1 systemd System and Service Manager
3 CHANGES WITH 246 in spe:
5 * The various programs included in systemd can now optionally output
6 their log messages on stderr prefixed with a timestamp, controlled by
7 the $SYSTEMD_LOG_TIME environment variable.
9 * A new boolean kernel command line option systemd.swap= has been
10 added, which may be used to turn off automatic activation of swap
11 devices, as listed in /etc/fstab.
13 * The CPUAffinity= setting in service unit files now supports a new
14 special value "numa". If used, the NUMA mask is copied into the CPU
17 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
19 * networkctl gained the new "forcerenew" command for forcing all DHCP
20 server clients to renew their lease. The interface "status" output
21 will now show numerous additional fields of information about an
22 interface. There are new "up" and "down" commands to bring specific
23 interfaces up or down.
25 * systemd-networkd's [IPv6Prefix] section in .network files gained a
26 new boolean setting Assign=. If enabled an address from the prefix is
27 automatically assigned to the interface.
29 * systemd-networkd's [Network] section gained a new setting
30 IPv6PDSubnetId= that allows explicit configuration of the preferred
31 subnet that networkd's Prefix Delegation logic assigns to an
34 * systemd-networkd gained support for configuring the HTB queuing
35 discipline in the [HierarchyTokenBucket] and
36 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
37 be configured in the [PFIFO] section, "GRED" in
38 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
39 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
40 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
41 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
42 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
43 "QFQ" in [QuickFairQueueingClass].
45 * systemd-networkd gained support for a new Termination= setting in the
46 [CAN] section for configuring the termination resistor. It also
47 gained a new ListenOnly= setting for controlling whether to only
48 listen on CAN interfaces, without interfering with traffic otherwise
49 (which is useful for debugging/monitoring CAN network
50 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
51 been added to configure various CAN-FD aspects.
53 * .link files managed by systemd-udevd gained options RxFlowControl=,
54 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
55 order to configure various flow control parameters. They also gained
56 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
57 frame ring buffer sizes.
59 * systemd-networkd's [DHCPv6] section gained a new WithoutRA= boolean
60 setting. If enabled, DHCPv6 will be attempted right-away without
61 requiring an Router Advertisement packet suggesting it
62 first. Conversely, the [IPv6AcceptRA] gained a boolean option
63 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
64 the RA packets suggest it.
66 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
67 which may be used to turn off use of the gateway information provided
68 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
69 used to configure how to process leases that lack a lifetime option.
71 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
72 setting SendVendorOption= allowing configuration of additional vendor
73 options to send in the DHCP requests/responses. The [DHCPv6] section
74 gained a new SendOption= setting for sending arbitrary DHCP
75 options. RequestOptions= has been added to request arbitrary options
76 from the server. UserClass= has been added to set the DHCP user class
79 * systemd-networkd's [DHCPServer] section gained a new set of options
80 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
81 information about these three protocols in the DHCP lease. It also
82 gained support for including "MUD" URLs ("Manufacturer Usage
83 Description"). Support for "MUD" URLs was also added to the LLDP
84 stack, configurable in the [LLDP] section in .network files.
86 * systemd-resolved's DNS= configuration option now optionally accepts
87 DNS server addresses suffixed by "#" followed by a host name. If
88 used, the DNS-over-TLS certificate is validated to match the
91 * systemd-resolved may be configured to forward single-label DNS names.
92 This is not standard-conformant, but may make sense in setups where
93 public DNS servers are not used.
95 * systemd-resolved's DNS-over-TLS support gained SNI validation.
97 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
98 systemd-coredump to save core files for suid processes. When saving
99 the core file, systemd-coredump will use the effective uid and gid of
100 the process that faulted.
102 * "systemctl list-units" and "systemctl list-machines" no longer hide
103 their first output column with --no-legend. To hide the first column,
106 * The service manager gained basic support for cgroup v2 freezer. Units
107 can now be suspended or resumed either using new systemctl verbs,
108 freeze and thaw respectively, or via D-Bus.
110 * systemd-udevd gained new configuration option timeout_signal= as well
111 as corresponding kernel command line option udev.timeout_signal=.
112 The option can be used to configure the UNIX signal that the main
113 daemon sends to the worker processes on timeout.
115 * A new sd-path.h API has been added to libsystemd. It provides a
116 simple API for retrieving various search paths and primary
117 directories for various resources.
119 * The sd-bus API gained a number of convenience functions that take
120 va_list arguments rather than "...". For example, there's now
121 sd_bus_call_methodv() to match sd_bus_call_method(). Previously,
122 these were missing since the calls are convenience calls only and
123 could be put together from the more low-level functions they build
126 * sd-bus vtable entries learnt a new flag SD_BUS_VTABLE_ABSOLUTE_OFFSET
127 which alters how the userdata pointer to pass to the callbacks is
128 determined. If the flag is set the offset field is converted as-is
129 into a pointer, without adding it to the object pointer the vtable is
132 * sd-bus now exposes four new functions:
133 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
134 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
135 validate strings to check if they qualify as various D-Bus concepts.
137 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
138 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
139 that simplify adding argument names to D-Bus methods and signals.
141 * systemctl gained a new "-P" switch that is a shortcut for "--value
144 * The expectations on user/group name syntax are now documented in
145 detail; documentation how classic home directories may be converted
146 into home directories managed by homed has been added; documentation
147 regarding integration of homed/userdb functionality in desktops has
150 https://systemd.io/USER_NAMES
151 https://systemd.io/CONVERTING_TO_HOMED
152 https://systemd.io/USERDB_AND_DESKTOPS
154 * systemd-run gained a new switch --slice-inherit. If specified the
155 unit it generates is placed in the same slice as the systemd-run
158 * service unit files now accept a new setting CoredumpFilter= which
159 allows configuration of the memory sections coredumps of the
160 service's processes shall include.
162 * coredumpctl gained a new --file= switch, matching the same one in
163 journalctl: a specific journal file may be specified to read the
166 * Various D-Bus APIs of systemd daemons now have man pages that
167 document the methods, signals and properties.
169 * journald.conf gained a new boolean setting Audit= that may be used to
170 control whether systemd-journald will enable audit during
173 * A new default .network file is now shipped that matches TUN/TAP
174 devices that begin with "vt-" in their name. Such interfaces will
175 have IP routing onto the host links set up automatically. This is
176 supposed to be used by VM managers to trivially acquire a network
177 interface which is fully set up for host communication, simply by
178 carefully picking an interface name to use.
180 * All D-Bus services shipped in systemd now implement the generic
181 LogControl1 D-Bus API which allows clients to change log level +
182 target of the service during runtime.
184 * systemd-nspawn's --resolv-conf= switch gained a number of new
185 supported values. Specifically, options starting with "replace-" are
186 like those prefixed "copy-" but replace any existing resolv.conf
187 file. And options ending in "-uplink" and "-stub" can now be used to
188 propagate other flavours of resolv.conf into the container (as
189 defined by systemd-resolved).
191 * systemd-binfmt gained a new switch --unregister for unregistering all
192 registered entries at once. This is now invoked automatically at
193 shutdown, so that binary formats registered with the "F" flag will
194 not block clean file system unmounting.
196 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
197 configuration files that support specifier expansion learnt six new
198 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
199 resolve to the various ID fields from /etc/os-release, %l resolves to
200 the "short" hostname of the system, i.e. the kernel configured
201 hostname, truncated at the first dot.
203 * systemd-notify's --pid= switch gained new values: "parent", "self",
204 "auto" for controlling which PID to send to the service manager: the
205 systemd-notify process' PID, or the one of the process invoking it.
207 * When sending a file descriptor (fd) to the service manager to keep
208 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
209 may be specified. If passed the service manager will refrain from
210 poll()ing on the file descriptor. Traditionally (and when the
211 parameter is not specified), the service manager will poll it for
212 POLLHUP or POLLERR events, and immediately close the fds in that
215 * A new call sd_notify_barrier() has been added to the sd-daemon.h
216 API. The call will block until all previously sent sd_notify()
217 messages have been processed by the service manager. This is useful
218 to remove races caused by a process already having disappeared at the
219 time a notification message is processed by the service manager,
220 making correct attribution impossible. The systemd-notify tool will
221 now make use of this call implicitly, but this can be turned off again
222 via the new --no-block switch.
224 * systemd-logind's Session bus object learnt a new method call
225 SetType() for temporarily updating the session type of an already
226 allocated session. This is useful for upgrading tty sessions to
227 graphical ones once a compositor is invoked.
229 * .mount units gained a new ReadWriteOnly= boolean option. If set
230 it will not be attempted to mount a file system read-only if mounting
231 in read-write mode doesn't succeed. An option x-systemd.rw-only is
232 available in /etc/fstab to control the same.
234 * coredumps collected by systemd-coredump may now be compressed using
237 * journalctl's "-o cat" output mode will now show one or more journal
238 fields specified with --output-fields= instead of unconditionally
239 MESSAGE=. This is useful to retrieve a very specific set of fields
240 without any decoration.
242 * systemd-socket-proxy gained a new switch --exit-idle-time= for
243 configuring an exit-on-idle time.
245 * systemd-homed's LUKS backend gained the ability to discard empty file
246 system blocks automatically when the user logs out. This is enabled
247 by default to ensure that home directories take minimal space when
248 logged out but get full size guarantees when logged in. This may be
249 controlled with the new --luks-offline-discard= switch to homectl.
251 * If systemd-homed detects that /home/ is encrypted as a whole it will
252 now default to the directory or subvolume backends instead of the
253 LUKS backend, in order to avoid double encryption. The default
254 storage and file system may now be configured explicitly, too, via
255 the new /etc/systemd/homed.conf configuration file.
257 * when systemd-journald's log stream is broken up into multiple lines
258 because the PID of the sender changed this is indicated in the
259 generated log records via the _LINE_BREAK=pid-change field.
261 * systemd-networkd's .netdev files now support a new setting
262 VLANProtocol= in the [Bridge] section that allows configuration of
263 the VLAN protocol to use.
265 * systemd-repart's --empty= setting gained a new value "create". If
266 specified a new empty regular disk image file is created under the
267 specified name. Its size may be specified with the new --size=
268 option. The latter is also supported without the "create" mode, in
269 order to grow existing disk image files to the specified size. These
270 two new options are useful when creating or manipulating disk images
271 instead of operating on actual block devices.
273 * systemd-repart drop-ins now support a new UUID= setting to control
274 the UUID to assign to a newly created partition.
276 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
279 * systemd-repart's Label= setting now support the usual, simple
282 * StandardError= and StandardOutput= in unit files no longer support
283 the "syslog" and "syslog-console" switches. They were long removed
284 from the documentation, but will now result in warnings when used,
285 and be converted to "journal" and "journal+console" automatically.
287 * systemd-networkd supports a new Group= setting in the [Link] section
288 of the .network files, to control the link group.
290 * Two new unit file settings
291 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
292 added. They may be used to check whether a specific file system path
293 resides on a block device that is encrypted on the block level
294 (i.e. using dm-crypt/LUKS).
296 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
297 has been added that may be used for simple environment checks. This
298 is particularly useful when passing in environment variables from a
299 container manager (or from PAM in case of the systemd --user
302 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
303 now automatically set to "Y" at boot, in order to enable pstore
304 generation for collection with systemd-pstore.
306 * New kernel command line options systemd.condition-needs-update= and
307 systemd.condition-first-boot= have been added, which override the
308 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
311 * A new kernel command line option systemd.clock-usec= has been added
312 that allows setting the system clock to the specified time in µs
313 since Jan 1st, 1970 early during boot. This is in particular useful
314 in order to make test cases more reliable.
316 * A new kernel command line option systemd.hostname= has been added
317 that allows controlling the hostname that is initialized early during
320 * The /etc/crypttab tmp option now optionally takes an argument
321 selecting the file system to use. Moreover, the default is now
322 changed from ext2 to ext4.
324 * There's a new /etc/crypttab option "keyfile-erase". If specified the
325 key file listed in the same line is removed after use, regardless if
326 volume activation was successful or not. This is useful if the key
327 file is only acquired transiently at runtime and shall be erased
328 before the system continues to boot.
330 * There's also a new /etc/crypttab option "try-empty-password". If
331 specified, before asking the user for a password it is attempted to
332 unlock the volume with an empty password. This is useful for
333 installing encrypted images whose password shall be set on first boot
334 instead of at installation time.
336 * systemd-cryptsetup will now attempt to load the keys to unlock
337 volumes with automatically from files in
338 /etc/cryptsetup-keys.d/<volume>.key and
339 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
341 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
342 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
345 * systemd-firstboot gained a new --root-password-hashed= parameter for
346 setting the root user's password as UNIX password hash. There's a new
347 --delete-root-password switch which instead of setting a password for
348 the root user, removes it so that log-in without a password is
349 permitted. There's now --force which if specified means any existing
350 configuration is overwritten by the specified settings. It also
351 gained a new --kernel-command-line= parameter which may be used to
352 set the /etc/kernel/cmdline file of an OS image.
354 * A new generator systemd-xdg-autostart-generator has been added. It
355 automatically generates systemd unit files from XDG autostart
356 .desktop files, and is useful for allowing systemd to manage services
357 defined that way safely and automatically.
359 * systemd will now log about all left-over processes remaining in a
360 unit when the unit is stopped. It will now warn about services using
361 KillMode=none, as this is generally an unsafe thing to make use of.
363 * .socket units gained a new boolean setting PassPacketInfo=. If
364 enabled, the kernel will attach additional per-packet metadata to all
365 packets read from the socket, as ancillary message. This controls the
366 IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
367 depending on socket type.
369 * A new boolean option AssignAcquiredDelegatedPrefixAddress= has been
370 added to the [DHCPv6] section of .network files. If enabled (which is
371 the default) an address from any acquired delegated prefix is
372 automatically chosen and assigned to the interface.
374 * "systemctl reboot" takes the option "--reboot-argument=".
375 The optional positional argument to "systemctl reboot" is now
376 being deprecated in favor of this option.
378 * Support for the .include syntax in unit files has been removed. The
379 concept has been obsolete for 6 years and we started warning about
380 its pending removal 2 years ago (also see NEWS file below). It's
383 * The BlackList= settings in .network files' [DHCPv4] and
384 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
385 are still understood to provide compatibility.
387 * systemd-journald gained support for zstd compression of large fields
388 in journal files. The hash tables in journal files have been hardened
389 against hash collisions. This is an incompatible change and means
390 that journal files created with new systemd versions are not readable
391 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
392 environment variable for systemd-journald.service is set to 0 this
393 new hardening functionality may be turned off, so that generated
394 journal files remain compatible with older journalctl
397 * Documentation for the on-disk Journal file format has been updated
398 and has now moved to:
400 https://systemd.io/JOURNAL_FILE_FORMAT
402 * systemd service units gained a new setting RootHash= which may be
403 used to specify the root hash for verity enabled disk images which
404 are specified in RootImage=. RootVerity= may be used to specify a
405 path to the Verity data matching a RootImage= file system. (The
406 latter is only useful for images that do not contain the Verity data
407 embedded into the same image that carries a GPT partition table
408 following the Discoverable Partition Specification). Similar,
409 systemd-nspawn gained a new switch --verity-data= that takes a path
410 to a file with the verity data of the disk image supplied in
411 --image=, if the image doesn't contain the verity data itself.
413 * systemd service units gained a new setting RootHashSignature= which
414 takes either a base64 encoded PKCS#7 signature of the root hash
415 specified with RootHash=, or a path to a file to read the signature
416 from. This allows validation of the root hash against public keys
417 available in the kernel keyring, and is only supported on recent
418 kernels (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been
419 added to systemd-nspawn and systemd-dissect
420 (--root-hash-sig=). Support for this mechanism has also been added to
423 * A new 'hwdb' file has been added that collects information about PCI
424 and USB devices that correctly support auto-suspend, on top of the
425 databases for this we import from the ChromiumOS project. If you have
426 a device that supports auto-suspend correctly and where it should be
427 enabled by default, please submit a patch that adds it to the
428 database (see hwdb.d/60-autosuspend.hwdb).
430 * The service manager (PID1) gained a new D-Bus method call
431 SetShowStatus() which may be used to control whether it shall show
432 boot-time status output on the console. This method has a similar
433 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
435 * PID 1 may now automatically load pre-compiled AppArmor policies from
436 /etc/apparmor/earlypolicy during early boot.
438 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
439 /etc/crypttab, during boot.
441 * systemd service unit files gained two new options
442 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
443 tune behaviour if a start or stop timeout is hit, i.e. whether to
444 terminate the service with SIGTERM, SIGABRT or SIGKILL.
446 * A kernel command line option "udev.blockdev_read_only" has been
447 added. If specified all hardware block devices that show up are
448 immediately marked as read-only by udev. This option is useful for
449 making sure that a specific boot under no circumstances modifies data
450 on disk. Use "blockdev --setrw" to undo the effect of this, per
453 * Most options in systemd that accept hexadecimal values prefixed with
454 0x in additional to the usual decimal notation now also support octal
455 notation when he 0o prefix is used and binary notation if the 0b
458 * "booctl" gained a new verb "reboot-to-firmware" that may be used
459 to query and change the firmware's reboot into firmware setup flag.
461 * journalctl will now include a clickable link in the default output for
462 each log message for which an URL with further documentation is
463 known. This is only supported on terminal emulators that support
464 clickable hyperlinks, and is turned off if a pager is used (since
465 "less" still doesn't support hyperlinks,
466 unfortunately). Documentation URLs may be included in log messages
467 either by including a DOCUMENTATION= journal field in it, or by
468 associating a journal message catalog entry with the log message's
469 MESSAGE_ID, which then carries a "Documentation:" tag.
471 * systemd-firstboot gained a new switch --kernel-command-line= that may
472 be used to initialize the /etc/kernel/cmdline file of the image. It
473 also gained a new switch --root-password-hashed= which is like
474 --root-password= but accepts a pre-hashed UNIX password as
475 argument. The new option --delete-root-password may be used to unset
476 any password for the root user (dangerous!). A new --force option may
477 be used to override any already set settings with the parameters
478 specified on the command line (by default, the tool will not override
479 what has already been set before, i.e. is purely incremental).
483 * A new tool "systemd-repart" has been added, that operates as an
484 idempotent declarative repartitioner for GPT partition tables.
485 Specifically, a set of partitions that must or may exist can be
486 configured via drop-in files, and during every boot the partition
487 table on disk is compared with these files, creating missing
488 partitions or growing existing ones based on configurable relative
489 and absolute size constraints. The tool is strictly incremental,
490 i.e. does not delete, shrink or move partitions, but only adds and
491 grows them. The primary use-case is OS images that ship in minimized
492 form, that on first boot are grown to the size of the underlying
493 block device or augmented with additional partitions. For example,
494 the root partition could be extended to cover the whole disk, or a
495 swap or /home partitions could be added on first boot. It can also be
496 used for systems that use an A/B update scheme but ship images with
497 just the A partition, with B added on first boot. The tool is
498 primarily intended to be run in the initrd, shortly before
499 transitioning into the host OS, but can also be run after the
500 transition took place. It automatically discovers the disk backing
501 the root file system, and should hence not require any additional
502 configuration besides the partition definition drop-ins. If no
503 configuration drop-ins are present, no action is taken.
505 * A new component "userdb" has been added, along with a small daemon
506 "systemd-userdb.service" and a client tool "userdbctl". The framework
507 allows defining rich user and group records in a JSON format,
508 extending on the classic "struct passwd" and "struct group"
509 structures. Various components in systemd have been updated to
510 process records in this format, including systemd-logind and
511 pam-systemd. The user records are intended to be extensible, and
512 allow setting various resource management, security and runtime
513 parameters that shall be applied to processes and sessions of the
514 user as they log in. This facility is intended to allow associating
515 such metadata directly with user/group records so that they can be
516 produced, extended and consumed in unified form. We hope that
517 eventually frameworks such as sssd will generate records this way, so
518 that for the first time resource management and various other
519 per-user settings can be configured in LDAP directories and then
520 provided to systemd (specifically to systemd-logind and pam-system)
521 to apply on login. For further details see:
523 https://systemd.io/USER_RECORD
524 https://systemd.io/GROUP_RECORD
525 https://systemd.io/USER_GROUP_API
527 * A small new service systemd-homed.service has been added, that may be
528 used to securely manage home directories with built-in encryption.
529 The complete user record data is unified with the home directory,
530 thus making home directories naturally migratable. Its primary
531 back-end is based on LUKS volumes, but fscrypt, plain directories,
532 and other storage schemes are also supported. This solves a couple of
533 problems we saw with traditional ways to manage home directories, in
534 particular when it comes to encryption. For further discussion of
535 this, see the video of Lennart's talk at AllSystemsGo! 2019:
537 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
539 For further details about the format and expectations on home
540 directories this new daemon makes, see:
542 https://systemd.io/HOME_DIRECTORY
544 * systemd-journald is now multi-instantiable. In addition to the main
545 instance systemd-journald.service there's now a template unit
546 systemd-journald@.service, with each instance defining a new named
547 log 'namespace' (whose name is specified via the instance part of the
548 unit name). A new unit file setting LogNamespace= has been added,
549 taking such a namespace name, that assigns services to the specified
550 log namespaces. As each log namespace is serviced by its own
551 independent journal daemon, this functionality may be used to improve
552 performance and increase isolation of applications, at the price of
553 losing global message ordering. Each instance of journald has a
554 separate set of configuration files, with possibly different disk
555 usage limitations and other settings.
557 journalctl now takes a new option --namespace= to show logs from a
558 specific log namespace. The sd-journal.h API gained
559 sd_journal_open_namespace() for opening the log stream of a specific
560 log namespace. systemd-journald also gained the ability to exit on
561 idle, which is useful in the context of log namespaces, as this means
562 log daemons for log namespaces can be activated automatically on
563 demand and will stop automatically when no longer used, minimizing
566 * When systemd-tmpfiles copies a file tree using the 'C' line type it
567 will now label every copied file according to the SELinux database.
569 * When systemd/PID 1 detects it is used in the initrd it will now boot
570 into initrd.target rather than default.target by default. This should
571 make it simpler to build initrds with systemd as for many cases the
572 only difference between a host OS image and an initrd image now is
573 the presence of the /etc/initrd-release file.
575 * A new kernel command line option systemd.cpu_affinity= is now
576 understood. It's equivalent to the CPUAffinity= option in
577 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
578 itself and the default for all other processes.
580 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
581 equivalent), the SELinux database is now reloaded, ensuring that
582 sockets and other file system objects are generated taking the new
583 database into account.
585 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
586 "quiet" has been changed to imply that instead of
587 "systemd.show-status=auto". In this mode, only messages about errors
588 and significant delays in boot are shown on the console.
590 * The sd-event.h API gained native support for the new Linux "pidfd"
591 concept. This permits watching processes using file descriptors
592 instead of PID numbers, which fixes a number of races and makes
593 process supervision more robust and efficient. All of systemd's
594 components will now use pidfds if the kernel supports it for process
595 watching, with the exception of PID 1 itself, unfortunately. We hope
596 to move PID 1 to exclusively using pidfds too eventually, but this
597 requires some more kernel work first. (Background: PID 1 watches
598 processes using waitid() with the P_ALL flag, and that does not play
599 together nicely with pidfds yet.)
601 * Closely related to this, the sd-event.h API gained two new calls
602 sd_event_source_send_child_signal() (for sending a signal to a
603 watched process) and sd_event_source_get_child_process_own() (for
604 marking a process so that it is killed automatically whenever the
605 event source watching it is freed).
607 * systemd-networkd gained support for configuring Token Bucket Filter
608 (TBF) parameters in its qdisc configuration support. Similarly,
609 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
610 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
612 * systemd-networkd gained support for Intermediate Functional Block
613 (IFB) network devices.
615 * systemd-networkd gained support for configuring multi-path IP routes,
616 using the new MultiPathRoute= setting in the [Route] section.
618 * systemd-networkd's DHCPv4 client has been updated to support a new
619 SendDecline= option. If enabled, duplicate address detection is done
620 after a DHCP offer is received from the server. If a conflict is
621 detected, the address is declined. The DHCPv4 client also gained
622 support for a new RouteMTUBytes= setting that allows to configure the
623 MTU size to be used for routes generated from DHCPv4 leases.
625 * The PrefixRoute= setting in systemd-networkd's [Address] section of
626 .network files has been deprecated, and replaced by AddPrefixRoute=,
627 with its sense inverted.
629 * The Gateway= setting of [Route] sections of .network files gained
630 support for a special new value "_dhcp". If set, the configured
631 static route uses the gateway host configured via DHCP.
633 * New User= and SuppressPrefixLength= settings have been implemented
634 for the [RoutingPolicyRule] section of .network files to configure
635 source routing based on UID ranges and prefix length, respectively.
637 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
638 D-Bus message object as "sensitive". Those objects are erased from
639 memory when they are freed. This concept is intended to be used for
640 messages that contain security sensitive data. A new flag
641 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
642 in sd-bus vtables, causing any incoming and outgoing messages of
643 those methods to be implicitly marked as "sensitive".
645 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
646 contents of a message (or parts thereof) to standard output for
649 * systemd-sysusers gained support for creating users with the primary
650 group named differently than the user.
652 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
653 gained support for growing XFS partitions. Previously it supported
654 only ext4 and btrfs partitions.
656 * The support for /etc/crypttab gained a new x-initrd.attach option. If
657 set, the specified encrypted volume is unlocked already in the
658 initrd. This concept corresponds to the x-initrd.mount option in
661 * systemd-cryptsetup gained native support for unlocking encrypted
662 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
663 encryption of volumes to YubiKeys. This is exposed in the new
664 pkcs11-uri= option in /etc/crypttab.
666 * The /etc/fstab support in systemd now supports two new mount options
667 x-systemd.{required,wanted}-by=, for explicitly configuring the units
668 that the specified mount shall be pulled in by, in place of
669 the usual local-fs.target/remote-fs.target.
671 * The https://systemd.io/ web site has been relaunched, directly
672 populated with most of the documentation included in the systemd
673 repository. systemd also acquired a new logo, thanks to Tobias
676 * systemd-udevd gained support for managing "alternative" network
677 interface names, as supported by new Linux kernels. For the first
678 time this permits assigning multiple (and longer!) names to a network
679 interface. systemd-udevd will now by default assign the names
680 generated via all supported naming schemes to each interface. This
681 may be further tweaked with .link files and the AlternativeName= and
682 AlternativeNamesPolicy= settings. Other components of systemd have
683 been updated to support the new alternative names wherever
684 appropriate. For example, systemd-nspawn will now generate
685 alternative interface names for the host-facing side of container
686 veth links based on the full container name without truncation.
688 * systemd-nspawn interface naming logic has been updated in another way
689 too: if the main interface name (i.e. as opposed to new-style
690 "alternative" names) based on the container name is truncated, a
691 simple hashing scheme is used to give different interface names to
692 multiple containers whose names all begin with the same prefix. Since
693 this changes the primary interface names pointing to containers if
694 truncation happens, the old scheme may still be requested by
695 selecting an older naming scheme, via the net.naming-scheme= kernel
698 * PrivateUsers= in service files now works in services run by the
699 systemd --user per-user instance of the service manager.
701 * A new per-service sandboxing option ProtectClock= has been added that
702 locks down write access to the system clock. It takes away device
703 node access to /dev/rtc as well as the system calls that set the
704 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
705 Note that this option does not affect access to auxiliary services
706 that allow changing the clock, for example access to
709 * The systemd-id128 tool gained a new "show" verb for listing or
710 resolving a number of well-known UUIDs/128bit IDs, currently mostly
711 GPT partition table types.
713 * The Discoverable Partitions Specification has been updated to support
714 /var and /var/tmp partition discovery. Support for this has been
715 added to systemd-gpt-auto-generator. For details see:
717 https://systemd.io/DISCOVERABLE_PARTITIONS
719 * "systemctl list-unit-files" has been updated to show a new column
720 with the suggested enablement state based on the vendor preset files
721 for the respective units.
723 * "systemctl" gained a new option "--with-dependencies". If specified
724 commands such as "systemctl status" or "systemctl cat" will now show
725 all specified units along with all units they depend on.
727 * networkctl gained support for showing per-interface logs in its
730 * systemd-networkd-wait-online gained support for specifying the maximum
731 operational state to wait for, and to wait for interfaces to
734 * The [Match] section of .link and .network files now supports a new
735 option PermanentMACAddress= which may be used to check against the
736 permanent MAC address of a network device even if a randomized MAC
739 * The [TrafficControlQueueingDiscipline] section in .network files has
740 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
741 dropped from the individual setting names.
743 * Any .link and .network files that have an empty [Match] section (this
744 also includes empty and commented-out files) will now be
745 rejected. systemd-udev and systemd-networkd started warning about
746 such files in version 243.
748 * systemd-logind will now validate access to the operation of changing
749 the virtual terminal via a polkit action. By default, only users
750 with at least one session on a local VT are granted permission.
752 * When systemd sets up PAM sessions that invoked service processes
753 shall run in, the pam_setcred() API is now invoked, thus permitting
754 PAM modules to set additional credentials for the processes.
756 * portablectl attach/detach verbs now accept --now and --enable options
757 to combine attachment with enablement and invocation, or detachment
758 with stopping and disablement.
760 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
761 fixed, which in turn exposed bugs in unit configuration of services
762 which have Type=oneshot and should only run once, but do not have
763 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
764 service may be started again after exiting successfully, for example
765 as a dependency in another transaction. Affected services included
766 some internal systemd services (most notably
767 systemd-vconsole-setup.service, which was updated to have
768 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
769 plymouth has been suitably updated or patched before upgrading to
770 this systemd release. See
771 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
772 additional discussion.
774 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
775 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
776 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
777 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
778 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
779 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
780 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
781 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
782 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
783 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
784 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
785 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
786 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
787 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
788 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
789 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
790 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
791 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
792 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
793 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
794 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
795 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
796 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
797 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
804 * Support for the cpuset cgroups v2 controller has been added.
805 Processes may be restricted to specific CPUs using the new
806 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
807 AllowedMemoryNodes= setting.
809 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
810 now be configured using a new RestartKillSignal= setting. This
811 allows units which signals to request termination to implement
812 different behaviour when stopping in preparation for a restart.
814 * "systemctl clean" may now be used also for socket, mount, and swap
817 * systemd will also read configuration options from the EFI variable
818 SystemdOptions. This may be used to configure systemd behaviour when
819 modifying the kernel command line is inconvenient, but configuration
820 on disk is read too late, for example for the options related to
821 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
822 set the EFI variable.
824 * systemd will now disable printk ratelimits in early boot. This should
825 allow us to capture more logs from the early boot phase where normal
826 storage is not available and the kernel ring buffer is used for
827 logging. Configuration on the kernel command line has higher priority
828 and overrides the systemd setting.
830 systemd programs which log to /dev/kmsg directly use internal
831 ratelimits to prevent runaway logging. (Normally this is only used
832 during early boot, so in practice this change has very little
835 * Unit files now support top level dropin directories of the form
836 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
837 that affects all corresponding unit files.
839 * systemctl gained support for 'stop --job-mode=triggering' which will
840 stop the specified unit and any units which could trigger it.
842 * Unit status display now includes units triggering and triggered by
843 the unit being shown.
845 * The RuntimeMaxSec= setting is now supported by scopes, not just
846 .service units. This is particularly useful for PAM sessions which
847 create a scope unit for the user login. systemd.runtime_max_sec=
848 setting may used with the pam_systemd module to limit the duration
849 of the PAM session, for example for time-limited logins.
851 * A new @pkey system call group is now defined to make it easier to
852 allow-list memory protection syscalls for containers and services
853 which need to use them.
855 * systemd-udevd: removed the 30s timeout for killing stale workers on
856 exit. systemd-udevd now waits for workers to finish. The hard-coded
857 exit timeout of 30s was too short for some large installations, where
858 driver initialization could be prematurely interrupted during initrd
859 processing if the root file system had been mounted and init was
860 preparing to switch root. If udevd is run without systemd and workers
861 are hanging while udevd receives an exit signal, udevd will now exit
862 when udev.event_timeout is reached for the last hanging worker. With
863 systemd, the exit timeout can additionally be configured using
864 TimeoutStopSec= in systemd-udevd.service.
866 * udev now provides a program (fido_id) that identifies FIDO CTAP1
867 ("U2F")/CTAP2 security tokens based on the usage declared in their
868 report and descriptor and outputs suitable environment variables.
869 This replaces the externally maintained allow lists of all known
870 security tokens that were used previously.
872 * Automatically generated autosuspend udev rules for allow-listed
873 devices have been imported from the Chromium OS project. This should
874 improve power saving with many more devices.
876 * udev gained a new "CONST{key}=value" setting that allows matching
877 against system-wide constants without forking a helper binary.
878 Currently "arch" and "virt" keys are supported.
880 * udev now opens CDROMs in non-exclusive mode when querying their
881 capabilities. This should fix issues where other programs trying to
882 use the CDROM cannot gain access to it, but carries a risk of
883 interfering with programs writing to the disk, if they did not open
884 the device in exclusive mode as they should.
886 * systemd-networkd does not create a default route for IPv4 link local
887 addressing anymore. The creation of the route was unexpected and was
888 breaking routing in various cases, but people who rely on it being
889 created implicitly will need to adjust. Such a route may be requested
890 with DefaultRouteOnDevice=yes.
892 Similarly, systemd-networkd will not assign a link-local IPv6 address
893 when IPv6 link-local routing is not enabled.
895 * Receive and transmit buffers may now be configured on links with
896 the new RxBufferSize= and TxBufferSize= settings.
898 * systemd-networkd may now advertise additional IPv6 routes. A new
899 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
902 * systemd-networkd may now configure "next hop" routes using the
903 [NextHop] section and Gateway= and Id= settings.
905 * systemd-networkd will now retain DHCP config on restarts by default
906 (but this may be overridden using the KeepConfiguration= setting).
907 The default for SendRelease= has been changed to true.
909 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
910 received from the server.
912 The client will use the received SIP server list if UseSIP=yes is
915 The client may be configured to request specific options from the
916 server using a new RequestOptions= setting.
918 The client may be configured to send arbitrary options to the server
919 using a new SendOption= setting.
921 A new IPServiceType= setting has been added to configure the "IP
922 service type" value used by the client.
924 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
925 request prefix hints in the DHCPv6 solicitation.
927 * The DHCPv4 server may be configured to send arbitrary options using
928 a new SendOption= setting.
930 * The DHCPv4 server may now be configured to emit SIP server list using
931 the new EmitSIP= and SIP= settings.
933 * systemd-networkd and networkctl may now renew DHCP leases on demand.
934 networkctl has a new 'networkctl renew' verb.
936 * systemd-networkd may now reconfigure links on demand. networkctl
937 gained two new verbs: "reload" will reload the configuration, and
938 "reconfigure DEVICE…" will reconfigure one or more devices.
940 * .network files may now match on SSID and BSSID of a wireless network,
941 i.e. the access point name and hardware address using the new SSID=
942 and BSSID= options. networkctl will display the current SSID and
943 BSSID for wireless links.
945 .network files may also match on the wireless network type using the
946 new WLANInterfaceType= option.
948 * systemd-networkd now includes default configuration that enables
949 link-local addressing when connected to an ad-hoc wireless network.
951 * systemd-networkd may configure the Traffic Control queueing
952 disciplines in the kernel using the new
953 [TrafficControlQueueingDiscipline] section and Parent=,
954 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
955 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
956 NetworkEmulatorDuplicateRate= settings.
958 * systemd-tmpfiles gained a new w+ setting to append to files.
960 * systemd-analyze dump will now report when the memory configuration in
961 the kernel does not match what systemd has configured (usually,
962 because some external program has modified the kernel configuration
965 * systemd-analyze gained a new --base-time= switch instructs the
966 'calendar' verb to resolve times relative to that timestamp instead
969 * journalctl --update-catalog now produces deterministic output (making
970 reproducible image builds easier).
972 * A new devicetree-overlay setting is now documented in the Boot Loader
975 * The default value of the WatchdogSec= setting used in systemd
976 services (the ones bundled with the project itself) may be set at
977 configuration time using the -Dservice-watchdog= setting. If set to
978 empty, the watchdogs will be disabled.
980 * systemd-resolved validates IP addresses in certificates now when GnuTLS
983 * libcryptsetup >= 2.0.1 is now required.
985 * A configuration option -Duser-path= may be used to override the $PATH
986 used by the user service manager. The default is again to use the same
987 path as the system manager.
989 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
990 outputting the 128bit IDs in UUID format (i.e. in the "canonical
993 * Service units gained a new sandboxing option ProtectKernelLogs= which
994 makes sure the program cannot get direct access to the kernel log
995 buffer anymore, i.e. the syslog() system call (not to be confused
996 with the API of the same name in libc, which is not affected), the
997 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
998 inaccessible to the service. It's recommended to enable this setting
999 for all services that should not be able to read from or write to the
1000 kernel log buffer, which are probably almost all.
1002 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
1003 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
1004 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
1005 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
1006 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
1007 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
1008 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
1009 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
1010 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
1011 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
1012 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
1013 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
1014 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
1015 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
1016 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
1017 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
1018 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
1019 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
1020 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
1021 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
1022 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
1023 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
1024 Zach Smith, Zbigniew Jędrzejewski-Szmek
1026 – Warsaw, 2019-11-29
1030 * This release enables unprivileged programs (i.e. requiring neither
1031 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
1032 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
1033 kernel for the whole UNIX group range, i.e. all processes. This
1034 change should be reasonably safe, as the kernel support for it was
1035 specifically implemented to allow safe access to ICMP Echo for
1036 processes lacking any privileges. If this is not desirable, it can be
1037 disabled again by setting the parameter to "1 0".
1039 * Previously, filters defined with SystemCallFilter= would have the
1040 effect that any calling of an offending system call would terminate
1041 the calling thread. This behaviour never made much sense, since
1042 killing individual threads of unsuspecting processes is likely to
1043 create more problems than it solves. With this release the default
1044 action changed from killing the thread to killing the whole
1045 process. For this to work correctly both a kernel version (>= 4.14)
1046 and a libseccomp version (>= 2.4.0) supporting this new seccomp
1047 action is required. If an older kernel or libseccomp is used the old
1048 behaviour continues to be used. This change does not affect any
1049 services that have no system call filters defined, or that use
1050 SystemCallErrorNumber= (and thus see EPERM or another error instead
1051 of being killed when calling an offending system call). Note that
1052 systemd documentation always claimed that the whole process is
1053 killed. With this change behaviour is thus adjusted to match the
1056 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
1057 4194304 by default, i.e. the full 22bit range the kernel allows, up
1058 from the old 16bit range. This should improve security and
1059 robustness, as PID collisions are made less likely (though certainly
1060 still possible). There are rumours this might create compatibility
1061 problems, though at this moment no practical ones are known to
1062 us. Downstream distributions are hence advised to undo this change in
1063 their builds if they are concerned about maximum compatibility, but
1064 for everybody else we recommend leaving the value bumped. Besides
1065 improving security and robustness this should also simplify things as
1066 the maximum number of allowed concurrent tasks was previously bounded
1067 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
1068 only a single knob is left ("kernel.threads-max"). There have been
1069 concerns that usability is affected by this change because larger PID
1070 numbers are harder to type, but we believe the change from 5 digits
1071 to 7 digits doesn't hamper usability.
1073 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
1074 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
1075 hierarchically set default memory protection values for a particular
1076 subtree of the unit hierarchy.
1078 * Memory protection directives can now take a value of zero, allowing
1079 explicit opting out of a default value propagated by an ancestor.
1081 * systemd now defaults to the "unified" cgroup hierarchy setup during
1082 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
1083 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
1084 change reflects the fact that cgroupsv2 support has matured
1085 substantially in both systemd and in the kernel, and is clearly the
1086 way forward. Downstream production distributions might want to
1087 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
1088 their builds as unfortunately the popular container managers have not
1089 caught up with the kernel API changes.
1091 * Man pages are not built by default anymore (html pages were already
1092 disabled by default), to make development builds quicker. When
1093 building systemd for a full installation with documentation, meson
1094 should be called with -Dman=true and/or -Dhtml=true as appropriate.
1095 The default was changed based on the assumption that quick one-off or
1096 repeated development builds are much more common than full optimized
1097 builds for installation, and people need to pass various other
1098 options to when doing "proper" builds anyway, so the gain from making
1099 development builds quicker is bigger than the one time disruption for
1102 Two scripts are created in the *build* directory to generate and
1103 preview man and html pages on demand, e.g.:
1105 build/man/man systemctl
1106 build/man/html systemd.index
1108 * libidn2 is used by default if both libidn2 and libidn are installed.
1109 Please use -Dlibidn=true if libidn is preferred.
1111 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
1112 big-endian machines. Before, bytes were written and read in native
1113 machine order as exposed by the native libc __cpu_mask interface.
1114 Now, little-endian order is always used (CPUs 0–7 are described by
1115 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
1116 This change fixes D-Bus calls that cross endianness boundary.
1118 The presentation format used for CPUAffinity= by "systemctl show" and
1119 "systemd-analyze dump" is changed to present CPU indices instead of
1120 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
1121 shown as CPUAffinity=03000000000000000000000000000… (on
1122 little-endian) or CPUAffinity=00000000000000300000000000000… (on
1123 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
1124 input format. The maximum integer that will be printed in the new
1125 format is 8191 (four digits), while the old format always used a very
1126 long number (with the length varying by architecture), so they can be
1127 unambiguously distinguished.
1129 * /usr/sbin/halt.local is no longer supported. Implementation in
1130 distributions was inconsistent and it seems this functionality was
1133 To replace this functionality, users should:
1134 - either define a new unit and make it a dependency of final.target
1135 (systemctl add-wants final.target my-halt-local.service)
1136 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
1137 and ensure that it accepts "halt", "poweroff", "reboot", and
1138 "kexec" as an argument, see the description in systemd-shutdown(8).
1140 * When a [Match] section in .link or .network file is empty (contains
1141 no match patterns), a warning will be emitted. Please add any "match
1142 all" pattern instead, e.g. OriginalName=* or Name=* in case all
1143 interfaces should really be matched.
1145 * A new setting NUMAPolicy= may be used to set process memory
1146 allocation policy. This setting can be specified in
1147 /etc/systemd/system.conf and hence will set the default policy for
1148 PID1. The default policy can be overridden on a per-service
1149 basis. The related setting NUMAMask= is used to specify NUMA node
1150 mask that should be associated with the selected policy.
1152 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
1153 generates when processes it manages are reaching their memory limits,
1154 and will place their units in a special state, and optionally kill or
1155 stop the whole unit.
1157 * The service manager will now expose bus properties for the IO
1158 resources used by units. This information is also shown in "systemctl
1159 status" now (for services that have IOAccounting=yes set). Moreover,
1160 the IO accounting data is included in the resource log message
1161 generated whenever a unit stops.
1163 * Units may now configure an explicit timeout to wait for when killed
1164 with SIGABRT, for example when a service watchdog is hit. Previously,
1165 the regular TimeoutStopSec= timeout was applied in this case too —
1166 now a separate timeout may be set using TimeoutAbortSec=.
1168 * Services may now send a special WATCHDOG=trigger message with
1169 sd_notify() to trigger an immediate "watchdog missed" event, and thus
1170 trigger service termination. This is useful both for testing watchdog
1171 handling, but also for defining error paths in services, that shall
1172 be handled the same way as watchdog events.
1174 * There are two new per-unit settings IPIngressFilterPath= and
1175 IPEgressFilterPath= which allow configuration of a BPF program
1176 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
1177 to apply to the IP packet ingress/egress path of all processes of a
1178 unit. This is useful to allow running systemd services with BPF
1179 programs set up externally.
1181 * systemctl gained a new "clean" verb for removing the state, cache,
1182 runtime or logs directories of a service while it is terminated. The
1183 new verb may also be used to remove the state maintained on disk for
1184 timer units that have Persistent= configured.
1186 * During the last phase of shutdown systemd will now automatically
1187 increase the log level configured in the "kernel.printk" sysctl so
1188 that any relevant loggable events happening during late shutdown are
1189 made visible. Previously, loggable events happening so late during
1190 shutdown were generally lost if the "kernel.printk" sysctl was set to
1191 high thresholds, as regular logging daemons are terminated at that
1192 time and thus nothing is written to disk.
1194 * If processes terminated during the last phase of shutdown do not exit
1195 quickly systemd will now show their names after a short time, to make
1196 debugging easier. After a longer timeout they are forcibly killed,
1199 * journalctl (and the other tools that display logs) will now highlight
1200 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
1201 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
1202 are now shown in blue color, to separate them visually from regular
1203 logs. References to configuration files are now turned into clickable
1204 links on terminals that support that.
1206 * systemd-journald will now stop logging to /var/log/journal during
1207 shutdown when /var/ is on a separate mount, so that it can be
1208 unmounted safely during shutdown.
1210 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
1212 * systemd-resolved "Cache=" configuration option in resolved.conf has
1213 been extended to also accept the 'no-negative' value. Previously,
1214 only a boolean option was allowed (yes/no), having yes as the
1215 default. If this option is set to 'no-negative', negative answers are
1216 not cached while the old cache heuristics are used positive answers.
1217 The default remains unchanged.
1219 * The predictable naming scheme for network devices now supports
1220 generating predictable names for "netdevsim" devices.
1222 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
1225 Those two changes form a new net.naming-policy-scheme= entry.
1226 Distributions which want to preserve naming stability may want to set
1227 the -Ddefault-net-naming-scheme= configuration option.
1229 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
1230 interfaces natively.
1232 * systemd-networkd's bridge FDB support now allows configuration of a
1233 destination address for each entry (Destination=), as well as the
1234 VXLAN VNI (VNI=), as well as an option to declare what an entry is
1235 associated with (AssociatedWith=).
1237 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
1238 option for configuring the maximum number of DHCP lease requests. It
1239 also learnt a new BlackList= option for deny-listing DHCP servers (a
1240 similar setting has also been added to the IPv6 RA client), as well
1241 as a SendRelease= option for configuring whether to send a DHCP
1242 RELEASE message when terminating.
1244 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
1245 separately in the [DHCPv4] and [DHCPv6] sections.
1247 * systemd-networkd's DHCP support will now optionally create an
1248 implicit host route to the DNS server specified in the DHCP lease, in
1249 addition to the routes listed explicitly in the lease. This should
1250 ensure that in multi-homed systems DNS traffic leaves the systems on
1251 the interface that acquired the DNS server information even if other
1252 routes such as default routes exist. This behaviour may be turned on
1253 with the new RoutesToDNS= option.
1255 * systemd-networkd's VXLAN support gained a new option
1256 GenericProtocolExtension= for enabling VXLAN Generic Protocol
1257 Extension support, as well as IPDoNotFragment= for setting the IP
1258 "Don't fragment" bit on outgoing packets. A similar option has been
1259 added to the GENEVE support.
1261 * In systemd-networkd's [Route] section you may now configure
1262 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
1263 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
1264 propagation. The Type= setting now supports local, broadcast,
1265 anycast, multicast, any, xresolve routes, too.
1267 * systemd-networkd's [Network] section learnt a new option
1268 DefaultRouteOnDevice= for automatically configuring a default route
1269 onto the network device.
1271 * systemd-networkd's bridging support gained two new options ProxyARP=
1272 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
1273 MulticastRouter= for configuring multicast routing behaviour. A new
1274 option MulticastIGMPVersion= may be used to change bridge's multicast
1275 Internet Group Management Protocol (IGMP) version.
1277 * systemd-networkd's FooOverUDP support gained the ability to configure
1278 local and peer IP addresses via Local= and Peer=. A new option
1279 PeerPort= may be used to configure the peer's IP port.
1281 * systemd-networkd's TUN support gained a new setting VnetHeader= for
1282 tweaking Generic Segment Offload support.
1284 * The address family for policy rules may be specified using the new
1285 Family= option in the [RoutingPolicyRule] section.
1287 * networkctl gained a new "delete" command for removing virtual network
1288 devices, as well as a new "--stats" switch for showing device
1291 * networkd.conf gained a new setting SpeedMeter= and
1292 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
1293 measured speed may be shown by 'networkctl status'.
1295 * "networkctl status" now displays MTU and queue lengths, and more
1296 detailed information about VXLAN and bridge devices.
1298 * systemd-networkd's .network and .link files gained a new Property=
1299 setting in the [Match] section, to match against devices with
1300 specific udev properties.
1302 * systemd-networkd's tunnel support gained a new option
1303 AssignToLoopback= for selecting whether to use the loopback device
1304 "lo" as underlying device.
1306 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
1307 been renamed to LinkLayerAddress=, and it now allows configuration of
1310 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
1311 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
1312 IPv6 configuration (static or DHCPv6) was found for a given
1313 interface. It will not touch the sysctl otherwise.
1315 * The order of entries is $PATH used by the user manager instance was
1316 changed to put bin/ entries before the corresponding sbin/ entries.
1317 It is recommended to not rely on this order, and only ever have one
1318 binary with a given name in the system paths under /usr.
1320 * A new tool systemd-network-generator has been added that may generate
1321 .network, .netdev and .link files from IP configuration specified on
1322 the kernel command line in the format used by Dracut.
1324 * The CriticalConnection= setting in .network files is now deprecated,
1325 and replaced by a new KeepConfiguration= setting which allows more
1326 detailed configuration of the IP configuration to keep in place.
1328 * systemd-analyze gained a few new verbs:
1330 - "systemd-analyze timestamp" parses and converts timestamps. This is
1331 similar to the existing "systemd-analyze calendar" command which
1332 does the same for recurring calendar events.
1334 - "systemd-analyze timespan" parses and converts timespans (i.e.
1335 durations as opposed to points in time).
1337 - "systemd-analyze condition" will parse and test ConditionXYZ=
1340 - "systemd-analyze exit-status" will parse and convert exit status
1341 codes to their names and back.
1343 - "systemd-analyze unit-files" will print a list of all unit
1344 file paths and unit aliases.
1346 * SuccessExitStatus=, RestartPreventExitStatus=, and
1347 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
1348 is equivalent to "65"). Those exit status name mappings may be
1349 displayed with the systemd-analyze exit-status verb describe above.
1351 * systemd-logind now exposes a per-session SetBrightness() bus call,
1352 which may be used to securely change the brightness of a kernel
1353 brightness device, if it belongs to the session's seat. By using this
1354 call unprivileged clients can make changes to "backlight" and "leds"
1355 devices securely with strict requirements on session membership.
1356 Desktop environments may use this to generically make brightness
1357 changes to such devices without shipping private SUID binaries or
1358 udev rules for that purpose.
1360 * "udevadm info" gained a --wait-for-initialization switch to wait for
1361 a device to be initialized.
1363 * systemd-hibernate-resume-generator will now look for resumeflags= on
1364 the kernel command line, which is similar to rootflags= and may be
1365 used to configure device timeout for the hibernation device.
1367 * sd-event learnt a new API call sd_event_source_disable_unref() for
1368 disabling and unref'ing an event source in a single function. A
1369 related call sd_event_source_disable_unrefp() has been added for use
1370 with gcc's cleanup extension.
1372 * The sd-id128.h public API gained a new definition
1373 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
1376 * "busctl introspect" gained a new switch --xml-interface for dumping
1377 XML introspection data unmodified.
1379 * PID 1 may now show the unit name instead of the unit description
1380 string in its status output during boot. This may be configured in
1381 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
1382 kernel command line option systemd.status_unit_format=.
1384 * PID 1 now understands a new option KExecWatchdogSec= in
1385 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
1386 Previously watchdog functionality was only available for regular
1387 reboots. The new setting defaults to off, because we don't know in
1388 the general case if the watchdog will be reset after kexec (some
1389 drivers do reset it, but not all), and the new userspace might not be
1390 configured to handle the watchdog.
1392 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
1393 RebootWatchdogSec= to more clearly communicate what it is about. The
1394 old name is still accepted for compatibility.
1396 * The systemd.debug_shell kernel command line option now optionally
1397 takes a tty name to spawn the debug shell on, which allows a
1398 different tty to be selected than the built-in default.
1400 * Service units gained a new ExecCondition= setting which will run
1401 before ExecStartPre= and either continue execution of the unit (for
1402 clean exit codes), stop execution without marking the unit failed
1403 (for exit codes 1 through 254), or stop execution and fail the unit
1404 (for exit code 255 or abnormal termination).
1406 * A new service systemd-pstore.service has been added that pulls data
1407 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
1410 * timedatectl gained new verbs for configuring per-interface NTP
1411 service configuration for systemd-timesyncd.
1413 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
1414 2019. (You can set non-UTF-8 locales though, if you know their name.)
1416 * If variable assignments in sysctl.d/ files are prefixed with "-" any
1417 failures to apply them are now ignored.
1419 * systemd-random-seed.service now optionally credits entropy when
1420 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
1421 true for the service to enable this behaviour, but please consult the
1422 documentation first, since this comes with a couple of caveats.
1424 * systemd-random-seed.service is now a synchronization point for full
1425 initialization of the kernel's entropy pool. Services that require
1426 /dev/urandom to be correctly initialized should be ordered after this
1429 * The systemd-boot boot loader has been updated to optionally maintain
1430 a random seed file in the EFI System Partition (ESP). During the boot
1431 phase, this random seed is read and updated with a new seed
1432 cryptographically derived from it. Another derived seed is passed to
1433 the OS. The latter seed is then credited to the kernel's entropy pool
1434 very early during userspace initialization (from PID 1). This allows
1435 systems to boot up with a fully initialized kernel entropy pool from
1436 earliest boot on, and thus entirely removes all entropy pool
1437 initialization delays from systems using systemd-boot. Special care
1438 is taken to ensure different seeds are derived on system images
1439 replicated to multiple systems. "bootctl status" will show whether
1440 a seed was received from the boot loader.
1442 * bootctl gained two new verbs:
1444 - "bootctl random-seed" will generate the file in ESP and an EFI
1445 variable to allow a random seed to be passed to the OS as described
1448 - "bootctl is-installed" checks whether systemd-boot is currently
1451 * bootctl will warn if it detects that boot entries are misconfigured
1452 (for example if the kernel image was removed without purging the
1455 * A new document has been added describing systemd's use and support
1456 for the kernel's entropy pool subsystem:
1458 https://systemd.io/RANDOM_SEEDS
1460 * When the system is hibernated the swap device to write the
1461 hibernation image to is now automatically picked from all available
1462 swap devices, preferring the swap device with the highest configured
1463 priority over all others, and picking the device with the most free
1464 space if there are multiple devices with the highest priority.
1466 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
1467 option that permits selecting the timeout how long to wait for a
1468 device with an encryption key before asking for the password.
1470 * IOWeight= has learnt to properly set the IO weight when using the
1471 BFQ scheduler officially found in kernels 5.0+.
1473 * A new mailing list has been created for reporting of security issues:
1474 systemd-security@redhat.com. For mode details, see
1475 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
1477 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
1478 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
1479 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
1480 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
1481 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
1482 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
1483 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
1484 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
1485 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
1486 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
1487 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
1488 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
1489 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
1490 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
1491 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
1492 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
1493 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
1494 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
1495 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
1496 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
1497 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
1498 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
1499 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
1500 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
1501 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
1502 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
1503 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
1504 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
1505 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
1506 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
1508 – Camerino, 2019-09-03
1512 * In .link files, MACAddressPolicy=persistent (the default) is changed
1513 to cover more devices. For devices like bridges, tun, tap, bond, and
1514 similar interfaces that do not have other identifying information,
1515 the interface name is used as the basis for persistent seed for MAC
1516 and IPv4LL addresses. The way that devices that were handled
1517 previously is not changed, and this change is about covering more
1518 devices then previously by the "persistent" policy.
1520 MACAddressPolicy=random may be used to force randomized MACs and
1521 IPv4LL addresses for a device if desired.
1523 Hint: the log output from udev (at debug level) was enhanced to
1524 clarify what policy is followed and which attributes are used.
1525 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
1526 may be used to view this.
1528 Hint: if a bridge interface is created without any slaves, and gains
1529 a slave later, then now the bridge does not inherit slave's MAC.
1530 To inherit slave's MAC, for example, create the following file:
1532 # /etc/systemd/network/98-bridge-inherit-mac.link
1537 MACAddressPolicy=none
1540 * The .device units generated by systemd-fstab-generator and other
1541 generators do not automatically pull in the corresponding .mount unit
1542 as a Wants= dependency. This means that simply plugging in the device
1543 will not cause the mount unit to be started automatically. But please
1544 note that the mount unit may be started for other reasons, in
1545 particular if it is part of local-fs.target, and any unit which
1546 (transitively) depends on local-fs.target is started.
1548 * networkctl list/status/lldp now accept globbing wildcards for network
1549 interface names to match against all existing interfaces.
1551 * The $PIDFILE environment variable is set to point the absolute path
1552 configured with PIDFile= for processes of that service.
1554 * The fallback DNS server list was augmented with Cloudflare public DNS
1555 servers. Use `-Ddns-servers=` to set a different fallback.
1557 * A new special target usb-gadget.target will be started automatically
1558 when a USB Device Controller is detected (which means that the system
1559 is a USB peripheral).
1561 * A new unit setting CPUQuotaPeriodSec= assigns the time period
1562 relatively to which the CPU time quota specified by CPUQuota= is
1565 * A new unit setting ProtectHostname= may be used to prevent services
1566 from modifying hostname information (even if they otherwise would
1567 have privileges to do so).
1569 * A new unit setting NetworkNamespacePath= may be used to specify a
1570 namespace for service or socket units through a path referring to a
1571 Linux network namespace pseudo-file.
1573 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
1574 have an effect on .socket units: when used the listening socket is
1575 created within the configured network namespace instead of the host
1578 * ExecStart= command lines in unit files may now be prefixed with ':'
1579 in which case environment variable substitution is
1580 disabled. (Supported for the other ExecXYZ= settings, too.)
1582 * .timer units gained two new boolean settings OnClockChange= and
1583 OnTimezoneChange= which may be used to also trigger a unit when the
1584 system clock is changed or the local timezone is
1585 modified. systemd-run has been updated to make these options easily
1586 accessible from the command line for transient timers.
1588 * Two new conditions for units have been added: ConditionMemory= may be
1589 used to conditionalize a unit based on installed system
1590 RAM. ConditionCPUs= may be used to conditionalize a unit based on
1591 installed CPU cores.
1593 * The @default system call filter group understood by SystemCallFilter=
1594 has been updated to include the new rseq() system call introduced in
1597 * A new time-set.target has been added that indicates that the system
1598 time has been set from a local source (possibly imprecise). The
1599 existing time-sync.target is stronger and indicates that the time has
1600 been synchronized with a precise external source. Services where
1601 approximate time is sufficient should use the new target.
1603 * "systemctl start" (and related commands) learnt a new
1604 --show-transaction option. If specified brief information about all
1605 jobs queued because of the requested operation is shown.
1607 * systemd-networkd recognizes a new operation state 'enslaved', used
1608 (instead of 'degraded' or 'carrier') for interfaces which form a
1609 bridge, bond, or similar, and an new 'degraded-carrier' operational
1610 state used for the bond or bridge master interface when one of the
1611 enslaved devices is not operational.
1613 * .network files learnt the new IgnoreCarrierLoss= option for leaving
1614 networks configured even if the carrier is lost.
1616 * The RequiredForOnline= setting in .network files may now specify a
1617 minimum operational state required for the interface to be considered
1618 "online" by systemd-networkd-wait-online. Related to this
1619 systemd-networkd-wait-online gained a new option --operational-state=
1620 to configure the same, and its --interface= option was updated to
1621 optionally also take an operational state specific for an interface.
1623 * systemd-networkd-wait-online gained a new setting --any for waiting
1624 for only one of the requested interfaces instead of all of them.
1626 * systemd-networkd now implements L2TP tunnels.
1628 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
1629 may be used to cause autonomous and onlink prefixes received in IPv6
1630 Router Advertisements to be ignored.
1632 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
1633 file settings may be used to tweak bridge behaviour.
1635 * The new TripleSampling= option in .network files may be used to
1636 configure CAN triple sampling.
1638 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
1639 used to point to private or preshared key for a WireGuard interface.
1641 * /etc/crypttab now supports the same-cpu-crypt and
1642 submit-from-crypt-cpus options to tweak encryption work scheduling
1645 * systemd-tmpfiles will now take a BSD file lock before operating on a
1646 contents of directory. This may be used to temporarily exclude
1647 directories from aging by taking the same lock (useful for example
1648 when extracting a tarball into /tmp or /var/tmp as a privileged user,
1649 which might create files with really old timestamps, which
1650 nevertheless should not be deleted). For further details, see:
1652 https://systemd.io/TEMPORARY_DIRECTORIES
1654 * systemd-tmpfiles' h line type gained support for the
1655 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
1656 controlling project quota inheritance.
1658 * sd-boot and bootctl now implement support for an Extended Boot Loader
1659 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
1660 addition to the ESP partition mounted to /efi or /boot/efi.
1661 Configuration file fragments, kernels, initrds and other EFI images
1662 to boot will be loaded from both the ESP and XBOOTLDR partitions.
1663 The XBOOTLDR partition was previously described by the Boot Loader
1664 Specification, but implementation was missing in sd-boot. Support for
1665 this concept allows using the sd-boot boot loader in more
1666 conservative scenarios where the boot loader itself is placed in the
1667 ESP but the kernels to boot (and their metadata) in a separate
1670 * A system may now be booted with systemd.volatile=overlay on the
1671 kernel command line, which causes the root file system to be set up
1672 an overlayfs mount combining the root-only root directory with a
1673 writable tmpfs. In this setup, the underlying root device is not
1674 modified, and any changes are lost at reboot.
1676 * Similar, systemd-nspawn can now boot containers with a volatile
1677 overlayfs root with the new --volatile=overlay switch.
1679 * systemd-nspawn can now consume OCI runtime bundles using a new
1680 --oci-bundle= option. This implementation is fully usable, with most
1681 features in the specification implemented, but since this a lot of
1682 new code and functionality, this feature should most likely not
1683 be used in production yet.
1685 * systemd-nspawn now supports various options described by the OCI
1686 runtime specification on the command-line and in .nspawn files:
1687 --inaccessible=/Inaccessible= may be used to mask parts of the file
1688 system tree, --console=/--pipe may be used to configure how standard
1689 input, output, and error are set up.
1691 * busctl learned the `emit` verb to generate D-Bus signals.
1693 * systemd-analyze cat-config may be used to gather and display
1694 configuration spread over multiple files, for example system and user
1695 presets, tmpfiles.d, sysusers.d, udev rules, etc.
1697 * systemd-analyze calendar now takes an optional new parameter
1698 --iterations= which may be used to show a maximum number of iterations
1699 the specified expression will elapse next.
1701 * The sd-bus C API gained support for naming method parameters in the
1704 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
1705 the reboot() system call expects.
1707 * journalctl learnt a new --cursor-file= option that points to a file
1708 from which a cursor should be loaded in the beginning and to which
1709 the updated cursor should be stored at the end.
1711 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
1712 detected by systemd-detect-virt (and may also be used in
1713 ConditionVirtualization=).
1715 * The behaviour of systemd-logind may now be modified with environment
1716 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
1717 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
1718 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
1719 skip the relevant operation completely (when set to false), or to
1720 create a flag file in /run/systemd (when set to true), instead of
1721 actually commencing the real operation when requested. The presence
1722 of /run/systemd/reboot-to-firmware-setup,
1723 /run/systemd/reboot-to-boot-loader-menu, and
1724 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
1725 boot loader implementations to replace some steps logind performs
1726 during reboot with their own operations.
1728 * systemctl can be used to request a reboot into the boot loader menu
1729 or a specific boot loader entry with the new --boot-load-menu= and
1730 --boot-loader-entry= options to a reboot command. (This requires a
1731 boot loader that supports this, for example sd-boot.)
1733 * kernel-install will no longer unconditionally create the output
1734 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
1735 snippets, but will do only if the machine-specific parent directory
1736 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
1737 to create this parent directory during sd-boot installation.
1739 This makes it easier to use kernel-install with plugins which support
1740 a different layout of the bootloader partitions (for example grub2).
1742 * During package installation (with `ninja install`), we would create
1743 symlinks for getty@tty1.service, systemd-networkd.service,
1744 systemd-networkd.socket, systemd-resolved.service,
1745 remote-cryptsetup.target, remote-fs.target,
1746 systemd-networkd-wait-online.service, and systemd-timesyncd.service
1747 in /etc, as if `systemctl enable` was called for those units, to make
1748 the system usable immediately after installation. Now this is not
1749 done anymore, and instead calling `systemctl preset-all` is
1750 recommended after the first installation of systemd.
1752 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
1753 is built on seccomp. When turned on creation of SUID/SGID files is
1756 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
1757 implied if DynamicUser= is turned on for a service. This hardens
1758 these services, so that they neither can benefit from nor create
1759 SUID/SGID executables. This is a minor compatibility breakage, given
1760 that when DynamicUser= was first introduced SUID/SGID behaviour was
1761 unaffected. However, the security benefit of these two options is
1762 substantial, and the setting is still relatively new, hence we opted
1763 to make it mandatory for services with dynamic users.
1765 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
1766 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
1767 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
1768 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
1769 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
1770 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
1771 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
1772 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
1773 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
1774 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
1775 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
1776 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
1777 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
1778 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
1779 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
1780 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
1781 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
1782 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1784 — Warsaw, 2019-04-11
1788 * The default locale can now be configured at compile time. Otherwise,
1789 a suitable default will be selected automatically (one of C.UTF-8,
1790 en_US.UTF-8, and C).
1792 * The version string shown by systemd and other tools now includes the
1793 git commit hash when built from git. An override may be specified
1794 during compilation, which is intended to be used by distributions to
1795 include the package release information.
1797 * systemd-cat can now filter standard input and standard error streams
1798 for different syslog priorities using the new --stderr-priority=
1801 * systemd-journald and systemd-journal-remote reject entries which
1802 contain too many fields (CVE-2018-16865) and set limits on the
1803 process' command line length (CVE-2018-16864).
1805 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
1808 * A new network device NamePolicy "keep" is implemented for link files,
1809 and used by default in 99-default.link (the fallback configuration
1810 provided by systemd). With this policy, if the network device name
1811 was already set by userspace, the device will not be renamed again.
1812 This matches the naming scheme that was implemented before
1813 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
1814 is also enabled by default, even if not specified. Effectively, this
1815 means that if naming-scheme >= 240 is specified, network devices will
1816 be renamed according to the configuration, even if they have been
1817 renamed already, if "keep" is not specified as the naming policy in
1818 the .link file. The 99-default.link file provided by systemd includes
1819 "keep" for backwards compatibility, but it is recommended for user
1820 installed .link files to *not* include it.
1822 The "kernel" policy, which keeps kernel names declared to be
1823 "persistent", now works again as documented.
1825 * kernel-install script now optionally takes the paths to one or more
1826 initrd files, and passes them to all plugins.
1828 * The mincore() system call has been dropped from the @system-service
1829 system call filter group, as it is pretty exotic and may potentially
1830 used for side-channel attacks.
1832 * -fPIE is dropped from compiler and linker options. Please specify
1833 -Db_pie=true option to meson to build position-independent
1834 executables. Note that the meson option is supported since meson-0.49.
1836 * The fs.protected_regular and fs.protected_fifos sysctls, which were
1837 added in Linux 4.19 to make some data spoofing attacks harder, are
1838 now enabled by default. While this will hopefully improve the
1839 security of most installations, it is technically a backwards
1840 incompatible change; to disable these sysctls again, place the
1841 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
1843 fs.protected_regular = 0
1844 fs.protected_fifos = 0
1846 Note that the similar hardlink and symlink protection has been
1847 enabled since v199, and may be disabled likewise.
1849 * The files read from the EnvironmentFile= setting in unit files now
1850 parse backslashes inside quotes literally, matching the behaviour of
1853 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
1854 now automatically become NOPs when run in a chroot() environment.
1856 * The tmpfiles.d/ "C" line type will now copy directory trees not only
1857 when the destination is so far missing, but also if it already exists
1858 as a directory and is empty. This is useful to cater for systems
1859 where directory trees are put together from multiple separate mount
1860 points but otherwise empty.
1862 * A new function sd_bus_close_unref() (and the associated
1863 sd_bus_close_unrefp()) has been added to libsystemd, that combines
1864 sd_bus_close() and sd_bus_unref() in one.
1866 * udevadm control learnt a new option for --ping for testing whether a
1867 systemd-udevd instance is running and reacting.
1869 * udevadm trigger learnt a new option for --wait-daemon for waiting
1870 systemd-udevd daemon to be initialized.
1872 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
1873 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
1874 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
1875 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
1876 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
1877 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
1878 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
1879 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
1880 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
1881 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
1882 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
1883 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
1884 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
1885 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
1886 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
1887 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
1888 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
1890 — Berlin, 2019-02-14
1894 * NoNewPrivileges=yes has been set for all long-running services
1895 implemented by systemd. Previously, this was problematic due to
1896 SELinux (as this would also prohibit the transition from PID1's label
1897 to the service's label). This restriction has since been lifted, but
1898 an SELinux policy update is required.
1899 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
1901 * DynamicUser=yes is dropped from systemd-networkd.service,
1902 systemd-resolved.service and systemd-timesyncd.service, which was
1903 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
1904 and since v236 for systemd-timesyncd.service. The users and groups
1905 systemd-network, systemd-resolve and systemd-timesync are created
1906 by systemd-sysusers again. Distributors or system administrators
1907 may need to create these users and groups if they not exist (or need
1908 to re-enable DynamicUser= for those units) while upgrading systemd.
1909 Also, the clock file for systemd-timesyncd may need to move from
1910 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
1912 * When unit files are loaded from disk, previously systemd would
1913 sometimes (depending on the unit loading order) load units from the
1914 target path of symlinks in .wants/ or .requires/ directories of other
1915 units. This meant that unit could be loaded from different paths
1916 depending on whether the unit was requested explicitly or as a
1917 dependency of another unit, not honouring the priority of directories
1918 in search path. It also meant that it was possible to successfully
1919 load and start units which are not found in the unit search path, as
1920 long as they were requested as a dependency and linked to from
1921 .wants/ or .requires/. The target paths of those symlinks are not
1922 used for loading units anymore and the unit file must be found in
1925 * A new service type has been added: Type=exec. It's very similar to
1926 Type=simple but ensures the service manager will wait for both fork()
1927 and execve() of the main service binary to complete before proceeding
1928 with follow-up units. This is primarily useful so that the manager
1929 propagates any errors in the preparation phase of service execution
1930 back to the job that requested the unit to be started. For example,
1931 consider a service that has ExecStart= set to a file system binary
1932 that doesn't exist. With Type=simple starting the unit would be
1933 considered instantly successful, as only fork() has to complete
1934 successfully and the manager does not wait for execve(), and hence
1935 its failure is seen "too late". With the new Type=exec service type
1936 starting the unit will fail, as the manager will wait for the
1937 execve() and notice its failure, which is then propagated back to the
1940 NOTE: with the next release 241 of systemd we intend to change the
1941 systemd-run tool to default to Type=exec for transient services
1942 started by it. This should be mostly safe, but in specific corner
1943 cases might result in problems, as the systemd-run tool will then
1944 block on NSS calls (such as user name look-ups due to User=) done
1945 between the fork() and execve(), which under specific circumstances
1946 might cause problems. It is recommended to specify "-p Type=simple"
1947 explicitly in the few cases where this applies. For regular,
1948 non-transient services (i.e. those defined with unit files on disk)
1949 we will continue to default to Type=simple.
1951 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
1952 userspace processes is set to 1024 (soft) and 4096
1953 (hard). Previously, systemd passed this on unmodified to all
1954 processes it forked off. With this systemd release the hard limit
1955 systemd passes on is increased to 512K, overriding the kernel's
1956 defaults and substantially increasing the number of simultaneous file
1957 descriptors unprivileged userspace processes can allocate. Note that
1958 the soft limit remains at 1024 for compatibility reasons: the
1959 traditional UNIX select() call cannot deal with file descriptors >=
1960 1024 and increasing the soft limit globally might thus result in
1961 programs unexpectedly allocating a high file descriptor and thus
1962 failing abnormally when attempting to use it with select() (of
1963 course, programs shouldn't use select() anymore, and prefer
1964 poll()/epoll, but the call unfortunately remains undeservedly popular
1965 at this time). This change reflects the fact that file descriptor
1966 handling in the Linux kernel has been optimized in more recent
1967 kernels and allocating large numbers of them should be much cheaper
1968 both in memory and in performance than it used to be. Programs that
1969 want to take benefit of the increased limit have to "opt-in" into
1970 high file descriptors explicitly by raising their soft limit. Of
1971 course, when they do that they must acknowledge that they cannot use
1972 select() anymore (and neither can any shared library they use — or
1973 any shared library used by any shared library they use and so on).
1974 Which default hard limit is most appropriate is of course hard to
1975 decide. However, given reports that ~300K file descriptors are used
1976 in real-life applications we believe 512K is sufficiently high as new
1977 default for now. Note that there are also reports that using very
1978 high hard limits (e.g. 1G) is problematic: some software allocates
1979 large arrays with one element for each potential file descriptor
1980 (Java, …) — a high hard limit thus triggers excessively large memory
1981 allocations in these applications. Hopefully, the new default of 512K
1982 is a good middle ground: higher than what real-life applications
1983 currently need, and low enough for avoid triggering excessively large
1984 allocations in problematic software. (And yes, somebody should fix
1987 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
1988 to the highest possible values, as separate accounting of file
1989 descriptors is no longer necessary, as memcg tracks them correctly as
1990 part of the memory accounting anyway. Thus, from the four limits on
1991 file descriptors currently enforced (fs.file-max, fs.nr_open,
1992 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
1993 and keep only the latter two. A set of build-time options
1994 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
1995 has been added to revert this change in behaviour, which might be
1996 an option for systems that turn off memcg in the kernel.
1998 * When no /etc/locale.conf file exists (and hence no locale settings
1999 are in place), systemd will now use the "C.UTF-8" locale by default,
2000 and set LANG= to it. This locale is supported by various
2001 distributions including Fedora, with clear indications that upstream
2002 glibc is going to make it available too. This locale enables UTF-8
2003 mode by default, which appears appropriate for 2018.
2005 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
2006 default. This effectively switches the RFC3704 Reverse Path filtering
2007 from Strict mode to Loose mode. This is more appropriate for hosts
2008 that have multiple links with routes to the same networks (e.g.
2009 a client with a Wi-Fi and Ethernet both connected to the internet).
2011 Consult the kernel documentation for details on this sysctl:
2012 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
2014 * CPUAccounting=yes no longer enables the CPU controller when using
2015 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
2016 statistics are now provided independently from the CPU controller.
2018 * Support for disabling a particular cgroup controller within a sub-tree
2019 has been added through the DisableControllers= directive.
2021 * cgroup_no_v1=all on the kernel command line now also implies
2022 using the unified cgroup hierarchy, unless one explicitly passes
2023 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
2025 * The new "MemoryMin=" unit file property may now be used to set the
2026 memory usage protection limit of processes invoked by the unit. This
2027 controls the cgroup v2 memory.min attribute. Similarly, the new
2028 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
2029 cgroup v2 io.latency cgroup property for configuring per-service I/O
2032 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
2033 to the cgroup v1 "devices" cgroup controller.
2035 * systemd-escape now is able to combine --unescape with --template. It
2036 also learnt a new option --instance for extracting and unescaping the
2037 instance part of a unit name.
2039 * sd-bus now provides the sd_bus_message_readv() which is similar to
2040 sd_bus_message_read() but takes a va_list object. The pair
2041 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
2042 has been added for configuring the default method call timeout to
2043 use. sd_bus_error_move() may be used to efficiently move the contents
2044 from one sd_bus_error structure to another, invalidating the
2045 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
2046 be used to control whether a bus connection object is automatically
2047 flushed when an sd-event loop is exited.
2049 * When processing classic BSD syslog log messages, journald will now
2050 save the original time-stamp string supplied in the new
2051 SYSLOG_TIMESTAMP= journal field. This permits consumers to
2052 reconstruct the original BSD syslog message more correctly.
2054 * StandardOutput=/StandardError= in service files gained support for
2055 new "append:…" parameters, for connecting STDOUT/STDERR of a service
2056 to a file, and appending to it.
2058 * The signal to use as last step of killing of unit processes is now
2059 configurable. Previously it was hard-coded to SIGKILL, which may now
2060 be overridden with the new KillSignal= setting. Note that this is the
2061 signal used when regular termination (i.e. SIGTERM) does not suffice.
2062 Similarly, the signal used when aborting a program in case of a
2063 watchdog timeout may now be configured too (WatchdogSignal=).
2065 * The XDG_SESSION_DESKTOP environment variable may now be configured in
2066 the pam_systemd argument line, using the new desktop= switch. This is
2067 useful to initialize it properly from a display manager without
2068 having to touch C code.
2070 * Most configuration options that previously accepted percentage values
2071 now also accept permille values with the '‰' suffix (instead of '%').
2073 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
2076 * systemd-resolved's configuration file resolved.conf gained a new
2077 option ReadEtcHosts= which may be used to turn off processing and
2078 honoring /etc/hosts entries.
2080 * The "--wait" switch may now be passed to "systemctl
2081 is-system-running", in which case the tool will synchronously wait
2082 until the system finished start-up.
2084 * hostnamed gained a new bus call to determine the DMI product UUID.
2086 * On x86-64 systemd will now prefer using the RDRAND processor
2087 instruction over /dev/urandom whenever it requires randomness that
2088 neither has to be crypto-grade nor should be reproducible. This
2089 should substantially reduce the amount of entropy systemd requests
2090 from the kernel during initialization on such systems, though not
2091 reduce it to zero. (Why not zero? systemd still needs to allocate
2092 UUIDs and such uniquely, which require high-quality randomness.)
2094 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
2095 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
2096 for forcing the "Other Information" bit in IPv6 RA messages. The
2097 bonding logic gained four new options AdActorSystemPriority=,
2098 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
2099 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
2100 shuffling of flows. The tunnel logic gained a new
2101 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
2102 Deployment. The policy rule logic gained four new options IPProtocol=,
2103 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
2104 support for the MulticastToUnicast= option. networkd also gained
2105 support for configuring static IPv4 ARP or IPv6 neighbor entries.
2107 * .preset files (as read by 'systemctl preset') may now be used to
2108 instantiate services.
2110 * /etc/crypttab now understands the sector-size= option to configure
2111 the sector size for an encrypted partition.
2113 * Key material for encrypted disks may now be placed on a formatted
2114 medium, and referenced from /etc/crypttab by the UUID of the file
2115 system, followed by "=" suffixed by the path to the key file.
2117 * The "collect" udev component has been removed without replacement, as
2118 it is neither used nor maintained.
2120 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
2121 LogsDirectory=, ConfigurationDirectory= settings are used in a
2122 service the executed processes will now receive a set of environment
2123 variables containing the full paths of these directories.
2124 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
2125 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
2126 are used. Note that these options may be used multiple times per
2127 service in which case the resulting paths will be concatenated and
2128 separated by colons.
2130 * Predictable interface naming has been extended to cover InfiniBand
2131 NICs. They will be exposed with an "ib" prefix.
2133 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
2134 which case the respective line failing is ignored.
2136 * .link files may now be used to configure the equivalent to the
2137 "ethtool advertise" commands.
2139 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
2140 alternative to libudev.h. Previously, the latter was just an internal
2141 wrapper around the former, but now these two APIs are exposed
2144 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
2145 which calculates an app-specific boot ID similar to how
2146 sd_id128_get_machine_app_specific() generates an app-specific machine
2149 * A new tool systemd-id128 has been added that can be used to determine
2150 and generate various 128bit IDs.
2152 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
2155 * systemd-hibernate-resume-generator will now honor the "noresume"
2156 kernel command line option, in which case it will bypass resuming
2157 from any hibernated image.
2159 * The systemd-sleep.conf configuration file gained new options
2160 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
2161 AllowHybridSleep= for prohibiting specific sleep modes even if the
2162 kernel exports them.
2164 * portablectl is now officially supported and has thus moved to
2167 * bootctl learnt the two new commands "set-default" and "set-oneshot"
2168 for setting the default boot loader item to boot to (either
2169 persistently or only for the next boot). This is currently only
2170 compatible with sd-boot, but may be implemented on other boot loaders
2171 too, that follow the boot loader interface. The updated interface is
2172 now documented here:
2174 https://systemd.io/BOOT_LOADER_INTERFACE
2176 * A new kernel command line option systemd.early_core_pattern= is now
2177 understood which may be used to influence the core_pattern PID 1
2178 installs during early boot.
2180 * busctl learnt two new options -j and --json= for outputting method
2181 call replies, properties and monitoring output in JSON.
2183 * journalctl's JSON output now supports simple ANSI coloring as well as
2184 a new "json-seq" mode for generating RFC7464 output.
2186 * Unit files now support the %g/%G specifiers that resolve to the UNIX
2187 group/GID of the service manager runs as, similar to the existing
2188 %u/%U specifiers that resolve to the UNIX user/UID.
2190 * systemd-logind learnt a new global configuration option
2191 UserStopDelaySec= that may be set in logind.conf. It specifies how
2192 long the systemd --user instance shall remain started after a user
2193 logs out. This is useful to speed up repetitive re-connections of the
2194 same user, as it means the user's service manager doesn't have to be
2195 stopped/restarted on each iteration, but can be reused between
2196 subsequent options. This setting defaults to 10s. systemd-logind also
2197 exports two new properties on its Manager D-Bus objects indicating
2198 whether the system's lid is currently closed, and whether the system
2201 * systemd gained support for a generic boot counting logic, which
2202 generically permits automatic reverting to older boot loader entries
2203 if newer updated ones don't work. The boot loader side is implemented
2204 in sd-boot, but is kept open for other boot loaders too. For details
2207 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
2209 * The SuccessAction=/FailureAction= unit file settings now learnt two
2210 new parameters: "exit" and "exit-force", which result in immediate
2211 exiting of the service manager, and are only useful in systemd --user
2212 and container environments.
2214 * Unit files gained support for a pair of options
2215 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
2216 exit status to use as service manager exit status when
2217 SuccessAction=/FailureAction= is set to exit or exit-force.
2219 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
2220 options may now be used to configure the log rate limiting applied by
2221 journald per-service.
2223 * systemd-analyze gained a new verb "timespan" for parsing and
2224 normalizing time span values (i.e. strings like "5min 7s 8us").
2226 * systemd-analyze also gained a new verb "security" for analyzing the
2227 security and sand-boxing settings of services in order to determine an
2228 "exposure level" for them, indicating whether a service would benefit
2229 from more sand-boxing options turned on for them.
2231 * "systemd-analyze syscall-filter" will now also show system calls
2232 supported by the local kernel but not included in any of the defined
2235 * .nspawn files now understand the Ephemeral= setting, matching the
2236 --ephemeral command line switch.
2238 * sd-event gained the new APIs sd_event_source_get_floating() and
2239 sd_event_source_set_floating() for controlling whether a specific
2240 event source is "floating", i.e. destroyed along with the even loop
2243 * Unit objects on D-Bus gained a new "Refs" property that lists all
2244 clients that currently have a reference on the unit (to ensure it is
2247 * The JoinControllers= option in system.conf is no longer supported, as
2248 it didn't work correctly, is hard to support properly, is legacy (as
2249 the concept only exists on cgroup v1) and apparently wasn't used.
2251 * Journal messages that are generated whenever a unit enters the failed
2252 state are now tagged with a unique MESSAGE_ID. Similarly, messages
2253 generated whenever a service process exits are now made recognizable,
2254 too. A tagged message is also emitted whenever a unit enters the
2255 "dead" state on success.
2257 * systemd-run gained a new switch --working-directory= for configuring
2258 the working directory of the service to start. A shortcut -d is
2259 equivalent, setting the working directory of the service to the
2260 current working directory of the invoking program. The new --shell
2261 (or just -S) option has been added for invoking the $SHELL of the
2262 caller as a service, and implies --pty --same-dir --wait --collect
2263 --service-type=exec. Or in other words, "systemd-run -S" is now the
2264 quickest way to quickly get an interactive in a fully clean and
2265 well-defined system service context.
2267 * machinectl gained a new verb "import-fs" for importing an OS tree
2268 from a directory. Moreover, when a directory or tarball is imported
2269 and single top-level directory found with the OS itself below the OS
2270 tree is automatically mangled and moved one level up.
2272 * systemd-importd will no longer set up an implicit btrfs loop-back
2273 file system on /var/lib/machines. If one is already set up, it will
2274 continue to be used.
2276 * A new generator "systemd-run-generator" has been added. It will
2277 synthesize a unit from one or more program command lines included in
2278 the kernel command line. This is very useful in container managers
2281 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
2283 This will run "systemd-nspawn" on an image, invoke the specified
2284 command line and immediately shut down the container again, returning
2285 the command line's exit code.
2287 * The block device locking logic is now documented:
2289 https://systemd.io/BLOCK_DEVICE_LOCKING
2291 * loginctl and machinectl now optionally output the various tables in
2292 JSON using the --output= switch. It is our intention to add similar
2293 support to systemctl and all other commands.
2295 * udevadm's query and trigger verb now optionally take a .device unit
2298 * systemd-udevd's network naming logic now understands a new
2299 net.naming-scheme= kernel command line switch, which may be used to
2300 pick a specific version of the naming scheme. This helps stabilizing
2301 interface names even as systemd/udev are updated and the naming logic
2304 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
2305 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
2306 initialize one to all 0xFF.
2308 * After loading the SELinux policy systemd will now recursively relabel
2309 all files and directories listed in
2310 /run/systemd/relabel-extra.d/*.relabel (which should be simple
2311 newline separated lists of paths) in addition to the ones it already
2312 implicitly relabels in /run, /dev and /sys. After the relabelling is
2313 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
2314 removed. This is useful to permit initrds (i.e. code running before
2315 the SELinux policy is in effect) to generate files in the host
2316 filesystem safely and ensure that the correct label is applied during
2317 the transition to the host OS.
2319 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
2320 mknod() handling in user namespaces. Previously mknod() would always
2321 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
2322 but device nodes generated that way cannot be opened, and attempts to
2323 open them result in EPERM. This breaks the "graceful fallback" logic
2324 in systemd's PrivateDevices= sand-boxing option. This option is
2325 implemented defensively, so that when systemd detects it runs in a
2326 restricted environment (such as a user namespace, or an environment
2327 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
2328 where device nodes cannot be created the effect of PrivateDevices= is
2329 bypassed (following the logic that 2nd-level sand-boxing is not
2330 essential if the system systemd runs in is itself already sand-boxed
2331 as a whole). This logic breaks with 4.18 in container managers where
2332 user namespacing is used: suddenly PrivateDevices= succeeds setting
2333 up a private /dev/ file system containing devices nodes — but when
2334 these are opened they don't work.
2336 At this point it is recommended that container managers utilizing
2337 user namespaces that intend to run systemd in the payload explicitly
2338 block mknod() with seccomp or similar, so that the graceful fallback
2341 We are very sorry for the breakage and the requirement to change
2342 container configurations for newer kernels. It's purely caused by an
2343 incompatible kernel change. The relevant kernel developers have been
2344 notified about this userspace breakage quickly, but they chose to
2347 * PermissionsStartOnly= setting is deprecated (but is still supported
2348 for backwards compatibility). The same functionality is provided by
2349 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
2352 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
2353 pam_systemd anymore.
2355 * The naming scheme for network devices was changed to always rename
2356 devices, even if they were already renamed by userspace. The "kernel"
2357 policy was changed to only apply as a fallback, if no other naming
2360 * The requirements to build systemd is bumped to meson-0.46 and
2363 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
2364 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
2365 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
2366 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
2367 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
2368 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
2369 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
2370 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
2371 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
2372 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
2373 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
2374 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
2375 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
2376 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
2377 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
2378 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
2379 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
2380 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
2381 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
2382 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
2383 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
2384 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
2385 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
2386 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
2387 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
2388 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
2389 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
2390 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
2391 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
2392 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
2393 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
2394 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
2395 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
2396 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
2397 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
2398 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
2399 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
2400 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
2401 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
2402 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
2403 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
2404 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
2405 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
2406 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
2407 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
2409 — Warsaw, 2018-12-21
2413 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
2414 builtin will name network interfaces differently than in previous
2415 versions for virtual network interfaces created with SR-IOV and NPAR
2416 and for devices where the PCI network controller device does not have
2417 a slot number associated.
2419 SR-IOV virtual devices are now named based on the name of the parent
2420 interface, with a suffix of "v<N>", where <N> is the virtual device
2421 number. Previously those virtual devices were named as if completely
2424 The ninth and later NPAR virtual devices will be named following the
2425 scheme used for the first eight NPAR partitions. Previously those
2426 devices were not renamed and the kernel default (eth<n>) was used.
2428 "net_id" will also generate names for PCI devices where the PCI
2429 network controller device does not have an associated slot number
2430 itself, but one of its parents does. Previously those devices were
2431 not renamed and the kernel default (eth<n>) was used.
2433 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
2434 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
2435 the unit. So, it is expected that the default behavior of
2436 systemd-logind is not changed. However, if distribution packagers or
2437 administrators disabled or modified IPAddressDeny= setting by a
2438 drop-in config file, then it may be necessary to update the file to
2439 re-enable AF_INET and AF_INET6 to support network user name services,
2442 * When the RestrictNamespaces= unit property is specified multiple
2443 times, then the specified types are merged now. Previously, only the
2444 last assignment was used. So, if distribution packagers or
2445 administrators modified the setting by a drop-in config file, then it
2446 may be necessary to update the file.
2448 * When OnFailure= is used in combination with Restart= on a service
2449 unit, then the specified units will no longer be triggered on
2450 failures that result in restarting. Previously, the specified units
2451 would be activated each time the unit failed, even when the unit was
2452 going to be restarted automatically. This behaviour contradicted the
2453 documentation. With this release the code is adjusted to match the
2456 * systemd-tmpfiles will now print a notice whenever it encounters
2457 tmpfiles.d/ lines referencing the /var/run/ directory. It will
2458 recommend reworking them to use the /run/ directory instead (for
2459 which /var/run/ is simply a symlinked compatibility alias). This way
2460 systemd-tmpfiles can properly detect line conflicts and merge lines
2461 referencing the same file by two paths, without having to access
2464 * systemctl disable/unmask/preset/preset-all cannot be used with
2465 --runtime. Previously this was allowed, but resulted in unintuitive
2466 behaviour that wasn't useful. systemctl disable/unmask will now undo
2467 both runtime and persistent enablement/masking, i.e. it will remove
2468 any relevant symlinks both in /run and /etc.
2470 * Note that all long-running system services shipped with systemd will
2471 now default to a system call allow list (rather than a deny list, as
2472 before). In particular, systemd-udevd will now enforce one too. For
2473 most cases this should be safe, however downstream distributions
2474 which disabled sandboxing of systemd-udevd (specifically the
2475 MountFlags= setting), might want to disable this security feature
2476 too, as the default allow-listing will prohibit all mount, swap,
2477 reboot and clock changing operations from udev rules.
2479 * sd-boot acquired new loader configuration settings to optionally turn
2480 off Windows and MacOS boot partition discovery as well as
2481 reboot-into-firmware menu items. It is also able to pick a better
2482 screen resolution for HiDPI systems, and now provides loader
2483 configuration settings to change the resolution explicitly.
2485 * systemd-resolved now supports DNS-over-TLS. It's still
2486 turned off by default, use DNSOverTLS=opportunistic to turn it on in
2487 resolved.conf. We intend to make this the default as soon as couple
2488 of additional techniques for optimizing the initial latency caused by
2489 establishing a TLS/TCP connection are implemented.
2491 * systemd-resolved.service and systemd-networkd.service now set
2492 DynamicUser=yes. The users systemd-resolve and systemd-network are
2493 not created by systemd-sysusers anymore.
2495 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
2496 that embed a network facing module into any process using getpwuid()
2497 or related call: the dynamic allocation of the user ID for
2498 systemd-resolved.service means the service manager has to check NSS
2499 if the user name is already taken when forking off the service. Since
2500 the user in the common case won't be defined in /etc/passwd the
2501 lookup is likely to trigger nss-ldap which in turn might use NSS to
2502 ask systemd-resolved for hostname lookups. This will hence result in
2503 a deadlock: a user name lookup in order to start
2504 systemd-resolved.service will result in a hostname lookup for which
2505 systemd-resolved.service needs to be started already. There are
2506 multiple ways to work around this problem: pre-allocate the
2507 "systemd-resolve" user on such systems, so that nss-ldap won't be
2508 triggered; or use a different NSS package that doesn't do networking
2509 in-process but provides a local asynchronous name cache; or configure
2510 the NSS package to avoid lookups for UIDs in the range `pkg-config
2511 systemd --variable=dynamicuidmin` … `pkg-config systemd
2512 --variable=dynamicuidmax`, so that it does not consider itself
2513 authoritative for the same UID range systemd allocates dynamic users
2516 * The systemd-resolve tool has been renamed to resolvectl (it also
2517 remains available under the old name, for compatibility), and its
2518 interface is now verb-based, similar in style to the other <xyz>ctl
2519 tools, such as systemctl or loginctl.
2521 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
2522 compatibility. It may be symlinked under the 'resolvconf' name, in
2523 which case it will take arguments and input compatible with the
2524 Debian and FreeBSD resolvconf tool.
2526 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
2527 where the system initially suspends, and after a timeout resumes and
2530 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
2531 set the client will only send a DUID as client identifier.
2533 * The nss-systemd glibc NSS module will now enumerate dynamic users and
2534 groups in effect. Previously, it could resolve UIDs/GIDs to user
2535 names/groups and vice versa, but did not support enumeration.
2537 * journald's Compress= configuration setting now optionally accepts a
2538 byte threshold value. All journal objects larger than this threshold
2539 will be compressed, smaller ones will not. Previously this threshold
2540 was not configurable and set to 512.
2542 * A new system.conf setting NoNewPrivileges= is now available which may
2543 be used to turn off acquisition of new privileges system-wide
2544 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
2545 for all its children). Note that turning this option on means setuid
2546 binaries and file system capabilities lose their special powers.
2547 While turning on this option is a big step towards a more secure
2548 system, doing so is likely to break numerous pre-existing UNIX tools,
2549 in particular su and sudo.
2551 * A new service systemd-time-sync-wait.service has been added. If
2552 enabled it will delay the time-sync.target unit at boot until time
2553 synchronization has been received from the network. This
2554 functionality is useful on systems lacking a local RTC or where it is
2555 acceptable that the boot process shall be delayed by external network
2558 * When hibernating, systemd will now inform the kernel of the image
2559 write offset, on kernels new enough to support this. This means swap
2560 files should work for hibernation now.
2562 * When loading unit files, systemd will now look for drop-in unit files
2563 extensions in additional places. Previously, for a unit file name
2564 "foo-bar-baz.service" it would look for dropin files in
2565 "foo-bar-baz.service.d/*.conf". Now, it will also look in
2566 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
2567 service name truncated after all inner dashes. This scheme allows
2568 writing drop-ins easily that apply to a whole set of unit files at
2569 once. It's particularly useful for mount and slice units (as their
2570 naming is prefix based), but is also useful for service and other
2571 units, for packages that install multiple unit files at once,
2572 following a strict naming regime of beginning the unit file name with
2573 the package's name. Two new specifiers are now supported in unit
2574 files to match this: %j and %J are replaced by the part of the unit
2575 name following the last dash.
2577 * Unit files and other configuration files that support specifier
2578 expansion now understand another three new specifiers: %T and %V will
2579 resolve to /tmp and /var/tmp respectively, or whatever temporary
2580 directory has been set for the calling user. %E will expand to either
2581 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
2583 * The ExecStart= lines of unit files are no longer required to
2584 reference absolute paths. If non-absolute paths are specified the
2585 specified binary name is searched within the service manager's
2586 built-in $PATH, which may be queried with 'systemd-path
2587 search-binaries-default'. It's generally recommended to continue to
2588 use absolute paths for all binaries specified in unit files.
2590 * Units gained a new load state "bad-setting", which is used when a
2591 unit file was loaded, but contained fatal errors which prevent it
2592 from being started (for example, a service unit has been defined
2593 lacking both ExecStart= and ExecStop= lines).
2595 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
2596 support alternative debuggers, for example lldb. The old name
2597 continues to be available however, for compatibility reasons. Use the
2598 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
2599 to pick an alternative debugger instead of the default gdb.
2601 * systemctl and the other tools will now output escape sequences that
2602 generate proper clickable hyperlinks in various terminal emulators
2603 where useful (for example, in the "systemctl status" output you can
2604 now click on the unit file name to quickly open it in the
2605 editor/viewer of your choice). Note that not all terminal emulators
2606 support this functionality yet, but many do. Unfortunately, the
2607 "less" pager doesn't support this yet, hence this functionality is
2608 currently automatically turned off when a pager is started (which
2609 happens quite often due to auto-paging). We hope to remove this
2610 limitation as soon as "less" learns these escape sequences. This new
2611 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
2612 environment variable. For details on these escape sequences see:
2613 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
2615 * networkd's .network files now support a new IPv6MTUBytes= option for
2616 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
2617 option in the [Route] section to configure the MTU to use for
2618 specific routes. It also gained support for configuration of the DHCP
2619 "UserClass" option through the new UserClass= setting. It gained
2620 three new options in the new [CAN] section for configuring CAN
2621 networks. The MULTICAST and ALLMULTI interface flags may now be
2622 controlled explicitly with the new Multicast= and AllMulticast=
2625 * networkd will now automatically make use of the kernel's route
2626 expiration feature, if it is available.
2628 * udevd's .link files now support setting the number of receive and
2629 transmit channels, using the RxChannels=, TxChannels=,
2630 OtherChannels=, CombinedChannels= settings.
2632 * Support for UDPSegmentationOffload= has been removed, given its
2633 limited support in hardware, and waning software support.
2635 * networkd's .netdev files now support creating "netdevsim" interfaces.
2637 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
2638 to query the unit belonging to a specific kernel control group.
2640 * systemd-analyze gained a new verb "cat-config", which may be used to
2641 dump the contents of any configuration file, with all its matching
2642 drop-in files added in, and honouring the usual search and masking
2643 logic applied to systemd configuration files. For example use
2644 "systemd-analyze cat-config systemd/system.conf" to get the complete
2645 system configuration file of systemd how it would be loaded by PID 1
2646 itself. Similar to this, various tools such as systemd-tmpfiles or
2647 systemd-sysusers, gained a new option "--cat-config", which does the
2648 corresponding operation for their own configuration settings. For
2649 example, "systemd-tmpfiles --cat-config" will now output the full
2650 list of tmpfiles.d/ lines in place.
2652 * timedatectl gained three new verbs: "show" shows bus properties of
2653 systemd-timedated, "timesync-status" shows the current NTP
2654 synchronization state of systemd-timesyncd, and "show-timesync"
2655 shows bus properties of systemd-timesyncd.
2657 * systemd-timesyncd gained a bus interface on which it exposes details
2660 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
2661 understood by systemd-timedated. It takes a colon-separated list of
2662 unit names of NTP client services. The list is used by
2663 "timedatectl set-ntp".
2665 * systemd-nspawn gained a new --rlimit= switch for setting initial
2666 resource limits for the container payload. There's a new switch
2667 --hostname= to explicitly override the container's hostname. A new
2668 --no-new-privileges= switch may be used to control the
2669 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
2670 --oom-score-adjust= switch controls the OOM scoring adjustment value
2671 for the payload. The new --cpu-affinity= switch controls the CPU
2672 affinity of the container payload. The new --resolv-conf= switch
2673 allows more detailed control of /etc/resolv.conf handling of the
2674 container. Similarly, the new --timezone= switch allows more detailed
2675 control of /etc/localtime handling of the container.
2677 * systemd-detect-virt gained a new --list switch, which will print a
2678 list of all currently known VM and container environments.
2680 * Support for "Portable Services" has been added, see
2681 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
2682 experimental, but this is expected to change soon. Reflecting this
2683 experimental state, the "portablectl" binary is not installed into
2684 /usr/bin yet. The binary has to be called with the full path
2685 /usr/lib/systemd/portablectl instead.
2687 * journalctl's and systemctl's -o switch now knows a new log output
2688 mode "with-unit". The output it generates is very similar to the
2689 regular "short" mode, but displays the unit name instead of the
2690 syslog tag for each log line. Also, the date is shown with timezone
2691 information. This mode is probably more useful than the classic
2692 "short" output mode for most purposes, except where pixel-perfect
2693 compatibility with classic /var/log/messages formatting is required.
2695 * A new --dump-bus-properties switch has been added to the systemd
2696 binary, which may be used to dump all supported D-Bus properties.
2697 (Options which are still supported, but are deprecated, are *not*
2700 * sd-bus gained a set of new calls:
2701 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
2702 enable/disable the "floating" state of a bus slot object,
2703 i.e. whether the slot object pins the bus it is allocated for into
2704 memory or if the bus slot object gets disconnected when the bus goes
2705 away. sd_bus_open_with_description(),
2706 sd_bus_open_user_with_description(),
2707 sd_bus_open_system_with_description() may be used to allocate bus
2708 objects and set their description string already during allocation.
2710 * sd-event gained support for watching inotify events from the event
2711 loop, in an efficient way, sharing inotify handles between multiple
2712 users. For this a new function sd_event_add_inotify() has been added.
2714 * sd-event and sd-bus gained support for calling special user-supplied
2715 destructor functions for userdata pointers associated with
2716 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
2717 functions sd_bus_slot_set_destroy_callback,
2718 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
2719 sd_bus_track_get_destroy_callback,
2720 sd_event_source_set_destroy_callback,
2721 sd_event_source_get_destroy_callback have been added.
2723 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
2725 * PID 1 will now automatically reschedule .timer units whenever the
2726 local timezone changes. (They previously got rescheduled
2727 automatically when the system clock changed.)
2729 * New documentation has been added to document cgroups delegation,
2730 portable services and the various code quality tools we have set up:
2732 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
2733 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
2734 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
2736 * The Boot Loader Specification has been added to the source tree.
2738 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
2740 While moving it into our source tree we have updated it and further
2741 changes are now accepted through the usual github PR workflow.
2743 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
2744 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
2745 earlier PAM modules. The data in these fields is used to initialize
2746 the session scope's resource properties. Thus external PAM modules
2747 may now configure per-session limits, for example sourced from
2748 external user databases.
2750 * socket units with Accept=yes will now maintain a "refused" counter in
2751 addition to the existing "accepted" counter, counting connections
2752 refused due to the enforced limits.
2754 * The "systemd-path search-binaries-default" command may now be use to
2755 query the default, built-in $PATH PID 1 will pass to the services it
2758 * A new unit file setting PrivateMounts= has been added. It's a boolean
2759 option. If enabled the unit's processes are invoked in their own file
2760 system namespace. Note that this behaviour is also implied if any
2761 other file system namespacing options (such as PrivateTmp=,
2762 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
2763 primarily useful for services that do not use any of the other file
2764 system namespacing options. One such service is systemd-udevd.service
2765 where this is now used by default.
2767 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
2768 when the system is booted in UEFI "secure mode".
2770 * A new unit "system-update-pre.target" is added, which defines an
2771 optional synchronization point for offline system updates, as
2772 implemented by the pre-existing "system-update.target" unit. It
2773 allows ordering services before the service that executes the actual
2774 update process in a generic way.
2776 * Systemd now emits warnings whenever .include syntax is used.
2778 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
2779 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
2780 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
2781 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
2782 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
2783 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
2784 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
2785 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
2786 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
2787 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
2788 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
2789 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
2790 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
2791 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
2792 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
2793 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
2794 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
2795 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
2796 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
2797 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
2798 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
2799 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
2800 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
2801 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
2802 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
2803 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
2804 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
2805 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
2806 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2808 — Berlin, 2018-06-22
2812 * The MemoryAccounting= unit property now defaults to on. After
2813 discussions with the upstream control group maintainers we learnt
2814 that the negative impact of cgroup memory accounting on current
2815 kernels is finally relatively minimal, so that it should be safe to
2816 enable this by default without affecting system performance. Besides
2817 memory accounting only task accounting is turned on by default, all
2818 other forms of resource accounting (CPU, IO, IP) remain off for now,
2819 because it's not clear yet that their impact is small enough to move
2820 from opt-in to opt-out. We recommend downstreams to leave memory
2821 accounting on by default if kernel 4.14 or higher is primarily
2822 used. On very resource constrained systems or when support for old
2823 kernels is a necessity, -Dmemory-accounting-default=false can be used
2824 to revert this change.
2826 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
2827 %udev_rules_update) and the journal catalog (%journal_catalog_update)
2828 from the upgrade scriptlets of individual packages now do nothing.
2829 Transfiletriggers have been added which will perform those updates
2830 once at the end of the transaction.
2832 Similar transfiletriggers have been added to execute any sysctl.d
2833 and binfmt.d rules. Thus, it should be unnecessary to provide any
2834 scriptlets to execute this configuration from package installation
2837 * systemd-sysusers gained a mode where the configuration to execute is
2838 specified on the command line, but this configuration is not executed
2839 directly, but instead it is merged with the configuration on disk,
2840 and the result is executed. This is useful for package installation
2841 scripts which want to create the user before installing any files on
2842 disk (in case some of those files are owned by that user), while
2843 still allowing local admin overrides.
2845 This functionality is exposed to rpm scriptlets through a new
2846 %sysusers_create_package macro. Old %sysusers_create and
2847 %sysusers_create_inline macros are deprecated.
2849 A transfiletrigger for sysusers.d configuration is now installed,
2850 which means that it should be unnecessary to call systemd-sysusers from
2851 package installation scripts, unless the package installs any files
2852 owned by those newly-created users, in which case
2853 %sysusers_create_package should be used.
2855 * Analogous change has been done for systemd-tmpfiles: it gained a mode
2856 where the command-line configuration is merged with the configuration
2857 on disk. This is exposed as the new %tmpfiles_create_package macro,
2858 and %tmpfiles_create is deprecated. A transfiletrigger is installed
2859 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
2860 from package installation scripts.
2862 * sysusers.d configuration for a user may now also specify the group
2863 number, in addition to the user number ("u username 123:456"), or
2864 without the user number ("u username -:456").
2866 * Configution items for systemd-sysusers can now be specified as
2867 positional arguments when the new --inline switch is used.
2869 * The login shell of users created through sysusers.d may now be
2870 specified (previously, it was always /bin/sh for root and
2871 /sbin/nologin for other users).
2873 * systemd-analyze gained a new --global switch to look at global user
2874 configuration. It also gained a unit-paths verb to list the unit load
2875 paths that are compiled into systemd (which can be used with
2876 --systemd, --user, or --global).
2878 * udevadm trigger gained a new --settle/-w option to wait for any
2879 triggered events to finish (but just those, and not any other events
2880 which are triggered meanwhile).
2882 * The action that systemd-logind takes when the lid is closed and the
2883 machine is connected to external power can now be configured using
2884 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
2885 was determined by HandleLidSwitch=, and, for backwards compatibility,
2886 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
2888 * journalctl will periodically call sd_journal_process() to make it
2889 resilient against inotify queue overruns when journal files are
2890 rotated very quickly.
2892 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
2893 sd_bus_get_n_queued_write — may be used to check the number of
2894 pending bus messages.
2896 * systemd gained a new
2897 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
2898 which can be used to migrate foreign processes to scope and service
2899 units. The primary user for this new API is systemd itself: the
2900 systemd --user instance uses this call of the systemd --system
2901 instance to migrate processes if it itself gets the request to
2902 migrate processes and the kernel refuses this due to access
2903 restrictions. Thanks to this "systemd-run --scope --user …" works
2904 again in pure cgroup v2 environments when invoked from the user
2907 * A new TemporaryFileSystem= setting can be used to mask out part of
2908 the real file system tree with tmpfs mounts. This may be combined
2909 with BindPaths= and BindReadOnlyPaths= to hide files or directories
2910 not relevant to the unit, while still allowing some paths lower in
2911 the tree to be accessed.
2913 ProtectHome=tmpfs may now be used to hide user home and runtime
2914 directories from units, in a way that is mostly equivalent to
2915 "TemporaryFileSystem=/home /run/user /root".
2917 * Non-service units are now started with KeyringMode=shared by default.
2918 This means that mount and swapon and other mount tools have access
2919 to keys in the main keyring.
2921 * /sys/fs/bpf is now mounted automatically.
2923 * QNX virtualization is now detected by systemd-detect-virt and may
2924 be used in ConditionVirtualization=.
2926 * IPAccounting= may now be enabled also for slice units.
2928 * A new -Dsplit-bin= build configuration switch may be used to specify
2929 whether bin and sbin directories are merged, or if they should be
2930 included separately in $PATH and various listings of executable
2931 directories. The build configuration scripts will try to autodetect
2932 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
2933 system, but distributions are encouraged to configure this
2936 * A new -Dok-color= build configuration switch may be used to change
2937 the colour of "OK" status messages.
2939 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
2940 PrivateNetwork=yes was buggy in previous versions of systemd. This
2941 means that after the upgrade and daemon-reexec, any such units must
2944 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
2945 will not exclude read-only files owned by root from cleanup.
2947 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
2948 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
2949 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
2950 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
2951 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
2952 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
2953 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
2954 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
2955 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
2956 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
2957 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
2958 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
2959 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2960 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
2961 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
2962 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
2964 — Warsaw, 2018-03-05
2968 * Some keyboards come with a zoom see-saw or rocker which until now got
2969 mapped to the Linux "zoomin/out" keys in hwdb. However, these
2970 keycodes are not recognized by any major desktop. They now produce
2971 Up/Down key events so that they can be used for scrolling.
2973 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
2974 slightly: previously, if an argument was specified for lines of this
2975 type (i.e. the right-most column was set) this string was appended to
2976 existing files each time systemd-tmpfiles was run. This behaviour was
2977 different from what the documentation said, and not particularly
2978 useful, as repeated systemd-tmpfiles invocations would not be
2979 idempotent and grow such files without bounds. With this release
2980 behaviour has been altered to match what the documentation says:
2981 lines of this type only have an effect if the indicated files don't
2982 exist yet, and only then the argument string is written to the file.
2984 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
2985 systemd-tmpfiles behaviour: previously, read-only files owned by root
2986 were always excluded from the file "aging" algorithm (i.e. the
2987 automatic clean-up of directories like /tmp based on
2988 atime/mtime/ctime). We intend to drop this restriction, and age files
2989 by default even when owned by root and read-only. This behaviour was
2990 inherited from older tools, but there have been requests to remove
2991 it, and it's not obvious why this restriction was made in the first
2992 place. Please speak up now, if you are aware of software that reqires
2993 this behaviour, otherwise we'll remove the restriction in v238.
2995 * A new environment variable $SYSTEMD_OFFLINE is now understood by
2996 systemctl. It takes a boolean argument. If on, systemctl assumes it
2997 operates on an "offline" OS tree, and will not attempt to talk to the
2998 service manager. Previously, this mode was implicitly enabled if a
2999 chroot() environment was detected, and this new environment variable
3000 now provides explicit control.
3002 * .path and .socket units may now be created transiently, too.
3003 Previously only service, mount, automount and timer units were
3004 supported as transient units. The systemd-run tool has been updated
3005 to expose this new functionality, you may hence use it now to bind
3006 arbitrary commands to path or socket activation on-the-fly from the
3007 command line. Moreover, almost all properties are now exposed for the
3008 unit types that already supported transient operation.
3010 * The systemd-mount command gained support for a new --owner= parameter
3011 which takes a user name, which is then resolved and included in uid=
3012 and gid= mount options string of the file system to mount.
3014 * A new unit condition ConditionControlGroupController= has been added
3015 that checks whether a specific cgroup controller is available.
3017 * Unit files, udev's .link files, and systemd-networkd's .netdev and
3018 .network files all gained support for a new condition
3019 ConditionKernelVersion= for checking against specific kernel
3022 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
3023 support for configuring device flags in the Flags= setting. In the
3024 same files, the [Tunnel] section gained support for configuring
3025 AllowLocalRemote=. The [Route] section in .network files gained
3026 support for configuring InitialCongestionWindow=,
3027 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
3028 understands RapidCommit=.
3030 * systemd-networkd's DHCPv6 support gained support for Prefix
3033 * sd-bus gained support for a new "watch-bind" feature. When this
3034 feature is enabled, an sd_bus connection may be set up to connect to
3035 an AF_UNIX socket in the file system as soon as it is created. This
3036 functionality is useful for writing early-boot services that
3037 automatically connect to the system bus as soon as it is started,
3038 without ugly time-based polling. systemd-networkd and
3039 systemd-resolved have been updated to make use of this
3040 functionality. busctl exposes this functionality in a new
3041 --watch-bind= command line switch.
3043 * sd-bus will now optionally synthesize a local "Connected" signal as
3044 soon as a D-Bus connection is set up fully. This message mirrors the
3045 already existing "Disconnected" signal which is synthesized when the
3046 connection is terminated. This signal is generally useful but
3047 particularly handy in combination with the "watch-bind" feature
3048 described above. Synthesizing of this message has to be requested
3049 explicitly through the new API call sd_bus_set_connected_signal(). In
3050 addition a new call sd_bus_is_ready() has been added that checks
3051 whether a connection is fully set up (i.e. between the "Connected" and
3052 "Disconnected" signals).
3054 * sd-bus gained two new calls sd_bus_request_name_async() and
3055 sd_bus_release_name_async() for asynchronously registering bus
3056 names. Similar, there is now sd_bus_add_match_async() for installing
3057 a signal match asynchronously. All of systemd's own services have
3058 been updated to make use of these calls. Doing these operations
3059 asynchronously has two benefits: it reduces the risk of deadlocks in
3060 case of cyclic dependencies between bus services, and it speeds up
3061 service initialization since synchronization points for bus
3062 round-trips are removed.
3064 * sd-bus gained two new calls sd_bus_match_signal() and
3065 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
3066 and sd_bus_add_match_async() but instead of taking a D-Bus match
3067 string take match fields as normal function parameters.
3069 * sd-bus gained two new calls sd_bus_set_sender() and
3070 sd_bus_message_set_sender() for setting the sender name of outgoing
3071 messages (either for all outgoing messages or for just one specific
3072 one). These calls are only useful in direct connections as on
3073 brokered connections the broker fills in the sender anyway,
3074 overwriting whatever the client filled in.
3076 * sd-event gained a new pseudo-handle that may be specified on all API
3077 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
3078 used this refers to the default event loop object of the calling
3079 thread. Note however that this does not implicitly allocate one —
3080 which has to be done prior by using sd_event_default(). Similarly
3081 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
3082 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
3083 to the default bus of the specified type of the calling thread. Here
3084 too this does not implicitly allocate bus connection objects, this
3085 has to be done prior with sd_bus_default() and friends.
3087 * sd-event gained a new call pair
3088 sd_event_source_{get|set}_io_fd_own(). This may be used to request
3089 automatic closure of the file descriptor an IO event source watches
3090 when the event source is destroyed.
3092 * systemd-networkd gained support for natively configuring WireGuard
3095 * In previous versions systemd synthesized user records both for the
3096 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
3097 internally. In order to simplify distribution-wide renames of the
3098 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
3099 new transitional flag file has been added: if
3100 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
3101 user and group record within the systemd codebase is disabled.
3103 * systemd-notify gained a new --uid= option for selecting the source
3104 user/UID to use for notification messages sent to the service
3107 * journalctl gained a new --grep= option to list only entries in which
3108 the message matches a certain pattern. By default matching is case
3109 insensitive if the pattern is lowercase, and case sensitive
3110 otherwise. Option --case-sensitive=yes|no can be used to override
3111 this an specify case sensitivity or case insensitivity.
3113 * There's now a "systemd-analyze service-watchdogs" command for printing
3114 the current state of the service runtime watchdog, and optionally
3115 enabling or disabling the per-service watchdogs system-wide if given a
3116 boolean argument (i.e. the concept you configure in WatchdogSec=), for
3117 debugging purposes. There's also a kernel command line option
3118 systemd.service_watchdogs= for controlling the same.
3120 * Two new "log-level" and "log-target" options for systemd-analyze were
3121 added that merge the now deprecated get-log-level, set-log-level and
3122 get-log-target, set-log-target pairs. The deprecated options are still
3123 understood for backwards compatibility. The two new options print the
3124 current value when no arguments are given, and set them when a
3125 level/target is given as an argument.
3127 * sysusers.d's "u" lines now optionally accept both a UID and a GID
3128 specification, separated by a ":" character, in order to create users
3129 where UID and GID do not match.
3131 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
3132 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
3133 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
3134 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
3135 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
3136 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
3137 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
3138 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
3139 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
3140 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
3141 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
3142 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
3143 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
3144 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
3145 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
3146 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
3147 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
3148 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
3149 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
3150 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
3157 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
3158 in v235 has been extended to also set the dummy.ko module option
3159 numdummies=0, preventing the kernel from automatically creating
3160 dummy0. All dummy interfaces must now be explicitly created.
3162 * Unknown '%' specifiers in configuration files are now rejected. This
3163 applies to units and tmpfiles.d configuration. Any percent characters
3164 that are followed by a letter or digit that are not supposed to be
3165 interpreted as the beginning of a specifier should be escaped by
3166 doubling ("%%"). (So "size=5%" is still accepted, as well as
3167 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
3168 valid specifiers today.)
3170 * systemd-resolved now maintains a new dynamic
3171 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
3172 recommended to make /etc/resolv.conf a symlink to it. This file
3173 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
3174 includes dynamically acquired search domains, achieving more correct
3175 DNS resolution by software that bypasses local DNS APIs such as NSS.
3177 * The "uaccess" udev tag has been dropped from /dev/kvm and
3178 /dev/dri/renderD*. These devices now have the 0666 permissions by
3179 default (but this may be changed at build-time). /dev/dri/renderD*
3180 will now be owned by the "render" group along with /dev/kfd.
3182 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
3183 systemd-journal-gatewayd.service and
3184 systemd-journal-upload.service. This means "nss-systemd" must be
3185 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
3186 services are resolved properly.
3188 * In /etc/fstab two new mount options are now understood:
3189 x-systemd.makefs and x-systemd.growfs. The former has the effect that
3190 the configured file system is formatted before it is mounted, the
3191 latter that the file system is resized to the full block device size
3192 after it is mounted (i.e. if the file system is smaller than the
3193 partition it resides on, it's grown). This is similar to the fsck
3194 logic in /etc/fstab, and pulls in systemd-makefs@.service and
3195 systemd-growfs@.service as necessary, similar to
3196 systemd-fsck@.service. Resizing is currently only supported on ext4
3199 * In systemd-networkd, the IPv6 RA logic now optionally may announce
3200 DNS server and domain information.
3202 * Support for the LUKS2 on-disk format for encrypted partitions has
3203 been added. This requires libcryptsetup2 during compilation and
3206 * The systemd --user instance will now signal "readiness" when its
3207 basic.target unit has been reached, instead of when the run queue ran
3208 empty for the first time.
3210 * Tmpfiles.d with user configuration are now also supported.
3211 systemd-tmpfiles gained a new --user switch, and snippets placed in
3212 ~/.config/user-tmpfiles.d/ and corresponding directories will be
3213 executed by systemd-tmpfiles --user running in the new
3214 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
3215 running in the user session.
3217 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
3218 %S resolves to the top-level state directory (/var/lib for the system
3219 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
3220 top-level cache directory (/var/cache for the system instance,
3221 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
3222 logs directory (/var/log for the system instance,
3223 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
3224 existing %t specifier, that resolves to the top-level runtime
3225 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
3228 * journalctl learnt a new parameter --output-fields= for limiting the
3229 set of journal fields to output in verbose and JSON output modes.
3231 * systemd-timesyncd's configuration file gained a new option
3232 RootDistanceMaxSec= for setting the maximum root distance of servers
3233 it'll use, as well as the new options PollIntervalMinSec= and
3234 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
3236 * bootctl gained a new command "list" for listing all available boot
3237 menu items on systems that follow the boot loader specification.
3239 * systemctl gained a new --dry-run switch that shows what would be done
3240 instead of doing it, and is currently supported by the shutdown and
3243 * ConditionSecurity= can now detect the TOMOYO security module.
3245 * Unit file [Install] sections are now also respected in unit drop-in
3246 files. This is intended to be used by drop-ins under /usr/lib/.
3248 * systemd-firstboot may now also set the initial keyboard mapping.
3250 * Udev "changed" events for devices which are exposed as systemd
3251 .device units are now propagated to units specified in
3252 ReloadPropagatedFrom= as reload requests.
3254 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
3255 unit template name (i.e. a name in the form of 'foobar@.service',
3256 without the instance component between the '@' and - the '.'), then
3257 the escaped sysfs path of the device is automatically used as the
3260 * SystemCallFilter= in unit files has been extended so that an "errno"
3261 can be specified individually for each system call. Example:
3262 SystemCallFilter=~uname:EILSEQ.
3264 * The cgroup delegation logic has been substantially updated. Delegate=
3265 now optionally takes a list of controllers (instead of a boolean, as
3266 before), which lists the controllers to delegate at least.
3268 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
3270 * A new LogLevelMax= setting configures the maximum log level any
3271 process of the service may log at (i.e. anything with a lesser
3272 priority than what is specified is automatically dropped). A new
3273 LogExtraFields= setting allows configuration of additional journal
3274 fields to attach to all log records generated by any of the unit's
3277 * New StandardInputData= and StandardInputText= settings along with the
3278 new option StandardInput=data may be used to configure textual or
3279 binary data that shall be passed to the executed service process via
3280 standard input, encoded in-line in the unit file.
3282 * StandardInput=, StandardOutput= and StandardError= may now be used to
3283 connect stdin/stdout/stderr of executed processes directly with a
3284 file or AF_UNIX socket in the file system, using the new "file:" option.
3286 * A new unit file option CollectMode= has been added, that allows
3287 tweaking the garbage collection logic for units. It may be used to
3288 tell systemd to garbage collect units that have failed automatically
3289 (normally it only GCs units that exited successfully). systemd-run
3290 and systemd-mount expose this new functionality with a new -G option.
3292 * "machinectl bind" may now be used to bind mount non-directories
3293 (i.e. regularfiles, devices, fifos, sockets).
3295 * systemd-analyze gained a new verb "calendar" for validating and
3296 testing calendar time specifications to use for OnCalendar= in timer
3297 units. Besides validating the expression it will calculate the next
3298 time the specified expression would elapse.
3300 * In addition to the pre-existing FailureAction= unit file setting
3301 there's now SuccessAction=, for configuring a shutdown action to
3302 execute when a unit completes successfully. This is useful in
3303 particular inside containers that shall terminate after some workload
3304 has been completed. Also, both options are now supported for all unit
3305 types, not just services.
3307 * networkds's IP rule support gained two new options
3308 IncomingInterface= and OutgoingInterface= for configuring the incoming
3309 and outgoing interfaces of configured rules. systemd-networkd also
3310 gained support for "vxcan" network devices.
3312 * networkd gained a new setting RequiredForOnline=, taking a
3313 boolean. If set, systemd-wait-online will take it into consideration
3314 when determining that the system is up, otherwise it will ignore the
3315 interface for this purpose.
3317 * The sd_notify() protocol gained support for a new operation: with
3318 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
3319 store again, ahead of POLLHUP or POLLERR when they are removed
3322 * A new document doc/UIDS-GIDS.md has been added to the source tree,
3323 that documents the UID/GID range and assignment assumptions and
3324 requirements of systemd.
3326 * The watchdog device PID 1 will ping may now be configured through the
3327 WatchdogDevice= configuration file setting, or by setting the
3328 systemd.watchdog_service= kernel commandline option.
3330 * systemd-resolved's gained support for registering DNS-SD services on
3331 the local network using MulticastDNS. Services may either be
3332 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
3333 the same dir below /run, /usr/lib), or through its D-Bus API.
3335 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
3336 extend the effective start, runtime, and stop time. The service must
3337 continue to send EXTEND_TIMEOUT_USEC within the period specified to
3338 prevent the service manager from making the service as timedout.
3340 * systemd-resolved's DNSSEC support gained support for RFC 8080
3341 (Ed25519 keys and signatures).
3343 * The systemd-resolve command line tool gained a new set of options
3344 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
3345 --set-nta= and --revert to configure per-interface DNS configuration
3346 dynamically during runtime. It's useful for pushing DNS information
3347 into systemd-resolved from DNS hook scripts that various interface
3348 managing software supports (such as pppd).
3350 * systemd-nspawn gained a new --network-namespace-path= command line
3351 option, which may be used to make a container join an existing
3352 network namespace, by specifying a path to a "netns" file.
3354 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
3355 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
3356 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
3357 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
3358 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
3359 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
3360 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
3361 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
3362 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
3363 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
3364 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
3365 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
3366 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
3367 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
3368 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
3369 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
3370 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
3371 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
3372 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
3373 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
3374 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
3375 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
3376 Jędrzejewski-Szmek, Zeal Jagannatha
3378 — Berlin, 2017-12-14
3382 * INCOMPATIBILITY: systemd-logind.service and other long-running
3383 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
3384 communication with the outside. This generally improves security of
3385 the system, and is in almost all cases a safe and good choice, as
3386 these services do not and should not provide any network-facing
3387 functionality. However, systemd-logind uses the glibc NSS API to
3388 query the user database. This creates problems on systems where NSS
3389 is set up to directly consult network services for user database
3390 lookups. In particular, this creates incompatibilities with the
3391 "nss-nis" module, which attempts to directly contact the NIS/YP
3392 network servers it is configured for, and will now consistently
3393 fail. In such cases, it is possible to turn off IP sandboxing for
3394 systemd-logind.service (set IPAddressDeny= in its [Service] section
3395 to the empty string, via a .d/ unit file drop-in). Downstream
3396 distributions might want to update their nss-nis packaging to include
3397 such a drop-in snippet, accordingly, to hide this incompatibility
3398 from the user. Another option is to make use of glibc's nscd service
3399 to proxy such network requests through a privilege-separated, minimal
3400 local caching daemon, or to switch to more modern technologies such
3401 sssd, whose NSS hook-ups generally do not involve direct network
3402 access. In general, we think it's definitely time to question the
3403 implementation choices of nss-nis, i.e. whether it's a good idea
3404 today to embed a network-facing loadable module into all local
3405 processes that need to query the user database, including the most
3406 trivial and benign ones, such as "ls". For more details about
3407 IPAddressDeny= see below.
3409 * A new modprobe.d drop-in is now shipped by default that sets the
3410 bonding module option max_bonds=0. This overrides the kernel default,
3411 to avoid conflicts and ambiguity as to whether or not bond0 should be
3412 managed by systemd-networkd or not. This resolves multiple issues
3413 with bond0 properties not being applied, when bond0 is configured
3414 with systemd-networkd. Distributors may choose to not package this,
3415 however in that case users will be prevented from correctly managing
3416 bond0 interface using systemd-networkd.
3418 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
3419 which print the logging level and target of the system manager. They
3420 complement the existing "set-log-level" and "set-log-target" verbs
3421 used to change those values.
3423 * journald.conf gained a new boolean setting ReadKMsg= which defaults
3424 to on. If turned off kernel log messages will not be read by
3425 systemd-journald or included in the logs. It also gained a new
3426 setting LineMax= for configuring the maximum line length in
3427 STDOUT/STDERR log streams. The new default for this value is 48K, up
3428 from the previous hardcoded 2048.
3430 * A new unit setting RuntimeDirectoryPreserve= has been added, which
3431 allows more detailed control of what to do with a runtime directory
3432 configured with RuntimeDirectory= (i.e. a directory below /run or
3433 $XDG_RUNTIME_DIR) after a unit is stopped.
3435 * The RuntimeDirectory= setting for units gained support for creating
3436 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
3437 one top-level directory.
3439 * Units gained new options StateDirectory=, CacheDirectory=,
3440 LogsDirectory= and ConfigurationDirectory= which are closely related
3441 to RuntimeDirectory= but manage per-service directories below
3442 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
3443 possible to write unit files which when activated automatically gain
3444 properly owned service specific directories in these locations, thus
3445 making unit files self-contained and increasing compatibility with
3446 stateless systems and factory reset where /etc or /var are
3447 unpopulated at boot. Matching these new settings there's also
3448 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
3449 ConfigurationDirectoryMode= for configuring the access mode of these
3450 directories. These settings are particularly useful in combination
3451 with DynamicUser=yes as they provide secure, properly-owned,
3452 writable, and stateful locations for storage, excluded from the
3453 sandbox that such services live in otherwise.
3455 * Automake support has been removed from this release. systemd is now
3458 * systemd-journald will now aggressively cache client metadata during
3459 runtime, speeding up log write performance under pressure. This comes
3460 at a small price though: as much of the metadata is read
3461 asynchronously from /proc/ (and isn't implicitly attached to log
3462 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
3463 metadata stored alongside a log entry might be slightly
3464 out-of-date. Previously it could only be slightly newer than the log
3465 message. The time window is small however, and given that the kernel
3466 is unlikely to be improved anytime soon in this regard, this appears
3469 * nss-myhostname/systemd-resolved will now by default synthesize an
3470 A/AAAA resource record for the "_gateway" hostname, pointing to the
3471 current default IP gateway. Previously it did that for the "gateway"
3472 name, hampering adoption, as some distributions wanted to leave that
3473 hostname open for local use. The old behaviour may still be
3474 requested at build time.
3476 * systemd-networkd's [Address] section in .network files gained a new
3477 Scope= setting for configuring the IP address scope. The [Network]
3478 section gained a new boolean setting ConfigureWithoutCarrier= that
3479 tells systemd-networkd to ignore link sensing when configuring the
3480 device. The [DHCP] section gained a new Anonymize= boolean option for
3481 turning on a number of options suggested in RFC 7844. A new
3482 [RoutingPolicyRule] section has been added for configuring the IP
3483 routing policy. The [Route] section has gained support for a new
3484 Type= setting which permits configuring
3485 blackhole/unreachable/prohibit routes.
3487 * The [VRF] section in .netdev files gained a new Table= setting for
3488 configuring the routing table to use. The [Tunnel] section gained a
3489 new Independent= boolean field for configuring tunnels independent of
3490 an underlying network interface. The [Bridge] section gained a new
3491 GroupForwardMask= option for configuration of propagation of link
3492 local frames between bridge ports.
3494 * The WakeOnLan= setting in .link files gained support for a number of
3495 new modes. A new TCP6SegmentationOffload= setting has been added for
3496 configuring TCP/IPv6 hardware segmentation offload.
3498 * The IPv6 RA sender implementation may now optionally send out RDNSS
3499 and RDNSSL records to supply DNS configuration to peers.
3501 * systemd-nspawn gained support for a new --system-call-filter= command
3502 line option for adding and removing entries in the default system
3503 call filter it applies. Moreover systemd-nspawn has been changed to
3504 implement a system call allow list instead of a deny list.
3506 * systemd-run gained support for a new --pipe command line option. If
3507 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
3508 are directly passed on to the activated transient service
3509 executable. This allows invoking arbitrary processes as systemd
3510 services (for example to take benefit of dependency management,
3511 accounting management, resource management or log management that is
3512 done automatically for services) — while still allowing them to be
3513 integrated in a classic UNIX shell pipeline.
3515 * When a service sends RELOAD=1 via sd_notify() and reload propagation
3516 using ReloadPropagationTo= is configured, a reload is now propagated
3517 to configured units. (Previously this was only done on explicitly
3518 requested reloads, using "systemctl reload" or an equivalent
3521 * For each service unit a restart counter is now kept: it is increased
3522 each time the service is restarted due to Restart=, and may be
3523 queried using "systemctl show -p NRestarts …".
3525 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
3526 @signal and @timer have been added, for usage with SystemCallFilter=
3527 in unit files and the new --system-call-filter= command line option
3528 of systemd-nspawn (see above).
3530 * ExecStart= lines in unit files gained two new modifiers: when a
3531 command line is prefixed with "!" the command will be executed as
3532 configured, except for the credentials applied by
3533 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
3534 "+", but does still apply namespacing options unlike "+". There's
3535 also "!!" now, which is mostly identical, but becomes a NOP on
3536 systems that support ambient capabilities. This is useful to write
3537 unit files that work with ambient capabilities where possible but
3538 automatically fall back to traditional privilege dropping mechanisms
3539 on systems where this is not supported.
3541 * ListenNetlink= settings in socket units now support RDMA netlink
3544 * A new unit file setting LockPersonality= has been added which permits
3545 locking down the chosen execution domain ("personality") of a service
3548 * A new special target "getty-pre.target" has been added, which is
3549 ordered before all text logins, and may be used to order services
3550 before textual logins acquire access to the console.
3552 * systemd will now attempt to load the virtio-rng.ko kernel module very
3553 early on if a VM environment supporting this is detected. This should
3554 improve entropy during early boot in virtualized environments.
3556 * A _netdev option is now supported in /etc/crypttab that operates in a
3557 similar way as the same option in /etc/fstab: it permits configuring
3558 encrypted devices that need to be ordered after the network is up.
3559 Following this logic, two new special targets
3560 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
3561 added that are to cryptsetup.target what remote-fs.target and
3562 remote-fs-pre.target are to local-fs.target.
3564 * Service units gained a new UnsetEnvironment= setting which permits
3565 unsetting specific environment variables for services that are
3566 normally passed to it (for example in order to mask out locale
3567 settings for specific services that can't deal with it).
3569 * Units acquired a new boolean option IPAccounting=. When turned on, IP
3570 traffic accounting (packet count as well as byte count) is done for
3571 the service, and shown as part of "systemctl status" or "systemd-run
3574 * Service units acquired two new options IPAddressAllow= and
3575 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
3576 for configuring a simple IP access control list for all sockets of
3577 the unit. These options are available also on .slice and .socket
3578 units, permitting flexible access list configuration for individual
3579 services as well as groups of services (as defined by a slice unit),
3580 including system-wide. Note that IP ACLs configured this way are
3581 enforced on every single IPv4 and IPv6 socket created by any process
3582 of the service unit, and apply to ingress as well as egress traffic.
3584 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
3585 structured log message is generated each time the unit is stopped,
3586 containing information about the consumed resources of this
3589 * A new setting KeyringMode= has been added to unit files, which may be
3590 used to control how the kernel keyring is set up for executed
3593 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
3594 "systemctl kexec" and "systemctl exit" are now always asynchronous in
3595 behaviour (that is: these commands return immediately after the
3596 operation was enqueued instead of waiting for the operation to
3597 complete). Previously, "systemctl poweroff" and "systemctl reboot"
3598 were asynchronous on systems using systemd-logind (i.e. almost
3599 always, and like they were on sysvinit), and the other three commands
3600 were unconditionally synchronous. With this release this is cleaned
3601 up, and callers will see the same asynchronous behaviour on all
3602 systems for all five operations.
3604 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
3607 * .timer units now accept calendar specifications in other timezones
3608 than UTC or the local timezone.
3610 * The tmpfiles snippet var.conf has been changed to create
3611 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
3612 the "utmp" group already, and it appears to be generally understood
3613 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
3614 databases. Previously this was implemented correctly for all these
3615 databases excepts btmp, which has been opened up like this now
3616 too. Note that while the other databases are world-readable
3617 (i.e. 0644), btmp is not and remains more restrictive.
3619 * The systemd-resolve tool gained a new --reset-server-features
3620 switch. When invoked like this systemd-resolved will forget
3621 everything it learnt about the features supported by the configured
3622 upstream DNS servers, and restarts the feature probing logic on the
3623 next resolver look-up for them at the highest feature level
3626 * The status dump systemd-resolved sends to the logs upon receiving
3627 SIGUSR1 now also includes information about all DNS servers it is
3628 configured to use, and the features levels it probed for them.
3630 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
3631 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
3632 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
3633 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
3634 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
3635 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
3636 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
3637 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
3638 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
3639 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
3640 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
3641 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
3642 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
3643 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
3644 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
3645 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
3646 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
3647 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
3648 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
3649 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
3651 — Berlin, 2017-10-06
3655 * Meson is now supported as build system in addition to Automake. It is
3656 our plan to remove Automake in one of our next releases, so that
3657 Meson becomes our exclusive build system. Hence, please start using
3658 the Meson build system in your downstream packaging. There's plenty
3659 of documentation around how to use Meson, the extremely brief
3662 ./autogen.sh && ./configure && make && sudo make install
3666 meson build && ninja -C build && sudo ninja -C build install
3668 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
3669 which permits configuring a timeout on the time a job is
3670 running. This is particularly useful for setting timeouts on jobs for
3673 * Unit files gained two new options ConditionUser= and ConditionGroup=
3674 for conditionalizing units based on the identity of the user/group
3675 running a systemd user instance.
3677 * systemd-networkd now understands a new FlowLabel= setting in the
3678 [VXLAN] section of .network files, as well as a Priority= in
3679 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
3680 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
3681 gained support for configuration of GENEVE links, and IPv6 address
3682 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
3684 * .link files now understand a new Port= setting.
3686 * systemd-networkd's DHCP support gained support for DHCP option 119
3687 (domain search list).
3689 * systemd-networkd gained support for serving IPv6 address ranges using
3690 the Router Advertisement protocol. The new .network configuration
3691 section [IPv6Prefix] may be used to configure the ranges to
3692 serve. This is implemented based on a new, minimal, native server
3693 implementation of RA.
3695 * journalctl's --output= switch gained support for a new parameter
3696 "short-iso-precise" for a mode where timestamps are shown as precise
3699 * systemd-udevd's "net_id" builtin may now generate stable network
3700 interface names from IBM PowerVM VIO devices as well as ACPI platform
3703 * MulticastDNS support in systemd-resolved may now be explicitly
3704 enabled/disabled using the new MulticastDNS= configuration file
3707 * systemd-resolved may now optionally use libidn2 instead of the libidn
3708 for processing internationalized domain names. Support for libidn2
3709 should be considered experimental and should not be enabled by
3712 * "machinectl pull-tar" and related call may now do verification of
3713 downloaded images using SUSE-style .sha256 checksum files in addition
3714 to the already existing support for validating using Ubuntu-style
3717 * sd-bus gained support for a new sd_bus_message_appendv() call which
3718 is va_list equivalent of sd_bus_message_append().
3720 * sd-boot gained support for validating images using SHIM/MOK.
3722 * The SMACK code learnt support for "onlycap".
3724 * systemd-mount --umount is now much smarter in figuring out how to
3725 properly unmount a device given its mount or device path.
3727 * The code to call libnss_dns as a fallback from libnss_resolve when
3728 the communication with systemd-resolved fails was removed. This
3729 fallback was redundant and interfered with the [!UNAVAIL=return]
3730 suffix. See nss-resolve(8) for the recommended configuration.
3732 * systemd-logind may now be restarted without losing state. It stores
3733 the file descriptors for devices it manages in the system manager
3734 using the FDSTORE= mechanism. Please note that further changes in
3735 other components may be required to make use of this (for example
3736 Xorg has code to listen for stops of systemd-logind and terminate
3737 itself when logind is stopped or restarted, in order to avoid using
3738 stale file descriptors for graphical devices, which is now
3739 counterproductive and must be reverted in order for restarts of
3740 systemd-logind to be safe. See
3741 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
3743 * All kernel-install plugins are called with the environment variable
3744 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
3745 /etc/machine-id. If the machine ID could not be determined,
3746 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
3747 anything in the entry directory (passed as the second argument) if
3748 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
3749 temporary directory is passed as the entry directory and removed
3750 after all the plugins exit.
3752 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
3753 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
3754 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
3755 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
3756 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
3757 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
3758 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
3759 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
3760 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
3761 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
3762 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
3763 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
3764 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
3765 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
3766 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
3767 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
3768 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
3769 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
3770 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
3771 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
3772 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
3773 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
3774 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
3775 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
3776 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
3777 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
3778 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
3779 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
3780 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
3783 — Berlin, 2017-07-12
3787 * The "hybrid" control group mode has been modified to improve
3788 compatibility with "legacy" cgroups-v1 setups. Specifically, the
3789 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
3790 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
3791 cgroups-v1 hierarchy), the only externally visible change being that
3792 the cgroups-v2 hierarchy is also mounted, to
3793 /sys/fs/cgroup/unified. This should provide a large degree of
3794 compatibility with "legacy" cgroups-v1, while taking benefit of the
3795 better management capabilities of cgroups-v2.
3797 * The default control group setup mode may be selected both a boot-time
3798 via a set of kernel command line parameters (specifically:
3799 systemd.unified_cgroup_hierarchy= and
3800 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
3801 default selected on the configure command line
3802 (--with-default-hierarchy=). The upstream default is "hybrid"
3803 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
3804 this will change in a future systemd version to be "unified" (pure
3805 cgroups-v2 mode). The third option for the compile time option is
3806 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
3807 distributions to default to "hybrid" mode for release distributions,
3808 starting with v233. We recommend "unified" for development
3809 distributions (specifically: distributions such as Fedora's rawhide)
3810 as that's where things are headed in the long run. Use "legacy" for
3811 greatest stability and compatibility only.
3813 * Note one current limitation of "unified" and "hybrid" control group
3814 setup modes: the kernel currently does not permit the systemd --user
3815 instance (i.e. unprivileged code) to migrate processes between two
3816 disconnected cgroup subtrees, even if both are managed and owned by
3817 the user. This effectively means "systemd-run --user --scope" doesn't
3818 work when invoked from outside of any "systemd --user" service or
3819 scope. Specifically, it is not supported from session scopes. We are
3820 working on fixing this in a future systemd version. (See #3388 for
3821 further details about this.)
3823 * DBus policy files are now installed into /usr rather than /etc. Make
3824 sure your system has dbus >= 1.9.18 running before upgrading to this
3825 version, or override the install path with --with-dbuspolicydir= .
3827 * All python scripts shipped with systemd (specifically: the various
3828 tests written in Python) now require Python 3.
3830 * systemd unit tests can now run standalone (without the source or
3831 build directories), and can be installed into /usr/lib/systemd/tests/
3832 with 'make install-tests'.
3834 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
3835 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
3838 * Support for the %c, %r, %R specifiers in unit files has been
3839 removed. Specifiers are not supposed to be dependent on configuration
3840 in the unit file itself (so that they resolve the same regardless
3841 where used in the unit files), but these specifiers were influenced
3842 by the Slice= option.
3844 * The shell invoked by debug-shell.service now defaults to /bin/sh in
3845 all cases. If distributions want to use a different shell for this
3846 purpose (for example Fedora's /sbin/sushell) they need to specify
3847 this explicitly at configure time using --with-debug-shell=.
3849 * The confirmation spawn prompt has been reworked to offer the
3852 (c)ontinue, proceed without asking anymore
3853 (D)ump, show the state of the unit
3854 (f)ail, don't execute the command and pretend it failed
3856 (i)nfo, show a short summary of the unit
3857 (j)obs, show jobs that are in progress
3858 (s)kip, don't execute the command and pretend it succeeded
3859 (y)es, execute the command
3861 The 'n' choice for the confirmation spawn prompt has been removed,
3862 because its meaning was confusing.
3864 The prompt may now also be redirected to an alternative console by
3865 specifying the console as parameter to systemd.confirm_spawn=.
3867 * Services of Type=notify require a READY=1 notification to be sent
3868 during startup. If no such message is sent, the service now fails,
3869 even if the main process exited with a successful exit code.
3871 * Services that fail to start up correctly now always have their
3872 ExecStopPost= commands executed. Previously, they'd enter "failed"
3873 state directly, without executing these commands.
3875 * The option MulticastDNS= of network configuration files has acquired
3876 an actual implementation. With MulticastDNS=yes a host can resolve
3877 names of remote hosts and reply to mDNS A and AAAA requests.
3879 * When units are about to be started an additional check is now done to
3880 ensure that all dependencies of type BindsTo= (when used in
3881 combination with After=) have been started.
3883 * systemd-analyze gained a new verb "syscall-filter" which shows which
3884 system call groups are defined for the SystemCallFilter= unit file
3885 setting, and which system calls they contain.
3887 * A new system call filter group "@filesystem" has been added,
3888 consisting of various file system related system calls. Group
3889 "@reboot" has been added, covering reboot, kexec and shutdown related
3890 calls. Finally, group "@swap" has been added covering swap
3891 configuration related calls.
3893 * A new unit file option RestrictNamespaces= has been added that may be
3894 used to restrict access to the various process namespace types the
3895 Linux kernel provides. Specifically, it may be used to take away the
3896 right for a service unit to create additional file system, network,
3897 user, and other namespaces. This sandboxing option is particularly
3898 relevant due to the high amount of recently discovered namespacing
3899 related vulnerabilities in the kernel.
3901 * systemd-udev's .link files gained support for a new AutoNegotiation=
3902 setting for configuring Ethernet auto-negotiation.
3904 * systemd-networkd's .network files gained support for a new
3905 ListenPort= setting in the [DHCP] section to explicitly configure the
3906 UDP client port the DHCP client shall listen on.
3908 * .network files gained a new Unmanaged= boolean setting for explicitly
3909 excluding one or more interfaces from management by systemd-networkd.
3911 * The systemd-networkd ProxyARP= option has been renamed to
3912 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
3913 renamed to ReduceARPProxy=. The old names continue to be available
3916 * systemd-networkd gained support for configuring IPv6 Proxy NDP
3917 addresses via the new IPv6ProxyNDPAddress= .network file setting.
3919 * systemd-networkd's bonding device support gained support for two new
3920 configuration options ActiveSlave= and PrimarySlave=.
3922 * The various options in the [Match] section of .network files gained
3923 support for negative matching.
3925 * New systemd-specific mount options are now understood in /etc/fstab:
3927 x-systemd.mount-timeout= may be used to configure the maximum
3928 permitted runtime of the mount command.
3930 x-systemd.device-bound may be set to bind a mount point to its
3931 backing device unit, in order to automatically remove a mount point
3932 if its backing device is unplugged. This option may also be
3933 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
3934 on the block device, which is now automatically set for all CDROM
3935 drives, so that mounted CDs are automatically unmounted when they are
3936 removed from the drive.
3938 x-systemd.after= and x-systemd.before= may be used to explicitly
3939 order a mount after or before another unit or mount point.
3941 * Enqueued start jobs for device units are now automatically garbage
3942 collected if there are no jobs waiting for them anymore.
3944 * systemctl list-jobs gained two new switches: with --after, for every
3945 queued job the jobs it's waiting for are shown; with --before the
3946 jobs which it's blocking are shown.
3948 * systemd-nspawn gained support for ephemeral boots from disk images
3949 (or in other words: --ephemeral and --image= may now be
3950 combined). Moreover, ephemeral boots are now supported for normal
3951 directories, even if the backing file system is not btrfs. Of course,
3952 if the file system does not support file system snapshots or
3953 reflinks, the initial copy operation will be relatively expensive, but
3954 this should still be suitable for many use cases.
3956 * Calendar time specifications in .timer units now support
3957 specifications relative to the end of a month by using "~" instead of
3958 "-" as separator between month and day. For example, "*-02~03" means
3959 "the third last day in February". In addition a new syntax for
3960 repeated events has been added using the "/" character. For example,
3961 "9..17/2:00" means "every two hours from 9am to 5pm".
3963 * systemd-socket-proxyd gained a new parameter --connections-max= for
3964 configuring the maximum number of concurrent connections.
3966 * sd-id128 gained a new API for generating unique IDs for the host in a
3967 way that does not leak the machine ID. Specifically,
3968 sd_id128_get_machine_app_specific() derives an ID based on the
3969 machine ID a in well-defined, non-reversible, stable way. This is
3970 useful whenever an identifier for the host is needed but where the
3971 identifier shall not be useful to identify the system beyond the
3972 scope of the application itself. (Internally this uses HMAC-SHA256 as
3973 keyed hash function using the machine ID as input.)
3975 * NotifyAccess= gained a new supported value "exec". When set
3976 notifications are accepted from all processes systemd itself invoked,
3977 including all control processes.
3979 * .nspawn files gained support for defining overlay mounts using the
3980 Overlay= and OverlayReadOnly= options. Previously this functionality
3981 was only available on the systemd-nspawn command line.
3983 * systemd-nspawn's --bind= and --overlay= options gained support for
3984 bind/overlay mounts whose source lies within the container tree by
3985 prefixing the source path with "+".
3987 * systemd-nspawn's --bind= and --overlay= options gained support for
3988 automatically allocating a temporary source directory in /var/tmp
3989 that is removed when the container dies. Specifically, if the source
3990 directory is specified as empty string this mechanism is selected. An
3991 example usage is --overlay=+/var::/var, which creates an overlay
3992 mount based on the original /var contained in the image, overlaid
3993 with a temporary directory in the host's /var/tmp. This way changes
3994 to /var are automatically flushed when the container shuts down.
3996 * systemd-nspawn --image= option does now permit raw file system block
3997 devices (in addition to images containing partition tables, as
4000 * The disk image dissection logic in systemd-nspawn gained support for
4001 automatically setting up LUKS encrypted as well as Verity protected
4002 partitions. When a container is booted from an encrypted image the
4003 passphrase is queried at start-up time. When a container with Verity
4004 data is started, the root hash is search in a ".roothash" file
4005 accompanying the disk image (alternatively, pass the root hash via
4006 the new --root-hash= command line option).
4008 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
4009 be used to dissect disk images the same way as systemd-nspawn does
4010 it, following the Bootable Partition Specification. It may even be
4011 used to mount disk images with complex partition setups (including
4012 LUKS and Verity partitions) to a local host directory, in order to
4013 inspect them. This tool is not considered public API (yet), and is
4014 thus not installed into /usr/bin. Please do not rely on its
4015 existence, since it might go away or be changed in later systemd
4018 * A new generator "systemd-verity-generator" has been added, similar in
4019 style to "systemd-cryptsetup-generator", permitting automatic setup of
4020 Verity root partitions when systemd boots up. In order to make use of
4021 this your partition setup should follow the Discoverable Partitions
4022 Specification, and the GPT partition ID of the root file system
4023 partition should be identical to the upper 128bit of the Verity root
4024 hash. The GPT partition ID of the Verity partition protecting it
4025 should be the lower 128bit of the Verity root hash. If the partition
4026 image follows this model it is sufficient to specify a single
4027 "roothash=" kernel command line argument to both configure which root
4028 image and verity partition to use as well as the root hash for
4029 it. Note that systemd-nspawn's Verity support follows the same
4030 semantics, meaning that disk images with proper Verity data in place
4031 may be booted in containers with systemd-nspawn as well as on
4032 physical systems via the verity generator. Also note that the "mkosi"
4033 tool available at https://github.com/systemd/mkosi has been updated
4034 to generate Verity protected disk images following this scheme. In
4035 fact, it has been updated to generate disk images that optionally
4036 implement a complete UEFI SecureBoot trust chain, involving a signed
4037 kernel and initrd image that incorporates such a root hash as well as
4038 a Verity-enabled root partition.
4040 * The hardware database (hwdb) udev supports has been updated to carry
4041 accelerometer quirks.
4043 * All system services are now run with a fresh kernel keyring set up
4044 for them. The invocation ID is stored by default in it, thus
4045 providing a safe, non-overridable way to determine the invocation
4048 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
4049 options for bind mounting arbitrary paths in a service-specific
4050 way. When these options are used, arbitrary host or service files and
4051 directories may be mounted to arbitrary locations in the service's
4054 * Documentation has been added that lists all of systemd's low-level
4055 environment variables:
4057 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
4059 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
4060 whether a specific socket file descriptor matches a specified socket
4063 * systemd-firstboot has been updated to check for the
4064 systemd.firstboot= kernel command line option. It accepts a boolean
4065 and when set to false the first boot questions are skipped.
4067 * systemd-fstab-generator has been updated to check for the
4068 systemd.volatile= kernel command line option, which either takes an
4069 optional boolean parameter or the special value "state". If used the
4070 system may be booted in a "volatile" boot mode. Specifically,
4071 "systemd.volatile" is used, the root directory will be mounted as
4072 tmpfs, and only /usr is mounted from the actual root file system. If
4073 "systemd.volatile=state" is used, the root directory will be mounted
4074 as usual, but /var is mounted as tmpfs. This concept provides similar
4075 functionality as systemd-nspawn's --volatile= option, but provides it
4076 on physical boots. Use this option for implementing stateless
4077 systems, or testing systems with all state and/or configuration reset
4078 to the defaults. (Note though that many distributions are not
4079 prepared to boot up without a populated /etc or /var, though.)
4081 * systemd-gpt-auto-generator gained support for LUKS encrypted root
4082 partitions. Previously it only supported LUKS encrypted partitions
4083 for all other uses, except for the root partition itself.
4085 * Socket units gained support for listening on AF_VSOCK sockets for
4086 communication in virtualized QEMU environments.
4088 * The "configure" script gained a new option --with-fallback-hostname=
4089 for specifying the fallback hostname to use if none is configured in
4090 /etc/hostname. For example, by specifying
4091 --with-fallback-hostname=fedora it is possible to default to a
4092 hostname of "fedora" on pristine installations.
4094 * systemd-cgls gained support for a new --unit= switch for listing only
4095 the control groups of a specific unit. Similar --user-unit= has been
4096 added for listing only the control groups of a specific user unit.
4098 * systemd-mount gained a new --umount switch for unmounting a mount or
4099 automount point (and all mount/automount points below it).
4101 * systemd will now refuse full configuration reloads (via systemctl
4102 daemon-reload and related calls) unless at least 16MiB of free space
4103 are available in /run. This is a safety precaution in order to ensure
4104 that generators can safely operate after the reload completed.
4106 * A new unit file option RootImage= has been added, which has a similar
4107 effect as RootDirectory= but mounts the service's root directory from
4108 a disk image instead of plain directory. This logic reuses the same
4109 image dissection and mount logic that systemd-nspawn already uses,
4110 and hence supports any disk images systemd-nspawn supports, including
4111 those following the Discoverable Partition Specification, as well as
4112 Verity enabled images. This option enables systemd to run system
4113 services directly off disk images acting as resource bundles,
4114 possibly even including full integrity data.
4116 * A new MountAPIVFS= unit file option has been added, taking a boolean
4117 argument. If enabled /proc, /sys and /dev (collectively called the
4118 "API VFS") will be mounted for the service. This is only relevant if
4119 RootDirectory= or RootImage= is used for the service, as these mounts
4120 are of course in place in the host mount namespace anyway.
4122 * systemd-nspawn gained support for a new --pivot-root= switch. If
4123 specified the root directory within the container image is pivoted to
4124 the specified mount point, while the original root disk is moved to a
4125 different place. This option enables booting of ostree images
4126 directly with systemd-nspawn.
4128 * The systemd build scripts will no longer complain if the NTP server
4129 addresses are not changed from the defaults. Google now supports
4130 these NTP servers officially. We still recommend downstreams to
4131 properly register an NTP pool with the NTP pool project though.
4133 * coredumpctl gained a new "--reverse" option for printing the list
4134 of coredumps in reverse order.
4136 * coredumpctl will now show additional information about truncated and
4137 inaccessible coredumps, as well as coredumps that are still being
4138 processed. It also gained a new --quiet switch for suppressing
4139 additional informational message in its output.
4141 * coredumpctl gained support for only showing coredumps newer and/or
4142 older than specific timestamps, using the new --since= and --until=
4143 options, reminiscent of journalctl's options by the same name.
4145 * The systemd-coredump logic has been improved so that it may be reused
4146 to collect backtraces in non-compiled languages, for example in
4147 scripting languages such as Python.
4149 * machinectl will now show the UID shift of local containers, if user
4150 namespacing is enabled for them.
4152 * systemd will now optionally run "environment generator" binaries at
4153 configuration load time. They may be used to add environment
4154 variables to the environment block passed to services invoked. One
4155 user environment generator is shipped by default that sets up
4156 environment variables based on files dropped into /etc/environment.d
4157 and ~/.config/environment.d/.
4159 * systemd-resolved now includes the new, recently published 2017 DNSSEC
4162 * hostnamed has been updated to report a new chassis type of
4163 "convertible" to cover "foldable" laptops that can both act as a
4164 tablet and as a laptop, such as various Lenovo Yoga devices.
4166 Contributions from: Adrián López, Alexander Galanin, Alexander
4167 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
4168 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
4169 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
4170 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
4171 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
4172 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
4173 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
4174 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
4175 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
4176 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
4177 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
4178 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
4179 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
4180 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
4181 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
4182 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
4183 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
4184 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
4185 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
4186 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
4187 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
4188 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
4189 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
4190 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
4191 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
4192 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
4193 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
4196 — Berlin, 2017-03-01
4200 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
4201 RestrictAddressFamilies= enabled. These sandboxing options should
4202 generally be compatible with the various external udev call-out
4203 binaries we are aware of, however there may be exceptions, in
4204 particular when exotic languages for these call-outs are used. In
4205 this case, consider turning off these settings locally.
4207 * The new RemoveIPC= option can be used to remove IPC objects owned by
4208 the user or group of a service when that service exits.
4210 * The new ProtectKernelModules= option can be used to disable explicit
4211 load and unload operations of kernel modules by a service. In
4212 addition access to /usr/lib/modules is removed if this option is set.
4214 * ProtectSystem= option gained a new value "strict", which causes the
4215 whole file system tree with the exception of /dev, /proc, and /sys,
4216 to be remounted read-only for a service.
4218 * The new ProtectKernelTunables= option can be used to disable
4219 modification of configuration files in /sys and /proc by a service.
4220 Various directories and files are remounted read-only, so access is
4221 restricted even if the file permissions would allow it.
4223 * The new ProtectControlGroups= option can be used to disable write
4224 access by a service to /sys/fs/cgroup.
4226 * Various systemd services have been hardened with
4227 ProtectKernelTunables=yes, ProtectControlGroups=yes,
4228 RestrictAddressFamilies=.
4230 * Support for dynamically creating users for the lifetime of a service
4231 has been added. If DynamicUser=yes is specified, user and group IDs
4232 will be allocated from the range 61184..65519 for the lifetime of the
4233 service. They can be resolved using the new nss-systemd.so NSS
4234 module. The module must be enabled in /etc/nsswitch.conf. Services
4235 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
4236 any resources allocated by the service will be cleaned up when the
4237 service exits. They also have ProtectHome=read-only and
4238 ProtectSystem=strict enabled, so they are not able to make any
4239 permanent modifications to the system.
4241 * The nss-systemd module also always resolves root and nobody, making
4242 it possible to have no /etc/passwd or /etc/group files in minimal
4243 container or chroot environments.
4245 * Services may be started with their own user namespace using the new
4246 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
4247 under which the service is running are mapped. All other users are
4250 * Support for the cgroup namespace has been added to systemd-nspawn. If
4251 supported by kernel, the container system started by systemd-nspawn
4252 will have its own view of the cgroup hierarchy. This new behaviour
4253 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
4255 * The new MemorySwapMax= option can be used to limit the maximum swap
4256 usage under the unified cgroup hierarchy.
4258 * Support for the CPU controller in the unified cgroup hierarchy has
4259 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
4260 options. This controller requires out-of-tree patches for the kernel
4261 and the support is provisional.
4263 * Mount and automount units may now be created transiently
4264 (i.e. dynamically at runtime via the bus API, instead of requiring
4265 unit files in the file system).
4267 * systemd-mount is a new tool which may mount file systems – much like
4268 mount(8), optionally pulling in additional dependencies through
4269 transient .mount and .automount units. For example, this tool
4270 automatically runs fsck on a backing block device before mounting,
4271 and allows the automount logic to be used dynamically from the
4272 command line for establishing mount points. This tool is particularly
4273 useful when dealing with removable media, as it will ensure fsck is
4274 run – if necessary – before the first access and that the file system
4275 is quickly unmounted after each access by utilizing the automount
4276 logic. This maximizes the chance that the file system on the
4277 removable media stays in a clean state, and if it isn't in a clean
4278 state is fixed automatically.
4280 * LazyUnmount=yes option for mount units has been added to expose the
4281 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
4284 * /efi will be used as the mount point of the EFI boot partition, if
4285 the directory is present, and the mount point was not configured
4286 through other means (e.g. fstab). If /efi directory does not exist,
4287 /boot will be used as before. This makes it easier to automatically
4288 mount the EFI partition on systems where /boot is used for something
4291 * When operating on GPT disk images for containers, systemd-nspawn will
4292 now mount the ESP to /boot or /efi according to the same rules as PID
4293 1 running on a host. This allows tools like "bootctl" to operate
4294 correctly within such containers, in order to make container images
4295 bootable on physical systems.
4297 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
4299 * Two new user session targets have been added to support running
4300 graphical sessions under the systemd --user instance:
4301 graphical-session.target and graphical-session-pre.target. See
4302 systemd.special(7) for a description of how those targets should be
4305 * The vconsole initialization code has been significantly reworked to
4306 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
4307 support unicode keymaps. Font and keymap configuration will now be
4308 copied to all allocated virtual consoles.
4310 * FreeBSD's bhyve virtualization is now detected.
4312 * Information recorded in the journal for core dumps now includes the
4313 contents of /proc/mountinfo and the command line of the process at
4314 the top of the process hierarchy (which is usually the init process
4317 * systemd-journal-gatewayd learned the --directory= option to serve
4318 files from the specified location.
4320 * journalctl --root=… can be used to peruse the journal in the
4321 /var/log/ directories inside of a container tree. This is similar to
4322 the existing --machine= option, but does not require the container to
4325 * The hardware database has been extended to support
4326 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
4329 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
4330 specify the click rate for mice which include a horizontal wheel with
4331 a click rate that is different than the one for the vertical wheel.
4333 * systemd-run gained a new --wait option that makes service execution
4334 synchronous. (Specifically, the command will not return until the
4335 specified service binary exited.)
4337 * systemctl gained a new --wait option that causes the start command to
4338 wait until the units being started have terminated again.
4340 * A new journal output mode "short-full" has been added which displays
4341 timestamps with abbreviated English day names and adds a timezone
4342 suffix. Those timestamps include more information than the default
4343 "short" output mode, and can be passed directly to journalctl's
4344 --since= and --until= options.
4346 * /etc/resolv.conf will be bind-mounted into containers started by
4347 systemd-nspawn, if possible, so any changes to resolv.conf contents
4348 are automatically propagated to the container.
4350 * The number of instances for socket-activated services originating
4351 from a single IP address can be limited with
4352 MaxConnectionsPerSource=, extending the existing setting of
4355 * systemd-networkd gained support for vcan ("Virtual CAN") interface
4358 * .netdev and .network configuration can now be extended through
4361 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
4362 Segmentation Offload, Generic Receive Offload, Large Receive Offload
4363 can be enabled and disabled using the new UDPSegmentationOffload=,
4364 TCPSegmentationOffload=, GenericSegmentationOffload=,
4365 GenericReceiveOffload=, LargeReceiveOffload= options in the
4366 [Link] section of .link files.
4368 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
4369 Port VLAN ID can be configured for bridge devices using the new STP=,
4370 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
4371 section of .netdev files.
4373 * The route table to which routes received over DHCP or RA should be
4374 added can be configured with the new RouteTable= option in the [DHCP]
4375 and [IPv6AcceptRA] sections of .network files.
4377 * The Address Resolution Protocol can be disabled on links managed by
4378 systemd-networkd using the ARP=no setting in the [Link] section of
4381 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
4382 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
4383 encode information about the result and exit codes of the current
4384 service runtime cycle.
4386 * systemd-sysctl will now configure kernel parameters in the order
4387 they occur in the configuration files. This matches what sysctl
4388 has been traditionally doing.
4390 * kernel-install "plugins" that are executed to perform various
4391 tasks after a new kernel is added and before an old one is removed
4392 can now return a special value to terminate the procedure and
4393 prevent any later plugins from running.
4395 * Journald's SplitMode=login setting has been deprecated. It has been
4396 removed from documentation, and its use is discouraged. In a future
4397 release it will be completely removed, and made equivalent to current
4398 default of SplitMode=uid.
4400 * Storage=both option setting in /etc/systemd/coredump.conf has been
4401 removed. With fast LZ4 compression storing the core dump twice is not
4404 * The --share-system systemd-nspawn option has been replaced with an
4405 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
4406 this functionality is discouraged. In addition the variables
4407 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
4408 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
4409 individual namespaces.
4411 * "machinectl list" now shows the IP address of running containers in
4412 the output, as well as OS release information.
4414 * "loginctl list" now shows the TTY of each session in the output.
4416 * sd-bus gained new API calls sd_bus_track_set_recursive(),
4417 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
4418 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
4419 tracking objects in a "recursive" mode, where a single client can be
4420 counted multiple times, if it takes multiple references.
4422 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
4423 sd_bus_get_exit_on_disconnect(). They may be used to make a
4424 process using sd-bus automatically exit if the bus connection is
4427 * Bus clients of the service manager may now "pin" loaded units into
4428 memory, by taking an explicit reference on them. This is useful to
4429 ensure the client can retrieve runtime data about the service even
4430 after the service completed execution. Taking such a reference is
4431 available only for privileged clients and should be helpful to watch
4432 running services in a race-free manner, and in particular collect
4433 information about exit statuses and results.
4435 * The nss-resolve module has been changed to strictly return UNAVAIL
4436 when communication via D-Bus with resolved failed, and NOTFOUND when
4437 a lookup completed but was negative. This means it is now possible to
4438 neatly configure fallbacks using nsswitch.conf result checking
4439 expressions. Taking benefit of this, the new recommended
4440 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
4442 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
4444 * A new setting CtrlAltDelBurstAction= has been added to
4445 /etc/systemd/system.conf which may be used to configure the precise
4446 behaviour if the user on the console presses Ctrl-Alt-Del more often
4447 than 7 times in 2s. Previously this would unconditionally result in
4448 an expedited, immediate reboot. With this new setting the precise
4449 operation may be configured in more detail, and also turned off
4452 * In .netdev files two new settings RemoteChecksumTx= and
4453 RemoteChecksumRx= are now understood that permit configuring the
4454 remote checksumming logic for VXLAN networks.
4456 * The service manager learnt a new "invocation ID" concept for invoked
4457 services. Each runtime cycle of a service will get a new invocation
4458 ID (a 128bit random UUID) assigned that identifies the current
4459 run of the service uniquely and globally. A new invocation ID
4460 is generated each time a service starts up. The journal will store
4461 the invocation ID of a service along with any logged messages, thus
4462 making the invocation ID useful for matching the online runtime of a
4463 service with the offline log data it generated in a safe way without
4464 relying on synchronized timestamps. In many ways this new service
4465 invocation ID concept is similar to the kernel's boot ID concept that
4466 uniquely and globally identifies the runtime of each boot. The
4467 invocation ID of a service is passed to the service itself via an
4468 environment variable ($INVOCATION_ID). A new bus call
4469 GetUnitByInvocationID() has been added that is similar to GetUnit()
4470 but instead of retrieving the bus path for a unit by its name
4471 retrieves it by its invocation ID. The returned path is valid only as
4472 long as the passed invocation ID is current.
4474 * systemd-resolved gained a new "DNSStubListener" setting in
4475 resolved.conf. It either takes a boolean value or the special values
4476 "udp" and "tcp", and configures whether to enable the stub DNS
4477 listener on 127.0.0.53:53.
4479 * IP addresses configured via networkd may now carry additional
4480 configuration settings supported by the kernel. New options include:
4481 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
4482 PrefixRoute=, AutoJoin=.
4484 * The PAM configuration fragment file for "user@.service" shipped with
4485 systemd (i.e. the --user instance of systemd) has been stripped to
4486 the minimum necessary to make the system boot. Previously, it
4487 contained Fedora-specific stanzas that did not apply to other
4488 distributions. It is expected that downstream distributions add
4489 additional configuration lines, matching their needs to this file,
4490 using it only as rough template of what systemd itself needs. Note
4491 that this reduced fragment does not even include an invocation of
4492 pam_limits which most distributions probably want to add, even though
4493 systemd itself does not need it. (There's also the new build time
4494 option --with-pamconfdir=no to disable installation of the PAM
4497 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
4498 capability is now also dropped from its set (in addition to
4499 CAP_SYS_MKNOD as before).
4501 * In service unit files it is now possible to connect a specific named
4502 file descriptor with stdin/stdout/stdout of an executed service. The
4503 name may be specified in matching .socket units using the
4504 FileDescriptorName= setting.
4506 * A number of journal settings may now be configured on the kernel
4507 command line. Specifically, the following options are now understood:
4508 systemd.journald.max_level_console=,
4509 systemd.journald.max_level_store=,
4510 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
4511 systemd.journald.max_level_wall=.
4513 * "systemctl is-enabled --full" will now show by which symlinks a unit
4514 file is enabled in the unit dependency tree.
4516 * Support for VeraCrypt encrypted partitions has been added to the
4517 "cryptsetup" logic and /etc/crypttab.
4519 * systemd-detect-virt gained support for a new --private-users switch
4520 that checks whether the invoking processes are running inside a user
4521 namespace. Similar, a new special value "private-users" for the
4522 existing ConditionVirtualization= setting has been added, permitting
4523 skipping of specific units in user namespace environments.
4525 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
4526 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
4527 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
4528 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
4529 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
4530 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
4531 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
4532 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
4533 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
4534 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
4535 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
4536 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
4537 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
4538 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
4539 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
4540 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
4541 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
4542 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
4543 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
4544 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
4545 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
4546 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
4547 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
4548 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
4549 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
4550 Jędrzejewski-Szmek, Zeal Jagannatha
4552 — Santa Fe, 2016-11-03
4556 * In service units the various ExecXYZ= settings have been extended
4557 with an additional special character as first argument of the
4558 assigned value: if the character '+' is used the specified command
4559 line it will be run with full privileges, regardless of User=,
4560 Group=, CapabilityBoundingSet= and similar options. The effect is
4561 similar to the existing PermissionsStartOnly= option, but allows
4562 configuration of this concept for each executed command line
4565 * Services may now alter the service watchdog timeout at runtime by
4566 sending a WATCHDOG_USEC= message via sd_notify().
4568 * MemoryLimit= and related unit settings now optionally take percentage
4569 specifications. The percentage is taken relative to the amount of
4570 physical memory in the system (or in case of containers, the assigned
4571 amount of memory). This allows scaling service resources neatly with
4572 the amount of RAM available on the system. Similarly, systemd-logind's
4573 RuntimeDirectorySize= option now also optionally takes percentage
4576 * In similar fashion TasksMax= takes percentage values now, too. The
4577 value is taken relative to the configured maximum number of processes
4578 on the system. The per-service task maximum has been changed to 15%
4579 using this functionality. (Effectively this is an increase of 512 →
4580 4915 for service units, given the kernel's default pid_max setting.)
4582 * Calendar time specifications in .timer units now understand a ".."
4583 syntax for time ranges. Example: "4..7:10" may now be used for
4584 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
4587 * The InaccessableDirectories=, ReadOnlyDirectories= and
4588 ReadWriteDirectories= unit file settings have been renamed to
4589 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
4590 applied to all kinds of file nodes, and not just directories, with
4591 the exception of symlinks. Specifically these settings may now be
4592 used on block and character device nodes, UNIX sockets and FIFOS as
4593 well as regular files. The old names of these settings remain
4594 available for compatibility.
4596 * systemd will now log about all service processes it kills forcibly
4597 (using SIGKILL) because they remained after the clean shutdown phase
4598 of the service completed. This should help identifying services that
4599 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
4600 systemd-logind's configuration a similar log message is generated for
4601 processes killed at the end of each session due to this setting.
4603 * systemd will now set the $JOURNAL_STREAM environment variable for all
4604 services whose stdout/stderr are connected to the Journal (which
4605 effectively means by default: all services). The variable contains
4606 the device and inode number of the file descriptor used for
4607 stdout/stderr. This may be used by invoked programs to detect whether
4608 their stdout/stderr is connected to the Journal, in which case they
4609 can switch over to direct Journal communication, thus being able to
4610 pass extended, structured metadata along with their log messages. As
4611 one example, this is now used by glib's logging primitives.
4613 * When using systemd's default tmp.mount unit for /tmp, the mount point
4614 will now be established with the "nosuid" and "nodev" options. This
4615 avoids privilege escalation attacks that put traps and exploits into
4616 /tmp. However, this might cause problems if you e. g. put container
4617 images or overlays into /tmp; if you need this, override tmp.mount's
4618 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
4621 * systemd now supports the "memory" cgroup controller also on
4624 * The systemd-cgtop tool now optionally takes a control group path as
4625 command line argument. If specified, the control group list shown is
4626 limited to subgroups of that group.
4628 * The SystemCallFilter= unit file setting gained support for
4629 pre-defined, named system call filter sets. For example
4630 SystemCallFilter=@clock is now an effective way to make all clock
4631 changing-related system calls unavailable to a service. A number of
4632 similar pre-defined groups are defined. Writing system call filters
4633 for system services is simplified substantially with this new
4634 concept. Accordingly, all of systemd's own, long-running services now
4635 enable system call filtering based on this, by default.
4637 * A new service setting MemoryDenyWriteExecute= has been added, taking
4638 a boolean value. If turned on, a service may no longer create memory
4639 mappings that are writable and executable at the same time. This
4640 enhances security for services where this is enabled as it becomes
4641 harder to dynamically write and then execute memory in exploited
4642 service processes. This option has been enabled for all of systemd's
4643 own long-running services.
4645 * A new RestrictRealtime= service setting has been added, taking a
4646 boolean argument. If set the service's processes may no longer
4647 acquire realtime scheduling. This improves security as realtime
4648 scheduling may otherwise be used to easily freeze the system.
4650 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
4651 value. This may be used for requesting that the system manager inside
4652 of the container reports start-up completion to nspawn which then
4653 propagates this notification further to the service manager
4654 supervising nspawn itself. A related option NotifyReady= in .nspawn
4655 files has been added too. This functionality allows ordering of the
4656 start-up of multiple containers using the usual systemd ordering
4659 * machinectl gained a new command "stop" that is an alias for
4662 * systemd-resolved gained support for contacting DNS servers on
4663 link-local IPv6 addresses.
4665 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
4666 its caches. A method call for requesting the same operation has been
4667 added to the bus API too, and is made available via "systemd-resolve
4670 * systemd-resolve gained a new --status switch. If passed a brief
4671 summary of the used DNS configuration with per-interface information
4674 * resolved.conf gained a new Cache= boolean option, defaulting to
4675 on. If turned off local DNS caching is disabled. This comes with a
4676 performance penalty in particular when DNSSEC is enabled. Note that
4677 resolved disables its internal caching implicitly anyway, when the
4678 configured DNS server is on a host-local IP address such as ::1 or
4679 127.0.0.1, thus automatically avoiding double local caching.
4681 * systemd-resolved now listens on the local IP address 127.0.0.53:53
4682 for DNS requests. This improves compatibility with local programs
4683 that do not use the libc NSS or systemd-resolved's bus APIs for name
4684 resolution. This minimal DNS service is only available to local
4685 programs and does not implement the full DNS protocol, but enough to
4686 cover local DNS clients. A new, static resolv.conf file, listing just
4687 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
4688 now recommended to make /etc/resolv.conf a symlink to this file in
4689 order to route all DNS lookups to systemd-resolved, regardless if
4690 done via NSS, the bus API or raw DNS packets. Note that this local
4691 DNS service is not as fully featured as the libc NSS or
4692 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
4693 used to deliver link-local address information (as this implies
4694 sending a local interface index along), LLMNR/mDNS support via this
4695 interface is severely restricted. It is thus strongly recommended for
4696 all applications to use the libc NSS API or native systemd-resolved
4699 * systemd-networkd's bridge support learned a new setting
4700 VLANFiltering= for controlling VLAN filtering. Moreover a new section
4701 in .network files has been added for configuring VLAN bridging in
4702 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
4704 * systemd-networkd's IPv6 Router Advertisement code now makes use of
4705 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
4706 now be acquired without relying on DHCPv6. Two new options
4707 UseDomains= and UseDNS= have been added to configure this behaviour.
4709 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
4710 renamed IPv6AcceptRA=, without altering its behaviour. The old
4711 setting name remains available for compatibility reasons.
4713 * The systemd-networkd VTI/VTI6 tunneling support gained new options
4714 Key=, InputKey= and OutputKey=.
4716 * systemd-networkd gained support for VRF ("Virtual Routing Function")
4717 interface configuration.
4719 * "systemctl edit" may now be used to create new unit files by
4720 specifying the --force switch.
4722 * sd-event gained a new function sd_event_get_iteration() for
4723 requesting the current iteration counter of the event loop. It starts
4724 at zero and is increased by one with each event loop iteration.
4726 * A new rpm macro %systemd_ordering is provided by the macros.systemd
4727 file. It can be used in lieu of %systemd_requires in packages which
4728 don't use any systemd functionality and are intended to be installed
4729 in minimal containers without systemd present. This macro provides
4730 ordering dependencies to ensure that if the package is installed in
4731 the same rpm transaction as systemd, systemd will be installed before
4732 the scriptlets for the package are executed, allowing unit presets
4735 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
4736 been added to simplify packaging of generators.
4738 * The os-release file gained VERSION_CODENAME field for the
4739 distribution nickname (e.g. VERSION_CODENAME=woody).
4741 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
4742 can be set to disable parsing of metadata and the creation
4743 of persistent symlinks for that device.
4745 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
4746 to make them available to logged-in users has been reverted.
4748 * Much of the common code of the various systemd components is now
4749 built into an internal shared library libsystemd-shared-231.so
4750 (incorporating the systemd version number in the name, to be updated
4751 with future releases) that the components link to. This should
4752 decrease systemd footprint both in memory during runtime and on
4753 disk. Note that the shared library is not for public use, and is
4754 neither API nor ABI stable, but is likely to change with every new
4755 released update. Packagers need to make sure that binaries
4756 linking to libsystemd-shared.so are updated in step with the
4759 * Configuration for "mkosi" is now part of the systemd
4760 repository. mkosi is a tool to easily build legacy-free OS images,
4761 and is available on github: https://github.com/systemd/mkosi. If
4762 "mkosi" is invoked in the build tree a new raw OS image is generated
4763 incorporating the systemd sources currently being worked on and a
4764 clean, fresh distribution installation. The generated OS image may be
4765 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
4766 UEFI PC. This functionality is particularly useful to easily test
4767 local changes made to systemd in a pristine, defined environment. See
4768 doc/HACKING for details.
4770 * configure learned the --with-support-url= option to specify the
4771 distribution's bugtracker.
4773 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
4774 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
4775 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
4776 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
4777 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
4778 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
4779 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
4780 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
4781 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
4782 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
4783 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
4784 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
4785 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
4786 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
4787 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
4788 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
4789 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
4790 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
4791 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
4793 — Berlin, 2016-07-25
4797 * DNSSEC is now turned on by default in systemd-resolved (in
4798 "allow-downgrade" mode), but may be turned off during compile time by
4799 passing "--with-default-dnssec=no" to "configure" (and of course,
4800 during runtime with DNSSEC= in resolved.conf). We recommend
4801 downstreams to leave this on at least during development cycles and
4802 report any issues with the DNSSEC logic upstream. We are very
4803 interested in collecting feedback about the DNSSEC validator and its
4804 limitations in the wild. Note however, that DNSSEC support is
4805 probably nothing downstreams should turn on in stable distros just
4806 yet, as it might create incompatibilities with a few DNS servers and
4807 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
4808 automatically whenever we detect such incompatible setups, but there
4809 might be systems we do not cover yet. Hence: please help us testing
4810 the DNSSEC code, leave this on where you can, report back, but then
4811 again don't consider turning this on in your stable, LTS or
4812 production release just yet. (Note that you have to enable
4813 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
4814 and its DNSSEC mode for hostname resolution from local
4817 * systemd-resolve conveniently resolves DANE records with the --tlsa
4818 option and OPENPGPKEY records with the --openpgp option. It also
4819 supports dumping raw DNS record data via the new --raw= switch.
4821 * systemd-logind will now by default terminate user processes that are
4822 part of the user session scope unit (session-XX.scope) when the user
4823 logs out. This behavior is controlled by the KillUserProcesses=
4824 setting in logind.conf, and the previous default of "no" is now
4825 changed to "yes". This means that user sessions will be properly
4826 cleaned up after, but additional steps are necessary to allow
4827 intentionally long-running processes to survive logout.
4829 While the user is logged in at least once, user@.service is running,
4830 and any service that should survive the end of any individual login
4831 session can be started at a user service or scope using systemd-run.
4832 systemd-run(1) man page has been extended with an example which shows
4833 how to run screen in a scope unit underneath user@.service. The same
4834 command works for tmux.
4836 After the user logs out of all sessions, user@.service will be
4837 terminated too, by default, unless the user has "lingering" enabled.
4838 To effectively allow users to run long-term tasks even if they are
4839 logged out, lingering must be enabled for them. See loginctl(1) for
4840 details. The default polkit policy was modified to allow users to
4841 set lingering for themselves without authentication.
4843 Previous defaults can be restored at compile time by the
4844 --without-kill-user-processes option to "configure".
4846 * systemd-logind gained new configuration settings SessionsMax= and
4847 InhibitorsMax=, both with a default of 8192. It will not register new
4848 user sessions or inhibitors above this limit.
4850 * systemd-logind will now reload configuration on SIGHUP.
4852 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
4853 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
4854 enable. Also, support for the "io" cgroup controller in the unified
4855 hierarchy has been added, so that the "memory", "pids" and "io" are
4856 now the controllers that are supported on the unified hierarchy.
4858 WARNING: it is not possible to use previous systemd versions with
4859 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
4860 is necessary to also update systemd in the initramfs if using the
4861 unified hierarchy. An updated SELinux policy is also required.
4863 * LLDP support has been extended, and both passive (receive-only) and
4864 active (sender) modes are supported. Passive mode ("routers-only") is
4865 enabled by default in systemd-networkd. Active LLDP mode is enabled
4866 by default for containers on the internal network. The "networkctl
4867 lldp" command may be used to list information gathered. "networkctl
4868 status" will also show basic LLDP information on connected peers now.
4870 * The IAID and DUID unique identifier sent in DHCP requests may now be
4871 configured for the system and each .network file managed by
4872 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
4874 * systemd-networkd gained support for configuring proxy ARP support for
4875 each interface, via the ProxyArp= setting in .network files. It also
4876 gained support for configuring the multicast querier feature of
4877 bridge devices, via the new MulticastQuerier= setting in .netdev
4878 files. Similarly, snooping on the IGMP traffic can be controlled
4879 via the new setting MulticastSnooping=.
4881 A new setting PreferredLifetime= has been added for addresses
4882 configured in .network file to configure the lifetime intended for an
4885 The systemd-networkd DHCP server gained the option EmitRouter=, which
4886 defaults to yes, to configure whether the DHCP Option 3 (Router)
4889 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
4890 systemd-socket-activate and installed into /usr/bin. It is now fully
4893 * systemd-journald now uses separate threads to flush changes to disk
4894 when closing journal files, thus reducing impact of slow disk I/O on
4895 logging performance.
4897 * The sd-journal API gained two new calls
4898 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
4899 can be used to open journal files using file descriptors instead of
4900 file or directory paths. sd_journal_open_container() has been
4901 deprecated, sd_journal_open_directory_fd() should be used instead
4902 with the flag SD_JOURNAL_OS_ROOT.
4904 * journalctl learned a new output mode "-o short-unix" that outputs log
4905 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
4906 UTC). It also gained support for a new --no-hostname setting to
4907 suppress the hostname column in the family of "short" output modes.
4909 * systemd-ask-password now optionally skips printing of the password to
4910 stdout with --no-output which can be useful in scripts.
4912 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
4913 (devices tagged with ID_MAKER_TOOL) are now tagged with
4914 "uaccess" and are available to logged in users.
4916 * The DeviceAllow= unit setting now supports specifiers (with "%").
4918 * "systemctl show" gained a new --value switch, which allows print a
4919 only the contents of a specific unit property, without also printing
4920 the property's name. Similar support was added to "show*" verbs
4921 of loginctl and machinectl that output "key=value" lists.
4923 * A new unit type "generated" was added for files dynamically generated
4924 by generator tools. Similarly, a new unit type "transient" is used
4925 for unit files created using the runtime API. "systemctl enable" will
4926 refuse to operate on such files.
4928 * A new command "systemctl revert" has been added that may be used to
4929 revert to the vendor version of a unit file, in case local changes
4930 have been made by adding drop-ins or overriding the unit file.
4932 * "machinectl clean" gained a new verb to automatically remove all or
4933 just hidden container images.
4935 * systemd-tmpfiles gained support for a new line type "e" for emptying
4936 directories, if they exist, without creating them if they don't.
4938 * systemd-nspawn gained support for automatically patching the UID/GIDs
4939 of the owners and the ACLs of all files and directories in a
4940 container tree to match the UID/GID user namespacing range selected
4941 for the container invocation. This mode is enabled via the new
4942 --private-users-chown switch. It also gained support for
4943 automatically choosing a free, previously unused UID/GID range when
4944 starting a container, via the new --private-users=pick setting (which
4945 implies --private-users-chown). Together, these options for the first
4946 time make user namespacing for nspawn containers fully automatic and
4947 thus deployable. The systemd-nspawn@.service template unit file has
4948 been changed to use this functionality by default.
4950 * systemd-nspawn gained a new --network-zone= switch, that allows
4951 creating ad-hoc virtual Ethernet links between multiple containers,
4952 that only exist as long as at least one container referencing them is
4953 running. This allows easy connecting of multiple containers with a
4954 common link that implements an Ethernet broadcast domain. Each of
4955 these network "zones" may be named relatively freely by the user, and
4956 may be referenced by any number of containers, but each container may
4957 only reference one of these "zones". On the lower level, this is
4958 implemented by an automatically managed bridge network interface for
4959 each zone, that is created when the first container referencing its
4960 zone is created and removed when the last one referencing its zone
4963 * The default start timeout may now be configured on the kernel command
4964 line via systemd.default_timeout_start_sec=. It was already
4965 configurable via the DefaultTimeoutStartSec= option in
4966 /etc/systemd/system.conf.
4968 * Socket units gained a new TriggerLimitIntervalSec= and
4969 TriggerLimitBurst= setting to configure a limit on the activation
4970 rate of the socket unit.
4972 * The LimitNICE= setting now optionally takes normal UNIX nice values
4973 in addition to the raw integer limit value. If the specified
4974 parameter is prefixed with "+" or "-" and is in the range -20..19 the
4975 value is understood as UNIX nice value. If not prefixed like this it
4976 is understood as raw RLIMIT_NICE limit.
4978 * Note that the effect of the PrivateDevices= unit file setting changed
4979 slightly with this release: the per-device /dev file system will be
4980 mounted read-only from this version on, and will have "noexec"
4981 set. This (minor) change of behavior might cause some (exceptional)
4982 legacy software to break, when PrivateDevices=yes is set for its
4983 service. Please leave PrivateDevices= off if you run into problems
4986 * systemd-bootchart has been split out to a separate repository:
4987 https://github.com/systemd/systemd-bootchart
4989 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
4990 merged into the kernel in its current form.
4992 * The compatibility libraries libsystemd-daemon.so,
4993 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
4994 which have been deprecated since systemd-209 have been removed along
4995 with the corresponding pkg-config files. All symbols provided by
4996 those libraries are provided by libsystemd.so.
4998 * The Capabilities= unit file setting has been removed (it is ignored
4999 for backwards compatibility). AmbientCapabilities= and
5000 CapabilityBoundingSet= should be used instead.
5002 * A new special target has been added, initrd-root-device.target,
5003 which creates a synchronization point for dependencies of the root
5004 device in early userspace. Initramfs builders must ensure that this
5005 target is now included in early userspace.
5007 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
5008 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
5009 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
5010 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
5011 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
5012 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
5013 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
5014 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
5015 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
5016 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
5017 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
5018 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
5019 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
5020 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
5021 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
5022 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
5023 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
5024 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
5025 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
5026 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
5027 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
5028 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
5029 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
5030 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
5031 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
5034 — Fairfax, 2016-05-21
5038 * The systemd-resolved DNS resolver service has gained a substantial
5039 set of new features, most prominently it may now act as a DNSSEC
5040 validating stub resolver. DNSSEC mode is currently turned off by
5041 default, but is expected to be turned on by default in one of the
5042 next releases. For now, we invite everybody to test the DNSSEC logic
5043 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
5044 service also gained a full set of D-Bus interfaces, including calls
5045 to configure DNS and DNSSEC settings per link (for use by external
5046 network management software). systemd-resolved and systemd-networkd
5047 now distinguish between "search" and "routing" domains. The former
5048 are used to qualify single-label names, the latter are used purely
5049 for routing lookups within certain domains to specific links.
5050 resolved now also synthesizes RRs for all entries from /etc/hosts.
5052 * The systemd-resolve tool (which is a client utility for
5053 systemd-resolved) has been improved considerably and is now fully
5054 supported and documented. Hence it has moved from /usr/lib/systemd to
5057 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
5060 * The coredump collection logic has been reworked: when a coredump is
5061 collected it is now written to disk, compressed and processed
5062 (including stacktrace extraction) from a new instantiated service
5063 systemd-coredump@.service, instead of directly from the
5064 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
5065 processing large coredumps can take up a substantial amount of
5066 resources and time, and this previously happened entirely outside of
5067 systemd's service supervision. With the new logic the core_pattern
5068 hook only does minimal metadata collection before passing off control
5069 to the new instantiated service, which is configured with a time
5070 limit, a nice level and other settings to minimize negative impact on
5071 the rest of the system. Also note that the new logic will honour the
5072 RLIMIT_CORE setting of the crashed process, which now allows users
5073 and processes to turn off coredumping for their processes by setting
5076 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
5077 and all forked processes by default. Previously, PID 1 would leave
5078 the setting at "0" for all processes, as set by the kernel. Note that
5079 the resource limit traditionally has no effect on the generated
5080 coredumps on the system if the /proc/sys/kernel/core_pattern hook
5081 logic is used. Since the limit is now honoured (see above) its
5082 default has been changed so that the coredumping logic is enabled by
5083 default for all processes, while allowing specific opt-out.
5085 * When the stacktrace is extracted from processes of system users, this
5086 is now done as "systemd-coredump" user, in order to sandbox this
5087 potentially security sensitive parsing operation. (Note that when
5088 processing coredumps of normal users this is done under the user ID
5089 of process that crashed, as before.) Packagers should take notice
5090 that it is now necessary to create the "systemd-coredump" system user
5091 and group at package installation time.
5093 * The systemd-activate socket activation testing tool gained support
5094 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
5095 and --seqpacket switches. It also has been extended to support both
5096 new-style and inetd-style file descriptor passing. Use the new
5097 --inetd switch to request inetd-style file descriptor passing.
5099 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
5100 variable, which takes a boolean value. If set to false, ANSI color
5101 output is disabled in the tools even when run on a terminal that
5104 * The VXLAN support in networkd now supports two new settings
5105 DestinationPort= and PortRange=.
5107 * A new systemd.machine_id= kernel command line switch has been added,
5108 that may be used to set the machine ID in /etc/machine-id if it is
5109 not initialized yet. This command line option has no effect if the
5110 file is already initialized.
5112 * systemd-nspawn gained a new --as-pid2 switch that invokes any
5113 specified command line as PID 2 rather than PID 1 in the
5114 container. In this mode PID 1 is a minimal stub init process that
5115 implements the special POSIX and Linux semantics of PID 1 regarding
5116 signal and child process management. Note that this stub init process
5117 is implemented in nspawn itself and requires no support from the
5118 container image. This new logic is useful to support running
5119 arbitrary commands in the container, as normal processes are
5120 generally not prepared to run as PID 1.
5122 * systemd-nspawn gained a new --chdir= switch for setting the current
5123 working directory for the process started in the container.
5125 * "journalctl /dev/sda" will now output all kernel log messages for
5126 specified device from the current boot, in addition to all devices
5127 that are parents of it. This should make log output about devices
5128 pretty useful, as long as kernel drivers attach enough metadata to
5129 the log messages. (The usual SATA drivers do.)
5131 * The sd-journal API gained two new calls
5132 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
5133 that report whether log data from /run or /var has been found.
5135 * journalctl gained a new switch "--fields" that prints all journal
5136 record field names currently in use in the journal. This is backed
5137 by two new sd-journal API calls sd_journal_enumerate_fields() and
5138 sd_journal_restart_fields().
5140 * Most configurable timeouts in systemd now expect an argument of
5141 "infinity" to turn them off, instead of "0" as before. The semantics
5142 from now on is that a timeout of "0" means "now", and "infinity"
5143 means "never". To maintain backwards compatibility, "0" continues to
5144 turn off previously existing timeout settings.
5146 * "systemctl reload-or-try-restart" has been renamed to "systemctl
5147 try-reload-or-restart" to clarify what it actually does: the "try"
5148 logic applies to both reloading and restarting, not just restarting.
5149 The old name continues to be accepted for compatibility.
5151 * On boot-up, when PID 1 detects that the system clock is behind the
5152 release date of the systemd version in use, the clock is now set
5153 to the latter. Previously, this was already done in timesyncd, in order
5154 to avoid running with clocks set to the various clock epochs such as
5155 1902, 1938 or 1970. With this change the logic is now done in PID 1
5156 in addition to timesyncd during early boot-up, so that it is enforced
5157 before the first process is spawned by systemd. Note that the logic
5158 in timesyncd remains, as it is more comprehensive and ensures
5159 clock monotonicity by maintaining a persistent timestamp file in
5160 /var. Since /var is generally not available in earliest boot or the
5161 initrd, this part of the logic remains in timesyncd, and is not done
5164 * Support for tweaking details in net_cls.class_id through the
5165 NetClass= configuration directive has been removed, as the kernel
5166 people have decided to deprecate that controller in cgroup v2.
5167 Userspace tools such as nftables are moving over to setting rules
5168 that are specific to the full cgroup path of a task, which obsoletes
5169 these controllers anyway. The NetClass= directive is kept around for
5170 legacy compatibility reasons. For a more in-depth description of the
5171 kernel change, please refer to the respective upstream commit:
5173 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
5175 * A new service setting RuntimeMaxSec= has been added that may be used
5176 to specify a maximum runtime for a service. If the timeout is hit, the
5177 service is terminated and put into a failure state.
5179 * A new service setting AmbientCapabilities= has been added. It allows
5180 configuration of additional Linux process capabilities that are
5181 passed to the activated processes. This is only available on very
5184 * The process resource limit settings in service units may now be used
5185 to configure hard and soft limits individually.
5187 * The various libsystemd APIs such as sd-bus or sd-event now publicly
5188 expose support for gcc's __attribute__((cleanup())) C extension.
5189 Specifically, for many object destructor functions alternative
5190 versions have been added that have names suffixed with "p" and take a
5191 pointer to a pointer to the object to destroy, instead of just a
5192 pointer to the object itself. This is useful because these destructor
5193 functions may be used directly as parameters to the cleanup
5194 construct. Internally, systemd has been a heavy user of this GCC
5195 extension for a long time, and with this change similar support is
5196 now available to consumers of the library outside of systemd. Note
5197 that by using this extension in your sources compatibility with old
5198 and strictly ANSI compatible C compilers is lost. However, all gcc or
5199 LLVM versions of recent years support this extension.
5201 * Timer units gained support for a new setting RandomizedDelaySec= that
5202 allows configuring some additional randomized delay to the configured
5203 time. This is useful to spread out timer events to avoid load peaks in
5204 clusters or larger setups.
5206 * Calendar time specifications now support sub-second accuracy.
5208 * Socket units now support listening on SCTP and UDP-lite protocol
5211 * The sd-event API now comes with a full set of man pages.
5213 * Older versions of systemd contained experimental support for
5214 compressing journal files and coredumps with the LZ4 compressor that
5215 was not compatible with the lz4 binary (due to API limitations of the
5216 lz4 library). This support has been removed; only support for files
5217 compatible with the lz4 binary remains. This LZ4 logic is now
5218 officially supported and no longer considered experimental.
5220 * The dkr image import logic has been removed again from importd. dkr's
5221 micro-services focus doesn't fit into the machine image focus of
5222 importd, and quickly got out of date with the upstream dkr API.
5224 * Creation of the /run/lock/lockdev/ directory was dropped from
5225 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
5226 been available for many years. If you still need this, you need to
5227 create your own tmpfiles.d config file with:
5229 d /run/lock/lockdev 0775 root lock -
5231 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
5232 and RebootArgument= have been moved from the [Service] section of
5233 unit files to [Unit], and they are now supported on all unit types,
5234 not just service units. Of course, systemd will continue to
5235 understand these settings also at the old location, in order to
5236 maintain compatibility.
5238 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
5239 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
5240 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
5241 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
5242 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
5243 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
5244 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
5245 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
5246 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
5247 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
5248 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
5249 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
5250 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
5251 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
5252 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
5253 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
5254 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
5255 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
5256 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
5258 — Berlin, 2016-02-11
5262 * A number of properties previously only settable in unit
5263 files are now also available as properties to set when
5264 creating transient units programmatically via the bus, as it
5265 is exposed with systemd-run's --property=
5266 setting. Specifically, these are: SyslogIdentifier=,
5267 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
5268 EnvironmentFile=, ReadWriteDirectories=,
5269 ReadOnlyDirectories=, InaccessibleDirectories=,
5270 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
5272 * When creating transient services via the bus API it is now
5273 possible to pass in a set of file descriptors to use as
5274 STDIN/STDOUT/STDERR for the invoked process.
5276 * Slice units may now be created transiently via the bus APIs,
5277 similar to the way service and scope units may already be
5278 created transiently.
5280 * Wherever systemd expects a calendar timestamp specification
5281 (like in journalctl's --since= and --until= switches) UTC
5282 timestamps are now supported. Timestamps suffixed with "UTC"
5283 are now considered to be in Universal Time Coordinated
5284 instead of the local timezone. Also, timestamps may now
5285 optionally be specified with sub-second accuracy. Both of
5286 these additions also apply to recurring calendar event
5287 specification, such as OnCalendar= in timer units.
5289 * journalctl gained a new "--sync" switch that asks the
5290 journal daemon to write all so far unwritten log messages to
5291 disk and sync the files, before returning.
5293 * systemd-tmpfiles learned two new line types "q" and "Q" that
5294 operate like "v", but also set up a basic btrfs quota
5295 hierarchy when used on a btrfs file system with quota
5298 * tmpfiles' "v", "q" and "Q" will now create a plain directory
5299 instead of a subvolume (even on a btrfs file system) if the
5300 root directory is a plain directory, and not a
5301 subvolume. This should simplify things with certain chroot()
5302 environments which are not aware of the concept of btrfs
5305 * systemd-detect-virt gained a new --chroot switch to detect
5306 whether execution takes place in a chroot() environment.
5308 * CPUAffinity= now takes CPU index ranges in addition to
5311 * The various memory-related resource limit settings (such as
5312 LimitAS=) now understand the usual K, M, G, ... suffixes to
5313 the base of 1024 (IEC). Similar, the time-related resource
5314 limit settings understand the usual min, h, day, ...
5317 * There's a new system.conf setting DefaultTasksMax= to
5318 control the default TasksMax= setting for services and
5319 scopes running on the system. (TasksMax= is the primary
5320 setting that exposes the "pids" cgroup controller on systemd
5321 and was introduced in the previous systemd release.) The
5322 setting now defaults to 512, which means services that are
5323 not explicitly configured otherwise will only be able to
5324 create 512 processes or threads at maximum, from this
5325 version on. Note that this means that thread- or
5326 process-heavy services might need to be reconfigured to set
5327 TasksMax= to a higher value. It is sufficient to set
5328 TasksMax= in these specific unit files to a higher value, or
5329 even "infinity". Similar, there's now a logind.conf setting
5330 UserTasksMax= that defaults to 4096 and limits the total
5331 number of processes or tasks each user may own
5332 concurrently. nspawn containers also have the TasksMax=
5333 value set by default now, to 8192. Note that all of this
5334 only has an effect if the "pids" cgroup controller is
5335 enabled in the kernel. The general benefit of these changes
5336 should be a more robust and safer system, that provides a
5337 certain amount of per-service fork() bomb protection.
5339 * systemd-nspawn gained the new --network-veth-extra= switch
5340 to define additional and arbitrarily-named virtual Ethernet
5341 links between the host and the container.
5343 * A new service execution setting PassEnvironment= has been
5344 added that allows importing select environment variables
5345 from PID1's environment block into the environment block of
5348 * Timer units gained support for a new RemainAfterElapse=
5349 setting which takes a boolean argument. It defaults to on,
5350 exposing behaviour unchanged to previous releases. If set to
5351 off, timer units are unloaded after they elapsed if they
5352 cannot elapse again. This is particularly useful for
5353 transient timer units, which shall not stay around longer
5354 than until they first elapse.
5356 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
5357 default now (the kernel default is 16). This is beneficial
5358 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
5359 allows substantially larger numbers of queued
5360 datagrams. This should increase the capability of systemd to
5361 parallelize boot-up, as logging and sd_notify() are unlikely
5362 to stall execution anymore. If you need to change the value
5363 from the new defaults, use the usual sysctl.d/ snippets.
5365 * The compression framing format used by the journal or
5366 coredump processing has changed to be in line with what the
5367 official LZ4 tools generate. LZ4 compression support in
5368 systemd was considered unsupported previously, as the format
5369 was not compatible with the normal tools. With this release
5370 this has changed now, and it is hence safe for downstream
5371 distributions to turn it on. While not compressing as well
5372 as the XZ, LZ4 is substantially faster, which makes
5373 it a good default choice for the compression logic in the
5374 journal and in coredump handling.
5376 * Any reference to /etc/mtab has been dropped from
5377 systemd. The file has been obsolete since a while, but
5378 systemd refused to work on systems where it was incorrectly
5379 set up (it should be a symlink or non-existent). Please make
5380 sure to update to util-linux 2.27.1 or newer in conjunction
5381 with this systemd release, which also drops any reference to
5382 /etc/mtab. If you maintain a distribution make sure that no
5383 software you package still references it, as this is a
5384 likely source of bugs. There's also a glibc bug pending,
5385 asking for removal of any reference to this obsolete file:
5387 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
5389 Note that only util-linux versions built with
5390 --enable-libmount-force-mountinfo are supported.
5392 * Support for the ".snapshot" unit type has been removed. This
5393 feature turned out to be little useful and little used, and
5394 has now been removed from the core and from systemctl.
5396 * The dependency types RequiresOverridable= and
5397 RequisiteOverridable= have been removed from systemd. They
5398 have been used only very sparingly to our knowledge and
5399 other options that provide a similar effect (such as
5400 systemctl --mode=ignore-dependencies) are much more useful
5401 and commonly used. Moreover, they were only half-way
5402 implemented as the option to control behaviour regarding
5403 these dependencies was never added to systemctl. By removing
5404 these dependency types the execution engine becomes a bit
5405 simpler. Unit files that use these dependencies should be
5406 changed to use the non-Overridable dependency types
5407 instead. In fact, when parsing unit files with these
5408 options, that's what systemd will automatically convert them
5409 too, but it will also warn, asking users to fix the unit
5410 files accordingly. Removal of these dependency types should
5411 only affect a negligible number of unit files in the wild.
5413 * Behaviour of networkd's IPForward= option changed
5414 (again). It will no longer maintain a per-interface setting,
5415 but propagate one way from interfaces where this is enabled
5416 to the global kernel setting. The global setting will be
5417 enabled when requested by a network that is set up, but
5418 never be disabled again. This change was made to make sure
5419 IPv4 and IPv6 behaviour regarding packet forwarding is
5420 similar (as the Linux IPv6 stack does not support
5421 per-interface control of this setting) and to minimize
5424 * In unit files the behaviour of %u, %U, %h, %s has
5425 changed. These specifiers will now unconditionally resolve
5426 to the various user database fields of the user that the
5427 systemd instance is running as, instead of the user
5428 configured in the specific unit via User=. Note that this
5429 effectively doesn't change much, as resolving of these
5430 specifiers was already turned off in the --system instance
5431 of systemd, as we cannot do NSS lookups from PID 1. In the
5432 --user instance of systemd these specifiers where correctly
5433 resolved, but hardly made any sense, since the user instance
5434 lacks privileges to do user switches anyway, and User= is
5435 hence useless. Moreover, even in the --user instance of
5436 systemd behaviour was awkward as it would only take settings
5437 from User= assignment placed before the specifier into
5438 account. In order to unify and simplify the logic around
5439 this the specifiers will now always resolve to the
5440 credentials of the user invoking the manager (which in case
5441 of PID 1 is the root user).
5443 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
5444 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
5445 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
5446 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
5447 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
5448 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
5449 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
5450 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
5451 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
5452 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
5453 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
5454 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
5455 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
5456 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
5459 — Berlin, 2015-11-18
5463 * systemd now depends on util-linux v2.27. More specifically,
5464 the newly added mount monitor feature in libmount now
5465 replaces systemd's former own implementation.
5467 * libmount mandates /etc/mtab not to be regular file, and
5468 systemd now enforces this condition at early boot.
5469 /etc/mtab has been deprecated and warned about for a very
5470 long time, so systems running systemd should already have
5471 stopped having this file around as anything else than a
5472 symlink to /proc/self/mounts.
5474 * Support for the "pids" cgroup controller has been added. It
5475 allows accounting the number of tasks in a cgroup and
5476 enforcing limits on it. This adds two new setting
5477 TasksAccounting= and TasksMax= to each unit, as well as a
5478 global option DefaultTasksAccounting=.
5480 * Support for the "net_cls" cgroup controller has been added.
5481 It allows assigning a net class ID to each task in the
5482 cgroup, which can then be used in firewall rules and traffic
5483 shaping configurations. Note that the kernel netfilter net
5484 class code does not currently work reliably for ingress
5485 packets on unestablished sockets.
5487 This adds a new config directive called NetClass= to CGroup
5488 enabled units. Allowed values are positive numbers for fixed
5489 assignments and "auto" for picking a free value
5492 * 'systemctl is-system-running' now returns 'offline' if the
5493 system is not booted with systemd. This command can now be
5494 used as a substitute for 'systemd-notify --booted'.
5496 * Watchdog timeouts have been increased to 3 minutes for all
5497 in-tree service files. Apparently, disk IO issues are more
5498 frequent than we hoped, and user reported >1 minute waiting
5501 * 'machine-id-commit' functionality has been merged into
5502 'machine-id-setup --commit'. The separate binary has been
5505 * The WorkingDirectory= directive in unit files may now be set
5506 to the special value '~'. In this case, the working
5507 directory is set to the home directory of the user
5508 configured in User=.
5510 * "machinectl shell" will now open the shell in the home
5511 directory of the selected user by default.
5513 * The CrashChVT= configuration file setting is renamed to
5514 CrashChangeVT=, following our usual logic of not
5515 abbreviating unnecessarily. The old directive is still
5516 supported for compat reasons. Also, this directive now takes
5517 an integer value between 1 and 63, or a boolean value. The
5518 formerly supported '-1' value for disabling stays around for
5521 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
5522 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
5523 RootDirectory= properties can now be set for transient
5526 * The systemd-analyze tool gained a new "set-log-target" verb
5527 to change the logging target the system manager logs to
5528 dynamically during runtime. This is similar to how
5529 "systemd-analyze set-log-level" already changes the log
5532 * In nspawn /sys is now mounted as tmpfs, with only a selected
5533 set of subdirectories mounted in from the real sysfs. This
5534 enhances security slightly, and is useful for ensuring user
5535 namespaces work correctly.
5537 * Support for USB FunctionFS activation has been added. This
5538 allows implementation of USB gadget services that are
5539 activated as soon as they are requested, so that they don't
5540 have to run continuously, similar to classic socket
5543 * The "systemctl exit" command now optionally takes an
5544 additional parameter that sets the exit code to return from
5545 the systemd manager when exiting. This is only relevant when
5546 running the systemd user instance, or when running the
5547 system instance in a container.
5549 * sd-bus gained the new API calls sd_bus_path_encode_many()
5550 and sd_bus_path_decode_many() that allow easy encoding and
5551 decoding of multiple identifier strings inside a D-Bus
5552 object path. Another new call sd_bus_default_flush_close()
5553 has been added to flush and close per-thread default
5556 * systemd-cgtop gained support for a -M/--machine= switch to
5557 show the control groups within a certain container only.
5559 * "systemctl kill" gained support for an optional --fail
5560 switch. If specified the requested operation will fail of no
5561 processes have been killed, because the unit had no
5562 processes attached, or similar.
5564 * A new systemd.crash_reboot=1 kernel command line option has
5565 been added that triggers a reboot after crashing. This can
5566 also be set through CrashReboot= in systemd.conf.
5568 * The RuntimeDirectory= setting now understands unit
5569 specifiers like %i or %f.
5571 * A new (still internal) library API sd-ipv4acd has been added,
5572 that implements address conflict detection for IPv4. It's
5573 based on code from sd-ipv4ll, and will be useful for
5574 detecting DHCP address conflicts.
5576 * File descriptors passed during socket activation may now be
5577 named. A new API sd_listen_fds_with_names() is added to
5578 access the names. The default names may be overridden,
5579 either in the .socket file using the FileDescriptorName=
5580 parameter, or by passing FDNAME= when storing the file
5581 descriptors using sd_notify().
5583 * systemd-networkd gained support for:
5585 - Setting the IPv6 Router Advertisement settings via
5586 IPv6AcceptRouterAdvertisements= in .network files.
5588 - Configuring the HelloTimeSec=, MaxAgeSec= and
5589 ForwardDelaySec= bridge parameters in .netdev files.
5591 - Configuring PreferredSource= for static routes in
5594 * The "ask-password" framework used to query for LUKS harddisk
5595 passwords or SSL passwords during boot gained support for
5596 caching passwords in the kernel keyring, if it is
5597 available. This makes sure that the user only has to type in
5598 a passphrase once if there are multiple objects to unlock
5599 with the same one. Previously, such password caching was
5600 available only when Plymouth was used; this moves the
5601 caching logic into the systemd codebase itself. The
5602 "systemd-ask-password" utility gained a new --keyname=
5603 switch to control which kernel keyring key to use for
5604 caching a password in. This functionality is also useful for
5605 enabling display managers such as gdm to automatically
5606 unlock the user's GNOME keyring if its passphrase, the
5607 user's password and the harddisk password are the same, if
5608 gdm-autologin is used.
5610 * When downloading tar or raw images using "machinectl
5611 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
5612 file is now also downloaded, if it is available and stored
5613 next to the image file.
5615 * Units of type ".socket" gained a new boolean setting
5616 Writable= which is only useful in conjunction with
5617 ListenSpecial=. If true, enables opening the specified
5618 special file in O_RDWR mode rather than O_RDONLY mode.
5620 * systemd-rfkill has been reworked to become a singleton
5621 service that is activated through /dev/rfkill on each rfkill
5622 state change and saves the settings to disk. This way,
5623 systemd-rfkill is now compatible with devices that exist
5624 only intermittendly, and even restores state if the previous
5625 system shutdown was abrupt rather than clean.
5627 * The journal daemon gained support for vacuuming old journal
5628 files controlled by the number of files that shall remain,
5629 in addition to the already existing control by size and by
5630 date. This is useful as journal interleaving performance
5631 degrades with too many separate journal files, and allows
5632 putting an effective limit on them. The new setting defaults
5633 to 100, but this may be changed by setting SystemMaxFiles=
5634 and RuntimeMaxFiles= in journald.conf. Also, the
5635 "journalctl" tool gained the new --vacuum-files= switch to
5636 manually vacuum journal files to leave only the specified
5637 number of files in place.
5639 * udev will now create /dev/disk/by-path links for ATA devices
5640 on kernels where that is supported.
5642 * Galician, Serbian, Turkish and Korean translations were added.
5644 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
5645 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
5646 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
5647 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
5648 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
5649 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
5650 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
5651 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
5652 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
5653 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
5654 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
5655 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
5656 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
5657 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
5658 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
5659 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
5660 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
5661 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
5663 — Berlin, 2015-10-07
5667 * The DHCP implementation of systemd-networkd gained a set of
5670 - The DHCP server now supports emitting DNS and NTP
5671 information. It may be enabled and configured via
5672 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
5673 and NTP information is enabled, but no servers are
5674 configured, the corresponding uplink information (if there
5675 is any) is propagated.
5677 - Server and client now support transmission and reception
5678 of timezone information. It can be configured via the
5679 newly introduced network options UseTimezone=,
5680 EmitTimezone=, and Timezone=. Transmission of timezone
5681 information is enabled between host and containers by
5682 default now: the container will change its local timezone
5683 to what the host has set.
5685 - Lease timeouts can now be configured via
5686 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
5688 - The DHCP server improved on the stability of
5689 leases. Clients are more likely to get the same lease
5690 information back, even if the server loses state.
5692 - The DHCP server supports two new configuration options to
5693 control the lease address pool metrics, PoolOffset= and
5696 * The encapsulation limit of tunnels in systemd-networkd may
5697 now be configured via 'EncapsulationLimit='. It allows
5698 modifying the maximum additional levels of encapsulation
5699 that are permitted to be prepended to a packet.
5701 * systemd now supports the concept of user buses replacing
5702 session buses, if used with dbus-1.10 (and enabled via dbus
5703 --enable-user-session). It previously only supported this on
5704 kdbus-enabled systems, and this release expands this to
5705 'dbus-daemon' systems.
5707 * systemd-networkd now supports predictable interface names
5710 * systemd now optionally supports the new Linux kernel
5711 "unified" control group hierarchy. If enabled via the kernel
5712 command-line option 'systemd.unified_cgroup_hierarchy=1',
5713 systemd will try to mount the unified cgroup hierarchy
5714 directly on /sys/fs/cgroup. If not enabled, or not
5715 available, systemd will fall back to the legacy cgroup
5716 hierarchy setup, as before. Host system and containers can
5717 mix and match legacy and unified hierarchies as they
5718 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
5719 environment variable to individually select the hierarchy to
5720 use for executed containers. By default, nspawn will use the
5721 unified hierarchy for the containers if the host uses the
5722 unified hierarchy, and the legacy hierarchy otherwise.
5723 Please note that at this point the unified hierarchy is an
5724 experimental kernel feature and is likely to change in one
5725 of the next kernel releases. Therefore, it should not be
5726 enabled by default in downstream distributions yet. The
5727 minimum required kernel version for the unified hierarchy to
5728 work is 4.2. Note that when the unified hierarchy is used
5729 for the first time delegated access to controllers is
5730 safe. Because of this systemd-nspawn containers will get
5731 access to controllers now, as will systemd user
5732 sessions. This means containers and user sessions may now
5733 manage their own resources, partitioning up what the system
5736 * A new special scope unit "init.scope" has been introduced
5737 that encapsulates PID 1 of the system. It may be used to
5738 determine resource usage and enforce resource limits on PID
5739 1 itself. PID 1 hence moved out of the root of the control
5742 * The cgtop tool gained support for filtering out kernel
5743 threads when counting tasks in a control group. Also, the
5744 count of processes is now recursively summed up by
5745 default. Two options -k and --recursive= have been added to
5746 revert to old behaviour. The tool has also been updated to
5747 work correctly in containers now.
5749 * systemd-nspawn's --bind= and --bind-ro= options have been
5750 extended to allow creation of non-recursive bind mounts.
5752 * libsystemd gained two new calls sd_pid_get_cgroup() and
5753 sd_peer_get_cgroup() which return the control group path of
5754 a process or peer of a connected AF_UNIX socket. This
5755 function call is particularly useful when implementing
5756 delegated subtrees support in the control group hierarchy.
5758 * The "sd-event" event loop API of libsystemd now supports
5759 correct dequeuing of real-time signals, without losing
5762 * When systemd requests a polkit decision when managing units it
5763 will now add additional fields to the request, including unit
5764 name and desired operation. This enables more powerful polkit
5765 policies, that make decisions depending on these parameters.
5767 * nspawn learnt support for .nspawn settings files, that may
5768 accompany the image files or directories of containers, and
5769 may contain additional settings for the container. This is
5770 an alternative to configuring container parameters via the
5771 nspawn command line.
5773 Contributions from: Cristian Rodríguez, Daniel Mack, David
5774 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
5775 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
5776 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
5777 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
5778 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
5779 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
5780 Andersen, Tom Gundersen, Torstein Husebø
5782 — Berlin, 2015-09-08
5786 * machinectl gained a new verb 'shell' which opens a fresh
5787 shell on the target container or the host. It is similar to
5788 the existing 'login' command of machinectl, but spawns the
5789 shell directly without prompting for username or
5790 password. The pseudo machine '.host' now refers to the local
5791 host and is used by default. Hence, 'machinectl shell' can
5792 be used as replacement for 'su -' which spawns a session as
5793 a fresh systemd unit in a way that is fully isolated from
5794 the originating session.
5796 * systemd-networkd learned to cope with private-zone DHCP
5797 options and allows other programs to query the values.
5799 * SELinux access control when enabling/disabling units is no
5800 longer enforced with this release. The previous implementation
5801 was incorrect, and a new corrected implementation is not yet
5802 available. As unit file operations are still protected via
5803 polkit and D-Bus policy this is not a security problem. Yet,
5804 distributions which care about optimal SELinux support should
5805 probably not stabilize on this release.
5807 * sd-bus gained support for matches of type "arg0has=", that
5808 test for membership of strings in string arrays sent in bus
5811 * systemd-resolved now dumps the contents of its DNS and LLMNR
5812 caches to the logs on reception of the SIGUSR1 signal. This
5813 is useful to debug DNS behaviour.
5815 * The coredumpctl tool gained a new --directory= option to
5816 operate on journal files in a specific directory.
5818 * "systemctl reboot" and related commands gained a new
5819 "--message=" option which may be used to set a free-text
5820 wall message when shutting down or rebooting the
5821 system. This message is also logged, which is useful for
5822 figuring out the reason for a reboot or shutdown a
5825 * The "systemd-resolve-host" tool's -i switch now takes
5826 network interface numbers as alternative to interface names.
5828 * A new unit file setting for services has been introduced:
5829 UtmpMode= allows configuration of how precisely systemd
5830 handles utmp and wtmp entries for the service if this is
5831 enabled. This allows writing services that appear similar to
5832 user sessions in the output of the "w", "who", "last" and
5835 * systemd-resolved will now locally synthesize DNS resource
5836 records for the "localhost" and "gateway" domains as well as
5837 the local hostname. This should ensure that clients querying
5838 RRs via resolved will get similar results as those going via
5839 NSS, if nss-myhostname is enabled.
5841 Contributions from: Alastair Hughes, Alex Crawford, Daniel
5842 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
5843 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
5844 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
5845 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
5846 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
5847 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
5848 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
5849 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
5850 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
5851 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
5852 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
5854 — Berlin, 2015-08-27
5858 * The systemd-efi-boot-generator functionality was merged into
5859 systemd-gpt-auto-generator.
5861 * systemd-networkd now supports Group Policy for vxlan
5862 devices. It can be enabled via the new boolean configuration
5863 option called 'GroupPolicyExtension='.
5865 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
5866 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
5867 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
5869 — Berlin, 2015-07-31
5873 * The python-systemd code has been removed from the systemd repository.
5874 A new repository has been created which accommodates the code from
5875 now on, and we kindly ask distributions to create a separate package
5876 for this: https://github.com/systemd/python-systemd
5878 * The systemd daemon will now reload its main configuration
5879 (/etc/systemd/system.conf) on daemon-reload.
5881 * sd-dhcp now exposes vendor specific extensions via
5882 sd_dhcp_lease_get_vendor_specific().
5884 * systemd-networkd gained a number of new configuration options.
5886 - A new boolean configuration option for TAP devices called
5887 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
5888 device, thus allowing to send and receive GSO packets.
5890 - A new tunnel configuration option called 'CopyDSCP='.
5891 If enabled, the DSCP field of ip6 tunnels is copied into the
5892 decapsulated packet.
5894 - A set of boolean bridge configuration options were added.
5895 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
5896 and 'UnicastFlood=' are now parsed by networkd and applied to the
5897 respective bridge link device via the respective IFLA_BRPORT_*
5900 - A new string configuration option to override the hostname sent
5901 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
5902 is true, networkd will use the configured hostname instead of the
5903 system hostname when sending DHCP requests.
5905 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
5906 networkd will configure the IPv6 flow-label of the tunnel device
5907 according to RFC2460.
5909 - The 'macvtap' virtual network devices are now supported, similar to
5910 the already supported 'macvlan' devices.
5912 * systemd-resolved now implements RFC5452 to improve resilience against
5913 cache poisoning. Additionally, source port randomization is enabled
5914 by default to further protect against DNS spoofing attacks.
5916 * nss-mymachines now supports translating UIDs and GIDs of running
5917 containers with user-namespaces enabled. If a container 'foo'
5918 translates a host uid 'UID' to the container uid 'TUID', then
5919 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
5920 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
5921 mapped as 'vg-foo-TGID'.
5923 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
5924 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
5925 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
5926 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
5927 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
5928 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
5929 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
5930 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
5931 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
5932 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
5934 — Berlin, 2015-07-29
5938 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
5939 There are no known issues with current sysfs, and udev does not need
5940 or should be used to work around such bugs.
5942 * udev does no longer enable USB HID power management. Several reports
5943 indicate, that some devices cannot handle that setting.
5945 * The udev accelerometer helper was removed. The functionality
5946 is now fully included in iio-sensor-proxy. But this means,
5947 older iio-sensor-proxy versions will no longer provide
5948 accelerometer/orientation data with this systemd version.
5949 Please upgrade iio-sensor-proxy to version 1.0.
5951 * networkd gained a new configuration option IPv6PrivacyExtensions=
5952 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
5953 for Stateless Address") on selected networks.
5955 * For the sake of fewer build-time dependencies and less code in the
5956 main repository, the python bindings are about to be removed in the
5957 next release. A new repository has been created which accommodates
5958 the code from now on, and we kindly ask distributions to create a
5959 separate package for this. The removal will take place in v223.
5961 https://github.com/systemd/python-systemd
5963 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
5964 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
5965 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
5966 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
5967 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
5968 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
5969 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
5970 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
5971 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5972 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
5974 — Berlin, 2015-07-07
5978 * The sd-bus.h and sd-event.h APIs have now been declared
5979 stable and have been added to the official interface of
5980 libsystemd.so. sd-bus implements an alternative D-Bus client
5981 library, that is relatively easy to use, very efficient and
5982 supports both classic D-Bus as well as kdbus as transport
5983 backend. sd-event is a generic event loop abstraction that
5984 is built around Linux epoll, but adds features such as event
5985 prioritization or efficient timer handling. Both APIs are good
5986 choices for C programs looking for a bus and/or event loop
5987 implementation that is minimal and does not have to be
5988 portable to other kernels.
5990 * kdbus support is no longer compile-time optional. It is now
5991 always built-in. However, it can still be disabled at
5992 runtime using the kdbus=0 kernel command line setting, and
5993 that setting may be changed to default to off, by specifying
5994 --disable-kdbus at build-time. Note though that the kernel
5995 command line setting has no effect if the kdbus.ko kernel
5996 module is not installed, in which case kdbus is (obviously)
5997 also disabled. We encourage all downstream distributions to
5998 begin testing kdbus by adding it to the kernel images in the
5999 development distributions, and leaving kdbus support in
6002 * The minimal required util-linux version has been bumped to
6005 * Support for chkconfig (--enable-chkconfig) was removed in
6006 favor of calling an abstraction tool
6007 /lib/systemd/systemd-sysv-install. This needs to be
6008 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
6009 in README for details.
6011 * If there's a systemd unit and a SysV init script for the
6012 same service name, and the user executes "systemctl enable"
6013 for it (or a related call), then this will now enable both
6014 (or execute the related operation on both), not just the
6017 * The libudev API documentation has been converted from gtkdoc
6020 * gudev has been removed from the systemd tree, it is now an
6023 * The systemd-cgtop tool learnt a new --raw switch to generate
6024 "raw" (machine parsable) output.
6026 * networkd's IPForwarding= .network file setting learnt the
6027 new setting "kernel", which ensures that networkd does not
6028 change the IP forwarding sysctl from the default kernel
6031 * The systemd-logind bus API now exposes a new boolean
6032 property "Docked" that reports whether logind considers the
6033 system "docked", i.e. connected to a docking station or not.
6035 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
6036 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
6037 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
6038 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
6039 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
6040 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
6041 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
6042 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
6043 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
6044 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
6045 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
6046 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
6047 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
6048 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
6049 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
6050 Fink, Zbigniew Jędrzejewski-Szmek
6052 — Berlin, 2015-06-19
6056 * The gudev library has been extracted into a separate repository
6057 available at: https://git.gnome.org/browse/libgudev/
6058 It is now managed as part of the Gnome project. Distributions
6059 are recommended to pass --disable-gudev to systemd and use
6060 gudev from the Gnome project instead. gudev is still included
6061 in systemd, for now. It will be removed soon, though. Please
6062 also see the announcement-thread on systemd-devel:
6063 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
6065 * systemd now exposes a CPUUsageNSec= property for each
6066 service unit on the bus, that contains the overall consumed
6067 CPU time of a service (the sum of what each process of the
6068 service consumed). This value is only available if
6069 CPUAccounting= is turned on for a service, and is then shown
6070 in the "systemctl status" output.
6072 * Support for configuring alternative mappings of the old SysV
6073 runlevels to systemd targets has been removed. They are now
6074 hardcoded in a way that runlevels 2, 3, 4 all map to
6075 multi-user.target and 5 to graphical.target (which
6076 previously was already the default behaviour).
6078 * The auto-mounter logic gained support for mount point
6079 expiry, using a new TimeoutIdleSec= setting in .automount
6080 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
6082 * The EFI System Partition (ESP) as mounted to /boot by
6083 systemd-efi-boot-generator will now be unmounted
6084 automatically after 2 minutes of not being used. This should
6085 minimize the risk of ESP corruptions.
6087 * New /etc/fstab options x-systemd.requires= and
6088 x-systemd.requires-mounts-for= are now supported to express
6089 additional dependencies for mounts. This is useful for
6090 journalling file systems that support external journal
6091 devices or overlay file systems that require underlying file
6092 systems to be mounted.
6094 * systemd does not support direct live-upgrades (via systemctl
6095 daemon-reexec) from versions older than v44 anymore. As no
6096 distribution we are aware of shipped such old versions in a
6097 stable release this should not be problematic.
6099 * When systemd forks off a new per-connection service instance
6100 it will now set the $REMOTE_ADDR environment variable to the
6101 remote IP address, and $REMOTE_PORT environment variable to
6102 the remote IP port. This behaviour is similar to the
6103 corresponding environment variables defined by CGI.
6105 * systemd-networkd gained support for uplink failure
6106 detection. The BindCarrier= option allows binding interface
6107 configuration dynamically to the link sense of other
6108 interfaces. This is useful to achieve behaviour like in
6111 * systemd-networkd gained support for configuring the DHCP
6112 client identifier to use when requesting leases.
6114 * systemd-networkd now has a per-network UseNTP= option to
6115 configure whether NTP server information acquired via DHCP
6116 is passed on to services like systemd-timesyncd.
6118 * systemd-networkd gained support for vti6 tunnels.
6120 * Note that systemd-networkd manages the sysctl variable
6121 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
6122 it is configured for since v219. The variable controls IP
6123 forwarding, and is a per-interface alternative to the global
6124 /proc/sys/net/ipv[46]/ip_forward. This setting is
6125 configurable in the IPForward= option, which defaults to
6126 "no". This means if networkd is used for an interface it is
6127 no longer sufficient to set the global sysctl option to turn
6128 on IP forwarding! Instead, the .network file option
6129 IPForward= needs to be turned on! Note that the
6130 implementation of this behaviour was broken in v219 and has
6133 * Many bonding and vxlan options are now configurable in
6136 * systemd-nspawn gained a new --property= setting to set unit
6137 properties for the container scope. This is useful for
6138 setting resource parameters (e.g. "CPUShares=500") on
6139 containers started from the command line.
6141 * systemd-nspawn gained a new --private-users= switch to make
6142 use of user namespacing available on recent Linux kernels.
6144 * systemd-nspawn may now be called as part of a shell pipeline
6145 in which case the pipes used for stdin and stdout are passed
6146 directly to the process invoked in the container, without
6147 indirection via a pseudo tty.
6149 * systemd-nspawn gained a new switch to control the UNIX
6150 signal to use when killing the init process of the container
6153 * systemd-nspawn gained a new --overlay= switch for mounting
6154 overlay file systems into the container using the new kernel
6157 * When a container image is imported via systemd-importd and
6158 the host file system is not btrfs, a loopback block device
6159 file is created in /var/lib/machines.raw with a btrfs file
6160 system inside. It is then mounted to /var/lib/machines to
6161 enable btrfs features for container management. The loopback
6162 file and btrfs file system is grown as needed when container
6163 images are imported via systemd-importd.
6165 * systemd-machined/systemd-importd gained support for btrfs
6166 quota, to enforce container disk space limits on disk. This
6167 is exposed in "machinectl set-limit".
6169 * systemd-importd now can import containers from local .tar,
6170 .raw and .qcow2 images, and export them to .tar and .raw. It
6171 can also import dkr v2 images now from the network (on top
6174 * systemd-importd gained support for verifying downloaded
6175 images with gpg2 (previously only gpg1 was supported).
6177 * systemd-machined, systemd-logind, systemd: most bus calls are
6178 now accessible to unprivileged processes via polkit. Also,
6179 systemd-logind will now allow users to kill their own sessions
6180 without further privileges or authorization.
6182 * systemd-shutdownd has been removed. This service was
6183 previously responsible for implementing scheduled shutdowns
6184 as exposed in /usr/bin/shutdown's time parameter. This
6185 functionality has now been moved into systemd-logind and is
6186 accessible via a bus interface.
6188 * "systemctl reboot" gained a new switch --firmware-setup that
6189 can be used to reboot into the EFI firmware setup, if that
6190 is available. systemd-logind now exposes an API on the bus
6191 to trigger such reboots, in case graphical desktop UIs want
6192 to cover this functionality.
6194 * "systemctl enable", "systemctl disable" and "systemctl mask"
6195 now support a new "--now" switch. If specified the units
6196 that are enabled will also be started, and the ones
6197 disabled/masked also stopped.
6199 * The Gummiboot EFI boot loader tool has been merged into
6200 systemd, and renamed to "systemd-boot". The bootctl tool has been
6201 updated to support systemd-boot.
6203 * An EFI kernel stub has been added that may be used to create
6204 kernel EFI binaries that contain not only the actual kernel,
6205 but also an initrd, boot splash, command line and OS release
6206 information. This combined binary can then be signed as a
6207 single image, so that the firmware can verify it all in one
6208 step. systemd-boot has special support for EFI binaries created
6209 like this and can extract OS release information from them
6210 and show them in the boot menu. This functionality is useful
6211 to implement cryptographically verified boot schemes.
6213 * Optional support has been added to systemd-fsck to pass
6214 fsck's progress report to an AF_UNIX socket in the file
6217 * udev will no longer create device symlinks for all block devices by
6218 default. A deny list for excluding special block devices from this
6219 logic has been turned into a allow list that requires picking block
6220 devices explicitly that require device symlinks.
6222 * A new (currently still internal) API sd-device.h has been
6223 added to libsystemd. This modernized API is supposed to
6224 replace libudev eventually. In fact, already much of libudev
6225 is now just a wrapper around sd-device.h.
6227 * A new hwdb database for storing metadata about pointing
6228 stick devices has been added.
6230 * systemd-tmpfiles gained support for setting file attributes
6231 similar to the "chattr" tool with new 'h' and 'H' lines.
6233 * systemd-journald will no longer unconditionally set the
6234 btrfs NOCOW flag on new journal files. This is instead done
6235 with tmpfiles snippet using the new 'h' line type. This
6236 allows easy disabling of this logic, by masking the
6237 journal-nocow.conf tmpfiles file.
6239 * systemd-journald will now translate audit message types to
6240 human readable identifiers when writing them to the
6241 journal. This should improve readability of audit messages.
6243 * The LUKS logic gained support for the offset= and skip=
6244 options in /etc/crypttab, as previously implemented by
6247 * /usr/lib/os-release gained a new optional field VARIANT= for
6248 distributions that support multiple variants (such as a
6249 desktop edition, a server edition, ...)
6251 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
6252 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
6253 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
6254 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
6255 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
6256 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
6257 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
6258 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
6259 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
6260 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
6261 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
6262 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
6263 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
6264 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
6265 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
6266 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
6267 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
6268 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
6269 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
6270 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
6271 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
6272 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
6273 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
6274 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
6275 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
6276 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
6277 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
6279 — Berlin, 2015-05-22
6283 * Introduce a new API "sd-hwdb.h" for querying the hardware
6284 metadata database. With this minimal interface one can query
6285 and enumerate the udev hwdb, decoupled from the old libudev
6286 library. libudev's interface for this is now only a wrapper
6287 around sd-hwdb. A new tool systemd-hwdb has been added to
6288 interface with and update the database.
6290 * When any of systemd's tools copies files (for example due to
6291 tmpfiles' C lines) a btrfs reflink will attempted first,
6292 before bytewise copying is done.
6294 * systemd-nspawn gained a new --ephemeral switch. When
6295 specified a btrfs snapshot is taken of the container's root
6296 directory, and immediately removed when the container
6297 terminates again. Thus, a container can be started whose
6298 changes never alter the container's root directory, and are
6299 lost on container termination. This switch can also be used
6300 for starting a container off the root file system of the
6301 host without affecting the host OS. This switch is only
6302 available on btrfs file systems.
6304 * systemd-nspawn gained a new --template= switch. It takes the
6305 path to a container tree to use as template for the tree
6306 specified via --directory=, should that directory be
6307 missing. This allows instantiating containers dynamically,
6308 on first run. This switch is only available on btrfs file
6311 * When a .mount unit refers to a mount point on which multiple
6312 mounts are stacked, and the .mount unit is stopped all of
6313 the stacked mount points will now be unmounted until no
6314 mount point remains.
6316 * systemd now has an explicit notion of supported and
6317 unsupported unit types. Jobs enqueued for unsupported unit
6318 types will now fail with an "unsupported" error code. More
6319 specifically .swap, .automount and .device units are not
6320 supported in containers, .busname units are not supported on
6321 non-kdbus systems. .swap and .automount are also not
6322 supported if their respective kernel compile time options
6325 * machinectl gained support for two new "copy-from" and
6326 "copy-to" commands for copying files from a running
6327 container to the host or vice versa.
6329 * machinectl gained support for a new "bind" command to bind
6330 mount host directories into local containers. This is
6331 currently only supported for nspawn containers.
6333 * networkd gained support for configuring bridge forwarding
6334 database entries (fdb) from .network files.
6336 * A new tiny daemon "systemd-importd" has been added that can
6337 download container images in tar, raw, qcow2 or dkr formats,
6338 and make them available locally in /var/lib/machines, so
6339 that they can run as nspawn containers. The daemon can GPG
6340 verify the downloads (not supported for dkr, since it has no
6341 provisions for verifying downloads). It will transparently
6342 decompress bz2, xz, gzip compressed downloads if necessary,
6343 and restore sparse files on disk. The daemon uses privilege
6344 separation to ensure the actual download logic runs with
6345 fewer privileges than the daemon itself. machinectl has
6346 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
6347 make the functionality of importd available to the
6348 user. With this in place the Fedora and Ubuntu "Cloud"
6349 images can be downloaded and booted as containers unmodified
6350 (the Fedora images lack the appropriate GPG signature files
6351 currently, so they cannot be verified, but this will change
6352 soon, hopefully). Note that downloading images is currently
6353 only fully supported on btrfs.
6355 * machinectl is now able to list container images found in
6356 /var/lib/machines, along with some metadata about sizes of
6357 disk and similar. If the directory is located on btrfs and
6358 quota is enabled, this includes quota display. A new command
6359 "image-status" has been added that shows additional
6360 information about images.
6362 * machinectl is now able to clone container images
6363 efficiently, if the underlying file system (btrfs) supports
6364 it, with the new "machinectl clone" command. It also
6365 gained commands for renaming and removing images, as well as
6366 marking them read-only or read-write (supported also on
6367 legacy file systems).
6369 * networkd gained support for collecting LLDP network
6370 announcements, from hardware that supports this. This is
6371 shown in networkctl output.
6373 * systemd-run gained support for a new -t (--pty) switch for
6374 invoking a binary on a pty whose input and output is
6375 connected to the invoking terminal. This allows executing
6376 processes as system services while interactively
6377 communicating with them via the terminal. Most interestingly
6378 this is supported across container boundaries. Invoking
6379 "systemd-run -t /bin/bash" is an alternative to running a
6380 full login session, the difference being that the former
6381 will not register a session, nor go through the PAM session
6384 * tmpfiles gained support for a new "v" line type for creating
6385 btrfs subvolumes. If the underlying file system is a legacy
6386 file system, this automatically degrades to creating a
6387 normal directory. Among others /var/lib/machines is now
6388 created like this at boot, should it be missing.
6390 * The directory /var/lib/containers/ has been deprecated and
6391 been replaced by /var/lib/machines. The term "machines" has
6392 been used in the systemd context as generic term for both
6393 VMs and containers, and hence appears more appropriate for
6394 this, as the directory can also contain raw images bootable
6397 * systemd-nspawn when invoked with -M but without --directory=
6398 or --image= is now capable of searching for the container
6399 root directory, subvolume or disk image automatically, in
6400 /var/lib/machines. systemd-nspawn@.service has been updated
6401 to make use of this, thus allowing it to be used for raw
6404 * A new machines.target unit has been introduced that is
6405 supposed to group all containers/VMs invoked as services on
6406 the system. systemd-nspawn@.service has been updated to
6407 integrate with that.
6409 * machinectl gained a new "start" command, for invoking a
6410 container as a service. "machinectl start foo" is mostly
6411 equivalent to "systemctl start systemd-nspawn@foo.service",
6412 but handles escaping in a nicer way.
6414 * systemd-nspawn will now mount most of the cgroupfs tree
6415 read-only into each container, with the exception of the
6416 container's own subtree in the name=systemd hierarchy.
6418 * journald now sets the special FS_NOCOW file flag for its
6419 journal files. This should improve performance on btrfs, by
6420 avoiding heavy fragmentation when journald's write-pattern
6421 is used on COW file systems. It degrades btrfs' data
6422 integrity guarantees for the files to the same levels as for
6423 ext3/ext4 however. This should be OK though as journald does
6424 its own data integrity checks and all its objects are
6425 checksummed on disk. Also, journald should handle btrfs disk
6426 full events a lot more gracefully now, by processing SIGBUS
6427 errors, and not relying on fallocate() anymore.
6429 * When journald detects that journal files it is writing to
6430 have been deleted it will immediately start new journal
6433 * systemd now provides a way to store file descriptors
6434 per-service in PID 1. This is useful for daemons to ensure
6435 that fds they require are not lost during a daemon
6436 restart. The fds are passed to the daemon on the next
6437 invocation in the same way socket activation fds are
6438 passed. This is now used by journald to ensure that the
6439 various sockets connected to all the system's stdout/stderr
6440 are not lost when journald is restarted. File descriptors
6441 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
6442 an extension to sd_notify(). Note that a limit is enforced
6443 on the number of fds a service can store in PID 1, and it
6444 defaults to 0, so that no fds may be stored, unless this is
6445 explicitly turned on.
6447 * The default TERM variable to use for units connected to a
6448 terminal, when no other value is explicitly is set is now
6449 vt220 rather than vt102. This should be fairly safe still,
6450 but allows PgUp/PgDn work.
6452 * The /etc/crypttab option header= as known from Debian is now
6455 * "loginctl user-status" and "loginctl session-status" will
6456 now show the last 10 lines of log messages of the
6457 user/session following the status output. Similar,
6458 "machinectl status" will show the last 10 log lines
6459 associated with a virtual machine or container
6460 service. (Note that this is usually not the log messages
6461 done in the VM/container itself, but simply what the
6462 container manager logs. For nspawn this includes all console
6465 * "loginctl session-status" without further argument will now
6466 show the status of the session of the caller. Similar,
6467 "lock-session", "unlock-session", "activate",
6468 "enable-linger", "disable-linger" may now be called without
6469 session/user parameter in which case they apply to the
6470 caller's session/user.
6472 * An X11 session scriptlet is now shipped that uploads
6473 $DISPLAY and $XAUTHORITY into the environment of the systemd
6474 --user daemon if a session begins. This should improve
6475 compatibility with X11 enabled applications run as systemd
6478 * Generators are now subject to masking via /etc and /run, the
6479 same way as unit files.
6481 * networkd .network files gained support for configuring
6482 per-link IPv4/IPv6 packet forwarding as well as IPv4
6483 masquerading. This is by default turned on for veth links to
6484 containers, as registered by systemd-nspawn. This means that
6485 nspawn containers run with --network-veth will now get
6486 automatic routed access to the host's networks without any
6487 further configuration or setup, as long as networkd runs on
6490 * systemd-nspawn gained the --port= (-p) switch to expose TCP
6491 or UDP posts of a container on the host. With this in place
6492 it is possible to run containers with private veth links
6493 (--network-veth), and have their functionality exposed on
6494 the host as if their services were running directly on the
6497 * systemd-nspawn's --network-veth switch now gained a short
6498 version "-n", since with the changes above it is now truly
6499 useful out-of-the-box. The systemd-nspawn@.service has been
6500 updated to make use of it too by default.
6502 * systemd-nspawn will now maintain a per-image R/W lock, to
6503 ensure that the same image is not started more than once
6504 writable. (It's OK to run an image multiple times
6505 simultaneously in read-only mode.)
6507 * systemd-nspawn's --image= option is now capable of
6508 dissecting and booting MBR and GPT disk images that contain
6509 only a single active Linux partition. Previously it
6510 supported only GPT disk images with proper GPT type
6511 IDs. This allows running cloud images from major
6512 distributions directly with systemd-nspawn, without
6515 * In addition to collecting mouse dpi data in the udev
6516 hardware database, there's now support for collecting angle
6517 information for mouse scroll wheels. The database is
6518 supposed to guarantee similar scrolling behavior on mice
6519 that it knows about. There's also support for collecting
6520 information about Touchpad types.
6522 * udev's input_id built-in will now also collect touch screen
6523 dimension data and attach it to probed devices.
6525 * /etc/os-release gained support for a Distribution Privacy
6528 * networkd gained support for creating "ipvlan", "gretap",
6529 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
6531 * systemd-tmpfiles gained support for "a" lines for setting
6534 * systemd-nspawn will now mount /tmp in the container to
6535 tmpfs, automatically.
6537 * systemd now exposes the memory.usage_in_bytes cgroup
6538 attribute and shows it for each service in the "systemctl
6539 status" output, if available.
6541 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
6542 immediate reboot is triggered. This useful if shutdown is
6543 hung and is unable to complete, to expedite the
6544 operation. Note that this kind of reboot will still unmount
6545 all file systems, and hence should not result in fsck being
6548 * A .device unit for an optical block device will now be
6549 considered active only when a medium is in the drive. Also,
6550 mount units are now bound to their backing devices thus
6551 triggering automatic unmounting when devices become
6552 unavailable. With this in place systemd will now
6553 automatically unmount left-over mounts when a CD-ROM is
6554 ejected or an USB stick is yanked from the system.
6556 * networkd-wait-online now has support for waiting for
6557 specific interfaces only (with globbing), and for giving up
6558 after a configurable timeout.
6560 * networkd now exits when idle. It will be automatically
6561 restarted as soon as interfaces show up, are removed or
6562 change state. networkd will stay around as long as there is
6563 at least one DHCP state machine or similar around, that keep
6566 * networkd may now configure IPv6 link-local addressing in
6567 addition to IPv4 link-local addressing.
6569 * The IPv6 "token" for use in SLAAC may now be configured for
6570 each .network interface in networkd.
6572 * Routes configured with networkd may now be assigned a scope
6575 * networkd's [Match] sections now support globbing and lists
6576 of multiple space-separated matches per item.
6578 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
6579 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
6580 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
6581 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
6582 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
6583 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
6584 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
6585 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
6586 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
6587 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
6588 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
6589 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
6590 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
6591 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
6592 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
6593 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
6594 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
6595 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
6596 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
6597 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
6598 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
6599 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
6600 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
6601 Hoffmann, Zbigniew Jędrzejewski-Szmek
6603 — Berlin, 2015-02-16
6607 * When querying unit file enablement status (for example via
6608 "systemctl is-enabled"), a new state "indirect" is now known
6609 which indicates that a unit might not be enabled itself, but
6610 another unit listed in its Also= setting might be.
6612 * Similar to the various existing ConditionXYZ= settings for
6613 units, there are now matching AssertXYZ= settings. While
6614 failing conditions cause a unit to be skipped, but its job
6615 to succeed, failing assertions declared like this will cause
6616 a unit start operation and its job to fail.
6618 * hostnamed now knows a new chassis type "embedded".
6620 * systemctl gained a new "edit" command. When used on a unit
6621 file, this allows extending unit files with .d/ drop-in
6622 configuration snippets or editing the full file (after
6623 copying it from /usr/lib to /etc). This will invoke the
6624 user's editor (as configured with $EDITOR), and reload the
6625 modified configuration after editing.
6627 * "systemctl status" now shows the suggested enablement state
6628 for a unit, as declared in the (usually vendor-supplied)
6629 system preset files.
6631 * nss-myhostname will now resolve the single-label hostname
6632 "gateway" to the locally configured default IP routing
6633 gateways, ordered by their metrics. This assigns a stable
6634 name to the used gateways, regardless which ones are
6635 currently configured. Note that the name will only be
6636 resolved after all other name sources (if nss-myhostname is
6637 configured properly) and should hence not negatively impact
6638 systems that use the single-label hostname "gateway" in
6641 * systemd-inhibit now allows filtering by mode when listing
6644 * Scope and service units gained a new "Delegate" boolean
6645 property, which, when set, allows processes running inside the
6646 unit to further partition resources. This is primarily
6647 useful for systemd user instances as well as container
6650 * journald will now pick up audit messages directly from
6651 the kernel, and log them like any other log message. The
6652 audit fields are split up and fully indexed. This means that
6653 journalctl in many ways is now a (nicer!) alternative to
6654 ausearch, the traditional audit client. Note that this
6655 implements only a minimal audit client. If you want the
6656 special audit modes like reboot-on-log-overflow, please use
6657 the traditional auditd instead, which can be used in
6658 parallel to journald.
6660 * The ConditionSecurity= unit file option now understands the
6661 special string "audit" to check whether auditing is
6664 * journalctl gained two new commands --vacuum-size= and
6665 --vacuum-time= to delete old journal files until the
6666 remaining ones take up no more than the specified size on disk,
6667 or are not older than the specified time.
6669 * A new, native PPPoE library has been added to sd-network,
6670 systemd's library of light-weight networking protocols. This
6671 library will be used in a future version of networkd to
6672 enable PPPoE communication without an external pppd daemon.
6674 * The busctl tool now understands a new "capture" verb that
6675 works similar to "monitor", but writes a packet capture
6676 trace to STDOUT that can be redirected to a file which is
6677 compatible with libcap's capture file format. This can then
6678 be loaded in Wireshark and similar tools to inspect bus
6681 * The busctl tool now understands a new "tree" verb that shows
6682 the object trees of a specific service on the bus, or of all
6685 * The busctl tool now understands a new "introspect" verb that
6686 shows all interfaces and members of objects on the bus,
6687 including their signature and values. This is particularly
6688 useful to get more information about bus objects shown by
6689 the new "busctl tree" command.
6691 * The busctl tool now understands new verbs "call",
6692 "set-property" and "get-property" for invoking bus method
6693 calls, setting and getting bus object properties in a
6696 * busctl gained a new --augment-creds= argument that controls
6697 whether the tool shall augment credential information it
6698 gets from the bus with data from /proc, in a possibly
6701 * nspawn's --link-journal= switch gained two new values
6702 "try-guest" and "try-host" that work like "guest" and
6703 "host", but do not fail if the host has no persistent
6704 journalling enabled. -j is now equivalent to
6705 --link-journal=try-guest.
6707 * macvlan network devices created by nspawn will now have
6708 stable MAC addresses.
6710 * A new SmackProcessLabel= unit setting has been added, which
6711 controls the SMACK security label processes forked off by
6712 the respective unit shall use.
6714 * If compiled with --enable-xkbcommon, systemd-localed will
6715 verify x11 keymap settings by compiling the given keymap. It
6716 will spew out warnings if the compilation fails. This
6717 requires libxkbcommon to be installed.
6719 * When a coredump is collected, a larger number of metadata
6720 fields is now collected and included in the journal records
6721 created for it. More specifically, control group membership,
6722 environment variables, memory maps, working directory,
6723 chroot directory, /proc/$PID/status, and a list of open file
6724 descriptors is now stored in the log entry.
6726 * The udev hwdb now contains DPI information for mice. For
6729 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
6731 * All systemd programs that read standalone configuration
6732 files in /etc now also support a corresponding series of
6733 .conf.d configuration directories in /etc/, /run/,
6734 /usr/local/lib/, /usr/lib/, and (if configured with
6735 --enable-split-usr) /lib/. In particular, the following
6736 configuration files now have corresponding configuration
6737 directories: system.conf user.conf, logind.conf,
6738 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
6739 resolved.conf, timesyncd.conf, journal-remote.conf, and
6740 journal-upload.conf. Note that distributions should use the
6741 configuration directories in /usr/lib/; the directories in
6742 /etc/ are reserved for the system administrator.
6744 * systemd-rfkill will no longer take the rfkill device name
6745 into account when storing rfkill state on disk, as the name
6746 might be dynamically assigned and not stable. Instead, the
6747 ID_PATH udev variable combined with the rfkill type (wlan,
6748 bluetooth, ...) is used.
6750 * A new service systemd-machine-id-commit.service has been
6751 added. When used on systems where /etc is read-only during
6752 boot, and /etc/machine-id is not initialized (but an empty
6753 file), this service will copy the temporary machine ID
6754 created as replacement into /etc after the system is fully
6755 booted up. This is useful for systems that are freshly
6756 installed with a non-initialized machine ID, but should get
6757 a fixed machine ID for subsequent boots.
6759 * networkd's .netdev files now provide a large set of
6760 configuration parameters for VXLAN devices. Similarly, the
6761 bridge port cost parameter is now configurable in .network
6762 files. There's also new support for configuring IP source
6763 routing. networkd .link files gained support for a new
6764 OriginalName= match that is useful to match against the
6765 original interface name the kernel assigned. .network files
6766 may include MTU= and MACAddress= fields for altering the MTU
6767 and MAC address while being connected to a specific network
6770 * The LUKS logic gained supported for configuring
6771 UUID-specific key files. There's also new support for naming
6772 LUKS device from the kernel command line, using the new
6773 luks.name= argument.
6775 * Timer units may now be transiently created via the bus API
6776 (this was previously already available for scope and service
6777 units). In addition it is now possible to create multiple
6778 transient units at the same time with a single bus call. The
6779 "systemd-run" tool has been updated to make use of this for
6780 running commands on a specified time, in at(1)-style.
6782 * tmpfiles gained support for "t" lines, for assigning
6783 extended attributes to files. Among other uses this may be
6784 used to assign SMACK labels to files.
6786 Contributions from: Alin Rauta, Alison Chaiken, Andrej
6787 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
6788 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
6789 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
6790 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
6791 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
6792 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
6793 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
6794 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
6795 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
6796 Michael Biebl, Michael Chapman, Michael Marineau, Michal
6797 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
6798 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
6799 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
6800 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
6801 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6802 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
6803 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
6805 — Berlin, 2014-12-10
6809 * journalctl gained the new options -t/--identifier= to match
6810 on the syslog identifier (aka "tag"), as well as --utc to
6811 show log timestamps in the UTC timezone. journalctl now also
6812 accepts -n/--lines=all to disable line capping in a pager.
6814 * journalctl gained a new switch, --flush, that synchronously
6815 flushes logs from /run/log/journal to /var/log/journal if
6816 persistent storage is enabled. systemd-journal-flush.service
6817 now waits until the operation is complete.
6819 * Services can notify the manager before they start a reload
6820 (by sending RELOADING=1) or shutdown (by sending
6821 STOPPING=1). This allows the manager to track and show the
6822 internal state of daemons and closes a race condition when
6823 the process is still running but has closed its D-Bus
6826 * Services with Type=oneshot do not have to have any ExecStart
6829 * User units are now loaded also from
6830 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
6831 /run/systemd/user directory that was already previously
6832 supported, but is under the control of the user.
6834 * Job timeouts (i.e. timeouts on the time a job that is
6835 queued stays in the run queue) can now optionally result in
6836 immediate reboot or power-off actions (JobTimeoutAction= and
6837 JobTimeoutRebootArgument=). This is useful on ".target"
6838 units, to limit the maximum time a target remains
6839 undispatched in the run queue, and to trigger an emergency
6840 operation in such a case. This is now used by default to
6841 turn off the system if boot-up (as defined by everything in
6842 basic.target) hangs and does not complete for at least
6843 15min. Also, if power-off or reboot hang for at least 30min
6844 an immediate power-off/reboot operation is triggered. This
6845 functionality is particularly useful to increase reliability
6846 on embedded devices, but also on laptops which might
6847 accidentally get powered on when carried in a backpack and
6848 whose boot stays stuck in a hard disk encryption passphrase
6851 * systemd-logind can be configured to also handle lid switch
6852 events even when the machine is docked or multiple displays
6853 are attached (HandleLidSwitchDocked= option).
6855 * A helper binary and a service have been added which can be
6856 used to resume from hibernation in the initramfs. A
6857 generator will parse the resume= option on the kernel
6858 command line to trigger resume.
6860 * A user console daemon systemd-consoled has been
6861 added. Currently, it is a preview, and will so far open a
6862 single terminal on each session of the user marked as
6863 Desktop=systemd-console.
6865 * Route metrics can be specified for DHCP routes added by
6868 * The SELinux context of socket-activated services can be set
6869 from the information provided by the networking stack
6870 (SELinuxContextFromNet= option).
6872 * Userspace firmware loading support has been removed and
6873 the minimum supported kernel version is thus bumped to 3.7.
6875 * Timeout for udev workers has been increased from 1 to 3
6876 minutes, but a warning will be printed after 1 minute to
6877 help diagnose kernel modules that take a long time to load.
6879 * Udev rules can now remove tags on devices with TAG-="foobar".
6881 * systemd's readahead implementation has been removed. In many
6882 circumstances it didn't give expected benefits even for
6883 rotational disk drives and was becoming less relevant in the
6884 age of SSDs. As none of the developers has been using
6885 rotating media anymore, and nobody stepped up to actively
6886 maintain this component of systemd it has now been removed.
6888 * Swap units can use Options= to specify discard options.
6889 Discard options specified for swaps in /etc/fstab are now
6892 * Docker containers are now detected as a separate type of
6895 * The Password Agent protocol gained support for queries where
6896 the user input is shown, useful e.g. for user names.
6897 systemd-ask-password gained a new --echo option to turn that
6900 * The default sysctl.d/ snippets will now set:
6902 net.core.default_qdisc = fq_codel
6904 This selects Fair Queuing Controlled Delay as the default
6905 queuing discipline for network interfaces. fq_codel helps
6906 fight the network bufferbloat problem. It is believed to be
6907 a good default with no tuning required for most workloads.
6908 Downstream distributions may override this choice. On 10Gbit
6909 servers that do not do forwarding, "fq" may perform better.
6910 Systems without a good clocksource should use "pfifo_fast".
6912 * If kdbus is enabled during build a new option BusPolicy= is
6913 available for service units, that allows locking all service
6914 processes into a stricter bus policy, in order to limit
6915 access to various bus services, or even hide most of them
6916 from the service's view entirely.
6918 * networkctl will now show the .network and .link file
6919 networkd has applied to a specific interface.
6921 * sd-login gained a new API call sd_session_get_desktop() to
6922 query which desktop environment has been selected for a
6925 * UNIX utmp support is now compile-time optional to support
6926 legacy-free systems.
6928 * systemctl gained two new commands "add-wants" and
6929 "add-requires" for pulling in units from specific targets
6932 * If the word "rescue" is specified on the kernel command line
6933 the system will now boot into rescue mode (aka
6934 rescue.target), which was previously available only by
6935 specifying "1" or "systemd.unit=rescue.target" on the kernel
6936 command line. This new kernel command line option nicely
6937 mirrors the already existing "emergency" kernel command line
6940 * New kernel command line options mount.usr=, mount.usrflags=,
6941 mount.usrfstype= have been added that match root=, rootflags=,
6942 rootfstype= but allow mounting a specific file system to
6945 * The $NOTIFY_SOCKET is now also passed to control processes of
6946 services, not only the main process.
6948 * This version reenables support for fsck's -l switch. This
6949 means at least version v2.25 of util-linux is required for
6950 operation, otherwise dead-locks on device nodes may
6951 occur. Again: you need to update util-linux to at least
6952 v2.25 when updating systemd to v217.
6954 * The "multi-seat-x" tool has been removed from systemd, as
6955 its functionality has been integrated into X servers 1.16,
6956 and the tool is hence redundant. It is recommended to update
6957 display managers invoking this tool to simply invoke X
6958 directly from now on, again.
6960 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
6961 message flag has been added for all of systemd's polkit
6962 authenticated method calls has been added. In particular this
6963 now allows optional interactive authorization via polkit for
6964 many of PID1's privileged operations such as unit file
6965 enabling and disabling.
6967 * "udevadm hwdb --update" learnt a new switch "--usr" for
6968 placing the rebuilt hardware database in /usr instead of
6969 /etc. When used only hardware database entries stored in
6970 /usr will be used, and any user database entries in /etc are
6971 ignored. This functionality is useful for vendors to ship a
6972 pre-built database on systems where local configuration is
6973 unnecessary or unlikely.
6975 * Calendar time specifications in .timer units now also
6976 understand the strings "semi-annually", "quarterly" and
6977 "minutely" as shortcuts (in addition to the preexisting
6978 "annually", "hourly", ...).
6980 * systemd-tmpfiles will now correctly create files in /dev
6981 at boot which are marked for creation only at boot. It is
6982 recommended to always create static device nodes with 'c!'
6983 and 'b!', so that they are created only at boot and not
6984 overwritten at runtime.
6986 * When the watchdog logic is used for a service (WatchdogSec=)
6987 and the watchdog timeout is hit the service will now be
6988 terminated with SIGABRT (instead of just SIGTERM), in order
6989 to make sure a proper coredump and backtrace is
6990 generated. This ensures that hanging services will result in
6991 similar coredump/backtrace behaviour as services that hit a
6994 Contributions from: Andreas Henriksson, Andrei Borzenkov,
6995 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
6996 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
6997 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
6998 Herrmann, David Sommerseth, David Strauss, Emil Renner
6999 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
7000 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
7001 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
7002 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
7003 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
7004 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
7005 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
7006 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
7007 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
7008 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
7009 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
7010 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
7011 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
7012 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
7013 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
7014 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
7017 — Berlin, 2014-10-28
7021 * timedated no longer reads NTP implementation unit names from
7022 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
7023 implementations should add a
7025 Conflicts=systemd-timesyncd.service
7027 to their unit files to take over and replace systemd's NTP
7028 default functionality.
7030 * systemd-sysusers gained a new line type "r" for configuring
7031 which UID/GID ranges to allocate system users/groups
7032 from. Lines of type "u" may now add an additional column
7033 that specifies the home directory for the system user to be
7034 created. Also, systemd-sysusers may now optionally read user
7035 information from STDIN instead of a file. This is useful for
7036 invoking it from RPM preinst scriptlets that need to create
7037 users before the first RPM file is installed since these
7038 files might need to be owned by them. A new
7039 %sysusers_create_inline RPM macro has been introduced to do
7040 just that. systemd-sysusers now updates the shadow files as
7041 well as the user/group databases, which should enhance
7042 compatibility with certain tools like grpck.
7044 * A number of bus APIs of PID 1 now optionally consult polkit to
7045 permit access for otherwise unprivileged clients under certain
7046 conditions. Note that this currently doesn't support
7047 interactive authentication yet, but this is expected to be
7048 added eventually, too.
7050 * /etc/machine-info now has new fields for configuring the
7051 deployment environment of the machine, as well as the
7052 location of the machine. hostnamectl has been updated with
7053 new command to update these fields.
7055 * systemd-timesyncd has been updated to automatically acquire
7056 NTP server information from systemd-networkd, which might
7057 have been discovered via DHCP.
7059 * systemd-resolved now includes a caching DNS stub resolver
7060 and a complete LLMNR name resolution implementation. A new
7061 NSS module "nss-resolve" has been added which can be used
7062 instead of glibc's own "nss-dns" to resolve hostnames via
7063 systemd-resolved. Hostnames, addresses and arbitrary RRs may
7064 be resolved via systemd-resolved D-Bus APIs. In contrast to
7065 the glibc internal resolver systemd-resolved is aware of
7066 multi-homed system, and keeps DNS server and caches separate
7067 and per-interface. Queries are sent simultaneously on all
7068 interfaces that have DNS servers configured, in order to
7069 properly handle VPNs and local LANs which might resolve
7070 separate sets of domain names. systemd-resolved may acquire
7071 DNS server information from systemd-networkd automatically,
7072 which in turn might have discovered them via DHCP. A tool
7073 "systemd-resolve-host" has been added that may be used to
7074 query the DNS logic in resolved. systemd-resolved implements
7075 IDNA and automatically uses IDNA or UTF-8 encoding depending
7076 on whether classic DNS or LLMNR is used as transport. In the
7077 next releases we intend to add a DNSSEC and mDNS/DNS-SD
7078 implementation to systemd-resolved.
7080 * A new NSS module nss-mymachines has been added, that
7081 automatically resolves the names of all local registered
7082 containers to their respective IP addresses.
7084 * A new client tool "networkctl" for systemd-networkd has been
7085 added. It currently is entirely passive and will query
7086 networking configuration from udev, rtnetlink and networkd,
7087 and present it to the user in a very friendly
7088 way. Eventually, we hope to extend it to become a full
7089 control utility for networkd.
7091 * .socket units gained a new DeferAcceptSec= setting that
7092 controls the kernels' TCP_DEFER_ACCEPT sockopt for
7093 TCP. Similarly, support for controlling TCP keep-alive
7094 settings has been added (KeepAliveTimeSec=,
7095 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
7096 turning off Nagle's algorithm on TCP has been added
7099 * logind learned a new session type "web", for use in projects
7100 like Cockpit which register web clients as PAM sessions.
7102 * timer units with at least one OnCalendar= setting will now
7103 be started only after time-sync.target has been
7104 reached. This way they will not elapse before the system
7105 clock has been corrected by a local NTP client or
7106 similar. This is particular useful on RTC-less embedded
7107 machines, that come up with an invalid system clock.
7109 * systemd-nspawn's --network-veth= switch should now result in
7110 stable MAC addresses for both the outer and the inner side
7113 * systemd-nspawn gained a new --volatile= switch for running
7114 container instances with /etc or /var unpopulated.
7116 * The kdbus client code has been updated to use the new Linux
7117 3.17 memfd subsystem instead of the old kdbus-specific one.
7119 * systemd-networkd's DHCP client and server now support
7120 FORCERENEW. There are also new configuration options to
7121 configure the vendor client identifier and broadcast mode
7124 * systemd will no longer inform the kernel about the current
7125 timezone, as this is necessarily incorrect and racy as the
7126 kernel has no understanding of DST and similar
7127 concepts. This hence means FAT timestamps will be always
7128 considered UTC, similar to what Android is already
7129 doing. Also, when the RTC is configured to the local time
7130 (rather than UTC) systemd will never synchronize back to it,
7131 as this might confuse Windows at a later boot.
7133 * systemd-analyze gained a new command "verify" for offline
7134 validation of unit files.
7136 * systemd-networkd gained support for a couple of additional
7137 settings for bonding networking setups. Also, the metric for
7138 statically configured routes may now be configured. For
7139 network interfaces where this is appropriate the peer IP
7140 address may now be configured.
7142 * systemd-networkd's DHCP client will no longer request
7143 broadcasting by default, as this tripped up some networks.
7144 For hardware where broadcast is required the feature should
7145 be switched back on using RequestBroadcast=yes.
7147 * systemd-networkd will now set up IPv4LL addresses (when
7148 enabled) even if DHCP is configured successfully.
7150 * udev will now default to respect network device names given
7151 by the kernel when the kernel indicates that these are
7152 predictable. This behavior can be tweaked by changing
7153 NamePolicy= in the relevant .link file.
7155 * A new library systemd-terminal has been added that
7156 implements full TTY stream parsing and rendering. This
7157 library is supposed to be used later on for implementing a
7158 full userspace VT subsystem, replacing the current kernel
7161 * A new tool systemd-journal-upload has been added to push
7162 journal data to a remote system running
7163 systemd-journal-remote.
7165 * journald will no longer forward all local data to another
7166 running syslog daemon. This change has been made because
7167 rsyslog (which appears to be the most commonly used syslog
7168 implementation these days) no longer makes use of this, and
7169 instead pulls the data out of the journal on its own. Since
7170 forwarding the messages to a non-existent syslog server is
7171 more expensive than we assumed we have now turned this
7172 off. If you run a syslog server that is not a recent rsyslog
7173 version, you have to turn this option on again
7174 (ForwardToSyslog= in journald.conf).
7176 * journald now optionally supports the LZ4 compressor for
7177 larger journal fields. This compressor should perform much
7178 better than XZ which was the previous default.
7180 * machinectl now shows the IP addresses of local containers,
7181 if it knows them, plus the interface name of the container.
7183 * A new tool "systemd-escape" has been added that makes it
7184 easy to escape strings to build unit names and similar.
7186 * sd_notify() messages may now include a new ERRNO= field
7187 which is parsed and collected by systemd and shown among the
7188 "systemctl status" output for a service.
7190 * A new component "systemd-firstboot" has been added that
7191 queries the most basic systemd information (timezone,
7192 hostname, root password) interactively on first
7193 boot. Alternatively it may also be used to provision these
7194 things offline on OS images installed into directories.
7196 * The default sysctl.d/ snippets will now set
7198 net.ipv4.conf.default.promote_secondaries=1
7200 This has the benefit of no flushing secondary IP addresses
7201 when primary addresses are removed.
7203 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
7204 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
7205 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
7206 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
7207 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
7208 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
7209 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
7210 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
7211 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
7212 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
7213 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
7214 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
7215 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
7216 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
7217 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
7219 — Berlin, 2014-08-19
7223 * A new tool systemd-sysusers has been added. This tool
7224 creates system users and groups in /etc/passwd and
7225 /etc/group, based on static declarative system user/group
7226 definitions in /usr/lib/sysusers.d/. This is useful to
7227 enable factory resets and volatile systems that boot up with
7228 an empty /etc directory, and thus need system users and
7229 groups created during early boot. systemd now also ships
7230 with two default sysusers.d/ files for the most basic
7231 users and groups systemd and the core operating system
7234 * A new tmpfiles snippet has been added that rebuilds the
7235 essential files in /etc on boot, should they be missing.
7237 * A directive for ensuring automatic clean-up of
7238 /var/cache/man/ has been removed from the default
7239 configuration. This line should now be shipped by the man
7240 implementation. The necessary change has been made to the
7241 man-db implementation. Note that you need to update your man
7242 implementation to one that ships this line, otherwise no
7243 automatic clean-up of /var/cache/man will take place.
7245 * A new condition ConditionNeedsUpdate= has been added that
7246 may conditionalize services to only run when /etc or /var
7247 are "older" than the vendor operating system resources in
7248 /usr. This is useful for reconstructing or updating /etc
7249 after an offline update of /usr or a factory reset, on the
7250 next reboot. Services that want to run once after such an
7251 update or reset should use this condition and order
7252 themselves before the new systemd-update-done.service, which
7253 will mark the two directories as fully updated. A number of
7254 service files have been added making use of this, to rebuild
7255 the udev hardware database, the journald message catalog and
7256 dynamic loader cache (ldconfig). The systemd-sysusers tool
7257 described above also makes use of this now. With this in
7258 place it is now possible to start up a minimal operating
7259 system with /etc empty cleanly. For more information on the
7260 concepts involved see this recent blog story:
7262 http://0pointer.de/blog/projects/stateless.html
7264 * A new system group "input" has been introduced, and all
7265 input device nodes get this group assigned. This is useful
7266 for system-level software to get access to input devices. It
7267 complements what is already done for "audio" and "video".
7269 * systemd-networkd learnt minimal DHCPv4 server support in
7270 addition to the existing DHCPv4 client support. It also
7271 learnt DHCPv6 client and IPv6 Router Solicitation client
7272 support. The DHCPv4 client gained support for static routes
7273 passed in from the server. Note that the [DHCPv4] section
7274 known in older systemd-networkd versions has been renamed to
7275 [DHCP] and is now also used by the DHCPv6 client. Existing
7276 .network files using settings of this section should be
7277 updated, though compatibility is maintained. Optionally, the
7278 client hostname may now be sent to the DHCP server.
7280 * networkd gained support for vxlan virtual networks as well
7281 as tun/tap and dummy devices.
7283 * networkd gained support for automatic allocation of address
7284 ranges for interfaces from a system-wide pool of
7285 addresses. This is useful for dynamically managing a large
7286 number of interfaces with a single network configuration
7287 file. In particular this is useful to easily assign
7288 appropriate IP addresses to the veth links of a large number
7289 of nspawn instances.
7291 * RPM macros for processing sysusers, sysctl and binfmt
7292 drop-in snippets at package installation time have been
7295 * The /etc/os-release file should now be placed in
7296 /usr/lib/os-release. The old location is automatically
7297 created as symlink. /usr/lib is the more appropriate
7298 location of this file, since it shall actually describe the
7299 vendor operating system shipped in /usr, and not the
7300 configuration stored in /etc.
7302 * .mount units gained a new boolean SloppyOptions= setting
7303 that maps to mount(8)'s -s option which enables permissive
7304 parsing of unknown mount options.
7306 * tmpfiles learnt a new "L+" directive which creates a symlink
7307 but (unlike "L") deletes a pre-existing file first, should
7308 it already exist and not already be the correct
7309 symlink. Similarly, "b+", "c+" and "p+" directives have been
7310 added as well, which create block and character devices, as
7311 well as fifos in the filesystem, possibly removing any
7312 pre-existing files of different types.
7314 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
7315 'argument' field (which so far specified the source to
7316 symlink/copy the files from) is now optional. If omitted the
7317 same file os copied from /usr/share/factory/ suffixed by the
7318 full destination path. This is useful for populating /etc
7319 with essential files, by copying them from vendor defaults
7320 shipped in /usr/share/factory/etc.
7322 * A new command "systemctl preset-all" has been added that
7323 applies the service preset settings to all installed unit
7324 files. A new switch --preset-mode= has been added that
7325 controls whether only enable or only disable operations
7328 * A new command "systemctl is-system-running" has been added
7329 that allows checking the overall state of the system, for
7330 example whether it is fully up and running.
7332 * When the system boots up with an empty /etc, the equivalent
7333 to "systemctl preset-all" is executed during early boot, to
7334 make sure all default services are enabled after a factory
7337 * systemd now contains a minimal preset file that enables the
7338 most basic services systemd ships by default.
7340 * Unit files' [Install] section gained a new DefaultInstance=
7341 field for defining the default instance to create if a
7342 template unit is enabled with no instance specified.
7344 * A new passive target cryptsetup-pre.target has been added
7345 that may be used by services that need to make they run and
7346 finish before the first LUKS cryptographic device is set up.
7348 * The /dev/loop-control and /dev/btrfs-control device nodes
7349 are now owned by the "disk" group by default, opening up
7350 access to this group.
7352 * systemd-coredump will now automatically generate a
7353 stack trace of all core dumps taking place on the system,
7354 based on elfutils' libdw library. This stack trace is logged
7357 * systemd-coredump may now optionally store coredumps directly
7358 on disk (in /var/lib/systemd/coredump, possibly compressed),
7359 instead of storing them unconditionally in the journal. This
7360 mode is the new default. A new configuration file
7361 /etc/systemd/coredump.conf has been added to configure this
7362 and other parameters of systemd-coredump.
7364 * coredumpctl gained a new "info" verb to show details about a
7365 specific coredump. A new switch "-1" has also been added
7366 that makes sure to only show information about the most
7367 recent entry instead of all entries. Also, as the tool is
7368 generally useful now the "systemd-" prefix of the binary
7369 name has been removed. Distributions that want to maintain
7370 compatibility with the old name should add a symlink from
7371 the old name to the new name.
7373 * journald's SplitMode= now defaults to "uid". This makes sure
7374 that unprivileged users can access their own coredumps with
7375 coredumpctl without restrictions.
7377 * New kernel command line options "systemd.wants=" (for
7378 pulling an additional unit during boot), "systemd.mask="
7379 (for masking a specific unit for the boot), and
7380 "systemd.debug-shell" (for enabling the debug shell on tty9)
7381 have been added. This is implemented in the new generator
7382 "systemd-debug-generator".
7384 * systemd-nspawn will now by default filter a couple of
7385 syscalls for containers, among them those required for
7386 kernel module loading, direct x86 IO port access, swap
7387 management, and kexec. Most importantly though
7388 open_by_handle_at() is now prohibited for containers,
7389 closing a hole similar to a recently discussed vulnerability
7390 in docker regarding access to files on file hierarchies the
7391 container should normally not have access to. Note that, for
7392 nspawn, we generally make no security claims anyway (and
7393 this is explicitly documented in the man page), so this is
7394 just a fix for one of the most obvious problems.
7396 * A new man page file-hierarchy(7) has been added that
7397 contains a minimized, modernized version of the file system
7398 layout systemd expects, similar in style to the FHS
7399 specification or hier(5). A new tool systemd-path(1) has
7400 been added to query many of these paths for the local
7403 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
7404 longer done. Since the directory now has a per-user size
7405 limit, and is cleaned on logout this appears unnecessary,
7406 in particular since this now brings the lifecycle of this
7407 directory closer in line with how IPC objects are handled.
7409 * systemd.pc now exports a number of additional directories,
7410 including $libdir (which is useful to identify the library
7411 path for the primary architecture of the system), and a
7412 couple of drop-in directories.
7414 * udev's predictable network interface names now use the dev_port
7415 sysfs attribute, introduced in linux 3.15 instead of dev_id to
7416 distinguish between ports of the same PCI function. dev_id should
7417 only be used for ports using the same HW address, hence the need
7420 * machined has been updated to export the OS version of a
7421 container (read from /etc/os-release and
7422 /usr/lib/os-release) on the bus. This is now shown in
7423 "machinectl status" for a machine.
7425 * A new service setting RestartForceExitStatus= has been
7426 added. If configured to a set of exit signals or process
7427 return values, the service will be restarted when the main
7428 daemon process exits with any of them, regardless of the
7431 * systemctl's -H switch for connecting to remote systemd
7432 machines has been extended so that it may be used to
7433 directly connect to a specific container on the
7434 host. "systemctl -H root@foobar:waldi" will now connect as
7435 user "root" to host "foobar", and then proceed directly to
7436 the container named "waldi". Note that currently you have to
7437 authenticate as user "root" for this to work, as entering
7438 containers is a privileged operation.
7440 Contributions from: Andreas Henriksson, Benjamin Steinwender,
7441 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
7442 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
7443 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
7444 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
7445 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
7446 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
7447 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
7448 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
7449 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
7450 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
7451 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
7453 — Berlin, 2014-07-03
7457 * As an experimental feature, udev now tries to lock the
7458 disk device node (flock(LOCK_SH|LOCK_NB)) while it
7459 executes events for the disk or any of its partitions.
7460 Applications like partitioning programs can lock the
7461 disk device node (flock(LOCK_EX)) and claim temporary
7462 device ownership that way; udev will entirely skip all event
7463 handling for this disk and its partitions. If the disk
7464 was opened for writing, the close will trigger a partition
7465 table rescan in udev's "watch" facility, and if needed
7466 synthesize "change" events for the disk and all its partitions.
7467 This is now unconditionally enabled, and if it turns out to
7468 cause major problems, we might turn it on only for specific
7469 devices, or might need to disable it entirely. Device Mapper
7470 devices are excluded from this logic.
7472 * We temporarily dropped the "-l" switch for fsck invocations,
7473 since they collide with the flock() logic above. util-linux
7474 upstream has been changed already to avoid this conflict,
7475 and we will re-add "-l" as soon as util-linux with this
7476 change has been released.
7478 * The dependency on libattr has been removed. Since a long
7479 time, the extended attribute calls have moved to glibc, and
7480 libattr is thus unnecessary.
7482 * Virtualization detection works without privileges now. This
7483 means the systemd-detect-virt binary no longer requires
7484 CAP_SYS_PTRACE file capabilities, and our daemons can run
7485 with fewer privileges.
7487 * systemd-networkd now runs under its own "systemd-network"
7488 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
7489 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
7490 loses the ability to write to files owned by root this way.
7492 * Similarly, systemd-resolved now runs under its own
7493 "systemd-resolve" user with no capabilities remaining.
7495 * Similarly, systemd-bus-proxyd now runs under its own
7496 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
7498 * systemd-networkd gained support for setting up "veth"
7499 virtual Ethernet devices for container connectivity, as well
7500 as GRE and VTI tunnels.
7502 * systemd-networkd will no longer automatically attempt to
7503 manually load kernel modules necessary for certain tunnel
7504 transports. Instead, it is assumed the kernel loads them
7505 automatically when required. This only works correctly on
7506 very new kernels. On older kernels, please consider adding
7507 the kernel modules to /etc/modules-load.d/ as a work-around.
7509 * The resolv.conf file systemd-resolved generates has been
7510 moved to /run/systemd/resolve/. If you have a symlink from
7511 /etc/resolv.conf, it might be necessary to correct it.
7513 * Two new service settings, ProtectHome= and ProtectSystem=,
7514 have been added. When enabled, they will make the user data
7515 (such as /home) inaccessible or read-only and the system
7516 (such as /usr) read-only, for specific services. This allows
7517 very light-weight per-service sandboxing to avoid
7518 modifications of user data or system files from
7519 services. These two new switches have been enabled for all
7520 of systemd's long-running services, where appropriate.
7522 * Socket units gained new SocketUser= and SocketGroup=
7523 settings to set the owner user and group of AF_UNIX sockets
7524 and FIFOs in the file system.
7526 * Socket units gained a new RemoveOnStop= setting. If enabled,
7527 all FIFOS and sockets in the file system will be removed
7528 when the specific socket unit is stopped.
7530 * Socket units gained a new Symlinks= setting. It takes a list
7531 of symlinks to create to file system sockets or FIFOs
7532 created by the specific Unix sockets. This is useful to
7533 manage symlinks to socket nodes with the same lifecycle as
7536 * The /dev/log socket and /dev/initctl FIFO have been moved to
7537 /run, and have been replaced by symlinks. This allows
7538 connecting to these facilities even if PrivateDevices=yes is
7539 used for a service (which makes /dev/log itself unavailable,
7540 but /run is left). This also has the benefit of ensuring
7541 that /dev only contains device nodes, directories and
7542 symlinks, and nothing else.
7544 * sd-daemon gained two new calls sd_pid_notify() and
7545 sd_pid_notifyf(). They are similar to sd_notify() and
7546 sd_notifyf(), but allow overriding of the source PID of
7547 notification messages if permissions permit this. This is
7548 useful to send notify messages on behalf of a different
7549 process (for example, the parent process). The
7550 systemd-notify tool has been updated to make use of this
7551 when sending messages (so that notification messages now
7552 originate from the shell script invoking systemd-notify and
7553 not the systemd-notify process itself. This should minimize
7554 a race where systemd fails to associate notification
7555 messages to services when the originating process already
7558 * A new "on-abnormal" setting for Restart= has been added. If
7559 set, it will result in automatic restarts on all "abnormal"
7560 reasons for a process to exit, which includes unclean
7561 signals, core dumps, timeouts and watchdog timeouts, but
7562 does not include clean and unclean exit codes or clean
7563 signals. Restart=on-abnormal is an alternative for
7564 Restart=on-failure for services that shall be able to
7565 terminate and avoid restarts on certain errors, by
7566 indicating so with an unclean exit code. Restart=on-failure
7567 or Restart=on-abnormal is now the recommended setting for
7568 all long-running services.
7570 * If the InaccessibleDirectories= service setting points to a
7571 mount point (or if there are any submounts contained within
7572 it), it is now attempted to completely unmount it, to make
7573 the file systems truly unavailable for the respective
7576 * The ReadOnlyDirectories= service setting and
7577 systemd-nspawn's --read-only parameter are now recursively
7578 applied to all submounts, too.
7580 * Mount units may now be created transiently via the bus APIs.
7582 * The support for SysV and LSB init scripts has been removed
7583 from the systemd daemon itself. Instead, it is now
7584 implemented as a generator that creates native systemd units
7585 from these scripts when needed. This enables us to remove a
7586 substantial amount of legacy code from PID 1, following the
7587 fact that many distributions only ship a very small number
7588 of LSB/SysV init scripts nowadays.
7590 * Privileged Xen (dom0) domains are not considered
7591 virtualization anymore by the virtualization detection
7592 logic. After all, they generally have unrestricted access to
7593 the hardware and usually are used to manage the unprivileged
7596 * systemd-tmpfiles gained a new "C" line type, for copying
7597 files or entire directories.
7599 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
7600 lines. So far, they have been non-globbing versions of the
7601 latter, and have thus been redundant. In future, it is
7602 recommended to only use "z". "m" has hence been removed
7603 from the documentation, even though it stays supported.
7605 * A tmpfiles snippet to recreate the most basic structure in
7606 /var has been added. This is enough to create the /var/run →
7607 /run symlink and create a couple of structural
7608 directories. This allows systems to boot up with an empty or
7609 volatile /var. Of course, while with this change, the core OS
7610 now is capable with dealing with a volatile /var, not all
7611 user services are ready for it. However, we hope that sooner
7612 or later, many service daemons will be changed upstream so
7613 that they are able to automatically create their necessary
7614 directories in /var at boot, should they be missing. This is
7615 the first step to allow state-less systems that only require
7616 the vendor image for /usr to boot.
7618 * systemd-nspawn has gained a new --tmpfs= switch to mount an
7619 empty tmpfs instance to a specific directory. This is
7620 particularly useful for making use of the automatic
7621 reconstruction of /var (see above), by passing --tmpfs=/var.
7623 * Access modes specified in tmpfiles snippets may now be
7624 prefixed with "~", which indicates that they shall be masked
7625 by whether the existing file or directory is currently
7626 writable, readable or executable at all. Also, if specified,
7627 the sgid/suid/sticky bits will be masked for all
7630 * A new passive target unit "network-pre.target" has been
7631 added which is useful for services that shall run before any
7632 network is configured, for example firewall scripts.
7634 * The "floppy" group that previously owned the /dev/fd*
7635 devices is no longer used. The "disk" group is now used
7636 instead. Distributions should probably deprecate usage of
7639 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
7640 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
7641 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
7642 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
7643 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
7644 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
7647 — Berlin, 2014-06-11
7651 * A new "systemd-timesyncd" daemon has been added for
7652 synchronizing the system clock across the network. It
7653 implements an SNTP client. In contrast to NTP
7654 implementations such as chrony or the NTP reference server,
7655 this only implements a client side, and does not bother with
7656 the full NTP complexity, focusing only on querying time from
7657 one remote server and synchronizing the local clock to
7658 it. Unless you intend to serve NTP to networked clients or
7659 want to connect to local hardware clocks, this simple NTP
7660 client should be more than appropriate for most
7661 installations. The daemon runs with minimal privileges, and
7662 has been hooked up with networkd to only operate when
7663 network connectivity is available. The daemon saves the
7664 current clock to disk every time a new NTP sync has been
7665 acquired, and uses this to possibly correct the system clock
7666 early at bootup, in order to accommodate for systems that
7667 lack an RTC such as the Raspberry Pi and embedded devices,
7668 and to make sure that time monotonically progresses on these
7669 systems, even if it is not always correct. To make use of
7670 this daemon, a new system user and group "systemd-timesync"
7671 needs to be created on installation of systemd.
7673 * The queue "seqnum" interface of libudev has been disabled, as
7674 it was generally incompatible with device namespacing as
7675 sequence numbers of devices go "missing" if the devices are
7676 part of a different namespace.
7678 * "systemctl list-timers" and "systemctl list-sockets" gained
7679 a --recursive switch for showing units of these types also
7680 for all local containers, similar in style to the already
7681 supported --recursive switch for "systemctl list-units".
7683 * A new RebootArgument= setting has been added for service
7684 units, which may be used to specify a kernel reboot argument
7685 to use when triggering reboots with StartLimitAction=.
7687 * A new FailureAction= setting has been added for service
7688 units which may be used to specify an operation to trigger
7689 when a service fails. This works similarly to
7690 StartLimitAction=, but unlike it, controls what is done
7691 immediately rather than only after several attempts to
7692 restart the service in question.
7694 * hostnamed got updated to also expose the kernel name,
7695 release, and version on the bus. This is useful for
7696 executing commands like hostnamectl with the -H switch.
7697 systemd-analyze makes use of this to properly display
7698 details when running non-locally.
7700 * The bootchart tool can now show cgroup information in the
7701 graphs it generates.
7703 * The CFS CPU quota cgroup attribute is now exposed for
7704 services. The new CPUQuota= switch has been added for this
7705 which takes a percentage value. Setting this will have the
7706 result that a service may never get more CPU time than the
7707 specified percentage, even if the machine is otherwise idle.
7709 * systemd-networkd learned IPIP and SIT tunnel support.
7711 * LSB init scripts exposing a dependency on $network will now
7712 get a dependency on network-online.target rather than simply
7713 network.target. This should bring LSB handling closer to
7714 what it was on SysV systems.
7716 * A new fsck.repair= kernel option has been added to control
7717 how fsck shall deal with unclean file systems at boot.
7719 * The (.ini) configuration file parser will now silently ignore
7720 sections whose names begin with "X-". This may be used to maintain
7721 application-specific extension sections in unit files.
7723 * machined gained a new API to query the IP addresses of
7724 registered containers. "machinectl status" has been updated
7725 to show these addresses in its output.
7727 * A new call sd_uid_get_display() has been added to the
7728 sd-login APIs for querying the "primary" session of a
7729 user. The "primary" session of the user is elected from the
7730 user's sessions and generally a graphical session is
7731 preferred over a text one.
7733 * A minimal systemd-resolved daemon has been added. It
7734 currently simply acts as a companion to systemd-networkd and
7735 manages resolv.conf based on per-interface DNS
7736 configuration, possibly supplied via DHCP. In the long run
7737 we hope to extend this into a local DNSSEC enabled DNS and
7740 * The systemd-networkd-wait-online tool is now enabled by
7741 default. It will delay network-online.target until a network
7742 connection has been configured. The tool primarily integrates
7743 with networkd, but will also make a best effort to make sense
7744 of network configuration performed in some other way.
7746 * Two new service options StartupCPUShares= and
7747 StartupBlockIOWeight= have been added that work similarly to
7748 CPUShares= and BlockIOWeight= however only apply during
7749 system startup. This is useful to prioritize certain services
7750 differently during bootup than during normal runtime.
7752 * hostnamed has been changed to prefer the statically
7753 configured hostname in /etc/hostname (unless set to
7754 'localhost' or empty) over any dynamic one supplied by
7755 dhcp. With this change, the rules for picking the hostname
7756 match more closely the rules of other configuration settings
7757 where the local administrator's configuration in /etc always
7758 overrides any other settings.
7760 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
7761 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
7762 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
7763 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
7764 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
7765 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
7766 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
7767 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
7768 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
7769 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
7770 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
7771 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
7772 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
7773 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
7774 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
7775 Lindskog, WaLyong Cho, Will Woods, Zbigniew
7778 — Beijing, 2014-05-28
7782 * When restoring the screen brightness at boot, stay away from
7783 the darkest setting or from the lowest 5% of the available
7784 range, depending on which is the larger value of both. This
7785 should effectively protect the user from rebooting into a
7786 black screen, should the brightness have been set to minimum
7789 * sd-login gained a new sd_machine_get_class() call to
7790 determine the class ("vm" or "container") of a machine
7791 registered with machined.
7793 * sd-login gained new calls
7794 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
7795 to query the identity of the peer of a local AF_UNIX
7796 connection. They operate similarly to their sd_pid_get_xyz()
7799 * PID 1 will now maintain a system-wide system state engine
7800 with the states "starting", "running", "degraded",
7801 "maintenance", "stopping". These states are bound to system
7802 startup, normal runtime, runtime with at least one failed
7803 service, rescue/emergency mode and system shutdown. This
7804 state is shown in the "systemctl status" output when no unit
7805 name is passed. It is useful to determine system state, in
7806 particularly when doing so for many systems or containers at
7809 * A new command "list-machines" has been added to "systemctl"
7810 that lists all local OS containers and shows their system
7811 state (see above), if systemd runs inside of them.
7813 * systemctl gained a new "-r" switch to recursively enumerate
7814 units on all local containers, when used with the
7815 "list-unit" command (which is the default one that is
7816 executed when no parameters are specified).
7818 * The GPT automatic partition discovery logic will now honour
7819 two GPT partition flags: one may be set on a partition to
7820 cause it to be mounted read-only, and the other may be set
7821 on a partition to ignore it during automatic discovery.
7823 * Two new GPT type UUIDs have been added for automatic root
7824 partition discovery, for 32-bit and 64-bit ARM. This is not
7825 particularly useful for discovering the root directory on
7826 these architectures during bare-metal boots (since UEFI is
7827 not common there), but still very useful to allow booting of
7828 ARM disk images in nspawn with the -i option.
7830 * MAC addresses of interfaces created with nspawn's
7831 --network-interface= switch will now be generated from the
7832 machine name, and thus be stable between multiple invocations
7835 * logind will now automatically remove all IPC objects owned
7836 by a user if she or he fully logs out. This makes sure that
7837 users who are logged out cannot continue to consume IPC
7838 resources. This covers SysV memory, semaphores and message
7839 queues as well as POSIX shared memory and message
7840 queues. Traditionally, SysV and POSIX IPC had no lifecycle
7841 limits. With this functionality, that is corrected. This may
7842 be turned off by using the RemoveIPC= switch of logind.conf.
7844 * The systemd-machine-id-setup and tmpfiles tools gained a
7845 --root= switch to operate on a specific root directory,
7848 * journald can now forward logged messages to the TTYs of all
7849 logged in users ("wall"). This is the default for all
7850 emergency messages now.
7852 * A new tool systemd-journal-remote has been added to stream
7853 journal log messages across the network.
7855 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
7856 controller trees are mounted into it. Note that the
7857 directories mounted beneath it are not read-only. This is a
7858 security measure and is particularly useful because glibc
7859 actually includes a search logic to pick any tmpfs it can
7860 find to implement shm_open() if /dev/shm is not available
7861 (which it might very well be in namespaced setups).
7863 * machinectl gained a new "poweroff" command to cleanly power
7864 down a local OS container.
7866 * The PrivateDevices= unit file setting will now also drop the
7867 CAP_MKNOD capability from the capability bound set, and
7868 imply DevicePolicy=closed.
7870 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
7871 comprehensively on all long-running systemd services where
7872 this is appropriate.
7874 * systemd-udevd will now run in a disassociated mount
7875 namespace. To mount directories from udev rules, make sure to
7876 pull in mount units via SYSTEMD_WANTS properties.
7878 * The kdbus support gained support for uploading policy into
7879 the kernel. sd-bus gained support for creating "monitoring"
7880 connections that can eavesdrop into all bus communication
7881 for debugging purposes.
7883 * Timestamps may now be specified in seconds since the UNIX
7884 epoch Jan 1st, 1970 by specifying "@" followed by the value
7887 * Native tcpwrap support in systemd has been removed. tcpwrap
7888 is old code, not really maintained anymore and has serious
7889 shortcomings, and better options such as firewalls
7890 exist. For setups that require tcpwrap usage, please
7891 consider invoking your socket-activated service via tcpd,
7892 like on traditional inetd.
7894 * A new system.conf configuration option
7895 DefaultTimerAccuracySec= has been added that controls the
7896 default AccuracySec= setting of .timer units.
7898 * Timer units gained a new WakeSystem= switch. If enabled,
7899 timers configured this way will cause the system to resume
7900 from system suspend (if the system supports that, which most
7903 * Timer units gained a new Persistent= switch. If enabled,
7904 timers configured this way will save to disk when they have
7905 been last triggered. This information is then used on next
7906 reboot to possible execute overdue timer events, that
7907 could not take place because the system was powered off.
7908 This enables simple anacron-like behaviour for timer units.
7910 * systemctl's "list-timers" will now also list the time a
7911 timer unit was last triggered in addition to the next time
7912 it will be triggered.
7914 * systemd-networkd will now assign predictable IPv4LL
7915 addresses to its local interfaces.
7917 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
7918 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
7919 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
7920 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
7921 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
7922 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
7923 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
7924 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
7927 — Berlin, 2014-03-25
7931 * A new unit file setting RestrictAddressFamilies= has been
7932 added to restrict which socket address families unit
7933 processes gain access to. This takes address family names
7934 like "AF_INET" or "AF_UNIX", and is useful to minimize the
7935 attack surface of services via exotic protocol stacks. This
7936 is built on seccomp system call filters.
7938 * Two new unit file settings RuntimeDirectory= and
7939 RuntimeDirectoryMode= have been added that may be used to
7940 manage a per-daemon runtime directories below /run. This is
7941 an alternative for setting up directory permissions with
7942 tmpfiles snippets, and has the advantage that the runtime
7943 directory's lifetime is bound to the daemon runtime and that
7944 the daemon starts up with an empty directory each time. This
7945 is particularly useful when writing services that drop
7946 privileges using the User= or Group= setting.
7948 * The DeviceAllow= unit setting now supports globbing for
7949 matching against device group names.
7951 * The systemd configuration file system.conf gained new
7952 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
7953 DefaultMemoryAccounting= to globally turn on/off accounting
7954 for specific resources (cgroups) for all units. These
7955 settings may still be overridden individually in each unit
7958 * systemd-gpt-auto-generator is now able to discover /srv and
7959 root partitions in addition to /home and swap partitions. It
7960 also supports LUKS-encrypted partitions now. With this in
7961 place, automatic discovery of partitions to mount following
7962 the Discoverable Partitions Specification
7963 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
7964 is now a lot more complete. This allows booting without
7965 /etc/fstab and without root= on the kernel command line on
7966 systems prepared appropriately.
7968 * systemd-nspawn gained a new --image= switch which allows
7969 booting up disk images and Linux installations on any block
7970 device that follow the Discoverable Partitions Specification
7971 (see above). This means that installations made with
7972 appropriately updated installers may now be started and
7973 deployed using container managers, completely
7974 unmodified. (We hope that libvirt-lxc will add support for
7975 this feature soon, too.)
7977 * systemd-nspawn gained a new --network-macvlan= setting to
7978 set up a private macvlan interface for the
7979 container. Similarly, systemd-networkd gained a new
7980 Kind=macvlan setting in .netdev files.
7982 * systemd-networkd now supports configuring local addresses
7985 * A new tool systemd-network-wait-online has been added to
7986 synchronously wait for network connectivity using
7989 * The sd-bus.h bus API gained a new sd_bus_track object for
7990 tracking the lifecycle of bus peers. Note that sd-bus.h is
7991 still not a public API though (unless you specify
7992 --enable-kdbus on the configure command line, which however
7993 voids your warranty and you get no API stability guarantee).
7995 * The $XDG_RUNTIME_DIR runtime directories for each user are
7996 now individual tmpfs instances, which has the benefit of
7997 introducing separate pools for each user, with individual
7998 size limits, and thus making sure that unprivileged clients
7999 can no longer negatively impact the system or other users by
8000 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
8001 RuntimeDirectorySize= has been introduced that allows
8002 controlling the default size limit for all users. It
8003 defaults to 10% of the available physical memory. This is no
8004 replacement for quotas on tmpfs though (which the kernel
8005 still does not support), as /dev/shm and /tmp are still
8006 shared resources used by both the system and unprivileged
8009 * logind will now automatically turn off automatic suspending
8010 on laptop lid close when more than one display is
8011 connected. This was previously expected to be implemented
8012 individually in desktop environments (such as GNOME),
8013 however has been added to logind now, in order to fix a
8014 boot-time race where a desktop environment might not have
8015 been started yet and thus not been able to take an inhibitor
8016 lock at the time where logind already suspends the system
8017 due to a closed lid.
8019 * logind will now wait at least 30s after each system
8020 suspend/resume cycle, and 3min after system boot before
8021 suspending the system due to a closed laptop lid. This
8022 should give USB docking stations and similar enough time to
8023 be probed and configured after system resume and boot in
8024 order to then act as suspend blocker.
8026 * systemd-run gained a new --property= setting which allows
8027 initialization of resource control properties (and others)
8028 for the created scope or service unit. Example: "systemd-run
8029 --property=BlockIOWeight=10 updatedb" may be used to run
8030 updatedb at a low block IO scheduling weight.
8032 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
8033 now also work in --scope mode.
8035 * When systemd is compiled with kdbus support, basic support
8036 for enforced policies is now in place. (Note that enabling
8037 kdbus still voids your warranty and no API compatibility
8040 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
8041 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
8042 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
8043 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
8044 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
8045 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
8046 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
8047 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
8048 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
8049 Zbigniew Jędrzejewski-Szmek
8051 — Berlin, 2014-03-12
8055 * systemd will now relabel /dev after loading the SMACK policy
8056 according to SMACK rules.
8058 * A new unit file option AppArmorProfile= has been added to
8059 set the AppArmor profile for the processes of a unit.
8061 * A new condition check ConditionArchitecture= has been added
8062 to conditionalize units based on the system architecture, as
8063 reported by uname()'s "machine" field.
8065 * systemd-networkd now supports matching on the system
8066 virtualization, architecture, kernel command line, hostname
8069 * logind is now a lot more aggressive when suspending the
8070 machine due to a closed laptop lid. Instead of acting only
8071 on the lid close action, it will continuously watch the lid
8072 status and act on it. This is useful for laptops where the
8073 power button is on the outside of the chassis so that it can
8074 be reached without opening the lid (such as the Lenovo
8075 Yoga). On those machines, logind will now immediately
8076 re-suspend the machine if the power button has been
8077 accidentally pressed while the laptop was suspended and in a
8078 backpack or similar.
8080 * logind will now watch SW_DOCK switches and inhibit reaction
8081 to the lid switch if it is pressed. This means that logind
8082 will not suspend the machine anymore if the lid is closed
8083 and the system is docked, if the laptop supports SW_DOCK
8084 notifications via the input layer. Note that ACPI docking
8085 stations do not generate this currently. Also note that this
8086 logic is usually not fully sufficient and Desktop
8087 Environments should take a lid switch inhibitor lock when an
8088 external display is connected, as systemd will not watch
8091 * nspawn will now make use of the devices cgroup controller by
8092 default, and only permit creation of and access to the usual
8093 API device nodes like /dev/null or /dev/random, as well as
8094 access to (but not creation of) the pty devices.
8096 * We will now ship a default .network file for
8097 systemd-networkd that automatically configures DHCP for
8098 network interfaces created by nspawn's --network-veth or
8099 --network-bridge= switches.
8101 * systemd will now understand the usual M, K, G, T suffixes
8102 according to SI conventions (i.e. to the base 1000) when
8103 referring to throughput and hardware metrics. It will stay
8104 with IEC conventions (i.e. to the base 1024) for software
8105 metrics, according to what is customary according to
8106 Wikipedia. We explicitly document which base applies for
8107 each configuration option.
8109 * The DeviceAllow= setting in unit files now supports a syntax to
8110 allow-list an entire group of devices node majors at once, based on
8111 the /proc/devices listing. For example, with the string "char-pts",
8112 it is now possible to allow-list all current and future pseudo-TTYs
8115 * sd-event learned a new "post" event source. Event sources of
8116 this type are triggered by the dispatching of any event
8117 source of a type that is not "post". This is useful for
8118 implementing clean-up and check event sources that are
8119 triggered by other work being done in the program.
8121 * systemd-networkd is no longer statically enabled, but uses
8122 the usual [Install] sections so that it can be
8123 enabled/disabled using systemctl. It still is enabled by
8126 * When creating a veth interface pair with systemd-nspawn, the
8127 host side will now be prefixed with "vb-" if
8128 --network-bridge= is used, and with "ve-" if --network-veth
8129 is used. This way, it is easy to distinguish these cases on
8130 the host, for example to apply different configuration to
8131 them with systemd-networkd.
8133 * The compatibility libraries for libsystemd-journal.so,
8134 libsystem-id128.so, libsystemd-login.so and
8135 libsystemd-daemon.so do not make use of IFUNC
8136 anymore. Instead, we now build libsystemd.so multiple times
8137 under these alternative names. This means that the footprint
8138 is drastically increased, but given that these are
8139 transitional compatibility libraries, this should not matter
8140 much. This change has been made necessary to support the ARM
8141 platform for these compatibility libraries, as the ARM
8142 toolchain is not really at the same level as the toolchain
8143 for other architectures like x86 and does not support
8144 IFUNC. Please make sure to use --enable-compat-libs only
8145 during a transitional period!
8147 * The .include syntax has been deprecated and is not documented
8148 anymore. Drop-in files in .d directories should be used instead.
8150 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
8151 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
8152 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
8153 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
8154 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
8155 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
8156 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
8157 Zbigniew Jędrzejewski-Szmek
8159 — Berlin, 2014-02-24
8163 * A new component "systemd-networkd" has been added that can
8164 be used to configure local network interfaces statically or
8165 via DHCP. It is capable of bringing up bridges, VLANs, and
8166 bonding. Currently, no hook-ups for interactive network
8167 configuration are provided. Use this for your initrd,
8168 container, embedded, or server setup if you need a simple,
8169 yet powerful, network configuration solution. This
8170 configuration subsystem is quite nifty, as it allows wildcard
8171 hotplug matching in interfaces. For example, with a single
8172 configuration snippet, you can configure that all Ethernet
8173 interfaces showing up are automatically added to a bridge,
8174 or similar. It supports link-sensing and more.
8176 * A new tool "systemd-socket-proxyd" has been added which can
8177 act as a bidirectional proxy for TCP sockets. This is
8178 useful for adding socket activation support to services that
8179 do not actually support socket activation, including virtual
8180 machines and the like.
8182 * Add a new tool to save/restore rfkill state on
8185 * Save/restore state of keyboard backlights in addition to
8186 display backlights on shutdown/boot.
8188 * udev learned a new SECLABEL{} construct to label device
8189 nodes with a specific security label when they appear. For
8190 now, only SECLABEL{selinux} is supported, but the syntax is
8191 prepared for additional security frameworks.
8193 * udev gained a new scheme to configure link-level attributes
8194 from files in /etc/systemd/network/*.link. These files can
8195 match against MAC address, device path, driver name and type,
8196 and will apply attributes like the naming policy, link speed,
8197 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
8198 address assignment policy (randomized, ...).
8200 * The configuration of network interface naming rules for
8201 "permanent interface names" has changed: a new NamePolicy=
8202 setting in the [Link] section of .link files determines the
8203 priority of possible naming schemes (onboard, slot, MAC,
8204 path). The default value of this setting is determined by
8205 /usr/lib/net/links/99-default.link. Old
8206 80-net-name-slot.rules udev configuration file has been
8207 removed, so local configuration overriding this file should
8208 be adapted to override 99-default.link instead.
8210 * When the User= switch is used in a unit file, also
8211 initialize $SHELL= based on the user database entry.
8213 * systemd no longer depends on libdbus. All communication is
8214 now done with sd-bus, systemd's low-level bus library
8217 * kdbus support has been added to PID 1 itself. When kdbus is
8218 enabled, this causes PID 1 to set up the system bus and
8219 enable support for a new ".busname" unit type that
8220 encapsulates bus name activation on kdbus. It works a little
8221 bit like ".socket" units, except for bus names. A new
8222 generator has been added that converts classic dbus1 service
8223 activation files automatically into native systemd .busname
8226 * sd-bus: add a light-weight vtable implementation that allows
8227 defining objects on the bus with a simple static const
8228 vtable array of its methods, signals and properties.
8230 * systemd will not generate or install static dbus
8231 introspection data anymore to /usr/share/dbus-1/interfaces,
8232 as the precise format of these files is unclear, and
8233 nothing makes use of it.
8235 * A proxy daemon is now provided to proxy clients connecting
8236 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
8237 compatibility with classic D-Bus.
8239 * A bus driver implementation has been added that supports the
8240 classic D-Bus bus driver calls on kdbus, also for
8241 compatibility purposes.
8243 * A new API "sd-event.h" has been added that implements a
8244 minimal event loop API built around epoll. It provides a
8245 couple of features that direct epoll usage is lacking:
8246 prioritization of events, scales to large numbers of timer
8247 events, per-event timer slack (accuracy), system-wide
8248 coalescing of timer events, exit handlers, watchdog
8249 supervision support using systemd's sd_notify() API, child
8252 * A new API "sd-rntl.h" has been added that provides an API
8253 around the route netlink interface of the kernel, similar in
8254 style to "sd-bus.h".
8256 * A new API "sd-dhcp-client.h" has been added that provides a
8257 small DHCPv4 client-side implementation. This is used by
8260 * There is a new kernel command line option
8261 "systemd.restore_state=0|1". When set to "0", none of the
8262 systemd tools will restore saved runtime state to hardware
8263 devices. More specifically, the rfkill and backlight states
8266 * The FsckPassNo= compatibility option in mount/service units
8267 has been removed. The fstab generator will now add the
8268 necessary dependencies automatically, and does not require
8269 PID1's support for that anymore.
8271 * journalctl gained a new switch, --list-boots, that lists
8272 recent boots with their times and boot IDs.
8274 * The various tools like systemctl, loginctl, timedatectl,
8275 busctl, systemd-run, ... have gained a new switch "-M" to
8276 connect to a specific, local OS container (as direct
8277 connection, without requiring SSH). This works on any
8278 container that is registered with machined, such as those
8279 created by libvirt-lxc or nspawn.
8281 * systemd-run and systemd-analyze also gained support for "-H"
8282 to connect to remote hosts via SSH. This is particularly
8283 useful for systemd-run because it enables queuing of jobs
8284 onto remote systems.
8286 * machinectl gained a new command "login" to open a getty
8287 login in any local container. This works with any container
8288 that is registered with machined (such as those created by
8289 libvirt-lxc or nspawn), and which runs systemd inside.
8291 * machinectl gained a new "reboot" command that may be used to
8292 trigger a reboot on a specific container that is registered
8293 with machined. This works on any container that runs an init
8294 system of some kind.
8296 * systemctl gained a new "list-timers" command to print a nice
8297 listing of installed timer units with the times they elapse
8300 * Alternative reboot() parameters may now be specified on the
8301 "systemctl reboot" command line and are passed to the
8302 reboot() system call.
8304 * systemctl gained a new --job-mode= switch to configure the
8305 mode to queue a job with. This is a more generic version of
8306 --fail, --irreversible, and --ignore-dependencies, which are
8307 still available but not advertised anymore.
8309 * /etc/systemd/system.conf gained new settings to configure
8310 various default timeouts of units, as well as the default
8311 start limit interval and burst. These may still be overridden
8314 * PID1 will now export on the bus profile data of the security
8315 policy upload process (such as the SELinux policy upload to
8318 * journald: when forwarding logs to the console, include
8319 timestamps (following the setting in
8320 /sys/module/printk/parameters/time).
8322 * OnCalendar= in timer units now understands the special
8323 strings "yearly" and "annually". (Both are equivalent)
8325 * The accuracy of timer units is now configurable with the new
8326 AccuracySec= setting. It defaults to 1min.
8328 * A new dependency type JoinsNamespaceOf= has been added that
8329 allows running two services within the same /tmp and network
8330 namespace, if PrivateNetwork= or PrivateTmp= are used.
8332 * A new command "cat" has been added to systemctl. It outputs
8333 the original unit file of a unit, and concatenates the
8334 contents of additional "drop-in" unit file snippets, so that
8335 the full configuration is shown.
8337 * systemctl now supports globbing on the various "list-xyz"
8338 commands, like "list-units" or "list-sockets", as well as on
8339 those commands which take multiple unit names.
8341 * journalctl's --unit= switch gained support for globbing.
8343 * All systemd daemons now make use of the watchdog logic so
8344 that systemd automatically notices when they hang.
8346 * If the $container_ttys environment variable is set,
8347 getty-generator will automatically spawn a getty for each
8348 listed tty. This is useful for container managers to request
8349 login gettys to be spawned on as many ttys as needed.
8351 * %h, %s, %U specifier support is not available anymore when
8352 used in unit files for PID 1. This is because NSS calls are
8353 not safe from PID 1. They stay available for --user
8354 instances of systemd, and as special case for the root user.
8356 * loginctl gained a new "--no-legend" switch to turn off output
8359 * The "sd-login.h" API gained three new calls:
8360 sd_session_is_remote(), sd_session_get_remote_user(),
8361 sd_session_get_remote_host() to query information about
8364 * The udev hardware database now also carries vendor/product
8365 information of SDIO devices.
8367 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
8368 determine whether watchdog notifications are requested by
8371 * Socket-activated per-connection services now include a
8372 short description of the connection parameters in the
8375 * tmpfiles gained a new "--boot" option. When this is not used,
8376 only lines where the command character is not suffixed with
8377 "!" are executed. When this option is specified, those
8378 options are executed too. This partitions tmpfiles
8379 directives into those that can be safely executed at any
8380 time, and those which should be run only at boot (for
8381 example, a line that creates /run/nologin).
8383 * A new API "sd-resolve.h" has been added which provides a simple
8384 asynchronous wrapper around glibc NSS hostname resolution
8385 calls, such as getaddrinfo(). In contrast to glibc's
8386 getaddrinfo_a(), it does not use signals. In contrast to most
8387 other asynchronous name resolution libraries, this one does
8388 not reimplement DNS, but reuses NSS, so that alternate
8389 hostname resolution systems continue to work, such as mDNS,
8390 LDAP, etc. This API is based on libasyncns, but it has been
8391 cleaned up for inclusion in systemd.
8393 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
8394 "sd-daemon.h" are no longer found in individual libraries
8395 libsystemd-journal.so, libsystemd-login.so,
8396 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
8397 merged them into a single library, libsystemd.so, which
8398 provides all symbols. The reason for this is cyclic
8399 dependencies, as these libraries tend to use each other's
8400 symbols. So far, we have managed to workaround that by linking
8401 a copy of a good part of our code into each of these
8402 libraries again and again, which, however, makes certain
8403 things hard to do, like sharing static variables. Also, it
8404 substantially increases footprint. With this change, there
8405 is only one library for the basic APIs systemd
8406 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
8407 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
8408 library as well, however are subject to the --enable-kdbus
8409 switch (see below). Note that "sd-dhcp-client.h" is not part
8410 of this library (this is because it only consumes, never
8411 provides, services of/to other APIs). To make the transition
8412 easy from the separate libraries to the unified one, we
8413 provide the --enable-compat-libs compile-time switch which
8414 will generate stub libraries that are compatible with the
8415 old ones but redirect all calls to the new one.
8417 * All of the kdbus logic and the new APIs "sd-bus.h",
8418 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
8419 and "sd-utf8.h" are compile-time optional via the
8420 "--enable-kdbus" switch, and they are not compiled in by
8421 default. To make use of kdbus, you have to explicitly enable
8422 the switch. Note however, that neither the kernel nor the
8423 userspace API for all of this is considered stable yet. We
8424 want to maintain the freedom to still change the APIs for
8425 now. By specifying this build-time switch, you acknowledge
8426 that you are aware of the instability of the current
8429 * Also, note that while kdbus is pretty much complete,
8430 it lacks one thing: proper policy support. This means you
8431 can build a fully working system with all features; however,
8432 it will be highly insecure. Policy support will be added in
8433 one of the next releases, at the same time that we will
8434 declare the APIs stable.
8436 * When the kernel command line argument "kdbus" is specified,
8437 systemd will automatically load the kdbus.ko kernel module. At
8438 this stage of development, it is only useful for testing kdbus
8439 and should not be used in production. Note: if "--enable-kdbus"
8440 is specified, and the kdbus.ko kernel module is available, and
8441 "kdbus" is added to the kernel command line, the entire system
8442 runs with kdbus instead of dbus-daemon, with the above mentioned
8443 problem of missing the system policy enforcement. Also a future
8444 version of kdbus.ko or a newer systemd will not be compatible with
8445 each other, and will unlikely be able to boot the machine if only
8446 one of them is updated.
8448 * systemctl gained a new "import-environment" command which
8449 uploads the caller's environment (or parts thereof) into the
8450 service manager so that it is inherited by services started
8451 by the manager. This is useful to upload variables like
8452 $DISPLAY into the user service manager.
8454 * A new PrivateDevices= switch has been added to service units
8455 which allows running a service with a namespaced /dev
8456 directory that does not contain any device nodes for
8457 physical devices. More specifically, it only includes devices
8458 such as /dev/null, /dev/urandom, and /dev/zero which are API
8461 * logind has been extended to support behaviour like VT
8462 switching on seats that do not support a VT. This makes
8463 multi-session available on seats that are not the first seat
8464 (seat0), and on systems where kernel support for VTs has
8465 been disabled at compile-time.
8467 * If a process holds a delay lock for system sleep or shutdown
8468 and fails to release it in time, we will now log its
8469 identity. This makes it easier to identify processes that
8470 cause slow suspends or power-offs.
8472 * When parsing /etc/crypttab, support for a new key-slot=
8473 option as supported by Debian is added. It allows indicating
8474 which LUKS slot to use on disk, speeding up key loading.
8476 * The sd_journal_sendv() API call has been checked and
8477 officially declared to be async-signal-safe so that it may
8478 be invoked from signal handlers for logging purposes.
8480 * Boot-time status output is now enabled automatically after a
8481 short timeout if boot does not progress, in order to give
8482 the user an indication what she or he is waiting for.
8484 * The boot-time output has been improved to show how much time
8485 remains until jobs expire.
8487 * The KillMode= switch in service units gained a new possible
8488 value "mixed". If set, and the unit is shut down, then the
8489 initial SIGTERM signal is sent only to the main daemon
8490 process, while the following SIGKILL signal is sent to
8491 all remaining processes of the service.
8493 * When a scope unit is registered, a new property "Controller"
8494 may be set. If set to a valid bus name, systemd will send a
8495 RequestStop() signal to this name when it would like to shut
8496 down the scope. This may be used to hook manager logic into
8497 the shutdown logic of scope units. Also, scope units may now
8498 be put in a special "abandoned" state, in which case the
8499 manager process which created them takes no further
8500 responsibilities for it.
8502 * When reading unit files, systemd will now verify
8503 the access mode of these files, and warn about certain
8504 suspicious combinations. This has been added to make it
8505 easier to track down packaging bugs where unit files are
8506 marked executable or world-writable.
8508 * systemd-nspawn gained a new "--setenv=" switch to set
8509 container-wide environment variables. The similar option in
8510 systemd-activate was renamed from "--environment=" to
8511 "--setenv=" for consistency.
8513 * systemd-nspawn has been updated to create a new kdbus domain
8514 for each container that is invoked, thus allowing each
8515 container to have its own set of system and user buses,
8516 independent of the host.
8518 * systemd-nspawn gained a new --drop-capability= switch to run
8519 the container with less capabilities than the default. Both
8520 --drop-capability= and --capability= now take the special
8521 string "all" for dropping or keeping all capabilities.
8523 * systemd-nspawn gained new switches for executing containers
8524 with specific SELinux labels set.
8526 * systemd-nspawn gained a new --quiet switch to not generate
8527 any additional output but the container's own console
8530 * systemd-nspawn gained a new --share-system switch to run a
8531 container without PID namespacing enabled.
8533 * systemd-nspawn gained a new --register= switch to control
8534 whether the container is registered with systemd-machined or
8535 not. This is useful for containers that do not run full
8536 OS images, but only specific apps.
8538 * systemd-nspawn gained a new --keep-unit which may be used
8539 when invoked as the only program from a service unit, and
8540 results in registration of the unit service itself in
8541 systemd-machined, instead of a newly opened scope unit.
8543 * systemd-nspawn gained a new --network-interface= switch for
8544 moving arbitrary interfaces to the container. The new
8545 --network-veth switch creates a virtual Ethernet connection
8546 between host and container. The new --network-bridge=
8547 switch then allows assigning the host side of this virtual
8548 Ethernet connection to a bridge device.
8550 * systemd-nspawn gained a new --personality= switch for
8551 setting the kernel personality for the container. This is
8552 useful when running a 32-bit container on a 64-bit host. A
8553 similar option Personality= is now also available for service
8556 * logind will now also track a "Desktop" identifier for each
8557 session which encodes the desktop environment of it. This is
8558 useful for desktop environments that want to identify
8559 multiple running sessions of itself easily.
8561 * A new SELinuxContext= setting for service units has been
8562 added that allows setting a specific SELinux execution
8563 context for a service.
8565 * Most systemd client tools will now honour $SYSTEMD_LESS for
8566 settings of the "less" pager. By default, these tools will
8567 override $LESS to allow certain operations to work, such as
8568 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
8569 influence this logic.
8571 * systemd's "seccomp" hook-up has been changed to make use of
8572 the libseccomp library instead of using its own
8573 implementation. This has benefits for portability among
8576 * For usage together with SystemCallFilter=, a new
8577 SystemCallErrorNumber= setting has been introduced that
8578 allows configuration of a system error number to be returned
8579 on filtered system calls, instead of immediately killing the
8580 process. Also, SystemCallArchitectures= has been added to
8581 limit access to system calls of a particular architecture
8582 (in order to turn off support for unused secondary
8583 architectures). There is also a global
8584 SystemCallArchitectures= setting in system.conf now to turn
8585 off support for non-native system calls system-wide.
8587 * systemd requires a kernel with a working name_to_handle_at(),
8588 please see the kernel config requirements in the README file.
8590 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
8591 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
8592 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
8593 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
8594 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
8595 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
8596 Elia Pinto, Florian Weimer, George McCollister, Goffredo
8597 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
8598 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
8599 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
8600 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
8601 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
8602 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
8603 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
8604 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
8605 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
8606 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
8607 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
8608 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
8609 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
8610 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
8611 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
8612 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
8613 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
8615 — Berlin, 2014-02-20
8619 * logind has gained support for facilitating privileged input
8620 and drm device access for unprivileged clients. This work is
8621 useful to allow Wayland display servers (and similar
8622 programs, such as kmscon) to run under the user's ID and
8623 access input and drm devices which are normally
8624 protected. When this is used (and the kernel is new enough)
8625 logind will "mute" IO on the file descriptors passed to
8626 Wayland as long as it is in the background and "unmute" it
8627 if it returns into the foreground. This allows secure
8628 session switching without allowing background sessions to
8629 eavesdrop on input and display data. This also introduces
8630 session switching support if VT support is turned off in the
8631 kernel, and on seats that are not seat0.
8633 * A new kernel command line option luks.options= is understood
8634 now which allows specifying LUKS options for usage for LUKS
8635 encrypted partitions specified with luks.uuid=.
8637 * tmpfiles.d(5) snippets may now use specifier expansion in
8638 path names. More specifically %m, %b, %H, %v, are now
8639 replaced by the local machine id, boot id, hostname, and
8640 kernel version number.
8642 * A new tmpfiles.d(5) command "m" has been introduced which
8643 may be used to change the owner/group/access mode of a file
8644 or directory if it exists, but do nothing if it does not.
8646 * This release removes high-level support for the
8647 MemorySoftLimit= cgroup setting. The underlying kernel
8648 cgroup attribute memory.soft_limit= is currently badly
8649 designed and likely to be removed from the kernel API in its
8650 current form, hence we should not expose it for now.
8652 * The memory.use_hierarchy cgroup attribute is now enabled for
8653 all cgroups systemd creates in the memory cgroup
8654 hierarchy. This option is likely to be come the built-in
8655 default in the kernel anyway, and the non-hierarchical mode
8656 never made much sense in the intrinsically hierarchical
8659 * A new field _SYSTEMD_SLICE= is logged along with all journal
8660 messages containing the slice a message was generated
8661 from. This is useful to allow easy per-customer filtering of
8662 logs among other things.
8664 * systemd-journald will no longer adjust the group of journal
8665 files it creates to the "systemd-journal" group. Instead we
8666 rely on the journal directory to be owned by the
8667 "systemd-journal" group, and its setgid bit set, so that the
8668 kernel file system layer will automatically enforce that
8669 journal files inherit this group assignment. The reason for
8670 this change is that we cannot allow NSS look-ups from
8671 journald which would be necessary to resolve
8672 "systemd-journal" to a numeric GID, because this might
8673 create deadlocks if NSS involves synchronous queries to
8674 other daemons (such as nscd, or sssd) which in turn are
8675 logging clients of journald and might block on it, which
8676 would then dead lock. A tmpfiles.d(5) snippet included in
8677 systemd will make sure the setgid bit and group are
8678 properly set on the journal directory if it exists on every
8679 boot. However, we recommend adjusting it manually after
8680 upgrades too (or from RPM scriptlets), so that the change is
8681 not delayed until next reboot.
8683 * Backlight and random seed files in /var/lib/ have moved into
8684 the /var/lib/systemd/ directory, in order to centralize all
8685 systemd generated files in one directory.
8687 * Boot time performance measurements (as displayed by
8688 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
8689 performance information if that's available to determine how
8690 much time BIOS and boot loader initialization required. With
8691 a sufficiently new BIOS you hence no longer need to boot
8692 with Gummiboot to get access to such information.
8694 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
8695 Cristian Rodríguez, Dave Reisner, David Herrmann, David
8696 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
8697 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
8698 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
8699 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
8700 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8702 — Berlin, 2013-10-02
8706 * The Restart= option for services now understands a new
8707 on-watchdog setting, which will restart the service
8708 automatically if the service stops sending out watchdog keep
8709 alive messages (as configured with WatchdogSec=).
8711 * The getty generator (which is responsible for bringing up a
8712 getty on configured serial consoles) will no longer only
8713 start a getty on the primary kernel console but on all
8714 others, too. This makes the order in which console= is
8715 specified on the kernel command line less important.
8717 * libsystemd-logind gained a new sd_session_get_vt() call to
8718 retrieve the VT number of a session.
8720 * If the option "tries=0" is set for an entry of /etc/crypttab
8721 its passphrase is queried indefinitely instead of any
8722 maximum number of tries.
8724 * If a service with a configure PID file terminates its PID
8725 file will now be removed automatically if it still exists
8726 afterwards. This should put an end to stale PID files.
8728 * systemd-run will now also take relative binary path names
8729 for execution and no longer insists on absolute paths.
8731 * InaccessibleDirectories= and ReadOnlyDirectories= now take
8732 paths that are optionally prefixed with "-" to indicate that
8733 it should not be considered a failure if they do not exist.
8735 * journalctl -o (and similar commands) now understands a new
8736 output mode "short-precise", it is similar to "short" but
8737 shows timestamps with usec accuracy.
8739 * The option "discard" (as known from Debian) is now
8740 synonymous to "allow-discards" in /etc/crypttab. In fact,
8741 "discard" is preferred now (since it is easier to remember
8744 * Some licensing clean-ups were made, so that more code is now
8745 LGPL-2.1 licensed than before.
8747 * A minimal tool to save/restore the display backlight
8748 brightness across reboots has been added. It will store the
8749 backlight setting as late as possible at shutdown, and
8750 restore it as early as possible during reboot.
8752 * A logic to automatically discover and enable home and swap
8753 partitions on GPT disks has been added. With this in place
8754 /etc/fstab becomes optional for many setups as systemd can
8755 discover certain partitions located on the root disk
8756 automatically. Home partitions are recognized under their
8757 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
8758 partitions are recognized under their GPT type ID
8759 0657fd6da4ab43c484e50933c84b4f4f.
8761 * systemd will no longer pass any environment from the kernel
8762 or initrd to system services. If you want to set an
8763 environment for all services, do so via the kernel command
8764 line systemd.setenv= assignment.
8766 * The systemd-sysctl tool no longer natively reads the file
8767 /etc/sysctl.conf. If desired, the file should be symlinked
8768 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
8769 legacy support by a symlink rather than built-in code, it
8770 also makes the otherwise hidden order of application of the
8771 different files visible. (Note that this partly reverts to a
8772 pre-198 application order of sysctl knobs!)
8774 * The "systemctl set-log-level" and "systemctl dump" commands
8775 have been moved to systemd-analyze.
8777 * systemd-run learned the new --remain-after-exit switch,
8778 which causes the scope unit not to be cleaned up
8779 automatically after the process terminated.
8781 * tmpfiles learned a new --exclude-prefix= switch to exclude
8782 certain paths from operation.
8784 * journald will now automatically flush all messages to disk
8785 as soon as a message at the log level CRIT, ALERT or EMERG
8788 Contributions from: Andrew Cook, Brandon Philips, Christian
8789 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
8790 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
8791 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
8792 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
8793 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
8794 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
8795 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
8796 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
8797 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
8798 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
8799 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
8800 William Giokas, Zbigniew Jędrzejewski-Szmek
8802 — Berlin, 2013-09-13
8806 * The documentation has been updated to cover the various new
8807 concepts introduced with 205.
8809 * Unit files now understand the new %v specifier which
8810 resolves to the kernel version string as returned by "uname
8813 * systemctl now supports filtering the unit list output by
8814 load state, active state and sub state, using the new
8817 * "systemctl status" will now show the results of the
8818 condition checks (like ConditionPathExists= and similar) of
8819 the last start attempts of the unit. They are also logged to
8822 * "journalctl -b" may now be used to look for boot output of a
8823 specific boot. Try "journalctl -b -1" for the previous boot,
8824 but the syntax is substantially more powerful.
8826 * "journalctl --show-cursor" has been added which prints the
8827 cursor string the last shown log line. This may then be used
8828 with the new "journalctl --after-cursor=" switch to continue
8829 browsing logs from that point on.
8831 * "journalctl --force" may now be used to force regeneration
8834 * Creation of "dead" device nodes has been moved from udev
8835 into kmod and tmpfiles. Previously, udev would read the kmod
8836 databases to pre-generate dead device nodes based on meta
8837 information contained in kernel modules, so that these would
8838 be auto-loaded on access rather then at boot. As this
8839 does not really have much to do with the exposing actual
8840 kernel devices to userspace this has always been slightly
8841 alien in the udev codebase. Following the new scheme kmod
8842 will now generate a runtime snippet for tmpfiles from the
8843 module meta information and it now is tmpfiles' job to the
8844 create the nodes. This also allows overriding access and
8845 other parameters for the nodes using the usual tmpfiles
8846 facilities. As side effect this allows us to remove the
8847 CAP_SYS_MKNOD capability bit from udevd entirely.
8849 * logind's device ACLs may now be applied to these "dead"
8850 devices nodes too, thus finally allowing managed access to
8851 devices such as /dev/snd/sequencer without loading the
8852 backing module right-away.
8854 * A new RPM macro has been added that may be used to apply
8855 tmpfiles configuration during package installation.
8857 * systemd-detect-virt and ConditionVirtualization= now can
8858 detect User-Mode-Linux machines (UML).
8860 * journald will now implicitly log the effective capabilities
8861 set of processes in the message metadata.
8863 * systemd-cryptsetup has gained support for TrueCrypt volumes.
8865 * The initrd interface has been simplified (more specifically,
8866 support for passing performance data via environment
8867 variables and fsck results via files in /run has been
8868 removed). These features were non-essential, and are
8869 nowadays available in a much nicer way by having systemd in
8870 the initrd serialize its state and have the hosts systemd
8871 deserialize it again.
8873 * The udev "keymap" data files and tools to apply keyboard
8874 specific mappings of scan to key codes, and force-release
8875 scan code lists have been entirely replaced by a udev
8876 "keyboard" builtin and a hwdb data file.
8878 * systemd will now honour the kernel's "quiet" command line
8879 argument also during late shutdown, resulting in a
8880 completely silent shutdown when used.
8882 * There's now an option to control the SO_REUSEPORT socket
8883 option in .socket units.
8885 * Instance units will now automatically get a per-template
8886 subslice of system.slice unless something else is explicitly
8887 configured. For example, instances of sshd@.service will now
8888 implicitly be placed in system-sshd.slice rather than
8889 system.slice as before.
8891 * Test coverage support may now be enabled at build time.
8893 Contributions from: Dave Reisner, Frederic Crozat, Harald
8894 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
8895 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
8896 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
8897 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
8898 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
8899 Giokas, Zbigniew Jędrzejewski-Szmek
8901 — Berlin, 2013-07-23
8905 * Two new unit types have been introduced:
8907 Scope units are very similar to service units, however, are
8908 created out of pre-existing processes — instead of PID 1
8909 forking off the processes. By using scope units it is
8910 possible for system services and applications to group their
8911 own child processes (worker processes) in a powerful way
8912 which then maybe used to organize them, or kill them
8913 together, or apply resource limits on them.
8915 Slice units may be used to partition system resources in an
8916 hierarchical fashion and then assign other units to them. By
8917 default there are now three slices: system.slice (for all
8918 system services), user.slice (for all user sessions),
8919 machine.slice (for VMs and containers).
8921 Slices and scopes have been introduced primarily in
8922 context of the work to move cgroup handling to a
8923 single-writer scheme, where only PID 1
8924 creates/removes/manages cgroups.
8926 * There's a new concept of "transient" units. In contrast to
8927 normal units these units are created via an API at runtime,
8928 not from configuration from disk. More specifically this
8929 means it is now possible to run arbitrary programs as
8930 independent services, with all execution parameters passed
8931 in via bus APIs rather than read from disk. Transient units
8932 make systemd substantially more dynamic then it ever was,
8933 and useful as a general batch manager.
8935 * logind has been updated to make use of scope and slice units
8936 for managing user sessions. As a user logs in he will get
8937 his own private slice unit, to which all sessions are added
8938 as scope units. We also added support for automatically
8939 adding an instance of user@.service for the user into the
8940 slice. Effectively logind will no longer create cgroup
8941 hierarchies on its own now, it will defer entirely to PID 1
8942 for this by means of scope, service and slice units. Since
8943 user sessions this way become entities managed by PID 1
8944 the output of "systemctl" is now a lot more comprehensive.
8946 * A new mini-daemon "systemd-machined" has been added which
8947 may be used by virtualization managers to register local
8948 VMs/containers. nspawn has been updated accordingly, and
8949 libvirt will be updated shortly. machined will collect a bit
8950 of meta information about the VMs/containers, and assign
8951 them their own scope unit (see above). The collected
8952 meta-data is then made available via the "machinectl" tool,
8953 and exposed in "ps" and similar tools. machined/machinectl
8954 is compile-time optional.
8956 * As discussed earlier, the low-level cgroup configuration
8957 options ControlGroup=, ControlGroupModify=,
8958 ControlGroupPersistent=, ControlGroupAttribute= have been
8959 removed. Please use high-level attribute settings instead as
8960 well as slice units.
8962 * A new bus call SetUnitProperties() has been added to alter
8963 various runtime parameters of a unit. This is primarily
8964 useful to alter cgroup parameters dynamically in a nice way,
8965 but will be extended later on to make more properties
8966 modifiable at runtime. systemctl gained a new set-properties
8967 command that wraps this call.
8969 * A new tool "systemd-run" has been added which can be used to
8970 run arbitrary command lines as transient services or scopes,
8971 while configuring a number of settings via the command
8972 line. This tool is currently very basic, however already
8973 very useful. We plan to extend this tool to even allow
8974 queuing of execution jobs with time triggers from the
8975 command line, similar in fashion to "at".
8977 * nspawn will now inform the user explicitly that kernels with
8978 audit enabled break containers, and suggest the user to turn
8981 * Support for detecting the IMA and AppArmor security
8982 frameworks with ConditionSecurity= has been added.
8984 * journalctl gained a new "-k" switch for showing only kernel
8985 messages, mimicking dmesg output; in addition to "--user"
8986 and "--system" switches for showing only user's own logs
8989 * systemd-delta can now show information about drop-in
8990 snippets extending unit files.
8992 * libsystemd-bus has been substantially updated but is still
8993 not available as public API.
8995 * systemd will now look for the "debug" argument on the kernel
8996 command line and enable debug logging, similar to what
8997 "systemd.log_level=debug" already did before.
8999 * "systemctl set-default", "systemctl get-default" has been
9000 added to configure the default.target symlink, which
9001 controls what to boot into by default.
9003 * "systemctl set-log-level" has been added as a convenient
9004 way to raise and lower systemd logging threshold.
9006 * "systemd-analyze plot" will now show the time the various
9007 generators needed for execution, as well as information
9008 about the unit file loading.
9010 * libsystemd-journal gained a new sd_journal_open_files() call
9011 for opening specific journal files. journactl also gained a
9012 new switch to expose this new functionality. Previously we
9013 only supported opening all files from a directory, or all
9014 files from the system, as opening individual files only is
9015 racy due to journal file rotation.
9017 * systemd gained the new DefaultEnvironment= setting in
9018 /etc/systemd/system.conf to set environment variables for
9021 * If a privileged process logs a journal message with the
9022 OBJECT_PID= field set, then journald will automatically
9023 augment this with additional OBJECT_UID=, OBJECT_GID=,
9024 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
9025 system services want to log events about specific client
9026 processes. journactl/systemctl has been updated to make use
9027 of this information if all log messages regarding a specific
9030 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
9031 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
9032 Reisner, David Coppa, David King, David Strauss, Eelco
9033 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
9034 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
9035 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
9036 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
9037 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
9038 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
9039 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
9040 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
9041 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
9042 Łukasz Stelmach, 장동준
9046 * The Python bindings gained some minimal support for the APIs
9047 exposed by libsystemd-logind.
9049 * ConditionSecurity= gained support for detecting SMACK. Since
9050 this condition already supports SELinux and AppArmor we only
9051 miss IMA for this. Patches welcome!
9053 Contributions from: Karol Lewandowski, Lennart Poettering,
9054 Zbigniew Jędrzejewski-Szmek
9058 * systemd-nspawn will now create /etc/resolv.conf if
9059 necessary, before bind-mounting the host's file onto it.
9061 * systemd-nspawn will now store meta information about a
9062 container on the container's cgroup as extended attribute
9063 fields, including the root directory.
9065 * The cgroup hierarchy has been reworked in many ways. All
9066 objects any of the components systemd creates in the cgroup
9067 tree are now suffixed. More specifically, user sessions are
9068 now placed in cgroups suffixed with ".session", users in
9069 cgroups suffixed with ".user", and nspawn containers in
9070 cgroups suffixed with ".nspawn". Furthermore, all cgroup
9071 names are now escaped in a simple scheme to avoid collision
9072 of userspace object names with kernel filenames. This work
9073 is preparation for making these objects relocatable in the
9074 cgroup tree, in order to allow easy resource partitioning of
9075 these objects without causing naming conflicts.
9077 * systemctl list-dependencies gained the new switches
9078 --plain, --reverse, --after and --before.
9080 * systemd-inhibit now shows the process name of processes that
9081 have taken an inhibitor lock.
9083 * nss-myhostname will now also resolve "localhost"
9084 implicitly. This makes /etc/hosts an optional file and
9085 nicely handles that on IPv6 ::1 maps to both "localhost" and
9088 * libsystemd-logind.so gained a new call
9089 sd_get_machine_names() to enumerate running containers and
9090 VMs (currently only supported by very new libvirt and
9091 nspawn). sd_login_monitor can now be used to watch
9092 VMs/containers coming and going.
9094 * .include is not allowed recursively anymore, and only in
9095 unit files. Usually it is better to use drop-in snippets in
9096 .d/*.conf anyway, as introduced with systemd 198.
9098 * systemd-analyze gained a new "critical-chain" command that
9099 determines the slowest chain of units run during system
9100 boot-up. It is very useful for tracking down where
9101 optimizing boot time is the most beneficial.
9103 * systemd will no longer allow manipulating service paths in
9104 the name=systemd:/system cgroup tree using ControlGroup= in
9105 units. (But is still fine with it in all other dirs.)
9107 * There's a new systemd-nspawn@.service service file that may
9108 be used to easily run nspawn containers as system
9109 services. With the container's root directory in
9110 /var/lib/container/foobar it is now sufficient to run
9111 "systemctl start systemd-nspawn@foobar.service" to boot it.
9113 * systemd-cgls gained a new parameter "--machine" to list only
9114 the processes within a certain container.
9116 * ConditionSecurity= now can check for "apparmor". We still
9117 are lacking checks for SMACK and IMA for this condition
9118 check though. Patches welcome!
9120 * A new configuration file /etc/systemd/sleep.conf has been
9121 added that may be used to configure which kernel operation
9122 systemd is supposed to execute when "suspend", "hibernate"
9123 or "hybrid-sleep" is requested. This makes the new kernel
9124 "freeze" state accessible to the user.
9126 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
9127 the passed argument if applicable.
9129 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
9130 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
9131 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
9132 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
9133 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
9134 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
9135 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
9140 * The output of 'systemctl list-jobs' got some polishing. The
9141 '--type=' argument may now be passed more than once. A new
9142 command 'systemctl list-sockets' has been added which shows
9143 a list of kernel sockets systemd is listening on with the
9144 socket units they belong to, plus the units these socket
9147 * The experimental libsystemd-bus library got substantial
9148 updates to work in conjunction with the (also experimental)
9149 kdbus kernel project. It works well enough to exchange
9150 messages with some sophistication. Note that kdbus is not
9151 ready yet, and the library is mostly an elaborate test case
9152 for now, and not installable.
9154 * systemd gained a new unit 'systemd-static-nodes.service'
9155 that generates static device nodes earlier during boot, and
9156 can run in conjunction with udev.
9158 * libsystemd-login gained a new call sd_pid_get_user_unit()
9159 to retrieve the user systemd unit a process is running
9160 in. This is useful for systems where systemd is used as
9163 * systemd-nspawn now places all containers in the new /machine
9164 top-level cgroup directory in the name=systemd
9165 hierarchy. libvirt will soon do the same, so that we get a
9166 uniform separation of /system, /user and /machine for system
9167 services, user processes and containers/virtual
9168 machines. This new cgroup hierarchy is also useful to stick
9169 stable names to specific container instances, which can be
9170 recognized later this way (this name may be controlled
9171 via systemd-nspawn's new -M switch). libsystemd-login also
9172 gained a new call sd_pid_get_machine_name() to retrieve the
9173 name of the container/VM a specific process belongs to.
9175 * bootchart can now store its data in the journal.
9177 * libsystemd-journal gained a new call
9178 sd_journal_add_conjunction() for AND expressions to the
9179 matching logic. This can be used to express more complex
9180 logical expressions.
9182 * journactl can now take multiple --unit= and --user-unit=
9185 * The cryptsetup logic now understands the "luks.key=" kernel
9186 command line switch for specifying a file to read the
9187 decryption key from. Also, if a configured key file is not
9188 found the tool will now automatically fall back to prompting
9191 * Python systemd.journal module was updated to wrap recently
9192 added functions from libsystemd-journal. The interface was
9193 changed to bring the low level interface in s.j._Reader
9194 closer to the C API, and the high level interface in
9195 s.j.Reader was updated to wrap and convert all data about
9198 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
9199 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
9200 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
9201 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
9202 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
9203 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9207 * journalctl --update-catalog now understands a new --root=
9208 option to operate on catalogs found in a different root
9211 * During shutdown after systemd has terminated all running
9212 services a final killing loop kills all remaining left-over
9213 processes. We will now print the name of these processes
9214 when we send SIGKILL to them, since this usually indicates a
9217 * If /etc/crypttab refers to password files stored on
9218 configured mount points automatic dependencies will now be
9219 generated to ensure the specific mount is established first
9220 before the key file is attempted to be read.
9222 * 'systemctl status' will now show information about the
9223 network sockets a socket unit is listening on.
9225 * 'systemctl status' will also shown information about any
9226 drop-in configuration file for units. (Drop-In configuration
9227 files in this context are files such as
9228 /etc/systemd/systemd/foobar.service.d/*.conf)
9230 * systemd-cgtop now optionally shows summed up CPU times of
9231 cgroups. Press '%' while running cgtop to switch between
9232 percentage and absolute mode. This is useful to determine
9233 which cgroups use up the most CPU time over the entire
9234 runtime of the system. systemd-cgtop has also been updated
9235 to be 'pipeable' for processing with further shell tools.
9237 * 'hostnamectl set-hostname' will now allow setting of FQDN
9240 * The formatting and parsing of time span values has been
9241 changed. The parser now understands fractional expressions
9242 such as "5.5h". The formatter will now output fractional
9243 expressions for all time spans under 1min, i.e. "5.123456s"
9244 rather than "5s 123ms 456us". For time spans under 1s
9245 millisecond values are shown, for those under 1ms
9246 microsecond values are shown. This should greatly improve
9247 all time-related output of systemd.
9249 * libsystemd-login and libsystemd-journal gained new
9250 functions for querying the poll() events mask and poll()
9251 timeout value for integration into arbitrary event
9254 * localectl gained the ability to list available X11 keymaps
9255 (models, layouts, variants, options).
9257 * 'systemd-analyze dot' gained the ability to filter for
9258 specific units via shell-style globs, to create smaller,
9259 more useful graphs. I.e. it is now possible to create simple
9260 graphs of all the dependencies between only target units, or
9261 of all units that Avahi has dependencies with.
9263 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
9264 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
9265 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
9266 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
9267 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
9268 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
9269 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
9273 * The boot-time readahead implementation for rotating media
9274 will now read the read-ahead data in multiple passes which
9275 consist of all read requests made in equidistant time
9276 intervals. This means instead of strictly reading read-ahead
9277 data in its physical order on disk we now try to find a
9278 middle ground between physical and access time order.
9280 * /etc/os-release files gained a new BUILD_ID= field for usage
9281 on operating systems that provide continuous builds of OS
9284 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
9285 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
9286 William Douglas, Zbigniew Jędrzejewski-Szmek
9290 * systemd-python gained an API exposing libsystemd-daemon.
9292 * The SMACK setup logic gained support for uploading CIPSO
9295 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
9296 ReadOnlyDirectories= and InaccessibleDirectories= has
9297 changed. The private /tmp and /var/tmp directories are now
9298 shared by all processes of a service (which means
9299 ExecStartPre= may now leave data in /tmp that ExecStart= of
9300 the same service can still access). When a service is
9301 stopped its temporary directories are immediately deleted
9302 (normal clean-up with tmpfiles is still done in addition to
9305 * By default, systemd will now set a couple of sysctl
9306 variables in the kernel: the safe sysrq options are turned
9307 on, IP route verification is turned on, and source routing
9308 disabled. The recently added hardlink and softlink
9309 protection of the kernel is turned on. These settings should
9310 be reasonably safe, and good defaults for all new systems.
9312 * The predictable network naming logic may now be turned off
9313 with a new kernel command line switch: net.ifnames=0.
9315 * A new libsystemd-bus module has been added that implements a
9316 pretty complete D-Bus client library. For details see:
9318 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
9320 * journald will now explicitly flush the journal files to disk
9321 at the latest 5min after each write. The file will then also
9322 be marked offline until the next write. This should increase
9323 reliability in case of a crash. The synchronization delay
9324 can be configured via SyncIntervalSec= in journald.conf.
9326 * There's a new remote-fs-setup.target unit that can be used
9327 to pull in specific services when at least one remote file
9328 system is to be mounted.
9330 * There are new targets timers.target and paths.target as
9331 canonical targets to pull user timer and path units in
9332 from. This complements sockets.target with a similar
9333 purpose for socket units.
9335 * libudev gained a new call udev_device_set_attribute_value()
9336 to set sysfs attributes of a device.
9338 * The udev daemon now sets the default number of worker
9339 processes executed in parallel based on the number of available
9340 CPUs instead of the amount of available RAM. This is supposed
9341 to provide a more reliable default and limit a too aggressive
9342 parallelism for setups with 1000s of devices connected.
9344 Contributions from: Auke Kok, Colin Walters, Cristian
9345 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
9346 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
9347 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
9348 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
9349 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
9350 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
9351 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
9352 Zbigniew Jędrzejewski-Szmek
9356 * Configuration of unit files may now be extended via drop-in
9357 files without having to edit/override the unit files
9358 themselves. More specifically, if the administrator wants to
9359 change one value for a service file foobar.service he can
9360 now do so by dropping in a configuration snippet into
9361 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
9362 will load all these snippets and apply them on top of the
9363 main unit configuration file, possibly extending or
9364 overriding its settings. Using these drop-in snippets is
9365 generally nicer than the two earlier options for changing
9366 unit files locally: copying the files from
9367 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
9368 them there; or creating a new file in /etc/systemd/system/
9369 that incorporates the original one via ".include". Drop-in
9370 snippets into these .d/ directories can be placed in any
9371 directory systemd looks for units in, and the usual
9372 overriding semantics between /usr/lib, /etc and /run apply
9375 * Most unit file settings which take lists of items can now be
9376 reset by assigning the empty string to them. For example,
9377 normally, settings such as Environment=FOO=BAR append a new
9378 environment variable assignment to the environment block,
9379 each time they are used. By assigning Environment= the empty
9380 string the environment block can be reset to empty. This is
9381 particularly useful with the .d/*.conf drop-in snippets
9382 mentioned above, since this adds the ability to reset list
9383 settings from vendor unit files via these drop-ins.
9385 * systemctl gained a new "list-dependencies" command for
9386 listing the dependencies of a unit recursively.
9388 * Inhibitors are now honored and listed by "systemctl
9389 suspend", "systemctl poweroff" (and similar) too, not only
9390 GNOME. These commands will also list active sessions by
9393 * Resource limits (as exposed by the various control group
9394 controllers) can now be controlled dynamically at runtime
9395 for all units. More specifically, you can now use a command
9396 like "systemctl set-cgroup-attr foobar.service cpu.shares
9397 2000" to alter the CPU shares a specific service gets. These
9398 settings are stored persistently on disk, and thus allow the
9399 administrator to easily adjust the resource usage of
9400 services with a few simple commands. This dynamic resource
9401 management logic is also available to other programs via the
9402 bus. Almost any kernel cgroup attribute and controller is
9405 * systemd-vconsole-setup will now copy all font settings to
9406 all allocated VTs, where it previously applied them only to
9409 * libsystemd-login gained the new sd_session_get_tty() API
9412 * This release drops support for a few legacy or
9413 distribution-specific LSB facility names when parsing init
9414 scripts: $x-display-manager, $mail-transfer-agent,
9415 $mail-transport-agent, $mail-transfer-agent, $smtp,
9416 $null. Also, the mail-transfer-agent.target unit backing
9417 this has been removed. Distributions which want to retain
9418 compatibility with this should carry the burden for
9419 supporting this themselves and patch support for these back
9420 in, if they really need to. Also, the facilities $syslog and
9421 $local_fs are now ignored, since systemd does not support
9422 early-boot LSB init scripts anymore, and these facilities
9423 are implied anyway for normal services. syslog.target has
9426 * There are new bus calls on PID1's Manager object for
9427 cancelling jobs, and removing snapshot units. Previously,
9428 both calls were only available on the Job and Snapshot
9431 * systemd-journal-gatewayd gained SSL support.
9433 * The various "environment" files, such as /etc/locale.conf
9434 now support continuation lines with a backslash ("\") as
9435 last character in the line, similarly in style (but different)
9436 to how this is supported in shells.
9438 * For normal user processes the _SYSTEMD_USER_UNIT= field is
9439 now implicitly appended to every log entry logged. systemctl
9440 has been updated to filter by this field when operating on a
9441 user systemd instance.
9443 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
9444 CAP_AUDIT_CONTROL capabilities to the capabilities set for
9445 the container. This makes it easier to boot unmodified
9446 Fedora systems in a container, which however still requires
9447 audit=0 to be passed on the kernel command line. Auditing in
9448 kernel and userspace is unfortunately still too broken in
9449 context of containers, hence we recommend compiling it out
9450 of the kernel or using audit=0. Hopefully this will be fixed
9451 one day for good in the kernel.
9453 * nspawn gained the new --bind= and --bind-ro= parameters to
9454 bind mount specific directories from the host into the
9457 * nspawn will now mount its own devpts file system instance
9458 into the container, in order not to leak pty devices from
9459 the host into the container.
9461 * systemd will now read the firmware boot time performance
9462 information from the EFI variables, if the used boot loader
9463 supports this, and takes it into account for boot performance
9464 analysis via "systemd-analyze". This is currently supported
9465 only in conjunction with Gummiboot, but could be supported
9466 by other boot loaders too. For details see:
9468 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
9470 * A new generator has been added that automatically mounts the
9471 EFI System Partition (ESP) to /boot, if that directory
9472 exists, is empty, and no other file system has been
9473 configured to be mounted there.
9475 * logind will now send out PrepareForSleep(false) out
9476 unconditionally, after coming back from suspend. This may be
9477 used by applications as asynchronous notification for
9478 system resume events.
9480 * "systemctl unlock-sessions" has been added, that allows
9481 unlocking the screens of all user sessions at once, similar
9482 to how "systemctl lock-sessions" already locked all users
9483 sessions. This is backed by a new D-Bus call UnlockSessions().
9485 * "loginctl seat-status" will now show the master device of a
9486 seat. (i.e. the device of a seat that needs to be around for
9487 the seat to be considered available, usually the graphics
9490 * tmpfiles gained a new "X" line type, that allows
9491 configuration of files and directories (with wildcards) that
9492 shall be excluded from automatic cleanup ("aging").
9494 * udev default rules set the device node permissions now only
9495 at "add" events, and do not change them any longer with a
9496 later "change" event.
9498 * The log messages for lid events and power/sleep keypresses
9499 now carry a message ID.
9501 * We now have a substantially larger unit test suite, but this
9502 continues to be work in progress.
9504 * udevadm hwdb gained a new --root= parameter to change the
9505 root directory to operate relative to.
9507 * logind will now issue a background sync() request to the kernel
9508 early at shutdown, so that dirty buffers are flushed to disk early
9509 instead of at the last moment, in order to optimize shutdown
9512 * A new bootctl tool has been added that is an interface for
9513 certain boot loader operations. This is currently a preview
9514 and is likely to be extended into a small mechanism daemon
9515 like timedated, localed, hostnamed, and can be used by
9516 graphical UIs to enumerate available boot options, and
9517 request boot into firmware operations.
9519 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
9520 the rest of the package. It also has been updated to work
9521 correctly in initrds.
9523 * polkit previously has been runtime optional, and is now also
9524 compile time optional via a configure switch.
9526 * systemd-analyze has been reimplemented in C. Also "systemctl
9527 dot" has moved into systemd-analyze.
9529 * "systemctl status" with no further parameters will now print
9530 the status of all active or failed units.
9532 * Operations such as "systemctl start" can now be executed
9533 with a new mode "--irreversible" which may be used to queue
9534 operations that cannot accidentally be reversed by a later
9535 job queuing. This is by default used to make shutdown
9536 requests more robust.
9538 * The Python API of systemd now gained a new module for
9539 reading journal files.
9541 * A new tool kernel-install has been added that can install
9542 kernel images according to the Boot Loader Specification:
9544 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
9546 * Boot time console output has been improved to provide
9547 animated boot time output for hanging jobs.
9549 * A new tool systemd-activate has been added which can be used
9550 to test socket activation with, directly from the command
9551 line. This should make it much easier to test and debug
9552 socket activation in daemons.
9554 * journalctl gained a new "--reverse" (or -r) option to show
9555 journal output in reverse order (i.e. newest line first).
9557 * journalctl gained a new "--pager-end" (or -e) option to jump
9558 to immediately jump to the end of the journal in the
9559 pager. This is only supported in conjunction with "less".
9561 * journalctl gained a new "--user-unit=" option, that works
9562 similarly to "--unit=" but filters for user units rather than
9565 * A number of unit files to ease adoption of systemd in
9566 initrds has been added. This moves some minimal logic from
9567 the various initrd implementations into systemd proper.
9569 * The journal files are now owned by a new group
9570 "systemd-journal", which exists specifically to allow access
9571 to the journal, and nothing else. Previously, we used the
9572 "adm" group for that, which however possibly covers more
9573 than just journal/log file access. This new group is now
9574 already used by systemd-journal-gatewayd to ensure this
9575 daemon gets access to the journal files and as little else
9576 as possible. Note that "make install" will also set FS ACLs
9577 up for /var/log/journal to give "adm" and "wheel" read
9578 access to it, in addition to "systemd-journal" which owns
9579 the journal files. We recommend that packaging scripts also
9580 add read access to "adm" + "wheel" to /var/log/journal, and
9581 all existing/future journal files. To normal users and
9582 administrators little changes, however packagers need to
9583 ensure to create the "systemd-journal" system group at
9584 package installation time.
9586 * The systemd-journal-gatewayd now runs as unprivileged user
9587 systemd-journal-gateway:systemd-journal-gateway. Packaging
9588 scripts need to create these system user/group at
9591 * timedated now exposes a new boolean property CanNTP that
9592 indicates whether a local NTP service is available or not.
9594 * systemd-detect-virt will now also detect xen PVs
9596 * The pstore file system is now mounted by default, if it is
9599 * In addition to the SELinux and IMA policies we will now also
9600 load SMACK policies at early boot.
9602 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
9603 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
9604 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
9605 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
9606 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
9607 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
9608 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
9609 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
9610 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
9611 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
9612 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
9613 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
9614 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
9615 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
9619 * Timer units now support calendar time events in addition to
9620 monotonic time events. That means you can now trigger a unit
9621 based on a calendar time specification such as "Thu,Fri
9622 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
9623 or fifth day of any month of the year 2013, given that it is
9624 a thursday or friday. This brings timer event support
9625 considerably closer to cron's capabilities. For details on
9626 the supported calendar time specification language see
9629 * udev now supports a number of different naming policies for
9630 network interfaces for predictable names, and a combination
9631 of these policies is now the default. Please see this wiki
9632 document for details:
9634 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
9636 * Auke Kok's bootchart implementation has been added to the
9637 systemd tree. It is an optional component that can graph the
9638 boot in quite some detail. It is one of the best bootchart
9639 implementations around and minimal in its code and
9642 * nss-myhostname has been integrated into the systemd source
9643 tree. nss-myhostname guarantees that the local hostname
9644 always stays resolvable via NSS. It has been a weak
9645 requirement of systemd-hostnamed since a long time, and
9646 since its code is actually trivial we decided to just
9647 include it in systemd's source tree. It can be turned off
9648 with a configure switch.
9650 * The read-ahead logic is now capable of properly detecting
9651 whether a btrfs file system is on SSD or rotating media, in
9652 order to optimize the read-ahead scheme. Previously, it was
9653 only capable of detecting this on traditional file systems
9656 * In udev, additional device properties are now read from the
9657 IAB in addition to the OUI database. Also, Bluetooth company
9658 identities are attached to the devices as well.
9660 * In service files %U may be used as specifier that is
9661 replaced by the configured user name of the service.
9663 * nspawn may now be invoked without a controlling TTY. This
9664 makes it suitable for invocation as its own service. This
9665 may be used to set up a simple containerized server system
9666 using only core OS tools.
9668 * systemd and nspawn can now accept socket file descriptors
9669 when they are started for socket activation. This enables
9670 implementation of socket activated nspawn
9671 containers. i.e. think about autospawning an entire OS image
9672 when the first SSH or HTTP connection is received. We expect
9673 that similar functionality will also be added to libvirt-lxc
9676 * journalctl will now suppress ANSI color codes when
9677 presenting log data.
9679 * systemctl will no longer show control group information for
9680 a unit if the control group is empty anyway.
9682 * logind can now automatically suspend/hibernate/shutdown the
9685 * /etc/machine-info and hostnamed now also expose the chassis
9686 type of the system. This can be used to determine whether
9687 the local system is a laptop, desktop, handset or
9688 tablet. This information may either be configured by the
9689 user/vendor or is automatically determined from ACPI and DMI
9690 information if possible.
9692 * A number of polkit actions are now bound together with "imply"
9693 rules. This should simplify creating UIs because many actions
9694 will now authenticate similar ones as well.
9696 * Unit files learnt a new condition ConditionACPower= which
9697 may be used to conditionalize a unit depending on whether an
9698 AC power source is connected or not, of whether the system
9699 is running on battery power.
9701 * systemctl gained a new "is-failed" verb that may be used in
9702 shell scripts and suchlike to check whether a specific unit
9703 is in the "failed" state.
9705 * The EnvironmentFile= setting in unit files now supports file
9706 globbing, and can hence be used to easily read a number of
9707 environment files at once.
9709 * systemd will no longer detect and recognize specific
9710 distributions. All distribution-specific #ifdeffery has been
9711 removed, systemd is now fully generic and
9712 distribution-agnostic. Effectively, not too much is lost as
9713 a lot of the code is still accessible via explicit configure
9714 switches. However, support for some distribution specific
9715 legacy configuration file formats has been dropped. We
9716 recommend distributions to simply adopt the configuration
9717 files everybody else uses now and convert the old
9718 configuration from packaging scripts. Most distributions
9719 already did that. If that's not possible or desirable,
9720 distributions are welcome to forward port the specific
9721 pieces of code locally from the git history.
9723 * When logging a message about a unit systemd will now always
9724 log the unit name in the message meta data.
9726 * localectl will now also discover system locale data that is
9727 not stored in locale archives, but directly unpacked.
9729 * logind will no longer unconditionally use framebuffer
9730 devices as seat masters, i.e. as devices that are required
9731 to be existing before a seat is considered preset. Instead,
9732 it will now look for all devices that are tagged as
9733 "seat-master" in udev. By default, framebuffer devices will
9734 be marked as such, but depending on local systems, other
9735 devices might be marked as well. This may be used to
9736 integrate graphics cards using closed source drivers (such
9737 as NVidia ones) more nicely into logind. Note however, that
9738 we recommend using the open source NVidia drivers instead,
9739 and no udev rules for the closed-source drivers will be
9740 shipped from us upstream.
9742 Contributions from: Adam Williamson, Alessandro Crismani, Auke
9743 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
9744 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
9745 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
9746 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
9747 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
9748 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
9749 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
9750 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
9751 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
9752 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
9753 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
9758 * udev gained support for loading additional device properties
9759 from an indexed database that is keyed by vendor/product IDs
9760 and similar device identifiers. For the beginning this
9761 "hwdb" is populated with data from the well-known PCI and
9762 USB database, but also includes PNP, ACPI and OID data. In
9763 the longer run this indexed database shall grow into
9764 becoming the one central database for non-essential
9765 userspace device metadata. Previously, data from the PCI/USB
9766 database was only attached to select devices, since the
9767 lookup was a relatively expensive operation due to O(n) time
9768 complexity (with n being the number of entries in the
9769 database). Since this is now O(1), we decided to add in this
9770 data for all devices where this is available, by
9771 default. Note that the indexed database needs to be rebuilt
9772 when new data files are installed. To achieve this you need
9773 to update your packaging scripts to invoke "udevadm hwdb
9774 --update" after installation of hwdb data files. For
9775 RPM-based distributions we introduced the new
9776 %udev_hwdb_update macro for this purpose.
9778 * The Journal gained support for the "Message Catalog", an
9779 indexed database to link up additional information with
9780 journal entries. For further details please check:
9782 https://www.freedesktop.org/wiki/Software/systemd/catalog
9784 The indexed message catalog database also needs to be
9785 rebuilt after installation of message catalog files. Use
9786 "journalctl --update-catalog" for this. For RPM-based
9787 distributions we introduced the %journal_catalog_update
9788 macro for this purpose.
9790 * The Python Journal bindings gained support for the standard
9791 Python logging framework.
9793 * The Journal API gained new functions for checking whether
9794 the underlying file system of a journal file is capable of
9795 properly reporting file change notifications, or whether
9796 applications that want to reflect journal changes "live"
9797 need to recheck journal files continuously in appropriate
9800 * It is now possible to set the "age" field for tmpfiles
9801 entries to 0, indicating that files matching this entry
9802 shall always be removed when the directories are cleaned up.
9804 * coredumpctl gained a new "gdb" verb which invokes gdb
9805 right-away on the selected coredump.
9807 * There's now support for "hybrid sleep" on kernels that
9808 support this, in addition to "suspend" and "hibernate". Use
9809 "systemctl hybrid-sleep" to make use of this.
9811 * logind's HandleSuspendKey= setting (and related settings)
9812 now gained support for a new "lock" setting to simply
9813 request the screen lock on all local sessions, instead of
9814 actually executing a suspend or hibernation.
9816 * systemd will now mount the EFI variables file system by
9819 * Socket units now gained support for configuration of the
9820 SMACK security label.
9822 * timedatectl will now output the time of the last and next
9823 daylight saving change.
9825 * We dropped support for various legacy and distro-specific
9826 concepts, such as insserv, early-boot SysV services
9827 (i.e. those for non-standard runlevels such as 'b' or 'S')
9828 or ArchLinux /etc/rc.conf support. We recommend the
9829 distributions who still need support this to either continue
9830 to maintain the necessary patches downstream, or find a
9831 different solution. (Talk to us if you have questions!)
9833 * Various systemd components will now bypass polkit checks for
9834 root and otherwise handle properly if polkit is not found to
9835 be around. This should fix most issues for polkit-less
9836 systems. Quite frankly this should have been this way since
9837 day one. It is absolutely our intention to make systemd work
9838 fine on polkit-less systems, and we consider it a bug if
9839 something does not work as it should if polkit is not around.
9841 * For embedded systems it is now possible to build udev and
9842 systemd without blkid and/or kmod support.
9844 * "systemctl switch-root" is now capable of switching root
9845 more than once. I.e. in addition to transitions from the
9846 initrd to the host OS it is now possible to transition to
9847 further OS images from the host. This is useful to implement
9848 offline updating tools.
9850 * Various other additions have been made to the RPM macros
9851 shipped with systemd. Use %udev_rules_update() after
9852 installing new udev rules files. %_udevhwdbdir,
9853 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
9854 %_sysctldir are now available which resolve to the right
9855 directories for packages to place various data files in.
9857 * journalctl gained the new --full switch (in addition to
9858 --all, to disable ellipsation for long messages.
9860 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
9861 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
9862 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
9863 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
9864 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
9865 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
9866 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
9867 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
9868 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
9872 * journalctl gained new --since= and --until= switches to
9873 filter by time. It also now supports nice filtering for
9874 units via --unit=/-u.
9876 * Type=oneshot services may use ExecReload= and do the
9879 * The journal daemon now supports time-based rotation and
9880 vacuuming, in addition to the usual disk-space based
9883 * The journal will now index the available field values for
9884 each field name. This enables clients to show pretty drop
9885 downs of available match values when filtering. The bash
9886 completion of journalctl has been updated
9887 accordingly. journalctl gained a new switch -F to list all
9888 values a certain field takes in the journal database.
9890 * More service events are now written as structured messages
9891 to the journal, and made recognizable via message IDs.
9893 * The timedated, localed and hostnamed mini-services which
9894 previously only provided support for changing time, locale
9895 and hostname settings from graphical DEs such as GNOME now
9896 also have a minimal (but very useful) text-based client
9897 utility each. This is probably the nicest way to changing
9898 these settings from the command line now, especially since
9899 it lists available options and is fully integrated with bash
9902 * There's now a new tool "systemd-coredumpctl" to list and
9903 extract coredumps from the journal.
9905 * We now install a README each in /var/log/ and
9906 /etc/rc.d/init.d explaining where the system logs and init
9907 scripts went. This hopefully should help folks who go to
9908 that dirs and look into the otherwise now empty void and
9909 scratch their heads.
9911 * When user-services are invoked (by systemd --user) the
9912 $MANAGERPID env var is set to the PID of systemd.
9914 * SIGRTMIN+24 when sent to a --user instance will now result
9915 in immediate termination of systemd.
9917 * gatewayd received numerous feature additions such as a
9918 "follow" mode, for live syncing and filtering.
9920 * browse.html now allows filtering and showing detailed
9921 information on specific entries. Keyboard navigation and
9922 mouse screen support has been added.
9924 * gatewayd/journalctl now supports HTML5/JSON
9925 Server-Sent-Events as output.
9927 * The SysV init script compatibility logic will now
9928 heuristically determine whether a script supports the
9929 "reload" verb, and only then make this available as
9932 * "systemctl status --follow" has been removed, use "journalctl
9935 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
9936 have been removed since they are hardly useful to be
9939 * And I'd like to take the opportunity to specifically mention
9940 Zbigniew for his great contributions. Zbigniew, you rock!
9942 Contributions from: Andrew Eikum, Christian Hesse, Colin
9943 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
9944 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
9945 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
9946 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
9947 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
9948 Jędrzejewski-Szmek, Сковорода Никита Андреевич
9952 * If /etc/vconsole.conf is non-existent or empty we will no
9953 longer load any console font or key map at boot by
9954 default. Instead the kernel defaults will be left
9955 intact. This is definitely the right thing to do, as no
9956 configuration should mean no configuration, and hard-coding
9957 font names that are different on all archs is probably a bad
9958 idea. Also, the kernel default key map and font should be
9959 good enough for most cases anyway, and mostly identical to
9960 the userspace fonts/key maps we previously overloaded them
9961 with. If distributions want to continue to default to a
9962 non-kernel font or key map they should ship a default
9963 /etc/vconsole.conf with the appropriate contents.
9965 Contributions from: Colin Walters, Daniel J Walsh, Dave
9966 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
9967 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9971 * journalctl gained a new --cursor= switch to show entries
9972 starting from the specified location in the journal.
9974 * We now enforce a size limit on journal entry fields exported
9975 with "-o json" in journalctl. Fields larger than 4K will be
9976 assigned null. This can be turned off with --all.
9978 * An (optional) journal gateway daemon is now available as
9979 "systemd-journal-gatewayd.service". This service provides
9980 access to the journal via HTTP and JSON. This functionality
9981 will be used to implement live log synchronization in both
9982 pull and push modes, but has various other users too, such
9983 as easy log access for debugging of embedded devices. Right
9984 now it is already useful to retrieve the journal via HTTP:
9986 # systemctl start systemd-journal-gatewayd.service
9987 # wget http://localhost:19531/entries
9989 This will download the journal contents in a
9990 /var/log/messages compatible format. The same as JSON:
9992 # curl -H"Accept: application/json" http://localhost:19531/entries
9994 This service is also accessible via a web browser where a
9995 single static HTML5 app is served that uses the JSON logic
9996 to enable the user to do some basic browsing of the
9997 journal. This will be extended later on. Here's an example
9998 screenshot of this app in its current state:
10000 http://0pointer.de/public/journal-gatewayd
10002 Contributions from: Kay Sievers, Lennart Poettering, Robert
10003 Milasan, Tom Gundersen
10007 * The bash completion logic is now available for journalctl
10010 * We do not mount the "cpuset" controller anymore together with
10011 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
10012 started if no parameters are assigned to it. "cpuset" hence
10013 broke code that assumed it could create "cpu" groups and
10016 * journalctl -f will now subscribe to terminal size changes,
10017 and line break accordingly.
10019 Contributions from: Dave Reisner, Kay Sievers, Lennart
10020 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
10024 * nspawn will now create a symlink /etc/localtime in the
10025 container environment, copying the host's timezone
10026 setting. Previously this has been done via a bind mount, but
10027 since symlinks cannot be bind mounted this has now been
10028 changed to create/update the appropriate symlink.
10030 * journalctl -n's line number argument is now optional, and
10031 will default to 10 if omitted.
10033 * journald will now log the maximum size the journal files may
10034 take up on disk. This is particularly useful if the default
10035 built-in logic of determining this parameter from the file
10036 system size is used. Use "systemctl status
10037 systemd-journald.service" to see this information.
10039 * The multi-seat X wrapper tool has been stripped down. As X
10040 is now capable of enumerating graphics devices via udev in a
10041 seat-aware way the wrapper is not strictly necessary
10042 anymore. A stripped down temporary stop-gap is still shipped
10043 until the upstream display managers have been updated to
10044 fully support the new X logic. Expect this wrapper to be
10045 removed entirely in one of the next releases.
10047 * HandleSleepKey= in logind.conf has been split up into
10048 HandleSuspendKey= and HandleHibernateKey=. The old setting
10049 is not available anymore. X11 and the kernel are
10050 distinguishing between these keys and we should too. This
10051 also means the inhibition lock for these keys has been split
10054 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
10055 Poettering, Lukas Nykryn, Václav Pavlín
10059 * Whenever a unit changes state we will now log this to the
10060 journal and show along the unit's own log output in
10061 "systemctl status".
10063 * ConditionPathIsMountPoint= can now properly detect bind
10064 mount points too. (Previously, a bind mount of one file
10065 system to another place in the same file system could not be
10066 detected as mount, since they shared struct stat's st_dev
10069 * We will now mount the cgroup controllers cpu, cpuacct,
10070 cpuset and the controllers net_cls, net_prio together by
10073 * nspawn containers will now have a virtualized boot
10074 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
10075 over with a randomized ID at container initialization). This
10076 has the effect of making "journalctl -b" do the right thing
10079 * The JSON output journal serialization has been updated not
10080 to generate "endless" list objects anymore, but rather one
10081 JSON object per line. This is more in line how most JSON
10082 parsers expect JSON objects. The new output mode
10083 "json-pretty" has been added to provide similar output, but
10084 neatly aligned for readability by humans.
10086 * We dropped all explicit sync() invocations in the shutdown
10087 code. The kernel does this implicitly anyway in the kernel
10088 reboot() syscall. halt(8)'s -n option is now a compatibility
10091 * We now support virtualized reboot() in containers, as
10092 supported by newer kernels. We will fall back to exit() if
10093 CAP_SYS_REBOOT is not available to the container. Also,
10094 nspawn makes use of this now and will actually reboot the
10095 container if the containerized OS asks for that.
10097 * journalctl will only show local log output by default
10098 now. Use --merge (-m) to show remote log output, too.
10100 * libsystemd-journal gained the new sd_journal_get_usage()
10101 call to determine the current disk usage of all journal
10102 files. This is exposed in the new "journalctl --disk-usage"
10105 * journald gained a new configuration setting SplitMode= in
10106 journald.conf which may be used to control how user journals
10107 are split off. See journald.conf(5) for details.
10109 * A new condition type ConditionFileNotEmpty= has been added.
10111 * tmpfiles' "w" lines now support file globbing, to write
10112 multiple files at once.
10114 * We added Python bindings for the journal submission
10115 APIs. More Python APIs for a number of selected APIs will
10116 likely follow. Note that we intend to add native bindings
10117 only for the Python language, as we consider it common
10118 enough to deserve bindings shipped within systemd. There are
10119 various projects outside of systemd that provide bindings
10120 for languages such as PHP or Lua.
10122 * Many conditions will now resolve specifiers such as %i. In
10123 addition, PathChanged= and related directives of .path units
10124 now support specifiers as well.
10126 * There's now a new RPM macro definition for the system preset
10129 * journald will now warn if it ca not forward a message to the
10130 syslog daemon because its socket is full.
10132 * timedated will no longer write or process /etc/timezone,
10133 except on Debian. As we do not support late mounted /usr
10134 anymore /etc/localtime always being a symlink is now safe,
10135 and hence the information in /etc/timezone is not necessary
10138 * logind will now always reserve one VT for a text getty (VT6
10139 by default). Previously if more than 6 X sessions where
10140 started they took up all the VTs with auto-spawned gettys,
10141 so that no text gettys were available anymore.
10143 * udev will now automatically inform the btrfs kernel logic
10144 about btrfs RAID components showing up. This should make
10145 simple hotplug based btrfs RAID assembly work.
10147 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
10148 (but not for its children which will stay at the kernel
10149 default). This should allow setups with a lot more listening
10152 * systemd will now always pass the configured timezone to the
10153 kernel at boot. timedated will do the same when the timezone
10156 * logind's inhibition logic has been updated. By default,
10157 logind will now handle the lid switch, the power and sleep
10158 keys all the time, even in graphical sessions. If DEs want
10159 to handle these events on their own they should take the new
10160 handle-power-key, handle-sleep-key and handle-lid-switch
10161 inhibitors during their runtime. A simple way to achieve
10162 that is to invoke the DE wrapped in an invocation of:
10164 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
10166 * Access to unit operations is now checked via SELinux taking
10167 the unit file label and client process label into account.
10169 * systemd will now notify the administrator in the journal
10170 when he over-mounts a non-empty directory.
10172 * There are new specifiers that are resolved in unit files,
10173 for the hostname (%H), the machine ID (%m) and the boot ID
10176 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
10177 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
10178 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
10179 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
10180 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
10181 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
10182 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
10186 * Support for reading structured kernel messages from
10187 /dev/kmsg has now been added and is enabled by default.
10189 * Support for reading kernel messages from /proc/kmsg has now
10190 been removed. If you want kernel messages in the journal
10191 make sure to run a recent kernel (>= 3.5) that supports
10192 reading structured messages from /dev/kmsg (see
10193 above). /proc/kmsg is now exclusive property of classic
10194 syslog daemons again.
10196 * The libudev API gained the new
10197 udev_device_new_from_device_id() call.
10199 * The logic for file system namespace (ReadOnlyDirectory=,
10200 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
10201 require pivot_root() anymore. This means fewer temporary
10202 directories are created below /tmp for this feature.
10204 * nspawn containers will now see and receive all submounts
10205 made on the host OS below the root file system of the
10208 * Forward Secure Sealing is now supported for Journal files,
10209 which provide cryptographical sealing of journal files so
10210 that attackers cannot alter log history anymore without this
10211 being detectable. Lennart will soon post a blog story about
10212 this explaining it in more detail.
10214 * There are two new service settings RestartPreventExitStatus=
10215 and SuccessExitStatus= which allow configuration of exit
10216 status (exit code or signal) which will be excepted from the
10217 restart logic, resp. consider successful.
10219 * journalctl gained the new --verify switch that can be used
10220 to check the integrity of the structure of journal files and
10221 (if Forward Secure Sealing is enabled) the contents of
10224 * nspawn containers will now be run with /dev/stdin, /dev/fd/
10225 and similar symlinks pre-created. This makes running shells
10226 as container init process a lot more fun.
10228 * The fstab support can now handle PARTUUID= and PARTLABEL=
10231 * A new ConditionHost= condition has been added to match
10232 against the hostname (with globs) and machine ID. This is
10233 useful for clusters where a single OS image is used to
10234 provision a large number of hosts which shall run slightly
10235 different sets of services.
10237 * Services which hit the restart limit will now be placed in a
10240 Contributions from: Bertram Poettering, Dave Reisner, Huang
10241 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
10242 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
10246 * When running in --user mode systemd will now become a
10247 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
10248 tree a lot more organized.
10250 * A new PartOf= unit dependency type has been introduced that
10251 may be used to group services in a natural way.
10253 * "systemctl enable" may now be used to enable instances of
10256 * journalctl now prints error log levels in red, and
10257 warning/notice log levels in bright white. It also supports
10258 filtering by log level now.
10260 * cgtop gained a new -n switch (similar to top), to configure
10261 the maximum number of iterations to run for. It also gained
10262 -b, to run in batch mode (accepting no input).
10264 * The suffix ".service" may now be omitted on most systemctl
10265 command lines involving service unit names.
10267 * There's a new bus call in logind to lock all sessions, as
10268 well as a loginctl verb for it "lock-sessions".
10270 * libsystemd-logind.so gained a new call sd_journal_perror()
10271 that works similar to libc perror() but logs to the journal
10272 and encodes structured information about the error number.
10274 * /etc/crypttab entries now understand the new keyfile-size=
10277 * shutdown(8) now can send a (configurable) wall message when
10278 a shutdown is cancelled.
10280 * The mount propagation mode for the root file system will now
10281 default to "shared", which is useful to make containers work
10282 nicely out-of-the-box so that they receive new mounts from
10283 the host. This can be undone locally by running "mount
10284 --make-rprivate /" if needed.
10286 * The prefdm.service file has been removed. Distributions
10287 should maintain this unit downstream if they intend to keep
10288 it around. However, we recommend writing normal unit files
10289 for display managers instead.
10291 * Since systemd is a crucial part of the OS we will now
10292 default to a number of compiler switches that improve
10293 security (hardening) such as read-only relocations, stack
10294 protection, and suchlike.
10296 * The TimeoutSec= setting for services is now split into
10297 TimeoutStartSec= and TimeoutStopSec= to allow configuration
10298 of individual time outs for the start and the stop phase of
10301 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
10302 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
10303 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
10304 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
10305 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
10306 Gundersen, Zbigniew Jędrzejewski-Szmek
10310 * The journal and id128 C APIs are now fully documented as man
10313 * Extra safety checks have been added when transitioning from
10314 the initial RAM disk to the main system to avoid accidental
10317 * /etc/crypttab entries now understand the new keyfile-offset=
10320 * systemctl -t can now be used to filter by unit load state.
10322 * The journal C API gained the new sd_journal_wait() call to
10323 make writing synchronous journal clients easier.
10325 * journalctl gained the new -D switch to show journals from a
10326 specific directory.
10328 * journalctl now displays a special marker between log
10329 messages of two different boots.
10331 * The journal is now explicitly flushed to /var via a service
10332 systemd-journal-flush.service, rather than implicitly simply
10333 by seeing /var/log/journal to be writable.
10335 * journalctl (and the journal C APIs) can now match for much
10336 more complex expressions, with alternatives and
10339 * When transitioning from the initial RAM disk to the main
10340 system we will now kill all processes in a killing spree to
10341 ensure no processes stay around by accident.
10343 * Three new specifiers may be used in unit files: %u, %h, %s
10344 resolve to the user name, user home directory resp. user
10345 shell. This is useful for running systemd user instances.
10347 * We now automatically rotate journal files if their data
10348 object hash table gets a fill level > 75%. We also size the
10349 hash table based on the configured maximum file size. This
10350 together should lower hash collisions drastically and thus
10351 speed things up a bit.
10353 * journalctl gained the new "--header" switch to introspect
10354 header data of journal files.
10356 * A new setting SystemCallFilters= has been added to services which may
10357 be used to apply deny lists or allow lists to system calls. This is
10358 based on SECCOMP Mode 2 of Linux 3.5.
10360 * nspawn gained a new --link-journal= switch (and quicker: -j)
10361 to link the container journal with the host. This makes it
10362 very easy to centralize log viewing on the host for all
10363 guests while still keeping the journal files separated.
10365 * Many bugfixes and optimizations
10367 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
10368 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
10369 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
10374 * Several tools now understand kernel command line arguments,
10375 which are only read when run in an initial RAM disk. They
10376 usually follow closely their normal counterparts, but are
10379 * There's a new tool to analyze the readahead files that are
10380 automatically generated at boot. Use:
10382 /usr/lib/systemd/systemd-readahead analyze /.readahead
10384 * We now provide an early debug shell on tty9 if this enabled. Use:
10386 systemctl enable debug-shell.service
10388 * All plymouth related units have been moved into the Plymouth
10389 package. Please make sure to upgrade your Plymouth version
10392 * systemd-tmpfiles now supports getting passed the basename of
10393 a configuration file only, in which case it will look for it
10394 in all appropriate directories automatically.
10396 * udevadm info now takes a /dev or /sys path as argument, and
10397 does the right thing. Example:
10399 udevadm info /dev/sda
10400 udevadm info /sys/class/block/sda
10402 * systemctl now prints a warning if a unit is stopped but a
10403 unit that might trigger it continues to run. Example: a
10404 service is stopped but the socket that activates it is left
10407 * "systemctl status" will now mention if the log output was
10408 shortened due to rotation since a service has been started.
10410 * The journal API now exposes functions to determine the
10411 "cutoff" times due to rotation.
10413 * journald now understands SIGUSR1 and SIGUSR2 for triggering
10414 immediately flushing of runtime logs to /var if possible,
10415 resp. for triggering immediate rotation of the journal
10418 * It is now considered an error if a service is attempted to
10419 be stopped that is not loaded.
10421 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
10423 * systemd-analyze now supports Python 3
10425 * tmpfiles now supports cleaning up directories via aging
10426 where the first level dirs are always kept around but
10427 directories beneath it automatically aged. This is enabled
10428 by prefixing the age field with '~'.
10430 * Seat objects now expose CanGraphical, CanTTY properties
10431 which is required to deal with very fast bootups where the
10432 display manager might be running before the graphics drivers
10433 completed initialization.
10435 * Seat objects now expose a State property.
10437 * We now include RPM macros for service enabling/disabling
10438 based on the preset logic. We recommend RPM based
10439 distributions to make use of these macros if possible. This
10440 makes it simpler to reuse RPM spec files across
10443 * We now make sure that the collected systemd unit name is
10444 always valid when services log to the journal via
10447 * There's a new man page kernel-command-line(7) detailing all
10448 command line options we understand.
10450 * The fstab generator may now be disabled at boot by passing
10451 fstab=0 on the kernel command line.
10453 * A new kernel command line option modules-load= is now understood
10454 to load a specific kernel module statically, early at boot.
10456 * Unit names specified on the systemctl command line are now
10457 automatically escaped as needed. Also, if file system or
10458 device paths are specified they are automatically turned
10459 into the appropriate mount or device unit names. Example:
10461 systemctl status /home
10462 systemctl status /dev/sda
10464 * The SysVConsole= configuration option has been removed from
10465 system.conf parsing.
10467 * The SysV search path is no longer exported on the D-Bus
10470 * The Names= option has been removed from unit file parsing.
10472 * There's a new man page bootup(7) detailing the boot process.
10474 * Every unit and every generator we ship with systemd now
10475 comes with full documentation. The self-explanatory boot is
10478 * A couple of services gained "systemd-" prefixes in their
10479 name if they wrap systemd code, rather than only external
10480 code. Among them fsck@.service which is now
10481 systemd-fsck@.service.
10483 * The HaveWatchdog property has been removed from the D-Bus
10486 * systemd.confirm_spawn= on the kernel command line should now
10489 * There's a new man page crypttab(5) which details all options
10490 we actually understand.
10492 * systemd-nspawn gained a new --capability= switch to pass
10493 additional capabilities to the container.
10495 * timedated will now read known NTP implementation unit names
10496 from /usr/lib/systemd/ntp-units.d/*.list,
10497 systemd-timedated-ntp.target has been removed.
10499 * journalctl gained a new switch "-b" that lists log data of
10500 the current boot only.
10502 * The notify socket is in the abstract namespace again, in
10503 order to support daemons which chroot() at start-up.
10505 * There is a new Storage= configuration option for journald
10506 which allows configuration of where log data should go. This
10507 also provides a way to disable journal logging entirely, so
10508 that data collected is only forwarded to the console, the
10509 kernel log buffer or another syslog implementation.
10511 * Many bugfixes and optimizations
10513 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
10514 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
10515 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
10516 Shawn Landden, Tom Gundersen
10520 * "systemctl help <unit>" now shows the man page if one is
10523 * Several new man pages have been added.
10525 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
10526 MaxLevelConsole= can now be specified in
10527 journald.conf. These options allow reducing the amount of
10528 data stored on disk or forwarded by the log level.
10530 * TimerSlackNSec= can now be specified in system.conf for
10531 PID1. This allows system-wide power savings.
10533 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
10534 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
10539 * logind is now capable of (optionally) handling power and
10540 sleep keys as well as the lid switch.
10542 * journalctl now understands the syntax "journalctl
10543 /usr/bin/avahi-daemon" to get all log output of a specific
10546 * CapabilityBoundingSet= in system.conf now also influences
10547 the capability bound set of usermode helpers of the kernel.
10549 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
10550 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
10551 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
10552 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
10556 * Note that we skipped 139 releases here in order to set the
10557 new version to something that is greater than both udev's
10558 and systemd's most recent version number.
10560 * udev: all udev sources are merged into the systemd source tree now.
10561 All future udev development will happen in the systemd tree. It
10562 is still fully supported to use the udev daemon and tools without
10563 systemd running, like in initramfs or other init systems. Building
10564 udev though, will require the *build* of the systemd tree, but
10565 udev can be properly *run* without systemd.
10567 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
10568 should be used to create dead device nodes as workarounds for broken
10571 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
10572 no longer supported. udev_monitor_new_from_netlink() needs to be
10573 used to subscribe to events.
10575 * udev: when udevd is started by systemd, processes which are left
10576 behind by forking them off of udev rules, are unconditionally cleaned
10577 up and killed now after the event handling has finished. Services or
10578 daemons must be started as systemd services. Services can be
10579 pulled-in by udev to get started, but they can no longer be directly
10580 forked by udev rules.
10582 * udev: the daemon binary is called systemd-udevd now and installed
10583 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
10584 to adapt to that, create symlink, or rename the binary after building
10587 * libudev no longer provides these symbols:
10588 udev_monitor_from_socket()
10589 udev_queue_get_failed_list_entry()
10590 udev_get_{dev,sys,run}_path()
10591 The versions number was bumped and symbol versioning introduced.
10593 * systemd-loginctl and systemd-journalctl have been renamed
10594 to loginctl and journalctl to match systemctl.
10596 * The config files: /etc/systemd/systemd-logind.conf and
10597 /etc/systemd/systemd-journald.conf have been renamed to
10598 logind.conf and journald.conf. Package updates should rename
10599 the files to the new names on upgrade.
10601 * For almost all files the license is now LGPL2.1+, changed
10602 from the previous GPL2.0+. Exceptions are some minor stuff
10603 of udev (which will be changed to LGPL2.1 eventually, too),
10604 and the MIT licensed sd-daemon.[ch] library that is suitable
10605 to be used as drop-in files.
10607 * systemd and logind now handle system sleep states, in
10608 particular suspending and hibernating.
10610 * logind now implements a sleep/shutdown/idle inhibiting logic
10611 suitable for a variety of uses. Soonishly Lennart will blog
10612 about this in more detail.
10614 * var-run.mount and var-lock.mount are no longer provided
10615 (which previously bind mounted these directories to their new
10616 places). Distributions which have not converted these
10617 directories to symlinks should consider stealing these files
10618 from git history and add them downstream.
10620 * We introduced the Documentation= field for units and added
10621 this to all our shipped units. This is useful to make it
10622 easier to explore the boot and the purpose of the various
10625 * All smaller setup units (such as
10626 systemd-vconsole-setup.service) now detect properly if they
10627 are run in a container and are skipped when
10628 appropriate. This guarantees an entirely noise-free boot in
10629 Linux container environments such as systemd-nspawn.
10631 * A framework for implementing offline system updates is now
10632 integrated, for details see:
10633 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
10635 * A new service type Type=idle is available now which helps us
10636 avoiding ugly interleaving of getty output and boot status
10639 * There's now a system-wide CapabilityBoundingSet= option to
10640 globally reduce the set of capabilities for the
10641 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
10642 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
10643 even CAP_NET_ADMIN system-wide for secure systems.
10645 * There are now system-wide DefaultLimitXXX= options to
10646 globally change the defaults of the various resource limits
10647 for all units started by PID 1.
10649 * Harald Hoyer's systemd test suite has been integrated into
10650 systemd which allows easy testing of systemd builds in qemu
10651 and nspawn. (This is really awesome! Ask us for details!)
10653 * The fstab parser is now implemented as generator, not inside
10656 * systemctl will now warn you if .mount units generated from
10657 /etc/fstab are out of date due to changes in fstab that
10658 have not been read by systemd yet.
10660 * systemd is now suitable for usage in initrds. Dracut has
10661 already been updated to make use of this. With this in place
10662 initrds get a slight bit faster but primarily are much
10663 easier to introspect and debug since "systemctl status" in
10664 the host system can be used to introspect initrd services,
10665 and the journal from the initrd is kept around too.
10667 * systemd-delta has been added, a tool to explore differences
10668 between user/admin configuration and vendor defaults.
10670 * PrivateTmp= now affects both /tmp and /var/tmp.
10672 * Boot time status messages are now much prettier and feature
10673 proper english language. Booting up systemd has never been
10676 * Read-ahead pack files now include the inode number of all
10677 files to pre-cache. When the inode changes the pre-caching
10678 is not attempted. This should be nicer to deal with updated
10679 packages which might result in changes of read-ahead
10682 * We now temporaritly lower the kernel's read_ahead_kb variable
10683 when collecting read-ahead data to ensure the kernel's
10684 built-in read-ahead does not add noise to our measurements
10685 of necessary blocks to pre-cache.
10687 * There's now RequiresMountsFor= to add automatic dependencies
10688 for all mounts necessary for a specific file system path.
10690 * MountAuto= and SwapAuto= have been removed from
10691 system.conf. Mounting file systems at boot has to take place
10694 * nspawn now learned a new switch --uuid= to set the machine
10695 ID on the command line.
10697 * nspawn now learned the -b switch to automatically search
10698 for an init system.
10700 * vt102 is now the default TERM for serial TTYs, upgraded from
10703 * systemd-logind now works on VT-less systems.
10705 * The build tree has been reorganized. The individual
10706 components now have directories of their own.
10708 * A new condition type ConditionPathIsReadWrite= is now available.
10710 * nspawn learned the new -C switch to create cgroups for the
10711 container in other hierarchies.
10713 * We now have support for hardware watchdogs, configurable in
10716 * The scheduled shutdown logic now has a public API.
10718 * We now mount /tmp as tmpfs by default, but this can be
10719 masked and /etc/fstab can override it.
10721 * Since udisks does not make use of /media anymore we are not
10722 mounting a tmpfs on it anymore.
10724 * journalctl gained a new --local switch to only interleave
10725 locally generated journal files.
10727 * We can now load the IMA policy at boot automatically.
10729 * The GTK tools have been split off into a systemd-ui.
10731 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
10732 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
10733 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
10734 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
10735 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
10736 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
10737 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
10738 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
10739 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
10744 * This is mostly a bugfix release
10746 * Support optional initialization of the machine ID from the
10747 KVM or container configured UUID.
10749 * Support immediate reboots with "systemctl reboot -ff"
10751 * Show /etc/os-release data in systemd-analyze output
10753 * Many bugfixes for the journal, including endianness fixes and
10754 ensuring that disk space enforcement works
10756 * sd-login.h is C++ compatible again
10758 * Extend the /etc/os-release format on request of the Debian
10761 * We now refuse non-UTF8 strings used in various configuration
10762 and unit files. This is done to ensure we do not pass invalid
10763 data over D-Bus or expose it elsewhere.
10765 * Register Mimo USB Screens as suitable for automatic seat
10768 * Read SELinux client context from journal clients in a race
10771 * Reorder configuration file lookup order. /etc now always
10772 overrides /run in order to allow the administrator to always
10773 and unconditionally override vendor-supplied or
10774 automatically generated data.
10776 * The various user visible bits of the journal now have man
10777 pages. We still lack man pages for the journal API calls
10780 * We now ship all man pages in HTML format again in the
10783 Contributions from: Dave Reisner, Dirk Eibach, Frederic
10784 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
10785 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
10790 * This is mostly a bugfix release
10792 * systems lacking /etc/os-release are no longer supported.
10794 * Various functionality updates to libsystemd-login.so
10796 * Track class of PAM logins to distinguish greeters from
10797 normal user logins.
10799 Contributions from: Kay Sievers, Lennart Poettering, Michael
10804 * This is an important bugfix release for v41.
10806 * Building man pages is now optional which should be useful
10807 for those building systemd from git but unwilling to install
10810 * Watchdog support for supervising services is now usable. In
10811 a future release support for hardware watchdogs
10812 (i.e. /dev/watchdog) will be added building on this.
10814 * Service start rate limiting is now configurable and can be
10815 turned off per service. When a start rate limit is hit a
10816 reboot can automatically be triggered.
10818 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
10820 Contributions from: Benjamin Franzke, Bill Nottingham,
10821 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
10822 Schmidt, Michał Górny, Piotr Drąg
10826 * The systemd binary is installed /usr/lib/systemd/systemd now;
10827 An existing /sbin/init symlink needs to be adapted with the
10830 * The code that loads kernel modules has been ported to invoke
10831 libkmod directly, instead of modprobe. This means we do not
10832 support systems with module-init-tools anymore.
10834 * Watchdog support is now already useful, but still not
10837 * A new kernel command line option systemd.setenv= is
10838 understood to set system wide environment variables
10839 dynamically at boot.
10841 * We now limit the set of capabilities of systemd-journald.
10843 * We now set SIGPIPE to ignore by default, since it only is
10844 useful in shell pipelines, and has little use in general
10845 code. This can be disabled with IgnoreSIPIPE=no in unit
10848 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
10849 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
10854 * This is mostly a bugfix release
10856 * We now expose the reason why a service failed in the
10857 "Result" D-Bus property.
10859 * Rudimentary service watchdog support (will be completed over
10860 the next few releases.)
10862 * When systemd forks off in order execute some service we will
10863 now immediately changes its argv[0] to reflect which process
10864 it will execute. This is useful to minimize the time window
10865 with a generic argv[0], which makes bootcharts more useful
10867 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
10868 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
10869 Mike Kazantsev, Ray Strode
10873 * This is mostly a test release, but incorporates many
10876 * New systemd-cgtop tool to show control groups by their
10879 * Linking against libacl for ACLs is optional again. If
10880 disabled, support tracking device access for active logins
10881 goes becomes unavailable, and so does access to the user
10882 journals by the respective users.
10884 * If a group "adm" exists, journal files are automatically
10885 owned by them, thus allow members of this group full access
10886 to the system journal as well as all user journals.
10888 * The journal now stores the SELinux context of the logging
10889 client for all entries.
10891 * Add C++ inclusion guards to all public headers
10893 * New output mode "cat" in the journal to print only text
10894 messages, without any meta data like date or time.
10896 * Include tiny X server wrapper as a temporary stop-gap to
10897 teach XOrg udev display enumeration. This is used by display
10898 managers such as gdm, and will go away as soon as XOrg
10899 learned native udev hotplugging for display devices.
10901 * Add new systemd-cat tool for executing arbitrary programs
10902 with STDERR/STDOUT connected to the journal. Can also act as
10903 BSD logger replacement, and does so by default.
10905 * Optionally store all locally generated coredumps in the
10906 journal along with meta data.
10908 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
10909 writing short strings to files (for usage for /sys), and for
10910 creating symlinks, character and block device nodes.
10912 * New unit file option ControlGroupPersistent= to make cgroups
10913 persistent, following the mechanisms outlined in
10914 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
10916 * Support multiple local RTCs in a sane way
10918 * No longer monopolize IO when replaying readahead data on
10919 rotating disks, since we might starve non-file-system IO to
10920 death, since fanotify() will not see accesses done by blkid,
10923 * Do not show kernel threads in systemd-cgls anymore, unless
10924 requested with new -k switch.
10926 Contributions from: Dan Horák, Kay Sievers, Lennart
10927 Poettering, Michal Schmidt
10931 * This is mostly a test release, but incorporates many
10934 * The git repository moved to:
10935 git://anongit.freedesktop.org/systemd/systemd
10936 ssh://git.freedesktop.org/git/systemd/systemd
10938 * First release with the journal
10939 http://0pointer.de/blog/projects/the-journal.html
10941 * The journal replaces both systemd-kmsg-syslogd and
10942 systemd-stdout-bridge.
10944 * New sd_pid_get_unit() API call in libsystemd-logind
10946 * Many systemadm clean-ups
10948 * Introduce remote-fs-pre.target which is ordered before all
10949 remote mounts and may be used to start services before all
10952 * Added Mageia support
10954 * Add bash completion for systemd-loginctl
10956 * Actively monitor PID file creation for daemons which exit in
10957 the parent process before having finished writing the PID
10958 file in the daemon process. Daemons which do this need to be
10959 fixed (i.e. PID file creation must have finished before the
10960 parent exits), but we now react a bit more gracefully to them.
10962 * Add colourful boot output, mimicking the well-known output
10963 of existing distributions.
10965 * New option PassCredentials= for socket units, for
10966 compatibility with a recent kernel ABI breakage.
10968 * /etc/rc.local is now hooked in via a generator binary, and
10969 thus will no longer act as synchronization point during
10972 * systemctl list-unit-files now supports --root=.
10974 * systemd-tmpfiles now understands two new commands: z, Z for
10975 relabelling files according to the SELinux database. This is
10976 useful to apply SELinux labels to specific files in /sys,
10977 among other things.
10979 * Output of SysV services is now forwarded to both the console
10980 and the journal by default, not only just the console.
10982 * New man pages for all APIs from libsystemd-login.
10984 * The build tree got reorganized and the build system is a
10985 lot more modular allowing embedded setups to specifically
10986 select the components of systemd they are interested in.
10988 * Support for Linux systems lacking the kernel VT subsystem is
10991 * configure's --with-rootdir= got renamed to
10992 --with-rootprefix= to follow the naming used by udev and
10995 * Unless specified otherwise we will now install to /usr instead
10996 of /usr/local by default.
10998 * Processes with '@' in argv[0][0] are now excluded from the
10999 final shut-down killing spree, following the logic explained
11001 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
11003 * All processes remaining in a service cgroup when we enter
11004 the START or START_PRE states are now killed with
11005 SIGKILL. That means it is no longer possible to spawn
11006 background processes from ExecStart= lines (which was never
11007 supported anyway, and bad style).
11009 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
11010 reloading of units together.
11012 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
11013 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
11014 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
11015 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
11016 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek