit
4086 from more sand-boxing options turned on for them.
4088 * "systemd-analyze syscall-filter" will now also show system calls
4089 supported by the local kernel but not included in any of the defined
4092 * .nspawn files now understand the Ephemeral= setting, matching the
4093 --ephemeral command line switch.
4095 * sd-event gained the new APIs sd_event_source_get_floating() and
4096 sd_event_source_set_floating() for controlling whether a specific
4097 event source is "floating", i.e. destroyed along with the even loop
4100 * Unit objects on D-Bus gained a new "Refs" property that lists all
4101 clients that currently have a reference on the unit (to ensure it is
4104 * The JoinControllers= option in system.conf is no longer supported, as
4105 it didn't work correctly, is hard to support properly, is legacy (as
4106 the concept only exists on cgroup v1) and apparently wasn't used.
4108 * Journal messages that are generated whenever a unit enters the failed
4109 state are now tagged with a unique MESSAGE_ID. Similarly, messages
4110 generated whenever a service process exits are now made recognizable,
4111 too. A tagged message is also emitted whenever a unit enters the
4112 "dead" state on success.
4114 * systemd-run gained a new switch --working-directory= for configuring
4115 the working directory of the service to start. A shortcut -d is
4116 equivalent, setting the working directory of the service to the
4117 current working directory of the invoking program. The new --shell
4118 (or just -S) option has been added for invoking the $SHELL of the
4119 caller as a service, and implies --pty --same-dir --wait --collect
4120 --service-type=exec. Or in other words, "systemd-run -S" is now the
4121 quickest way to quickly get an interactive in a fully clean and
4122 well-defined system service context.
4124 * machinectl gained a new verb "import-fs" for importing an OS tree
4125 from a directory. Moreover, when a directory or tarball is imported
4126 and single top-level directory found with the OS itself below the OS
4127 tree is automatically mangled and moved one level up.
4129 * systemd-importd will no longer set up an implicit btrfs loop-back
4130 file system on /var/lib/machines. If one is already set up, it will
4131 continue to be used.
4133 * A new generator "systemd-run-generator" has been added. It will
4134 synthesize a unit from one or more program command lines included in
4135 the kernel command line. This is very useful in container managers
4138 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
4140 This will run "systemd-nspawn" on an image, invoke the specified
4141 command line and immediately shut down the container again, returning
4142 the command line's exit code.
4144 * The block device locking logic is now documented:
4146 https://systemd.io/BLOCK_DEVICE_LOCKING
4148 * loginctl and machinectl now optionally output the various tables in
4149 JSON using the --output= switch. It is our intention to add similar
4150 support to systemctl and all other commands.
4152 * udevadm's query and trigger verb now optionally take a .device unit
4155 * systemd-udevd's network naming logic now understands a new
4156 net.naming-scheme= kernel command line switch, which may be used to
4157 pick a specific version of the naming scheme. This helps stabilizing
4158 interface names even as systemd/udev are updated and the naming logic
4161 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
4162 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
4163 initialize one to all 0xFF.
4165 * After loading the SELinux policy systemd will now recursively relabel
4166 all files and directories listed in
4167 /run/systemd/relabel-extra.d/*.relabel (which should be simple
4168 newline separated lists of paths) in addition to the ones it already
4169 implicitly relabels in /run, /dev and /sys. After the relabelling is
4170 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
4171 removed. This is useful to permit initrds (i.e. code running before
4172 the SELinux policy is in effect) to generate files in the host
4173 filesystem safely and ensure that the correct label is applied during
4174 the transition to the host OS.
4176 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
4177 mknod() handling in user namespaces. Previously mknod() would always
4178 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
4179 but device nodes generated that way cannot be opened, and attempts to
4180 open them result in EPERM. This breaks the "graceful fallback" logic
4181 in systemd's PrivateDevices= sand-boxing option. This option is
4182 implemented defensively, so that when systemd detects it runs in a
4183 restricted environment (such as a user namespace, or an environment
4184 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
4185 where device nodes cannot be created the effect of PrivateDevices= is
4186 bypassed (following the logic that 2nd-level sand-boxing is not
4187 essential if the system systemd runs in is itself already sand-boxed
4188 as a whole). This logic breaks with 4.18 in container managers where
4189 user namespacing is used: suddenly PrivateDevices= succeeds setting
4190 up a private /dev/ file system containing devices nodes — but when
4191 these are opened they don't work.
4193 At this point it is recommended that container managers utilizing
4194 user namespaces that intend to run systemd in the payload explicitly
4195 block mknod() with seccomp or similar, so that the graceful fallback
4198 We are very sorry for the breakage and the requirement to change
4199 container configurations for newer kernels. It's purely caused by an
4200 incompatible kernel change. The relevant kernel developers have been
4201 notified about this userspace breakage quickly, but they chose to
4204 * PermissionsStartOnly= setting is deprecated (but is still supported
4205 for backwards compatibility). The same functionality is provided by
4206 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
4209 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
4210 pam_systemd anymore.
4212 * The naming scheme for network devices was changed to always rename
4213 devices, even if they were already renamed by userspace. The "kernel"
4214 policy was changed to only apply as a fallback, if no other naming
4217 * The requirements to build systemd is bumped to meson-0.46 and
4220 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
4221 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
4222 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
4223 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
4224 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
4225 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
4226 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
4227 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
4228 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
4229 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
4230 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
4231 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
4232 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
4233 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
4234 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
4235 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
4236 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
4237 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
4238 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
4239 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
4240 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
4241 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
4242 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
4243 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
4244 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
4245 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
4246 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
4247 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
4248 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
4249 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
4250 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
4251 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
4252 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
4253 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
4254 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
4255 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
4256 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
4257 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
4258 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
4259 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
4260 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
4261 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
4262 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
4263 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
4264 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
4266 — Warsaw, 2018-12-21
4270 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
4271 builtin will name network interfaces differently than in previous
4272 versions for virtual network interfaces created with SR-IOV and NPAR
4273 and for devices where the PCI network controller device does not have
4274 a slot number associated.
4276 SR-IOV virtual devices are now named based on the name of the parent
4277 interface, with a suffix of "v<N>", where <N> is the virtual device
4278 number. Previously those virtual devices were named as if completely
4281 The ninth and later NPAR virtual devices will be named following the
4282 scheme used for the first eight NPAR partitions. Previously those
4283 devices were not renamed and the kernel default (eth<n>) was used.
4285 "net_id" will also generate names for PCI devices where the PCI
4286 network controller device does not have an associated slot number
4287 itself, but one of its parents does. Previously those devices were
4288 not renamed and the kernel default (eth<n>) was used.
4290 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
4291 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
4292 the unit. So, it is expected that the default behavior of
4293 systemd-logind is not changed. However, if distribution packagers or
4294 administrators disabled or modified IPAddressDeny= setting by a
4295 drop-in config file, then it may be necessary to update the file to
4296 re-enable AF_INET and AF_INET6 to support network user name services,
4299 * When the RestrictNamespaces= unit property is specified multiple
4300 times, then the specified types are merged now. Previously, only the
4301 last assignment was used. So, if distribution packagers or
4302 administrators modified the setting by a drop-in config file, then it
4303 may be necessary to update the file.
4305 * When OnFailure= is used in combination with Restart= on a service
4306 unit, then the specified units will no longer be triggered on
4307 failures that result in restarting. Previously, the specified units
4308 would be activated each time the unit failed, even when the unit was
4309 going to be restarted automatically. This behaviour contradicted the
4310 documentation. With this release the code is adjusted to match the
4313 * systemd-tmpfiles will now print a notice whenever it encounters
4314 tmpfiles.d/ lines referencing the /var/run/ directory. It will
4315 recommend reworking them to use the /run/ directory instead (for
4316 which /var/run/ is simply a symlinked compatibility alias). This way
4317 systemd-tmpfiles can properly detect line conflicts and merge lines
4318 referencing the same file by two paths, without having to access
4321 * systemctl disable/unmask/preset/preset-all cannot be used with
4322 --runtime. Previously this was allowed, but resulted in unintuitive
4323 behaviour that wasn't useful. systemctl disable/unmask will now undo
4324 both runtime and persistent enablement/masking, i.e. it will remove
4325 any relevant symlinks both in /run and /etc.
4327 * Note that all long-running system services shipped with systemd will
4328 now default to a system call allow list (rather than a deny list, as
4329 before). In particular, systemd-udevd will now enforce one too. For
4330 most cases this should be safe, however downstream distributions
4331 which disabled sandboxing of systemd-udevd (specifically the
4332 MountFlags= setting), might want to disable this security feature
4333 too, as the default allow-listing will prohibit all mount, swap,
4334 reboot and clock changing operations from udev rules.
4336 * sd-boot acquired new loader configuration settings to optionally turn
4337 off Windows and MacOS boot partition discovery as well as
4338 reboot-into-firmware menu items. It is also able to pick a better
4339 screen resolution for HiDPI systems, and now provides loader
4340 configuration settings to change the resolution explicitly.
4342 * systemd-resolved now supports DNS-over-TLS. It's still
4343 turned off by default, use DNSOverTLS=opportunistic to turn it on in
4344 resolved.conf. We intend to make this the default as soon as couple
4345 of additional techniques for optimizing the initial latency caused by
4346 establishing a TLS/TCP connection are implemented.
4348 * systemd-resolved.service and systemd-networkd.service now set
4349 DynamicUser=yes. The users systemd-resolve and systemd-network are
4350 not created by systemd-sysusers anymore.
4352 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
4353 that embed a network facing module into any process using getpwuid()
4354 or related call: the dynamic allocation of the user ID for
4355 systemd-resolved.service means the service manager has to check NSS
4356 if the user name is already taken when forking off the service. Since
4357 the user in the common case won't be defined in /etc/passwd the
4358 lookup is likely to trigger nss-ldap which in turn might use NSS to
4359 ask systemd-resolved for hostname lookups. This will hence result in
4360 a deadlock: a user name lookup in order to start
4361 systemd-resolved.service will result in a hostname lookup for which
4362 systemd-resolved.service needs to be started already. There are
4363 multiple ways to work around this problem: pre-allocate the
4364 "systemd-resolve" user on such systems, so that nss-ldap won't be
4365 triggered; or use a different NSS package that doesn't do networking
4366 in-process but provides a local asynchronous name cache; or configure
4367 the NSS package to avoid lookups for UIDs in the range `pkg-config
4368 systemd --variable=dynamicuidmin` … `pkg-config systemd
4369 --variable=dynamicuidmax`, so that it does not consider itself
4370 authoritative for the same UID range systemd allocates dynamic users
4373 * The systemd-resolve tool has been renamed to resolvectl (it also
4374 remains available under the old name, for compatibility), and its
4375 interface is now verb-based, similar in style to the other <xyz>ctl
4376 tools, such as systemctl or loginctl.
4378 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
4379 compatibility. It may be symlinked under the 'resolvconf' name, in
4380 which case it will take arguments and input compatible with the
4381 Debian and FreeBSD resolvconf tool.
4383 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
4384 where the system initially suspends, and after a timeout resumes and
4387 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
4388 set the client will only send a DUID as client identifier.
4390 * The nss-systemd glibc NSS module will now enumerate dynamic users and
4391 groups in effect. Previously, it could resolve UIDs/GIDs to user
4392 names/groups and vice versa, but did not support enumeration.
4394 * journald's Compress= configuration setting now optionally accepts a
4395 byte threshold value. All journal objects larger than this threshold
4396 will be compressed, smaller ones will not. Previously this threshold
4397 was not configurable and set to 512.
4399 * A new system.conf setting NoNewPrivileges= is now available which may
4400 be used to turn off acquisition of new privileges system-wide
4401 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
4402 for all its children). Note that turning this option on means setuid
4403 binaries and file system capabilities lose their special powers.
4404 While turning on this option is a big step towards a more secure
4405 system, doing so is likely to break numerous pre-existing UNIX tools,
4406 in particular su and sudo.
4408 * A new service systemd-time-sync-wait.service has been added. If
4409 enabled it will delay the time-sync.target unit at boot until time
4410 synchronization has been received from the network. This
4411 functionality is useful on systems lacking a local RTC or where it is
4412 acceptable that the boot process shall be delayed by external network
4415 * When hibernating, systemd will now inform the kernel of the image
4416 write offset, on kernels new enough to support this. This means swap
4417 files should work for hibernation now.
4419 * When loading unit files, systemd will now look for drop-in unit files
4420 extensions in additional places. Previously, for a unit file name
4421 "foo-bar-baz.service" it would look for dropin files in
4422 "foo-bar-baz.service.d/*.conf". Now, it will also look in
4423 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
4424 service name truncated after all inner dashes. This scheme allows
4425 writing drop-ins easily that apply to a whole set of unit files at
4426 once. It's particularly useful for mount and slice units (as their
4427 naming is prefix based), but is also useful for service and other
4428 units, for packages that install multiple unit files at once,
4429 following a strict naming regime of beginning the unit file name with
4430 the package's name. Two new specifiers are now supported in unit
4431 files to match this: %j and %J are replaced by the part of the unit
4432 name following the last dash.
4434 * Unit files and other configuration files that support specifier
4435 expansion now understand another three new specifiers: %T and %V will
4436 resolve to /tmp and /var/tmp respectively, or whatever temporary
4437 directory has been set for the calling user. %E will expand to either
4438 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
4440 * The ExecStart= lines of unit files are no longer required to
4441 reference absolute paths. If non-absolute paths are specified the
4442 specified binary name is searched within the service manager's
4443 built-in $PATH, which may be queried with 'systemd-path
4444 search-binaries-default'. It's generally recommended to continue to
4445 use absolute paths for all binaries specified in unit files.
4447 * Units gained a new load state "bad-setting", which is used when a
4448 unit file was loaded, but contained fatal errors which prevent it
4449 from being started (for example, a service unit has been defined
4450 lacking both ExecStart= and ExecStop= lines).
4452 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
4453 support alternative debuggers, for example lldb. The old name
4454 continues to be available however, for compatibility reasons. Use the
4455 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
4456 to pick an alternative debugger instead of the default gdb.
4458 * systemctl and the other tools will now output escape sequences that
4459 generate proper clickable hyperlinks in various terminal emulators
4460 where useful (for example, in the "systemctl status" output you can
4461 now click on the unit file name to quickly open it in the
4462 editor/viewer of your choice). Note that not all terminal emulators
4463 support this functionality yet, but many do. Unfortunately, the
4464 "less" pager doesn't support this yet, hence this functionality is
4465 currently automatically turned off when a pager is started (which
4466 happens quite often due to auto-paging). We hope to remove this
4467 limitation as soon as "less" learns these escape sequences. This new
4468 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
4469 environment variable. For details on these escape sequences see:
4470 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
4472 * networkd's .network files now support a new IPv6MTUBytes= option for
4473 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
4474 option in the [Route] section to configure the MTU to use for
4475 specific routes. It also gained support for configuration of the DHCP
4476 "UserClass" option through the new UserClass= setting. It gained
4477 three new options in the new [CAN] section for configuring CAN
4478 networks. The MULTICAST and ALLMULTI interface flags may now be
4479 controlled explicitly with the new Multicast= and AllMulticast=
4482 * networkd will now automatically make use of the kernel's route
4483 expiration feature, if it is available.
4485 * udevd's .link files now support setting the number of receive and
4486 transmit channels, using the RxChannels=, TxChannels=,
4487 OtherChannels=, CombinedChannels= settings.
4489 * Support for UDPSegmentationOffload= has been removed, given its
4490 limited support in hardware, and waning software support.
4492 * networkd's .netdev files now support creating "netdevsim" interfaces.
4494 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
4495 to query the unit belonging to a specific kernel control group.
4497 * systemd-analyze gained a new verb "cat-config", which may be used to
4498 dump the contents of any configuration file, with all its matching
4499 drop-in files added in, and honouring the usual search and masking
4500 logic applied to systemd configuration files. For example use
4501 "systemd-analyze cat-config systemd/system.conf" to get the complete
4502 system configuration file of systemd how it would be loaded by PID 1
4503 itself. Similar to this, various tools such as systemd-tmpfiles or
4504 systemd-sysusers, gained a new option "--cat-config", which does the
4505 corresponding operation for their own configuration settings. For
4506 example, "systemd-tmpfiles --cat-config" will now output the full
4507 list of tmpfiles.d/ lines in place.
4509 * timedatectl gained three new verbs: "show" shows bus properties of
4510 systemd-timedated, "timesync-status" shows the current NTP
4511 synchronization state of systemd-timesyncd, and "show-timesync"
4512 shows bus properties of systemd-timesyncd.
4514 * systemd-timesyncd gained a bus interface on which it exposes details
4517 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
4518 understood by systemd-timedated. It takes a colon-separated list of
4519 unit names of NTP client services. The list is used by
4520 "timedatectl set-ntp".
4522 * systemd-nspawn gained a new --rlimit= switch for setting initial
4523 resource limits for the container payload. There's a new switch
4524 --hostname= to explicitly override the container's hostname. A new
4525 --no-new-privileges= switch may be used to control the
4526 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
4527 --oom-score-adjust= switch controls the OOM scoring adjustment value
4528 for the payload. The new --cpu-affinity= switch controls the CPU
4529 affinity of the container payload. The new --resolv-conf= switch
4530 allows more detailed control of /etc/resolv.conf handling of the
4531 container. Similarly, the new --timezone= switch allows more detailed
4532 control of /etc/localtime handling of the container.
4534 * systemd-detect-virt gained a new --list switch, which will print a
4535 list of all currently known VM and container environments.
4537 * Support for "Portable Services" has been added, see
4538 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
4539 experimental, but this is expected to change soon. Reflecting this
4540 experimental state, the "portablectl" binary is not installed into
4541 /usr/bin yet. The binary has to be called with the full path
4542 /usr/lib/systemd/portablectl instead.
4544 * journalctl's and systemctl's -o switch now knows a new log output
4545 mode "with-unit". The output it generates is very similar to the
4546 regular "short" mode, but displays the unit name instead of the
4547 syslog tag for each log line. Also, the date is shown with timezone
4548 information. This mode is probably more useful than the classic
4549 "short" output mode for most purposes, except where pixel-perfect
4550 compatibility with classic /var/log/messages formatting is required.
4552 * A new --dump-bus-properties switch has been added to the systemd
4553 binary, which may be used to dump all supported D-Bus properties.
4554 (Options which are still supported, but are deprecated, are *not*
4557 * sd-bus gained a set of new calls:
4558 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
4559 enable/disable the "floating" state of a bus slot object,
4560 i.e. whether the slot object pins the bus it is allocated for into
4561 memory or if the bus slot object gets disconnected when the bus goes
4562 away. sd_bus_open_with_description(),
4563 sd_bus_open_user_with_description(),
4564 sd_bus_open_system_with_description() may be used to allocate bus
4565 objects and set their description string already during allocation.
4567 * sd-event gained support for watching inotify events from the event
4568 loop, in an efficient way, sharing inotify handles between multiple
4569 users. For this a new function sd_event_add_inotify() has been added.
4571 * sd-event and sd-bus gained support for calling special user-supplied
4572 destructor functions for userdata pointers associated with
4573 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
4574 functions sd_bus_slot_set_destroy_callback,
4575 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
4576 sd_bus_track_get_destroy_callback,
4577 sd_event_source_set_destroy_callback,
4578 sd_event_source_get_destroy_callback have been added.
4580 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
4582 * PID 1 will now automatically reschedule .timer units whenever the
4583 local timezone changes. (They previously got rescheduled
4584 automatically when the system clock changed.)
4586 * New documentation has been added to document cgroups delegation,
4587 portable services and the various code quality tools we have set up:
4589 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
4590 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
4591 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
4593 * The Boot Loader Specification has been added to the source tree.
4595 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
4597 While moving it into our source tree we have updated it and further
4598 changes are now accepted through the usual github PR workflow.
4600 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
4601 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
4602 earlier PAM modules. The data in these fields is used to initialize
4603 the session scope's resource properties. Thus external PAM modules
4604 may now configure per-session limits, for example sourced from
4605 external user databases.
4607 * socket units with Accept=yes will now maintain a "refused" counter in
4608 addition to the existing "accepted" counter, counting connections
4609 refused due to the enforced limits.
4611 * The "systemd-path search-binaries-default" command may now be use to
4612 query the default, built-in $PATH PID 1 will pass to the services it
4615 * A new unit file setting PrivateMounts= has been added. It's a boolean
4616 option. If enabled the unit's processes are invoked in their own file
4617 system namespace. Note that this behaviour is also implied if any
4618 other file system namespacing options (such as PrivateTmp=,
4619 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
4620 primarily useful for services that do not use any of the other file
4621 system namespacing options. One such service is systemd-udevd.service
4622 where this is now used by default.
4624 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
4625 when the system is booted in UEFI "secure mode".
4627 * A new unit "system-update-pre.target" is added, which defines an
4628 optional synchronization point for offline system updates, as
4629 implemented by the pre-existing "system-update.target" unit. It
4630 allows ordering services before the service that executes the actual
4631 update process in a generic way.
4633 * Systemd now emits warnings whenever .include syntax is used.
4635 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
4636 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
4637 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
4638 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
4639 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
4640 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
4641 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
4642 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
4643 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
4644 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
4645 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
4646 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
4647 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
4648 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
4649 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
4650 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
4651 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
4652 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
4653 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
4654 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
4655 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
4656 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
4657 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
4658 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
4659 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
4660 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
4661 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
4662 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
4663 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
4665 — Berlin, 2018-06-22
4669 * The MemoryAccounting= unit property now defaults to on. After
4670 discussions with the upstream control group maintainers we learnt
4671 that the negative impact of cgroup memory accounting on current
4672 kernels is finally relatively minimal, so that it should be safe to
4673 enable this by default without affecting system performance. Besides
4674 memory accounting only task accounting is turned on by default, all
4675 other forms of resource accounting (CPU, IO, IP) remain off for now,
4676 because it's not clear yet that their impact is small enough to move
4677 from opt-in to opt-out. We recommend downstreams to leave memory
4678 accounting on by default if kernel 4.14 or higher is primarily
4679 used. On very resource constrained systems or when support for old
4680 kernels is a necessity, -Dmemory-accounting-default=false can be used
4681 to revert this change.
4683 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
4684 %udev_rules_update) and the journal catalog (%journal_catalog_update)
4685 from the upgrade scriptlets of individual packages now do nothing.
4686 Transfiletriggers have been added which will perform those updates
4687 once at the end of the transaction.
4689 Similar transfiletriggers have been added to execute any sysctl.d
4690 and binfmt.d rules. Thus, it should be unnecessary to provide any
4691 scriptlets to execute this configuration from package installation
4694 * systemd-sysusers gained a mode where the configuration to execute is
4695 specified on the command line, but this configuration is not executed
4696 directly, but instead it is merged with the configuration on disk,
4697 and the result is executed. This is useful for package installation
4698 scripts which want to create the user before installing any files on
4699 disk (in case some of those files are owned by that user), while
4700 still allowing local admin overrides.
4702 This functionality is exposed to rpm scriptlets through a new
4703 %sysusers_create_package macro. Old %sysusers_create and
4704 %sysusers_create_inline macros are deprecated.
4706 A transfiletrigger for sysusers.d configuration is now installed,
4707 which means that it should be unnecessary to call systemd-sysusers from
4708 package installation scripts, unless the package installs any files
4709 owned by those newly-created users, in which case
4710 %sysusers_create_package should be used.
4712 * Analogous change has been done for systemd-tmpfiles: it gained a mode
4713 where the command-line configuration is merged with the configuration
4714 on disk. This is exposed as the new %tmpfiles_create_package macro,
4715 and %tmpfiles_create is deprecated. A transfiletrigger is installed
4716 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
4717 from package installation scripts.
4719 * sysusers.d configuration for a user may now also specify the group
4720 number, in addition to the user number ("u username 123:456"), or
4721 without the user number ("u username -:456").
4723 * Configution items for systemd-sysusers can now be specified as
4724 positional arguments when the new --inline switch is used.
4726 * The login shell of users created through sysusers.d may now be
4727 specified (previously, it was always /bin/sh for root and
4728 /sbin/nologin for other users).
4730 * systemd-analyze gained a new --global switch to look at global user
4731 configuration. It also gained a unit-paths verb to list the unit load
4732 paths that are compiled into systemd (which can be used with
4733 --systemd, --user, or --global).
4735 * udevadm trigger gained a new --settle/-w option to wait for any
4736 triggered events to finish (but just those, and not any other events
4737 which are triggered meanwhile).
4739 * The action that systemd-logind takes when the lid is closed and the
4740 machine is connected to external power can now be configured using
4741 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
4742 was determined by HandleLidSwitch=, and, for backwards compatibility,
4743 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
4745 * journalctl will periodically call sd_journal_process() to make it
4746 resilient against inotify queue overruns when journal files are
4747 rotated very quickly.
4749 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
4750 sd_bus_get_n_queued_write — may be used to check the number of
4751 pending bus messages.
4753 * systemd gained a new
4754 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
4755 which can be used to migrate foreign processes to scope and service
4756 units. The primary user for this new API is systemd itself: the
4757 systemd --user instance uses this call of the systemd --system
4758 instance to migrate processes if it itself gets the request to
4759 migrate processes and the kernel refuses this due to access
4760 restrictions. Thanks to this "systemd-run --scope --user …" works
4761 again in pure cgroup v2 environments when invoked from the user
4764 * A new TemporaryFileSystem= setting can be used to mask out part of
4765 the real file system tree with tmpfs mounts. This may be combined
4766 with BindPaths= and BindReadOnlyPaths= to hide files or directories
4767 not relevant to the unit, while still allowing some paths lower in
4768 the tree to be accessed.
4770 ProtectHome=tmpfs may now be used to hide user home and runtime
4771 directories from units, in a way that is mostly equivalent to
4772 "TemporaryFileSystem=/home /run/user /root".
4774 * Non-service units are now started with KeyringMode=shared by default.
4775 This means that mount and swapon and other mount tools have access
4776 to keys in the main keyring.
4778 * /sys/fs/bpf is now mounted automatically.
4780 * QNX virtualization is now detected by systemd-detect-virt and may
4781 be used in ConditionVirtualization=.
4783 * IPAccounting= may now be enabled also for slice units.
4785 * A new -Dsplit-bin= build configuration switch may be used to specify
4786 whether bin and sbin directories are merged, or if they should be
4787 included separately in $PATH and various listings of executable
4788 directories. The build configuration scripts will try to autodetect
4789 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
4790 system, but distributions are encouraged to configure this
4793 * A new -Dok-color= build configuration switch may be used to change
4794 the colour of "OK" status messages.
4796 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
4797 PrivateNetwork=yes was buggy in previous versions of systemd. This
4798 means that after the upgrade and daemon-reexec, any such units must
4801 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
4802 will not exclude read-only files owned by root from cleanup.
4804 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
4805 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
4806 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
4807 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
4808 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
4809 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
4810 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
4811 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
4812 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
4813 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
4814 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
4815 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
4816 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
4817 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
4818 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
4819 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
4821 — Warsaw, 2018-03-05
4825 * Some keyboards come with a zoom see-saw or rocker which until now got
4826 mapped to the Linux "zoomin/out" keys in hwdb. However, these
4827 keycodes are not recognized by any major desktop. They now produce
4828 Up/Down key events so that they can be used for scrolling.
4830 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
4831 slightly: previously, if an argument was specified for lines of this
4832 type (i.e. the right-most column was set) this string was appended to
4833 existing files each time systemd-tmpfiles was run. This behaviour was
4834 different from what the documentation said, and not particularly
4835 useful, as repeated systemd-tmpfiles invocations would not be
4836 idempotent and grow such files without bounds. With this release
4837 behaviour has been altered to match what the documentation says:
4838 lines of this type only have an effect if the indicated files don't
4839 exist yet, and only then the argument string is written to the file.
4841 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
4842 systemd-tmpfiles behaviour: previously, read-only files owned by root
4843 were always excluded from the file "aging" algorithm (i.e. the
4844 automatic clean-up of directories like /tmp based on
4845 atime/mtime/ctime). We intend to drop this restriction, and age files
4846 by default even when owned by root and read-only. This behaviour was
4847 inherited from older tools, but there have been requests to remove
4848 it, and it's not obvious why this restriction was made in the first
4849 place. Please speak up now, if you are aware of software that requires
4850 this behaviour, otherwise we'll remove the restriction in v238.
4852 * A new environment variable $SYSTEMD_OFFLINE is now understood by
4853 systemctl. It takes a boolean argument. If on, systemctl assumes it
4854 operates on an "offline" OS tree, and will not attempt to talk to the
4855 service manager. Previously, this mode was implicitly enabled if a
4856 chroot() environment was detected, and this new environment variable
4857 now provides explicit control.
4859 * .path and .socket units may now be created transiently, too.
4860 Previously only service, mount, automount and timer units were
4861 supported as transient units. The systemd-run tool has been updated
4862 to expose this new functionality, you may hence use it now to bind
4863 arbitrary commands to path or socket activation on-the-fly from the
4864 command line. Moreover, almost all properties are now exposed for the
4865 unit types that already supported transient operation.
4867 * The systemd-mount command gained support for a new --owner= parameter
4868 which takes a user name, which is then resolved and included in uid=
4869 and gid= mount options string of the file system to mount.
4871 * A new unit condition ConditionControlGroupController= has been added
4872 that checks whether a specific cgroup controller is available.
4874 * Unit files, udev's .link files, and systemd-networkd's .netdev and
4875 .network files all gained support for a new condition
4876 ConditionKernelVersion= for checking against specific kernel
4879 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
4880 support for configuring device flags in the Flags= setting. In the
4881 same files, the [Tunnel] section gained support for configuring
4882 AllowLocalRemote=. The [Route] section in .network files gained
4883 support for configuring InitialCongestionWindow=,
4884 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
4885 understands RapidCommit=.
4887 * systemd-networkd's DHCPv6 support gained support for Prefix
4890 * sd-bus gained support for a new "watch-bind" feature. When this
4891 feature is enabled, an sd_bus connection may be set up to connect to
4892 an AF_UNIX socket in the file system as soon as it is created. This
4893 functionality is useful for writing early-boot services that
4894 automatically connect to the system bus as soon as it is started,
4895 without ugly time-based polling. systemd-networkd and
4896 systemd-resolved have been updated to make use of this
4897 functionality. busctl exposes this functionality in a new
4898 --watch-bind= command line switch.
4900 * sd-bus will now optionally synthesize a local "Connected" signal as
4901 soon as a D-Bus connection is set up fully. This message mirrors the
4902 already existing "Disconnected" signal which is synthesized when the
4903 connection is terminated. This signal is generally useful but
4904 particularly handy in combination with the "watch-bind" feature
4905 described above. Synthesizing of this message has to be requested
4906 explicitly through the new API call sd_bus_set_connected_signal(). In
4907 addition a new call sd_bus_is_ready() has been added that checks
4908 whether a connection is fully set up (i.e. between the "Connected" and
4909 "Disconnected" signals).
4911 * sd-bus gained two new calls sd_bus_request_name_async() and
4912 sd_bus_release_name_async() for asynchronously registering bus
4913 names. Similar, there is now sd_bus_add_match_async() for installing
4914 a signal match asynchronously. All of systemd's own services have
4915 been updated to make use of these calls. Doing these operations
4916 asynchronously has two benefits: it reduces the risk of deadlocks in
4917 case of cyclic dependencies between bus services, and it speeds up
4918 service initialization since synchronization points for bus
4919 round-trips are removed.
4921 * sd-bus gained two new calls sd_bus_match_signal() and
4922 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
4923 and sd_bus_add_match_async() but instead of taking a D-Bus match
4924 string take match fields as normal function parameters.
4926 * sd-bus gained two new calls sd_bus_set_sender() and
4927 sd_bus_message_set_sender() for setting the sender name of outgoing
4928 messages (either for all outgoing messages or for just one specific
4929 one). These calls are only useful in direct connections as on
4930 brokered connections the broker fills in the sender anyway,
4931 overwriting whatever the client filled in.
4933 * sd-event gained a new pseudo-handle that may be specified on all API
4934 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
4935 used this refers to the default event loop object of the calling
4936 thread. Note however that this does not implicitly allocate one —
4937 which has to be done prior by using sd_event_default(). Similarly
4938 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
4939 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
4940 to the default bus of the specified type of the calling thread. Here
4941 too this does not implicitly allocate bus connection objects, this
4942 has to be done prior with sd_bus_default() and friends.
4944 * sd-event gained a new call pair
4945 sd_event_source_{get|set}_io_fd_own(). This may be used to request
4946 automatic closure of the file descriptor an IO event source watches
4947 when the event source is destroyed.
4949 * systemd-networkd gained support for natively configuring WireGuard
4952 * In previous versions systemd synthesized user records both for the
4953 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
4954 internally. In order to simplify distribution-wide renames of the
4955 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
4956 new transitional flag file has been added: if
4957 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
4958 user and group record within the systemd codebase is disabled.
4960 * systemd-notify gained a new --uid= option for selecting the source
4961 user/UID to use for notification messages sent to the service
4964 * journalctl gained a new --grep= option to list only entries in which
4965 the message matches a certain pattern. By default matching is case
4966 insensitive if the pattern is lowercase, and case sensitive
4967 otherwise. Option --case-sensitive=yes|no can be used to override
4968 this an specify case sensitivity or case insensitivity.
4970 * There's now a "systemd-analyze service-watchdogs" command for printing
4971 the current state of the service runtime watchdog, and optionally
4972 enabling or disabling the per-service watchdogs system-wide if given a
4973 boolean argument (i.e. the concept you configure in WatchdogSec=), for
4974 debugging purposes. There's also a kernel command line option
4975 systemd.service_watchdogs= for controlling the same.
4977 * Two new "log-level" and "log-target" options for systemd-analyze were
4978 added that merge the now deprecated get-log-level, set-log-level and
4979 get-log-target, set-log-target pairs. The deprecated options are still
4980 understood for backwards compatibility. The two new options print the
4981 current value when no arguments are given, and set them when a
4982 level/target is given as an argument.
4984 * sysusers.d's "u" lines now optionally accept both a UID and a GID
4985 specification, separated by a ":" character, in order to create users
4986 where UID and GID do not match.
4988 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
4989 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
4990 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
4991 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
4992 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
4993 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
4994 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
4995 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
4996 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
4997 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
4998 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
4999 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
5000 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
5001 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
5002 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
5003 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
5004 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
5005 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
5006 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
5007 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
5014 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
5015 in v235 has been extended to also set the dummy.ko module option
5016 numdummies=0, preventing the kernel from automatically creating
5017 dummy0. All dummy interfaces must now be explicitly created.
5019 * Unknown '%' specifiers in configuration files are now rejected. This
5020 applies to units and tmpfiles.d configuration. Any percent characters
5021 that are followed by a letter or digit that are not supposed to be
5022 interpreted as the beginning of a specifier should be escaped by
5023 doubling ("%%"). (So "size=5%" is still accepted, as well as
5024 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
5025 valid specifiers today.)
5027 * systemd-resolved now maintains a new dynamic
5028 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
5029 recommended to make /etc/resolv.conf a symlink to it. This file
5030 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
5031 includes dynamically acquired search domains, achieving more correct
5032 DNS resolution by software that bypasses local DNS APIs such as NSS.
5034 * The "uaccess" udev tag has been dropped from /dev/kvm and
5035 /dev/dri/renderD*. These devices now have the 0666 permissions by
5036 default (but this may be changed at build-time). /dev/dri/renderD*
5037 will now be owned by the "render" group along with /dev/kfd.
5039 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
5040 systemd-journal-gatewayd.service and
5041 systemd-journal-upload.service. This means "nss-systemd" must be
5042 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
5043 services are resolved properly.
5045 * In /etc/fstab two new mount options are now understood:
5046 x-systemd.makefs and x-systemd.growfs. The former has the effect that
5047 the configured file system is formatted before it is mounted, the
5048 latter that the file system is resized to the full block device size
5049 after it is mounted (i.e. if the file system is smaller than the
5050 partition it resides on, it's grown). This is similar to the fsck
5051 logic in /etc/fstab, and pulls in systemd-makefs@.service and
5052 systemd-growfs@.service as necessary, similar to
5053 systemd-fsck@.service. Resizing is currently only supported on ext4
5056 * In systemd-networkd, the IPv6 RA logic now optionally may announce
5057 DNS server and domain information.
5059 * Support for the LUKS2 on-disk format for encrypted partitions has
5060 been added. This requires libcryptsetup2 during compilation and
5063 * The systemd --user instance will now signal "readiness" when its
5064 basic.target unit has been reached, instead of when the run queue ran
5065 empty for the first time.
5067 * Tmpfiles.d with user configuration are now also supported.
5068 systemd-tmpfiles gained a new --user switch, and snippets placed in
5069 ~/.config/user-tmpfiles.d/ and corresponding directories will be
5070 executed by systemd-tmpfiles --user running in the new
5071 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
5072 running in the user session.
5074 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
5075 %S resolves to the top-level state directory (/var/lib for the system
5076 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
5077 top-level cache directory (/var/cache for the system instance,
5078 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
5079 logs directory (/var/log for the system instance,
5080 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
5081 existing %t specifier, that resolves to the top-level runtime
5082 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
5085 * journalctl learnt a new parameter --output-fields= for limiting the
5086 set of journal fields to output in verbose and JSON output modes.
5088 * systemd-timesyncd's configuration file gained a new option
5089 RootDistanceMaxSec= for setting the maximum root distance of servers
5090 it'll use, as well as the new options PollIntervalMinSec= and
5091 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
5093 * bootctl gained a new command "list" for listing all available boot
5094 menu items on systems that follow the boot loader specification.
5096 * systemctl gained a new --dry-run switch that shows what would be done
5097 instead of doing it, and is currently supported by the shutdown and
5100 * ConditionSecurity= can now detect the TOMOYO security module.
5102 * Unit file [Install] sections are now also respected in unit drop-in
5103 files. This is intended to be used by drop-ins under /usr/lib/.
5105 * systemd-firstboot may now also set the initial keyboard mapping.
5107 * Udev "changed" events for devices which are exposed as systemd
5108 .device units are now propagated to units specified in
5109 ReloadPropagatedFrom= as reload requests.
5111 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
5112 unit template name (i.e. a name in the form of 'foobar@.service',
5113 without the instance component between the '@' and - the '.'), then
5114 the escaped sysfs path of the device is automatically used as the
5117 * SystemCallFilter= in unit files has been extended so that an "errno"
5118 can be specified individually for each system call. Example:
5119 SystemCallFilter=~uname:EILSEQ.
5121 * The cgroup delegation logic has been substantially updated. Delegate=
5122 now optionally takes a list of controllers (instead of a boolean, as
5123 before), which lists the controllers to delegate at least.
5125 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
5127 * A new LogLevelMax= setting configures the maximum log level any
5128 process of the service may log at (i.e. anything with a lesser
5129 priority than what is specified is automatically dropped). A new
5130 LogExtraFields= setting allows configuration of additional journal
5131 fields to attach to all log records generated by any of the unit's
5134 * New StandardInputData= and StandardInputText= settings along with the
5135 new option StandardInput=data may be used to configure textual or
5136 binary data that shall be passed to the executed service process via
5137 standard input, encoded in-line in the unit file.
5139 * StandardInput=, StandardOutput= and StandardError= may now be used to
5140 connect stdin/stdout/stderr of executed processes directly with a
5141 file or AF_UNIX socket in the file system, using the new "file:" option.
5143 * A new unit file option CollectMode= has been added, that allows
5144 tweaking the garbage collection logic for units. It may be used to
5145 tell systemd to garbage collect units that have failed automatically
5146 (normally it only GCs units that exited successfully). systemd-run
5147 and systemd-mount expose this new functionality with a new -G option.
5149 * "machinectl bind" may now be used to bind mount non-directories
5150 (i.e. regularfiles, devices, fifos, sockets).
5152 * systemd-analyze gained a new verb "calendar" for validating and
5153 testing calendar time specifications to use for OnCalendar= in timer
5154 units. Besides validating the expression it will calculate the next
5155 time the specified expression would elapse.
5157 * In addition to the pre-existing FailureAction= unit file setting
5158 there's now SuccessAction=, for configuring a shutdown action to
5159 execute when a unit completes successfully. This is useful in
5160 particular inside containers that shall terminate after some workload
5161 has been completed. Also, both options are now supported for all unit
5162 types, not just services.
5164 * networkds's IP rule support gained two new options
5165 IncomingInterface= and OutgoingInterface= for configuring the incoming
5166 and outgoing interfaces of configured rules. systemd-networkd also
5167 gained support for "vxcan" network devices.
5169 * networkd gained a new setting RequiredForOnline=, taking a
5170 boolean. If set, systemd-wait-online will take it into consideration
5171 when determining that the system is up, otherwise it will ignore the
5172 interface for this purpose.
5174 * The sd_notify() protocol gained support for a new operation: with
5175 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
5176 store again, ahead of POLLHUP or POLLERR when they are removed
5179 * A new document doc/UIDS-GIDS.md has been added to the source tree,
5180 that documents the UID/GID range and assignment assumptions and
5181 requirements of systemd.
5183 * The watchdog device PID 1 will ping may now be configured through the
5184 WatchdogDevice= configuration file setting, or by setting the
5185 systemd.watchdog_service= kernel commandline option.
5187 * systemd-resolved's gained support for registering DNS-SD services on
5188 the local network using MulticastDNS. Services may either be
5189 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
5190 the same dir below /run, /usr/lib), or through its D-Bus API.
5192 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
5193 extend the effective start, runtime, and stop time. The service must
5194 continue to send EXTEND_TIMEOUT_USEC within the period specified to
5195 prevent the service manager from making the service as timedout.
5197 * systemd-resolved's DNSSEC support gained support for RFC 8080
5198 (Ed25519 keys and signatures).
5200 * The systemd-resolve command line tool gained a new set of options
5201 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
5202 --set-nta= and --revert to configure per-interface DNS configuration
5203 dynamically during runtime. It's useful for pushing DNS information
5204 into systemd-resolved from DNS hook scripts that various interface
5205 managing software supports (such as pppd).
5207 * systemd-nspawn gained a new --network-namespace-path= command line
5208 option, which may be used to make a container join an existing
5209 network namespace, by specifying a path to a "netns" file.
5211 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
5212 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
5213 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
5214 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
5215 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
5216 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
5217 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
5218 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
5219 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
5220 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
5221 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
5222 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
5223 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
5224 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
5225 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
5226 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
5227 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
5228 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
5229 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
5230 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
5231 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
5232 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
5233 Jędrzejewski-Szmek, Zeal Jagannatha
5235 — Berlin, 2017-12-14
5239 * INCOMPATIBILITY: systemd-logind.service and other long-running
5240 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
5241 communication with the outside. This generally improves security of
5242 the system, and is in almost all cases a safe and good choice, as
5243 these services do not and should not provide any network-facing
5244 functionality. However, systemd-logind uses the glibc NSS API to
5245 query the user database. This creates problems on systems where NSS
5246 is set up to directly consult network services for user database
5247 lookups. In particular, this creates incompatibilities with the
5248 "nss-nis" module, which attempts to directly contact the NIS/YP
5249 network servers it is configured for, and will now consistently
5250 fail. In such cases, it is possible to turn off IP sandboxing for
5251 systemd-logind.service (set IPAddressDeny= in its [Service] section
5252 to the empty string, via a .d/ unit file drop-in). Downstream
5253 distributions might want to update their nss-nis packaging to include
5254 such a drop-in snippet, accordingly, to hide this incompatibility
5255 from the user. Another option is to make use of glibc's nscd service
5256 to proxy such network requests through a privilege-separated, minimal
5257 local caching daemon, or to switch to more modern technologies such
5258 sssd, whose NSS hook-ups generally do not involve direct network
5259 access. In general, we think it's definitely time to question the
5260 implementation choices of nss-nis, i.e. whether it's a good idea
5261 today to embed a network-facing loadable module into all local
5262 processes that need to query the user database, including the most
5263 trivial and benign ones, such as "ls". For more details about
5264 IPAddressDeny= see below.
5266 * A new modprobe.d drop-in is now shipped by default that sets the
5267 bonding module option max_bonds=0. This overrides the kernel default,
5268 to avoid conflicts and ambiguity as to whether or not bond0 should be
5269 managed by systemd-networkd or not. This resolves multiple issues
5270 with bond0 properties not being applied, when bond0 is configured
5271 with systemd-networkd. Distributors may choose to not package this,
5272 however in that case users will be prevented from correctly managing
5273 bond0 interface using systemd-networkd.
5275 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
5276 which print the logging level and target of the system manager. They
5277 complement the existing "set-log-level" and "set-log-target" verbs
5278 used to change those values.
5280 * journald.conf gained a new boolean setting ReadKMsg= which defaults
5281 to on. If turned off kernel log messages will not be read by
5282 systemd-journald or included in the logs. It also gained a new
5283 setting LineMax= for configuring the maximum line length in
5284 STDOUT/STDERR log streams. The new default for this value is 48K, up
5285 from the previous hardcoded 2048.
5287 * A new unit setting RuntimeDirectoryPreserve= has been added, which
5288 allows more detailed control of what to do with a runtime directory
5289 configured with RuntimeDirectory= (i.e. a directory below /run or
5290 $XDG_RUNTIME_DIR) after a unit is stopped.
5292 * The RuntimeDirectory= setting for units gained support for creating
5293 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
5294 one top-level directory.
5296 * Units gained new options StateDirectory=, CacheDirectory=,
5297 LogsDirectory= and ConfigurationDirectory= which are closely related
5298 to RuntimeDirectory= but manage per-service directories below
5299 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
5300 possible to write unit files which when activated automatically gain
5301 properly owned service specific directories in these locations, thus
5302 making unit files self-contained and increasing compatibility with
5303 stateless systems and factory reset where /etc or /var are
5304 unpopulated at boot. Matching these new settings there's also
5305 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
5306 ConfigurationDirectoryMode= for configuring the access mode of these
5307 directories. These settings are particularly useful in combination
5308 with DynamicUser=yes as they provide secure, properly-owned,
5309 writable, and stateful locations for storage, excluded from the
5310 sandbox that such services live in otherwise.
5312 * Automake support has been removed from this release. systemd is now
5315 * systemd-journald will now aggressively cache client metadata during
5316 runtime, speeding up log write performance under pressure. This comes
5317 at a small price though: as much of the metadata is read
5318 asynchronously from /proc/ (and isn't implicitly attached to log
5319 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
5320 metadata stored alongside a log entry might be slightly
5321 out-of-date. Previously it could only be slightly newer than the log
5322 message. The time window is small however, and given that the kernel
5323 is unlikely to be improved anytime soon in this regard, this appears
5326 * nss-myhostname/systemd-resolved will now by default synthesize an
5327 A/AAAA resource record for the "_gateway" hostname, pointing to the
5328 current default IP gateway. Previously it did that for the "gateway"
5329 name, hampering adoption, as some distributions wanted to leave that
5330 hostname open for local use. The old behaviour may still be
5331 requested at build time.
5333 * systemd-networkd's [Address] section in .network files gained a new
5334 Scope= setting for configuring the IP address scope. The [Network]
5335 section gained a new boolean setting ConfigureWithoutCarrier= that
5336 tells systemd-networkd to ignore link sensing when configuring the
5337 device. The [DHCP] section gained a new Anonymize= boolean option for
5338 turning on a number of options suggested in RFC 7844. A new
5339 [RoutingPolicyRule] section has been added for configuring the IP
5340 routing policy. The [Route] section has gained support for a new
5341 Type= setting which permits configuring
5342 blackhole/unreachable/prohibit routes.
5344 * The [VRF] section in .netdev files gained a new Table= setting for
5345 configuring the routing table to use. The [Tunnel] section gained a
5346 new Independent= boolean field for configuring tunnels independent of
5347 an underlying network interface. The [Bridge] section gained a new
5348 GroupForwardMask= option for configuration of propagation of link
5349 local frames between bridge ports.
5351 * The WakeOnLan= setting in .link files gained support for a number of
5352 new modes. A new TCP6SegmentationOffload= setting has been added for
5353 configuring TCP/IPv6 hardware segmentation offload.
5355 * The IPv6 RA sender implementation may now optionally send out RDNSS
5356 and RDNSSL records to supply DNS configuration to peers.
5358 * systemd-nspawn gained support for a new --system-call-filter= command
5359 line option for adding and removing entries in the default system
5360 call filter it applies. Moreover systemd-nspawn has been changed to
5361 implement a system call allow list instead of a deny list.
5363 * systemd-run gained support for a new --pipe command line option. If
5364 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
5365 are directly passed on to the activated transient service
5366 executable. This allows invoking arbitrary processes as systemd
5367 services (for example to take benefit of dependency management,
5368 accounting management, resource management or log management that is
5369 done automatically for services) — while still allowing them to be
5370 integrated in a classic UNIX shell pipeline.
5372 * When a service sends RELOAD=1 via sd_notify() and reload propagation
5373 using ReloadPropagationTo= is configured, a reload is now propagated
5374 to configured units. (Previously this was only done on explicitly
5375 requested reloads, using "systemctl reload" or an equivalent
5378 * For each service unit a restart counter is now kept: it is increased
5379 each time the service is restarted due to Restart=, and may be
5380 queried using "systemctl show -p NRestarts …".
5382 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
5383 @signal and @timer have been added, for usage with SystemCallFilter=
5384 in unit files and the new --system-call-filter= command line option
5385 of systemd-nspawn (see above).
5387 * ExecStart= lines in unit files gained two new modifiers: when a
5388 command line is prefixed with "!" the command will be executed as
5389 configured, except for the credentials applied by
5390 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
5391 "+", but does still apply namespacing options unlike "+". There's
5392 also "!!" now, which is mostly identical, but becomes a NOP on
5393 systems that support ambient capabilities. This is useful to write
5394 unit files that work with ambient capabilities where possible but
5395 automatically fall back to traditional privilege dropping mechanisms
5396 on systems where this is not supported.
5398 * ListenNetlink= settings in socket units now support RDMA netlink
5401 * A new unit file setting LockPersonality= has been added which permits
5402 locking down the chosen execution domain ("personality") of a service
5405 * A new special target "getty-pre.target" has been added, which is
5406 ordered before all text logins, and may be used to order services
5407 before textual logins acquire access to the console.
5409 * systemd will now attempt to load the virtio-rng.ko kernel module very
5410 early on if a VM environment supporting this is detected. This should
5411 improve entropy during early boot in virtualized environments.
5413 * A _netdev option is now supported in /etc/crypttab that operates in a
5414 similar way as the same option in /etc/fstab: it permits configuring
5415 encrypted devices that need to be ordered after the network is up.
5416 Following this logic, two new special targets
5417 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
5418 added that are to cryptsetup.target what remote-fs.target and
5419 remote-fs-pre.target are to local-fs.target.
5421 * Service units gained a new UnsetEnvironment= setting which permits
5422 unsetting specific environment variables for services that are
5423 normally passed to it (for example in order to mask out locale
5424 settings for specific services that can't deal with it).
5426 * Units acquired a new boolean option IPAccounting=. When turned on, IP
5427 traffic accounting (packet count as well as byte count) is done for
5428 the service, and shown as part of "systemctl status" or "systemd-run
5431 * Service units acquired two new options IPAddressAllow= and
5432 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
5433 for configuring a simple IP access control list for all sockets of
5434 the unit. These options are available also on .slice and .socket
5435 units, permitting flexible access list configuration for individual
5436 services as well as groups of services (as defined by a slice unit),
5437 including system-wide. Note that IP ACLs configured this way are
5438 enforced on every single IPv4 and IPv6 socket created by any process
5439 of the service unit, and apply to ingress as well as egress traffic.
5441 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
5442 structured log message is generated each time the unit is stopped,
5443 containing information about the consumed resources of this
5446 * A new setting KeyringMode= has been added to unit files, which may be
5447 used to control how the kernel keyring is set up for executed
5450 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
5451 "systemctl kexec" and "systemctl exit" are now always asynchronous in
5452 behaviour (that is: these commands return immediately after the
5453 operation was enqueued instead of waiting for the operation to
5454 complete). Previously, "systemctl poweroff" and "systemctl reboot"
5455 were asynchronous on systems using systemd-logind (i.e. almost
5456 always, and like they were on sysvinit), and the other three commands
5457 were unconditionally synchronous. With this release this is cleaned
5458 up, and callers will see the same asynchronous behaviour on all
5459 systems for all five operations.
5461 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
5464 * .timer units now accept calendar specifications in other timezones
5465 than UTC or the local timezone.
5467 * The tmpfiles snippet var.conf has been changed to create
5468 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
5469 the "utmp" group already, and it appears to be generally understood
5470 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
5471 databases. Previously this was implemented correctly for all these
5472 databases excepts btmp, which has been opened up like this now
5473 too. Note that while the other databases are world-readable
5474 (i.e. 0644), btmp is not and remains more restrictive.
5476 * The systemd-resolve tool gained a new --reset-server-features
5477 switch. When invoked like this systemd-resolved will forget
5478 everything it learnt about the features supported by the configured
5479 upstream DNS servers, and restarts the feature probing logic on the
5480 next resolver look-up for them at the highest feature level
5483 * The status dump systemd-resolved sends to the logs upon receiving
5484 SIGUSR1 now also includes information about all DNS servers it is
5485 configured to use, and the features levels it probed for them.
5487 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
5488 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
5489 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
5490 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
5491 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
5492 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
5493 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
5494 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
5495 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
5496 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
5497 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
5498 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
5499 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
5500 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
5501 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
5502 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
5503 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
5504 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
5505 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
5506 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
5508 — Berlin, 2017-10-06
5512 * Meson is now supported as build system in addition to Automake. It is
5513 our plan to remove Automake in one of our next releases, so that
5514 Meson becomes our exclusive build system. Hence, please start using
5515 the Meson build system in your downstream packaging. There's plenty
5516 of documentation around how to use Meson, the extremely brief
5519 ./autogen.sh && ./configure && make && sudo make install
5523 meson build && ninja -C build && sudo ninja -C build install
5525 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
5526 which permits configuring a timeout on the time a job is
5527 running. This is particularly useful for setting timeouts on jobs for
5530 * Unit files gained two new options ConditionUser= and ConditionGroup=
5531 for conditionalizing units based on the identity of the user/group
5532 running a systemd user instance.
5534 * systemd-networkd now understands a new FlowLabel= setting in the
5535 [VXLAN] section of .network files, as well as a Priority= in
5536 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
5537 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
5538 gained support for configuration of GENEVE links, and IPv6 address
5539 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
5541 * .link files now understand a new Port= setting.
5543 * systemd-networkd's DHCP support gained support for DHCP option 119
5544 (domain search list).
5546 * systemd-networkd gained support for serving IPv6 address ranges using
5547 the Router Advertisement protocol. The new .network configuration
5548 section [IPv6Prefix] may be used to configure the ranges to
5549 serve. This is implemented based on a new, minimal, native server
5550 implementation of RA.
5552 * journalctl's --output= switch gained support for a new parameter
5553 "short-iso-precise" for a mode where timestamps are shown as precise
5556 * systemd-udevd's "net_id" builtin may now generate stable network
5557 interface names from IBM PowerVM VIO devices as well as ACPI platform
5560 * MulticastDNS support in systemd-resolved may now be explicitly
5561 enabled/disabled using the new MulticastDNS= configuration file
5564 * systemd-resolved may now optionally use libidn2 instead of the libidn
5565 for processing internationalized domain names. Support for libidn2
5566 should be considered experimental and should not be enabled by
5569 * "machinectl pull-tar" and related call may now do verification of
5570 downloaded images using SUSE-style .sha256 checksum files in addition
5571 to the already existing support for validating using Ubuntu-style
5574 * sd-bus gained support for a new sd_bus_message_appendv() call which
5575 is va_list equivalent of sd_bus_message_append().
5577 * sd-boot gained support for validating images using SHIM/MOK.
5579 * The SMACK code learnt support for "onlycap".
5581 * systemd-mount --umount is now much smarter in figuring out how to
5582 properly unmount a device given its mount or device path.
5584 * The code to call libnss_dns as a fallback from libnss_resolve when
5585 the communication with systemd-resolved fails was removed. This
5586 fallback was redundant and interfered with the [!UNAVAIL=return]
5587 suffix. See nss-resolve(8) for the recommended configuration.
5589 * systemd-logind may now be restarted without losing state. It stores
5590 the file descriptors for devices it manages in the system manager
5591 using the FDSTORE= mechanism. Please note that further changes in
5592 other components may be required to make use of this (for example
5593 Xorg has code to listen for stops of systemd-logind and terminate
5594 itself when logind is stopped or restarted, in order to avoid using
5595 stale file descriptors for graphical devices, which is now
5596 counterproductive and must be reverted in order for restarts of
5597 systemd-logind to be safe. See
5598 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
5600 * All kernel-install plugins are called with the environment variable
5601 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
5602 /etc/machine-id. If the machine ID could not be determined,
5603 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
5604 anything in the entry directory (passed as the second argument) if
5605 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
5606 temporary directory is passed as the entry directory and removed
5607 after all the plugins exit.
5609 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
5610 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
5611 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
5612 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
5613 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
5614 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
5615 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
5616 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
5617 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
5618 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
5619 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
5620 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
5621 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
5622 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
5623 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
5624 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
5625 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
5626 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
5627 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
5628 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
5629 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
5630 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
5631 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
5632 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
5633 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
5634 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
5635 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
5636 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
5637 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
5640 — Berlin, 2017-07-12
5644 * The "hybrid" control group mode has been modified to improve
5645 compatibility with "legacy" cgroups-v1 setups. Specifically, the
5646 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
5647 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
5648 cgroups-v1 hierarchy), the only externally visible change being that
5649 the cgroups-v2 hierarchy is also mounted, to
5650 /sys/fs/cgroup/unified. This should provide a large degree of
5651 compatibility with "legacy" cgroups-v1, while taking benefit of the
5652 better management capabilities of cgroups-v2.
5654 * The default control group setup mode may be selected both a boot-time
5655 via a set of kernel command line parameters (specifically:
5656 systemd.unified_cgroup_hierarchy= and
5657 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
5658 default selected on the configure command line
5659 (--with-default-hierarchy=). The upstream default is "hybrid"
5660 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
5661 this will change in a future systemd version to be "unified" (pure
5662 cgroups-v2 mode). The third option for the compile time option is
5663 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
5664 distributions to default to "hybrid" mode for release distributions,
5665 starting with v233. We recommend "unified" for development
5666 distributions (specifically: distributions such as Fedora's rawhide)
5667 as that's where things are headed in the long run. Use "legacy" for
5668 greatest stability and compatibility only.
5670 * Note one current limitation of "unified" and "hybrid" control group
5671 setup modes: the kernel currently does not permit the systemd --user
5672 instance (i.e. unprivileged code) to migrate processes between two
5673 disconnected cgroup subtrees, even if both are managed and owned by
5674 the user. This effectively means "systemd-run --user --scope" doesn't
5675 work when invoked from outside of any "systemd --user" service or
5676 scope. Specifically, it is not supported from session scopes. We are
5677 working on fixing this in a future systemd version. (See #3388 for
5678 further details about this.)
5680 * DBus policy files are now installed into /usr rather than /etc. Make
5681 sure your system has dbus >= 1.9.18 running before upgrading to this
5682 version, or override the install path with --with-dbuspolicydir= .
5684 * All python scripts shipped with systemd (specifically: the various
5685 tests written in Python) now require Python 3.
5687 * systemd unit tests can now run standalone (without the source or
5688 build directories), and can be installed into /usr/lib/systemd/tests/
5689 with 'make install-tests'.
5691 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
5692 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
5695 * Support for the %c, %r, %R specifiers in unit files has been
5696 removed. Specifiers are not supposed to be dependent on configuration
5697 in the unit file itself (so that they resolve the same regardless
5698 where used in the unit files), but these specifiers were influenced
5699 by the Slice= option.
5701 * The shell invoked by debug-shell.service now defaults to /bin/sh in
5702 all cases. If distributions want to use a different shell for this
5703 purpose (for example Fedora's /sbin/sushell) they need to specify
5704 this explicitly at configure time using --with-debug-shell=.
5706 * The confirmation spawn prompt has been reworked to offer the
5709 (c)ontinue, proceed without asking anymore
5710 (D)ump, show the state of the unit
5711 (f)ail, don't execute the command and pretend it failed
5713 (i)nfo, show a short summary of the unit
5714 (j)obs, show jobs that are in progress
5715 (s)kip, don't execute the command and pretend it succeeded
5716 (y)es, execute the command
5718 The 'n' choice for the confirmation spawn prompt has been removed,
5719 because its meaning was confusing.
5721 The prompt may now also be redirected to an alternative console by
5722 specifying the console as parameter to systemd.confirm_spawn=.
5724 * Services of Type=notify require a READY=1 notification to be sent
5725 during startup. If no such message is sent, the service now fails,
5726 even if the main process exited with a successful exit code.
5728 * Services that fail to start up correctly now always have their
5729 ExecStopPost= commands executed. Previously, they'd enter "failed"
5730 state directly, without executing these commands.
5732 * The option MulticastDNS= of network configuration files has acquired
5733 an actual implementation. With MulticastDNS=yes a host can resolve
5734 names of remote hosts and reply to mDNS A and AAAA requests.
5736 * When units are about to be started an additional check is now done to
5737 ensure that all dependencies of type BindsTo= (when used in
5738 combination with After=) have been started.
5740 * systemd-analyze gained a new verb "syscall-filter" which shows which
5741 system call groups are defined for the SystemCallFilter= unit file
5742 setting, and which system calls they contain.
5744 * A new system call filter group "@filesystem" has been added,
5745 consisting of various file system related system calls. Group
5746 "@reboot" has been added, covering reboot, kexec and shutdown related
5747 calls. Finally, group "@swap" has been added covering swap
5748 configuration related calls.
5750 * A new unit file option RestrictNamespaces= has been added that may be
5751 used to restrict access to the various process namespace types the
5752 Linux kernel provides. Specifically, it may be used to take away the
5753 right for a service unit to create additional file system, network,
5754 user, and other namespaces. This sandboxing option is particularly
5755 relevant due to the high amount of recently discovered namespacing
5756 related vulnerabilities in the kernel.
5758 * systemd-udev's .link files gained support for a new AutoNegotiation=
5759 setting for configuring Ethernet auto-negotiation.
5761 * systemd-networkd's .network files gained support for a new
5762 ListenPort= setting in the [DHCP] section to explicitly configure the
5763 UDP client port the DHCP client shall listen on.
5765 * .network files gained a new Unmanaged= boolean setting for explicitly
5766 excluding one or more interfaces from management by systemd-networkd.
5768 * The systemd-networkd ProxyARP= option has been renamed to
5769 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
5770 renamed to ReduceARPProxy=. The old names continue to be available
5773 * systemd-networkd gained support for configuring IPv6 Proxy NDP
5774 addresses via the new IPv6ProxyNDPAddress= .network file setting.
5776 * systemd-networkd's bonding device support gained support for two new
5777 configuration options ActiveSlave= and PrimarySlave=.
5779 * The various options in the [Match] section of .network files gained
5780 support for negative matching.
5782 * New systemd-specific mount options are now understood in /etc/fstab:
5784 x-systemd.mount-timeout= may be used to configure the maximum
5785 permitted runtime of the mount command.
5787 x-systemd.device-bound may be set to bind a mount point to its
5788 backing device unit, in order to automatically remove a mount point
5789 if its backing device is unplugged. This option may also be
5790 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
5791 on the block device, which is now automatically set for all CDROM
5792 drives, so that mounted CDs are automatically unmounted when they are
5793 removed from the drive.
5795 x-systemd.after= and x-systemd.before= may be used to explicitly
5796 order a mount after or before another unit or mount point.
5798 * Enqueued start jobs for device units are now automatically garbage
5799 collected if there are no jobs waiting for them anymore.
5801 * systemctl list-jobs gained two new switches: with --after, for every
5802 queued job the jobs it's waiting for are shown; with --before the
5803 jobs which it's blocking are shown.
5805 * systemd-nspawn gained support for ephemeral boots from disk images
5806 (or in other words: --ephemeral and --image= may now be
5807 combined). Moreover, ephemeral boots are now supported for normal
5808 directories, even if the backing file system is not btrfs. Of course,
5809 if the file system does not support file system snapshots or
5810 reflinks, the initial copy operation will be relatively expensive, but
5811 this should still be suitable for many use cases.
5813 * Calendar time specifications in .timer units now support
5814 specifications relative to the end of a month by using "~" instead of
5815 "-" as separator between month and day. For example, "*-02~03" means
5816 "the third last day in February". In addition a new syntax for
5817 repeated events has been added using the "/" character. For example,
5818 "9..17/2:00" means "every two hours from 9am to 5pm".
5820 * systemd-socket-proxyd gained a new parameter --connections-max= for
5821 configuring the maximum number of concurrent connections.
5823 * sd-id128 gained a new API for generating unique IDs for the host in a
5824 way that does not leak the machine ID. Specifically,
5825 sd_id128_get_machine_app_specific() derives an ID based on the
5826 machine ID in a well-defined, non-reversible, stable way. This is
5827 useful whenever an identifier for the host is needed but where the
5828 identifier shall not be useful to identify the system beyond the
5829 scope of the application itself. (Internally this uses HMAC-SHA256 as
5830 keyed hash function using the machine ID as input.)
5832 * NotifyAccess= gained a new supported value "exec". When set
5833 notifications are accepted from all processes systemd itself invoked,
5834 including all control processes.
5836 * .nspawn files gained support for defining overlay mounts using the
5837 Overlay= and OverlayReadOnly= options. Previously this functionality
5838 was only available on the systemd-nspawn command line.
5840 * systemd-nspawn's --bind= and --overlay= options gained support for
5841 bind/overlay mounts whose source lies within the container tree by
5842 prefixing the source path with "+".
5844 * systemd-nspawn's --bind= and --overlay= options gained support for
5845 automatically allocating a temporary source directory in /var/tmp
5846 that is removed when the container dies. Specifically, if the source
5847 directory is specified as empty string this mechanism is selected. An
5848 example usage is --overlay=+/var::/var, which creates an overlay
5849 mount based on the original /var contained in the image, overlaid
5850 with a temporary directory in the host's /var/tmp. This way changes
5851 to /var are automatically flushed when the container shuts down.
5853 * systemd-nspawn --image= option does now permit raw file system block
5854 devices (in addition to images containing partition tables, as
5857 * The disk image dissection logic in systemd-nspawn gained support for
5858 automatically setting up LUKS encrypted as well as Verity protected
5859 partitions. When a container is booted from an encrypted image the
5860 passphrase is queried at start-up time. When a container with Verity
5861 data is started, the root hash is search in a ".roothash" file
5862 accompanying the disk image (alternatively, pass the root hash via
5863 the new --root-hash= command line option).
5865 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
5866 be used to dissect disk images the same way as systemd-nspawn does
5867 it, following the Bootable Partition Specification. It may even be
5868 used to mount disk images with complex partition setups (including
5869 LUKS and Verity partitions) to a local host directory, in order to
5870 inspect them. This tool is not considered public API (yet), and is
5871 thus not installed into /usr/bin. Please do not rely on its
5872 existence, since it might go away or be changed in later systemd
5875 * A new generator "systemd-verity-generator" has been added, similar in
5876 style to "systemd-cryptsetup-generator", permitting automatic setup of
5877 Verity root partitions when systemd boots up. In order to make use of
5878 this your partition setup should follow the Discoverable Partitions
5879 Specification, and the GPT partition ID of the root file system
5880 partition should be identical to the upper 128bit of the Verity root
5881 hash. The GPT partition ID of the Verity partition protecting it
5882 should be the lower 128bit of the Verity root hash. If the partition
5883 image follows this model it is sufficient to specify a single
5884 "roothash=" kernel command line argument to both configure which root
5885 image and verity partition to use as well as the root hash for
5886 it. Note that systemd-nspawn's Verity support follows the same
5887 semantics, meaning that disk images with proper Verity data in place
5888 may be booted in containers with systemd-nspawn as well as on
5889 physical systems via the verity generator. Also note that the "mkosi"
5890 tool available at https://github.com/systemd/mkosi has been updated
5891 to generate Verity protected disk images following this scheme. In
5892 fact, it has been updated to generate disk images that optionally
5893 implement a complete UEFI SecureBoot trust chain, involving a signed
5894 kernel and initrd image that incorporates such a root hash as well as
5895 a Verity-enabled root partition.
5897 * The hardware database (hwdb) udev supports has been updated to carry
5898 accelerometer quirks.
5900 * All system services are now run with a fresh kernel keyring set up
5901 for them. The invocation ID is stored by default in it, thus
5902 providing a safe, non-overridable way to determine the invocation
5905 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
5906 options for bind mounting arbitrary paths in a service-specific
5907 way. When these options are used, arbitrary host or service files and
5908 directories may be mounted to arbitrary locations in the service's
5911 * Documentation has been added that lists all of systemd's low-level
5912 environment variables:
5914 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
5916 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
5917 whether a specific socket file descriptor matches a specified socket
5920 * systemd-firstboot has been updated to check for the
5921 systemd.firstboot= kernel command line option. It accepts a boolean
5922 and when set to false the first boot questions are skipped.
5924 * systemd-fstab-generator has been updated to check for the
5925 systemd.volatile= kernel command line option, which either takes an
5926 optional boolean parameter or the special value "state". If used the
5927 system may be booted in a "volatile" boot mode. Specifically,
5928 "systemd.volatile" is used, the root directory will be mounted as
5929 tmpfs, and only /usr is mounted from the actual root file system. If
5930 "systemd.volatile=state" is used, the root directory will be mounted
5931 as usual, but /var is mounted as tmpfs. This concept provides similar
5932 functionality as systemd-nspawn's --volatile= option, but provides it
5933 on physical boots. Use this option for implementing stateless
5934 systems, or testing systems with all state and/or configuration reset
5935 to the defaults. (Note though that many distributions are not
5936 prepared to boot up without a populated /etc or /var, though.)
5938 * systemd-gpt-auto-generator gained support for LUKS encrypted root
5939 partitions. Previously it only supported LUKS encrypted partitions
5940 for all other uses, except for the root partition itself.
5942 * Socket units gained support for listening on AF_VSOCK sockets for
5943 communication in virtualized QEMU environments.
5945 * The "configure" script gained a new option --with-fallback-hostname=
5946 for specifying the fallback hostname to use if none is configured in
5947 /etc/hostname. For example, by specifying
5948 --with-fallback-hostname=fedora it is possible to default to a
5949 hostname of "fedora" on pristine installations.
5951 * systemd-cgls gained support for a new --unit= switch for listing only
5952 the control groups of a specific unit. Similar --user-unit= has been
5953 added for listing only the control groups of a specific user unit.
5955 * systemd-mount gained a new --umount switch for unmounting a mount or
5956 automount point (and all mount/automount points below it).
5958 * systemd will now refuse full configuration reloads (via systemctl
5959 daemon-reload and related calls) unless at least 16MiB of free space
5960 are available in /run. This is a safety precaution in order to ensure
5961 that generators can safely operate after the reload completed.
5963 * A new unit file option RootImage= has been added, which has a similar
5964 effect as RootDirectory= but mounts the service's root directory from
5965 a disk image instead of plain directory. This logic reuses the same
5966 image dissection and mount logic that systemd-nspawn already uses,
5967 and hence supports any disk images systemd-nspawn supports, including
5968 those following the Discoverable Partition Specification, as well as
5969 Verity enabled images. This option enables systemd to run system
5970 services directly off disk images acting as resource bundles,
5971 possibly even including full integrity data.
5973 * A new MountAPIVFS= unit file option has been added, taking a boolean
5974 argument. If enabled /proc, /sys and /dev (collectively called the
5975 "API VFS") will be mounted for the service. This is only relevant if
5976 RootDirectory= or RootImage= is used for the service, as these mounts
5977 are of course in place in the host mount namespace anyway.
5979 * systemd-nspawn gained support for a new --pivot-root= switch. If
5980 specified the root directory within the container image is pivoted to
5981 the specified mount point, while the original root disk is moved to a
5982 different place. This option enables booting of ostree images
5983 directly with systemd-nspawn.
5985 * The systemd build scripts will no longer complain if the NTP server
5986 addresses are not changed from the defaults. Google now supports
5987 these NTP servers officially. We still recommend downstreams to
5988 properly register an NTP pool with the NTP pool project though.
5990 * coredumpctl gained a new "--reverse" option for printing the list
5991 of coredumps in reverse order.
5993 * coredumpctl will now show additional information about truncated and
5994 inaccessible coredumps, as well as coredumps that are still being
5995 processed. It also gained a new --quiet switch for suppressing
5996 additional informational message in its output.
5998 * coredumpctl gained support for only showing coredumps newer and/or
5999 older than specific timestamps, using the new --since= and --until=
6000 options, reminiscent of journalctl's options by the same name.
6002 * The systemd-coredump logic has been improved so that it may be reused
6003 to collect backtraces in non-compiled languages, for example in
6004 scripting languages such as Python.
6006 * machinectl will now show the UID shift of local containers, if user
6007 namespacing is enabled for them.
6009 * systemd will now optionally run "environment generator" binaries at
6010 configuration load time. They may be used to add environment
6011 variables to the environment block passed to services invoked. One
6012 user environment generator is shipped by default that sets up
6013 environment variables based on files dropped into /etc/environment.d
6014 and ~/.config/environment.d/.
6016 * systemd-resolved now includes the new, recently published 2017 DNSSEC
6019 * hostnamed has been updated to report a new chassis type of
6020 "convertible" to cover "foldable" laptops that can both act as a
6021 tablet and as a laptop, such as various Lenovo Yoga devices.
6023 Contributions from: Adrián López, Alexander Galanin, Alexander
6024 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
6025 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
6026 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
6027 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
6028 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
6029 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
6030 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
6031 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
6032 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
6033 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
6034 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
6035 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
6036 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
6037 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
6038 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
6039 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
6040 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
6041 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
6042 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
6043 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
6044 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
6045 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
6046 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
6047 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
6048 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
6049 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
6050 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
6053 — Berlin, 2017-03-01
6057 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
6058 RestrictAddressFamilies= enabled. These sandboxing options should
6059 generally be compatible with the various external udev call-out
6060 binaries we are aware of, however there may be exceptions, in
6061 particular when exotic languages for these call-outs are used. In
6062 this case, consider turning off these settings locally.
6064 * The new RemoveIPC= option can be used to remove IPC objects owned by
6065 the user or group of a service when that service exits.
6067 * The new ProtectKernelModules= option can be used to disable explicit
6068 load and unload operations of kernel modules by a service. In
6069 addition access to /usr/lib/modules is removed if this option is set.
6071 * ProtectSystem= option gained a new value "strict", which causes the
6072 whole file system tree with the exception of /dev, /proc, and /sys,
6073 to be remounted read-only for a service.
6075 * The new ProtectKernelTunables= option can be used to disable
6076 modification of configuration files in /sys and /proc by a service.
6077 Various directories and files are remounted read-only, so access is
6078 restricted even if the file permissions would allow it.
6080 * The new ProtectControlGroups= option can be used to disable write
6081 access by a service to /sys/fs/cgroup.
6083 * Various systemd services have been hardened with
6084 ProtectKernelTunables=yes, ProtectControlGroups=yes,
6085 RestrictAddressFamilies=.
6087 * Support for dynamically creating users for the lifetime of a service
6088 has been added. If DynamicUser=yes is specified, user and group IDs
6089 will be allocated from the range 61184…65519 for the lifetime of the
6090 service. They can be resolved using the new nss-systemd.so NSS
6091 module. The module must be enabled in /etc/nsswitch.conf. Services
6092 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
6093 any resources allocated by the service will be cleaned up when the
6094 service exits. They also have ProtectHome=read-only and
6095 ProtectSystem=strict enabled, so they are not able to make any
6096 permanent modifications to the system.
6098 * The nss-systemd module also always resolves root and nobody, making
6099 it possible to have no /etc/passwd or /etc/group files in minimal
6100 container or chroot environments.
6102 * Services may be started with their own user namespace using the new
6103 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
6104 under which the service is running are mapped. All other users are
6107 * Support for the cgroup namespace has been added to systemd-nspawn. If
6108 supported by kernel, the container system started by systemd-nspawn
6109 will have its own view of the cgroup hierarchy. This new behaviour
6110 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
6112 * The new MemorySwapMax= option can be used to limit the maximum swap
6113 usage under the unified cgroup hierarchy.
6115 * Support for the CPU controller in the unified cgroup hierarchy has
6116 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
6117 options. This controller requires out-of-tree patches for the kernel
6118 and the support is provisional.
6120 * Mount and automount units may now be created transiently
6121 (i.e. dynamically at runtime via the bus API, instead of requiring
6122 unit files in the file system).
6124 * systemd-mount is a new tool which may mount file systems – much like
6125 mount(8), optionally pulling in additional dependencies through
6126 transient .mount and .automount units. For example, this tool
6127 automatically runs fsck on a backing block device before mounting,
6128 and allows the automount logic to be used dynamically from the
6129 command line for establishing mount points. This tool is particularly
6130 useful when dealing with removable media, as it will ensure fsck is
6131 run – if necessary – before the first access and that the file system
6132 is quickly unmounted after each access by utilizing the automount
6133 logic. This maximizes the chance that the file system on the
6134 removable media stays in a clean state, and if it isn't in a clean
6135 state is fixed automatically.
6137 * LazyUnmount=yes option for mount units has been added to expose the
6138 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
6141 * /efi will be used as the mount point of the EFI boot partition, if
6142 the directory is present, and the mount point was not configured
6143 through other means (e.g. fstab). If /efi directory does not exist,
6144 /boot will be used as before. This makes it easier to automatically
6145 mount the EFI partition on systems where /boot is used for something
6148 * When operating on GPT disk images for containers, systemd-nspawn will
6149 now mount the ESP to /boot or /efi according to the same rules as PID
6150 1 running on a host. This allows tools like "bootctl" to operate
6151 correctly within such containers, in order to make container images
6152 bootable on physical systems.
6154 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
6156 * Two new user session targets have been added to support running
6157 graphical sessions under the systemd --user instance:
6158 graphical-session.target and graphical-session-pre.target. See
6159 systemd.special(7) for a description of how those targets should be
6162 * The vconsole initialization code has been significantly reworked to
6163 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
6164 support unicode keymaps. Font and keymap configuration will now be
6165 copied to all allocated virtual consoles.
6167 * FreeBSD's bhyve virtualization is now detected.
6169 * Information recorded in the journal for core dumps now includes the
6170 contents of /proc/mountinfo and the command line of the process at
6171 the top of the process hierarchy (which is usually the init process
6174 * systemd-journal-gatewayd learned the --directory= option to serve
6175 files from the specified location.
6177 * journalctl --root=… can be used to peruse the journal in the
6178 /var/log/ directories inside of a container tree. This is similar to
6179 the existing --machine= option, but does not require the container to
6182 * The hardware database has been extended to support
6183 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
6186 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
6187 specify the click rate for mice which include a horizontal wheel with
6188 a click rate that is different than the one for the vertical wheel.
6190 * systemd-run gained a new --wait option that makes service execution
6191 synchronous. (Specifically, the command will not return until the
6192 specified service binary exited.)
6194 * systemctl gained a new --wait option that causes the start command to
6195 wait until the units being started have terminated again.
6197 * A new journal output mode "short-full" has been added which displays
6198 timestamps with abbreviated English day names and adds a timezone
6199 suffix. Those timestamps include more information than the default
6200 "short" output mode, and can be passed directly to journalctl's
6201 --since= and --until= options.
6203 * /etc/resolv.conf will be bind-mounted into containers started by
6204 systemd-nspawn, if possible, so any changes to resolv.conf contents
6205 are automatically propagated to the container.
6207 * The number of instances for socket-activated services originating
6208 from a single IP address can be limited with
6209 MaxConnectionsPerSource=, extending the existing setting of
6212 * systemd-networkd gained support for vcan ("Virtual CAN") interface
6215 * .netdev and .network configuration can now be extended through
6218 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
6219 Segmentation Offload, Generic Receive Offload, Large Receive Offload
6220 can be enabled and disabled using the new UDPSegmentationOffload=,
6221 TCPSegmentationOffload=, GenericSegmentationOffload=,
6222 GenericReceiveOffload=, LargeReceiveOffload= options in the
6223 [Link] section of .link files.
6225 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
6226 Port VLAN ID can be configured for bridge devices using the new STP=,
6227 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
6228 section of .netdev files.
6230 * The route table to which routes received over DHCP or RA should be
6231 added can be configured with the new RouteTable= option in the [DHCP]
6232 and [IPv6AcceptRA] sections of .network files.
6234 * The Address Resolution Protocol can be disabled on links managed by
6235 systemd-networkd using the ARP=no setting in the [Link] section of
6238 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
6239 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
6240 encode information about the result and exit codes of the current
6241 service runtime cycle.
6243 * systemd-sysctl will now configure kernel parameters in the order
6244 they occur in the configuration files. This matches what sysctl
6245 has been traditionally doing.
6247 * kernel-install "plugins" that are executed to perform various
6248 tasks after a new kernel is added and before an old one is removed
6249 can now return a special value to terminate the procedure and
6250 prevent any later plugins from running.
6252 * Journald's SplitMode=login setting has been deprecated. It has been
6253 removed from documentation, and its use is discouraged. In a future
6254 release it will be completely removed, and made equivalent to current
6255 default of SplitMode=uid.
6257 * Storage=both option setting in /etc/systemd/coredump.conf has been
6258 removed. With fast LZ4 compression storing the core dump twice is not
6261 * The --share-system systemd-nspawn option has been replaced with an
6262 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
6263 this functionality is discouraged. In addition the variables
6264 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
6265 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
6266 individual namespaces.
6268 * "machinectl list" now shows the IP address of running containers in
6269 the output, as well as OS release information.
6271 * "loginctl list" now shows the TTY of each session in the output.
6273 * sd-bus gained new API calls sd_bus_track_set_recursive(),
6274 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
6275 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
6276 tracking objects in a "recursive" mode, where a single client can be
6277 counted multiple times, if it takes multiple references.
6279 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
6280 sd_bus_get_exit_on_disconnect(). They may be used to make a
6281 process using sd-bus automatically exit if the bus connection is
6284 * Bus clients of the service manager may now "pin" loaded units into
6285 memory, by taking an explicit reference on them. This is useful to
6286 ensure the client can retrieve runtime data about the service even
6287 after the service completed execution. Taking such a reference is
6288 available only for privileged clients and should be helpful to watch
6289 running services in a race-free manner, and in particular collect
6290 information about exit statuses and results.
6292 * The nss-resolve module has been changed to strictly return UNAVAIL
6293 when communication via D-Bus with resolved failed, and NOTFOUND when
6294 a lookup completed but was negative. This means it is now possible to
6295 neatly configure fallbacks using nsswitch.conf result checking
6296 expressions. Taking benefit of this, the new recommended
6297 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
6299 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
6301 * A new setting CtrlAltDelBurstAction= has been added to
6302 /etc/systemd/system.conf which may be used to configure the precise
6303 behaviour if the user on the console presses Ctrl-Alt-Del more often
6304 than 7 times in 2s. Previously this would unconditionally result in
6305 an expedited, immediate reboot. With this new setting the precise
6306 operation may be configured in more detail, and also turned off
6309 * In .netdev files two new settings RemoteChecksumTx= and
6310 RemoteChecksumRx= are now understood that permit configuring the
6311 remote checksumming logic for VXLAN networks.
6313 * The service manager learnt a new "invocation ID" concept for invoked
6314 services. Each runtime cycle of a service will get a new invocation
6315 ID (a 128bit random UUID) assigned that identifies the current
6316 run of the service uniquely and globally. A new invocation ID
6317 is generated each time a service starts up. The journal will store
6318 the invocation ID of a service along with any logged messages, thus
6319 making the invocation ID useful for matching the online runtime of a
6320 service with the offline log data it generated in a safe way without
6321 relying on synchronized timestamps. In many ways this new service
6322 invocation ID concept is similar to the kernel's boot ID concept that
6323 uniquely and globally identifies the runtime of each boot. The
6324 invocation ID of a service is passed to the service itself via an
6325 environment variable ($INVOCATION_ID). A new bus call
6326 GetUnitByInvocationID() has been added that is similar to GetUnit()
6327 but instead of retrieving the bus path for a unit by its name
6328 retrieves it by its invocation ID. The returned path is valid only as
6329 long as the passed invocation ID is current.
6331 * systemd-resolved gained a new "DNSStubListener" setting in
6332 resolved.conf. It either takes a boolean value or the special values
6333 "udp" and "tcp", and configures whether to enable the stub DNS
6334 listener on 127.0.0.53:53.
6336 * IP addresses configured via networkd may now carry additional
6337 configuration settings supported by the kernel. New options include:
6338 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
6339 PrefixRoute=, AutoJoin=.
6341 * The PAM configuration fragment file for "user@.service" shipped with
6342 systemd (i.e. the --user instance of systemd) has been stripped to
6343 the minimum necessary to make the system boot. Previously, it
6344 contained Fedora-specific stanzas that did not apply to other
6345 distributions. It is expected that downstream distributions add
6346 additional configuration lines, matching their needs to this file,
6347 using it only as rough template of what systemd itself needs. Note
6348 that this reduced fragment does not even include an invocation of
6349 pam_limits which most distributions probably want to add, even though
6350 systemd itself does not need it. (There's also the new build time
6351 option --with-pamconfdir=no to disable installation of the PAM
6354 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
6355 capability is now also dropped from its set (in addition to
6356 CAP_SYS_MKNOD as before).
6358 * In service unit files it is now possible to connect a specific named
6359 file descriptor with stdin/stdout/stdout of an executed service. The
6360 name may be specified in matching .socket units using the
6361 FileDescriptorName= setting.
6363 * A number of journal settings may now be configured on the kernel
6364 command line. Specifically, the following options are now understood:
6365 systemd.journald.max_level_console=,
6366 systemd.journald.max_level_store=,
6367 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
6368 systemd.journald.max_level_wall=.
6370 * "systemctl is-enabled --full" will now show by which symlinks a unit
6371 file is enabled in the unit dependency tree.
6373 * Support for VeraCrypt encrypted partitions has been added to the
6374 "cryptsetup" logic and /etc/crypttab.
6376 * systemd-detect-virt gained support for a new --private-users switch
6377 that checks whether the invoking processes are running inside a user
6378 namespace. Similar, a new special value "private-users" for the
6379 existing ConditionVirtualization= setting has been added, permitting
6380 skipping of specific units in user namespace environments.
6382 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
6383 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
6384 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
6385 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
6386 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
6387 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
6388 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
6389 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
6390 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
6391 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
6392 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
6393 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
6394 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
6395 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
6396 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
6397 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
6398 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
6399 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
6400 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
6401 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
6402 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
6403 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
6404 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
6405 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
6406 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
6407 Jędrzejewski-Szmek, Zeal Jagannatha
6409 — Santa Fe, 2016-11-03
6413 * In service units the various ExecXYZ= settings have been extended
6414 with an additional special character as first argument of the
6415 assigned value: if the character '+' is used the specified command
6416 line it will be run with full privileges, regardless of User=,
6417 Group=, CapabilityBoundingSet= and similar options. The effect is
6418 similar to the existing PermissionsStartOnly= option, but allows
6419 configuration of this concept for each executed command line
6422 * Services may now alter the service watchdog timeout at runtime by
6423 sending a WATCHDOG_USEC= message via sd_notify().
6425 * MemoryLimit= and related unit settings now optionally take percentage
6426 specifications. The percentage is taken relative to the amount of
6427 physical memory in the system (or in case of containers, the assigned
6428 amount of memory). This allows scaling service resources neatly with
6429 the amount of RAM available on the system. Similarly, systemd-logind's
6430 RuntimeDirectorySize= option now also optionally takes percentage
6433 * In similar fashion TasksMax= takes percentage values now, too. The
6434 value is taken relative to the configured maximum number of processes
6435 on the system. The per-service task maximum has been changed to 15%
6436 using this functionality. (Effectively this is an increase of 512 →
6437 4915 for service units, given the kernel's default pid_max setting.)
6439 * Calendar time specifications in .timer units now understand a ".."
6440 syntax for time ranges. Example: "4..7:10" may now be used for
6441 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
6444 * The InaccessableDirectories=, ReadOnlyDirectories= and
6445 ReadWriteDirectories= unit file settings have been renamed to
6446 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
6447 applied to all kinds of file nodes, and not just directories, with
6448 the exception of symlinks. Specifically these settings may now be
6449 used on block and character device nodes, UNIX sockets and FIFOS as
6450 well as regular files. The old names of these settings remain
6451 available for compatibility.
6453 * systemd will now log about all service processes it kills forcibly
6454 (using SIGKILL) because they remained after the clean shutdown phase
6455 of the service completed. This should help identifying services that
6456 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
6457 systemd-logind's configuration a similar log message is generated for
6458 processes killed at the end of each session due to this setting.
6460 * systemd will now set the $JOURNAL_STREAM environment variable for all
6461 services whose stdout/stderr are connected to the Journal (which
6462 effectively means by default: all services). The variable contains
6463 the device and inode number of the file descriptor used for
6464 stdout/stderr. This may be used by invoked programs to detect whether
6465 their stdout/stderr is connected to the Journal, in which case they
6466 can switch over to direct Journal communication, thus being able to
6467 pass extended, structured metadata along with their log messages. As
6468 one example, this is now used by glib's logging primitives.
6470 * When using systemd's default tmp.mount unit for /tmp, the mount point
6471 will now be established with the "nosuid" and "nodev" options. This
6472 avoids privilege escalation attacks that put traps and exploits into
6473 /tmp. However, this might cause problems if you e. g. put container
6474 images or overlays into /tmp; if you need this, override tmp.mount's
6475 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
6478 * systemd now supports the "memory" cgroup controller also on
6481 * The systemd-cgtop tool now optionally takes a control group path as
6482 command line argument. If specified, the control group list shown is
6483 limited to subgroups of that group.
6485 * The SystemCallFilter= unit file setting gained support for
6486 pre-defined, named system call filter sets. For example
6487 SystemCallFilter=@clock is now an effective way to make all clock
6488 changing-related system calls unavailable to a service. A number of
6489 similar pre-defined groups are defined. Writing system call filters
6490 for system services is simplified substantially with this new
6491 concept. Accordingly, all of systemd's own, long-running services now
6492 enable system call filtering based on this, by default.
6494 * A new service setting MemoryDenyWriteExecute= has been added, taking
6495 a boolean value. If turned on, a service may no longer create memory
6496 mappings that are writable and executable at the same time. This
6497 enhances security for services where this is enabled as it becomes
6498 harder to dynamically write and then execute memory in exploited
6499 service processes. This option has been enabled for all of systemd's
6500 own long-running services.
6502 * A new RestrictRealtime= service setting has been added, taking a
6503 boolean argument. If set the service's processes may no longer
6504 acquire realtime scheduling. This improves security as realtime
6505 scheduling may otherwise be used to easily freeze the system.
6507 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
6508 value. This may be used for requesting that the system manager inside
6509 of the container reports start-up completion to nspawn which then
6510 propagates this notification further to the service manager
6511 supervising nspawn itself. A related option NotifyReady= in .nspawn
6512 files has been added too. This functionality allows ordering of the
6513 start-up of multiple containers using the usual systemd ordering
6516 * machinectl gained a new command "stop" that is an alias for
6519 * systemd-resolved gained support for contacting DNS servers on
6520 link-local IPv6 addresses.
6522 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
6523 its caches. A method call for requesting the same operation has been
6524 added to the bus API too, and is made available via "systemd-resolve
6527 * systemd-resolve gained a new --status switch. If passed a brief
6528 summary of the used DNS configuration with per-interface information
6531 * resolved.conf gained a new Cache= boolean option, defaulting to
6532 on. If turned off local DNS caching is disabled. This comes with a
6533 performance penalty in particular when DNSSEC is enabled. Note that
6534 resolved disables its internal caching implicitly anyway, when the
6535 configured DNS server is on a host-local IP address such as ::1 or
6536 127.0.0.1, thus automatically avoiding double local caching.
6538 * systemd-resolved now listens on the local IP address 127.0.0.53:53
6539 for DNS requests. This improves compatibility with local programs
6540 that do not use the libc NSS or systemd-resolved's bus APIs for name
6541 resolution. This minimal DNS service is only available to local
6542 programs and does not implement the full DNS protocol, but enough to
6543 cover local DNS clients. A new, static resolv.conf file, listing just
6544 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
6545 now recommended to make /etc/resolv.conf a symlink to this file in
6546 order to route all DNS lookups to systemd-resolved, regardless if
6547 done via NSS, the bus API or raw DNS packets. Note that this local
6548 DNS service is not as fully featured as the libc NSS or
6549 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
6550 used to deliver link-local address information (as this implies
6551 sending a local interface index along), LLMNR/mDNS support via this
6552 interface is severely restricted. It is thus strongly recommended for
6553 all applications to use the libc NSS API or native systemd-resolved
6556 * systemd-networkd's bridge support learned a new setting
6557 VLANFiltering= for controlling VLAN filtering. Moreover a new section
6558 in .network files has been added for configuring VLAN bridging in
6559 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
6561 * systemd-networkd's IPv6 Router Advertisement code now makes use of
6562 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
6563 now be acquired without relying on DHCPv6. Two new options
6564 UseDomains= and UseDNS= have been added to configure this behaviour.
6566 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
6567 renamed IPv6AcceptRA=, without altering its behaviour. The old
6568 setting name remains available for compatibility reasons.
6570 * The systemd-networkd VTI/VTI6 tunneling support gained new options
6571 Key=, InputKey= and OutputKey=.
6573 * systemd-networkd gained support for VRF ("Virtual Routing Function")
6574 interface configuration.
6576 * "systemctl edit" may now be used to create new unit files by
6577 specifying the --force switch.
6579 * sd-event gained a new function sd_event_get_iteration() for
6580 requesting the current iteration counter of the event loop. It starts
6581 at zero and is increased by one with each event loop iteration.
6583 * A new rpm macro %systemd_ordering is provided by the macros.systemd
6584 file. It can be used in lieu of %systemd_requires in packages which
6585 don't use any systemd functionality and are intended to be installed
6586 in minimal containers without systemd present. This macro provides
6587 ordering dependencies to ensure that if the package is installed in
6588 the same rpm transaction as systemd, systemd will be installed before
6589 the scriptlets for the package are executed, allowing unit presets
6592 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
6593 been added to simplify packaging of generators.
6595 * The os-release file gained VERSION_CODENAME field for the
6596 distribution nickname (e.g. VERSION_CODENAME=woody).
6598 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
6599 can be set to disable parsing of metadata and the creation
6600 of persistent symlinks for that device.
6602 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
6603 to make them available to logged-in users has been reverted.
6605 * Much of the common code of the various systemd components is now
6606 built into an internal shared library libsystemd-shared-231.so
6607 (incorporating the systemd version number in the name, to be updated
6608 with future releases) that the components link to. This should
6609 decrease systemd footprint both in memory during runtime and on
6610 disk. Note that the shared library is not for public use, and is
6611 neither API nor ABI stable, but is likely to change with every new
6612 released update. Packagers need to make sure that binaries
6613 linking to libsystemd-shared.so are updated in step with the
6616 * Configuration for "mkosi" is now part of the systemd
6617 repository. mkosi is a tool to easily build legacy-free OS images,
6618 and is available on github: https://github.com/systemd/mkosi. If
6619 "mkosi" is invoked in the build tree a new raw OS image is generated
6620 incorporating the systemd sources currently being worked on and a
6621 clean, fresh distribution installation. The generated OS image may be
6622 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
6623 UEFI PC. This functionality is particularly useful to easily test
6624 local changes made to systemd in a pristine, defined environment. See
6625 doc/HACKING for details.
6627 * configure learned the --with-support-url= option to specify the
6628 distribution's bugtracker.
6630 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
6631 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
6632 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
6633 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
6634 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
6635 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
6636 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
6637 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
6638 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
6639 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
6640 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
6641 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
6642 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
6643 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
6644 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
6645 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
6646 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
6647 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
6648 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
6650 — Berlin, 2016-07-25
6654 * DNSSEC is now turned on by default in systemd-resolved (in
6655 "allow-downgrade" mode), but may be turned off during compile time by
6656 passing "--with-default-dnssec=no" to "configure" (and of course,
6657 during runtime with DNSSEC= in resolved.conf). We recommend
6658 downstreams to leave this on at least during development cycles and
6659 report any issues with the DNSSEC logic upstream. We are very
6660 interested in collecting feedback about the DNSSEC validator and its
6661 limitations in the wild. Note however, that DNSSEC support is
6662 probably nothing downstreams should turn on in stable distros just
6663 yet, as it might create incompatibilities with a few DNS servers and
6664 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
6665 automatically whenever we detect such incompatible setups, but there
6666 might be systems we do not cover yet. Hence: please help us testing
6667 the DNSSEC code, leave this on where you can, report back, but then
6668 again don't consider turning this on in your stable, LTS or
6669 production release just yet. (Note that you have to enable
6670 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
6671 and its DNSSEC mode for hostname resolution from local
6674 * systemd-resolve conveniently resolves DANE records with the --tlsa
6675 option and OPENPGPKEY records with the --openpgp option. It also
6676 supports dumping raw DNS record data via the new --raw= switch.
6678 * systemd-logind will now by default terminate user processes that are
6679 part of the user session scope unit (session-XX.scope) when the user
6680 logs out. This behavior is controlled by the KillUserProcesses=
6681 setting in logind.conf, and the previous default of "no" is now
6682 changed to "yes". This means that user sessions will be properly
6683 cleaned up after, but additional steps are necessary to allow
6684 intentionally long-running processes to survive logout.
6686 While the user is logged in at least once, user@.service is running,
6687 and any service that should survive the end of any individual login
6688 session can be started at a user service or scope using systemd-run.
6689 systemd-run(1) man page has been extended with an example which shows
6690 how to run screen in a scope unit underneath user@.service. The same
6691 command works for tmux.
6693 After the user logs out of all sessions, user@.service will be
6694 terminated too, by default, unless the user has "lingering" enabled.
6695 To effectively allow users to run long-term tasks even if they are
6696 logged out, lingering must be enabled for them. See loginctl(1) for
6697 details. The default polkit policy was modified to allow users to
6698 set lingering for themselves without authentication.
6700 Previous defaults can be restored at compile time by the
6701 --without-kill-user-processes option to "configure".
6703 * systemd-logind gained new configuration settings SessionsMax= and
6704 InhibitorsMax=, both with a default of 8192. It will not register new
6705 user sessions or inhibitors above this limit.
6707 * systemd-logind will now reload configuration on SIGHUP.
6709 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
6710 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
6711 enable. Also, support for the "io" cgroup controller in the unified
6712 hierarchy has been added, so that the "memory", "pids" and "io" are
6713 now the controllers that are supported on the unified hierarchy.
6715 WARNING: it is not possible to use previous systemd versions with
6716 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
6717 is necessary to also update systemd in the initramfs if using the
6718 unified hierarchy. An updated SELinux policy is also required.
6720 * LLDP support has been extended, and both passive (receive-only) and
6721 active (sender) modes are supported. Passive mode ("routers-only") is
6722 enabled by default in systemd-networkd. Active LLDP mode is enabled
6723 by default for containers on the internal network. The "networkctl
6724 lldp" command may be used to list information gathered. "networkctl
6725 status" will also show basic LLDP information on connected peers now.
6727 * The IAID and DUID unique identifier sent in DHCP requests may now be
6728 configured for the system and each .network file managed by
6729 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
6731 * systemd-networkd gained support for configuring proxy ARP support for
6732 each interface, via the ProxyArp= setting in .network files. It also
6733 gained support for configuring the multicast querier feature of
6734 bridge devices, via the new MulticastQuerier= setting in .netdev
6735 files. Similarly, snooping on the IGMP traffic can be controlled
6736 via the new setting MulticastSnooping=.
6738 A new setting PreferredLifetime= has been added for addresses
6739 configured in .network file to configure the lifetime intended for an
6742 The systemd-networkd DHCP server gained the option EmitRouter=, which
6743 defaults to yes, to configure whether the DHCP Option 3 (Router)
6746 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
6747 systemd-socket-activate and installed into /usr/bin. It is now fully
6750 * systemd-journald now uses separate threads to flush changes to disk
6751 when closing journal files, thus reducing impact of slow disk I/O on
6752 logging performance.
6754 * The sd-journal API gained two new calls
6755 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
6756 can be used to open journal files using file descriptors instead of
6757 file or directory paths. sd_journal_open_container() has been
6758 deprecated, sd_journal_open_directory_fd() should be used instead
6759 with the flag SD_JOURNAL_OS_ROOT.
6761 * journalctl learned a new output mode "-o short-unix" that outputs log
6762 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
6763 UTC). It also gained support for a new --no-hostname setting to
6764 suppress the hostname column in the family of "short" output modes.
6766 * systemd-ask-password now optionally skips printing of the password to
6767 stdout with --no-output which can be useful in scripts.
6769 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
6770 (devices tagged with ID_MAKER_TOOL) are now tagged with
6771 "uaccess" and are available to logged in users.
6773 * The DeviceAllow= unit setting now supports specifiers (with "%").
6775 * "systemctl show" gained a new --value switch, which allows print a
6776 only the contents of a specific unit property, without also printing
6777 the property's name. Similar support was added to "show*" verbs
6778 of loginctl and machinectl that output "key=value" lists.
6780 * A new unit type "generated" was added for files dynamically generated
6781 by generator tools. Similarly, a new unit type "transient" is used
6782 for unit files created using the runtime API. "systemctl enable" will
6783 refuse to operate on such files.
6785 * A new command "systemctl revert" has been added that may be used to
6786 revert to the vendor version of a unit file, in case local changes
6787 have been made by adding drop-ins or overriding the unit file.
6789 * "machinectl clean" gained a new verb to automatically remove all or
6790 just hidden container images.
6792 * systemd-tmpfiles gained support for a new line type "e" for emptying
6793 directories, if they exist, without creating them if they don't.
6795 * systemd-nspawn gained support for automatically patching the UID/GIDs
6796 of the owners and the ACLs of all files and directories in a
6797 container tree to match the UID/GID user namespacing range selected
6798 for the container invocation. This mode is enabled via the new
6799 --private-users-chown switch. It also gained support for
6800 automatically choosing a free, previously unused UID/GID range when
6801 starting a container, via the new --private-users=pick setting (which
6802 implies --private-users-chown). Together, these options for the first
6803 time make user namespacing for nspawn containers fully automatic and
6804 thus deployable. The systemd-nspawn@.service template unit file has
6805 been changed to use this functionality by default.
6807 * systemd-nspawn gained a new --network-zone= switch, that allows
6808 creating ad-hoc virtual Ethernet links between multiple containers,
6809 that only exist as long as at least one container referencing them is
6810 running. This allows easy connecting of multiple containers with a
6811 common link that implements an Ethernet broadcast domain. Each of
6812 these network "zones" may be named relatively freely by the user, and
6813 may be referenced by any number of containers, but each container may
6814 only reference one of these "zones". On the lower level, this is
6815 implemented by an automatically managed bridge network interface for
6816 each zone, that is created when the first container referencing its
6817 zone is created and removed when the last one referencing its zone
6820 * The default start timeout may now be configured on the kernel command
6821 line via systemd.default_timeout_start_sec=. It was already
6822 configurable via the DefaultTimeoutStartSec= option in
6823 /etc/systemd/system.conf.
6825 * Socket units gained a new TriggerLimitIntervalSec= and
6826 TriggerLimitBurst= setting to configure a limit on the activation
6827 rate of the socket unit.
6829 * The LimitNICE= setting now optionally takes normal UNIX nice values
6830 in addition to the raw integer limit value. If the specified
6831 parameter is prefixed with "+" or "-" and is in the range -20…19 the
6832 value is understood as UNIX nice value. If not prefixed like this it
6833 is understood as raw RLIMIT_NICE limit.
6835 * Note that the effect of the PrivateDevices= unit file setting changed
6836 slightly with this release: the per-device /dev file system will be
6837 mounted read-only from this version on, and will have "noexec"
6838 set. This (minor) change of behavior might cause some (exceptional)
6839 legacy software to break, when PrivateDevices=yes is set for its
6840 service. Please leave PrivateDevices= off if you run into problems
6843 * systemd-bootchart has been split out to a separate repository:
6844 https://github.com/systemd/systemd-bootchart
6846 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
6847 merged into the kernel in its current form.
6849 * The compatibility libraries libsystemd-daemon.so,
6850 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
6851 which have been deprecated since systemd-209 have been removed along
6852 with the corresponding pkg-config files. All symbols provided by
6853 those libraries are provided by libsystemd.so.
6855 * The Capabilities= unit file setting has been removed (it is ignored
6856 for backwards compatibility). AmbientCapabilities= and
6857 CapabilityBoundingSet= should be used instead.
6859 * A new special target has been added, initrd-root-device.target,
6860 which creates a synchronization point for dependencies of the root
6861 device in early userspace. Initramfs builders must ensure that this
6862 target is now included in early userspace.
6864 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
6865 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
6866 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
6867 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
6868 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
6869 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
6870 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
6871 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
6872 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
6873 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
6874 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
6875 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
6876 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
6877 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
6878 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
6879 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
6880 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
6881 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
6882 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
6883 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
6884 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
6885 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
6886 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
6887 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
6888 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
6891 — Fairfax, 2016-05-21
6895 * The systemd-resolved DNS resolver service has gained a substantial
6896 set of new features, most prominently it may now act as a DNSSEC
6897 validating stub resolver. DNSSEC mode is currently turned off by
6898 default, but is expected to be turned on by default in one of the
6899 next releases. For now, we invite everybody to test the DNSSEC logic
6900 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
6901 service also gained a full set of D-Bus interfaces, including calls
6902 to configure DNS and DNSSEC settings per link (for use by external
6903 network management software). systemd-resolved and systemd-networkd
6904 now distinguish between "search" and "routing" domains. The former
6905 are used to qualify single-label names, the latter are used purely
6906 for routing lookups within certain domains to specific links.
6907 resolved now also synthesizes RRs for all entries from /etc/hosts.
6909 * The systemd-resolve tool (which is a client utility for
6910 systemd-resolved) has been improved considerably and is now fully
6911 supported and documented. Hence it has moved from /usr/lib/systemd to
6914 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
6917 * The coredump collection logic has been reworked: when a coredump is
6918 collected it is now written to disk, compressed and processed
6919 (including stacktrace extraction) from a new instantiated service
6920 systemd-coredump@.service, instead of directly from the
6921 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
6922 processing large coredumps can take up a substantial amount of
6923 resources and time, and this previously happened entirely outside of
6924 systemd's service supervision. With the new logic the core_pattern
6925 hook only does minimal metadata collection before passing off control
6926 to the new instantiated service, which is configured with a time
6927 limit, a nice level and other settings to minimize negative impact on
6928 the rest of the system. Also note that the new logic will honour the
6929 RLIMIT_CORE setting of the crashed process, which now allows users
6930 and processes to turn off coredumping for their processes by setting
6933 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
6934 and all forked processes by default. Previously, PID 1 would leave
6935 the setting at "0" for all processes, as set by the kernel. Note that
6936 the resource limit traditionally has no effect on the generated
6937 coredumps on the system if the /proc/sys/kernel/core_pattern hook
6938 logic is used. Since the limit is now honoured (see above) its
6939 default has been changed so that the coredumping logic is enabled by
6940 default for all processes, while allowing specific opt-out.
6942 * When the stacktrace is extracted from processes of system users, this
6943 is now done as "systemd-coredump" user, in order to sandbox this
6944 potentially security sensitive parsing operation. (Note that when
6945 processing coredumps of normal users this is done under the user ID
6946 of process that crashed, as before.) Packagers should take notice
6947 that it is now necessary to create the "systemd-coredump" system user
6948 and group at package installation time.
6950 * The systemd-activate socket activation testing tool gained support
6951 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
6952 and --seqpacket switches. It also has been extended to support both
6953 new-style and inetd-style file descriptor passing. Use the new
6954 --inetd switch to request inetd-style file descriptor passing.
6956 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
6957 variable, which takes a boolean value. If set to false, ANSI color
6958 output is disabled in the tools even when run on a terminal that
6961 * The VXLAN support in networkd now supports two new settings
6962 DestinationPort= and PortRange=.
6964 * A new systemd.machine_id= kernel command line switch has been added,
6965 that may be used to set the machine ID in /etc/machine-id if it is
6966 not initialized yet. This command line option has no effect if the
6967 file is already initialized.
6969 * systemd-nspawn gained a new --as-pid2 switch that invokes any
6970 specified command line as PID 2 rather than PID 1 in the
6971 container. In this mode PID 1 is a minimal stub init process that
6972 implements the special POSIX and Linux semantics of PID 1 regarding
6973 signal and child process management. Note that this stub init process
6974 is implemented in nspawn itself and requires no support from the
6975 container image. This new logic is useful to support running
6976 arbitrary commands in the container, as normal processes are
6977 generally not prepared to run as PID 1.
6979 * systemd-nspawn gained a new --chdir= switch for setting the current
6980 working directory for the process started in the container.
6982 * "journalctl /dev/sda" will now output all kernel log messages for
6983 specified device from the current boot, in addition to all devices
6984 that are parents of it. This should make log output about devices
6985 pretty useful, as long as kernel drivers attach enough metadata to
6986 the log messages. (The usual SATA drivers do.)
6988 * The sd-journal API gained two new calls
6989 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
6990 that report whether log data from /run or /var has been found.
6992 * journalctl gained a new switch "--fields" that prints all journal
6993 record field names currently in use in the journal. This is backed
6994 by two new sd-journal API calls sd_journal_enumerate_fields() and
6995 sd_journal_restart_fields().
6997 * Most configurable timeouts in systemd now expect an argument of
6998 "infinity" to turn them off, instead of "0" as before. The semantics
6999 from now on is that a timeout of "0" means "now", and "infinity"
7000 means "never". To maintain backwards compatibility, "0" continues to
7001 turn off previously existing timeout settings.
7003 * "systemctl reload-or-try-restart" has been renamed to "systemctl
7004 try-reload-or-restart" to clarify what it actually does: the "try"
7005 logic applies to both reloading and restarting, not just restarting.
7006 The old name continues to be accepted for compatibility.
7008 * On boot-up, when PID 1 detects that the system clock is behind the
7009 release date of the systemd version in use, the clock is now set
7010 to the latter. Previously, this was already done in timesyncd, in order
7011 to avoid running with clocks set to the various clock epochs such as
7012 1902, 1938 or 1970. With this change the logic is now done in PID 1
7013 in addition to timesyncd during early boot-up, so that it is enforced
7014 before the first process is spawned by systemd. Note that the logic
7015 in timesyncd remains, as it is more comprehensive and ensures
7016 clock monotonicity by maintaining a persistent timestamp file in
7017 /var. Since /var is generally not available in earliest boot or the
7018 initrd, this part of the logic remains in timesyncd, and is not done
7021 * Support for tweaking details in net_cls.class_id through the
7022 NetClass= configuration directive has been removed, as the kernel
7023 people have decided to deprecate that controller in cgroup v2.
7024 Userspace tools such as nftables are moving over to setting rules
7025 that are specific to the full cgroup path of a task, which obsoletes
7026 these controllers anyway. The NetClass= directive is kept around for
7027 legacy compatibility reasons. For a more in-depth description of the
7028 kernel change, please refer to the respective upstream commit:
7030 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
7032 * A new service setting RuntimeMaxSec= has been added that may be used
7033 to specify a maximum runtime for a service. If the timeout is hit, the
7034 service is terminated and put into a failure state.
7036 * A new service setting AmbientCapabilities= has been added. It allows
7037 configuration of additional Linux process capabilities that are
7038 passed to the activated processes. This is only available on very
7041 * The process resource limit settings in service units may now be used
7042 to configure hard and soft limits individually.
7044 * The various libsystemd APIs such as sd-bus or sd-event now publicly
7045 expose support for gcc's __attribute__((cleanup())) C extension.
7046 Specifically, for many object destructor functions alternative
7047 versions have been added that have names suffixed with "p" and take a
7048 pointer to a pointer to the object to destroy, instead of just a
7049 pointer to the object itself. This is useful because these destructor
7050 functions may be used directly as parameters to the cleanup
7051 construct. Internally, systemd has been a heavy user of this GCC
7052 extension for a long time, and with this change similar support is
7053 now available to consumers of the library outside of systemd. Note
7054 that by using this extension in your sources compatibility with old
7055 and strictly ANSI compatible C compilers is lost. However, all gcc or
7056 LLVM versions of recent years support this extension.
7058 * Timer units gained support for a new setting RandomizedDelaySec= that
7059 allows configuring some additional randomized delay to the configured
7060 time. This is useful to spread out timer events to avoid load peaks in
7061 clusters or larger setups.
7063 * Calendar time specifications now support sub-second accuracy.
7065 * Socket units now support listening on SCTP and UDP-lite protocol
7068 * The sd-event API now comes with a full set of man pages.
7070 * Older versions of systemd contained experimental support for
7071 compressing journal files and coredumps with the LZ4 compressor that
7072 was not compatible with the lz4 binary (due to API limitations of the
7073 lz4 library). This support has been removed; only support for files
7074 compatible with the lz4 binary remains. This LZ4 logic is now
7075 officially supported and no longer considered experimental.
7077 * The dkr image import logic has been removed again from importd. dkr's
7078 micro-services focus doesn't fit into the machine image focus of
7079 importd, and quickly got out of date with the upstream dkr API.
7081 * Creation of the /run/lock/lockdev/ directory was dropped from
7082 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
7083 been available for many years. If you still need this, you need to
7084 create your own tmpfiles.d config file with:
7086 d /run/lock/lockdev 0775 root lock -
7088 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
7089 and RebootArgument= have been moved from the [Service] section of
7090 unit files to [Unit], and they are now supported on all unit types,
7091 not just service units. Of course, systemd will continue to
7092 understand these settings also at the old location, in order to
7093 maintain compatibility.
7095 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
7096 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
7097 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
7098 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
7099 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
7100 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
7101 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
7102 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
7103 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
7104 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
7105 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
7106 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
7107 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
7108 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
7109 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
7110 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
7111 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
7112 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
7113 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
7115 — Berlin, 2016-02-11
7119 * A number of properties previously only settable in unit
7120 files are now also available as properties to set when
7121 creating transient units programmatically via the bus, as it
7122 is exposed with systemd-run's --property=
7123 setting. Specifically, these are: SyslogIdentifier=,
7124 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
7125 EnvironmentFile=, ReadWriteDirectories=,
7126 ReadOnlyDirectories=, InaccessibleDirectories=,
7127 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
7129 * When creating transient services via the bus API it is now
7130 possible to pass in a set of file descriptors to use as
7131 STDIN/STDOUT/STDERR for the invoked process.
7133 * Slice units may now be created transiently via the bus APIs,
7134 similar to the way service and scope units may already be
7135 created transiently.
7137 * Wherever systemd expects a calendar timestamp specification
7138 (like in journalctl's --since= and --until= switches) UTC
7139 timestamps are now supported. Timestamps suffixed with "UTC"
7140 are now considered to be in Universal Time Coordinated
7141 instead of the local timezone. Also, timestamps may now
7142 optionally be specified with sub-second accuracy. Both of
7143 these additions also apply to recurring calendar event
7144 specification, such as OnCalendar= in timer units.
7146 * journalctl gained a new "--sync" switch that asks the
7147 journal daemon to write all so far unwritten log messages to
7148 disk and sync the files, before returning.
7150 * systemd-tmpfiles learned two new line types "q" and "Q" that
7151 operate like "v", but also set up a basic btrfs quota
7152 hierarchy when used on a btrfs file system with quota
7155 * tmpfiles' "v", "q" and "Q" will now create a plain directory
7156 instead of a subvolume (even on a btrfs file system) if the
7157 root directory is a plain directory, and not a
7158 subvolume. This should simplify things with certain chroot()
7159 environments which are not aware of the concept of btrfs
7162 * systemd-detect-virt gained a new --chroot switch to detect
7163 whether execution takes place in a chroot() environment.
7165 * CPUAffinity= now takes CPU index ranges in addition to
7168 * The various memory-related resource limit settings (such as
7169 LimitAS=) now understand the usual K, M, G, … suffixes to
7170 the base of 1024 (IEC). Similar, the time-related resource
7171 limit settings understand the usual min, h, day, … suffixes
7174 * There's a new system.conf setting DefaultTasksMax= to
7175 control the default TasksMax= setting for services and
7176 scopes running on the system. (TasksMax= is the primary
7177 setting that exposes the "pids" cgroup controller on systemd
7178 and was introduced in the previous systemd release.) The
7179 setting now defaults to 512, which means services that are
7180 not explicitly configured otherwise will only be able to
7181 create 512 processes or threads at maximum, from this
7182 version on. Note that this means that thread- or
7183 process-heavy services might need to be reconfigured to set
7184 TasksMax= to a higher value. It is sufficient to set
7185 TasksMax= in these specific unit files to a higher value, or
7186 even "infinity". Similar, there's now a logind.conf setting
7187 UserTasksMax= that defaults to 4096 and limits the total
7188 number of processes or tasks each user may own
7189 concurrently. nspawn containers also have the TasksMax=
7190 value set by default now, to 8192. Note that all of this
7191 only has an effect if the "pids" cgroup controller is
7192 enabled in the kernel. The general benefit of these changes
7193 should be a more robust and safer system, that provides a
7194 certain amount of per-service fork() bomb protection.
7196 * systemd-nspawn gained the new --network-veth-extra= switch
7197 to define additional and arbitrarily-named virtual Ethernet
7198 links between the host and the container.
7200 * A new service execution setting PassEnvironment= has been
7201 added that allows importing select environment variables
7202 from PID1's environment block into the environment block of
7205 * Timer units gained support for a new RemainAfterElapse=
7206 setting which takes a boolean argument. It defaults to on,
7207 exposing behaviour unchanged to previous releases. If set to
7208 off, timer units are unloaded after they elapsed if they
7209 cannot elapse again. This is particularly useful for
7210 transient timer units, which shall not stay around longer
7211 than until they first elapse.
7213 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
7214 default now (the kernel default is 16). This is beneficial
7215 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
7216 allows substantially larger numbers of queued
7217 datagrams. This should increase the capability of systemd to
7218 parallelize boot-up, as logging and sd_notify() are unlikely
7219 to stall execution anymore. If you need to change the value
7220 from the new defaults, use the usual sysctl.d/ snippets.
7222 * The compression framing format used by the journal or
7223 coredump processing has changed to be in line with what the
7224 official LZ4 tools generate. LZ4 compression support in
7225 systemd was considered unsupported previously, as the format
7226 was not compatible with the normal tools. With this release
7227 this has changed now, and it is hence safe for downstream
7228 distributions to turn it on. While not compressing as well
7229 as the XZ, LZ4 is substantially faster, which makes
7230 it a good default choice for the compression logic in the
7231 journal and in coredump handling.
7233 * Any reference to /etc/mtab has been dropped from
7234 systemd. The file has been obsolete since a while, but
7235 systemd refused to work on systems where it was incorrectly
7236 set up (it should be a symlink or non-existent). Please make
7237 sure to update to util-linux 2.27.1 or newer in conjunction
7238 with this systemd release, which also drops any reference to
7239 /etc/mtab. If you maintain a distribution make sure that no
7240 software you package still references it, as this is a
7241 likely source of bugs. There's also a glibc bug pending,
7242 asking for removal of any reference to this obsolete file:
7244 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
7246 Note that only util-linux versions built with
7247 --enable-libmount-force-mountinfo are supported.
7249 * Support for the ".snapshot" unit type has been removed. This
7250 feature turned out to be little useful and little used, and
7251 has now been removed from the core and from systemctl.
7253 * The dependency types RequiresOverridable= and
7254 RequisiteOverridable= have been removed from systemd. They
7255 have been used only very sparingly to our knowledge and
7256 other options that provide a similar effect (such as
7257 systemctl --mode=ignore-dependencies) are much more useful
7258 and commonly used. Moreover, they were only half-way
7259 implemented as the option to control behaviour regarding
7260 these dependencies was never added to systemctl. By removing
7261 these dependency types the execution engine becomes a bit
7262 simpler. Unit files that use these dependencies should be
7263 changed to use the non-Overridable dependency types
7264 instead. In fact, when parsing unit files with these
7265 options, that's what systemd will automatically convert them
7266 too, but it will also warn, asking users to fix the unit
7267 files accordingly. Removal of these dependency types should
7268 only affect a negligible number of unit files in the wild.
7270 * Behaviour of networkd's IPForward= option changed
7271 (again). It will no longer maintain a per-interface setting,
7272 but propagate one way from interfaces where this is enabled
7273 to the global kernel setting. The global setting will be
7274 enabled when requested by a network that is set up, but
7275 never be disabled again. This change was made to make sure
7276 IPv4 and IPv6 behaviour regarding packet forwarding is
7277 similar (as the Linux IPv6 stack does not support
7278 per-interface control of this setting) and to minimize
7281 * In unit files the behaviour of %u, %U, %h, %s has
7282 changed. These specifiers will now unconditionally resolve
7283 to the various user database fields of the user that the
7284 systemd instance is running as, instead of the user
7285 configured in the specific unit via User=. Note that this
7286 effectively doesn't change much, as resolving of these
7287 specifiers was already turned off in the --system instance
7288 of systemd, as we cannot do NSS lookups from PID 1. In the
7289 --user instance of systemd these specifiers where correctly
7290 resolved, but hardly made any sense, since the user instance
7291 lacks privileges to do user switches anyway, and User= is
7292 hence useless. Moreover, even in the --user instance of
7293 systemd behaviour was awkward as it would only take settings
7294 from User= assignment placed before the specifier into
7295 account. In order to unify and simplify the logic around
7296 this the specifiers will now always resolve to the
7297 credentials of the user invoking the manager (which in case
7298 of PID 1 is the root user).
7300 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
7301 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
7302 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
7303 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
7304 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
7305 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
7306 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
7307 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
7308 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
7309 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
7310 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
7311 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
7312 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
7313 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
7316 — Berlin, 2015-11-18
7320 * systemd now depends on util-linux v2.27. More specifically,
7321 the newly added mount monitor feature in libmount now
7322 replaces systemd's former own implementation.
7324 * libmount mandates /etc/mtab not to be regular file, and
7325 systemd now enforces this condition at early boot.
7326 /etc/mtab has been deprecated and warned about for a very
7327 long time, so systems running systemd should already have
7328 stopped having this file around as anything else than a
7329 symlink to /proc/self/mounts.
7331 * Support for the "pids" cgroup controller has been added. It
7332 allows accounting the number of tasks in a cgroup and
7333 enforcing limits on it. This adds two new setting
7334 TasksAccounting= and TasksMax= to each unit, as well as a
7335 global option DefaultTasksAccounting=.
7337 * Support for the "net_cls" cgroup controller has been added.
7338 It allows assigning a net class ID to each task in the
7339 cgroup, which can then be used in firewall rules and traffic
7340 shaping configurations. Note that the kernel netfilter net
7341 class code does not currently work reliably for ingress
7342 packets on unestablished sockets.
7344 This adds a new config directive called NetClass= to CGroup
7345 enabled units. Allowed values are positive numbers for fixed
7346 assignments and "auto" for picking a free value
7349 * 'systemctl is-system-running' now returns 'offline' if the
7350 system is not booted with systemd. This command can now be
7351 used as a substitute for 'systemd-notify --booted'.
7353 * Watchdog timeouts have been increased to 3 minutes for all
7354 in-tree service files. Apparently, disk IO issues are more
7355 frequent than we hoped, and user reported >1 minute waiting
7358 * 'machine-id-commit' functionality has been merged into
7359 'machine-id-setup --commit'. The separate binary has been
7362 * The WorkingDirectory= directive in unit files may now be set
7363 to the special value '~'. In this case, the working
7364 directory is set to the home directory of the user
7365 configured in User=.
7367 * "machinectl shell" will now open the shell in the home
7368 directory of the selected user by default.
7370 * The CrashChVT= configuration file setting is renamed to
7371 CrashChangeVT=, following our usual logic of not
7372 abbreviating unnecessarily. The old directive is still
7373 supported for compat reasons. Also, this directive now takes
7374 an integer value between 1 and 63, or a boolean value. The
7375 formerly supported '-1' value for disabling stays around for
7378 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
7379 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
7380 RootDirectory= properties can now be set for transient
7383 * The systemd-analyze tool gained a new "set-log-target" verb
7384 to change the logging target the system manager logs to
7385 dynamically during runtime. This is similar to how
7386 "systemd-analyze set-log-level" already changes the log
7389 * In nspawn /sys is now mounted as tmpfs, with only a selected
7390 set of subdirectories mounted in from the real sysfs. This
7391 enhances security slightly, and is useful for ensuring user
7392 namespaces work correctly.
7394 * Support for USB FunctionFS activation has been added. This
7395 allows implementation of USB gadget services that are
7396 activated as soon as they are requested, so that they don't
7397 have to run continuously, similar to classic socket
7400 * The "systemctl exit" command now optionally takes an
7401 additional parameter that sets the exit code to return from
7402 the systemd manager when exiting. This is only relevant when
7403 running the systemd user instance, or when running the
7404 system instance in a container.
7406 * sd-bus gained the new API calls sd_bus_path_encode_many()
7407 and sd_bus_path_decode_many() that allow easy encoding and
7408 decoding of multiple identifier strings inside a D-Bus
7409 object path. Another new call sd_bus_default_flush_close()
7410 has been added to flush and close per-thread default
7413 * systemd-cgtop gained support for a -M/--machine= switch to
7414 show the control groups within a certain container only.
7416 * "systemctl kill" gained support for an optional --fail
7417 switch. If specified the requested operation will fail of no
7418 processes have been killed, because the unit had no
7419 processes attached, or similar.
7421 * A new systemd.crash_reboot=1 kernel command line option has
7422 been added that triggers a reboot after crashing. This can
7423 also be set through CrashReboot= in systemd.conf.
7425 * The RuntimeDirectory= setting now understands unit
7426 specifiers like %i or %f.
7428 * A new (still internal) library API sd-ipv4acd has been added,
7429 that implements address conflict detection for IPv4. It's
7430 based on code from sd-ipv4ll, and will be useful for
7431 detecting DHCP address conflicts.
7433 * File descriptors passed during socket activation may now be
7434 named. A new API sd_listen_fds_with_names() is added to
7435 access the names. The default names may be overridden,
7436 either in the .socket file using the FileDescriptorName=
7437 parameter, or by passing FDNAME= when storing the file
7438 descriptors using sd_notify().
7440 * systemd-networkd gained support for:
7442 - Setting the IPv6 Router Advertisement settings via
7443 IPv6AcceptRouterAdvertisements= in .network files.
7445 - Configuring the HelloTimeSec=, MaxAgeSec= and
7446 ForwardDelaySec= bridge parameters in .netdev files.
7448 - Configuring PreferredSource= for static routes in
7451 * The "ask-password" framework used to query for LUKS harddisk
7452 passwords or SSL passwords during boot gained support for
7453 caching passwords in the kernel keyring, if it is
7454 available. This makes sure that the user only has to type in
7455 a passphrase once if there are multiple objects to unlock
7456 with the same one. Previously, such password caching was
7457 available only when Plymouth was used; this moves the
7458 caching logic into the systemd codebase itself. The
7459 "systemd-ask-password" utility gained a new --keyname=
7460 switch to control which kernel keyring key to use for
7461 caching a password in. This functionality is also useful for
7462 enabling display managers such as gdm to automatically
7463 unlock the user's GNOME keyring if its passphrase, the
7464 user's password and the harddisk password are the same, if
7465 gdm-autologin is used.
7467 * When downloading tar or raw images using "machinectl
7468 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
7469 file is now also downloaded, if it is available and stored
7470 next to the image file.
7472 * Units of type ".socket" gained a new boolean setting
7473 Writable= which is only useful in conjunction with
7474 ListenSpecial=. If true, enables opening the specified
7475 special file in O_RDWR mode rather than O_RDONLY mode.
7477 * systemd-rfkill has been reworked to become a singleton
7478 service that is activated through /dev/rfkill on each rfkill
7479 state change and saves the settings to disk. This way,
7480 systemd-rfkill is now compatible with devices that exist
7481 only intermittendly, and even restores state if the previous
7482 system shutdown was abrupt rather than clean.
7484 * The journal daemon gained support for vacuuming old journal
7485 files controlled by the number of files that shall remain,
7486 in addition to the already existing control by size and by
7487 date. This is useful as journal interleaving performance
7488 degrades with too many separate journal files, and allows
7489 putting an effective limit on them. The new setting defaults
7490 to 100, but this may be changed by setting SystemMaxFiles=
7491 and RuntimeMaxFiles= in journald.conf. Also, the
7492 "journalctl" tool gained the new --vacuum-files= switch to
7493 manually vacuum journal files to leave only the specified
7494 number of files in place.
7496 * udev will now create /dev/disk/by-path links for ATA devices
7497 on kernels where that is supported.
7499 * Galician, Serbian, Turkish and Korean translations were added.
7501 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
7502 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
7503 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
7504 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
7505 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
7506 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
7507 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
7508 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
7509 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
7510 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
7511 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
7512 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
7513 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
7514 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
7515 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
7516 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
7517 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
7518 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
7520 — Berlin, 2015-10-07
7524 * The DHCP implementation of systemd-networkd gained a set of
7527 - The DHCP server now supports emitting DNS and NTP
7528 information. It may be enabled and configured via
7529 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
7530 and NTP information is enabled, but no servers are
7531 configured, the corresponding uplink information (if there
7532 is any) is propagated.
7534 - Server and client now support transmission and reception
7535 of timezone information. It can be configured via the
7536 newly introduced network options UseTimezone=,
7537 EmitTimezone=, and Timezone=. Transmission of timezone
7538 information is enabled between host and containers by
7539 default now: the container will change its local timezone
7540 to what the host has set.
7542 - Lease timeouts can now be configured via
7543 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
7545 - The DHCP server improved on the stability of
7546 leases. Clients are more likely to get the same lease
7547 information back, even if the server loses state.
7549 - The DHCP server supports two new configuration options to
7550 control the lease address pool metrics, PoolOffset= and
7553 * The encapsulation limit of tunnels in systemd-networkd may
7554 now be configured via 'EncapsulationLimit='. It allows
7555 modifying the maximum additional levels of encapsulation
7556 that are permitted to be prepended to a packet.
7558 * systemd now supports the concept of user buses replacing
7559 session buses, if used with dbus-1.10 (and enabled via dbus
7560 --enable-user-session). It previously only supported this on
7561 kdbus-enabled systems, and this release expands this to
7562 'dbus-daemon' systems.
7564 * systemd-networkd now supports predictable interface names
7567 * systemd now optionally supports the new Linux kernel
7568 "unified" control group hierarchy. If enabled via the kernel
7569 command-line option 'systemd.unified_cgroup_hierarchy=1',
7570 systemd will try to mount the unified cgroup hierarchy
7571 directly on /sys/fs/cgroup. If not enabled, or not
7572 available, systemd will fall back to the legacy cgroup
7573 hierarchy setup, as before. Host system and containers can
7574 mix and match legacy and unified hierarchies as they
7575 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
7576 environment variable to individually select the hierarchy to
7577 use for executed containers. By default, nspawn will use the
7578 unified hierarchy for the containers if the host uses the
7579 unified hierarchy, and the legacy hierarchy otherwise.
7580 Please note that at this point the unified hierarchy is an
7581 experimental kernel feature and is likely to change in one
7582 of the next kernel releases. Therefore, it should not be
7583 enabled by default in downstream distributions yet. The
7584 minimum required kernel version for the unified hierarchy to
7585 work is 4.2. Note that when the unified hierarchy is used
7586 for the first time delegated access to controllers is
7587 safe. Because of this systemd-nspawn containers will get
7588 access to controllers now, as will systemd user
7589 sessions. This means containers and user sessions may now
7590 manage their own resources, partitioning up what the system
7593 * A new special scope unit "init.scope" has been introduced
7594 that encapsulates PID 1 of the system. It may be used to
7595 determine resource usage and enforce resource limits on PID
7596 1 itself. PID 1 hence moved out of the root of the control
7599 * The cgtop tool gained support for filtering out kernel
7600 threads when counting tasks in a control group. Also, the
7601 count of processes is now recursively summed up by
7602 default. Two options -k and --recursive= have been added to
7603 revert to old behaviour. The tool has also been updated to
7604 work correctly in containers now.
7606 * systemd-nspawn's --bind= and --bind-ro= options have been
7607 extended to allow creation of non-recursive bind mounts.
7609 * libsystemd gained two new calls sd_pid_get_cgroup() and
7610 sd_peer_get_cgroup() which return the control group path of
7611 a process or peer of a connected AF_UNIX socket. This
7612 function call is particularly useful when implementing
7613 delegated subtrees support in the control group hierarchy.
7615 * The "sd-event" event loop API of libsystemd now supports
7616 correct dequeuing of real-time signals, without losing
7619 * When systemd requests a polkit decision when managing units it
7620 will now add additional fields to the request, including unit
7621 name and desired operation. This enables more powerful polkit
7622 policies, that make decisions depending on these parameters.
7624 * nspawn learnt support for .nspawn settings files, that may
7625 accompany the image files or directories of containers, and
7626 may contain additional settings for the container. This is
7627 an alternative to configuring container parameters via the
7628 nspawn command line.
7630 Contributions from: Cristian Rodríguez, Daniel Mack, David
7631 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
7632 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
7633 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
7634 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
7635 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
7636 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
7637 Andersen, Tom Gundersen, Torstein Husebø
7639 — Berlin, 2015-09-08
7643 * machinectl gained a new verb 'shell' which opens a fresh
7644 shell on the target container or the host. It is similar to
7645 the existing 'login' command of machinectl, but spawns the
7646 shell directly without prompting for username or
7647 password. The pseudo machine '.host' now refers to the local
7648 host and is used by default. Hence, 'machinectl shell' can
7649 be used as replacement for 'su -' which spawns a session as
7650 a fresh systemd unit in a way that is fully isolated from
7651 the originating session.
7653 * systemd-networkd learned to cope with private-zone DHCP
7654 options and allows other programs to query the values.
7656 * SELinux access control when enabling/disabling units is no
7657 longer enforced with this release. The previous implementation
7658 was incorrect, and a new corrected implementation is not yet
7659 available. As unit file operations are still protected via
7660 polkit and D-Bus policy this is not a security problem. Yet,
7661 distributions which care about optimal SELinux support should
7662 probably not stabilize on this release.
7664 * sd-bus gained support for matches of type "arg0has=", that
7665 test for membership of strings in string arrays sent in bus
7668 * systemd-resolved now dumps the contents of its DNS and LLMNR
7669 caches to the logs on reception of the SIGUSR1 signal. This
7670 is useful to debug DNS behaviour.
7672 * The coredumpctl tool gained a new --directory= option to
7673 operate on journal files in a specific directory.
7675 * "systemctl reboot" and related commands gained a new
7676 "--message=" option which may be used to set a free-text
7677 wall message when shutting down or rebooting the
7678 system. This message is also logged, which is useful for
7679 figuring out the reason for a reboot or shutdown a
7682 * The "systemd-resolve-host" tool's -i switch now takes
7683 network interface numbers as alternative to interface names.
7685 * A new unit file setting for services has been introduced:
7686 UtmpMode= allows configuration of how precisely systemd
7687 handles utmp and wtmp entries for the service if this is
7688 enabled. This allows writing services that appear similar to
7689 user sessions in the output of the "w", "who", "last" and
7692 * systemd-resolved will now locally synthesize DNS resource
7693 records for the "localhost" and "gateway" domains as well as
7694 the local hostname. This should ensure that clients querying
7695 RRs via resolved will get similar results as those going via
7696 NSS, if nss-myhostname is enabled.
7698 Contributions from: Alastair Hughes, Alex Crawford, Daniel
7699 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
7700 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
7701 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
7702 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
7703 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
7704 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
7705 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
7706 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
7707 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
7708 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
7709 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
7711 — Berlin, 2015-08-27
7715 * The systemd-efi-boot-generator functionality was merged into
7716 systemd-gpt-auto-generator.
7718 * systemd-networkd now supports Group Policy for vxlan
7719 devices. It can be enabled via the new boolean configuration
7720 option called 'GroupPolicyExtension='.
7722 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
7723 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
7724 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
7726 — Berlin, 2015-07-31
7730 * The python-systemd code has been removed from the systemd repository.
7731 A new repository has been created which accommodates the code from
7732 now on, and we kindly ask distributions to create a separate package
7733 for this: https://github.com/systemd/python-systemd
7735 * The systemd daemon will now reload its main configuration
7736 (/etc/systemd/system.conf) on daemon-reload.
7738 * sd-dhcp now exposes vendor specific extensions via
7739 sd_dhcp_lease_get_vendor_specific().
7741 * systemd-networkd gained a number of new configuration options.
7743 - A new boolean configuration option for TAP devices called
7744 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
7745 device, thus allowing to send and receive GSO packets.
7747 - A new tunnel configuration option called 'CopyDSCP='.
7748 If enabled, the DSCP field of ip6 tunnels is copied into the
7749 decapsulated packet.
7751 - A set of boolean bridge configuration options were added.
7752 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
7753 and 'UnicastFlood=' are now parsed by networkd and applied to the
7754 respective bridge link device via the respective IFLA_BRPORT_*
7757 - A new string configuration option to override the hostname sent
7758 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
7759 is true, networkd will use the configured hostname instead of the
7760 system hostname when sending DHCP requests.
7762 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
7763 networkd will configure the IPv6 flow-label of the tunnel device
7764 according to RFC2460.
7766 - The 'macvtap' virtual network devices are now supported, similar to
7767 the already supported 'macvlan' devices.
7769 * systemd-resolved now implements RFC5452 to improve resilience against
7770 cache poisoning. Additionally, source port randomization is enabled
7771 by default to further protect against DNS spoofing attacks.
7773 * nss-mymachines now supports translating UIDs and GIDs of running
7774 containers with user-namespaces enabled. If a container 'foo'
7775 translates a host uid 'UID' to the container uid 'TUID', then
7776 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
7777 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
7778 mapped as 'vg-foo-TGID'.
7780 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
7781 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
7782 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
7783 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
7784 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
7785 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
7786 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
7787 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
7788 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
7789 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
7791 — Berlin, 2015-07-29
7795 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
7796 There are no known issues with current sysfs, and udev does not need
7797 or should be used to work around such bugs.
7799 * udev does no longer enable USB HID power management. Several reports
7800 indicate, that some devices cannot handle that setting.
7802 * The udev accelerometer helper was removed. The functionality
7803 is now fully included in iio-sensor-proxy. But this means,
7804 older iio-sensor-proxy versions will no longer provide
7805 accelerometer/orientation data with this systemd version.
7806 Please upgrade iio-sensor-proxy to version 1.0.
7808 * networkd gained a new configuration option IPv6PrivacyExtensions=
7809 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
7810 for Stateless Address") on selected networks.
7812 * For the sake of fewer build-time dependencies and less code in the
7813 main repository, the python bindings are about to be removed in the
7814 next release. A new repository has been created which accommodates
7815 the code from now on, and we kindly ask distributions to create a
7816 separate package for this. The removal will take place in v223.
7818 https://github.com/systemd/python-systemd
7820 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
7821 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
7822 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
7823 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
7824 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
7825 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
7826 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
7827 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
7828 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
7829 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
7831 — Berlin, 2015-07-07
7835 * The sd-bus.h and sd-event.h APIs have now been declared
7836 stable and have been added to the official interface of
7837 libsystemd.so. sd-bus implements an alternative D-Bus client
7838 library, that is relatively easy to use, very efficient and
7839 supports both classic D-Bus as well as kdbus as transport
7840 backend. sd-event is a generic event loop abstraction that
7841 is built around Linux epoll, but adds features such as event
7842 prioritization or efficient timer handling. Both APIs are good
7843 choices for C programs looking for a bus and/or event loop
7844 implementation that is minimal and does not have to be
7845 portable to other kernels.
7847 * kdbus support is no longer compile-time optional. It is now
7848 always built-in. However, it can still be disabled at
7849 runtime using the kdbus=0 kernel command line setting, and
7850 that setting may be changed to default to off, by specifying
7851 --disable-kdbus at build-time. Note though that the kernel
7852 command line setting has no effect if the kdbus.ko kernel
7853 module is not installed, in which case kdbus is (obviously)
7854 also disabled. We encourage all downstream distributions to
7855 begin testing kdbus by adding it to the kernel images in the
7856 development distributions, and leaving kdbus support in
7859 * The minimal required util-linux version has been bumped to
7862 * Support for chkconfig (--enable-chkconfig) was removed in
7863 favor of calling an abstraction tool
7864 /lib/systemd/systemd-sysv-install. This needs to be
7865 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
7866 in README for details.
7868 * If there's a systemd unit and a SysV init script for the
7869 same service name, and the user executes "systemctl enable"
7870 for it (or a related call), then this will now enable both
7871 (or execute the related operation on both), not just the
7874 * The libudev API documentation has been converted from gtkdoc
7877 * gudev has been removed from the systemd tree, it is now an
7880 * The systemd-cgtop tool learnt a new --raw switch to generate
7881 "raw" (machine parsable) output.
7883 * networkd's IPForwarding= .network file setting learnt the
7884 new setting "kernel", which ensures that networkd does not
7885 change the IP forwarding sysctl from the default kernel
7888 * The systemd-logind bus API now exposes a new boolean
7889 property "Docked" that reports whether logind considers the
7890 system "docked", i.e. connected to a docking station or not.
7892 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
7893 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
7894 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
7895 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
7896 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
7897 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
7898 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
7899 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
7900 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
7901 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
7902 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
7903 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
7904 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
7905 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
7906 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
7907 Fink, Zbigniew Jędrzejewski-Szmek
7909 — Berlin, 2015-06-19
7913 * The gudev library has been extracted into a separate repository
7914 available at: https://git.gnome.org/browse/libgudev/
7915 It is now managed as part of the Gnome project. Distributions
7916 are recommended to pass --disable-gudev to systemd and use
7917 gudev from the Gnome project instead. gudev is still included
7918 in systemd, for now. It will be removed soon, though. Please
7919 also see the announcement-thread on systemd-devel:
7920 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
7922 * systemd now exposes a CPUUsageNSec= property for each
7923 service unit on the bus, that contains the overall consumed
7924 CPU time of a service (the sum of what each process of the
7925 service consumed). This value is only available if
7926 CPUAccounting= is turned on for a service, and is then shown
7927 in the "systemctl status" output.
7929 * Support for configuring alternative mappings of the old SysV
7930 runlevels to systemd targets has been removed. They are now
7931 hardcoded in a way that runlevels 2, 3, 4 all map to
7932 multi-user.target and 5 to graphical.target (which
7933 previously was already the default behaviour).
7935 * The auto-mounter logic gained support for mount point
7936 expiry, using a new TimeoutIdleSec= setting in .automount
7937 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
7939 * The EFI System Partition (ESP) as mounted to /boot by
7940 systemd-efi-boot-generator will now be unmounted
7941 automatically after 2 minutes of not being used. This should
7942 minimize the risk of ESP corruptions.
7944 * New /etc/fstab options x-systemd.requires= and
7945 x-systemd.requires-mounts-for= are now supported to express
7946 additional dependencies for mounts. This is useful for
7947 journaling file systems that support external journal
7948 devices or overlay file systems that require underlying file
7949 systems to be mounted.
7951 * systemd does not support direct live-upgrades (via systemctl
7952 daemon-reexec) from versions older than v44 anymore. As no
7953 distribution we are aware of shipped such old versions in a
7954 stable release this should not be problematic.
7956 * When systemd forks off a new per-connection service instance
7957 it will now set the $REMOTE_ADDR environment variable to the
7958 remote IP address, and $REMOTE_PORT environment variable to
7959 the remote IP port. This behaviour is similar to the
7960 corresponding environment variables defined by CGI.
7962 * systemd-networkd gained support for uplink failure
7963 detection. The BindCarrier= option allows binding interface
7964 configuration dynamically to the link sense of other
7965 interfaces. This is useful to achieve behaviour like in
7968 * systemd-networkd gained support for configuring the DHCP
7969 client identifier to use when requesting leases.
7971 * systemd-networkd now has a per-network UseNTP= option to
7972 configure whether NTP server information acquired via DHCP
7973 is passed on to services like systemd-timesyncd.
7975 * systemd-networkd gained support for vti6 tunnels.
7977 * Note that systemd-networkd manages the sysctl variable
7978 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
7979 it is configured for since v219. The variable controls IP
7980 forwarding, and is a per-interface alternative to the global
7981 /proc/sys/net/ipv[46]/ip_forward. This setting is
7982 configurable in the IPForward= option, which defaults to
7983 "no". This means if networkd is used for an interface it is
7984 no longer sufficient to set the global sysctl option to turn
7985 on IP forwarding! Instead, the .network file option
7986 IPForward= needs to be turned on! Note that the
7987 implementation of this behaviour was broken in v219 and has
7990 * Many bonding and vxlan options are now configurable in
7993 * systemd-nspawn gained a new --property= setting to set unit
7994 properties for the container scope. This is useful for
7995 setting resource parameters (e.g. "CPUShares=500") on
7996 containers started from the command line.
7998 * systemd-nspawn gained a new --private-users= switch to make
7999 use of user namespacing available on recent Linux kernels.
8001 * systemd-nspawn may now be called as part of a shell pipeline
8002 in which case the pipes used for stdin and stdout are passed
8003 directly to the process invoked in the container, without
8004 indirection via a pseudo tty.
8006 * systemd-nspawn gained a new switch to control the UNIX
8007 signal to use when killing the init process of the container
8010 * systemd-nspawn gained a new --overlay= switch for mounting
8011 overlay file systems into the container using the new kernel
8014 * When a container image is imported via systemd-importd and
8015 the host file system is not btrfs, a loopback block device
8016 file is created in /var/lib/machines.raw with a btrfs file
8017 system inside. It is then mounted to /var/lib/machines to
8018 enable btrfs features for container management. The loopback
8019 file and btrfs file system is grown as needed when container
8020 images are imported via systemd-importd.
8022 * systemd-machined/systemd-importd gained support for btrfs
8023 quota, to enforce container disk space limits on disk. This
8024 is exposed in "machinectl set-limit".
8026 * systemd-importd now can import containers from local .tar,
8027 .raw and .qcow2 images, and export them to .tar and .raw. It
8028 can also import dkr v2 images now from the network (on top
8031 * systemd-importd gained support for verifying downloaded
8032 images with gpg2 (previously only gpg1 was supported).
8034 * systemd-machined, systemd-logind, systemd: most bus calls are
8035 now accessible to unprivileged processes via polkit. Also,
8036 systemd-logind will now allow users to kill their own sessions
8037 without further privileges or authorization.
8039 * systemd-shutdownd has been removed. This service was
8040 previously responsible for implementing scheduled shutdowns
8041 as exposed in /usr/bin/shutdown's time parameter. This
8042 functionality has now been moved into systemd-logind and is
8043 accessible via a bus interface.
8045 * "systemctl reboot" gained a new switch --firmware-setup that
8046 can be used to reboot into the EFI firmware setup, if that
8047 is available. systemd-logind now exposes an API on the bus
8048 to trigger such reboots, in case graphical desktop UIs want
8049 to cover this functionality.
8051 * "systemctl enable", "systemctl disable" and "systemctl mask"
8052 now support a new "--now" switch. If specified the units
8053 that are enabled will also be started, and the ones
8054 disabled/masked also stopped.
8056 * The Gummiboot EFI boot loader tool has been merged into
8057 systemd, and renamed to "systemd-boot". The bootctl tool has been
8058 updated to support systemd-boot.
8060 * An EFI kernel stub has been added that may be used to create
8061 kernel EFI binaries that contain not only the actual kernel,
8062 but also an initrd, boot splash, command line and OS release
8063 information. This combined binary can then be signed as a
8064 single image, so that the firmware can verify it all in one
8065 step. systemd-boot has special support for EFI binaries created
8066 like this and can extract OS release information from them
8067 and show them in the boot menu. This functionality is useful
8068 to implement cryptographically verified boot schemes.
8070 * Optional support has been added to systemd-fsck to pass
8071 fsck's progress report to an AF_UNIX socket in the file
8074 * udev will no longer create device symlinks for all block devices by
8075 default. A deny list for excluding special block devices from this
8076 logic has been turned into an allow list that requires picking block
8077 devices explicitly that require device symlinks.
8079 * A new (currently still internal) API sd-device.h has been
8080 added to libsystemd. This modernized API is supposed to
8081 replace libudev eventually. In fact, already much of libudev
8082 is now just a wrapper around sd-device.h.
8084 * A new hwdb database for storing metadata about pointing
8085 stick devices has been added.
8087 * systemd-tmpfiles gained support for setting file attributes
8088 similar to the "chattr" tool with new 'h' and 'H' lines.
8090 * systemd-journald will no longer unconditionally set the
8091 btrfs NOCOW flag on new journal files. This is instead done
8092 with tmpfiles snippet using the new 'h' line type. This
8093 allows easy disabling of this logic, by masking the
8094 journal-nocow.conf tmpfiles file.
8096 * systemd-journald will now translate audit message types to
8097 human readable identifiers when writing them to the
8098 journal. This should improve readability of audit messages.
8100 * The LUKS logic gained support for the offset= and skip=
8101 options in /etc/crypttab, as previously implemented by
8104 * /usr/lib/os-release gained a new optional field VARIANT= for
8105 distributions that support multiple variants (such as a
8106 desktop edition, a server edition, …)
8108 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
8109 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
8110 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
8111 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
8112 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
8113 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
8114 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
8115 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
8116 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
8117 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
8118 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
8119 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
8120 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
8121 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
8122 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
8123 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
8124 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
8125 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
8126 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
8127 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
8128 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
8129 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
8130 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
8131 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
8132 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
8133 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
8134 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
8136 — Berlin, 2015-05-22
8140 * Introduce a new API "sd-hwdb.h" for querying the hardware
8141 metadata database. With this minimal interface one can query
8142 and enumerate the udev hwdb, decoupled from the old libudev
8143 library. libudev's interface for this is now only a wrapper
8144 around sd-hwdb. A new tool systemd-hwdb has been added to
8145 interface with and update the database.
8147 * When any of systemd's tools copies files (for example due to
8148 tmpfiles' C lines) a btrfs reflink will attempted first,
8149 before bytewise copying is done.
8151 * systemd-nspawn gained a new --ephemeral switch. When
8152 specified a btrfs snapshot is taken of the container's root
8153 directory, and immediately removed when the container
8154 terminates again. Thus, a container can be started whose
8155 changes never alter the container's root directory, and are
8156 lost on container termination. This switch can also be used
8157 for starting a container off the root file system of the
8158 host without affecting the host OS. This switch is only
8159 available on btrfs file systems.
8161 * systemd-nspawn gained a new --template= switch. It takes the
8162 path to a container tree to use as template for the tree
8163 specified via --directory=, should that directory be
8164 missing. This allows instantiating containers dynamically,
8165 on first run. This switch is only available on btrfs file
8168 * When a .mount unit refers to a mount point on which multiple
8169 mounts are stacked, and the .mount unit is stopped all of
8170 the stacked mount points will now be unmounted until no
8171 mount point remains.
8173 * systemd now has an explicit notion of supported and
8174 unsupported unit types. Jobs enqueued for unsupported unit
8175 types will now fail with an "unsupported" error code. More
8176 specifically .swap, .automount and .device units are not
8177 supported in containers, .busname units are not supported on
8178 non-kdbus systems. .swap and .automount are also not
8179 supported if their respective kernel compile time options
8182 * machinectl gained support for two new "copy-from" and
8183 "copy-to" commands for copying files from a running
8184 container to the host or vice versa.
8186 * machinectl gained support for a new "bind" command to bind
8187 mount host directories into local containers. This is
8188 currently only supported for nspawn containers.
8190 * networkd gained support for configuring bridge forwarding
8191 database entries (fdb) from .network files.
8193 * A new tiny daemon "systemd-importd" has been added that can
8194 download container images in tar, raw, qcow2 or dkr formats,
8195 and make them available locally in /var/lib/machines, so
8196 that they can run as nspawn containers. The daemon can GPG
8197 verify the downloads (not supported for dkr, since it has no
8198 provisions for verifying downloads). It will transparently
8199 decompress bz2, xz, gzip compressed downloads if necessary,
8200 and restore sparse files on disk. The daemon uses privilege
8201 separation to ensure the actual download logic runs with
8202 fewer privileges than the daemon itself. machinectl has
8203 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
8204 make the functionality of importd available to the
8205 user. With this in place the Fedora and Ubuntu "Cloud"
8206 images can be downloaded and booted as containers unmodified
8207 (the Fedora images lack the appropriate GPG signature files
8208 currently, so they cannot be verified, but this will change
8209 soon, hopefully). Note that downloading images is currently
8210 only fully supported on btrfs.
8212 * machinectl is now able to list container images found in
8213 /var/lib/machines, along with some metadata about sizes of
8214 disk and similar. If the directory is located on btrfs and
8215 quota is enabled, this includes quota display. A new command
8216 "image-status" has been added that shows additional
8217 information about images.
8219 * machinectl is now able to clone container images
8220 efficiently, if the underlying file system (btrfs) supports
8221 it, with the new "machinectl clone" command. It also
8222 gained commands for renaming and removing images, as well as
8223 marking them read-only or read-write (supported also on
8224 legacy file systems).
8226 * networkd gained support for collecting LLDP network
8227 announcements, from hardware that supports this. This is
8228 shown in networkctl output.
8230 * systemd-run gained support for a new -t (--pty) switch for
8231 invoking a binary on a pty whose input and output is
8232 connected to the invoking terminal. This allows executing
8233 processes as system services while interactively
8234 communicating with them via the terminal. Most interestingly
8235 this is supported across container boundaries. Invoking
8236 "systemd-run -t /bin/bash" is an alternative to running a
8237 full login session, the difference being that the former
8238 will not register a session, nor go through the PAM session
8241 * tmpfiles gained support for a new "v" line type for creating
8242 btrfs subvolumes. If the underlying file system is a legacy
8243 file system, this automatically degrades to creating a
8244 normal directory. Among others /var/lib/machines is now
8245 created like this at boot, should it be missing.
8247 * The directory /var/lib/containers/ has been deprecated and
8248 been replaced by /var/lib/machines. The term "machines" has
8249 been used in the systemd context as generic term for both
8250 VMs and containers, and hence appears more appropriate for
8251 this, as the directory can also contain raw images bootable
8254 * systemd-nspawn when invoked with -M but without --directory=
8255 or --image= is now capable of searching for the container
8256 root directory, subvolume or disk image automatically, in
8257 /var/lib/machines. systemd-nspawn@.service has been updated
8258 to make use of this, thus allowing it to be used for raw
8261 * A new machines.target unit has been introduced that is
8262 supposed to group all containers/VMs invoked as services on
8263 the system. systemd-nspawn@.service has been updated to
8264 integrate with that.
8266 * machinectl gained a new "start" command, for invoking a
8267 container as a service. "machinectl start foo" is mostly
8268 equivalent to "systemctl start systemd-nspawn@foo.service",
8269 but handles escaping in a nicer way.
8271 * systemd-nspawn will now mount most of the cgroupfs tree
8272 read-only into each container, with the exception of the
8273 container's own subtree in the name=systemd hierarchy.
8275 * journald now sets the special FS_NOCOW file flag for its
8276 journal files. This should improve performance on btrfs, by
8277 avoiding heavy fragmentation when journald's write-pattern
8278 is used on COW file systems. It degrades btrfs' data
8279 integrity guarantees for the files to the same levels as for
8280 ext3/ext4 however. This should be OK though as journald does
8281 its own data integrity checks and all its objects are
8282 checksummed on disk. Also, journald should handle btrfs disk
8283 full events a lot more gracefully now, by processing SIGBUS
8284 errors, and not relying on fallocate() anymore.
8286 * When journald detects that journal files it is writing to
8287 have been deleted it will immediately start new journal
8290 * systemd now provides a way to store file descriptors
8291 per-service in PID 1. This is useful for daemons to ensure
8292 that fds they require are not lost during a daemon
8293 restart. The fds are passed to the daemon on the next
8294 invocation in the same way socket activation fds are
8295 passed. This is now used by journald to ensure that the
8296 various sockets connected to all the system's stdout/stderr
8297 are not lost when journald is restarted. File descriptors
8298 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
8299 an extension to sd_notify(). Note that a limit is enforced
8300 on the number of fds a service can store in PID 1, and it
8301 defaults to 0, so that no fds may be stored, unless this is
8302 explicitly turned on.
8304 * The default TERM variable to use for units connected to a
8305 terminal, when no other value is explicitly is set is now
8306 vt220 rather than vt102. This should be fairly safe still,
8307 but allows PgUp/PgDn work.
8309 * The /etc/crypttab option header= as known from Debian is now
8312 * "loginctl user-status" and "loginctl session-status" will
8313 now show the last 10 lines of log messages of the
8314 user/session following the status output. Similar,
8315 "machinectl status" will show the last 10 log lines
8316 associated with a virtual machine or container
8317 service. (Note that this is usually not the log messages
8318 done in the VM/container itself, but simply what the
8319 container manager logs. For nspawn this includes all console
8322 * "loginctl session-status" without further argument will now
8323 show the status of the session of the caller. Similar,
8324 "lock-session", "unlock-session", "activate",
8325 "enable-linger", "disable-linger" may now be called without
8326 session/user parameter in which case they apply to the
8327 caller's session/user.
8329 * An X11 session scriptlet is now shipped that uploads
8330 $DISPLAY and $XAUTHORITY into the environment of the systemd
8331 --user daemon if a session begins. This should improve
8332 compatibility with X11 enabled applications run as systemd
8335 * Generators are now subject to masking via /etc and /run, the
8336 same way as unit files.
8338 * networkd .network files gained support for configuring
8339 per-link IPv4/IPv6 packet forwarding as well as IPv4
8340 masquerading. This is by default turned on for veth links to
8341 containers, as registered by systemd-nspawn. This means that
8342 nspawn containers run with --network-veth will now get
8343 automatic routed access to the host's networks without any
8344 further configuration or setup, as long as networkd runs on
8347 * systemd-nspawn gained the --port= (-p) switch to expose TCP
8348 or UDP posts of a container on the host. With this in place
8349 it is possible to run containers with private veth links
8350 (--network-veth), and have their functionality exposed on
8351 the host as if their services were running directly on the
8354 * systemd-nspawn's --network-veth switch now gained a short
8355 version "-n", since with the changes above it is now truly
8356 useful out-of-the-box. The systemd-nspawn@.service has been
8357 updated to make use of it too by default.
8359 * systemd-nspawn will now maintain a per-image R/W lock, to
8360 ensure that the same image is not started more than once
8361 writable. (It's OK to run an image multiple times
8362 simultaneously in read-only mode.)
8364 * systemd-nspawn's --image= option is now capable of
8365 dissecting and booting MBR and GPT disk images that contain
8366 only a single active Linux partition. Previously it
8367 supported only GPT disk images with proper GPT type
8368 IDs. This allows running cloud images from major
8369 distributions directly with systemd-nspawn, without
8372 * In addition to collecting mouse dpi data in the udev
8373 hardware database, there's now support for collecting angle
8374 information for mouse scroll wheels. The database is
8375 supposed to guarantee similar scrolling behavior on mice
8376 that it knows about. There's also support for collecting
8377 information about Touchpad types.
8379 * udev's input_id built-in will now also collect touch screen
8380 dimension data and attach it to probed devices.
8382 * /etc/os-release gained support for a Distribution Privacy
8385 * networkd gained support for creating "ipvlan", "gretap",
8386 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
8388 * systemd-tmpfiles gained support for "a" lines for setting
8391 * systemd-nspawn will now mount /tmp in the container to
8392 tmpfs, automatically.
8394 * systemd now exposes the memory.usage_in_bytes cgroup
8395 attribute and shows it for each service in the "systemctl
8396 status" output, if available.
8398 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
8399 immediate reboot is triggered. This useful if shutdown is
8400 hung and is unable to complete, to expedite the
8401 operation. Note that this kind of reboot will still unmount
8402 all file systems, and hence should not result in fsck being
8405 * A .device unit for an optical block device will now be
8406 considered active only when a medium is in the drive. Also,
8407 mount units are now bound to their backing devices thus
8408 triggering automatic unmounting when devices become
8409 unavailable. With this in place systemd will now
8410 automatically unmount left-over mounts when a CD-ROM is
8411 ejected or an USB stick is yanked from the system.
8413 * networkd-wait-online now has support for waiting for
8414 specific interfaces only (with globbing), and for giving up
8415 after a configurable timeout.
8417 * networkd now exits when idle. It will be automatically
8418 restarted as soon as interfaces show up, are removed or
8419 change state. networkd will stay around as long as there is
8420 at least one DHCP state machine or similar around, that keep
8423 * networkd may now configure IPv6 link-local addressing in
8424 addition to IPv4 link-local addressing.
8426 * The IPv6 "token" for use in SLAAC may now be configured for
8427 each .network interface in networkd.
8429 * Routes configured with networkd may now be assigned a scope
8432 * networkd's [Match] sections now support globbing and lists
8433 of multiple space-separated matches per item.
8435 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
8436 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
8437 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
8438 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
8439 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
8440 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
8441 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
8442 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
8443 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
8444 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
8445 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
8446 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
8447 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
8448 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
8449 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
8450 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
8451 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
8452 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
8453 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
8454 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
8455 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
8456 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
8457 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
8458 Hoffmann, Zbigniew Jędrzejewski-Szmek
8460 — Berlin, 2015-02-16
8464 * When querying unit file enablement status (for example via
8465 "systemctl is-enabled"), a new state "indirect" is now known
8466 which indicates that a unit might not be enabled itself, but
8467 another unit listed in its Also= setting might be.
8469 * Similar to the various existing ConditionXYZ= settings for
8470 units, there are now matching AssertXYZ= settings. While
8471 failing conditions cause a unit to be skipped, but its job
8472 to succeed, failing assertions declared like this will cause
8473 a unit start operation and its job to fail.
8475 * hostnamed now knows a new chassis type "embedded".
8477 * systemctl gained a new "edit" command. When used on a unit
8478 file, this allows extending unit files with .d/ drop-in
8479 configuration snippets or editing the full file (after
8480 copying it from /usr/lib to /etc). This will invoke the
8481 user's editor (as configured with $EDITOR), and reload the
8482 modified configuration after editing.
8484 * "systemctl status" now shows the suggested enablement state
8485 for a unit, as declared in the (usually vendor-supplied)
8486 system preset files.
8488 * nss-myhostname will now resolve the single-label hostname
8489 "gateway" to the locally configured default IP routing
8490 gateways, ordered by their metrics. This assigns a stable
8491 name to the used gateways, regardless which ones are
8492 currently configured. Note that the name will only be
8493 resolved after all other name sources (if nss-myhostname is
8494 configured properly) and should hence not negatively impact
8495 systems that use the single-label hostname "gateway" in
8498 * systemd-inhibit now allows filtering by mode when listing
8501 * Scope and service units gained a new "Delegate" boolean
8502 property, which, when set, allows processes running inside the
8503 unit to further partition resources. This is primarily
8504 useful for systemd user instances as well as container
8507 * journald will now pick up audit messages directly from
8508 the kernel, and log them like any other log message. The
8509 audit fields are split up and fully indexed. This means that
8510 journalctl in many ways is now a (nicer!) alternative to
8511 ausearch, the traditional audit client. Note that this
8512 implements only a minimal audit client. If you want the
8513 special audit modes like reboot-on-log-overflow, please use
8514 the traditional auditd instead, which can be used in
8515 parallel to journald.
8517 * The ConditionSecurity= unit file option now understands the
8518 special string "audit" to check whether auditing is
8521 * journalctl gained two new commands --vacuum-size= and
8522 --vacuum-time= to delete old journal files until the
8523 remaining ones take up no more than the specified size on disk,
8524 or are not older than the specified time.
8526 * A new, native PPPoE library has been added to sd-network,
8527 systemd's library of light-weight networking protocols. This
8528 library will be used in a future version of networkd to
8529 enable PPPoE communication without an external pppd daemon.
8531 * The busctl tool now understands a new "capture" verb that
8532 works similar to "monitor", but writes a packet capture
8533 trace to STDOUT that can be redirected to a file which is
8534 compatible with libcap's capture file format. This can then
8535 be loaded in Wireshark and similar tools to inspect bus
8538 * The busctl tool now understands a new "tree" verb that shows
8539 the object trees of a specific service on the bus, or of all
8542 * The busctl tool now understands a new "introspect" verb that
8543 shows all interfaces and members of objects on the bus,
8544 including their signature and values. This is particularly
8545 useful to get more information about bus objects shown by
8546 the new "busctl tree" command.
8548 * The busctl tool now understands new verbs "call",
8549 "set-property" and "get-property" for invoking bus method
8550 calls, setting and getting bus object properties in a
8553 * busctl gained a new --augment-creds= argument that controls
8554 whether the tool shall augment credential information it
8555 gets from the bus with data from /proc, in a possibly
8558 * nspawn's --link-journal= switch gained two new values
8559 "try-guest" and "try-host" that work like "guest" and
8560 "host", but do not fail if the host has no persistent
8561 journaling enabled. -j is now equivalent to
8562 --link-journal=try-guest.
8564 * macvlan network devices created by nspawn will now have
8565 stable MAC addresses.
8567 * A new SmackProcessLabel= unit setting has been added, which
8568 controls the SMACK security label processes forked off by
8569 the respective unit shall use.
8571 * If compiled with --enable-xkbcommon, systemd-localed will
8572 verify x11 keymap settings by compiling the given keymap. It
8573 will spew out warnings if the compilation fails. This
8574 requires libxkbcommon to be installed.
8576 * When a coredump is collected, a larger number of metadata
8577 fields is now collected and included in the journal records
8578 created for it. More specifically, control group membership,
8579 environment variables, memory maps, working directory,
8580 chroot directory, /proc/$PID/status, and a list of open file
8581 descriptors is now stored in the log entry.
8583 * The udev hwdb now contains DPI information for mice. For
8586 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
8588 * All systemd programs that read standalone configuration
8589 files in /etc now also support a corresponding series of
8590 .conf.d configuration directories in /etc/, /run/,
8591 /usr/local/lib/, /usr/lib/, and (if configured with
8592 --enable-split-usr) /lib/. In particular, the following
8593 configuration files now have corresponding configuration
8594 directories: system.conf user.conf, logind.conf,
8595 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
8596 resolved.conf, timesyncd.conf, journal-remote.conf, and
8597 journal-upload.conf. Note that distributions should use the
8598 configuration directories in /usr/lib/; the directories in
8599 /etc/ are reserved for the system administrator.
8601 * systemd-rfkill will no longer take the rfkill device name
8602 into account when storing rfkill state on disk, as the name
8603 might be dynamically assigned and not stable. Instead, the
8604 ID_PATH udev variable combined with the rfkill type (wlan,
8605 bluetooth, …) is used.
8607 * A new service systemd-machine-id-commit.service has been
8608 added. When used on systems where /etc is read-only during
8609 boot, and /etc/machine-id is not initialized (but an empty
8610 file), this service will copy the temporary machine ID
8611 created as replacement into /etc after the system is fully
8612 booted up. This is useful for systems that are freshly
8613 installed with a non-initialized machine ID, but should get
8614 a fixed machine ID for subsequent boots.
8616 * networkd's .netdev files now provide a large set of
8617 configuration parameters for VXLAN devices. Similarly, the
8618 bridge port cost parameter is now configurable in .network
8619 files. There's also new support for configuring IP source
8620 routing. networkd .link files gained support for a new
8621 OriginalName= match that is useful to match against the
8622 original interface name the kernel assigned. .network files
8623 may include MTU= and MACAddress= fields for altering the MTU
8624 and MAC address while being connected to a specific network
8627 * The LUKS logic gained supported for configuring
8628 UUID-specific key files. There's also new support for naming
8629 LUKS device from the kernel command line, using the new
8630 luks.name= argument.
8632 * Timer units may now be transiently created via the bus API
8633 (this was previously already available for scope and service
8634 units). In addition it is now possible to create multiple
8635 transient units at the same time with a single bus call. The
8636 "systemd-run" tool has been updated to make use of this for
8637 running commands on a specified time, in at(1)-style.
8639 * tmpfiles gained support for "t" lines, for assigning
8640 extended attributes to files. Among other uses this may be
8641 used to assign SMACK labels to files.
8643 Contributions from: Alin Rauta, Alison Chaiken, Andrej
8644 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
8645 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
8646 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
8647 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
8648 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
8649 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
8650 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
8651 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
8652 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
8653 Michael Biebl, Michael Chapman, Michael Marineau, Michal
8654 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
8655 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
8656 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
8657 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
8658 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
8659 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
8660 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
8662 — Berlin, 2014-12-10
8666 * journalctl gained the new options -t/--identifier= to match
8667 on the syslog identifier (aka "tag"), as well as --utc to
8668 show log timestamps in the UTC timezone. journalctl now also
8669 accepts -n/--lines=all to disable line capping in a pager.
8671 * journalctl gained a new switch, --flush, that synchronously
8672 flushes logs from /run/log/journal to /var/log/journal if
8673 persistent storage is enabled. systemd-journal-flush.service
8674 now waits until the operation is complete.
8676 * Services can notify the manager before they start a reload
8677 (by sending RELOADING=1) or shutdown (by sending
8678 STOPPING=1). This allows the manager to track and show the
8679 internal state of daemons and closes a race condition when
8680 the process is still running but has closed its D-Bus
8683 * Services with Type=oneshot do not have to have any ExecStart
8686 * User units are now loaded also from
8687 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
8688 /run/systemd/user directory that was already previously
8689 supported, but is under the control of the user.
8691 * Job timeouts (i.e. timeouts on the time a job that is
8692 queued stays in the run queue) can now optionally result in
8693 immediate reboot or power-off actions (JobTimeoutAction= and
8694 JobTimeoutRebootArgument=). This is useful on ".target"
8695 units, to limit the maximum time a target remains
8696 undispatched in the run queue, and to trigger an emergency
8697 operation in such a case. This is now used by default to
8698 turn off the system if boot-up (as defined by everything in
8699 basic.target) hangs and does not complete for at least
8700 15min. Also, if power-off or reboot hang for at least 30min
8701 an immediate power-off/reboot operation is triggered. This
8702 functionality is particularly useful to increase reliability
8703 on embedded devices, but also on laptops which might
8704 accidentally get powered on when carried in a backpack and
8705 whose boot stays stuck in a hard disk encryption passphrase
8708 * systemd-logind can be configured to also handle lid switch
8709 events even when the machine is docked or multiple displays
8710 are attached (HandleLidSwitchDocked= option).
8712 * A helper binary and a service have been added which can be
8713 used to resume from hibernation in the initramfs. A
8714 generator will parse the resume= option on the kernel
8715 command line to trigger resume.
8717 * A user console daemon systemd-consoled has been
8718 added. Currently, it is a preview, and will so far open a
8719 single terminal on each session of the user marked as
8720 Desktop=systemd-console.
8722 * Route metrics can be specified for DHCP routes added by
8725 * The SELinux context of socket-activated services can be set
8726 from the information provided by the networking stack
8727 (SELinuxContextFromNet= option).
8729 * Userspace firmware loading support has been removed and
8730 the minimum supported kernel version is thus bumped to 3.7.
8732 * Timeout for udev workers has been increased from 1 to 3
8733 minutes, but a warning will be printed after 1 minute to
8734 help diagnose kernel modules that take a long time to load.
8736 * Udev rules can now remove tags on devices with TAG-="foobar".
8738 * systemd's readahead implementation has been removed. In many
8739 circumstances it didn't give expected benefits even for
8740 rotational disk drives and was becoming less relevant in the
8741 age of SSDs. As none of the developers has been using
8742 rotating media anymore, and nobody stepped up to actively
8743 maintain this component of systemd it has now been removed.
8745 * Swap units can use Options= to specify discard options.
8746 Discard options specified for swaps in /etc/fstab are now
8749 * Docker containers are now detected as a separate type of
8752 * The Password Agent protocol gained support for queries where
8753 the user input is shown, useful e.g. for user names.
8754 systemd-ask-password gained a new --echo option to turn that
8757 * The default sysctl.d/ snippets will now set:
8759 net.core.default_qdisc = fq_codel
8761 This selects Fair Queuing Controlled Delay as the default
8762 queuing discipline for network interfaces. fq_codel helps
8763 fight the network bufferbloat problem. It is believed to be
8764 a good default with no tuning required for most workloads.
8765 Downstream distributions may override this choice. On 10Gbit
8766 servers that do not do forwarding, "fq" may perform better.
8767 Systems without a good clocksource should use "pfifo_fast".
8769 * If kdbus is enabled during build a new option BusPolicy= is
8770 available for service units, that allows locking all service
8771 processes into a stricter bus policy, in order to limit
8772 access to various bus services, or even hide most of them
8773 from the service's view entirely.
8775 * networkctl will now show the .network and .link file
8776 networkd has applied to a specific interface.
8778 * sd-login gained a new API call sd_session_get_desktop() to
8779 query which desktop environment has been selected for a
8782 * UNIX utmp support is now compile-time optional to support
8783 legacy-free systems.
8785 * systemctl gained two new commands "add-wants" and
8786 "add-requires" for pulling in units from specific targets
8789 * If the word "rescue" is specified on the kernel command line
8790 the system will now boot into rescue mode (aka
8791 rescue.target), which was previously available only by
8792 specifying "1" or "systemd.unit=rescue.target" on the kernel
8793 command line. This new kernel command line option nicely
8794 mirrors the already existing "emergency" kernel command line
8797 * New kernel command line options mount.usr=, mount.usrflags=,
8798 mount.usrfstype= have been added that match root=, rootflags=,
8799 rootfstype= but allow mounting a specific file system to
8802 * The $NOTIFY_SOCKET is now also passed to control processes of
8803 services, not only the main process.
8805 * This version reenables support for fsck's -l switch. This
8806 means at least version v2.25 of util-linux is required for
8807 operation, otherwise dead-locks on device nodes may
8808 occur. Again: you need to update util-linux to at least
8809 v2.25 when updating systemd to v217.
8811 * The "multi-seat-x" tool has been removed from systemd, as
8812 its functionality has been integrated into X servers 1.16,
8813 and the tool is hence redundant. It is recommended to update
8814 display managers invoking this tool to simply invoke X
8815 directly from now on, again.
8817 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
8818 message flag has been added for all of systemd's polkit
8819 authenticated method calls has been added. In particular this
8820 now allows optional interactive authorization via polkit for
8821 many of PID1's privileged operations such as unit file
8822 enabling and disabling.
8824 * "udevadm hwdb --update" learnt a new switch "--usr" for
8825 placing the rebuilt hardware database in /usr instead of
8826 /etc. When used only hardware database entries stored in
8827 /usr will be used, and any user database entries in /etc are
8828 ignored. This functionality is useful for vendors to ship a
8829 pre-built database on systems where local configuration is
8830 unnecessary or unlikely.
8832 * Calendar time specifications in .timer units now also
8833 understand the strings "semi-annually", "quarterly" and
8834 "minutely" as shortcuts (in addition to the preexisting
8835 "annually", "hourly", …).
8837 * systemd-tmpfiles will now correctly create files in /dev
8838 at boot which are marked for creation only at boot. It is
8839 recommended to always create static device nodes with 'c!'
8840 and 'b!', so that they are created only at boot and not
8841 overwritten at runtime.
8843 * When the watchdog logic is used for a service (WatchdogSec=)
8844 and the watchdog timeout is hit the service will now be
8845 terminated with SIGABRT (instead of just SIGTERM), in order
8846 to make sure a proper coredump and backtrace is
8847 generated. This ensures that hanging services will result in
8848 similar coredump/backtrace behaviour as services that hit a
8851 Contributions from: Andreas Henriksson, Andrei Borzenkov,
8852 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
8853 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
8854 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
8855 Herrmann, David Sommerseth, David Strauss, Emil Renner
8856 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
8857 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
8858 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
8859 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
8860 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
8861 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
8862 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
8863 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
8864 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
8865 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
8866 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
8867 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
8868 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
8869 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
8870 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
8871 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
8874 — Berlin, 2014-10-28
8878 * timedated no longer reads NTP implementation unit names from
8879 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
8880 implementations should add a
8882 Conflicts=systemd-timesyncd.service
8884 to their unit files to take over and replace systemd's NTP
8885 default functionality.
8887 * systemd-sysusers gained a new line type "r" for configuring
8888 which UID/GID ranges to allocate system users/groups
8889 from. Lines of type "u" may now add an additional column
8890 that specifies the home directory for the system user to be
8891 created. Also, systemd-sysusers may now optionally read user
8892 information from STDIN instead of a file. This is useful for
8893 invoking it from RPM preinst scriptlets that need to create
8894 users before the first RPM file is installed since these
8895 files might need to be owned by them. A new
8896 %sysusers_create_inline RPM macro has been introduced to do
8897 just that. systemd-sysusers now updates the shadow files as
8898 well as the user/group databases, which should enhance
8899 compatibility with certain tools like grpck.
8901 * A number of bus APIs of PID 1 now optionally consult polkit to
8902 permit access for otherwise unprivileged clients under certain
8903 conditions. Note that this currently doesn't support
8904 interactive authentication yet, but this is expected to be
8905 added eventually, too.
8907 * /etc/machine-info now has new fields for configuring the
8908 deployment environment of the machine, as well as the
8909 location of the machine. hostnamectl has been updated with
8910 new command to update these fields.
8912 * systemd-timesyncd has been updated to automatically acquire
8913 NTP server information from systemd-networkd, which might
8914 have been discovered via DHCP.
8916 * systemd-resolved now includes a caching DNS stub resolver
8917 and a complete LLMNR name resolution implementation. A new
8918 NSS module "nss-resolve" has been added which can be used
8919 instead of glibc's own "nss-dns" to resolve hostnames via
8920 systemd-resolved. Hostnames, addresses and arbitrary RRs may
8921 be resolved via systemd-resolved D-Bus APIs. In contrast to
8922 the glibc internal resolver systemd-resolved is aware of
8923 multi-homed system, and keeps DNS server and caches separate
8924 and per-interface. Queries are sent simultaneously on all
8925 interfaces that have DNS servers configured, in order to
8926 properly handle VPNs and local LANs which might resolve
8927 separate sets of domain names. systemd-resolved may acquire
8928 DNS server information from systemd-networkd automatically,
8929 which in turn might have discovered them via DHCP. A tool
8930 "systemd-resolve-host" has been added that may be used to
8931 query the DNS logic in resolved. systemd-resolved implements
8932 IDNA and automatically uses IDNA or UTF-8 encoding depending
8933 on whether classic DNS or LLMNR is used as transport. In the
8934 next releases we intend to add a DNSSEC and mDNS/DNS-SD
8935 implementation to systemd-resolved.
8937 * A new NSS module nss-mymachines has been added, that
8938 automatically resolves the names of all local registered
8939 containers to their respective IP addresses.
8941 * A new client tool "networkctl" for systemd-networkd has been
8942 added. It currently is entirely passive and will query
8943 networking configuration from udev, rtnetlink and networkd,
8944 and present it to the user in a very friendly
8945 way. Eventually, we hope to extend it to become a full
8946 control utility for networkd.
8948 * .socket units gained a new DeferAcceptSec= setting that
8949 controls the kernels' TCP_DEFER_ACCEPT sockopt for
8950 TCP. Similarly, support for controlling TCP keep-alive
8951 settings has been added (KeepAliveTimeSec=,
8952 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
8953 turning off Nagle's algorithm on TCP has been added
8956 * logind learned a new session type "web", for use in projects
8957 like Cockpit which register web clients as PAM sessions.
8959 * timer units with at least one OnCalendar= setting will now
8960 be started only after time-sync.target has been
8961 reached. This way they will not elapse before the system
8962 clock has been corrected by a local NTP client or
8963 similar. This is particular useful on RTC-less embedded
8964 machines, that come up with an invalid system clock.
8966 * systemd-nspawn's --network-veth= switch should now result in
8967 stable MAC addresses for both the outer and the inner side
8970 * systemd-nspawn gained a new --volatile= switch for running
8971 container instances with /etc or /var unpopulated.
8973 * The kdbus client code has been updated to use the new Linux
8974 3.17 memfd subsystem instead of the old kdbus-specific one.
8976 * systemd-networkd's DHCP client and server now support
8977 FORCERENEW. There are also new configuration options to
8978 configure the vendor client identifier and broadcast mode
8981 * systemd will no longer inform the kernel about the current
8982 timezone, as this is necessarily incorrect and racy as the
8983 kernel has no understanding of DST and similar
8984 concepts. This hence means FAT timestamps will be always
8985 considered UTC, similar to what Android is already
8986 doing. Also, when the RTC is configured to the local time
8987 (rather than UTC) systemd will never synchronize back to it,
8988 as this might confuse Windows at a later boot.
8990 * systemd-analyze gained a new command "verify" for offline
8991 validation of unit files.
8993 * systemd-networkd gained support for a couple of additional
8994 settings for bonding networking setups. Also, the metric for
8995 statically configured routes may now be configured. For
8996 network interfaces where this is appropriate the peer IP
8997 address may now be configured.
8999 * systemd-networkd's DHCP client will no longer request
9000 broadcasting by default, as this tripped up some networks.
9001 For hardware where broadcast is required the feature should
9002 be switched back on using RequestBroadcast=yes.
9004 * systemd-networkd will now set up IPv4LL addresses (when
9005 enabled) even if DHCP is configured successfully.
9007 * udev will now default to respect network device names given
9008 by the kernel when the kernel indicates that these are
9009 predictable. This behavior can be tweaked by changing
9010 NamePolicy= in the relevant .link file.
9012 * A new library systemd-terminal has been added that
9013 implements full TTY stream parsing and rendering. This
9014 library is supposed to be used later on for implementing a
9015 full userspace VT subsystem, replacing the current kernel
9018 * A new tool systemd-journal-upload has been added to push
9019 journal data to a remote system running
9020 systemd-journal-remote.
9022 * journald will no longer forward all local data to another
9023 running syslog daemon. This change has been made because
9024 rsyslog (which appears to be the most commonly used syslog
9025 implementation these days) no longer makes use of this, and
9026 instead pulls the data out of the journal on its own. Since
9027 forwarding the messages to a non-existent syslog server is
9028 more expensive than we assumed we have now turned this
9029 off. If you run a syslog server that is not a recent rsyslog
9030 version, you have to turn this option on again
9031 (ForwardToSyslog= in journald.conf).
9033 * journald now optionally supports the LZ4 compressor for
9034 larger journal fields. This compressor should perform much
9035 better than XZ which was the previous default.
9037 * machinectl now shows the IP addresses of local containers,
9038 if it knows them, plus the interface name of the container.
9040 * A new tool "systemd-escape" has been added that makes it
9041 easy to escape strings to build unit names and similar.
9043 * sd_notify() messages may now include a new ERRNO= field
9044 which is parsed and collected by systemd and shown among the
9045 "systemctl status" output for a service.
9047 * A new component "systemd-firstboot" has been added that
9048 queries the most basic systemd information (timezone,
9049 hostname, root password) interactively on first
9050 boot. Alternatively it may also be used to provision these
9051 things offline on OS images installed into directories.
9053 * The default sysctl.d/ snippets will now set
9055 net.ipv4.conf.default.promote_secondaries=1
9057 This has the benefit of no flushing secondary IP addresses
9058 when primary addresses are removed.
9060 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
9061 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
9062 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
9063 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
9064 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
9065 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
9066 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
9067 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
9068 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
9069 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
9070 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
9071 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
9072 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
9073 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
9074 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
9076 — Berlin, 2014-08-19
9080 * A new tool systemd-sysusers has been added. This tool
9081 creates system users and groups in /etc/passwd and
9082 /etc/group, based on static declarative system user/group
9083 definitions in /usr/lib/sysusers.d/. This is useful to
9084 enable factory resets and volatile systems that boot up with
9085 an empty /etc directory, and thus need system users and
9086 groups created during early boot. systemd now also ships
9087 with two default sysusers.d/ files for the most basic
9088 users and groups systemd and the core operating system
9091 * A new tmpfiles snippet has been added that rebuilds the
9092 essential files in /etc on boot, should they be missing.
9094 * A directive for ensuring automatic clean-up of
9095 /var/cache/man/ has been removed from the default
9096 configuration. This line should now be shipped by the man
9097 implementation. The necessary change has been made to the
9098 man-db implementation. Note that you need to update your man
9099 implementation to one that ships this line, otherwise no
9100 automatic clean-up of /var/cache/man will take place.
9102 * A new condition ConditionNeedsUpdate= has been added that
9103 may conditionalize services to only run when /etc or /var
9104 are "older" than the vendor operating system resources in
9105 /usr. This is useful for reconstructing or updating /etc
9106 after an offline update of /usr or a factory reset, on the
9107 next reboot. Services that want to run once after such an
9108 update or reset should use this condition and order
9109 themselves before the new systemd-update-done.service, which
9110 will mark the two directories as fully updated. A number of
9111 service files have been added making use of this, to rebuild
9112 the udev hardware database, the journald message catalog and
9113 dynamic loader cache (ldconfig). The systemd-sysusers tool
9114 described above also makes use of this now. With this in
9115 place it is now possible to start up a minimal operating
9116 system with /etc empty cleanly. For more information on the
9117 concepts involved see this recent blog story:
9119 http://0pointer.de/blog/projects/stateless.html
9121 * A new system group "input" has been introduced, and all
9122 input device nodes get this group assigned. This is useful
9123 for system-level software to get access to input devices. It
9124 complements what is already done for "audio" and "video".
9126 * systemd-networkd learnt minimal DHCPv4 server support in
9127 addition to the existing DHCPv4 client support. It also
9128 learnt DHCPv6 client and IPv6 Router Solicitation client
9129 support. The DHCPv4 client gained support for static routes
9130 passed in from the server. Note that the [DHCPv4] section
9131 known in older systemd-networkd versions has been renamed to
9132 [DHCP] and is now also used by the DHCPv6 client. Existing
9133 .network files using settings of this section should be
9134 updated, though compatibility is maintained. Optionally, the
9135 client hostname may now be sent to the DHCP server.
9137 * networkd gained support for vxlan virtual networks as well
9138 as tun/tap and dummy devices.
9140 * networkd gained support for automatic allocation of address
9141 ranges for interfaces from a system-wide pool of
9142 addresses. This is useful for dynamically managing a large
9143 number of interfaces with a single network configuration
9144 file. In particular this is useful to easily assign
9145 appropriate IP addresses to the veth links of a large number
9146 of nspawn instances.
9148 * RPM macros for processing sysusers, sysctl and binfmt
9149 drop-in snippets at package installation time have been
9152 * The /etc/os-release file should now be placed in
9153 /usr/lib/os-release. The old location is automatically
9154 created as symlink. /usr/lib is the more appropriate
9155 location of this file, since it shall actually describe the
9156 vendor operating system shipped in /usr, and not the
9157 configuration stored in /etc.
9159 * .mount units gained a new boolean SloppyOptions= setting
9160 that maps to mount(8)'s -s option which enables permissive
9161 parsing of unknown mount options.
9163 * tmpfiles learnt a new "L+" directive which creates a symlink
9164 but (unlike "L") deletes a pre-existing file first, should
9165 it already exist and not already be the correct
9166 symlink. Similarly, "b+", "c+" and "p+" directives have been
9167 added as well, which create block and character devices, as
9168 well as fifos in the filesystem, possibly removing any
9169 pre-existing files of different types.
9171 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
9172 'argument' field (which so far specified the source to
9173 symlink/copy the files from) is now optional. If omitted the
9174 same file os copied from /usr/share/factory/ suffixed by the
9175 full destination path. This is useful for populating /etc
9176 with essential files, by copying them from vendor defaults
9177 shipped in /usr/share/factory/etc.
9179 * A new command "systemctl preset-all" has been added that
9180 applies the service preset settings to all installed unit
9181 files. A new switch --preset-mode= has been added that
9182 controls whether only enable or only disable operations
9185 * A new command "systemctl is-system-running" has been added
9186 that allows checking the overall state of the system, for
9187 example whether it is fully up and running.
9189 * When the system boots up with an empty /etc, the equivalent
9190 to "systemctl preset-all" is executed during early boot, to
9191 make sure all default services are enabled after a factory
9194 * systemd now contains a minimal preset file that enables the
9195 most basic services systemd ships by default.
9197 * Unit files' [Install] section gained a new DefaultInstance=
9198 field for defining the default instance to create if a
9199 template unit is enabled with no instance specified.
9201 * A new passive target cryptsetup-pre.target has been added
9202 that may be used by services that need to make they run and
9203 finish before the first LUKS cryptographic device is set up.
9205 * The /dev/loop-control and /dev/btrfs-control device nodes
9206 are now owned by the "disk" group by default, opening up
9207 access to this group.
9209 * systemd-coredump will now automatically generate a
9210 stack trace of all core dumps taking place on the system,
9211 based on elfutils' libdw library. This stack trace is logged
9214 * systemd-coredump may now optionally store coredumps directly
9215 on disk (in /var/lib/systemd/coredump, possibly compressed),
9216 instead of storing them unconditionally in the journal. This
9217 mode is the new default. A new configuration file
9218 /etc/systemd/coredump.conf has been added to configure this
9219 and other parameters of systemd-coredump.
9221 * coredumpctl gained a new "info" verb to show details about a
9222 specific coredump. A new switch "-1" has also been added
9223 that makes sure to only show information about the most
9224 recent entry instead of all entries. Also, as the tool is
9225 generally useful now the "systemd-" prefix of the binary
9226 name has been removed. Distributions that want to maintain
9227 compatibility with the old name should add a symlink from
9228 the old name to the new name.
9230 * journald's SplitMode= now defaults to "uid". This makes sure
9231 that unprivileged users can access their own coredumps with
9232 coredumpctl without restrictions.
9234 * New kernel command line options "systemd.wants=" (for
9235 pulling an additional unit during boot), "systemd.mask="
9236 (for masking a specific unit for the boot), and
9237 "systemd.debug-shell" (for enabling the debug shell on tty9)
9238 have been added. This is implemented in the new generator
9239 "systemd-debug-generator".
9241 * systemd-nspawn will now by default filter a couple of
9242 syscalls for containers, among them those required for
9243 kernel module loading, direct x86 IO port access, swap
9244 management, and kexec. Most importantly though
9245 open_by_handle_at() is now prohibited for containers,
9246 closing a hole similar to a recently discussed vulnerability
9247 in docker regarding access to files on file hierarchies the
9248 container should normally not have access to. Note that, for
9249 nspawn, we generally make no security claims anyway (and
9250 this is explicitly documented in the man page), so this is
9251 just a fix for one of the most obvious problems.
9253 * A new man page file-hierarchy(7) has been added that
9254 contains a minimized, modernized version of the file system
9255 layout systemd expects, similar in style to the FHS
9256 specification or hier(5). A new tool systemd-path(1) has
9257 been added to query many of these paths for the local
9260 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
9261 longer done. Since the directory now has a per-user size
9262 limit, and is cleaned on logout this appears unnecessary,
9263 in particular since this now brings the lifecycle of this
9264 directory closer in line with how IPC objects are handled.
9266 * systemd.pc now exports a number of additional directories,
9267 including $libdir (which is useful to identify the library
9268 path for the primary architecture of the system), and a
9269 couple of drop-in directories.
9271 * udev's predictable network interface names now use the dev_port
9272 sysfs attribute, introduced in linux 3.15 instead of dev_id to
9273 distinguish between ports of the same PCI function. dev_id should
9274 only be used for ports using the same HW address, hence the need
9277 * machined has been updated to export the OS version of a
9278 container (read from /etc/os-release and
9279 /usr/lib/os-release) on the bus. This is now shown in
9280 "machinectl status" for a machine.
9282 * A new service setting RestartForceExitStatus= has been
9283 added. If configured to a set of exit signals or process
9284 return values, the service will be restarted when the main
9285 daemon process exits with any of them, regardless of the
9288 * systemctl's -H switch for connecting to remote systemd
9289 machines has been extended so that it may be used to
9290 directly connect to a specific container on the
9291 host. "systemctl -H root@foobar:waldi" will now connect as
9292 user "root" to host "foobar", and then proceed directly to
9293 the container named "waldi". Note that currently you have to
9294 authenticate as user "root" for this to work, as entering
9295 containers is a privileged operation.
9297 Contributions from: Andreas Henriksson, Benjamin Steinwender,
9298 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
9299 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
9300 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
9301 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
9302 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
9303 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
9304 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
9305 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
9306 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
9307 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
9308 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
9310 — Berlin, 2014-07-03
9314 * As an experimental feature, udev now tries to lock the
9315 disk device node (flock(LOCK_SH|LOCK_NB)) while it
9316 executes events for the disk or any of its partitions.
9317 Applications like partitioning programs can lock the
9318 disk device node (flock(LOCK_EX)) and claim temporary
9319 device ownership that way; udev will entirely skip all event
9320 handling for this disk and its partitions. If the disk
9321 was opened for writing, the close will trigger a partition
9322 table rescan in udev's "watch" facility, and if needed
9323 synthesize "change" events for the disk and all its partitions.
9324 This is now unconditionally enabled, and if it turns out to
9325 cause major problems, we might turn it on only for specific
9326 devices, or might need to disable it entirely. Device Mapper
9327 devices are excluded from this logic.
9329 * We temporarily dropped the "-l" switch for fsck invocations,
9330 since they collide with the flock() logic above. util-linux
9331 upstream has been changed already to avoid this conflict,
9332 and we will re-add "-l" as soon as util-linux with this
9333 change has been released.
9335 * The dependency on libattr has been removed. Since a long
9336 time, the extended attribute calls have moved to glibc, and
9337 libattr is thus unnecessary.
9339 * Virtualization detection works without privileges now. This
9340 means the systemd-detect-virt binary no longer requires
9341 CAP_SYS_PTRACE file capabilities, and our daemons can run
9342 with fewer privileges.
9344 * systemd-networkd now runs under its own "systemd-network"
9345 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
9346 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
9347 loses the ability to write to files owned by root this way.
9349 * Similarly, systemd-resolved now runs under its own
9350 "systemd-resolve" user with no capabilities remaining.
9352 * Similarly, systemd-bus-proxyd now runs under its own
9353 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
9355 * systemd-networkd gained support for setting up "veth"
9356 virtual Ethernet devices for container connectivity, as well
9357 as GRE and VTI tunnels.
9359 * systemd-networkd will no longer automatically attempt to
9360 manually load kernel modules necessary for certain tunnel
9361 transports. Instead, it is assumed the kernel loads them
9362 automatically when required. This only works correctly on
9363 very new kernels. On older kernels, please consider adding
9364 the kernel modules to /etc/modules-load.d/ as a work-around.
9366 * The resolv.conf file systemd-resolved generates has been
9367 moved to /run/systemd/resolve/. If you have a symlink from
9368 /etc/resolv.conf, it might be necessary to correct it.
9370 * Two new service settings, ProtectHome= and ProtectSystem=,
9371 have been added. When enabled, they will make the user data
9372 (such as /home) inaccessible or read-only and the system
9373 (such as /usr) read-only, for specific services. This allows
9374 very light-weight per-service sandboxing to avoid
9375 modifications of user data or system files from
9376 services. These two new switches have been enabled for all
9377 of systemd's long-running services, where appropriate.
9379 * Socket units gained new SocketUser= and SocketGroup=
9380 settings to set the owner user and group of AF_UNIX sockets
9381 and FIFOs in the file system.
9383 * Socket units gained a new RemoveOnStop= setting. If enabled,
9384 all FIFOS and sockets in the file system will be removed
9385 when the specific socket unit is stopped.
9387 * Socket units gained a new Symlinks= setting. It takes a list
9388 of symlinks to create to file system sockets or FIFOs
9389 created by the specific Unix sockets. This is useful to
9390 manage symlinks to socket nodes with the same lifecycle as
9393 * The /dev/log socket and /dev/initctl FIFO have been moved to
9394 /run, and have been replaced by symlinks. This allows
9395 connecting to these facilities even if PrivateDevices=yes is
9396 used for a service (which makes /dev/log itself unavailable,
9397 but /run is left). This also has the benefit of ensuring
9398 that /dev only contains device nodes, directories and
9399 symlinks, and nothing else.
9401 * sd-daemon gained two new calls sd_pid_notify() and
9402 sd_pid_notifyf(). They are similar to sd_notify() and
9403 sd_notifyf(), but allow overriding of the source PID of
9404 notification messages if permissions permit this. This is
9405 useful to send notify messages on behalf of a different
9406 process (for example, the parent process). The
9407 systemd-notify tool has been updated to make use of this
9408 when sending messages (so that notification messages now
9409 originate from the shell script invoking systemd-notify and
9410 not the systemd-notify process itself. This should minimize
9411 a race where systemd fails to associate notification
9412 messages to services when the originating process already
9415 * A new "on-abnormal" setting for Restart= has been added. If
9416 set, it will result in automatic restarts on all "abnormal"
9417 reasons for a process to exit, which includes unclean
9418 signals, core dumps, timeouts and watchdog timeouts, but
9419 does not include clean and unclean exit codes or clean
9420 signals. Restart=on-abnormal is an alternative for
9421 Restart=on-failure for services that shall be able to
9422 terminate and avoid restarts on certain errors, by
9423 indicating so with an unclean exit code. Restart=on-failure
9424 or Restart=on-abnormal is now the recommended setting for
9425 all long-running services.
9427 * If the InaccessibleDirectories= service setting points to a
9428 mount point (or if there are any submounts contained within
9429 it), it is now attempted to completely unmount it, to make
9430 the file systems truly unavailable for the respective
9433 * The ReadOnlyDirectories= service setting and
9434 systemd-nspawn's --read-only parameter are now recursively
9435 applied to all submounts, too.
9437 * Mount units may now be created transiently via the bus APIs.
9439 * The support for SysV and LSB init scripts has been removed
9440 from the systemd daemon itself. Instead, it is now
9441 implemented as a generator that creates native systemd units
9442 from these scripts when needed. This enables us to remove a
9443 substantial amount of legacy code from PID 1, following the
9444 fact that many distributions only ship a very small number
9445 of LSB/SysV init scripts nowadays.
9447 * Privileged Xen (dom0) domains are not considered
9448 virtualization anymore by the virtualization detection
9449 logic. After all, they generally have unrestricted access to
9450 the hardware and usually are used to manage the unprivileged
9453 * systemd-tmpfiles gained a new "C" line type, for copying
9454 files or entire directories.
9456 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
9457 lines. So far, they have been non-globbing versions of the
9458 latter, and have thus been redundant. In future, it is
9459 recommended to only use "z". "m" has hence been removed
9460 from the documentation, even though it stays supported.
9462 * A tmpfiles snippet to recreate the most basic structure in
9463 /var has been added. This is enough to create the /var/run →
9464 /run symlink and create a couple of structural
9465 directories. This allows systems to boot up with an empty or
9466 volatile /var. Of course, while with this change, the core OS
9467 now is capable with dealing with a volatile /var, not all
9468 user services are ready for it. However, we hope that sooner
9469 or later, many service daemons will be changed upstream so
9470 that they are able to automatically create their necessary
9471 directories in /var at boot, should they be missing. This is
9472 the first step to allow state-less systems that only require
9473 the vendor image for /usr to boot.
9475 * systemd-nspawn has gained a new --tmpfs= switch to mount an
9476 empty tmpfs instance to a specific directory. This is
9477 particularly useful for making use of the automatic
9478 reconstruction of /var (see above), by passing --tmpfs=/var.
9480 * Access modes specified in tmpfiles snippets may now be
9481 prefixed with "~", which indicates that they shall be masked
9482 by whether the existing file or directory is currently
9483 writable, readable or executable at all. Also, if specified,
9484 the sgid/suid/sticky bits will be masked for all
9487 * A new passive target unit "network-pre.target" has been
9488 added which is useful for services that shall run before any
9489 network is configured, for example firewall scripts.
9491 * The "floppy" group that previously owned the /dev/fd*
9492 devices is no longer used. The "disk" group is now used
9493 instead. Distributions should probably deprecate usage of
9496 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
9497 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
9498 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
9499 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
9500 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
9501 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
9504 — Berlin, 2014-06-11
9508 * A new "systemd-timesyncd" daemon has been added for
9509 synchronizing the system clock across the network. It
9510 implements an SNTP client. In contrast to NTP
9511 implementations such as chrony or the NTP reference server,
9512 this only implements a client side, and does not bother with
9513 the full NTP complexity, focusing only on querying time from
9514 one remote server and synchronizing the local clock to
9515 it. Unless you intend to serve NTP to networked clients or
9516 want to connect to local hardware clocks, this simple NTP
9517 client should be more than appropriate for most
9518 installations. The daemon runs with minimal privileges, and
9519 has been hooked up with networkd to only operate when
9520 network connectivity is available. The daemon saves the
9521 current clock to disk every time a new NTP sync has been
9522 acquired, and uses this to possibly correct the system clock
9523 early at bootup, in order to accommodate for systems that
9524 lack an RTC such as the Raspberry Pi and embedded devices,
9525 and to make sure that time monotonically progresses on these
9526 systems, even if it is not always correct. To make use of
9527 this daemon, a new system user and group "systemd-timesync"
9528 needs to be created on installation of systemd.
9530 * The queue "seqnum" interface of libudev has been disabled, as
9531 it was generally incompatible with device namespacing as
9532 sequence numbers of devices go "missing" if the devices are
9533 part of a different namespace.
9535 * "systemctl list-timers" and "systemctl list-sockets" gained
9536 a --recursive switch for showing units of these types also
9537 for all local containers, similar in style to the already
9538 supported --recursive switch for "systemctl list-units".
9540 * A new RebootArgument= setting has been added for service
9541 units, which may be used to specify a kernel reboot argument
9542 to use when triggering reboots with StartLimitAction=.
9544 * A new FailureAction= setting has been added for service
9545 units which may be used to specify an operation to trigger
9546 when a service fails. This works similarly to
9547 StartLimitAction=, but unlike it, controls what is done
9548 immediately rather than only after several attempts to
9549 restart the service in question.
9551 * hostnamed got updated to also expose the kernel name,
9552 release, and version on the bus. This is useful for
9553 executing commands like hostnamectl with the -H switch.
9554 systemd-analyze makes use of this to properly display
9555 details when running non-locally.
9557 * The bootchart tool can now show cgroup information in the
9558 graphs it generates.
9560 * The CFS CPU quota cgroup attribute is now exposed for
9561 services. The new CPUQuota= switch has been added for this
9562 which takes a percentage value. Setting this will have the
9563 result that a service may never get more CPU time than the
9564 specified percentage, even if the machine is otherwise idle.
9566 * systemd-networkd learned IPIP and SIT tunnel support.
9568 * LSB init scripts exposing a dependency on $network will now
9569 get a dependency on network-online.target rather than simply
9570 network.target. This should bring LSB handling closer to
9571 what it was on SysV systems.
9573 * A new fsck.repair= kernel option has been added to control
9574 how fsck shall deal with unclean file systems at boot.
9576 * The (.ini) configuration file parser will now silently ignore
9577 sections whose names begin with "X-". This may be used to maintain
9578 application-specific extension sections in unit files.
9580 * machined gained a new API to query the IP addresses of
9581 registered containers. "machinectl status" has been updated
9582 to show these addresses in its output.
9584 * A new call sd_uid_get_display() has been added to the
9585 sd-login APIs for querying the "primary" session of a
9586 user. The "primary" session of the user is elected from the
9587 user's sessions and generally a graphical session is
9588 preferred over a text one.
9590 * A minimal systemd-resolved daemon has been added. It
9591 currently simply acts as a companion to systemd-networkd and
9592 manages resolv.conf based on per-interface DNS
9593 configuration, possibly supplied via DHCP. In the long run
9594 we hope to extend this into a local DNSSEC enabled DNS and
9597 * The systemd-networkd-wait-online tool is now enabled by
9598 default. It will delay network-online.target until a network
9599 connection has been configured. The tool primarily integrates
9600 with networkd, but will also make a best effort to make sense
9601 of network configuration performed in some other way.
9603 * Two new service options StartupCPUShares= and
9604 StartupBlockIOWeight= have been added that work similarly to
9605 CPUShares= and BlockIOWeight= however only apply during
9606 system startup. This is useful to prioritize certain services
9607 differently during bootup than during normal runtime.
9609 * hostnamed has been changed to prefer the statically
9610 configured hostname in /etc/hostname (unless set to
9611 'localhost' or empty) over any dynamic one supplied by
9612 dhcp. With this change, the rules for picking the hostname
9613 match more closely the rules of other configuration settings
9614 where the local administrator's configuration in /etc always
9615 overrides any other settings.
9617 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
9618 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
9619 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
9620 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
9621 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
9622 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
9623 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
9624 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
9625 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
9626 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
9627 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
9628 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
9629 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
9630 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
9631 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
9632 Lindskog, WaLyong Cho, Will Woods, Zbigniew
9635 — Beijing, 2014-05-28
9639 * When restoring the screen brightness at boot, stay away from
9640 the darkest setting or from the lowest 5% of the available
9641 range, depending on which is the larger value of both. This
9642 should effectively protect the user from rebooting into a
9643 black screen, should the brightness have been set to minimum
9646 * sd-login gained a new sd_machine_get_class() call to
9647 determine the class ("vm" or "container") of a machine
9648 registered with machined.
9650 * sd-login gained new calls
9651 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
9652 to query the identity of the peer of a local AF_UNIX
9653 connection. They operate similarly to their sd_pid_get_xyz()
9656 * PID 1 will now maintain a system-wide system state engine
9657 with the states "starting", "running", "degraded",
9658 "maintenance", "stopping". These states are bound to system
9659 startup, normal runtime, runtime with at least one failed
9660 service, rescue/emergency mode and system shutdown. This
9661 state is shown in the "systemctl status" output when no unit
9662 name is passed. It is useful to determine system state, in
9663 particularly when doing so for many systems or containers at
9666 * A new command "list-machines" has been added to "systemctl"
9667 that lists all local OS containers and shows their system
9668 state (see above), if systemd runs inside of them.
9670 * systemctl gained a new "-r" switch to recursively enumerate
9671 units on all local containers, when used with the
9672 "list-unit" command (which is the default one that is
9673 executed when no parameters are specified).
9675 * The GPT automatic partition discovery logic will now honour
9676 two GPT partition flags: one may be set on a partition to
9677 cause it to be mounted read-only, and the other may be set
9678 on a partition to ignore it during automatic discovery.
9680 * Two new GPT type UUIDs have been added for automatic root
9681 partition discovery, for 32-bit and 64-bit ARM. This is not
9682 particularly useful for discovering the root directory on
9683 these architectures during bare-metal boots (since UEFI is
9684 not common there), but still very useful to allow booting of
9685 ARM disk images in nspawn with the -i option.
9687 * MAC addresses of interfaces created with nspawn's
9688 --network-interface= switch will now be generated from the
9689 machine name, and thus be stable between multiple invocations
9692 * logind will now automatically remove all IPC objects owned
9693 by a user if she or he fully logs out. This makes sure that
9694 users who are logged out cannot continue to consume IPC
9695 resources. This covers SysV memory, semaphores and message
9696 queues as well as POSIX shared memory and message
9697 queues. Traditionally, SysV and POSIX IPC had no lifecycle
9698 limits. With this functionality, that is corrected. This may
9699 be turned off by using the RemoveIPC= switch of logind.conf.
9701 * The systemd-machine-id-setup and tmpfiles tools gained a
9702 --root= switch to operate on a specific root directory,
9705 * journald can now forward logged messages to the TTYs of all
9706 logged in users ("wall"). This is the default for all
9707 emergency messages now.
9709 * A new tool systemd-journal-remote has been added to stream
9710 journal log messages across the network.
9712 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
9713 controller trees are mounted into it. Note that the
9714 directories mounted beneath it are not read-only. This is a
9715 security measure and is particularly useful because glibc
9716 actually includes a search logic to pick any tmpfs it can
9717 find to implement shm_open() if /dev/shm is not available
9718 (which it might very well be in namespaced setups).
9720 * machinectl gained a new "poweroff" command to cleanly power
9721 down a local OS container.
9723 * The PrivateDevices= unit file setting will now also drop the
9724 CAP_MKNOD capability from the capability bound set, and
9725 imply DevicePolicy=closed.
9727 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
9728 comprehensively on all long-running systemd services where
9729 this is appropriate.
9731 * systemd-udevd will now run in a disassociated mount
9732 namespace. To mount directories from udev rules, make sure to
9733 pull in mount units via SYSTEMD_WANTS properties.
9735 * The kdbus support gained support for uploading policy into
9736 the kernel. sd-bus gained support for creating "monitoring"
9737 connections that can eavesdrop into all bus communication
9738 for debugging purposes.
9740 * Timestamps may now be specified in seconds since the UNIX
9741 epoch Jan 1st, 1970 by specifying "@" followed by the value
9744 * Native tcpwrap support in systemd has been removed. tcpwrap
9745 is old code, not really maintained anymore and has serious
9746 shortcomings, and better options such as firewalls
9747 exist. For setups that require tcpwrap usage, please
9748 consider invoking your socket-activated service via tcpd,
9749 like on traditional inetd.
9751 * A new system.conf configuration option
9752 DefaultTimerAccuracySec= has been added that controls the
9753 default AccuracySec= setting of .timer units.
9755 * Timer units gained a new WakeSystem= switch. If enabled,
9756 timers configured this way will cause the system to resume
9757 from system suspend (if the system supports that, which most
9760 * Timer units gained a new Persistent= switch. If enabled,
9761 timers configured this way will save to disk when they have
9762 been last triggered. This information is then used on next
9763 reboot to possible execute overdue timer events, that
9764 could not take place because the system was powered off.
9765 This enables simple anacron-like behaviour for timer units.
9767 * systemctl's "list-timers" will now also list the time a
9768 timer unit was last triggered in addition to the next time
9769 it will be triggered.
9771 * systemd-networkd will now assign predictable IPv4LL
9772 addresses to its local interfaces.
9774 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
9775 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
9776 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
9777 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
9778 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
9779 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
9780 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
9781 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
9784 — Berlin, 2014-03-25
9788 * A new unit file setting RestrictAddressFamilies= has been
9789 added to restrict which socket address families unit
9790 processes gain access to. This takes address family names
9791 like "AF_INET" or "AF_UNIX", and is useful to minimize the
9792 attack surface of services via exotic protocol stacks. This
9793 is built on seccomp system call filters.
9795 * Two new unit file settings RuntimeDirectory= and
9796 RuntimeDirectoryMode= have been added that may be used to
9797 manage a per-daemon runtime directories below /run. This is
9798 an alternative for setting up directory permissions with
9799 tmpfiles snippets, and has the advantage that the runtime
9800 directory's lifetime is bound to the daemon runtime and that
9801 the daemon starts up with an empty directory each time. This
9802 is particularly useful when writing services that drop
9803 privileges using the User= or Group= setting.
9805 * The DeviceAllow= unit setting now supports globbing for
9806 matching against device group names.
9808 * The systemd configuration file system.conf gained new
9809 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
9810 DefaultMemoryAccounting= to globally turn on/off accounting
9811 for specific resources (cgroups) for all units. These
9812 settings may still be overridden individually in each unit
9815 * systemd-gpt-auto-generator is now able to discover /srv and
9816 root partitions in addition to /home and swap partitions. It
9817 also supports LUKS-encrypted partitions now. With this in
9818 place, automatic discovery of partitions to mount following
9819 the Discoverable Partitions Specification
9820 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
9821 is now a lot more complete. This allows booting without
9822 /etc/fstab and without root= on the kernel command line on
9823 systems prepared appropriately.
9825 * systemd-nspawn gained a new --image= switch which allows
9826 booting up disk images and Linux installations on any block
9827 device that follow the Discoverable Partitions Specification
9828 (see above). This means that installations made with
9829 appropriately updated installers may now be started and
9830 deployed using container managers, completely
9831 unmodified. (We hope that libvirt-lxc will add support for
9832 this feature soon, too.)
9834 * systemd-nspawn gained a new --network-macvlan= setting to
9835 set up a private macvlan interface for the
9836 container. Similarly, systemd-networkd gained a new
9837 Kind=macvlan setting in .netdev files.
9839 * systemd-networkd now supports configuring local addresses
9842 * A new tool systemd-network-wait-online has been added to
9843 synchronously wait for network connectivity using
9846 * The sd-bus.h bus API gained a new sd_bus_track object for
9847 tracking the lifecycle of bus peers. Note that sd-bus.h is
9848 still not a public API though (unless you specify
9849 --enable-kdbus on the configure command line, which however
9850 voids your warranty and you get no API stability guarantee).
9852 * The $XDG_RUNTIME_DIR runtime directories for each user are
9853 now individual tmpfs instances, which has the benefit of
9854 introducing separate pools for each user, with individual
9855 size limits, and thus making sure that unprivileged clients
9856 can no longer negatively impact the system or other users by
9857 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
9858 RuntimeDirectorySize= has been introduced that allows
9859 controlling the default size limit for all users. It
9860 defaults to 10% of the available physical memory. This is no
9861 replacement for quotas on tmpfs though (which the kernel
9862 still does not support), as /dev/shm and /tmp are still
9863 shared resources used by both the system and unprivileged
9866 * logind will now automatically turn off automatic suspending
9867 on laptop lid close when more than one display is
9868 connected. This was previously expected to be implemented
9869 individually in desktop environments (such as GNOME),
9870 however has been added to logind now, in order to fix a
9871 boot-time race where a desktop environment might not have
9872 been started yet and thus not been able to take an inhibitor
9873 lock at the time where logind already suspends the system
9874 due to a closed lid.
9876 * logind will now wait at least 30s after each system
9877 suspend/resume cycle, and 3min after system boot before
9878 suspending the system due to a closed laptop lid. This
9879 should give USB docking stations and similar enough time to
9880 be probed and configured after system resume and boot in
9881 order to then act as suspend blocker.
9883 * systemd-run gained a new --property= setting which allows
9884 initialization of resource control properties (and others)
9885 for the created scope or service unit. Example: "systemd-run
9886 --property=BlockIOWeight=10 updatedb" may be used to run
9887 updatedb at a low block IO scheduling weight.
9889 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
9890 now also work in --scope mode.
9892 * When systemd is compiled with kdbus support, basic support
9893 for enforced policies is now in place. (Note that enabling
9894 kdbus still voids your warranty and no API compatibility
9897 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
9898 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
9899 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
9900 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
9901 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
9902 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
9903 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
9904 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
9905 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
9906 Zbigniew Jędrzejewski-Szmek
9908 — Berlin, 2014-03-12
9912 * systemd will now relabel /dev after loading the SMACK policy
9913 according to SMACK rules.
9915 * A new unit file option AppArmorProfile= has been added to
9916 set the AppArmor profile for the processes of a unit.
9918 * A new condition check ConditionArchitecture= has been added
9919 to conditionalize units based on the system architecture, as
9920 reported by uname()'s "machine" field.
9922 * systemd-networkd now supports matching on the system
9923 virtualization, architecture, kernel command line, hostname
9926 * logind is now a lot more aggressive when suspending the
9927 machine due to a closed laptop lid. Instead of acting only
9928 on the lid close action, it will continuously watch the lid
9929 status and act on it. This is useful for laptops where the
9930 power button is on the outside of the chassis so that it can
9931 be reached without opening the lid (such as the Lenovo
9932 Yoga). On those machines, logind will now immediately
9933 re-suspend the machine if the power button has been
9934 accidentally pressed while the laptop was suspended and in a
9935 backpack or similar.
9937 * logind will now watch SW_DOCK switches and inhibit reaction
9938 to the lid switch if it is pressed. This means that logind
9939 will not suspend the machine anymore if the lid is closed
9940 and the system is docked, if the laptop supports SW_DOCK
9941 notifications via the input layer. Note that ACPI docking
9942 stations do not generate this currently. Also note that this
9943 logic is usually not fully sufficient and Desktop
9944 Environments should take a lid switch inhibitor lock when an
9945 external display is connected, as systemd will not watch
9948 * nspawn will now make use of the devices cgroup controller by
9949 default, and only permit creation of and access to the usual
9950 API device nodes like /dev/null or /dev/random, as well as
9951 access to (but not creation of) the pty devices.
9953 * We will now ship a default .network file for
9954 systemd-networkd that automatically configures DHCP for
9955 network interfaces created by nspawn's --network-veth or
9956 --network-bridge= switches.
9958 * systemd will now understand the usual M, K, G, T suffixes
9959 according to SI conventions (i.e. to the base 1000) when
9960 referring to throughput and hardware metrics. It will stay
9961 with IEC conventions (i.e. to the base 1024) for software
9962 metrics, according to what is customary according to
9963 Wikipedia. We explicitly document which base applies for
9964 each configuration option.
9966 * The DeviceAllow= setting in unit files now supports a syntax to
9967 allow-list an entire group of devices node majors at once, based on
9968 the /proc/devices listing. For example, with the string "char-pts",
9969 it is now possible to allow-list all current and future pseudo-TTYs
9972 * sd-event learned a new "post" event source. Event sources of
9973 this type are triggered by the dispatching of any event
9974 source of a type that is not "post". This is useful for
9975 implementing clean-up and check event sources that are
9976 triggered by other work being done in the program.
9978 * systemd-networkd is no longer statically enabled, but uses
9979 the usual [Install] sections so that it can be
9980 enabled/disabled using systemctl. It still is enabled by
9983 * When creating a veth interface pair with systemd-nspawn, the
9984 host side will now be prefixed with "vb-" if
9985 --network-bridge= is used, and with "ve-" if --network-veth
9986 is used. This way, it is easy to distinguish these cases on
9987 the host, for example to apply different configuration to
9988 them with systemd-networkd.
9990 * The compatibility libraries for libsystemd-journal.so,
9991 libsystem-id128.so, libsystemd-login.so and
9992 libsystemd-daemon.so do not make use of IFUNC
9993 anymore. Instead, we now build libsystemd.so multiple times
9994 under these alternative names. This means that the footprint
9995 is drastically increased, but given that these are
9996 transitional compatibility libraries, this should not matter
9997 much. This change has been made necessary to support the ARM
9998 platform for these compatibility libraries, as the ARM
9999 toolchain is not really at the same level as the toolchain
10000 for other architectures like x86 and does not support
10001 IFUNC. Please make sure to use --enable-compat-libs only
10002 during a transitional period!
10004 * The .include syntax has been deprecated and is not documented
10005 anymore. Drop-in files in .d directories should be used instead.
10007 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
10008 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
10009 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
10010 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
10011 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
10012 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
10013 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
10014 Zbigniew Jędrzejewski-Szmek
10016 — Berlin, 2014-02-24
10020 * A new component "systemd-networkd" has been added that can
10021 be used to configure local network interfaces statically or
10022 via DHCP. It is capable of bringing up bridges, VLANs, and
10023 bonding. Currently, no hook-ups for interactive network
10024 configuration are provided. Use this for your initrd,
10025 container, embedded, or server setup if you need a simple,
10026 yet powerful, network configuration solution. This
10027 configuration subsystem is quite nifty, as it allows wildcard
10028 hotplug matching in interfaces. For example, with a single
10029 configuration snippet, you can configure that all Ethernet
10030 interfaces showing up are automatically added to a bridge,
10031 or similar. It supports link-sensing and more.
10033 * A new tool "systemd-socket-proxyd" has been added which can
10034 act as a bidirectional proxy for TCP sockets. This is
10035 useful for adding socket activation support to services that
10036 do not actually support socket activation, including virtual
10037 machines and the like.
10039 * Add a new tool to save/restore rfkill state on
10042 * Save/restore state of keyboard backlights in addition to
10043 display backlights on shutdown/boot.
10045 * udev learned a new SECLABEL{} construct to label device
10046 nodes with a specific security label when they appear. For
10047 now, only SECLABEL{selinux} is supported, but the syntax is
10048 prepared for additional security frameworks.
10050 * udev gained a new scheme to configure link-level attributes
10051 from files in /etc/systemd/network/*.link. These files can
10052 match against MAC address, device path, driver name and type,
10053 and will apply attributes like the naming policy, link speed,
10054 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
10055 address assignment policy (randomized, …).
10057 * The configuration of network interface naming rules for
10058 "permanent interface names" has changed: a new NamePolicy=
10059 setting in the [Link] section of .link files determines the
10060 priority of possible naming schemes (onboard, slot, MAC,
10061 path). The default value of this setting is determined by
10062 /usr/lib/net/links/99-default.link. Old
10063 80-net-name-slot.rules udev configuration file has been
10064 removed, so local configuration overriding this file should
10065 be adapted to override 99-default.link instead.
10067 * When the User= switch is used in a unit file, also
10068 initialize $SHELL= based on the user database entry.
10070 * systemd no longer depends on libdbus. All communication is
10071 now done with sd-bus, systemd's low-level bus library
10074 * kdbus support has been added to PID 1 itself. When kdbus is
10075 enabled, this causes PID 1 to set up the system bus and
10076 enable support for a new ".busname" unit type that
10077 encapsulates bus name activation on kdbus. It works a little
10078 bit like ".socket" units, except for bus names. A new
10079 generator has been added that converts classic dbus1 service
10080 activation files automatically into native systemd .busname
10081 and .service units.
10083 * sd-bus: add a light-weight vtable implementation that allows
10084 defining objects on the bus with a simple static const
10085 vtable array of its methods, signals and properties.
10087 * systemd will not generate or install static dbus
10088 introspection data anymore to /usr/share/dbus-1/interfaces,
10089 as the precise format of these files is unclear, and
10090 nothing makes use of it.
10092 * A proxy daemon is now provided to proxy clients connecting
10093 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
10094 compatibility with classic D-Bus.
10096 * A bus driver implementation has been added that supports the
10097 classic D-Bus bus driver calls on kdbus, also for
10098 compatibility purposes.
10100 * A new API "sd-event.h" has been added that implements a
10101 minimal event loop API built around epoll. It provides a
10102 couple of features that direct epoll usage is lacking:
10103 prioritization of events, scales to large numbers of timer
10104 events, per-event timer slack (accuracy), system-wide
10105 coalescing of timer events, exit handlers, watchdog
10106 supervision support using systemd's sd_notify() API, child
10109 * A new API "sd-rntl.h" has been added that provides an API
10110 around the route netlink interface of the kernel, similar in
10111 style to "sd-bus.h".
10113 * A new API "sd-dhcp-client.h" has been added that provides a
10114 small DHCPv4 client-side implementation. This is used by
10115 "systemd-networkd".
10117 * There is a new kernel command line option
10118 "systemd.restore_state=0|1". When set to "0", none of the
10119 systemd tools will restore saved runtime state to hardware
10120 devices. More specifically, the rfkill and backlight states
10123 * The FsckPassNo= compatibility option in mount/service units
10124 has been removed. The fstab generator will now add the
10125 necessary dependencies automatically, and does not require
10126 PID1's support for that anymore.
10128 * journalctl gained a new switch, --list-boots, that lists
10129 recent boots with their times and boot IDs.
10131 * The various tools like systemctl, loginctl, timedatectl,
10132 busctl, systemd-run, … have gained a new switch "-M" to
10133 connect to a specific, local OS container (as direct
10134 connection, without requiring SSH). This works on any
10135 container that is registered with machined, such as those
10136 created by libvirt-lxc or nspawn.
10138 * systemd-run and systemd-analyze also gained support for "-H"
10139 to connect to remote hosts via SSH. This is particularly
10140 useful for systemd-run because it enables queuing of jobs
10141 onto remote systems.
10143 * machinectl gained a new command "login" to open a getty
10144 login in any local container. This works with any container
10145 that is registered with machined (such as those created by
10146 libvirt-lxc or nspawn), and which runs systemd inside.
10148 * machinectl gained a new "reboot" command that may be used to
10149 trigger a reboot on a specific container that is registered
10150 with machined. This works on any container that runs an init
10151 system of some kind.
10153 * systemctl gained a new "list-timers" command to print a nice
10154 listing of installed timer units with the times they elapse
10157 * Alternative reboot() parameters may now be specified on the
10158 "systemctl reboot" command line and are passed to the
10159 reboot() system call.
10161 * systemctl gained a new --job-mode= switch to configure the
10162 mode to queue a job with. This is a more generic version of
10163 --fail, --irreversible, and --ignore-dependencies, which are
10164 still available but not advertised anymore.
10166 * /etc/systemd/system.conf gained new settings to configure
10167 various default timeouts of units, as well as the default
10168 start limit interval and burst. These may still be overridden
10171 * PID1 will now export on the bus profile data of the security
10172 policy upload process (such as the SELinux policy upload to
10175 * journald: when forwarding logs to the console, include
10176 timestamps (following the setting in
10177 /sys/module/printk/parameters/time).
10179 * OnCalendar= in timer units now understands the special
10180 strings "yearly" and "annually". (Both are equivalent)
10182 * The accuracy of timer units is now configurable with the new
10183 AccuracySec= setting. It defaults to 1min.
10185 * A new dependency type JoinsNamespaceOf= has been added that
10186 allows running two services within the same /tmp and network
10187 namespace, if PrivateNetwork= or PrivateTmp= are used.
10189 * A new command "cat" has been added to systemctl. It outputs
10190 the original unit file of a unit, and concatenates the
10191 contents of additional "drop-in" unit file snippets, so that
10192 the full configuration is shown.
10194 * systemctl now supports globbing on the various "list-xyz"
10195 commands, like "list-units" or "list-sockets", as well as on
10196 those commands which take multiple unit names.
10198 * journalctl's --unit= switch gained support for globbing.
10200 * All systemd daemons now make use of the watchdog logic so
10201 that systemd automatically notices when they hang.
10203 * If the $container_ttys environment variable is set,
10204 getty-generator will automatically spawn a getty for each
10205 listed tty. This is useful for container managers to request
10206 login gettys to be spawned on as many ttys as needed.
10208 * %h, %s, %U specifier support is not available anymore when
10209 used in unit files for PID 1. This is because NSS calls are
10210 not safe from PID 1. They stay available for --user
10211 instances of systemd, and as special case for the root user.
10213 * loginctl gained a new "--no-legend" switch to turn off output
10214 of the legend text.
10216 * The "sd-login.h" API gained three new calls:
10217 sd_session_is_remote(), sd_session_get_remote_user(),
10218 sd_session_get_remote_host() to query information about
10221 * The udev hardware database now also carries vendor/product
10222 information of SDIO devices.
10224 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
10225 determine whether watchdog notifications are requested by
10226 the system manager.
10228 * Socket-activated per-connection services now include a
10229 short description of the connection parameters in the
10232 * tmpfiles gained a new "--boot" option. When this is not used,
10233 only lines where the command character is not suffixed with
10234 "!" are executed. When this option is specified, those
10235 options are executed too. This partitions tmpfiles
10236 directives into those that can be safely executed at any
10237 time, and those which should be run only at boot (for
10238 example, a line that creates /run/nologin).
10240 * A new API "sd-resolve.h" has been added which provides a simple
10241 asynchronous wrapper around glibc NSS hostname resolution
10242 calls, such as getaddrinfo(). In contrast to glibc's
10243 getaddrinfo_a(), it does not use signals. In contrast to most
10244 other asynchronous name resolution libraries, this one does
10245 not reimplement DNS, but reuses NSS, so that alternate
10246 hostname resolution systems continue to work, such as mDNS,
10247 LDAP, etc. This API is based on libasyncns, but it has been
10248 cleaned up for inclusion in systemd.
10250 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
10251 "sd-daemon.h" are no longer found in individual libraries
10252 libsystemd-journal.so, libsystemd-login.so,
10253 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
10254 merged them into a single library, libsystemd.so, which
10255 provides all symbols. The reason for this is cyclic
10256 dependencies, as these libraries tend to use each other's
10257 symbols. So far, we have managed to workaround that by linking
10258 a copy of a good part of our code into each of these
10259 libraries again and again, which, however, makes certain
10260 things hard to do, like sharing static variables. Also, it
10261 substantially increases footprint. With this change, there
10262 is only one library for the basic APIs systemd
10263 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
10264 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
10265 library as well, however are subject to the --enable-kdbus
10266 switch (see below). Note that "sd-dhcp-client.h" is not part
10267 of this library (this is because it only consumes, never
10268 provides, services of/to other APIs). To make the transition
10269 easy from the separate libraries to the unified one, we
10270 provide the --enable-compat-libs compile-time switch which
10271 will generate stub libraries that are compatible with the
10272 old ones but redirect all calls to the new one.
10274 * All of the kdbus logic and the new APIs "sd-bus.h",
10275 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
10276 and "sd-utf8.h" are compile-time optional via the
10277 "--enable-kdbus" switch, and they are not compiled in by
10278 default. To make use of kdbus, you have to explicitly enable
10279 the switch. Note however, that neither the kernel nor the
10280 userspace API for all of this is considered stable yet. We
10281 want to maintain the freedom to still change the APIs for
10282 now. By specifying this build-time switch, you acknowledge
10283 that you are aware of the instability of the current
10286 * Also, note that while kdbus is pretty much complete,
10287 it lacks one thing: proper policy support. This means you
10288 can build a fully working system with all features; however,
10289 it will be highly insecure. Policy support will be added in
10290 one of the next releases, at the same time that we will
10291 declare the APIs stable.
10293 * When the kernel command line argument "kdbus" is specified,
10294 systemd will automatically load the kdbus.ko kernel module. At
10295 this stage of development, it is only useful for testing kdbus
10296 and should not be used in production. Note: if "--enable-kdbus"
10297 is specified, and the kdbus.ko kernel module is available, and
10298 "kdbus" is added to the kernel command line, the entire system
10299 runs with kdbus instead of dbus-daemon, with the above mentioned
10300 problem of missing the system policy enforcement. Also a future
10301 version of kdbus.ko or a newer systemd will not be compatible with
10302 each other, and will unlikely be able to boot the machine if only
10303 one of them is updated.
10305 * systemctl gained a new "import-environment" command which
10306 uploads the caller's environment (or parts thereof) into the
10307 service manager so that it is inherited by services started
10308 by the manager. This is useful to upload variables like
10309 $DISPLAY into the user service manager.
10311 * A new PrivateDevices= switch has been added to service units
10312 which allows running a service with a namespaced /dev
10313 directory that does not contain any device nodes for
10314 physical devices. More specifically, it only includes devices
10315 such as /dev/null, /dev/urandom, and /dev/zero which are API
10318 * logind has been extended to support behaviour like VT
10319 switching on seats that do not support a VT. This makes
10320 multi-session available on seats that are not the first seat
10321 (seat0), and on systems where kernel support for VTs has
10322 been disabled at compile-time.
10324 * If a process holds a delay lock for system sleep or shutdown
10325 and fails to release it in time, we will now log its
10326 identity. This makes it easier to identify processes that
10327 cause slow suspends or power-offs.
10329 * When parsing /etc/crypttab, support for a new key-slot=
10330 option as supported by Debian is added. It allows indicating
10331 which LUKS slot to use on disk, speeding up key loading.
10333 * The sd_journal_sendv() API call has been checked and
10334 officially declared to be async-signal-safe so that it may
10335 be invoked from signal handlers for logging purposes.
10337 * Boot-time status output is now enabled automatically after a
10338 short timeout if boot does not progress, in order to give
10339 the user an indication what she or he is waiting for.
10341 * The boot-time output has been improved to show how much time
10342 remains until jobs expire.
10344 * The KillMode= switch in service units gained a new possible
10345 value "mixed". If set, and the unit is shut down, then the
10346 initial SIGTERM signal is sent only to the main daemon
10347 process, while the following SIGKILL signal is sent to
10348 all remaining processes of the service.
10350 * When a scope unit is registered, a new property "Controller"
10351 may be set. If set to a valid bus name, systemd will send a
10352 RequestStop() signal to this name when it would like to shut
10353 down the scope. This may be used to hook manager logic into
10354 the shutdown logic of scope units. Also, scope units may now
10355 be put in a special "abandoned" state, in which case the
10356 manager process which created them takes no further
10357 responsibilities for it.
10359 * When reading unit files, systemd will now verify
10360 the access mode of these files, and warn about certain
10361 suspicious combinations. This has been added to make it
10362 easier to track down packaging bugs where unit files are
10363 marked executable or world-writable.
10365 * systemd-nspawn gained a new "--setenv=" switch to set
10366 container-wide environment variables. The similar option in
10367 systemd-activate was renamed from "--environment=" to
10368 "--setenv=" for consistency.
10370 * systemd-nspawn has been updated to create a new kdbus domain
10371 for each container that is invoked, thus allowing each
10372 container to have its own set of system and user buses,
10373 independent of the host.
10375 * systemd-nspawn gained a new --drop-capability= switch to run
10376 the container with less capabilities than the default. Both
10377 --drop-capability= and --capability= now take the special
10378 string "all" for dropping or keeping all capabilities.
10380 * systemd-nspawn gained new switches for executing containers
10381 with specific SELinux labels set.
10383 * systemd-nspawn gained a new --quiet switch to not generate
10384 any additional output but the container's own console
10387 * systemd-nspawn gained a new --share-system switch to run a
10388 container without PID namespacing enabled.
10390 * systemd-nspawn gained a new --register= switch to control
10391 whether the container is registered with systemd-machined or
10392 not. This is useful for containers that do not run full
10393 OS images, but only specific apps.
10395 * systemd-nspawn gained a new --keep-unit which may be used
10396 when invoked as the only program from a service unit, and
10397 results in registration of the unit service itself in
10398 systemd-machined, instead of a newly opened scope unit.
10400 * systemd-nspawn gained a new --network-interface= switch for
10401 moving arbitrary interfaces to the container. The new
10402 --network-veth switch creates a virtual Ethernet connection
10403 between host and container. The new --network-bridge=
10404 switch then allows assigning the host side of this virtual
10405 Ethernet connection to a bridge device.
10407 * systemd-nspawn gained a new --personality= switch for
10408 setting the kernel personality for the container. This is
10409 useful when running a 32-bit container on a 64-bit host. A
10410 similar option Personality= is now also available for service
10413 * logind will now also track a "Desktop" identifier for each
10414 session which encodes the desktop environment of it. This is
10415 useful for desktop environments that want to identify
10416 multiple running sessions of itself easily.
10418 * A new SELinuxContext= setting for service units has been
10419 added that allows setting a specific SELinux execution
10420 context for a service.
10422 * Most systemd client tools will now honour $SYSTEMD_LESS for
10423 settings of the "less" pager. By default, these tools will
10424 override $LESS to allow certain operations to work, such as
10425 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
10426 influence this logic.
10428 * systemd's "seccomp" hook-up has been changed to make use of
10429 the libseccomp library instead of using its own
10430 implementation. This has benefits for portability among
10433 * For usage together with SystemCallFilter=, a new
10434 SystemCallErrorNumber= setting has been introduced that
10435 allows configuration of a system error number to be returned
10436 on filtered system calls, instead of immediately killing the
10437 process. Also, SystemCallArchitectures= has been added to
10438 limit access to system calls of a particular architecture
10439 (in order to turn off support for unused secondary
10440 architectures). There is also a global
10441 SystemCallArchitectures= setting in system.conf now to turn
10442 off support for non-native system calls system-wide.
10444 * systemd requires a kernel with a working name_to_handle_at(),
10445 please see the kernel config requirements in the README file.
10447 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
10448 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
10449 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
10450 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
10451 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
10452 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
10453 Elia Pinto, Florian Weimer, George McCollister, Goffredo
10454 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
10455 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
10456 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
10457 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
10458 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
10459 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
10460 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
10461 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
10462 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
10463 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
10464 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
10465 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
10466 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
10467 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
10468 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
10469 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
10470 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
10472 — Berlin, 2014-02-20
10476 * logind has gained support for facilitating privileged input
10477 and drm device access for unprivileged clients. This work is
10478 useful to allow Wayland display servers (and similar
10479 programs, such as kmscon) to run under the user's ID and
10480 access input and drm devices which are normally
10481 protected. When this is used (and the kernel is new enough)
10482 logind will "mute" IO on the file descriptors passed to
10483 Wayland as long as it is in the background and "unmute" it
10484 if it returns into the foreground. This allows secure
10485 session switching without allowing background sessions to
10486 eavesdrop on input and display data. This also introduces
10487 session switching support if VT support is turned off in the
10488 kernel, and on seats that are not seat0.
10490 * A new kernel command line option luks.options= is understood
10491 now which allows specifying LUKS options for usage for LUKS
10492 encrypted partitions specified with luks.uuid=.
10494 * tmpfiles.d(5) snippets may now use specifier expansion in
10495 path names. More specifically %m, %b, %H, %v, are now
10496 replaced by the local machine id, boot id, hostname, and
10497 kernel version number.
10499 * A new tmpfiles.d(5) command "m" has been introduced which
10500 may be used to change the owner/group/access mode of a file
10501 or directory if it exists, but do nothing if it does not.
10503 * This release removes high-level support for the
10504 MemorySoftLimit= cgroup setting. The underlying kernel
10505 cgroup attribute memory.soft_limit= is currently badly
10506 designed and likely to be removed from the kernel API in its
10507 current form, hence we should not expose it for now.
10509 * The memory.use_hierarchy cgroup attribute is now enabled for
10510 all cgroups systemd creates in the memory cgroup
10511 hierarchy. This option is likely to be come the built-in
10512 default in the kernel anyway, and the non-hierarchical mode
10513 never made much sense in the intrinsically hierarchical
10516 * A new field _SYSTEMD_SLICE= is logged along with all journal
10517 messages containing the slice a message was generated
10518 from. This is useful to allow easy per-customer filtering of
10519 logs among other things.
10521 * systemd-journald will no longer adjust the group of journal
10522 files it creates to the "systemd-journal" group. Instead we
10523 rely on the journal directory to be owned by the
10524 "systemd-journal" group, and its setgid bit set, so that the
10525 kernel file system layer will automatically enforce that
10526 journal files inherit this group assignment. The reason for
10527 this change is that we cannot allow NSS look-ups from
10528 journald which would be necessary to resolve
10529 "systemd-journal" to a numeric GID, because this might
10530 create deadlocks if NSS involves synchronous queries to
10531 other daemons (such as nscd, or sssd) which in turn are
10532 logging clients of journald and might block on it, which
10533 would then dead lock. A tmpfiles.d(5) snippet included in
10534 systemd will make sure the setgid bit and group are
10535 properly set on the journal directory if it exists on every
10536 boot. However, we recommend adjusting it manually after
10537 upgrades too (or from RPM scriptlets), so that the change is
10538 not delayed until next reboot.
10540 * Backlight and random seed files in /var/lib/ have moved into
10541 the /var/lib/systemd/ directory, in order to centralize all
10542 systemd generated files in one directory.
10544 * Boot time performance measurements (as displayed by
10545 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
10546 performance information if that's available to determine how
10547 much time BIOS and boot loader initialization required. With
10548 a sufficiently new BIOS you hence no longer need to boot
10549 with Gummiboot to get access to such information.
10551 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
10552 Cristian Rodríguez, Dave Reisner, David Herrmann, David
10553 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
10554 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
10555 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
10556 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
10557 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
10559 — Berlin, 2013-10-02
10563 * The Restart= option for services now understands a new
10564 on-watchdog setting, which will restart the service
10565 automatically if the service stops sending out watchdog keep
10566 alive messages (as configured with WatchdogSec=).
10568 * The getty generator (which is responsible for bringing up a
10569 getty on configured serial consoles) will no longer only
10570 start a getty on the primary kernel console but on all
10571 others, too. This makes the order in which console= is
10572 specified on the kernel command line less important.
10574 * libsystemd-logind gained a new sd_session_get_vt() call to
10575 retrieve the VT number of a session.
10577 * If the option "tries=0" is set for an entry of /etc/crypttab
10578 its passphrase is queried indefinitely instead of any
10579 maximum number of tries.
10581 * If a service with a configure PID file terminates its PID
10582 file will now be removed automatically if it still exists
10583 afterwards. This should put an end to stale PID files.
10585 * systemd-run will now also take relative binary path names
10586 for execution and no longer insists on absolute paths.
10588 * InaccessibleDirectories= and ReadOnlyDirectories= now take
10589 paths that are optionally prefixed with "-" to indicate that
10590 it should not be considered a failure if they do not exist.
10592 * journalctl -o (and similar commands) now understands a new
10593 output mode "short-precise", it is similar to "short" but
10594 shows timestamps with usec accuracy.
10596 * The option "discard" (as known from Debian) is now
10597 synonymous to "allow-discards" in /etc/crypttab. In fact,
10598 "discard" is preferred now (since it is easier to remember
10601 * Some licensing clean-ups were made, so that more code is now
10602 LGPL-2.1 licensed than before.
10604 * A minimal tool to save/restore the display backlight
10605 brightness across reboots has been added. It will store the
10606 backlight setting as late as possible at shutdown, and
10607 restore it as early as possible during reboot.
10609 * A logic to automatically discover and enable home and swap
10610 partitions on GPT disks has been added. With this in place
10611 /etc/fstab becomes optional for many setups as systemd can
10612 discover certain partitions located on the root disk
10613 automatically. Home partitions are recognized under their
10614 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
10615 partitions are recognized under their GPT type ID
10616 0657fd6da4ab43c484e50933c84b4f4f.
10618 * systemd will no longer pass any environment from the kernel
10619 or initrd to system services. If you want to set an
10620 environment for all services, do so via the kernel command
10621 line systemd.setenv= assignment.
10623 * The systemd-sysctl tool no longer natively reads the file
10624 /etc/sysctl.conf. If desired, the file should be symlinked
10625 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
10626 legacy support by a symlink rather than built-in code, it
10627 also makes the otherwise hidden order of application of the
10628 different files visible. (Note that this partly reverts to a
10629 pre-198 application order of sysctl knobs!)
10631 * The "systemctl set-log-level" and "systemctl dump" commands
10632 have been moved to systemd-analyze.
10634 * systemd-run learned the new --remain-after-exit switch,
10635 which causes the scope unit not to be cleaned up
10636 automatically after the process terminated.
10638 * tmpfiles learned a new --exclude-prefix= switch to exclude
10639 certain paths from operation.
10641 * journald will now automatically flush all messages to disk
10642 as soon as a message at the log level CRIT, ALERT or EMERG
10645 Contributions from: Andrew Cook, Brandon Philips, Christian
10646 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
10647 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
10648 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
10649 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
10650 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
10651 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
10652 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
10653 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
10654 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
10655 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
10656 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
10657 William Giokas, Zbigniew Jędrzejewski-Szmek
10659 — Berlin, 2013-09-13
10663 * The documentation has been updated to cover the various new
10664 concepts introduced with 205.
10666 * Unit files now understand the new %v specifier which
10667 resolves to the kernel version string as returned by "uname
10670 * systemctl now supports filtering the unit list output by
10671 load state, active state and sub state, using the new
10672 --state= parameter.
10674 * "systemctl status" will now show the results of the
10675 condition checks (like ConditionPathExists= and similar) of
10676 the last start attempts of the unit. They are also logged to
10679 * "journalctl -b" may now be used to look for boot output of a
10680 specific boot. Try "journalctl -b -1" for the previous boot,
10681 but the syntax is substantially more powerful.
10683 * "journalctl --show-cursor" has been added which prints the
10684 cursor string the last shown log line. This may then be used
10685 with the new "journalctl --after-cursor=" switch to continue
10686 browsing logs from that point on.
10688 * "journalctl --force" may now be used to force regeneration
10691 * Creation of "dead" device nodes has been moved from udev
10692 into kmod and tmpfiles. Previously, udev would read the kmod
10693 databases to pre-generate dead device nodes based on meta
10694 information contained in kernel modules, so that these would
10695 be auto-loaded on access rather then at boot. As this
10696 does not really have much to do with the exposing actual
10697 kernel devices to userspace this has always been slightly
10698 alien in the udev codebase. Following the new scheme kmod
10699 will now generate a runtime snippet for tmpfiles from the
10700 module meta information and it now is tmpfiles' job to the
10701 create the nodes. This also allows overriding access and
10702 other parameters for the nodes using the usual tmpfiles
10703 facilities. As side effect this allows us to remove the
10704 CAP_SYS_MKNOD capability bit from udevd entirely.
10706 * logind's device ACLs may now be applied to these "dead"
10707 devices nodes too, thus finally allowing managed access to
10708 devices such as /dev/snd/sequencer without loading the
10709 backing module right-away.
10711 * A new RPM macro has been added that may be used to apply
10712 tmpfiles configuration during package installation.
10714 * systemd-detect-virt and ConditionVirtualization= now can
10715 detect User-Mode-Linux machines (UML).
10717 * journald will now implicitly log the effective capabilities
10718 set of processes in the message metadata.
10720 * systemd-cryptsetup has gained support for TrueCrypt volumes.
10722 * The initrd interface has been simplified (more specifically,
10723 support for passing performance data via environment
10724 variables and fsck results via files in /run has been
10725 removed). These features were non-essential, and are
10726 nowadays available in a much nicer way by having systemd in
10727 the initrd serialize its state and have the hosts systemd
10728 deserialize it again.
10730 * The udev "keymap" data files and tools to apply keyboard
10731 specific mappings of scan to key codes, and force-release
10732 scan code lists have been entirely replaced by a udev
10733 "keyboard" builtin and a hwdb data file.
10735 * systemd will now honour the kernel's "quiet" command line
10736 argument also during late shutdown, resulting in a
10737 completely silent shutdown when used.
10739 * There's now an option to control the SO_REUSEPORT socket
10740 option in .socket units.
10742 * Instance units will now automatically get a per-template
10743 subslice of system.slice unless something else is explicitly
10744 configured. For example, instances of sshd@.service will now
10745 implicitly be placed in system-sshd.slice rather than
10746 system.slice as before.
10748 * Test coverage support may now be enabled at build time.
10750 Contributions from: Dave Reisner, Frederic Crozat, Harald
10751 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
10752 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
10753 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
10754 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
10755 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
10756 Giokas, Zbigniew Jędrzejewski-Szmek
10758 — Berlin, 2013-07-23
10762 * Two new unit types have been introduced:
10764 Scope units are very similar to service units, however, are
10765 created out of pre-existing processes — instead of PID 1
10766 forking off the processes. By using scope units it is
10767 possible for system services and applications to group their
10768 own child processes (worker processes) in a powerful way
10769 which then maybe used to organize them, or kill them
10770 together, or apply resource limits on them.
10772 Slice units may be used to partition system resources in an
10773 hierarchical fashion and then assign other units to them. By
10774 default there are now three slices: system.slice (for all
10775 system services), user.slice (for all user sessions),
10776 machine.slice (for VMs and containers).
10778 Slices and scopes have been introduced primarily in
10779 context of the work to move cgroup handling to a
10780 single-writer scheme, where only PID 1
10781 creates/removes/manages cgroups.
10783 * There's a new concept of "transient" units. In contrast to
10784 normal units these units are created via an API at runtime,
10785 not from configuration from disk. More specifically this
10786 means it is now possible to run arbitrary programs as
10787 independent services, with all execution parameters passed
10788 in via bus APIs rather than read from disk. Transient units
10789 make systemd substantially more dynamic then it ever was,
10790 and useful as a general batch manager.
10792 * logind has been updated to make use of scope and slice units
10793 for managing user sessions. As a user logs in he will get
10794 his own private slice unit, to which all sessions are added
10795 as scope units. We also added support for automatically
10796 adding an instance of user@.service for the user into the
10797 slice. Effectively logind will no longer create cgroup
10798 hierarchies on its own now, it will defer entirely to PID 1
10799 for this by means of scope, service and slice units. Since
10800 user sessions this way become entities managed by PID 1
10801 the output of "systemctl" is now a lot more comprehensive.
10803 * A new mini-daemon "systemd-machined" has been added which
10804 may be used by virtualization managers to register local
10805 VMs/containers. nspawn has been updated accordingly, and
10806 libvirt will be updated shortly. machined will collect a bit
10807 of meta information about the VMs/containers, and assign
10808 them their own scope unit (see above). The collected
10809 meta-data is then made available via the "machinectl" tool,
10810 and exposed in "ps" and similar tools. machined/machinectl
10811 is compile-time optional.
10813 * As discussed earlier, the low-level cgroup configuration
10814 options ControlGroup=, ControlGroupModify=,
10815 ControlGroupPersistent=, ControlGroupAttribute= have been
10816 removed. Please use high-level attribute settings instead as
10817 well as slice units.
10819 * A new bus call SetUnitProperties() has been added to alter
10820 various runtime parameters of a unit. This is primarily
10821 useful to alter cgroup parameters dynamically in a nice way,
10822 but will be extended later on to make more properties
10823 modifiable at runtime. systemctl gained a new set-properties
10824 command that wraps this call.
10826 * A new tool "systemd-run" has been added which can be used to
10827 run arbitrary command lines as transient services or scopes,
10828 while configuring a number of settings via the command
10829 line. This tool is currently very basic, however already
10830 very useful. We plan to extend this tool to even allow
10831 queuing of execution jobs with time triggers from the
10832 command line, similar in fashion to "at".
10834 * nspawn will now inform the user explicitly that kernels with
10835 audit enabled break containers, and suggest the user to turn
10838 * Support for detecting the IMA and AppArmor security
10839 frameworks with ConditionSecurity= has been added.
10841 * journalctl gained a new "-k" switch for showing only kernel
10842 messages, mimicking dmesg output; in addition to "--user"
10843 and "--system" switches for showing only user's own logs
10846 * systemd-delta can now show information about drop-in
10847 snippets extending unit files.
10849 * libsystemd-bus has been substantially updated but is still
10850 not available as public API.
10852 * systemd will now look for the "debug" argument on the kernel
10853 command line and enable debug logging, similar to what
10854 "systemd.log_level=debug" already did before.
10856 * "systemctl set-default", "systemctl get-default" has been
10857 added to configure the default.target symlink, which
10858 controls what to boot into by default.
10860 * "systemctl set-log-level" has been added as a convenient
10861 way to raise and lower systemd logging threshold.
10863 * "systemd-analyze plot" will now show the time the various
10864 generators needed for execution, as well as information
10865 about the unit file loading.
10867 * libsystemd-journal gained a new sd_journal_open_files() call
10868 for opening specific journal files. journactl also gained a
10869 new switch to expose this new functionality. Previously we
10870 only supported opening all files from a directory, or all
10871 files from the system, as opening individual files only is
10872 racy due to journal file rotation.
10874 * systemd gained the new DefaultEnvironment= setting in
10875 /etc/systemd/system.conf to set environment variables for
10878 * If a privileged process logs a journal message with the
10879 OBJECT_PID= field set, then journald will automatically
10880 augment this with additional OBJECT_UID=, OBJECT_GID=,
10881 OBJECT_COMM=, OBJECT_EXE=, … fields. This is useful if
10882 system services want to log events about specific client
10883 processes. journactl/systemctl has been updated to make use
10884 of this information if all log messages regarding a specific
10887 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
10888 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
10889 Reisner, David Coppa, David King, David Strauss, Eelco
10890 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
10891 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
10892 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
10893 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
10894 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
10895 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
10896 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
10897 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
10898 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
10899 Łukasz Stelmach, 장동준
10903 * The Python bindings gained some minimal support for the APIs
10904 exposed by libsystemd-logind.
10906 * ConditionSecurity= gained support for detecting SMACK. Since
10907 this condition already supports SELinux and AppArmor we only
10908 miss IMA for this. Patches welcome!
10910 Contributions from: Karol Lewandowski, Lennart Poettering,
10911 Zbigniew Jędrzejewski-Szmek
10915 * systemd-nspawn will now create /etc/resolv.conf if
10916 necessary, before bind-mounting the host's file onto it.
10918 * systemd-nspawn will now store meta information about a
10919 container on the container's cgroup as extended attribute
10920 fields, including the root directory.
10922 * The cgroup hierarchy has been reworked in many ways. All
10923 objects any of the components systemd creates in the cgroup
10924 tree are now suffixed. More specifically, user sessions are
10925 now placed in cgroups suffixed with ".session", users in
10926 cgroups suffixed with ".user", and nspawn containers in
10927 cgroups suffixed with ".nspawn". Furthermore, all cgroup
10928 names are now escaped in a simple scheme to avoid collision
10929 of userspace object names with kernel filenames. This work
10930 is preparation for making these objects relocatable in the
10931 cgroup tree, in order to allow easy resource partitioning of
10932 these objects without causing naming conflicts.
10934 * systemctl list-dependencies gained the new switches
10935 --plain, --reverse, --after and --before.
10937 * systemd-inhibit now shows the process name of processes that
10938 have taken an inhibitor lock.
10940 * nss-myhostname will now also resolve "localhost"
10941 implicitly. This makes /etc/hosts an optional file and
10942 nicely handles that on IPv6 ::1 maps to both "localhost" and
10943 the local hostname.
10945 * libsystemd-logind.so gained a new call
10946 sd_get_machine_names() to enumerate running containers and
10947 VMs (currently only supported by very new libvirt and
10948 nspawn). sd_login_monitor can now be used to watch
10949 VMs/containers coming and going.
10951 * .include is not allowed recursively anymore, and only in
10952 unit files. Usually it is better to use drop-in snippets in
10953 .d/*.conf anyway, as introduced with systemd 198.
10955 * systemd-analyze gained a new "critical-chain" command that
10956 determines the slowest chain of units run during system
10957 boot-up. It is very useful for tracking down where
10958 optimizing boot time is the most beneficial.
10960 * systemd will no longer allow manipulating service paths in
10961 the name=systemd:/system cgroup tree using ControlGroup= in
10962 units. (But is still fine with it in all other dirs.)
10964 * There's a new systemd-nspawn@.service service file that may
10965 be used to easily run nspawn containers as system
10966 services. With the container's root directory in
10967 /var/lib/container/foobar it is now sufficient to run
10968 "systemctl start systemd-nspawn@foobar.service" to boot it.
10970 * systemd-cgls gained a new parameter "--machine" to list only
10971 the processes within a certain container.
10973 * ConditionSecurity= now can check for "apparmor". We still
10974 are lacking checks for SMACK and IMA for this condition
10975 check though. Patches welcome!
10977 * A new configuration file /etc/systemd/sleep.conf has been
10978 added that may be used to configure which kernel operation
10979 systemd is supposed to execute when "suspend", "hibernate"
10980 or "hybrid-sleep" is requested. This makes the new kernel
10981 "freeze" state accessible to the user.
10983 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
10984 the passed argument if applicable.
10986 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
10987 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
10988 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
10989 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
10990 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
10991 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
10992 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
10997 * The output of 'systemctl list-jobs' got some polishing. The
10998 '--type=' argument may now be passed more than once. A new
10999 command 'systemctl list-sockets' has been added which shows
11000 a list of kernel sockets systemd is listening on with the
11001 socket units they belong to, plus the units these socket
11004 * The experimental libsystemd-bus library got substantial
11005 updates to work in conjunction with the (also experimental)
11006 kdbus kernel project. It works well enough to exchange
11007 messages with some sophistication. Note that kdbus is not
11008 ready yet, and the library is mostly an elaborate test case
11009 for now, and not installable.
11011 * systemd gained a new unit 'systemd-static-nodes.service'
11012 that generates static device nodes earlier during boot, and
11013 can run in conjunction with udev.
11015 * libsystemd-login gained a new call sd_pid_get_user_unit()
11016 to retrieve the user systemd unit a process is running
11017 in. This is useful for systems where systemd is used as
11020 * systemd-nspawn now places all containers in the new /machine
11021 top-level cgroup directory in the name=systemd
11022 hierarchy. libvirt will soon do the same, so that we get a
11023 uniform separation of /system, /user and /machine for system
11024 services, user processes and containers/virtual
11025 machines. This new cgroup hierarchy is also useful to stick
11026 stable names to specific container instances, which can be
11027 recognized later this way (this name may be controlled
11028 via systemd-nspawn's new -M switch). libsystemd-login also
11029 gained a new call sd_pid_get_machine_name() to retrieve the
11030 name of the container/VM a specific process belongs to.
11032 * bootchart can now store its data in the journal.
11034 * libsystemd-journal gained a new call
11035 sd_journal_add_conjunction() for AND expressions to the
11036 matching logic. This can be used to express more complex
11037 logical expressions.
11039 * journactl can now take multiple --unit= and --user-unit=
11042 * The cryptsetup logic now understands the "luks.key=" kernel
11043 command line switch for specifying a file to read the
11044 decryption key from. Also, if a configured key file is not
11045 found the tool will now automatically fall back to prompting
11048 * Python systemd.journal module was updated to wrap recently
11049 added functions from libsystemd-journal. The interface was
11050 changed to bring the low level interface in s.j._Reader
11051 closer to the C API, and the high level interface in
11052 s.j.Reader was updated to wrap and convert all data about
11055 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
11056 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
11057 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
11058 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
11059 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
11060 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
11064 * journalctl --update-catalog now understands a new --root=
11065 option to operate on catalogs found in a different root
11068 * During shutdown after systemd has terminated all running
11069 services a final killing loop kills all remaining left-over
11070 processes. We will now print the name of these processes
11071 when we send SIGKILL to them, since this usually indicates a
11074 * If /etc/crypttab refers to password files stored on
11075 configured mount points automatic dependencies will now be
11076 generated to ensure the specific mount is established first
11077 before the key file is attempted to be read.
11079 * 'systemctl status' will now show information about the
11080 network sockets a socket unit is listening on.
11082 * 'systemctl status' will also shown information about any
11083 drop-in configuration file for units. (Drop-In configuration
11084 files in this context are files such as
11085 /etc/systemd/system/foobar.service.d/*.conf)
11087 * systemd-cgtop now optionally shows summed up CPU times of
11088 cgroups. Press '%' while running cgtop to switch between
11089 percentage and absolute mode. This is useful to determine
11090 which cgroups use up the most CPU time over the entire
11091 runtime of the system. systemd-cgtop has also been updated
11092 to be 'pipeable' for processing with further shell tools.
11094 * 'hostnamectl set-hostname' will now allow setting of FQDN
11097 * The formatting and parsing of time span values has been
11098 changed. The parser now understands fractional expressions
11099 such as "5.5h". The formatter will now output fractional
11100 expressions for all time spans under 1min, i.e. "5.123456s"
11101 rather than "5s 123ms 456us". For time spans under 1s
11102 millisecond values are shown, for those under 1ms
11103 microsecond values are shown. This should greatly improve
11104 all time-related output of systemd.
11106 * libsystemd-login and libsystemd-journal gained new
11107 functions for querying the poll() events mask and poll()
11108 timeout value for integration into arbitrary event
11111 * localectl gained the ability to list available X11 keymaps
11112 (models, layouts, variants, options).
11114 * 'systemd-analyze dot' gained the ability to filter for
11115 specific units via shell-style globs, to create smaller,
11116 more useful graphs. I.e. it is now possible to create simple
11117 graphs of all the dependencies between only target units, or
11118 of all units that Avahi has dependencies with.
11120 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
11121 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
11122 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
11123 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
11124 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
11125 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
11126 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
11130 * The boot-time readahead implementation for rotating media
11131 will now read the read-ahead data in multiple passes which
11132 consist of all read requests made in equidistant time
11133 intervals. This means instead of strictly reading read-ahead
11134 data in its physical order on disk we now try to find a
11135 middle ground between physical and access time order.
11137 * /etc/os-release files gained a new BUILD_ID= field for usage
11138 on operating systems that provide continuous builds of OS
11141 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
11142 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
11143 William Douglas, Zbigniew Jędrzejewski-Szmek
11147 * systemd-python gained an API exposing libsystemd-daemon.
11149 * The SMACK setup logic gained support for uploading CIPSO
11152 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
11153 ReadOnlyDirectories= and InaccessibleDirectories= has
11154 changed. The private /tmp and /var/tmp directories are now
11155 shared by all processes of a service (which means
11156 ExecStartPre= may now leave data in /tmp that ExecStart= of
11157 the same service can still access). When a service is
11158 stopped its temporary directories are immediately deleted
11159 (normal clean-up with tmpfiles is still done in addition to
11162 * By default, systemd will now set a couple of sysctl
11163 variables in the kernel: the safe sysrq options are turned
11164 on, IP route verification is turned on, and source routing
11165 disabled. The recently added hardlink and softlink
11166 protection of the kernel is turned on. These settings should
11167 be reasonably safe, and good defaults for all new systems.
11169 * The predictable network naming logic may now be turned off
11170 with a new kernel command line switch: net.ifnames=0.
11172 * A new libsystemd-bus module has been added that implements a
11173 pretty complete D-Bus client library. For details see:
11175 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
11177 * journald will now explicitly flush the journal files to disk
11178 at the latest 5min after each write. The file will then also
11179 be marked offline until the next write. This should increase
11180 reliability in case of a crash. The synchronization delay
11181 can be configured via SyncIntervalSec= in journald.conf.
11183 * There's a new remote-fs-setup.target unit that can be used
11184 to pull in specific services when at least one remote file
11185 system is to be mounted.
11187 * There are new targets timers.target and paths.target as
11188 canonical targets to pull user timer and path units in
11189 from. This complements sockets.target with a similar
11190 purpose for socket units.
11192 * libudev gained a new call udev_device_set_attribute_value()
11193 to set sysfs attributes of a device.
11195 * The udev daemon now sets the default number of worker
11196 processes executed in parallel based on the number of available
11197 CPUs instead of the amount of available RAM. This is supposed
11198 to provide a more reliable default and limit a too aggressive
11199 parallelism for setups with 1000s of devices connected.
11201 Contributions from: Auke Kok, Colin Walters, Cristian
11202 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
11203 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
11204 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
11205 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
11206 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
11207 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
11208 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
11209 Zbigniew Jędrzejewski-Szmek
11213 * Configuration of unit files may now be extended via drop-in
11214 files without having to edit/override the unit files
11215 themselves. More specifically, if the administrator wants to
11216 change one value for a service file foobar.service he can
11217 now do so by dropping in a configuration snippet into
11218 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
11219 will load all these snippets and apply them on top of the
11220 main unit configuration file, possibly extending or
11221 overriding its settings. Using these drop-in snippets is
11222 generally nicer than the two earlier options for changing
11223 unit files locally: copying the files from
11224 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
11225 them there; or creating a new file in /etc/systemd/system/
11226 that incorporates the original one via ".include". Drop-in
11227 snippets into these .d/ directories can be placed in any
11228 directory systemd looks for units in, and the usual
11229 overriding semantics between /usr/lib, /etc and /run apply
11232 * Most unit file settings which take lists of items can now be
11233 reset by assigning the empty string to them. For example,
11234 normally, settings such as Environment=FOO=BAR append a new
11235 environment variable assignment to the environment block,
11236 each time they are used. By assigning Environment= the empty
11237 string the environment block can be reset to empty. This is
11238 particularly useful with the .d/*.conf drop-in snippets
11239 mentioned above, since this adds the ability to reset list
11240 settings from vendor unit files via these drop-ins.
11242 * systemctl gained a new "list-dependencies" command for
11243 listing the dependencies of a unit recursively.
11245 * Inhibitors are now honored and listed by "systemctl
11246 suspend", "systemctl poweroff" (and similar) too, not only
11247 GNOME. These commands will also list active sessions by
11250 * Resource limits (as exposed by the various control group
11251 controllers) can now be controlled dynamically at runtime
11252 for all units. More specifically, you can now use a command
11253 like "systemctl set-cgroup-attr foobar.service cpu.shares
11254 2000" to alter the CPU shares a specific service gets. These
11255 settings are stored persistently on disk, and thus allow the
11256 administrator to easily adjust the resource usage of
11257 services with a few simple commands. This dynamic resource
11258 management logic is also available to other programs via the
11259 bus. Almost any kernel cgroup attribute and controller is
11262 * systemd-vconsole-setup will now copy all font settings to
11263 all allocated VTs, where it previously applied them only to
11266 * libsystemd-login gained the new sd_session_get_tty() API
11269 * This release drops support for a few legacy or
11270 distribution-specific LSB facility names when parsing init
11271 scripts: $x-display-manager, $mail-transfer-agent,
11272 $mail-transport-agent, $mail-transfer-agent, $smtp,
11273 $null. Also, the mail-transfer-agent.target unit backing
11274 this has been removed. Distributions which want to retain
11275 compatibility with this should carry the burden for
11276 supporting this themselves and patch support for these back
11277 in, if they really need to. Also, the facilities $syslog and
11278 $local_fs are now ignored, since systemd does not support
11279 early-boot LSB init scripts anymore, and these facilities
11280 are implied anyway for normal services. syslog.target has
11283 * There are new bus calls on PID1's Manager object for
11284 cancelling jobs, and removing snapshot units. Previously,
11285 both calls were only available on the Job and Snapshot
11286 objects themselves.
11288 * systemd-journal-gatewayd gained SSL support.
11290 * The various "environment" files, such as /etc/locale.conf
11291 now support continuation lines with a backslash ("\") as
11292 last character in the line, similarly in style (but different)
11293 to how this is supported in shells.
11295 * For normal user processes the _SYSTEMD_USER_UNIT= field is
11296 now implicitly appended to every log entry logged. systemctl
11297 has been updated to filter by this field when operating on a
11298 user systemd instance.
11300 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
11301 CAP_AUDIT_CONTROL capabilities to the capabilities set for
11302 the container. This makes it easier to boot unmodified
11303 Fedora systems in a container, which however still requires
11304 audit=0 to be passed on the kernel command line. Auditing in
11305 kernel and userspace is unfortunately still too broken in
11306 context of containers, hence we recommend compiling it out
11307 of the kernel or using audit=0. Hopefully this will be fixed
11308 one day for good in the kernel.
11310 * nspawn gained the new --bind= and --bind-ro= parameters to
11311 bind mount specific directories from the host into the
11314 * nspawn will now mount its own devpts file system instance
11315 into the container, in order not to leak pty devices from
11316 the host into the container.
11318 * systemd will now read the firmware boot time performance
11319 information from the EFI variables, if the used boot loader
11320 supports this, and takes it into account for boot performance
11321 analysis via "systemd-analyze". This is currently supported
11322 only in conjunction with Gummiboot, but could be supported
11323 by other boot loaders too. For details see:
11325 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
11327 * A new generator has been added that automatically mounts the
11328 EFI System Partition (ESP) to /boot, if that directory
11329 exists, is empty, and no other file system has been
11330 configured to be mounted there.
11332 * logind will now send out PrepareForSleep(false) out
11333 unconditionally, after coming back from suspend. This may be
11334 used by applications as asynchronous notification for
11335 system resume events.
11337 * "systemctl unlock-sessions" has been added, that allows
11338 unlocking the screens of all user sessions at once, similar
11339 to how "systemctl lock-sessions" already locked all users
11340 sessions. This is backed by a new D-Bus call UnlockSessions().
11342 * "loginctl seat-status" will now show the master device of a
11343 seat. (i.e. the device of a seat that needs to be around for
11344 the seat to be considered available, usually the graphics
11347 * tmpfiles gained a new "X" line type, that allows
11348 configuration of files and directories (with wildcards) that
11349 shall be excluded from automatic cleanup ("aging").
11351 * udev default rules set the device node permissions now only
11352 at "add" events, and do not change them any longer with a
11353 later "change" event.
11355 * The log messages for lid events and power/sleep keypresses
11356 now carry a message ID.
11358 * We now have a substantially larger unit test suite, but this
11359 continues to be work in progress.
11361 * udevadm hwdb gained a new --root= parameter to change the
11362 root directory to operate relative to.
11364 * logind will now issue a background sync() request to the kernel
11365 early at shutdown, so that dirty buffers are flushed to disk early
11366 instead of at the last moment, in order to optimize shutdown
11369 * A new bootctl tool has been added that is an interface for
11370 certain boot loader operations. This is currently a preview
11371 and is likely to be extended into a small mechanism daemon
11372 like timedated, localed, hostnamed, and can be used by
11373 graphical UIs to enumerate available boot options, and
11374 request boot into firmware operations.
11376 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
11377 the rest of the package. It also has been updated to work
11378 correctly in initrds.
11380 * polkit previously has been runtime optional, and is now also
11381 compile time optional via a configure switch.
11383 * systemd-analyze has been reimplemented in C. Also "systemctl
11384 dot" has moved into systemd-analyze.
11386 * "systemctl status" with no further parameters will now print
11387 the status of all active or failed units.
11389 * Operations such as "systemctl start" can now be executed
11390 with a new mode "--irreversible" which may be used to queue
11391 operations that cannot accidentally be reversed by a later
11392 job queuing. This is by default used to make shutdown
11393 requests more robust.
11395 * The Python API of systemd now gained a new module for
11396 reading journal files.
11398 * A new tool kernel-install has been added that can install
11399 kernel images according to the Boot Loader Specification:
11401 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
11403 * Boot time console output has been improved to provide
11404 animated boot time output for hanging jobs.
11406 * A new tool systemd-activate has been added which can be used
11407 to test socket activation with, directly from the command
11408 line. This should make it much easier to test and debug
11409 socket activation in daemons.
11411 * journalctl gained a new "--reverse" (or -r) option to show
11412 journal output in reverse order (i.e. newest line first).
11414 * journalctl gained a new "--pager-end" (or -e) option to jump
11415 to immediately jump to the end of the journal in the
11416 pager. This is only supported in conjunction with "less".
11418 * journalctl gained a new "--user-unit=" option, that works
11419 similarly to "--unit=" but filters for user units rather than
11422 * A number of unit files to ease adoption of systemd in
11423 initrds has been added. This moves some minimal logic from
11424 the various initrd implementations into systemd proper.
11426 * The journal files are now owned by a new group
11427 "systemd-journal", which exists specifically to allow access
11428 to the journal, and nothing else. Previously, we used the
11429 "adm" group for that, which however possibly covers more
11430 than just journal/log file access. This new group is now
11431 already used by systemd-journal-gatewayd to ensure this
11432 daemon gets access to the journal files and as little else
11433 as possible. Note that "make install" will also set FS ACLs
11434 up for /var/log/journal to give "adm" and "wheel" read
11435 access to it, in addition to "systemd-journal" which owns
11436 the journal files. We recommend that packaging scripts also
11437 add read access to "adm" + "wheel" to /var/log/journal, and
11438 all existing/future journal files. To normal users and
11439 administrators little changes, however packagers need to
11440 ensure to create the "systemd-journal" system group at
11441 package installation time.
11443 * The systemd-journal-gatewayd now runs as unprivileged user
11444 systemd-journal-gateway:systemd-journal-gateway. Packaging
11445 scripts need to create these system user/group at
11448 * timedated now exposes a new boolean property CanNTP that
11449 indicates whether a local NTP service is available or not.
11451 * systemd-detect-virt will now also detect xen PVs
11453 * The pstore file system is now mounted by default, if it is
11456 * In addition to the SELinux and IMA policies we will now also
11457 load SMACK policies at early boot.
11459 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
11460 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
11461 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
11462 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
11463 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
11464 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
11465 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
11466 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
11467 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
11468 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
11469 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
11470 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
11471 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
11472 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
11476 * Timer units now support calendar time events in addition to
11477 monotonic time events. That means you can now trigger a unit
11478 based on a calendar time specification such as "Thu,Fri
11479 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
11480 or fifth day of any month of the year 2013, given that it is
11481 a thursday or friday. This brings timer event support
11482 considerably closer to cron's capabilities. For details on
11483 the supported calendar time specification language see
11486 * udev now supports a number of different naming policies for
11487 network interfaces for predictable names, and a combination
11488 of these policies is now the default. Please see this wiki
11489 document for details:
11491 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
11493 * Auke Kok's bootchart implementation has been added to the
11494 systemd tree. It is an optional component that can graph the
11495 boot in quite some detail. It is one of the best bootchart
11496 implementations around and minimal in its code and
11499 * nss-myhostname has been integrated into the systemd source
11500 tree. nss-myhostname guarantees that the local hostname
11501 always stays resolvable via NSS. It has been a weak
11502 requirement of systemd-hostnamed since a long time, and
11503 since its code is actually trivial we decided to just
11504 include it in systemd's source tree. It can be turned off
11505 with a configure switch.
11507 * The read-ahead logic is now capable of properly detecting
11508 whether a btrfs file system is on SSD or rotating media, in
11509 order to optimize the read-ahead scheme. Previously, it was
11510 only capable of detecting this on traditional file systems
11513 * In udev, additional device properties are now read from the
11514 IAB in addition to the OUI database. Also, Bluetooth company
11515 identities are attached to the devices as well.
11517 * In service files %U may be used as specifier that is
11518 replaced by the configured user name of the service.
11520 * nspawn may now be invoked without a controlling TTY. This
11521 makes it suitable for invocation as its own service. This
11522 may be used to set up a simple containerized server system
11523 using only core OS tools.
11525 * systemd and nspawn can now accept socket file descriptors
11526 when they are started for socket activation. This enables
11527 implementation of socket activated nspawn
11528 containers. i.e. think about autospawning an entire OS image
11529 when the first SSH or HTTP connection is received. We expect
11530 that similar functionality will also be added to libvirt-lxc
11533 * journalctl will now suppress ANSI color codes when
11534 presenting log data.
11536 * systemctl will no longer show control group information for
11537 a unit if the control group is empty anyway.
11539 * logind can now automatically suspend/hibernate/shutdown the
11542 * /etc/machine-info and hostnamed now also expose the chassis
11543 type of the system. This can be used to determine whether
11544 the local system is a laptop, desktop, handset or
11545 tablet. This information may either be configured by the
11546 user/vendor or is automatically determined from ACPI and DMI
11547 information if possible.
11549 * A number of polkit actions are now bound together with "imply"
11550 rules. This should simplify creating UIs because many actions
11551 will now authenticate similar ones as well.
11553 * Unit files learnt a new condition ConditionACPower= which
11554 may be used to conditionalize a unit depending on whether an
11555 AC power source is connected or not, of whether the system
11556 is running on battery power.
11558 * systemctl gained a new "is-failed" verb that may be used in
11559 shell scripts and suchlike to check whether a specific unit
11560 is in the "failed" state.
11562 * The EnvironmentFile= setting in unit files now supports file
11563 globbing, and can hence be used to easily read a number of
11564 environment files at once.
11566 * systemd will no longer detect and recognize specific
11567 distributions. All distribution-specific #ifdeffery has been
11568 removed, systemd is now fully generic and
11569 distribution-agnostic. Effectively, not too much is lost as
11570 a lot of the code is still accessible via explicit configure
11571 switches. However, support for some distribution specific
11572 legacy configuration file formats has been dropped. We
11573 recommend distributions to simply adopt the configuration
11574 files everybody else uses now and convert the old
11575 configuration from packaging scripts. Most distributions
11576 already did that. If that's not possible or desirable,
11577 distributions are welcome to forward port the specific
11578 pieces of code locally from the git history.
11580 * When logging a message about a unit systemd will now always
11581 log the unit name in the message meta data.
11583 * localectl will now also discover system locale data that is
11584 not stored in locale archives, but directly unpacked.
11586 * logind will no longer unconditionally use framebuffer
11587 devices as seat masters, i.e. as devices that are required
11588 to be existing before a seat is considered preset. Instead,
11589 it will now look for all devices that are tagged as
11590 "seat-master" in udev. By default, framebuffer devices will
11591 be marked as such, but depending on local systems, other
11592 devices might be marked as well. This may be used to
11593 integrate graphics cards using closed source drivers (such
11594 as NVidia ones) more nicely into logind. Note however, that
11595 we recommend using the open source NVidia drivers instead,
11596 and no udev rules for the closed-source drivers will be
11597 shipped from us upstream.
11599 Contributions from: Adam Williamson, Alessandro Crismani, Auke
11600 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
11601 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
11602 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
11603 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
11604 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
11605 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
11606 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
11607 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
11608 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
11609 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
11610 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
11615 * udev gained support for loading additional device properties
11616 from an indexed database that is keyed by vendor/product IDs
11617 and similar device identifiers. For the beginning this
11618 "hwdb" is populated with data from the well-known PCI and
11619 USB database, but also includes PNP, ACPI and OID data. In
11620 the longer run this indexed database shall grow into
11621 becoming the one central database for non-essential
11622 userspace device metadata. Previously, data from the PCI/USB
11623 database was only attached to select devices, since the
11624 lookup was a relatively expensive operation due to O(n) time
11625 complexity (with n being the number of entries in the
11626 database). Since this is now O(1), we decided to add in this
11627 data for all devices where this is available, by
11628 default. Note that the indexed database needs to be rebuilt
11629 when new data files are installed. To achieve this you need
11630 to update your packaging scripts to invoke "udevadm hwdb
11631 --update" after installation of hwdb data files. For
11632 RPM-based distributions we introduced the new
11633 %udev_hwdb_update macro for this purpose.
11635 * The Journal gained support for the "Message Catalog", an
11636 indexed database to link up additional information with
11637 journal entries. For further details please check:
11639 https://www.freedesktop.org/wiki/Software/systemd/catalog
11641 The indexed message catalog database also needs to be
11642 rebuilt after installation of message catalog files. Use
11643 "journalctl --update-catalog" for this. For RPM-based
11644 distributions we introduced the %journal_catalog_update
11645 macro for this purpose.
11647 * The Python Journal bindings gained support for the standard
11648 Python logging framework.
11650 * The Journal API gained new functions for checking whether
11651 the underlying file system of a journal file is capable of
11652 properly reporting file change notifications, or whether
11653 applications that want to reflect journal changes "live"
11654 need to recheck journal files continuously in appropriate
11657 * It is now possible to set the "age" field for tmpfiles
11658 entries to 0, indicating that files matching this entry
11659 shall always be removed when the directories are cleaned up.
11661 * coredumpctl gained a new "gdb" verb which invokes gdb
11662 right-away on the selected coredump.
11664 * There's now support for "hybrid sleep" on kernels that
11665 support this, in addition to "suspend" and "hibernate". Use
11666 "systemctl hybrid-sleep" to make use of this.
11668 * logind's HandleSuspendKey= setting (and related settings)
11669 now gained support for a new "lock" setting to simply
11670 request the screen lock on all local sessions, instead of
11671 actually executing a suspend or hibernation.
11673 * systemd will now mount the EFI variables file system by
11676 * Socket units now gained support for configuration of the
11677 SMACK security label.
11679 * timedatectl will now output the time of the last and next
11680 daylight saving change.
11682 * We dropped support for various legacy and distro-specific
11683 concepts, such as insserv, early-boot SysV services
11684 (i.e. those for non-standard runlevels such as 'b' or 'S')
11685 or ArchLinux /etc/rc.conf support. We recommend the
11686 distributions who still need support this to either continue
11687 to maintain the necessary patches downstream, or find a
11688 different solution. (Talk to us if you have questions!)
11690 * Various systemd components will now bypass polkit checks for
11691 root and otherwise handle properly if polkit is not found to
11692 be around. This should fix most issues for polkit-less
11693 systems. Quite frankly this should have been this way since
11694 day one. It is absolutely our intention to make systemd work
11695 fine on polkit-less systems, and we consider it a bug if
11696 something does not work as it should if polkit is not around.
11698 * For embedded systems it is now possible to build udev and
11699 systemd without blkid and/or kmod support.
11701 * "systemctl switch-root" is now capable of switching root
11702 more than once. I.e. in addition to transitions from the
11703 initrd to the host OS it is now possible to transition to
11704 further OS images from the host. This is useful to implement
11705 offline updating tools.
11707 * Various other additions have been made to the RPM macros
11708 shipped with systemd. Use %udev_rules_update() after
11709 installing new udev rules files. %_udevhwdbdir,
11710 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
11711 %_sysctldir are now available which resolve to the right
11712 directories for packages to place various data files in.
11714 * journalctl gained the new --full switch (in addition to
11715 --all, to disable ellipsation for long messages.
11717 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
11718 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
11719 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
11720 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
11721 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
11722 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
11723 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
11724 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
11725 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
11729 * journalctl gained new --since= and --until= switches to
11730 filter by time. It also now supports nice filtering for
11731 units via --unit=/-u.
11733 * Type=oneshot services may use ExecReload= and do the
11736 * The journal daemon now supports time-based rotation and
11737 vacuuming, in addition to the usual disk-space based
11740 * The journal will now index the available field values for
11741 each field name. This enables clients to show pretty drop
11742 downs of available match values when filtering. The bash
11743 completion of journalctl has been updated
11744 accordingly. journalctl gained a new switch -F to list all
11745 values a certain field takes in the journal database.
11747 * More service events are now written as structured messages
11748 to the journal, and made recognizable via message IDs.
11750 * The timedated, localed and hostnamed mini-services which
11751 previously only provided support for changing time, locale
11752 and hostname settings from graphical DEs such as GNOME now
11753 also have a minimal (but very useful) text-based client
11754 utility each. This is probably the nicest way to changing
11755 these settings from the command line now, especially since
11756 it lists available options and is fully integrated with bash
11759 * There's now a new tool "systemd-coredumpctl" to list and
11760 extract coredumps from the journal.
11762 * We now install a README each in /var/log/ and
11763 /etc/rc.d/init.d explaining where the system logs and init
11764 scripts went. This hopefully should help folks who go to
11765 that dirs and look into the otherwise now empty void and
11766 scratch their heads.
11768 * When user-services are invoked (by systemd --user) the
11769 $MANAGERPID env var is set to the PID of systemd.
11771 * SIGRTMIN+24 when sent to a --user instance will now result
11772 in immediate termination of systemd.
11774 * gatewayd received numerous feature additions such as a
11775 "follow" mode, for live syncing and filtering.
11777 * browse.html now allows filtering and showing detailed
11778 information on specific entries. Keyboard navigation and
11779 mouse screen support has been added.
11781 * gatewayd/journalctl now supports HTML5/JSON
11782 Server-Sent-Events as output.
11784 * The SysV init script compatibility logic will now
11785 heuristically determine whether a script supports the
11786 "reload" verb, and only then make this available as
11787 "systemctl reload".
11789 * "systemctl status --follow" has been removed, use "journalctl
11792 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
11793 have been removed since they are hardly useful to be
11796 * And I'd like to take the opportunity to specifically mention
11797 Zbigniew for his great contributions. Zbigniew, you rock!
11799 Contributions from: Andrew Eikum, Christian Hesse, Colin
11800 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
11801 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
11802 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
11803 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
11804 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
11805 Jędrzejewski-Szmek, Сковорода Никита Андреевич
11809 * If /etc/vconsole.conf is non-existent or empty we will no
11810 longer load any console font or key map at boot by
11811 default. Instead the kernel defaults will be left
11812 intact. This is definitely the right thing to do, as no
11813 configuration should mean no configuration, and hard-coding
11814 font names that are different on all archs is probably a bad
11815 idea. Also, the kernel default key map and font should be
11816 good enough for most cases anyway, and mostly identical to
11817 the userspace fonts/key maps we previously overloaded them
11818 with. If distributions want to continue to default to a
11819 non-kernel font or key map they should ship a default
11820 /etc/vconsole.conf with the appropriate contents.
11822 Contributions from: Colin Walters, Daniel J Walsh, Dave
11823 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
11824 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
11828 * journalctl gained a new --cursor= switch to show entries
11829 starting from the specified location in the journal.
11831 * We now enforce a size limit on journal entry fields exported
11832 with "-o json" in journalctl. Fields larger than 4K will be
11833 assigned null. This can be turned off with --all.
11835 * An (optional) journal gateway daemon is now available as
11836 "systemd-journal-gatewayd.service". This service provides
11837 access to the journal via HTTP and JSON. This functionality
11838 will be used to implement live log synchronization in both
11839 pull and push modes, but has various other users too, such
11840 as easy log access for debugging of embedded devices. Right
11841 now it is already useful to retrieve the journal via HTTP:
11843 # systemctl start systemd-journal-gatewayd.service
11844 # wget http://localhost:19531/entries
11846 This will download the journal contents in a
11847 /var/log/messages compatible format. The same as JSON:
11849 # curl -H"Accept: application/json" http://localhost:19531/entries
11851 This service is also accessible via a web browser where a
11852 single static HTML5 app is served that uses the JSON logic
11853 to enable the user to do some basic browsing of the
11854 journal. This will be extended later on. Here's an example
11855 screenshot of this app in its current state:
11857 http://0pointer.de/public/journal-gatewayd
11859 Contributions from: Kay Sievers, Lennart Poettering, Robert
11860 Milasan, Tom Gundersen
11864 * The bash completion logic is now available for journalctl
11867 * We do not mount the "cpuset" controller anymore together with
11868 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
11869 started if no parameters are assigned to it. "cpuset" hence
11870 broke code that assumed it could create "cpu" groups and
11873 * journalctl -f will now subscribe to terminal size changes,
11874 and line break accordingly.
11876 Contributions from: Dave Reisner, Kay Sievers, Lennart
11877 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
11881 * nspawn will now create a symlink /etc/localtime in the
11882 container environment, copying the host's timezone
11883 setting. Previously this has been done via a bind mount, but
11884 since symlinks cannot be bind mounted this has now been
11885 changed to create/update the appropriate symlink.
11887 * journalctl -n's line number argument is now optional, and
11888 will default to 10 if omitted.
11890 * journald will now log the maximum size the journal files may
11891 take up on disk. This is particularly useful if the default
11892 built-in logic of determining this parameter from the file
11893 system size is used. Use "systemctl status
11894 systemd-journald.service" to see this information.
11896 * The multi-seat X wrapper tool has been stripped down. As X
11897 is now capable of enumerating graphics devices via udev in a
11898 seat-aware way the wrapper is not strictly necessary
11899 anymore. A stripped down temporary stop-gap is still shipped
11900 until the upstream display managers have been updated to
11901 fully support the new X logic. Expect this wrapper to be
11902 removed entirely in one of the next releases.
11904 * HandleSleepKey= in logind.conf has been split up into
11905 HandleSuspendKey= and HandleHibernateKey=. The old setting
11906 is not available anymore. X11 and the kernel are
11907 distinguishing between these keys and we should too. This
11908 also means the inhibition lock for these keys has been split
11911 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
11912 Poettering, Lukas Nykryn, Václav Pavlín
11916 * Whenever a unit changes state we will now log this to the
11917 journal and show along the unit's own log output in
11918 "systemctl status".
11920 * ConditionPathIsMountPoint= can now properly detect bind
11921 mount points too. (Previously, a bind mount of one file
11922 system to another place in the same file system could not be
11923 detected as mount, since they shared struct stat's st_dev
11926 * We will now mount the cgroup controllers cpu, cpuacct,
11927 cpuset and the controllers net_cls, net_prio together by
11930 * nspawn containers will now have a virtualized boot
11931 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
11932 over with a randomized ID at container initialization). This
11933 has the effect of making "journalctl -b" do the right thing
11936 * The JSON output journal serialization has been updated not
11937 to generate "endless" list objects anymore, but rather one
11938 JSON object per line. This is more in line how most JSON
11939 parsers expect JSON objects. The new output mode
11940 "json-pretty" has been added to provide similar output, but
11941 neatly aligned for readability by humans.
11943 * We dropped all explicit sync() invocations in the shutdown
11944 code. The kernel does this implicitly anyway in the kernel
11945 reboot() syscall. halt(8)'s -n option is now a compatibility
11948 * We now support virtualized reboot() in containers, as
11949 supported by newer kernels. We will fall back to exit() if
11950 CAP_SYS_REBOOT is not available to the container. Also,
11951 nspawn makes use of this now and will actually reboot the
11952 container if the containerized OS asks for that.
11954 * journalctl will only show local log output by default
11955 now. Use --merge (-m) to show remote log output, too.
11957 * libsystemd-journal gained the new sd_journal_get_usage()
11958 call to determine the current disk usage of all journal
11959 files. This is exposed in the new "journalctl --disk-usage"
11962 * journald gained a new configuration setting SplitMode= in
11963 journald.conf which may be used to control how user journals
11964 are split off. See journald.conf(5) for details.
11966 * A new condition type ConditionFileNotEmpty= has been added.
11968 * tmpfiles' "w" lines now support file globbing, to write
11969 multiple files at once.
11971 * We added Python bindings for the journal submission
11972 APIs. More Python APIs for a number of selected APIs will
11973 likely follow. Note that we intend to add native bindings
11974 only for the Python language, as we consider it common
11975 enough to deserve bindings shipped within systemd. There are
11976 various projects outside of systemd that provide bindings
11977 for languages such as PHP or Lua.
11979 * Many conditions will now resolve specifiers such as %i. In
11980 addition, PathChanged= and related directives of .path units
11981 now support specifiers as well.
11983 * There's now a new RPM macro definition for the system preset
11986 * journald will now warn if it ca not forward a message to the
11987 syslog daemon because its socket is full.
11989 * timedated will no longer write or process /etc/timezone,
11990 except on Debian. As we do not support late mounted /usr
11991 anymore /etc/localtime always being a symlink is now safe,
11992 and hence the information in /etc/timezone is not necessary
11995 * logind will now always reserve one VT for a text getty (VT6
11996 by default). Previously if more than 6 X sessions where
11997 started they took up all the VTs with auto-spawned gettys,
11998 so that no text gettys were available anymore.
12000 * udev will now automatically inform the btrfs kernel logic
12001 about btrfs RAID components showing up. This should make
12002 simple hotplug based btrfs RAID assembly work.
12004 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
12005 (but not for its children which will stay at the kernel
12006 default). This should allow setups with a lot more listening
12009 * systemd will now always pass the configured timezone to the
12010 kernel at boot. timedated will do the same when the timezone
12013 * logind's inhibition logic has been updated. By default,
12014 logind will now handle the lid switch, the power and sleep
12015 keys all the time, even in graphical sessions. If DEs want
12016 to handle these events on their own they should take the new
12017 handle-power-key, handle-sleep-key and handle-lid-switch
12018 inhibitors during their runtime. A simple way to achieve
12019 that is to invoke the DE wrapped in an invocation of:
12021 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch …
12023 * Access to unit operations is now checked via SELinux taking
12024 the unit file label and client process label into account.
12026 * systemd will now notify the administrator in the journal
12027 when he over-mounts a non-empty directory.
12029 * There are new specifiers that are resolved in unit files,
12030 for the hostname (%H), the machine ID (%m) and the boot ID
12033 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
12034 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
12035 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
12036 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
12037 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
12038 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
12039 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
12043 * Support for reading structured kernel messages from
12044 /dev/kmsg has now been added and is enabled by default.
12046 * Support for reading kernel messages from /proc/kmsg has now
12047 been removed. If you want kernel messages in the journal
12048 make sure to run a recent kernel (>= 3.5) that supports
12049 reading structured messages from /dev/kmsg (see
12050 above). /proc/kmsg is now exclusive property of classic
12051 syslog daemons again.
12053 * The libudev API gained the new
12054 udev_device_new_from_device_id() call.
12056 * The logic for file system namespace (ReadOnlyDirectory=,
12057 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
12058 require pivot_root() anymore. This means fewer temporary
12059 directories are created below /tmp for this feature.
12061 * nspawn containers will now see and receive all submounts
12062 made on the host OS below the root file system of the
12065 * Forward Secure Sealing is now supported for Journal files,
12066 which provide cryptographical sealing of journal files so
12067 that attackers cannot alter log history anymore without this
12068 being detectable. Lennart will soon post a blog story about
12069 this explaining it in more detail.
12071 * There are two new service settings RestartPreventExitStatus=
12072 and SuccessExitStatus= which allow configuration of exit
12073 status (exit code or signal) which will be excepted from the
12074 restart logic, resp. consider successful.
12076 * journalctl gained the new --verify switch that can be used
12077 to check the integrity of the structure of journal files and
12078 (if Forward Secure Sealing is enabled) the contents of
12081 * nspawn containers will now be run with /dev/stdin, /dev/fd/
12082 and similar symlinks pre-created. This makes running shells
12083 as container init process a lot more fun.
12085 * The fstab support can now handle PARTUUID= and PARTLABEL=
12088 * A new ConditionHost= condition has been added to match
12089 against the hostname (with globs) and machine ID. This is
12090 useful for clusters where a single OS image is used to
12091 provision a large number of hosts which shall run slightly
12092 different sets of services.
12094 * Services which hit the restart limit will now be placed in a
12097 Contributions from: Bertram Poettering, Dave Reisner, Huang
12098 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
12099 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
12103 * When running in --user mode systemd will now become a
12104 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
12105 tree a lot more organized.
12107 * A new PartOf= unit dependency type has been introduced that
12108 may be used to group services in a natural way.
12110 * "systemctl enable" may now be used to enable instances of
12113 * journalctl now prints error log levels in red, and
12114 warning/notice log levels in bright white. It also supports
12115 filtering by log level now.
12117 * cgtop gained a new -n switch (similar to top), to configure
12118 the maximum number of iterations to run for. It also gained
12119 -b, to run in batch mode (accepting no input).
12121 * The suffix ".service" may now be omitted on most systemctl
12122 command lines involving service unit names.
12124 * There's a new bus call in logind to lock all sessions, as
12125 well as a loginctl verb for it "lock-sessions".
12127 * libsystemd-logind.so gained a new call sd_journal_perror()
12128 that works similar to libc perror() but logs to the journal
12129 and encodes structured information about the error number.
12131 * /etc/crypttab entries now understand the new keyfile-size=
12134 * shutdown(8) now can send a (configurable) wall message when
12135 a shutdown is cancelled.
12137 * The mount propagation mode for the root file system will now
12138 default to "shared", which is useful to make containers work
12139 nicely out-of-the-box so that they receive new mounts from
12140 the host. This can be undone locally by running "mount
12141 --make-rprivate /" if needed.
12143 * The prefdm.service file has been removed. Distributions
12144 should maintain this unit downstream if they intend to keep
12145 it around. However, we recommend writing normal unit files
12146 for display managers instead.
12148 * Since systemd is a crucial part of the OS we will now
12149 default to a number of compiler switches that improve
12150 security (hardening) such as read-only relocations, stack
12151 protection, and suchlike.
12153 * The TimeoutSec= setting for services is now split into
12154 TimeoutStartSec= and TimeoutStopSec= to allow configuration
12155 of individual time outs for the start and the stop phase of
12158 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
12159 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
12160 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
12161 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
12162 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
12163 Gundersen, Zbigniew Jędrzejewski-Szmek
12167 * The journal and id128 C APIs are now fully documented as man
12170 * Extra safety checks have been added when transitioning from
12171 the initial RAM disk to the main system to avoid accidental
12174 * /etc/crypttab entries now understand the new keyfile-offset=
12177 * systemctl -t can now be used to filter by unit load state.
12179 * The journal C API gained the new sd_journal_wait() call to
12180 make writing synchronous journal clients easier.
12182 * journalctl gained the new -D switch to show journals from a
12183 specific directory.
12185 * journalctl now displays a special marker between log
12186 messages of two different boots.
12188 * The journal is now explicitly flushed to /var via a service
12189 systemd-journal-flush.service, rather than implicitly simply
12190 by seeing /var/log/journal to be writable.
12192 * journalctl (and the journal C APIs) can now match for much
12193 more complex expressions, with alternatives and
12196 * When transitioning from the initial RAM disk to the main
12197 system we will now kill all processes in a killing spree to
12198 ensure no processes stay around by accident.
12200 * Three new specifiers may be used in unit files: %u, %h, %s
12201 resolve to the user name, user home directory resp. user
12202 shell. This is useful for running systemd user instances.
12204 * We now automatically rotate journal files if their data
12205 object hash table gets a fill level > 75%. We also size the
12206 hash table based on the configured maximum file size. This
12207 together should lower hash collisions drastically and thus
12208 speed things up a bit.
12210 * journalctl gained the new "--header" switch to introspect
12211 header data of journal files.
12213 * A new setting SystemCallFilters= has been added to services which may
12214 be used to apply deny lists or allow lists to system calls. This is
12215 based on SECCOMP Mode 2 of Linux 3.5.
12217 * nspawn gained a new --link-journal= switch (and quicker: -j)
12218 to link the container journal with the host. This makes it
12219 very easy to centralize log viewing on the host for all
12220 guests while still keeping the journal files separated.
12222 * Many bugfixes and optimizations
12224 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
12225 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
12226 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
12231 * Several tools now understand kernel command line arguments,
12232 which are only read when run in an initial RAM disk. They
12233 usually follow closely their normal counterparts, but are
12236 * There's a new tool to analyze the readahead files that are
12237 automatically generated at boot. Use:
12239 /usr/lib/systemd/systemd-readahead analyze /.readahead
12241 * We now provide an early debug shell on tty9 if this enabled. Use:
12243 systemctl enable debug-shell.service
12245 * All plymouth related units have been moved into the Plymouth
12246 package. Please make sure to upgrade your Plymouth version
12249 * systemd-tmpfiles now supports getting passed the basename of
12250 a configuration file only, in which case it will look for it
12251 in all appropriate directories automatically.
12253 * udevadm info now takes a /dev or /sys path as argument, and
12254 does the right thing. Example:
12256 udevadm info /dev/sda
12257 udevadm info /sys/class/block/sda
12259 * systemctl now prints a warning if a unit is stopped but a
12260 unit that might trigger it continues to run. Example: a
12261 service is stopped but the socket that activates it is left
12264 * "systemctl status" will now mention if the log output was
12265 shortened due to rotation since a service has been started.
12267 * The journal API now exposes functions to determine the
12268 "cutoff" times due to rotation.
12270 * journald now understands SIGUSR1 and SIGUSR2 for triggering
12271 immediately flushing of runtime logs to /var if possible,
12272 resp. for triggering immediate rotation of the journal
12275 * It is now considered an error if a service is attempted to
12276 be stopped that is not loaded.
12278 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
12280 * systemd-analyze now supports Python 3
12282 * tmpfiles now supports cleaning up directories via aging
12283 where the first level dirs are always kept around but
12284 directories beneath it automatically aged. This is enabled
12285 by prefixing the age field with '~'.
12287 * Seat objects now expose CanGraphical, CanTTY properties
12288 which is required to deal with very fast bootups where the
12289 display manager might be running before the graphics drivers
12290 completed initialization.
12292 * Seat objects now expose a State property.
12294 * We now include RPM macros for service enabling/disabling
12295 based on the preset logic. We recommend RPM based
12296 distributions to make use of these macros if possible. This
12297 makes it simpler to reuse RPM spec files across
12300 * We now make sure that the collected systemd unit name is
12301 always valid when services log to the journal via
12304 * There's a new man page kernel-command-line(7) detailing all
12305 command line options we understand.
12307 * The fstab generator may now be disabled at boot by passing
12308 fstab=0 on the kernel command line.
12310 * A new kernel command line option modules-load= is now understood
12311 to load a specific kernel module statically, early at boot.
12313 * Unit names specified on the systemctl command line are now
12314 automatically escaped as needed. Also, if file system or
12315 device paths are specified they are automatically turned
12316 into the appropriate mount or device unit names. Example:
12318 systemctl status /home
12319 systemctl status /dev/sda
12321 * The SysVConsole= configuration option has been removed from
12322 system.conf parsing.
12324 * The SysV search path is no longer exported on the D-Bus
12327 * The Names= option has been removed from unit file parsing.
12329 * There's a new man page bootup(7) detailing the boot process.
12331 * Every unit and every generator we ship with systemd now
12332 comes with full documentation. The self-explanatory boot is
12335 * A couple of services gained "systemd-" prefixes in their
12336 name if they wrap systemd code, rather than only external
12337 code. Among them fsck@.service which is now
12338 systemd-fsck@.service.
12340 * The HaveWatchdog property has been removed from the D-Bus
12343 * systemd.confirm_spawn= on the kernel command line should now
12346 * There's a new man page crypttab(5) which details all options
12347 we actually understand.
12349 * systemd-nspawn gained a new --capability= switch to pass
12350 additional capabilities to the container.
12352 * timedated will now read known NTP implementation unit names
12353 from /usr/lib/systemd/ntp-units.d/*.list,
12354 systemd-timedated-ntp.target has been removed.
12356 * journalctl gained a new switch "-b" that lists log data of
12357 the current boot only.
12359 * The notify socket is in the abstract namespace again, in
12360 order to support daemons which chroot() at start-up.
12362 * There is a new Storage= configuration option for journald
12363 which allows configuration of where log data should go. This
12364 also provides a way to disable journal logging entirely, so
12365 that data collected is only forwarded to the console, the
12366 kernel log buffer or another syslog implementation.
12368 * Many bugfixes and optimizations
12370 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
12371 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
12372 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
12373 Shawn Landden, Tom Gundersen
12377 * "systemctl help <unit>" now shows the man page if one is
12380 * Several new man pages have been added.
12382 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
12383 MaxLevelConsole= can now be specified in
12384 journald.conf. These options allow reducing the amount of
12385 data stored on disk or forwarded by the log level.
12387 * TimerSlackNSec= can now be specified in system.conf for
12388 PID1. This allows system-wide power savings.
12390 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
12391 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
12396 * logind is now capable of (optionally) handling power and
12397 sleep keys as well as the lid switch.
12399 * journalctl now understands the syntax "journalctl
12400 /usr/bin/avahi-daemon" to get all log output of a specific
12403 * CapabilityBoundingSet= in system.conf now also influences
12404 the capability bound set of usermode helpers of the kernel.
12406 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
12407 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
12408 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
12409 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
12413 * Note that we skipped 139 releases here in order to set the
12414 new version to something that is greater than both udev's
12415 and systemd's most recent version number.
12417 * udev: all udev sources are merged into the systemd source tree now.
12418 All future udev development will happen in the systemd tree. It
12419 is still fully supported to use the udev daemon and tools without
12420 systemd running, like in initramfs or other init systems. Building
12421 udev though, will require the *build* of the systemd tree, but
12422 udev can be properly *run* without systemd.
12424 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
12425 should be used to create dead device nodes as workarounds for broken
12428 * udev: RUN+="socket:…" and udev_monitor_new_from_socket() is
12429 no longer supported. udev_monitor_new_from_netlink() needs to be
12430 used to subscribe to events.
12432 * udev: when udevd is started by systemd, processes which are left
12433 behind by forking them off of udev rules, are unconditionally cleaned
12434 up and killed now after the event handling has finished. Services or
12435 daemons must be started as systemd services. Services can be
12436 pulled-in by udev to get started, but they can no longer be directly
12437 forked by udev rules.
12439 * udev: the daemon binary is called systemd-udevd now and installed
12440 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
12441 to adapt to that, create symlink, or rename the binary after building
12444 * libudev no longer provides these symbols:
12445 udev_monitor_from_socket()
12446 udev_queue_get_failed_list_entry()
12447 udev_get_{dev,sys,run}_path()
12448 The versions number was bumped and symbol versioning introduced.
12450 * systemd-loginctl and systemd-journalctl have been renamed
12451 to loginctl and journalctl to match systemctl.
12453 * The config files: /etc/systemd/systemd-logind.conf and
12454 /etc/systemd/systemd-journald.conf have been renamed to
12455 logind.conf and journald.conf. Package updates should rename
12456 the files to the new names on upgrade.
12458 * For almost all files the license is now LGPL2.1+, changed
12459 from the previous GPL2.0+. Exceptions are some minor stuff
12460 of udev (which will be changed to LGPL2.1 eventually, too),
12461 and the MIT licensed sd-daemon.[ch] library that is suitable
12462 to be used as drop-in files.
12464 * systemd and logind now handle system sleep states, in
12465 particular suspending and hibernating.
12467 * logind now implements a sleep/shutdown/idle inhibiting logic
12468 suitable for a variety of uses. Soonishly Lennart will blog
12469 about this in more detail.
12471 * var-run.mount and var-lock.mount are no longer provided
12472 (which previously bind mounted these directories to their new
12473 places). Distributions which have not converted these
12474 directories to symlinks should consider stealing these files
12475 from git history and add them downstream.
12477 * We introduced the Documentation= field for units and added
12478 this to all our shipped units. This is useful to make it
12479 easier to explore the boot and the purpose of the various
12482 * All smaller setup units (such as
12483 systemd-vconsole-setup.service) now detect properly if they
12484 are run in a container and are skipped when
12485 appropriate. This guarantees an entirely noise-free boot in
12486 Linux container environments such as systemd-nspawn.
12488 * A framework for implementing offline system updates is now
12489 integrated, for details see:
12490 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
12492 * A new service type Type=idle is available now which helps us
12493 avoiding ugly interleaving of getty output and boot status
12496 * There's now a system-wide CapabilityBoundingSet= option to
12497 globally reduce the set of capabilities for the
12498 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
12499 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
12500 even CAP_NET_ADMIN system-wide for secure systems.
12502 * There are now system-wide DefaultLimitXXX= options to
12503 globally change the defaults of the various resource limits
12504 for all units started by PID 1.
12506 * Harald Hoyer's systemd test suite has been integrated into
12507 systemd which allows easy testing of systemd builds in qemu
12508 and nspawn. (This is really awesome! Ask us for details!)
12510 * The fstab parser is now implemented as generator, not inside
12513 * systemctl will now warn you if .mount units generated from
12514 /etc/fstab are out of date due to changes in fstab that
12515 have not been read by systemd yet.
12517 * systemd is now suitable for usage in initrds. Dracut has
12518 already been updated to make use of this. With this in place
12519 initrds get a slight bit faster but primarily are much
12520 easier to introspect and debug since "systemctl status" in
12521 the host system can be used to introspect initrd services,
12522 and the journal from the initrd is kept around too.
12524 * systemd-delta has been added, a tool to explore differences
12525 between user/admin configuration and vendor defaults.
12527 * PrivateTmp= now affects both /tmp and /var/tmp.
12529 * Boot time status messages are now much prettier and feature
12530 proper english language. Booting up systemd has never been
12533 * Read-ahead pack files now include the inode number of all
12534 files to pre-cache. When the inode changes the pre-caching
12535 is not attempted. This should be nicer to deal with updated
12536 packages which might result in changes of read-ahead
12539 * We now temporaritly lower the kernel's read_ahead_kb variable
12540 when collecting read-ahead data to ensure the kernel's
12541 built-in read-ahead does not add noise to our measurements
12542 of necessary blocks to pre-cache.
12544 * There's now RequiresMountsFor= to add automatic dependencies
12545 for all mounts necessary for a specific file system path.
12547 * MountAuto= and SwapAuto= have been removed from
12548 system.conf. Mounting file systems at boot has to take place
12551 * nspawn now learned a new switch --uuid= to set the machine
12552 ID on the command line.
12554 * nspawn now learned the -b switch to automatically search
12555 for an init system.
12557 * vt102 is now the default TERM for serial TTYs, upgraded from
12560 * systemd-logind now works on VT-less systems.
12562 * The build tree has been reorganized. The individual
12563 components now have directories of their own.
12565 * A new condition type ConditionPathIsReadWrite= is now available.
12567 * nspawn learned the new -C switch to create cgroups for the
12568 container in other hierarchies.
12570 * We now have support for hardware watchdogs, configurable in
12573 * The scheduled shutdown logic now has a public API.
12575 * We now mount /tmp as tmpfs by default, but this can be
12576 masked and /etc/fstab can override it.
12578 * Since udisks does not make use of /media anymore we are not
12579 mounting a tmpfs on it anymore.
12581 * journalctl gained a new --local switch to only interleave
12582 locally generated journal files.
12584 * We can now load the IMA policy at boot automatically.
12586 * The GTK tools have been split off into a systemd-ui.
12588 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
12589 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
12590 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
12591 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
12592 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
12593 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
12594 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
12595 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
12596 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
12601 * This is mostly a bugfix release
12603 * Support optional initialization of the machine ID from the
12604 KVM or container configured UUID.
12606 * Support immediate reboots with "systemctl reboot -ff"
12608 * Show /etc/os-release data in systemd-analyze output
12610 * Many bugfixes for the journal, including endianness fixes and
12611 ensuring that disk space enforcement works
12613 * sd-login.h is C++ compatible again
12615 * Extend the /etc/os-release format on request of the Debian
12618 * We now refuse non-UTF8 strings used in various configuration
12619 and unit files. This is done to ensure we do not pass invalid
12620 data over D-Bus or expose it elsewhere.
12622 * Register Mimo USB Screens as suitable for automatic seat
12625 * Read SELinux client context from journal clients in a race
12628 * Reorder configuration file lookup order. /etc now always
12629 overrides /run in order to allow the administrator to always
12630 and unconditionally override vendor-supplied or
12631 automatically generated data.
12633 * The various user visible bits of the journal now have man
12634 pages. We still lack man pages for the journal API calls
12637 * We now ship all man pages in HTML format again in the
12640 Contributions from: Dave Reisner, Dirk Eibach, Frederic
12641 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
12642 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
12647 * This is mostly a bugfix release
12649 * systems lacking /etc/os-release are no longer supported.
12651 * Various functionality updates to libsystemd-login.so
12653 * Track class of PAM logins to distinguish greeters from
12654 normal user logins.
12656 Contributions from: Kay Sievers, Lennart Poettering, Michael
12661 * This is an important bugfix release for v41.
12663 * Building man pages is now optional which should be useful
12664 for those building systemd from git but unwilling to install
12667 * Watchdog support for supervising services is now usable. In
12668 a future release support for hardware watchdogs
12669 (i.e. /dev/watchdog) will be added building on this.
12671 * Service start rate limiting is now configurable and can be
12672 turned off per service. When a start rate limit is hit a
12673 reboot can automatically be triggered.
12675 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
12677 Contributions from: Benjamin Franzke, Bill Nottingham,
12678 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
12679 Schmidt, Michał Górny, Piotr Drąg
12683 * The systemd binary is installed /usr/lib/systemd/systemd now;
12684 An existing /sbin/init symlink needs to be adapted with the
12687 * The code that loads kernel modules has been ported to invoke
12688 libkmod directly, instead of modprobe. This means we do not
12689 support systems with module-init-tools anymore.
12691 * Watchdog support is now already useful, but still not
12694 * A new kernel command line option systemd.setenv= is
12695 understood to set system wide environment variables
12696 dynamically at boot.
12698 * We now limit the set of capabilities of systemd-journald.
12700 * We now set SIGPIPE to ignore by default, since it only is
12701 useful in shell pipelines, and has little use in general
12702 code. This can be disabled with IgnoreSIPIPE=no in unit
12705 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
12706 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
12711 * This is mostly a bugfix release
12713 * We now expose the reason why a service failed in the
12714 "Result" D-Bus property.
12716 * Rudimentary service watchdog support (will be completed over
12717 the next few releases.)
12719 * When systemd forks off in order execute some service we will
12720 now immediately changes its argv[0] to reflect which process
12721 it will execute. This is useful to minimize the time window
12722 with a generic argv[0], which makes bootcharts more useful
12724 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
12725 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
12726 Mike Kazantsev, Ray Strode
12730 * This is mostly a test release, but incorporates many
12733 * New systemd-cgtop tool to show control groups by their
12736 * Linking against libacl for ACLs is optional again. If
12737 disabled, support tracking device access for active logins
12738 goes becomes unavailable, and so does access to the user
12739 journals by the respective users.
12741 * If a group "adm" exists, journal files are automatically
12742 owned by them, thus allow members of this group full access
12743 to the system journal as well as all user journals.
12745 * The journal now stores the SELinux context of the logging
12746 client for all entries.
12748 * Add C++ inclusion guards to all public headers
12750 * New output mode "cat" in the journal to print only text
12751 messages, without any meta data like date or time.
12753 * Include tiny X server wrapper as a temporary stop-gap to
12754 teach XOrg udev display enumeration. This is used by display
12755 managers such as gdm, and will go away as soon as XOrg
12756 learned native udev hotplugging for display devices.
12758 * Add new systemd-cat tool for executing arbitrary programs
12759 with STDERR/STDOUT connected to the journal. Can also act as
12760 BSD logger replacement, and does so by default.
12762 * Optionally store all locally generated coredumps in the
12763 journal along with meta data.
12765 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
12766 writing short strings to files (for usage for /sys), and for
12767 creating symlinks, character and block device nodes.
12769 * New unit file option ControlGroupPersistent= to make cgroups
12770 persistent, following the mechanisms outlined in
12771 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
12773 * Support multiple local RTCs in a sane way
12775 * No longer monopolize IO when replaying readahead data on
12776 rotating disks, since we might starve non-file-system IO to
12777 death, since fanotify() will not see accesses done by blkid,
12780 * Do not show kernel threads in systemd-cgls anymore, unless
12781 requested with new -k switch.
12783 Contributions from: Dan Horák, Kay Sievers, Lennart
12784 Poettering, Michal Schmidt
12788 * This is mostly a test release, but incorporates many
12791 * The git repository moved to:
12792 git://anongit.freedesktop.org/systemd/systemd
12793 ssh://git.freedesktop.org/git/systemd/systemd
12795 * First release with the journal
12796 http://0pointer.de/blog/projects/the-journal.html
12798 * The journal replaces both systemd-kmsg-syslogd and
12799 systemd-stdout-bridge.
12801 * New sd_pid_get_unit() API call in libsystemd-logind
12803 * Many systemadm clean-ups
12805 * Introduce remote-fs-pre.target which is ordered before all
12806 remote mounts and may be used to start services before all
12809 * Added Mageia support
12811 * Add bash completion for systemd-loginctl
12813 * Actively monitor PID file creation for daemons which exit in
12814 the parent process before having finished writing the PID
12815 file in the daemon process. Daemons which do this need to be
12816 fixed (i.e. PID file creation must have finished before the
12817 parent exits), but we now react a bit more gracefully to them.
12819 * Add colourful boot output, mimicking the well-known output
12820 of existing distributions.
12822 * New option PassCredentials= for socket units, for
12823 compatibility with a recent kernel ABI breakage.
12825 * /etc/rc.local is now hooked in via a generator binary, and
12826 thus will no longer act as synchronization point during
12829 * systemctl list-unit-files now supports --root=.
12831 * systemd-tmpfiles now understands two new commands: z, Z for
12832 relabelling files according to the SELinux database. This is
12833 useful to apply SELinux labels to specific files in /sys,
12834 among other things.
12836 * Output of SysV services is now forwarded to both the console
12837 and the journal by default, not only just the console.
12839 * New man pages for all APIs from libsystemd-login.
12841 * The build tree got reorganized and the build system is a
12842 lot more modular allowing embedded setups to specifically
12843 select the components of systemd they are interested in.
12845 * Support for Linux systems lacking the kernel VT subsystem is
12848 * configure's --with-rootdir= got renamed to
12849 --with-rootprefix= to follow the naming used by udev and
12852 * Unless specified otherwise we will now install to /usr instead
12853 of /usr/local by default.
12855 * Processes with '@' in argv[0][0] are now excluded from the
12856 final shut-down killing spree, following the logic explained
12858 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
12860 * All processes remaining in a service cgroup when we enter
12861 the START or START_PRE states are now killed with
12862 SIGKILL. That means it is no longer possible to spawn
12863 background processes from ExecStart= lines (which was never
12864 supported anyway, and bad style).
12866 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
12867 reloading of units together.
12869 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
12870 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
12871 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
12872 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
12873 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek