1 systemd System and Service Manager
5 * The service manager gained basic support for cgroup v2 freezer. Units
6 can now be suspended or resumed either using new systemctl verbs,
7 freeze and thaw respectively, or via D-Bus.
9 * PID 1 may now automatically load pre-compiled AppArmor policies from
10 /etc/apparmor/earlypolicy during early boot.
12 * The CPUAffinity= setting in service unit files now supports a new
13 special value "numa" that causes the CPU affinity masked to be set
14 based on the NUMA mask.
16 * systemd will now log about all left-over processes remaining in a
17 unit when the unit is stopped. It will now warn about services using
18 KillMode=none, as this is generally an unsafe thing to make use of.
20 * Two new unit file settings
21 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
22 added. They may be used to check whether a specific file system path
23 resides on a block device that is encrypted on the block level
24 (i.e. using dm-crypt/LUKS).
26 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
27 has been added that may be used for simple environment checks. This
28 is particularly useful when passing in environment variables from a
29 container manager (or from PAM in case of the systemd --user
32 * .service unit files now accept a new setting CoredumpFilter= which
33 allows configuration of the memory sections coredumps of the
34 service's processes shall include.
36 * .mount units gained a new ReadWriteOnly= boolean option. If set
37 it will not be attempted to mount a file system read-only if mounting
38 in read-write mode doesn't succeed. An option x-systemd.rw-only is
39 available in /etc/fstab to control the same.
41 * .socket units gained a new boolean setting PassPacketInfo=. If
42 enabled, the kernel will attach additional per-packet metadata to all
43 packets read from the socket, as an ancillary message. This controls
44 the IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
45 depending on socket type.
47 * .service units gained a new setting RootHash= which may be used to
48 specify the root hash for verity enabled disk images which are
49 specified in RootImage=. RootVerity= may be used to specify a path to
50 the Verity data matching a RootImage= file system. (The latter is
51 only useful for images that do not contain the Verity data embedded
52 into the same image that carries a GPT partition table following the
53 Discoverable Partition Specification). Similarly, systemd-nspawn
54 gained a new switch --verity-data= that takes a path to a file with
55 the verity data of the disk image supplied in --image=, if the image
56 doesn't contain the verity data itself.
58 * .service units gained a new setting RootHashSignature= which takes
59 either a base64 encoded PKCS#7 signature of the root hash specified
60 with RootHash=, or a path to a file to read the signature from. This
61 allows validation of the root hash against public keys available in
62 the kernel keyring, and is only supported on recent kernels
63 (>= 5.4)/libcryptsetup (>= 2.30). A similar switch has been added to
64 systemd-nspawn and systemd-dissect (--root-hash-sig=). Support for
65 this mechanism has also been added to systemd-veritysetup.
67 * .service unit files gained two new options
68 TimeoutStartFailureMode=/TimeoutStopFailureMode= that may be used to
69 tune behaviour if a start or stop timeout is hit, i.e. whether to
70 terminate the service with SIGTERM, SIGABRT or SIGKILL.
72 * Most options in systemd that accept hexadecimal values prefixed with
73 0x in additional to the usual decimal notation now also support octal
74 notation when the 0o prefix is used and binary notation if the 0b
77 * Various command line parameters and configuration file settings that
78 configure key or certificate files now optionally take paths to
79 AF_UNIX sockets in the file system. If configured that way a stream
80 connection is made to the socket and the required data read from
81 it. This is a simple and natural extension to the existing regular
82 file logic, and permits other software to provide keys or
83 certificates via simple IPC services, for example when unencrypted
84 storage on disk is not desired. Specifically, systemd-networkd's
85 Wireguard and MACSEC key file settings as well as
86 systemd-journal-gatewayd's and systemd-journal-remote's PEM
87 key/certificate parameters support this now.
89 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
90 configuration files that support specifier expansion learnt six new
91 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
92 resolve to the various ID fields from /etc/os-release, %l resolves to
93 the "short" hostname of the system, i.e. the hostname configured in
94 the kernel truncated at the first dot.
96 * Support for the .include syntax in unit files has been removed. The
97 concept has been obsolete for 6 years and we started warning about
98 its pending removal 2 years ago (also see NEWS file below). It's
101 * StandardError= and StandardOutput= in unit files no longer support
102 the "syslog" and "syslog-console" switches. They were long removed
103 from the documentation, but will now result in warnings when used,
104 and be converted to "journal" and "journal+console" automatically.
106 * If the service setting User= is set to the "nobody" user, a warning
107 message is now written to the logs (but the value is nonetheless
108 accepted). Setting User=nobody is unsafe, since the primary purpose
109 of the "nobody" user is to own all files whose owner cannot be mapped
110 locally. It's in particular used by the NFS subsystem and in user
111 namespacing. By running a service under this user's UID it might get
112 read and even write access to all these otherwise unmappable files,
113 which is quite likely a major security problem.
115 * tmpfs mounts automatically created by systemd (/tmp, /run, /dev/shm,
116 and others) now have a size and inode limits applied (50% of RAM for
117 /tmp and /dev/shm, 10% of RAM for other mounts, etc.)
119 * nss-mymachines lost support for resolution of users and groups, and
120 now only does resolution of hostnames. This functionality is now
121 provided by nss-systemd. Thus, the 'mymachines' entry should be
122 removed from the 'passwd:' and 'group:' lines in /etc/nsswitch.conf
123 (and 'systemd' added if it is not already there).
125 * A new kernel command line option systemd.hostname= has been added
126 that allows controlling the hostname that is initialized early during
129 * A kernel command line option "udev.blockdev_read_only" has been
130 added. If specified all hardware block devices that show up are
131 immediately marked as read-only by udev. This option is useful for
132 making sure that a specific boot under no circumstances modifies data
133 on disk. Use "blockdev --setrw" to undo the effect of this, per
136 * A new boolean kernel command line option systemd.swap= has been
137 added, which may be used to turn off automatic activation of swap
138 devices listed in /etc/fstab.
140 * New kernel command line options systemd.condition-needs-update= and
141 systemd.condition-first-boot= have been added, which override the
142 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
145 * A new kernel command line option systemd.clock-usec= has been added
146 that allows setting the system clock to the specified time in µs
147 since Jan 1st, 1970 early during boot. This is in particular useful
148 in order to make test cases more reliable.
150 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
151 systemd-coredump to save core files for suid processes. When saving
152 the core file, systemd-coredump will use the effective uid and gid of
153 the process that faulted.
155 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
156 now automatically set to "Y" at boot, in order to enable pstore
157 generation for collection with systemd-pstore.
159 * A new 'hwdb' file has been added that collects information about PCI
160 and USB devices that correctly support auto-suspend, on top of the
161 databases for this we import from the ChromiumOS project. If you have
162 a device that supports auto-suspend correctly and where it should be
163 enabled by default, please submit a patch that adds it to the
164 database (see /usr/lib/udev/hwdb.d/60-autosuspend.hwdb).
166 * systemd-udevd gained the new configuration option timeout_signal= as well
167 as a corresponding kernel command line option udev.timeout_signal=.
168 The option can be used to configure the UNIX signal that the main
169 daemon sends to the worker processes on timeout. Setting the signal
170 to SIGABRT is useful for debugging.
172 * .link files managed by systemd-udevd gained options RxFlowControl=,
173 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
174 order to configure various flow control parameters. They also gained
175 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
176 frame ring buffer sizes.
178 * networkd.conf gained a new boolean setting ManageForeignRoutes=. If
179 enabled systemd-networkd manages all routes configured by other tools.
181 * .network files managed by systemd-networkd gained a new section
182 [SR-IOV], in order to configure SR-IOV capable network devices.
184 * systemd-networkd's [IPv6Prefix] section in .network files gained a
185 new boolean setting Assign=. If enabled an address from the prefix is
186 automatically assigned to the interface.
188 * systemd-networkd gained a new section [DHCPv6PrefixDelegation] which
189 controls delegated prefixes assigned by DHCPv6 client. The section
190 has three settings: SubnetID=, Assign=, and Token=. The setting
191 SubnetID= allows explicit configuration of the preferred subnet that
192 systemd-networkd's Prefix Delegation logic assigns to interfaces. If
193 Assign= is enabled (which is the default) an address from any acquired
194 delegated prefix is automatically chosen and assigned to the
195 interface. The setting Token= specifies an optional address generation
198 * systemd-networkd's [Network] section gained a new setting
199 IPv4AcceptLocal=. If enabled the interface accepts packets with local
202 * systemd-networkd gained support for configuring the HTB queuing
203 discipline in the [HierarchyTokenBucket] and
204 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
205 be configured in the [PFIFO] section, "GRED" in
206 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
207 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
208 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
209 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
210 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
211 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
213 * systemd-networkd gained support for a new Termination= setting in the
214 [CAN] section for configuring the termination resistor. It also
215 gained a new ListenOnly= setting for controlling whether to only
216 listen on CAN interfaces, without interfering with traffic otherwise
217 (which is useful for debugging/monitoring CAN network
218 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
219 been added to configure various CAN-FD aspects.
221 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
222 When enabled, DHCPv6 will be attempted right-away without requiring an
223 Router Advertisement packet suggesting it first (i.e. without the 'M'
224 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
225 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
226 the RA packets suggest it.
228 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
229 which may be used to turn off use of the gateway information provided
230 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
231 used to configure how to process leases that lack a lifetime option.
233 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
234 setting SendVendorOption= allowing configuration of additional vendor
235 options to send in the DHCP requests/responses. The [DHCPv6] section
236 gained a new SendOption= setting for sending arbitrary DHCP
237 options. RequestOptions= has been added to request arbitrary options
238 from the server. UserClass= has been added to set the DHCP user class
241 * systemd-networkd's [DHCPServer] section gained a new set of options
242 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
243 information about these three protocols in the DHCP lease. It also
244 gained support for including "MUD" URLs ("Manufacturer Usage
245 Description"). Support for "MUD" URLs was also added to the LLDP
246 stack, configurable in the [LLDP] section in .network files.
248 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
249 mode. Also, the sections now support a new setting SourceMACAddress=.
251 * systemd-networkd's .netdev files now support a new setting
252 VLANProtocol= in the [Bridge] section that allows configuration of
253 the VLAN protocol to use.
255 * systemd-networkd supports a new Group= setting in the [Link] section
256 of the .network files, to control the link group.
258 * systemd-networkd's [Network] section gained a new
259 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
260 link local address is generated.
262 * A new default .network file is now shipped that matches TUN/TAP
263 devices that begin with "vt-" in their name. Such interfaces will
264 have IP routing onto the host links set up automatically. This is
265 supposed to be used by VM managers to trivially acquire a network
266 interface which is fully set up for host communication, simply by
267 carefully picking an interface name to use.
269 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
270 which sets the route priority for routes specified by the DHCP server.
272 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
273 which configures the vendor class information sent to DHCP server.
275 * The BlackList= settings in .network files' [DHCPv4] and
276 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
277 are still understood to provide compatibility.
279 * networkctl gained the new "forcerenew" command for forcing all DHCP
280 server clients to renew their lease. The interface "status" output
281 will now show numerous additional fields of information about an
282 interface. There are new "up" and "down" commands to bring specific
283 interfaces up or down.
285 * systemd-resolved's DNS= configuration option now optionally accepts a
286 port number (after ":") and a host name (after "#"). When the host
287 name is specified, the DNS-over-TLS certificate is validated to match
288 the specified hostname. Additionally, in case of IPv6 addresses, an
289 interface may be specified (after "%").
291 * systemd-resolved may be configured to forward single-label DNS names.
292 This is not standard-conformant, but may make sense in setups where
293 public DNS servers are not used.
295 * systemd-resolved's DNS-over-TLS support gained SNI validation.
297 * systemd-nspawn's --resolv-conf= switch gained a number of new
298 supported values. Specifically, options starting with "replace-" are
299 like those prefixed "copy-" but replace any existing resolv.conf
300 file. And options ending in "-uplink" and "-stub" can now be used to
301 propagate other flavours of resolv.conf into the container (as
302 defined by systemd-resolved).
304 * The various programs included in systemd can now optionally output
305 their log messages on stderr prefixed with a timestamp, controlled by
306 the $SYSTEMD_LOG_TIME environment variable.
308 * systemctl gained a new "-P" switch that is a shortcut for "--value
311 * "systemctl list-units" and "systemctl list-machines" no longer hide
312 their first output column with --no-legend. To hide the first column,
315 * "systemctl reboot" takes the option "--reboot-argument=".
316 The optional positional argument to "systemctl reboot" is now
317 being deprecated in favor of this option.
319 * systemd-run gained a new switch --slice-inherit. If specified the
320 unit it generates is placed in the same slice as the systemd-run
323 * systemd-journald gained support for zstd compression of large fields
324 in journal files. The hash tables in journal files have been hardened
325 against hash collisions. This is an incompatible change and means
326 that journal files created with new systemd versions are not readable
327 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
328 environment variable for systemd-journald.service is set to 0 this
329 new hardening functionality may be turned off, so that generated
330 journal files remain compatible with older journalctl
333 * journalctl will now include a clickable link in the default output for
334 each log message for which an URL with further documentation is
335 known. This is only supported on terminal emulators that support
336 clickable hyperlinks, and is turned off if a pager is used (since
337 "less" still doesn't support hyperlinks,
338 unfortunately). Documentation URLs may be included in log messages
339 either by including a DOCUMENTATION= journal field in it, or by
340 associating a journal message catalog entry with the log message's
341 MESSAGE_ID, which then carries a "Documentation:" tag.
343 * journald.conf gained a new boolean setting Audit= that may be used to
344 control whether systemd-journald will enable audit during
347 * when systemd-journald's log stream is broken up into multiple lines
348 because the PID of the sender changed this is indicated in the
349 generated log records via the _LINE_BREAK=pid-change field.
351 * journalctl's "-o cat" output mode will now show one or more journal
352 fields specified with --output-fields= instead of unconditionally
353 MESSAGE=. This is useful to retrieve a very specific set of fields
354 without any decoration.
356 * The sd-journal.h API gained two new functions:
357 sd_journal_enumerate_available_unique() and
358 sd_journal_enumerate_available_data() that operate like their
359 counterparts that lack the _available_ in the name, but skip items
360 that cannot be read and processed by the local implementation
361 (i.e. are compressed in an unsupported format or such),
363 * coredumpctl gained a new --file= switch, matching the same one in
364 journalctl: a specific journal file may be specified to read the
367 * coredumps collected by systemd-coredump may now be compressed using
370 * systemd-binfmt gained a new switch --unregister for unregistering all
371 registered entries at once. This is now invoked automatically at
372 shutdown, so that binary formats registered with the "F" flag will
373 not block clean file system unmounting.
375 * systemd-notify's --pid= switch gained new values: "parent", "self",
376 "auto" for controlling which PID to send to the service manager: the
377 systemd-notify process' PID, or the one of the process invoking it.
379 * systemd-logind's Session bus object learnt a new method call
380 SetType() for temporarily updating the session type of an already
381 allocated session. This is useful for upgrading tty sessions to
382 graphical ones once a compositor is invoked.
384 * systemd-socket-proxy gained a new switch --exit-idle-time= for
385 configuring an exit-on-idle time.
387 * systemd-repart's --empty= setting gained a new value "create". If
388 specified a new empty regular disk image file is created under the
389 specified name. Its size may be specified with the new --size=
390 option. The latter is also supported without the "create" mode, in
391 order to grow existing disk image files to the specified size. These
392 two new options are useful when creating or manipulating disk images
393 instead of operating on actual block devices.
395 * systemd-repart drop-ins now support a new UUID= setting to control
396 the UUID to assign to a newly created partition.
398 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
401 * systemd-repart's Label= setting now support the usual, simple
404 * systemd-homed's LUKS backend gained the ability to discard empty file
405 system blocks automatically when the user logs out. This is enabled
406 by default to ensure that home directories take minimal space when
407 logged out but get full size guarantees when logged in. This may be
408 controlled with the new --luks-offline-discard= switch to homectl.
410 * If systemd-homed detects that /home/ is encrypted as a whole it will
411 now default to the directory or subvolume backends instead of the
412 LUKS backend, in order to avoid double encryption. The default
413 storage and file system may now be configured explicitly, too, via
414 the new /etc/systemd/homed.conf configuration file.
416 * systemd-homed now supports unlocking home directories with FIDO2
417 security tokens that support the 'hmac-secret' extension, in addition
418 to the existing support for PKCS#11 security token unlocking
419 support. Note that many recent hardware security tokens support both
420 interfaces. The FIDO2 support is accessible via homectl's
421 --fido2-device= option.
423 * homectl's --pkcs11-uri= setting now accepts two special parameters:
424 if "auto" is specified and only one suitable PKCS#11 security token
425 is plugged in, its URL is automatically determined and enrolled for
426 unlocking the home directory. If "list" is specified a brief table of
427 suitable PKCS#11 security tokens is shown. Similar, the new
428 --fido2-device= option also supports these two special values, for
429 automatically selecting and listing suitable FIDO2 devices.
431 * The /etc/crypttab tmp option now optionally takes an argument
432 selecting the file system to use. Moreover, the default is now
433 changed from ext2 to ext4.
435 * There's a new /etc/crypttab option "keyfile-erase". If specified the
436 key file listed in the same line is removed after use, regardless if
437 volume activation was successful or not. This is useful if the key
438 file is only acquired transiently at runtime and shall be erased
439 before the system continues to boot.
441 * There's also a new /etc/crypttab option "try-empty-password". If
442 specified, before asking the user for a password it is attempted to
443 unlock the volume with an empty password. This is useful for
444 installing encrypted images whose password shall be set on first boot
445 instead of at installation time.
447 * systemd-cryptsetup will now attempt to load the keys to unlock
448 volumes with automatically from files in
449 /etc/cryptsetup-keys.d/<volume>.key and
450 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
452 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
453 /etc/crypttab, during boot.
455 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
456 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
459 * A new generator systemd-xdg-autostart-generator has been added. It
460 generates systemd unit files from XDG autostart .desktop files, and
461 may be used to let the systemd user instance manage services that are
462 started automatically as part of the desktop session.
464 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
465 to query and change the firmware's 'reboot into firmware' setup flag.
467 * systemd-firstboot gained a new switch --kernel-command-line= that may
468 be used to initialize the /etc/kernel/cmdline file of the image. It
469 also gained a new switch --root-password-hashed= which is like
470 --root-password= but accepts a pre-hashed UNIX password as
471 argument. The new option --delete-root-password may be used to unset
472 any password for the root user (dangerous!). The --root-shell= switch
473 may be used to control the shell to use for the root account. A new
474 --force option may be used to override any already set settings with
475 the parameters specified on the command line (by default, the tool
476 will not override what has already been set before, i.e. is purely
479 * systemd-firstboot gained support for a new --image= switch, which is
480 similar to --root= but accepts the path to a disk image file, on
481 which it then operates.
483 * A new sd-path.h API has been added to libsystemd. It provides a
484 simple API for retrieving various search paths and primary
485 directories for various resources.
487 * A new call sd_notify_barrier() has been added to the sd-daemon.h
488 API. The call will block until all previously sent sd_notify()
489 messages have been processed by the service manager. This is useful
490 to remove races caused by a process already having disappeared at the
491 time a notification message is processed by the service manager,
492 making correct attribution impossible. The systemd-notify tool will
493 now make use of this call implicitly, but this can be turned off again
494 via the new --no-block switch.
496 * When sending a file descriptor (fd) to the service manager to keep
497 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
498 may be specified. If passed the service manager will refrain from
499 poll()ing on the file descriptor. Traditionally (and when the
500 parameter is not specified), the service manager will poll it for
501 POLLHUP or POLLERR events, and immediately close the fds in that
504 * The service manager (PID1) gained a new D-Bus method call
505 SetShowStatus() which may be used to control whether it shall show
506 boot-time status output on the console. This method has a similar
507 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
509 * The sd-bus API gained a number of convenience functions that take
510 va_list arguments rather than "...". For example, there's now
511 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
512 it easier to build wrappers that accept variadic arguments and want
513 to pass a ready va_list structure to sd-bus.
515 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
516 flag which alters how the userdata pointer to pass to the callbacks
517 is determined. When the flag is set, the offset field is converted
518 as-is into a pointer, without adding it to the object pointer the
519 vtable is associated with.
521 * sd-bus now exposes four new functions:
522 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
523 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
524 validate strings to check if they qualify as various D-Bus concepts.
526 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
527 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
528 that simplify adding argument names to D-Bus methods and signals.
530 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
532 * Various D-Bus APIs of systemd daemons now have man pages that
533 document the methods, signals and properties.
535 * The expectations on user/group name syntax are now documented in
536 detail; documentation on how classic home directories may be
537 converted into home directories managed by homed has been added;
538 documentation regarding integration of homed/userdb functionality in
539 desktops has been added:
541 https://systemd.io/USER_NAMES
542 https://systemd.io/CONVERTING_TO_HOMED
543 https://systemd.io/USERDB_AND_DESKTOPS
545 * Documentation for the on-disk Journal file format has been updated
546 and has now moved to:
548 https://systemd.io/JOURNAL_FILE_FORMAT
550 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
551 has been extended by a set of environment variables that expose
552 select fields from the host's os-release file to the container
553 payload. Similarly, host's os-release files can be mounted into the
554 container underneath /run/host. Together, those mechanisms provide a
555 standardized way to expose information about the host to the
556 container payload. Both interfaces are implemented in systemd-nspawn.
558 * All D-Bus services shipped in systemd now implement the generic
559 LogControl1 D-Bus API which allows clients to change log level +
560 target of the service during runtime.
562 * Only relevant for developers: the mkosi.default symlink has been
563 dropped from version control. Please create a symlink to one of the
564 distribution-specific defaults in .mkosi/ based on your preference.
566 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
567 Malafeev, Amitanand.Chikorde, Alin Popa, Alvin Šipraga, Amos Bird,
568 Andreas Rammhold, AndreRH, Andrew Doran, Anita Zhang, Ankit Jain,
569 antznin, Arnaud Ferraris, Arthur Moraes do Lago, Arusekk, Balaji
570 Punnuru, Balint Reczey, Bastien Nocera, bemarek, Benjamin Berg,
571 Benjamin Dahlhoff, Benjamin Robin, Chris Down, Chris Kerr, Christian
572 Göttsche, Christian Hesse, Christian Oder, Ciprian Hacman, Clinton Roy,
573 codicodi, Corey Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan,
574 Daniel Fullmer, Daniel Rusek, Dan Streetman, Dave Reisner, David
575 Edmundson, David Wood, Denis Pronin, Diego Escalante Urrelo, Dimitri
576 John Ledkov, dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel
577 Garette, Eric Anderson, Eric DeVolder, Evgeny Vereshchagin,
578 ExtinctFire, fangxiuning, Ferran Pallarès Roca, Filipe Brandenburger,
579 Filippo Falezza, Finn, Florian Klink, Florian Mayer, Franck Bui,
580 Frantisek Sumsal, gaurav, Georg Müller, Gergely Polonkai, Giedrius
581 Statkevičius, Gigadoc2, gogogogi, Gaurav Singh, gzjsgdsb, Hans de
582 Goede, Haochen Tong, ianhi, ignapk, Jakov Smolic, James T. Lee, Jan
583 Janssen, Jan Klötzke, Jan Palus, Jay Burger, Jeremy Cline, Jérémy
584 Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro, Joerg Behrmann, Jörg
585 Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny Levinsen, Kevin
586 Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus, Lénaïc Huard,
587 Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca BRUNO, Lucas
588 Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz Stelmach, Maciej
589 S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel Holtmann, Marc
590 Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt Ranostay, Maxim
591 Fomin, MaxVerevkin, Michael Biebl, Michael Chapman, Michael Gubbels,
592 Michael Marley, Michał Bartoszkiewicz, Michal Koutný, Michal Sekletár,
593 Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml, Motiejus Jakštys,
594 nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas Hambüchen, Norbert
595 Lange, Paul Cercueil, pelzvieh, Peter Hutterer, Piero La Terza, Pieter
596 Lexis, Piotr Drąg, Rafael Fontenelle, Richard Petri, Ronan Pigott, Ross
597 Lagerwall, Rubens Figueiredo, satmandu, Sean-StarLabs, Sebastian
598 Jennen, sterlinghughes, Surhud More, Susant Sahani, szb512, Thomas
599 Haller, Tobias Hunger, Tom, Tomáš Pospíšek, Tomer Shechner, Tom Hughes,
600 Topi Miettinen, Tudor Roman, Uwe Kleine-König, Valery0xff, Vito Caputo,
601 Vladimir Panteleev, Vladyslav Tronko, Wen Yang, Yegor Vialov, Yigal
602 Korman, Yi Gao, YmrDtnJu, Yuri Chornoivan, Yu Watanabe, Zbigniew
603 Jędrzejewski-Szmek, Zhu Li, Дамјан Георгиевски, наб
609 * A new tool "systemd-repart" has been added, that operates as an
610 idempotent declarative repartitioner for GPT partition tables.
611 Specifically, a set of partitions that must or may exist can be
612 configured via drop-in files, and during every boot the partition
613 table on disk is compared with these files, creating missing
614 partitions or growing existing ones based on configurable relative
615 and absolute size constraints. The tool is strictly incremental,
616 i.e. does not delete, shrink or move partitions, but only adds and
617 grows them. The primary use-case is OS images that ship in minimized
618 form, that on first boot are grown to the size of the underlying
619 block device or augmented with additional partitions. For example,
620 the root partition could be extended to cover the whole disk, or a
621 swap or /home partitions could be added on first boot. It can also be
622 used for systems that use an A/B update scheme but ship images with
623 just the A partition, with B added on first boot. The tool is
624 primarily intended to be run in the initrd, shortly before
625 transitioning into the host OS, but can also be run after the
626 transition took place. It automatically discovers the disk backing
627 the root file system, and should hence not require any additional
628 configuration besides the partition definition drop-ins. If no
629 configuration drop-ins are present, no action is taken.
631 * A new component "userdb" has been added, along with a small daemon
632 "systemd-userdb.service" and a client tool "userdbctl". The framework
633 allows defining rich user and group records in a JSON format,
634 extending on the classic "struct passwd" and "struct group"
635 structures. Various components in systemd have been updated to
636 process records in this format, including systemd-logind and
637 pam-systemd. The user records are intended to be extensible, and
638 allow setting various resource management, security and runtime
639 parameters that shall be applied to processes and sessions of the
640 user as they log in. This facility is intended to allow associating
641 such metadata directly with user/group records so that they can be
642 produced, extended and consumed in unified form. We hope that
643 eventually frameworks such as sssd will generate records this way, so
644 that for the first time resource management and various other
645 per-user settings can be configured in LDAP directories and then
646 provided to systemd (specifically to systemd-logind and pam-system)
647 to apply on login. For further details see:
649 https://systemd.io/USER_RECORD
650 https://systemd.io/GROUP_RECORD
651 https://systemd.io/USER_GROUP_API
653 * A small new service systemd-homed.service has been added, that may be
654 used to securely manage home directories with built-in encryption.
655 The complete user record data is unified with the home directory,
656 thus making home directories naturally migratable. Its primary
657 back-end is based on LUKS volumes, but fscrypt, plain directories,
658 and other storage schemes are also supported. This solves a couple of
659 problems we saw with traditional ways to manage home directories, in
660 particular when it comes to encryption. For further discussion of
661 this, see the video of Lennart's talk at AllSystemsGo! 2019:
663 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
665 For further details about the format and expectations on home
666 directories this new daemon makes, see:
668 https://systemd.io/HOME_DIRECTORY
670 * systemd-journald is now multi-instantiable. In addition to the main
671 instance systemd-journald.service there's now a template unit
672 systemd-journald@.service, with each instance defining a new named
673 log 'namespace' (whose name is specified via the instance part of the
674 unit name). A new unit file setting LogNamespace= has been added,
675 taking such a namespace name, that assigns services to the specified
676 log namespaces. As each log namespace is serviced by its own
677 independent journal daemon, this functionality may be used to improve
678 performance and increase isolation of applications, at the price of
679 losing global message ordering. Each instance of journald has a
680 separate set of configuration files, with possibly different disk
681 usage limitations and other settings.
683 journalctl now takes a new option --namespace= to show logs from a
684 specific log namespace. The sd-journal.h API gained
685 sd_journal_open_namespace() for opening the log stream of a specific
686 log namespace. systemd-journald also gained the ability to exit on
687 idle, which is useful in the context of log namespaces, as this means
688 log daemons for log namespaces can be activated automatically on
689 demand and will stop automatically when no longer used, minimizing
692 * When systemd-tmpfiles copies a file tree using the 'C' line type it
693 will now label every copied file according to the SELinux database.
695 * When systemd/PID 1 detects it is used in the initrd it will now boot
696 into initrd.target rather than default.target by default. This should
697 make it simpler to build initrds with systemd as for many cases the
698 only difference between a host OS image and an initrd image now is
699 the presence of the /etc/initrd-release file.
701 * A new kernel command line option systemd.cpu_affinity= is now
702 understood. It's equivalent to the CPUAffinity= option in
703 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
704 itself and the default for all other processes.
706 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
707 equivalent), the SELinux database is now reloaded, ensuring that
708 sockets and other file system objects are generated taking the new
709 database into account.
711 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
712 "quiet" has been changed to imply that instead of
713 "systemd.show-status=auto". In this mode, only messages about errors
714 and significant delays in boot are shown on the console.
716 * The sd-event.h API gained native support for the new Linux "pidfd"
717 concept. This permits watching processes using file descriptors
718 instead of PID numbers, which fixes a number of races and makes
719 process supervision more robust and efficient. All of systemd's
720 components will now use pidfds if the kernel supports it for process
721 watching, with the exception of PID 1 itself, unfortunately. We hope
722 to move PID 1 to exclusively using pidfds too eventually, but this
723 requires some more kernel work first. (Background: PID 1 watches
724 processes using waitid() with the P_ALL flag, and that does not play
725 together nicely with pidfds yet.)
727 * Closely related to this, the sd-event.h API gained two new calls
728 sd_event_source_send_child_signal() (for sending a signal to a
729 watched process) and sd_event_source_get_child_process_own() (for
730 marking a process so that it is killed automatically whenever the
731 event source watching it is freed).
733 * systemd-networkd gained support for configuring Token Bucket Filter
734 (TBF) parameters in its qdisc configuration support. Similarly,
735 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
736 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
738 * systemd-networkd gained support for Intermediate Functional Block
739 (IFB) network devices.
741 * systemd-networkd gained support for configuring multi-path IP routes,
742 using the new MultiPathRoute= setting in the [Route] section.
744 * systemd-networkd's DHCPv4 client has been updated to support a new
745 SendDecline= option. If enabled, duplicate address detection is done
746 after a DHCP offer is received from the server. If a conflict is
747 detected, the address is declined. The DHCPv4 client also gained
748 support for a new RouteMTUBytes= setting that allows to configure the
749 MTU size to be used for routes generated from DHCPv4 leases.
751 * The PrefixRoute= setting in systemd-networkd's [Address] section of
752 .network files has been deprecated, and replaced by AddPrefixRoute=,
753 with its sense inverted.
755 * The Gateway= setting of [Route] sections of .network files gained
756 support for a special new value "_dhcp". If set, the configured
757 static route uses the gateway host configured via DHCP.
759 * New User= and SuppressPrefixLength= settings have been implemented
760 for the [RoutingPolicyRule] section of .network files to configure
761 source routing based on UID ranges and prefix length, respectively.
763 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
764 D-Bus message object as "sensitive". Those objects are erased from
765 memory when they are freed. This concept is intended to be used for
766 messages that contain security sensitive data. A new flag
767 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
768 in sd-bus vtables, causing any incoming and outgoing messages of
769 those methods to be implicitly marked as "sensitive".
771 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
772 contents of a message (or parts thereof) to standard output for
775 * systemd-sysusers gained support for creating users with the primary
776 group named differently than the user.
778 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
779 gained support for growing XFS partitions. Previously it supported
780 only ext4 and btrfs partitions.
782 * The support for /etc/crypttab gained a new x-initrd.attach option. If
783 set, the specified encrypted volume is unlocked already in the
784 initrd. This concept corresponds to the x-initrd.mount option in
787 * systemd-cryptsetup gained native support for unlocking encrypted
788 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
789 encryption of volumes to YubiKeys. This is exposed in the new
790 pkcs11-uri= option in /etc/crypttab.
792 * The /etc/fstab support in systemd now supports two new mount options
793 x-systemd.{required,wanted}-by=, for explicitly configuring the units
794 that the specified mount shall be pulled in by, in place of
795 the usual local-fs.target/remote-fs.target.
797 * The https://systemd.io/ web site has been relaunched, directly
798 populated with most of the documentation included in the systemd
799 repository. systemd also acquired a new logo, thanks to Tobias
802 * systemd-udevd gained support for managing "alternative" network
803 interface names, as supported by new Linux kernels. For the first
804 time this permits assigning multiple (and longer!) names to a network
805 interface. systemd-udevd will now by default assign the names
806 generated via all supported naming schemes to each interface. This
807 may be further tweaked with .link files and the AlternativeName= and
808 AlternativeNamesPolicy= settings. Other components of systemd have
809 been updated to support the new alternative names wherever
810 appropriate. For example, systemd-nspawn will now generate
811 alternative interface names for the host-facing side of container
812 veth links based on the full container name without truncation.
814 * systemd-nspawn interface naming logic has been updated in another way
815 too: if the main interface name (i.e. as opposed to new-style
816 "alternative" names) based on the container name is truncated, a
817 simple hashing scheme is used to give different interface names to
818 multiple containers whose names all begin with the same prefix. Since
819 this changes the primary interface names pointing to containers if
820 truncation happens, the old scheme may still be requested by
821 selecting an older naming scheme, via the net.naming-scheme= kernel
824 * PrivateUsers= in service files now works in services run by the
825 systemd --user per-user instance of the service manager.
827 * A new per-service sandboxing option ProtectClock= has been added that
828 locks down write access to the system clock. It takes away device
829 node access to /dev/rtc as well as the system calls that set the
830 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
831 Note that this option does not affect access to auxiliary services
832 that allow changing the clock, for example access to
835 * The systemd-id128 tool gained a new "show" verb for listing or
836 resolving a number of well-known UUIDs/128bit IDs, currently mostly
837 GPT partition table types.
839 * The Discoverable Partitions Specification has been updated to support
840 /var and /var/tmp partition discovery. Support for this has been
841 added to systemd-gpt-auto-generator. For details see:
843 https://systemd.io/DISCOVERABLE_PARTITIONS
845 * "systemctl list-unit-files" has been updated to show a new column
846 with the suggested enablement state based on the vendor preset files
847 for the respective units.
849 * "systemctl" gained a new option "--with-dependencies". If specified
850 commands such as "systemctl status" or "systemctl cat" will now show
851 all specified units along with all units they depend on.
853 * networkctl gained support for showing per-interface logs in its
856 * systemd-networkd-wait-online gained support for specifying the maximum
857 operational state to wait for, and to wait for interfaces to
860 * The [Match] section of .link and .network files now supports a new
861 option PermanentMACAddress= which may be used to check against the
862 permanent MAC address of a network device even if a randomized MAC
865 * The [TrafficControlQueueingDiscipline] section in .network files has
866 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
867 dropped from the individual setting names.
869 * Any .link and .network files that have an empty [Match] section (this
870 also includes empty and commented-out files) will now be
871 rejected. systemd-udev and systemd-networkd started warning about
872 such files in version 243.
874 * systemd-logind will now validate access to the operation of changing
875 the virtual terminal via a polkit action. By default, only users
876 with at least one session on a local VT are granted permission.
878 * When systemd sets up PAM sessions that invoked service processes
879 shall run in, the pam_setcred() API is now invoked, thus permitting
880 PAM modules to set additional credentials for the processes.
882 * portablectl attach/detach verbs now accept --now and --enable options
883 to combine attachment with enablement and invocation, or detachment
884 with stopping and disablement.
886 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
887 fixed, which in turn exposed bugs in unit configuration of services
888 which have Type=oneshot and should only run once, but do not have
889 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
890 service may be started again after exiting successfully, for example
891 as a dependency in another transaction. Affected services included
892 some internal systemd services (most notably
893 systemd-vconsole-setup.service, which was updated to have
894 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
895 plymouth has been suitably updated or patched before upgrading to
896 this systemd release. See
897 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
898 additional discussion.
900 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
901 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
902 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
903 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
904 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
905 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
906 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
907 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
908 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
909 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
910 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
911 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
912 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
913 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
914 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
915 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
916 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
917 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
918 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
919 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
920 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
921 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
922 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
923 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
930 * Support for the cpuset cgroups v2 controller has been added.
931 Processes may be restricted to specific CPUs using the new
932 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
933 AllowedMemoryNodes= setting.
935 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
936 now be configured using a new RestartKillSignal= setting. This
937 allows units which signals to request termination to implement
938 different behaviour when stopping in preparation for a restart.
940 * "systemctl clean" may now be used also for socket, mount, and swap
943 * systemd will also read configuration options from the EFI variable
944 SystemdOptions. This may be used to configure systemd behaviour when
945 modifying the kernel command line is inconvenient, but configuration
946 on disk is read too late, for example for the options related to
947 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
948 set the EFI variable.
950 * systemd will now disable printk ratelimits in early boot. This should
951 allow us to capture more logs from the early boot phase where normal
952 storage is not available and the kernel ring buffer is used for
953 logging. Configuration on the kernel command line has higher priority
954 and overrides the systemd setting.
956 systemd programs which log to /dev/kmsg directly use internal
957 ratelimits to prevent runaway logging. (Normally this is only used
958 during early boot, so in practice this change has very little
961 * Unit files now support top level dropin directories of the form
962 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
963 that affects all corresponding unit files.
965 * systemctl gained support for 'stop --job-mode=triggering' which will
966 stop the specified unit and any units which could trigger it.
968 * Unit status display now includes units triggering and triggered by
969 the unit being shown.
971 * The RuntimeMaxSec= setting is now supported by scopes, not just
972 .service units. This is particularly useful for PAM sessions which
973 create a scope unit for the user login. systemd.runtime_max_sec=
974 setting may used with the pam_systemd module to limit the duration
975 of the PAM session, for example for time-limited logins.
977 * A new @pkey system call group is now defined to make it easier to
978 allow-list memory protection syscalls for containers and services
979 which need to use them.
981 * systemd-udevd: removed the 30s timeout for killing stale workers on
982 exit. systemd-udevd now waits for workers to finish. The hard-coded
983 exit timeout of 30s was too short for some large installations, where
984 driver initialization could be prematurely interrupted during initrd
985 processing if the root file system had been mounted and init was
986 preparing to switch root. If udevd is run without systemd and workers
987 are hanging while udevd receives an exit signal, udevd will now exit
988 when udev.event_timeout is reached for the last hanging worker. With
989 systemd, the exit timeout can additionally be configured using
990 TimeoutStopSec= in systemd-udevd.service.
992 * udev now provides a program (fido_id) that identifies FIDO CTAP1
993 ("U2F")/CTAP2 security tokens based on the usage declared in their
994 report and descriptor and outputs suitable environment variables.
995 This replaces the externally maintained allow lists of all known
996 security tokens that were used previously.
998 * Automatically generated autosuspend udev rules for allow-listed
999 devices have been imported from the Chromium OS project. This should
1000 improve power saving with many more devices.
1002 * udev gained a new "CONST{key}=value" setting that allows matching
1003 against system-wide constants without forking a helper binary.
1004 Currently "arch" and "virt" keys are supported.
1006 * udev now opens CDROMs in non-exclusive mode when querying their
1007 capabilities. This should fix issues where other programs trying to
1008 use the CDROM cannot gain access to it, but carries a risk of
1009 interfering with programs writing to the disk, if they did not open
1010 the device in exclusive mode as they should.
1012 * systemd-networkd does not create a default route for IPv4 link local
1013 addressing anymore. The creation of the route was unexpected and was
1014 breaking routing in various cases, but people who rely on it being
1015 created implicitly will need to adjust. Such a route may be requested
1016 with DefaultRouteOnDevice=yes.
1018 Similarly, systemd-networkd will not assign a link-local IPv6 address
1019 when IPv6 link-local routing is not enabled.
1021 * Receive and transmit buffers may now be configured on links with
1022 the new RxBufferSize= and TxBufferSize= settings.
1024 * systemd-networkd may now advertise additional IPv6 routes. A new
1025 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
1028 * systemd-networkd may now configure "next hop" routes using the
1029 [NextHop] section and Gateway= and Id= settings.
1031 * systemd-networkd will now retain DHCP config on restarts by default
1032 (but this may be overridden using the KeepConfiguration= setting).
1033 The default for SendRelease= has been changed to true.
1035 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
1036 received from the server.
1038 The client will use the received SIP server list if UseSIP=yes is
1041 The client may be configured to request specific options from the
1042 server using a new RequestOptions= setting.
1044 The client may be configured to send arbitrary options to the server
1045 using a new SendOption= setting.
1047 A new IPServiceType= setting has been added to configure the "IP
1048 service type" value used by the client.
1050 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
1051 request prefix hints in the DHCPv6 solicitation.
1053 * The DHCPv4 server may be configured to send arbitrary options using
1054 a new SendOption= setting.
1056 * The DHCPv4 server may now be configured to emit SIP server list using
1057 the new EmitSIP= and SIP= settings.
1059 * systemd-networkd and networkctl may now renew DHCP leases on demand.
1060 networkctl has a new 'networkctl renew' verb.
1062 * systemd-networkd may now reconfigure links on demand. networkctl
1063 gained two new verbs: "reload" will reload the configuration, and
1064 "reconfigure DEVICE…" will reconfigure one or more devices.
1066 * .network files may now match on SSID and BSSID of a wireless network,
1067 i.e. the access point name and hardware address using the new SSID=
1068 and BSSID= options. networkctl will display the current SSID and
1069 BSSID for wireless links.
1071 .network files may also match on the wireless network type using the
1072 new WLANInterfaceType= option.
1074 * systemd-networkd now includes default configuration that enables
1075 link-local addressing when connected to an ad-hoc wireless network.
1077 * systemd-networkd may configure the Traffic Control queueing
1078 disciplines in the kernel using the new
1079 [TrafficControlQueueingDiscipline] section and Parent=,
1080 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
1081 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
1082 NetworkEmulatorDuplicateRate= settings.
1084 * systemd-tmpfiles gained a new w+ setting to append to files.
1086 * systemd-analyze dump will now report when the memory configuration in
1087 the kernel does not match what systemd has configured (usually,
1088 because some external program has modified the kernel configuration
1091 * systemd-analyze gained a new --base-time= switch instructs the
1092 'calendar' verb to resolve times relative to that timestamp instead
1093 of the present time.
1095 * journalctl --update-catalog now produces deterministic output (making
1096 reproducible image builds easier).
1098 * A new devicetree-overlay setting is now documented in the Boot Loader
1101 * The default value of the WatchdogSec= setting used in systemd
1102 services (the ones bundled with the project itself) may be set at
1103 configuration time using the -Dservice-watchdog= setting. If set to
1104 empty, the watchdogs will be disabled.
1106 * systemd-resolved validates IP addresses in certificates now when GnuTLS
1109 * libcryptsetup >= 2.0.1 is now required.
1111 * A configuration option -Duser-path= may be used to override the $PATH
1112 used by the user service manager. The default is again to use the same
1113 path as the system manager.
1115 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
1116 outputting the 128bit IDs in UUID format (i.e. in the "canonical
1119 * Service units gained a new sandboxing option ProtectKernelLogs= which
1120 makes sure the program cannot get direct access to the kernel log
1121 buffer anymore, i.e. the syslog() system call (not to be confused
1122 with the API of the same name in libc, which is not affected), the
1123 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
1124 inaccessible to the service. It's recommended to enable this setting
1125 for all services that should not be able to read from or write to the
1126 kernel log buffer, which are probably almost all.
1128 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
1129 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
1130 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
1131 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
1132 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
1133 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
1134 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
1135 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
1136 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
1137 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
1138 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
1139 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
1140 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
1141 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
1142 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
1143 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
1144 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
1145 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
1146 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
1147 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
1148 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
1149 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
1150 Zach Smith, Zbigniew Jędrzejewski-Szmek
1152 – Warsaw, 2019-11-29
1156 * This release enables unprivileged programs (i.e. requiring neither
1157 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
1158 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
1159 kernel for the whole UNIX group range, i.e. all processes. This
1160 change should be reasonably safe, as the kernel support for it was
1161 specifically implemented to allow safe access to ICMP Echo for
1162 processes lacking any privileges. If this is not desirable, it can be
1163 disabled again by setting the parameter to "1 0".
1165 * Previously, filters defined with SystemCallFilter= would have the
1166 effect that any calling of an offending system call would terminate
1167 the calling thread. This behaviour never made much sense, since
1168 killing individual threads of unsuspecting processes is likely to
1169 create more problems than it solves. With this release the default
1170 action changed from killing the thread to killing the whole
1171 process. For this to work correctly both a kernel version (>= 4.14)
1172 and a libseccomp version (>= 2.4.0) supporting this new seccomp
1173 action is required. If an older kernel or libseccomp is used the old
1174 behaviour continues to be used. This change does not affect any
1175 services that have no system call filters defined, or that use
1176 SystemCallErrorNumber= (and thus see EPERM or another error instead
1177 of being killed when calling an offending system call). Note that
1178 systemd documentation always claimed that the whole process is
1179 killed. With this change behaviour is thus adjusted to match the
1182 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
1183 4194304 by default, i.e. the full 22bit range the kernel allows, up
1184 from the old 16bit range. This should improve security and
1185 robustness, as PID collisions are made less likely (though certainly
1186 still possible). There are rumours this might create compatibility
1187 problems, though at this moment no practical ones are known to
1188 us. Downstream distributions are hence advised to undo this change in
1189 their builds if they are concerned about maximum compatibility, but
1190 for everybody else we recommend leaving the value bumped. Besides
1191 improving security and robustness this should also simplify things as
1192 the maximum number of allowed concurrent tasks was previously bounded
1193 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
1194 only a single knob is left ("kernel.threads-max"). There have been
1195 concerns that usability is affected by this change because larger PID
1196 numbers are harder to type, but we believe the change from 5 digits
1197 to 7 digits doesn't hamper usability.
1199 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
1200 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
1201 hierarchically set default memory protection values for a particular
1202 subtree of the unit hierarchy.
1204 * Memory protection directives can now take a value of zero, allowing
1205 explicit opting out of a default value propagated by an ancestor.
1207 * systemd now defaults to the "unified" cgroup hierarchy setup during
1208 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
1209 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
1210 change reflects the fact that cgroupsv2 support has matured
1211 substantially in both systemd and in the kernel, and is clearly the
1212 way forward. Downstream production distributions might want to
1213 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
1214 their builds as unfortunately the popular container managers have not
1215 caught up with the kernel API changes.
1217 * Man pages are not built by default anymore (html pages were already
1218 disabled by default), to make development builds quicker. When
1219 building systemd for a full installation with documentation, meson
1220 should be called with -Dman=true and/or -Dhtml=true as appropriate.
1221 The default was changed based on the assumption that quick one-off or
1222 repeated development builds are much more common than full optimized
1223 builds for installation, and people need to pass various other
1224 options to when doing "proper" builds anyway, so the gain from making
1225 development builds quicker is bigger than the one time disruption for
1228 Two scripts are created in the *build* directory to generate and
1229 preview man and html pages on demand, e.g.:
1231 build/man/man systemctl
1232 build/man/html systemd.index
1234 * libidn2 is used by default if both libidn2 and libidn are installed.
1235 Please use -Dlibidn=true if libidn is preferred.
1237 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
1238 big-endian machines. Before, bytes were written and read in native
1239 machine order as exposed by the native libc __cpu_mask interface.
1240 Now, little-endian order is always used (CPUs 0–7 are described by
1241 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
1242 This change fixes D-Bus calls that cross endianness boundary.
1244 The presentation format used for CPUAffinity= by "systemctl show" and
1245 "systemd-analyze dump" is changed to present CPU indices instead of
1246 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
1247 shown as CPUAffinity=03000000000000000000000000000… (on
1248 little-endian) or CPUAffinity=00000000000000300000000000000… (on
1249 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
1250 input format. The maximum integer that will be printed in the new
1251 format is 8191 (four digits), while the old format always used a very
1252 long number (with the length varying by architecture), so they can be
1253 unambiguously distinguished.
1255 * /usr/sbin/halt.local is no longer supported. Implementation in
1256 distributions was inconsistent and it seems this functionality was
1259 To replace this functionality, users should:
1260 - either define a new unit and make it a dependency of final.target
1261 (systemctl add-wants final.target my-halt-local.service)
1262 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
1263 and ensure that it accepts "halt", "poweroff", "reboot", and
1264 "kexec" as an argument, see the description in systemd-shutdown(8).
1266 * When a [Match] section in .link or .network file is empty (contains
1267 no match patterns), a warning will be emitted. Please add any "match
1268 all" pattern instead, e.g. OriginalName=* or Name=* in case all
1269 interfaces should really be matched.
1271 * A new setting NUMAPolicy= may be used to set process memory
1272 allocation policy. This setting can be specified in
1273 /etc/systemd/system.conf and hence will set the default policy for
1274 PID1. The default policy can be overridden on a per-service
1275 basis. The related setting NUMAMask= is used to specify NUMA node
1276 mask that should be associated with the selected policy.
1278 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
1279 generates when processes it manages are reaching their memory limits,
1280 and will place their units in a special state, and optionally kill or
1281 stop the whole unit.
1283 * The service manager will now expose bus properties for the IO
1284 resources used by units. This information is also shown in "systemctl
1285 status" now (for services that have IOAccounting=yes set). Moreover,
1286 the IO accounting data is included in the resource log message
1287 generated whenever a unit stops.
1289 * Units may now configure an explicit timeout to wait for when killed
1290 with SIGABRT, for example when a service watchdog is hit. Previously,
1291 the regular TimeoutStopSec= timeout was applied in this case too —
1292 now a separate timeout may be set using TimeoutAbortSec=.
1294 * Services may now send a special WATCHDOG=trigger message with
1295 sd_notify() to trigger an immediate "watchdog missed" event, and thus
1296 trigger service termination. This is useful both for testing watchdog
1297 handling, but also for defining error paths in services, that shall
1298 be handled the same way as watchdog events.
1300 * There are two new per-unit settings IPIngressFilterPath= and
1301 IPEgressFilterPath= which allow configuration of a BPF program
1302 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
1303 to apply to the IP packet ingress/egress path of all processes of a
1304 unit. This is useful to allow running systemd services with BPF
1305 programs set up externally.
1307 * systemctl gained a new "clean" verb for removing the state, cache,
1308 runtime or logs directories of a service while it is terminated. The
1309 new verb may also be used to remove the state maintained on disk for
1310 timer units that have Persistent= configured.
1312 * During the last phase of shutdown systemd will now automatically
1313 increase the log level configured in the "kernel.printk" sysctl so
1314 that any relevant loggable events happening during late shutdown are
1315 made visible. Previously, loggable events happening so late during
1316 shutdown were generally lost if the "kernel.printk" sysctl was set to
1317 high thresholds, as regular logging daemons are terminated at that
1318 time and thus nothing is written to disk.
1320 * If processes terminated during the last phase of shutdown do not exit
1321 quickly systemd will now show their names after a short time, to make
1322 debugging easier. After a longer timeout they are forcibly killed,
1325 * journalctl (and the other tools that display logs) will now highlight
1326 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
1327 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
1328 are now shown in blue color, to separate them visually from regular
1329 logs. References to configuration files are now turned into clickable
1330 links on terminals that support that.
1332 * systemd-journald will now stop logging to /var/log/journal during
1333 shutdown when /var/ is on a separate mount, so that it can be
1334 unmounted safely during shutdown.
1336 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
1338 * systemd-resolved "Cache=" configuration option in resolved.conf has
1339 been extended to also accept the 'no-negative' value. Previously,
1340 only a boolean option was allowed (yes/no), having yes as the
1341 default. If this option is set to 'no-negative', negative answers are
1342 not cached while the old cache heuristics are used positive answers.
1343 The default remains unchanged.
1345 * The predictable naming scheme for network devices now supports
1346 generating predictable names for "netdevsim" devices.
1348 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
1351 Those two changes form a new net.naming-policy-scheme= entry.
1352 Distributions which want to preserve naming stability may want to set
1353 the -Ddefault-net-naming-scheme= configuration option.
1355 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
1356 interfaces natively.
1358 * systemd-networkd's bridge FDB support now allows configuration of a
1359 destination address for each entry (Destination=), as well as the
1360 VXLAN VNI (VNI=), as well as an option to declare what an entry is
1361 associated with (AssociatedWith=).
1363 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
1364 option for configuring the maximum number of DHCP lease requests. It
1365 also learnt a new BlackList= option for deny-listing DHCP servers (a
1366 similar setting has also been added to the IPv6 RA client), as well
1367 as a SendRelease= option for configuring whether to send a DHCP
1368 RELEASE message when terminating.
1370 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
1371 separately in the [DHCPv4] and [DHCPv6] sections.
1373 * systemd-networkd's DHCP support will now optionally create an
1374 implicit host route to the DNS server specified in the DHCP lease, in
1375 addition to the routes listed explicitly in the lease. This should
1376 ensure that in multi-homed systems DNS traffic leaves the systems on
1377 the interface that acquired the DNS server information even if other
1378 routes such as default routes exist. This behaviour may be turned on
1379 with the new RoutesToDNS= option.
1381 * systemd-networkd's VXLAN support gained a new option
1382 GenericProtocolExtension= for enabling VXLAN Generic Protocol
1383 Extension support, as well as IPDoNotFragment= for setting the IP
1384 "Don't fragment" bit on outgoing packets. A similar option has been
1385 added to the GENEVE support.
1387 * In systemd-networkd's [Route] section you may now configure
1388 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
1389 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
1390 propagation. The Type= setting now supports local, broadcast,
1391 anycast, multicast, any, xresolve routes, too.
1393 * systemd-networkd's [Network] section learnt a new option
1394 DefaultRouteOnDevice= for automatically configuring a default route
1395 onto the network device.
1397 * systemd-networkd's bridging support gained two new options ProxyARP=
1398 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
1399 MulticastRouter= for configuring multicast routing behaviour. A new
1400 option MulticastIGMPVersion= may be used to change bridge's multicast
1401 Internet Group Management Protocol (IGMP) version.
1403 * systemd-networkd's FooOverUDP support gained the ability to configure
1404 local and peer IP addresses via Local= and Peer=. A new option
1405 PeerPort= may be used to configure the peer's IP port.
1407 * systemd-networkd's TUN support gained a new setting VnetHeader= for
1408 tweaking Generic Segment Offload support.
1410 * The address family for policy rules may be specified using the new
1411 Family= option in the [RoutingPolicyRule] section.
1413 * networkctl gained a new "delete" command for removing virtual network
1414 devices, as well as a new "--stats" switch for showing device
1417 * networkd.conf gained a new setting SpeedMeter= and
1418 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
1419 measured speed may be shown by 'networkctl status'.
1421 * "networkctl status" now displays MTU and queue lengths, and more
1422 detailed information about VXLAN and bridge devices.
1424 * systemd-networkd's .network and .link files gained a new Property=
1425 setting in the [Match] section, to match against devices with
1426 specific udev properties.
1428 * systemd-networkd's tunnel support gained a new option
1429 AssignToLoopback= for selecting whether to use the loopback device
1430 "lo" as underlying device.
1432 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
1433 been renamed to LinkLayerAddress=, and it now allows configuration of
1436 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
1437 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
1438 IPv6 configuration (static or DHCPv6) was found for a given
1439 interface. It will not touch the sysctl otherwise.
1441 * The order of entries is $PATH used by the user manager instance was
1442 changed to put bin/ entries before the corresponding sbin/ entries.
1443 It is recommended to not rely on this order, and only ever have one
1444 binary with a given name in the system paths under /usr.
1446 * A new tool systemd-network-generator has been added that may generate
1447 .network, .netdev and .link files from IP configuration specified on
1448 the kernel command line in the format used by Dracut.
1450 * The CriticalConnection= setting in .network files is now deprecated,
1451 and replaced by a new KeepConfiguration= setting which allows more
1452 detailed configuration of the IP configuration to keep in place.
1454 * systemd-analyze gained a few new verbs:
1456 - "systemd-analyze timestamp" parses and converts timestamps. This is
1457 similar to the existing "systemd-analyze calendar" command which
1458 does the same for recurring calendar events.
1460 - "systemd-analyze timespan" parses and converts timespans (i.e.
1461 durations as opposed to points in time).
1463 - "systemd-analyze condition" will parse and test ConditionXYZ=
1466 - "systemd-analyze exit-status" will parse and convert exit status
1467 codes to their names and back.
1469 - "systemd-analyze unit-files" will print a list of all unit
1470 file paths and unit aliases.
1472 * SuccessExitStatus=, RestartPreventExitStatus=, and
1473 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
1474 is equivalent to "65"). Those exit status name mappings may be
1475 displayed with the systemd-analyze exit-status verb describe above.
1477 * systemd-logind now exposes a per-session SetBrightness() bus call,
1478 which may be used to securely change the brightness of a kernel
1479 brightness device, if it belongs to the session's seat. By using this
1480 call unprivileged clients can make changes to "backlight" and "leds"
1481 devices securely with strict requirements on session membership.
1482 Desktop environments may use this to generically make brightness
1483 changes to such devices without shipping private SUID binaries or
1484 udev rules for that purpose.
1486 * "udevadm info" gained a --wait-for-initialization switch to wait for
1487 a device to be initialized.
1489 * systemd-hibernate-resume-generator will now look for resumeflags= on
1490 the kernel command line, which is similar to rootflags= and may be
1491 used to configure device timeout for the hibernation device.
1493 * sd-event learnt a new API call sd_event_source_disable_unref() for
1494 disabling and unref'ing an event source in a single function. A
1495 related call sd_event_source_disable_unrefp() has been added for use
1496 with gcc's cleanup extension.
1498 * The sd-id128.h public API gained a new definition
1499 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
1502 * "busctl introspect" gained a new switch --xml-interface for dumping
1503 XML introspection data unmodified.
1505 * PID 1 may now show the unit name instead of the unit description
1506 string in its status output during boot. This may be configured in
1507 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
1508 kernel command line option systemd.status_unit_format=.
1510 * PID 1 now understands a new option KExecWatchdogSec= in
1511 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
1512 Previously watchdog functionality was only available for regular
1513 reboots. The new setting defaults to off, because we don't know in
1514 the general case if the watchdog will be reset after kexec (some
1515 drivers do reset it, but not all), and the new userspace might not be
1516 configured to handle the watchdog.
1518 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
1519 RebootWatchdogSec= to more clearly communicate what it is about. The
1520 old name is still accepted for compatibility.
1522 * The systemd.debug_shell kernel command line option now optionally
1523 takes a tty name to spawn the debug shell on, which allows a
1524 different tty to be selected than the built-in default.
1526 * Service units gained a new ExecCondition= setting which will run
1527 before ExecStartPre= and either continue execution of the unit (for
1528 clean exit codes), stop execution without marking the unit failed
1529 (for exit codes 1 through 254), or stop execution and fail the unit
1530 (for exit code 255 or abnormal termination).
1532 * A new service systemd-pstore.service has been added that pulls data
1533 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
1536 * timedatectl gained new verbs for configuring per-interface NTP
1537 service configuration for systemd-timesyncd.
1539 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
1540 2019. (You can set non-UTF-8 locales though, if you know their name.)
1542 * If variable assignments in sysctl.d/ files are prefixed with "-" any
1543 failures to apply them are now ignored.
1545 * systemd-random-seed.service now optionally credits entropy when
1546 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
1547 true for the service to enable this behaviour, but please consult the
1548 documentation first, since this comes with a couple of caveats.
1550 * systemd-random-seed.service is now a synchronization point for full
1551 initialization of the kernel's entropy pool. Services that require
1552 /dev/urandom to be correctly initialized should be ordered after this
1555 * The systemd-boot boot loader has been updated to optionally maintain
1556 a random seed file in the EFI System Partition (ESP). During the boot
1557 phase, this random seed is read and updated with a new seed
1558 cryptographically derived from it. Another derived seed is passed to
1559 the OS. The latter seed is then credited to the kernel's entropy pool
1560 very early during userspace initialization (from PID 1). This allows
1561 systems to boot up with a fully initialized kernel entropy pool from
1562 earliest boot on, and thus entirely removes all entropy pool
1563 initialization delays from systems using systemd-boot. Special care
1564 is taken to ensure different seeds are derived on system images
1565 replicated to multiple systems. "bootctl status" will show whether
1566 a seed was received from the boot loader.
1568 * bootctl gained two new verbs:
1570 - "bootctl random-seed" will generate the file in ESP and an EFI
1571 variable to allow a random seed to be passed to the OS as described
1574 - "bootctl is-installed" checks whether systemd-boot is currently
1577 * bootctl will warn if it detects that boot entries are misconfigured
1578 (for example if the kernel image was removed without purging the
1581 * A new document has been added describing systemd's use and support
1582 for the kernel's entropy pool subsystem:
1584 https://systemd.io/RANDOM_SEEDS
1586 * When the system is hibernated the swap device to write the
1587 hibernation image to is now automatically picked from all available
1588 swap devices, preferring the swap device with the highest configured
1589 priority over all others, and picking the device with the most free
1590 space if there are multiple devices with the highest priority.
1592 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
1593 option that permits selecting the timeout how long to wait for a
1594 device with an encryption key before asking for the password.
1596 * IOWeight= has learnt to properly set the IO weight when using the
1597 BFQ scheduler officially found in kernels 5.0+.
1599 * A new mailing list has been created for reporting of security issues:
1600 systemd-security@redhat.com. For mode details, see
1601 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
1603 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
1604 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
1605 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
1606 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
1607 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
1608 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
1609 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
1610 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
1611 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
1612 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
1613 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
1614 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
1615 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
1616 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
1617 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
1618 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
1619 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
1620 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
1621 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
1622 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
1623 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
1624 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
1625 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
1626 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
1627 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
1628 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
1629 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
1630 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
1631 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
1632 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
1634 – Camerino, 2019-09-03
1638 * In .link files, MACAddressPolicy=persistent (the default) is changed
1639 to cover more devices. For devices like bridges, tun, tap, bond, and
1640 similar interfaces that do not have other identifying information,
1641 the interface name is used as the basis for persistent seed for MAC
1642 and IPv4LL addresses. The way that devices that were handled
1643 previously is not changed, and this change is about covering more
1644 devices then previously by the "persistent" policy.
1646 MACAddressPolicy=random may be used to force randomized MACs and
1647 IPv4LL addresses for a device if desired.
1649 Hint: the log output from udev (at debug level) was enhanced to
1650 clarify what policy is followed and which attributes are used.
1651 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
1652 may be used to view this.
1654 Hint: if a bridge interface is created without any slaves, and gains
1655 a slave later, then now the bridge does not inherit slave's MAC.
1656 To inherit slave's MAC, for example, create the following file:
1658 # /etc/systemd/network/98-bridge-inherit-mac.link
1663 MACAddressPolicy=none
1666 * The .device units generated by systemd-fstab-generator and other
1667 generators do not automatically pull in the corresponding .mount unit
1668 as a Wants= dependency. This means that simply plugging in the device
1669 will not cause the mount unit to be started automatically. But please
1670 note that the mount unit may be started for other reasons, in
1671 particular if it is part of local-fs.target, and any unit which
1672 (transitively) depends on local-fs.target is started.
1674 * networkctl list/status/lldp now accept globbing wildcards for network
1675 interface names to match against all existing interfaces.
1677 * The $PIDFILE environment variable is set to point the absolute path
1678 configured with PIDFile= for processes of that service.
1680 * The fallback DNS server list was augmented with Cloudflare public DNS
1681 servers. Use `-Ddns-servers=` to set a different fallback.
1683 * A new special target usb-gadget.target will be started automatically
1684 when a USB Device Controller is detected (which means that the system
1685 is a USB peripheral).
1687 * A new unit setting CPUQuotaPeriodSec= assigns the time period
1688 relatively to which the CPU time quota specified by CPUQuota= is
1691 * A new unit setting ProtectHostname= may be used to prevent services
1692 from modifying hostname information (even if they otherwise would
1693 have privileges to do so).
1695 * A new unit setting NetworkNamespacePath= may be used to specify a
1696 namespace for service or socket units through a path referring to a
1697 Linux network namespace pseudo-file.
1699 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
1700 have an effect on .socket units: when used the listening socket is
1701 created within the configured network namespace instead of the host
1704 * ExecStart= command lines in unit files may now be prefixed with ':'
1705 in which case environment variable substitution is
1706 disabled. (Supported for the other ExecXYZ= settings, too.)
1708 * .timer units gained two new boolean settings OnClockChange= and
1709 OnTimezoneChange= which may be used to also trigger a unit when the
1710 system clock is changed or the local timezone is
1711 modified. systemd-run has been updated to make these options easily
1712 accessible from the command line for transient timers.
1714 * Two new conditions for units have been added: ConditionMemory= may be
1715 used to conditionalize a unit based on installed system
1716 RAM. ConditionCPUs= may be used to conditionalize a unit based on
1717 installed CPU cores.
1719 * The @default system call filter group understood by SystemCallFilter=
1720 has been updated to include the new rseq() system call introduced in
1723 * A new time-set.target has been added that indicates that the system
1724 time has been set from a local source (possibly imprecise). The
1725 existing time-sync.target is stronger and indicates that the time has
1726 been synchronized with a precise external source. Services where
1727 approximate time is sufficient should use the new target.
1729 * "systemctl start" (and related commands) learnt a new
1730 --show-transaction option. If specified brief information about all
1731 jobs queued because of the requested operation is shown.
1733 * systemd-networkd recognizes a new operation state 'enslaved', used
1734 (instead of 'degraded' or 'carrier') for interfaces which form a
1735 bridge, bond, or similar, and an new 'degraded-carrier' operational
1736 state used for the bond or bridge master interface when one of the
1737 enslaved devices is not operational.
1739 * .network files learnt the new IgnoreCarrierLoss= option for leaving
1740 networks configured even if the carrier is lost.
1742 * The RequiredForOnline= setting in .network files may now specify a
1743 minimum operational state required for the interface to be considered
1744 "online" by systemd-networkd-wait-online. Related to this
1745 systemd-networkd-wait-online gained a new option --operational-state=
1746 to configure the same, and its --interface= option was updated to
1747 optionally also take an operational state specific for an interface.
1749 * systemd-networkd-wait-online gained a new setting --any for waiting
1750 for only one of the requested interfaces instead of all of them.
1752 * systemd-networkd now implements L2TP tunnels.
1754 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
1755 may be used to cause autonomous and onlink prefixes received in IPv6
1756 Router Advertisements to be ignored.
1758 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
1759 file settings may be used to tweak bridge behaviour.
1761 * The new TripleSampling= option in .network files may be used to
1762 configure CAN triple sampling.
1764 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
1765 used to point to private or preshared key for a WireGuard interface.
1767 * /etc/crypttab now supports the same-cpu-crypt and
1768 submit-from-crypt-cpus options to tweak encryption work scheduling
1771 * systemd-tmpfiles will now take a BSD file lock before operating on a
1772 contents of directory. This may be used to temporarily exclude
1773 directories from aging by taking the same lock (useful for example
1774 when extracting a tarball into /tmp or /var/tmp as a privileged user,
1775 which might create files with really old timestamps, which
1776 nevertheless should not be deleted). For further details, see:
1778 https://systemd.io/TEMPORARY_DIRECTORIES
1780 * systemd-tmpfiles' h line type gained support for the
1781 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
1782 controlling project quota inheritance.
1784 * sd-boot and bootctl now implement support for an Extended Boot Loader
1785 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
1786 addition to the ESP partition mounted to /efi or /boot/efi.
1787 Configuration file fragments, kernels, initrds and other EFI images
1788 to boot will be loaded from both the ESP and XBOOTLDR partitions.
1789 The XBOOTLDR partition was previously described by the Boot Loader
1790 Specification, but implementation was missing in sd-boot. Support for
1791 this concept allows using the sd-boot boot loader in more
1792 conservative scenarios where the boot loader itself is placed in the
1793 ESP but the kernels to boot (and their metadata) in a separate
1796 * A system may now be booted with systemd.volatile=overlay on the
1797 kernel command line, which causes the root file system to be set up
1798 an overlayfs mount combining the root-only root directory with a
1799 writable tmpfs. In this setup, the underlying root device is not
1800 modified, and any changes are lost at reboot.
1802 * Similar, systemd-nspawn can now boot containers with a volatile
1803 overlayfs root with the new --volatile=overlay switch.
1805 * systemd-nspawn can now consume OCI runtime bundles using a new
1806 --oci-bundle= option. This implementation is fully usable, with most
1807 features in the specification implemented, but since this a lot of
1808 new code and functionality, this feature should most likely not
1809 be used in production yet.
1811 * systemd-nspawn now supports various options described by the OCI
1812 runtime specification on the command-line and in .nspawn files:
1813 --inaccessible=/Inaccessible= may be used to mask parts of the file
1814 system tree, --console=/--pipe may be used to configure how standard
1815 input, output, and error are set up.
1817 * busctl learned the `emit` verb to generate D-Bus signals.
1819 * systemd-analyze cat-config may be used to gather and display
1820 configuration spread over multiple files, for example system and user
1821 presets, tmpfiles.d, sysusers.d, udev rules, etc.
1823 * systemd-analyze calendar now takes an optional new parameter
1824 --iterations= which may be used to show a maximum number of iterations
1825 the specified expression will elapse next.
1827 * The sd-bus C API gained support for naming method parameters in the
1830 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
1831 the reboot() system call expects.
1833 * journalctl learnt a new --cursor-file= option that points to a file
1834 from which a cursor should be loaded in the beginning and to which
1835 the updated cursor should be stored at the end.
1837 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
1838 detected by systemd-detect-virt (and may also be used in
1839 ConditionVirtualization=).
1841 * The behaviour of systemd-logind may now be modified with environment
1842 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
1843 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
1844 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
1845 skip the relevant operation completely (when set to false), or to
1846 create a flag file in /run/systemd (when set to true), instead of
1847 actually commencing the real operation when requested. The presence
1848 of /run/systemd/reboot-to-firmware-setup,
1849 /run/systemd/reboot-to-boot-loader-menu, and
1850 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
1851 boot loader implementations to replace some steps logind performs
1852 during reboot with their own operations.
1854 * systemctl can be used to request a reboot into the boot loader menu
1855 or a specific boot loader entry with the new --boot-load-menu= and
1856 --boot-loader-entry= options to a reboot command. (This requires a
1857 boot loader that supports this, for example sd-boot.)
1859 * kernel-install will no longer unconditionally create the output
1860 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
1861 snippets, but will do only if the machine-specific parent directory
1862 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
1863 to create this parent directory during sd-boot installation.
1865 This makes it easier to use kernel-install with plugins which support
1866 a different layout of the bootloader partitions (for example grub2).
1868 * During package installation (with `ninja install`), we would create
1869 symlinks for getty@tty1.service, systemd-networkd.service,
1870 systemd-networkd.socket, systemd-resolved.service,
1871 remote-cryptsetup.target, remote-fs.target,
1872 systemd-networkd-wait-online.service, and systemd-timesyncd.service
1873 in /etc, as if `systemctl enable` was called for those units, to make
1874 the system usable immediately after installation. Now this is not
1875 done anymore, and instead calling `systemctl preset-all` is
1876 recommended after the first installation of systemd.
1878 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
1879 is built on seccomp. When turned on creation of SUID/SGID files is
1882 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
1883 implied if DynamicUser= is turned on for a service. This hardens
1884 these services, so that they neither can benefit from nor create
1885 SUID/SGID executables. This is a minor compatibility breakage, given
1886 that when DynamicUser= was first introduced SUID/SGID behaviour was
1887 unaffected. However, the security benefit of these two options is
1888 substantial, and the setting is still relatively new, hence we opted
1889 to make it mandatory for services with dynamic users.
1891 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
1892 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
1893 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
1894 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
1895 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
1896 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
1897 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
1898 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
1899 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
1900 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
1901 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
1902 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
1903 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
1904 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
1905 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
1906 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
1907 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
1908 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1910 — Warsaw, 2019-04-11
1914 * The default locale can now be configured at compile time. Otherwise,
1915 a suitable default will be selected automatically (one of C.UTF-8,
1916 en_US.UTF-8, and C).
1918 * The version string shown by systemd and other tools now includes the
1919 git commit hash when built from git. An override may be specified
1920 during compilation, which is intended to be used by distributions to
1921 include the package release information.
1923 * systemd-cat can now filter standard input and standard error streams
1924 for different syslog priorities using the new --stderr-priority=
1927 * systemd-journald and systemd-journal-remote reject entries which
1928 contain too many fields (CVE-2018-16865) and set limits on the
1929 process' command line length (CVE-2018-16864).
1931 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
1934 * A new network device NamePolicy "keep" is implemented for link files,
1935 and used by default in 99-default.link (the fallback configuration
1936 provided by systemd). With this policy, if the network device name
1937 was already set by userspace, the device will not be renamed again.
1938 This matches the naming scheme that was implemented before
1939 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
1940 is also enabled by default, even if not specified. Effectively, this
1941 means that if naming-scheme >= 240 is specified, network devices will
1942 be renamed according to the configuration, even if they have been
1943 renamed already, if "keep" is not specified as the naming policy in
1944 the .link file. The 99-default.link file provided by systemd includes
1945 "keep" for backwards compatibility, but it is recommended for user
1946 installed .link files to *not* include it.
1948 The "kernel" policy, which keeps kernel names declared to be
1949 "persistent", now works again as documented.
1951 * kernel-install script now optionally takes the paths to one or more
1952 initrd files, and passes them to all plugins.
1954 * The mincore() system call has been dropped from the @system-service
1955 system call filter group, as it is pretty exotic and may potentially
1956 used for side-channel attacks.
1958 * -fPIE is dropped from compiler and linker options. Please specify
1959 -Db_pie=true option to meson to build position-independent
1960 executables. Note that the meson option is supported since meson-0.49.
1962 * The fs.protected_regular and fs.protected_fifos sysctls, which were
1963 added in Linux 4.19 to make some data spoofing attacks harder, are
1964 now enabled by default. While this will hopefully improve the
1965 security of most installations, it is technically a backwards
1966 incompatible change; to disable these sysctls again, place the
1967 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
1969 fs.protected_regular = 0
1970 fs.protected_fifos = 0
1972 Note that the similar hardlink and symlink protection has been
1973 enabled since v199, and may be disabled likewise.
1975 * The files read from the EnvironmentFile= setting in unit files now
1976 parse backslashes inside quotes literally, matching the behaviour of
1979 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
1980 now automatically become NOPs when run in a chroot() environment.
1982 * The tmpfiles.d/ "C" line type will now copy directory trees not only
1983 when the destination is so far missing, but also if it already exists
1984 as a directory and is empty. This is useful to cater for systems
1985 where directory trees are put together from multiple separate mount
1986 points but otherwise empty.
1988 * A new function sd_bus_close_unref() (and the associated
1989 sd_bus_close_unrefp()) has been added to libsystemd, that combines
1990 sd_bus_close() and sd_bus_unref() in one.
1992 * udevadm control learnt a new option for --ping for testing whether a
1993 systemd-udevd instance is running and reacting.
1995 * udevadm trigger learnt a new option for --wait-daemon for waiting
1996 systemd-udevd daemon to be initialized.
1998 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
1999 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
2000 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
2001 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
2002 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
2003 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
2004 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
2005 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
2006 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
2007 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
2008 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
2009 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
2010 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
2011 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
2012 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
2013 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
2014 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
2016 — Berlin, 2019-02-14
2020 * NoNewPrivileges=yes has been set for all long-running services
2021 implemented by systemd. Previously, this was problematic due to
2022 SELinux (as this would also prohibit the transition from PID1's label
2023 to the service's label). This restriction has since been lifted, but
2024 an SELinux policy update is required.
2025 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
2027 * DynamicUser=yes is dropped from systemd-networkd.service,
2028 systemd-resolved.service and systemd-timesyncd.service, which was
2029 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
2030 and since v236 for systemd-timesyncd.service. The users and groups
2031 systemd-network, systemd-resolve and systemd-timesync are created
2032 by systemd-sysusers again. Distributors or system administrators
2033 may need to create these users and groups if they not exist (or need
2034 to re-enable DynamicUser= for those units) while upgrading systemd.
2035 Also, the clock file for systemd-timesyncd may need to move from
2036 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
2038 * When unit files are loaded from disk, previously systemd would
2039 sometimes (depending on the unit loading order) load units from the
2040 target path of symlinks in .wants/ or .requires/ directories of other
2041 units. This meant that unit could be loaded from different paths
2042 depending on whether the unit was requested explicitly or as a
2043 dependency of another unit, not honouring the priority of directories
2044 in search path. It also meant that it was possible to successfully
2045 load and start units which are not found in the unit search path, as
2046 long as they were requested as a dependency and linked to from
2047 .wants/ or .requires/. The target paths of those symlinks are not
2048 used for loading units anymore and the unit file must be found in
2051 * A new service type has been added: Type=exec. It's very similar to
2052 Type=simple but ensures the service manager will wait for both fork()
2053 and execve() of the main service binary to complete before proceeding
2054 with follow-up units. This is primarily useful so that the manager
2055 propagates any errors in the preparation phase of service execution
2056 back to the job that requested the unit to be started. For example,
2057 consider a service that has ExecStart= set to a file system binary
2058 that doesn't exist. With Type=simple starting the unit would be
2059 considered instantly successful, as only fork() has to complete
2060 successfully and the manager does not wait for execve(), and hence
2061 its failure is seen "too late". With the new Type=exec service type
2062 starting the unit will fail, as the manager will wait for the
2063 execve() and notice its failure, which is then propagated back to the
2066 NOTE: with the next release 241 of systemd we intend to change the
2067 systemd-run tool to default to Type=exec for transient services
2068 started by it. This should be mostly safe, but in specific corner
2069 cases might result in problems, as the systemd-run tool will then
2070 block on NSS calls (such as user name look-ups due to User=) done
2071 between the fork() and execve(), which under specific circumstances
2072 might cause problems. It is recommended to specify "-p Type=simple"
2073 explicitly in the few cases where this applies. For regular,
2074 non-transient services (i.e. those defined with unit files on disk)
2075 we will continue to default to Type=simple.
2077 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
2078 userspace processes is set to 1024 (soft) and 4096
2079 (hard). Previously, systemd passed this on unmodified to all
2080 processes it forked off. With this systemd release the hard limit
2081 systemd passes on is increased to 512K, overriding the kernel's
2082 defaults and substantially increasing the number of simultaneous file
2083 descriptors unprivileged userspace processes can allocate. Note that
2084 the soft limit remains at 1024 for compatibility reasons: the
2085 traditional UNIX select() call cannot deal with file descriptors >=
2086 1024 and increasing the soft limit globally might thus result in
2087 programs unexpectedly allocating a high file descriptor and thus
2088 failing abnormally when attempting to use it with select() (of
2089 course, programs shouldn't use select() anymore, and prefer
2090 poll()/epoll, but the call unfortunately remains undeservedly popular
2091 at this time). This change reflects the fact that file descriptor
2092 handling in the Linux kernel has been optimized in more recent
2093 kernels and allocating large numbers of them should be much cheaper
2094 both in memory and in performance than it used to be. Programs that
2095 want to take benefit of the increased limit have to "opt-in" into
2096 high file descriptors explicitly by raising their soft limit. Of
2097 course, when they do that they must acknowledge that they cannot use
2098 select() anymore (and neither can any shared library they use — or
2099 any shared library used by any shared library they use and so on).
2100 Which default hard limit is most appropriate is of course hard to
2101 decide. However, given reports that ~300K file descriptors are used
2102 in real-life applications we believe 512K is sufficiently high as new
2103 default for now. Note that there are also reports that using very
2104 high hard limits (e.g. 1G) is problematic: some software allocates
2105 large arrays with one element for each potential file descriptor
2106 (Java, …) — a high hard limit thus triggers excessively large memory
2107 allocations in these applications. Hopefully, the new default of 512K
2108 is a good middle ground: higher than what real-life applications
2109 currently need, and low enough for avoid triggering excessively large
2110 allocations in problematic software. (And yes, somebody should fix
2113 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
2114 to the highest possible values, as separate accounting of file
2115 descriptors is no longer necessary, as memcg tracks them correctly as
2116 part of the memory accounting anyway. Thus, from the four limits on
2117 file descriptors currently enforced (fs.file-max, fs.nr_open,
2118 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
2119 and keep only the latter two. A set of build-time options
2120 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
2121 has been added to revert this change in behaviour, which might be
2122 an option for systems that turn off memcg in the kernel.
2124 * When no /etc/locale.conf file exists (and hence no locale settings
2125 are in place), systemd will now use the "C.UTF-8" locale by default,
2126 and set LANG= to it. This locale is supported by various
2127 distributions including Fedora, with clear indications that upstream
2128 glibc is going to make it available too. This locale enables UTF-8
2129 mode by default, which appears appropriate for 2018.
2131 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
2132 default. This effectively switches the RFC3704 Reverse Path filtering
2133 from Strict mode to Loose mode. This is more appropriate for hosts
2134 that have multiple links with routes to the same networks (e.g.
2135 a client with a Wi-Fi and Ethernet both connected to the internet).
2137 Consult the kernel documentation for details on this sysctl:
2138 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
2140 * CPUAccounting=yes no longer enables the CPU controller when using
2141 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
2142 statistics are now provided independently from the CPU controller.
2144 * Support for disabling a particular cgroup controller within a sub-tree
2145 has been added through the DisableControllers= directive.
2147 * cgroup_no_v1=all on the kernel command line now also implies
2148 using the unified cgroup hierarchy, unless one explicitly passes
2149 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
2151 * The new "MemoryMin=" unit file property may now be used to set the
2152 memory usage protection limit of processes invoked by the unit. This
2153 controls the cgroup v2 memory.min attribute. Similarly, the new
2154 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
2155 cgroup v2 io.latency cgroup property for configuring per-service I/O
2158 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
2159 to the cgroup v1 "devices" cgroup controller.
2161 * systemd-escape now is able to combine --unescape with --template. It
2162 also learnt a new option --instance for extracting and unescaping the
2163 instance part of a unit name.
2165 * sd-bus now provides the sd_bus_message_readv() which is similar to
2166 sd_bus_message_read() but takes a va_list object. The pair
2167 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
2168 has been added for configuring the default method call timeout to
2169 use. sd_bus_error_move() may be used to efficiently move the contents
2170 from one sd_bus_error structure to another, invalidating the
2171 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
2172 be used to control whether a bus connection object is automatically
2173 flushed when an sd-event loop is exited.
2175 * When processing classic BSD syslog log messages, journald will now
2176 save the original time-stamp string supplied in the new
2177 SYSLOG_TIMESTAMP= journal field. This permits consumers to
2178 reconstruct the original BSD syslog message more correctly.
2180 * StandardOutput=/StandardError= in service files gained support for
2181 new "append:…" parameters, for connecting STDOUT/STDERR of a service
2182 to a file, and appending to it.
2184 * The signal to use as last step of killing of unit processes is now
2185 configurable. Previously it was hard-coded to SIGKILL, which may now
2186 be overridden with the new KillSignal= setting. Note that this is the
2187 signal used when regular termination (i.e. SIGTERM) does not suffice.
2188 Similarly, the signal used when aborting a program in case of a
2189 watchdog timeout may now be configured too (WatchdogSignal=).
2191 * The XDG_SESSION_DESKTOP environment variable may now be configured in
2192 the pam_systemd argument line, using the new desktop= switch. This is
2193 useful to initialize it properly from a display manager without
2194 having to touch C code.
2196 * Most configuration options that previously accepted percentage values
2197 now also accept permille values with the '‰' suffix (instead of '%').
2199 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
2202 * systemd-resolved's configuration file resolved.conf gained a new
2203 option ReadEtcHosts= which may be used to turn off processing and
2204 honoring /etc/hosts entries.
2206 * The "--wait" switch may now be passed to "systemctl
2207 is-system-running", in which case the tool will synchronously wait
2208 until the system finished start-up.
2210 * hostnamed gained a new bus call to determine the DMI product UUID.
2212 * On x86-64 systemd will now prefer using the RDRAND processor
2213 instruction over /dev/urandom whenever it requires randomness that
2214 neither has to be crypto-grade nor should be reproducible. This
2215 should substantially reduce the amount of entropy systemd requests
2216 from the kernel during initialization on such systems, though not
2217 reduce it to zero. (Why not zero? systemd still needs to allocate
2218 UUIDs and such uniquely, which require high-quality randomness.)
2220 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
2221 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
2222 for forcing the "Other Information" bit in IPv6 RA messages. The
2223 bonding logic gained four new options AdActorSystemPriority=,
2224 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
2225 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
2226 shuffling of flows. The tunnel logic gained a new
2227 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
2228 Deployment. The policy rule logic gained four new options IPProtocol=,
2229 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
2230 support for the MulticastToUnicast= option. networkd also gained
2231 support for configuring static IPv4 ARP or IPv6 neighbor entries.
2233 * .preset files (as read by 'systemctl preset') may now be used to
2234 instantiate services.
2236 * /etc/crypttab now understands the sector-size= option to configure
2237 the sector size for an encrypted partition.
2239 * Key material for encrypted disks may now be placed on a formatted
2240 medium, and referenced from /etc/crypttab by the UUID of the file
2241 system, followed by "=" suffixed by the path to the key file.
2243 * The "collect" udev component has been removed without replacement, as
2244 it is neither used nor maintained.
2246 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
2247 LogsDirectory=, ConfigurationDirectory= settings are used in a
2248 service the executed processes will now receive a set of environment
2249 variables containing the full paths of these directories.
2250 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
2251 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
2252 are used. Note that these options may be used multiple times per
2253 service in which case the resulting paths will be concatenated and
2254 separated by colons.
2256 * Predictable interface naming has been extended to cover InfiniBand
2257 NICs. They will be exposed with an "ib" prefix.
2259 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
2260 which case the respective line failing is ignored.
2262 * .link files may now be used to configure the equivalent to the
2263 "ethtool advertise" commands.
2265 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
2266 alternative to libudev.h. Previously, the latter was just an internal
2267 wrapper around the former, but now these two APIs are exposed
2270 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
2271 which calculates an app-specific boot ID similar to how
2272 sd_id128_get_machine_app_specific() generates an app-specific machine
2275 * A new tool systemd-id128 has been added that can be used to determine
2276 and generate various 128bit IDs.
2278 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
2281 * systemd-hibernate-resume-generator will now honor the "noresume"
2282 kernel command line option, in which case it will bypass resuming
2283 from any hibernated image.
2285 * The systemd-sleep.conf configuration file gained new options
2286 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
2287 AllowHybridSleep= for prohibiting specific sleep modes even if the
2288 kernel exports them.
2290 * portablectl is now officially supported and has thus moved to
2293 * bootctl learnt the two new commands "set-default" and "set-oneshot"
2294 for setting the default boot loader item to boot to (either
2295 persistently or only for the next boot). This is currently only
2296 compatible with sd-boot, but may be implemented on other boot loaders
2297 too, that follow the boot loader interface. The updated interface is
2298 now documented here:
2300 https://systemd.io/BOOT_LOADER_INTERFACE
2302 * A new kernel command line option systemd.early_core_pattern= is now
2303 understood which may be used to influence the core_pattern PID 1
2304 installs during early boot.
2306 * busctl learnt two new options -j and --json= for outputting method
2307 call replies, properties and monitoring output in JSON.
2309 * journalctl's JSON output now supports simple ANSI coloring as well as
2310 a new "json-seq" mode for generating RFC7464 output.
2312 * Unit files now support the %g/%G specifiers that resolve to the UNIX
2313 group/GID of the service manager runs as, similar to the existing
2314 %u/%U specifiers that resolve to the UNIX user/UID.
2316 * systemd-logind learnt a new global configuration option
2317 UserStopDelaySec= that may be set in logind.conf. It specifies how
2318 long the systemd --user instance shall remain started after a user
2319 logs out. This is useful to speed up repetitive re-connections of the
2320 same user, as it means the user's service manager doesn't have to be
2321 stopped/restarted on each iteration, but can be reused between
2322 subsequent options. This setting defaults to 10s. systemd-logind also
2323 exports two new properties on its Manager D-Bus objects indicating
2324 whether the system's lid is currently closed, and whether the system
2327 * systemd gained support for a generic boot counting logic, which
2328 generically permits automatic reverting to older boot loader entries
2329 if newer updated ones don't work. The boot loader side is implemented
2330 in sd-boot, but is kept open for other boot loaders too. For details
2333 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
2335 * The SuccessAction=/FailureAction= unit file settings now learnt two
2336 new parameters: "exit" and "exit-force", which result in immediate
2337 exiting of the service manager, and are only useful in systemd --user
2338 and container environments.
2340 * Unit files gained support for a pair of options
2341 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
2342 exit status to use as service manager exit status when
2343 SuccessAction=/FailureAction= is set to exit or exit-force.
2345 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
2346 options may now be used to configure the log rate limiting applied by
2347 journald per-service.
2349 * systemd-analyze gained a new verb "timespan" for parsing and
2350 normalizing time span values (i.e. strings like "5min 7s 8us").
2352 * systemd-analyze also gained a new verb "security" for analyzing the
2353 security and sand-boxing settings of services in order to determine an
2354 "exposure level" for them, indicating whether a service would benefit
2355 from more sand-boxing options turned on for them.
2357 * "systemd-analyze syscall-filter" will now also show system calls
2358 supported by the local kernel but not included in any of the defined
2361 * .nspawn files now understand the Ephemeral= setting, matching the
2362 --ephemeral command line switch.
2364 * sd-event gained the new APIs sd_event_source_get_floating() and
2365 sd_event_source_set_floating() for controlling whether a specific
2366 event source is "floating", i.e. destroyed along with the even loop
2369 * Unit objects on D-Bus gained a new "Refs" property that lists all
2370 clients that currently have a reference on the unit (to ensure it is
2373 * The JoinControllers= option in system.conf is no longer supported, as
2374 it didn't work correctly, is hard to support properly, is legacy (as
2375 the concept only exists on cgroup v1) and apparently wasn't used.
2377 * Journal messages that are generated whenever a unit enters the failed
2378 state are now tagged with a unique MESSAGE_ID. Similarly, messages
2379 generated whenever a service process exits are now made recognizable,
2380 too. A tagged message is also emitted whenever a unit enters the
2381 "dead" state on success.
2383 * systemd-run gained a new switch --working-directory= for configuring
2384 the working directory of the service to start. A shortcut -d is
2385 equivalent, setting the working directory of the service to the
2386 current working directory of the invoking program. The new --shell
2387 (or just -S) option has been added for invoking the $SHELL of the
2388 caller as a service, and implies --pty --same-dir --wait --collect
2389 --service-type=exec. Or in other words, "systemd-run -S" is now the
2390 quickest way to quickly get an interactive in a fully clean and
2391 well-defined system service context.
2393 * machinectl gained a new verb "import-fs" for importing an OS tree
2394 from a directory. Moreover, when a directory or tarball is imported
2395 and single top-level directory found with the OS itself below the OS
2396 tree is automatically mangled and moved one level up.
2398 * systemd-importd will no longer set up an implicit btrfs loop-back
2399 file system on /var/lib/machines. If one is already set up, it will
2400 continue to be used.
2402 * A new generator "systemd-run-generator" has been added. It will
2403 synthesize a unit from one or more program command lines included in
2404 the kernel command line. This is very useful in container managers
2407 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
2409 This will run "systemd-nspawn" on an image, invoke the specified
2410 command line and immediately shut down the container again, returning
2411 the command line's exit code.
2413 * The block device locking logic is now documented:
2415 https://systemd.io/BLOCK_DEVICE_LOCKING
2417 * loginctl and machinectl now optionally output the various tables in
2418 JSON using the --output= switch. It is our intention to add similar
2419 support to systemctl and all other commands.
2421 * udevadm's query and trigger verb now optionally take a .device unit
2424 * systemd-udevd's network naming logic now understands a new
2425 net.naming-scheme= kernel command line switch, which may be used to
2426 pick a specific version of the naming scheme. This helps stabilizing
2427 interface names even as systemd/udev are updated and the naming logic
2430 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
2431 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
2432 initialize one to all 0xFF.
2434 * After loading the SELinux policy systemd will now recursively relabel
2435 all files and directories listed in
2436 /run/systemd/relabel-extra.d/*.relabel (which should be simple
2437 newline separated lists of paths) in addition to the ones it already
2438 implicitly relabels in /run, /dev and /sys. After the relabelling is
2439 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
2440 removed. This is useful to permit initrds (i.e. code running before
2441 the SELinux policy is in effect) to generate files in the host
2442 filesystem safely and ensure that the correct label is applied during
2443 the transition to the host OS.
2445 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
2446 mknod() handling in user namespaces. Previously mknod() would always
2447 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
2448 but device nodes generated that way cannot be opened, and attempts to
2449 open them result in EPERM. This breaks the "graceful fallback" logic
2450 in systemd's PrivateDevices= sand-boxing option. This option is
2451 implemented defensively, so that when systemd detects it runs in a
2452 restricted environment (such as a user namespace, or an environment
2453 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
2454 where device nodes cannot be created the effect of PrivateDevices= is
2455 bypassed (following the logic that 2nd-level sand-boxing is not
2456 essential if the system systemd runs in is itself already sand-boxed
2457 as a whole). This logic breaks with 4.18 in container managers where
2458 user namespacing is used: suddenly PrivateDevices= succeeds setting
2459 up a private /dev/ file system containing devices nodes — but when
2460 these are opened they don't work.
2462 At this point it is recommended that container managers utilizing
2463 user namespaces that intend to run systemd in the payload explicitly
2464 block mknod() with seccomp or similar, so that the graceful fallback
2467 We are very sorry for the breakage and the requirement to change
2468 container configurations for newer kernels. It's purely caused by an
2469 incompatible kernel change. The relevant kernel developers have been
2470 notified about this userspace breakage quickly, but they chose to
2473 * PermissionsStartOnly= setting is deprecated (but is still supported
2474 for backwards compatibility). The same functionality is provided by
2475 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
2478 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
2479 pam_systemd anymore.
2481 * The naming scheme for network devices was changed to always rename
2482 devices, even if they were already renamed by userspace. The "kernel"
2483 policy was changed to only apply as a fallback, if no other naming
2486 * The requirements to build systemd is bumped to meson-0.46 and
2489 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
2490 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
2491 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
2492 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
2493 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
2494 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
2495 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
2496 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
2497 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
2498 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
2499 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
2500 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
2501 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
2502 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
2503 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
2504 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
2505 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
2506 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
2507 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
2508 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
2509 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
2510 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
2511 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
2512 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
2513 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
2514 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
2515 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
2516 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
2517 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
2518 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
2519 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
2520 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
2521 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
2522 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
2523 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
2524 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
2525 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
2526 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
2527 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
2528 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
2529 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
2530 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
2531 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
2532 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
2533 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
2535 — Warsaw, 2018-12-21
2539 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
2540 builtin will name network interfaces differently than in previous
2541 versions for virtual network interfaces created with SR-IOV and NPAR
2542 and for devices where the PCI network controller device does not have
2543 a slot number associated.
2545 SR-IOV virtual devices are now named based on the name of the parent
2546 interface, with a suffix of "v<N>", where <N> is the virtual device
2547 number. Previously those virtual devices were named as if completely
2550 The ninth and later NPAR virtual devices will be named following the
2551 scheme used for the first eight NPAR partitions. Previously those
2552 devices were not renamed and the kernel default (eth<n>) was used.
2554 "net_id" will also generate names for PCI devices where the PCI
2555 network controller device does not have an associated slot number
2556 itself, but one of its parents does. Previously those devices were
2557 not renamed and the kernel default (eth<n>) was used.
2559 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
2560 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
2561 the unit. So, it is expected that the default behavior of
2562 systemd-logind is not changed. However, if distribution packagers or
2563 administrators disabled or modified IPAddressDeny= setting by a
2564 drop-in config file, then it may be necessary to update the file to
2565 re-enable AF_INET and AF_INET6 to support network user name services,
2568 * When the RestrictNamespaces= unit property is specified multiple
2569 times, then the specified types are merged now. Previously, only the
2570 last assignment was used. So, if distribution packagers or
2571 administrators modified the setting by a drop-in config file, then it
2572 may be necessary to update the file.
2574 * When OnFailure= is used in combination with Restart= on a service
2575 unit, then the specified units will no longer be triggered on
2576 failures that result in restarting. Previously, the specified units
2577 would be activated each time the unit failed, even when the unit was
2578 going to be restarted automatically. This behaviour contradicted the
2579 documentation. With this release the code is adjusted to match the
2582 * systemd-tmpfiles will now print a notice whenever it encounters
2583 tmpfiles.d/ lines referencing the /var/run/ directory. It will
2584 recommend reworking them to use the /run/ directory instead (for
2585 which /var/run/ is simply a symlinked compatibility alias). This way
2586 systemd-tmpfiles can properly detect line conflicts and merge lines
2587 referencing the same file by two paths, without having to access
2590 * systemctl disable/unmask/preset/preset-all cannot be used with
2591 --runtime. Previously this was allowed, but resulted in unintuitive
2592 behaviour that wasn't useful. systemctl disable/unmask will now undo
2593 both runtime and persistent enablement/masking, i.e. it will remove
2594 any relevant symlinks both in /run and /etc.
2596 * Note that all long-running system services shipped with systemd will
2597 now default to a system call allow list (rather than a deny list, as
2598 before). In particular, systemd-udevd will now enforce one too. For
2599 most cases this should be safe, however downstream distributions
2600 which disabled sandboxing of systemd-udevd (specifically the
2601 MountFlags= setting), might want to disable this security feature
2602 too, as the default allow-listing will prohibit all mount, swap,
2603 reboot and clock changing operations from udev rules.
2605 * sd-boot acquired new loader configuration settings to optionally turn
2606 off Windows and MacOS boot partition discovery as well as
2607 reboot-into-firmware menu items. It is also able to pick a better
2608 screen resolution for HiDPI systems, and now provides loader
2609 configuration settings to change the resolution explicitly.
2611 * systemd-resolved now supports DNS-over-TLS. It's still
2612 turned off by default, use DNSOverTLS=opportunistic to turn it on in
2613 resolved.conf. We intend to make this the default as soon as couple
2614 of additional techniques for optimizing the initial latency caused by
2615 establishing a TLS/TCP connection are implemented.
2617 * systemd-resolved.service and systemd-networkd.service now set
2618 DynamicUser=yes. The users systemd-resolve and systemd-network are
2619 not created by systemd-sysusers anymore.
2621 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
2622 that embed a network facing module into any process using getpwuid()
2623 or related call: the dynamic allocation of the user ID for
2624 systemd-resolved.service means the service manager has to check NSS
2625 if the user name is already taken when forking off the service. Since
2626 the user in the common case won't be defined in /etc/passwd the
2627 lookup is likely to trigger nss-ldap which in turn might use NSS to
2628 ask systemd-resolved for hostname lookups. This will hence result in
2629 a deadlock: a user name lookup in order to start
2630 systemd-resolved.service will result in a hostname lookup for which
2631 systemd-resolved.service needs to be started already. There are
2632 multiple ways to work around this problem: pre-allocate the
2633 "systemd-resolve" user on such systems, so that nss-ldap won't be
2634 triggered; or use a different NSS package that doesn't do networking
2635 in-process but provides a local asynchronous name cache; or configure
2636 the NSS package to avoid lookups for UIDs in the range `pkg-config
2637 systemd --variable=dynamicuidmin` … `pkg-config systemd
2638 --variable=dynamicuidmax`, so that it does not consider itself
2639 authoritative for the same UID range systemd allocates dynamic users
2642 * The systemd-resolve tool has been renamed to resolvectl (it also
2643 remains available under the old name, for compatibility), and its
2644 interface is now verb-based, similar in style to the other <xyz>ctl
2645 tools, such as systemctl or loginctl.
2647 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
2648 compatibility. It may be symlinked under the 'resolvconf' name, in
2649 which case it will take arguments and input compatible with the
2650 Debian and FreeBSD resolvconf tool.
2652 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
2653 where the system initially suspends, and after a timeout resumes and
2656 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
2657 set the client will only send a DUID as client identifier.
2659 * The nss-systemd glibc NSS module will now enumerate dynamic users and
2660 groups in effect. Previously, it could resolve UIDs/GIDs to user
2661 names/groups and vice versa, but did not support enumeration.
2663 * journald's Compress= configuration setting now optionally accepts a
2664 byte threshold value. All journal objects larger than this threshold
2665 will be compressed, smaller ones will not. Previously this threshold
2666 was not configurable and set to 512.
2668 * A new system.conf setting NoNewPrivileges= is now available which may
2669 be used to turn off acquisition of new privileges system-wide
2670 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
2671 for all its children). Note that turning this option on means setuid
2672 binaries and file system capabilities lose their special powers.
2673 While turning on this option is a big step towards a more secure
2674 system, doing so is likely to break numerous pre-existing UNIX tools,
2675 in particular su and sudo.
2677 * A new service systemd-time-sync-wait.service has been added. If
2678 enabled it will delay the time-sync.target unit at boot until time
2679 synchronization has been received from the network. This
2680 functionality is useful on systems lacking a local RTC or where it is
2681 acceptable that the boot process shall be delayed by external network
2684 * When hibernating, systemd will now inform the kernel of the image
2685 write offset, on kernels new enough to support this. This means swap
2686 files should work for hibernation now.
2688 * When loading unit files, systemd will now look for drop-in unit files
2689 extensions in additional places. Previously, for a unit file name
2690 "foo-bar-baz.service" it would look for dropin files in
2691 "foo-bar-baz.service.d/*.conf". Now, it will also look in
2692 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
2693 service name truncated after all inner dashes. This scheme allows
2694 writing drop-ins easily that apply to a whole set of unit files at
2695 once. It's particularly useful for mount and slice units (as their
2696 naming is prefix based), but is also useful for service and other
2697 units, for packages that install multiple unit files at once,
2698 following a strict naming regime of beginning the unit file name with
2699 the package's name. Two new specifiers are now supported in unit
2700 files to match this: %j and %J are replaced by the part of the unit
2701 name following the last dash.
2703 * Unit files and other configuration files that support specifier
2704 expansion now understand another three new specifiers: %T and %V will
2705 resolve to /tmp and /var/tmp respectively, or whatever temporary
2706 directory has been set for the calling user. %E will expand to either
2707 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
2709 * The ExecStart= lines of unit files are no longer required to
2710 reference absolute paths. If non-absolute paths are specified the
2711 specified binary name is searched within the service manager's
2712 built-in $PATH, which may be queried with 'systemd-path
2713 search-binaries-default'. It's generally recommended to continue to
2714 use absolute paths for all binaries specified in unit files.
2716 * Units gained a new load state "bad-setting", which is used when a
2717 unit file was loaded, but contained fatal errors which prevent it
2718 from being started (for example, a service unit has been defined
2719 lacking both ExecStart= and ExecStop= lines).
2721 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
2722 support alternative debuggers, for example lldb. The old name
2723 continues to be available however, for compatibility reasons. Use the
2724 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
2725 to pick an alternative debugger instead of the default gdb.
2727 * systemctl and the other tools will now output escape sequences that
2728 generate proper clickable hyperlinks in various terminal emulators
2729 where useful (for example, in the "systemctl status" output you can
2730 now click on the unit file name to quickly open it in the
2731 editor/viewer of your choice). Note that not all terminal emulators
2732 support this functionality yet, but many do. Unfortunately, the
2733 "less" pager doesn't support this yet, hence this functionality is
2734 currently automatically turned off when a pager is started (which
2735 happens quite often due to auto-paging). We hope to remove this
2736 limitation as soon as "less" learns these escape sequences. This new
2737 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
2738 environment variable. For details on these escape sequences see:
2739 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
2741 * networkd's .network files now support a new IPv6MTUBytes= option for
2742 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
2743 option in the [Route] section to configure the MTU to use for
2744 specific routes. It also gained support for configuration of the DHCP
2745 "UserClass" option through the new UserClass= setting. It gained
2746 three new options in the new [CAN] section for configuring CAN
2747 networks. The MULTICAST and ALLMULTI interface flags may now be
2748 controlled explicitly with the new Multicast= and AllMulticast=
2751 * networkd will now automatically make use of the kernel's route
2752 expiration feature, if it is available.
2754 * udevd's .link files now support setting the number of receive and
2755 transmit channels, using the RxChannels=, TxChannels=,
2756 OtherChannels=, CombinedChannels= settings.
2758 * Support for UDPSegmentationOffload= has been removed, given its
2759 limited support in hardware, and waning software support.
2761 * networkd's .netdev files now support creating "netdevsim" interfaces.
2763 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
2764 to query the unit belonging to a specific kernel control group.
2766 * systemd-analyze gained a new verb "cat-config", which may be used to
2767 dump the contents of any configuration file, with all its matching
2768 drop-in files added in, and honouring the usual search and masking
2769 logic applied to systemd configuration files. For example use
2770 "systemd-analyze cat-config systemd/system.conf" to get the complete
2771 system configuration file of systemd how it would be loaded by PID 1
2772 itself. Similar to this, various tools such as systemd-tmpfiles or
2773 systemd-sysusers, gained a new option "--cat-config", which does the
2774 corresponding operation for their own configuration settings. For
2775 example, "systemd-tmpfiles --cat-config" will now output the full
2776 list of tmpfiles.d/ lines in place.
2778 * timedatectl gained three new verbs: "show" shows bus properties of
2779 systemd-timedated, "timesync-status" shows the current NTP
2780 synchronization state of systemd-timesyncd, and "show-timesync"
2781 shows bus properties of systemd-timesyncd.
2783 * systemd-timesyncd gained a bus interface on which it exposes details
2786 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
2787 understood by systemd-timedated. It takes a colon-separated list of
2788 unit names of NTP client services. The list is used by
2789 "timedatectl set-ntp".
2791 * systemd-nspawn gained a new --rlimit= switch for setting initial
2792 resource limits for the container payload. There's a new switch
2793 --hostname= to explicitly override the container's hostname. A new
2794 --no-new-privileges= switch may be used to control the
2795 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
2796 --oom-score-adjust= switch controls the OOM scoring adjustment value
2797 for the payload. The new --cpu-affinity= switch controls the CPU
2798 affinity of the container payload. The new --resolv-conf= switch
2799 allows more detailed control of /etc/resolv.conf handling of the
2800 container. Similarly, the new --timezone= switch allows more detailed
2801 control of /etc/localtime handling of the container.
2803 * systemd-detect-virt gained a new --list switch, which will print a
2804 list of all currently known VM and container environments.
2806 * Support for "Portable Services" has been added, see
2807 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
2808 experimental, but this is expected to change soon. Reflecting this
2809 experimental state, the "portablectl" binary is not installed into
2810 /usr/bin yet. The binary has to be called with the full path
2811 /usr/lib/systemd/portablectl instead.
2813 * journalctl's and systemctl's -o switch now knows a new log output
2814 mode "with-unit". The output it generates is very similar to the
2815 regular "short" mode, but displays the unit name instead of the
2816 syslog tag for each log line. Also, the date is shown with timezone
2817 information. This mode is probably more useful than the classic
2818 "short" output mode for most purposes, except where pixel-perfect
2819 compatibility with classic /var/log/messages formatting is required.
2821 * A new --dump-bus-properties switch has been added to the systemd
2822 binary, which may be used to dump all supported D-Bus properties.
2823 (Options which are still supported, but are deprecated, are *not*
2826 * sd-bus gained a set of new calls:
2827 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
2828 enable/disable the "floating" state of a bus slot object,
2829 i.e. whether the slot object pins the bus it is allocated for into
2830 memory or if the bus slot object gets disconnected when the bus goes
2831 away. sd_bus_open_with_description(),
2832 sd_bus_open_user_with_description(),
2833 sd_bus_open_system_with_description() may be used to allocate bus
2834 objects and set their description string already during allocation.
2836 * sd-event gained support for watching inotify events from the event
2837 loop, in an efficient way, sharing inotify handles between multiple
2838 users. For this a new function sd_event_add_inotify() has been added.
2840 * sd-event and sd-bus gained support for calling special user-supplied
2841 destructor functions for userdata pointers associated with
2842 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
2843 functions sd_bus_slot_set_destroy_callback,
2844 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
2845 sd_bus_track_get_destroy_callback,
2846 sd_event_source_set_destroy_callback,
2847 sd_event_source_get_destroy_callback have been added.
2849 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
2851 * PID 1 will now automatically reschedule .timer units whenever the
2852 local timezone changes. (They previously got rescheduled
2853 automatically when the system clock changed.)
2855 * New documentation has been added to document cgroups delegation,
2856 portable services and the various code quality tools we have set up:
2858 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
2859 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
2860 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
2862 * The Boot Loader Specification has been added to the source tree.
2864 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
2866 While moving it into our source tree we have updated it and further
2867 changes are now accepted through the usual github PR workflow.
2869 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
2870 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
2871 earlier PAM modules. The data in these fields is used to initialize
2872 the session scope's resource properties. Thus external PAM modules
2873 may now configure per-session limits, for example sourced from
2874 external user databases.
2876 * socket units with Accept=yes will now maintain a "refused" counter in
2877 addition to the existing "accepted" counter, counting connections
2878 refused due to the enforced limits.
2880 * The "systemd-path search-binaries-default" command may now be use to
2881 query the default, built-in $PATH PID 1 will pass to the services it
2884 * A new unit file setting PrivateMounts= has been added. It's a boolean
2885 option. If enabled the unit's processes are invoked in their own file
2886 system namespace. Note that this behaviour is also implied if any
2887 other file system namespacing options (such as PrivateTmp=,
2888 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
2889 primarily useful for services that do not use any of the other file
2890 system namespacing options. One such service is systemd-udevd.service
2891 where this is now used by default.
2893 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
2894 when the system is booted in UEFI "secure mode".
2896 * A new unit "system-update-pre.target" is added, which defines an
2897 optional synchronization point for offline system updates, as
2898 implemented by the pre-existing "system-update.target" unit. It
2899 allows ordering services before the service that executes the actual
2900 update process in a generic way.
2902 * Systemd now emits warnings whenever .include syntax is used.
2904 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
2905 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
2906 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
2907 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
2908 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
2909 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
2910 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
2911 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
2912 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
2913 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
2914 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
2915 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
2916 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
2917 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
2918 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
2919 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
2920 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
2921 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
2922 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
2923 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
2924 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
2925 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
2926 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
2927 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
2928 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
2929 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
2930 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
2931 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
2932 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2934 — Berlin, 2018-06-22
2938 * The MemoryAccounting= unit property now defaults to on. After
2939 discussions with the upstream control group maintainers we learnt
2940 that the negative impact of cgroup memory accounting on current
2941 kernels is finally relatively minimal, so that it should be safe to
2942 enable this by default without affecting system performance. Besides
2943 memory accounting only task accounting is turned on by default, all
2944 other forms of resource accounting (CPU, IO, IP) remain off for now,
2945 because it's not clear yet that their impact is small enough to move
2946 from opt-in to opt-out. We recommend downstreams to leave memory
2947 accounting on by default if kernel 4.14 or higher is primarily
2948 used. On very resource constrained systems or when support for old
2949 kernels is a necessity, -Dmemory-accounting-default=false can be used
2950 to revert this change.
2952 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
2953 %udev_rules_update) and the journal catalog (%journal_catalog_update)
2954 from the upgrade scriptlets of individual packages now do nothing.
2955 Transfiletriggers have been added which will perform those updates
2956 once at the end of the transaction.
2958 Similar transfiletriggers have been added to execute any sysctl.d
2959 and binfmt.d rules. Thus, it should be unnecessary to provide any
2960 scriptlets to execute this configuration from package installation
2963 * systemd-sysusers gained a mode where the configuration to execute is
2964 specified on the command line, but this configuration is not executed
2965 directly, but instead it is merged with the configuration on disk,
2966 and the result is executed. This is useful for package installation
2967 scripts which want to create the user before installing any files on
2968 disk (in case some of those files are owned by that user), while
2969 still allowing local admin overrides.
2971 This functionality is exposed to rpm scriptlets through a new
2972 %sysusers_create_package macro. Old %sysusers_create and
2973 %sysusers_create_inline macros are deprecated.
2975 A transfiletrigger for sysusers.d configuration is now installed,
2976 which means that it should be unnecessary to call systemd-sysusers from
2977 package installation scripts, unless the package installs any files
2978 owned by those newly-created users, in which case
2979 %sysusers_create_package should be used.
2981 * Analogous change has been done for systemd-tmpfiles: it gained a mode
2982 where the command-line configuration is merged with the configuration
2983 on disk. This is exposed as the new %tmpfiles_create_package macro,
2984 and %tmpfiles_create is deprecated. A transfiletrigger is installed
2985 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
2986 from package installation scripts.
2988 * sysusers.d configuration for a user may now also specify the group
2989 number, in addition to the user number ("u username 123:456"), or
2990 without the user number ("u username -:456").
2992 * Configution items for systemd-sysusers can now be specified as
2993 positional arguments when the new --inline switch is used.
2995 * The login shell of users created through sysusers.d may now be
2996 specified (previously, it was always /bin/sh for root and
2997 /sbin/nologin for other users).
2999 * systemd-analyze gained a new --global switch to look at global user
3000 configuration. It also gained a unit-paths verb to list the unit load
3001 paths that are compiled into systemd (which can be used with
3002 --systemd, --user, or --global).
3004 * udevadm trigger gained a new --settle/-w option to wait for any
3005 triggered events to finish (but just those, and not any other events
3006 which are triggered meanwhile).
3008 * The action that systemd-logind takes when the lid is closed and the
3009 machine is connected to external power can now be configured using
3010 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
3011 was determined by HandleLidSwitch=, and, for backwards compatibility,
3012 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
3014 * journalctl will periodically call sd_journal_process() to make it
3015 resilient against inotify queue overruns when journal files are
3016 rotated very quickly.
3018 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
3019 sd_bus_get_n_queued_write — may be used to check the number of
3020 pending bus messages.
3022 * systemd gained a new
3023 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
3024 which can be used to migrate foreign processes to scope and service
3025 units. The primary user for this new API is systemd itself: the
3026 systemd --user instance uses this call of the systemd --system
3027 instance to migrate processes if it itself gets the request to
3028 migrate processes and the kernel refuses this due to access
3029 restrictions. Thanks to this "systemd-run --scope --user …" works
3030 again in pure cgroup v2 environments when invoked from the user
3033 * A new TemporaryFileSystem= setting can be used to mask out part of
3034 the real file system tree with tmpfs mounts. This may be combined
3035 with BindPaths= and BindReadOnlyPaths= to hide files or directories
3036 not relevant to the unit, while still allowing some paths lower in
3037 the tree to be accessed.
3039 ProtectHome=tmpfs may now be used to hide user home and runtime
3040 directories from units, in a way that is mostly equivalent to
3041 "TemporaryFileSystem=/home /run/user /root".
3043 * Non-service units are now started with KeyringMode=shared by default.
3044 This means that mount and swapon and other mount tools have access
3045 to keys in the main keyring.
3047 * /sys/fs/bpf is now mounted automatically.
3049 * QNX virtualization is now detected by systemd-detect-virt and may
3050 be used in ConditionVirtualization=.
3052 * IPAccounting= may now be enabled also for slice units.
3054 * A new -Dsplit-bin= build configuration switch may be used to specify
3055 whether bin and sbin directories are merged, or if they should be
3056 included separately in $PATH and various listings of executable
3057 directories. The build configuration scripts will try to autodetect
3058 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
3059 system, but distributions are encouraged to configure this
3062 * A new -Dok-color= build configuration switch may be used to change
3063 the colour of "OK" status messages.
3065 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
3066 PrivateNetwork=yes was buggy in previous versions of systemd. This
3067 means that after the upgrade and daemon-reexec, any such units must
3070 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
3071 will not exclude read-only files owned by root from cleanup.
3073 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
3074 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
3075 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
3076 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
3077 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
3078 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
3079 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
3080 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
3081 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
3082 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
3083 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
3084 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
3085 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
3086 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
3087 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
3088 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
3090 — Warsaw, 2018-03-05
3094 * Some keyboards come with a zoom see-saw or rocker which until now got
3095 mapped to the Linux "zoomin/out" keys in hwdb. However, these
3096 keycodes are not recognized by any major desktop. They now produce
3097 Up/Down key events so that they can be used for scrolling.
3099 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
3100 slightly: previously, if an argument was specified for lines of this
3101 type (i.e. the right-most column was set) this string was appended to
3102 existing files each time systemd-tmpfiles was run. This behaviour was
3103 different from what the documentation said, and not particularly
3104 useful, as repeated systemd-tmpfiles invocations would not be
3105 idempotent and grow such files without bounds. With this release
3106 behaviour has been altered to match what the documentation says:
3107 lines of this type only have an effect if the indicated files don't
3108 exist yet, and only then the argument string is written to the file.
3110 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
3111 systemd-tmpfiles behaviour: previously, read-only files owned by root
3112 were always excluded from the file "aging" algorithm (i.e. the
3113 automatic clean-up of directories like /tmp based on
3114 atime/mtime/ctime). We intend to drop this restriction, and age files
3115 by default even when owned by root and read-only. This behaviour was
3116 inherited from older tools, but there have been requests to remove
3117 it, and it's not obvious why this restriction was made in the first
3118 place. Please speak up now, if you are aware of software that reqires
3119 this behaviour, otherwise we'll remove the restriction in v238.
3121 * A new environment variable $SYSTEMD_OFFLINE is now understood by
3122 systemctl. It takes a boolean argument. If on, systemctl assumes it
3123 operates on an "offline" OS tree, and will not attempt to talk to the
3124 service manager. Previously, this mode was implicitly enabled if a
3125 chroot() environment was detected, and this new environment variable
3126 now provides explicit control.
3128 * .path and .socket units may now be created transiently, too.
3129 Previously only service, mount, automount and timer units were
3130 supported as transient units. The systemd-run tool has been updated
3131 to expose this new functionality, you may hence use it now to bind
3132 arbitrary commands to path or socket activation on-the-fly from the
3133 command line. Moreover, almost all properties are now exposed for the
3134 unit types that already supported transient operation.
3136 * The systemd-mount command gained support for a new --owner= parameter
3137 which takes a user name, which is then resolved and included in uid=
3138 and gid= mount options string of the file system to mount.
3140 * A new unit condition ConditionControlGroupController= has been added
3141 that checks whether a specific cgroup controller is available.
3143 * Unit files, udev's .link files, and systemd-networkd's .netdev and
3144 .network files all gained support for a new condition
3145 ConditionKernelVersion= for checking against specific kernel
3148 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
3149 support for configuring device flags in the Flags= setting. In the
3150 same files, the [Tunnel] section gained support for configuring
3151 AllowLocalRemote=. The [Route] section in .network files gained
3152 support for configuring InitialCongestionWindow=,
3153 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
3154 understands RapidCommit=.
3156 * systemd-networkd's DHCPv6 support gained support for Prefix
3159 * sd-bus gained support for a new "watch-bind" feature. When this
3160 feature is enabled, an sd_bus connection may be set up to connect to
3161 an AF_UNIX socket in the file system as soon as it is created. This
3162 functionality is useful for writing early-boot services that
3163 automatically connect to the system bus as soon as it is started,
3164 without ugly time-based polling. systemd-networkd and
3165 systemd-resolved have been updated to make use of this
3166 functionality. busctl exposes this functionality in a new
3167 --watch-bind= command line switch.
3169 * sd-bus will now optionally synthesize a local "Connected" signal as
3170 soon as a D-Bus connection is set up fully. This message mirrors the
3171 already existing "Disconnected" signal which is synthesized when the
3172 connection is terminated. This signal is generally useful but
3173 particularly handy in combination with the "watch-bind" feature
3174 described above. Synthesizing of this message has to be requested
3175 explicitly through the new API call sd_bus_set_connected_signal(). In
3176 addition a new call sd_bus_is_ready() has been added that checks
3177 whether a connection is fully set up (i.e. between the "Connected" and
3178 "Disconnected" signals).
3180 * sd-bus gained two new calls sd_bus_request_name_async() and
3181 sd_bus_release_name_async() for asynchronously registering bus
3182 names. Similar, there is now sd_bus_add_match_async() for installing
3183 a signal match asynchronously. All of systemd's own services have
3184 been updated to make use of these calls. Doing these operations
3185 asynchronously has two benefits: it reduces the risk of deadlocks in
3186 case of cyclic dependencies between bus services, and it speeds up
3187 service initialization since synchronization points for bus
3188 round-trips are removed.
3190 * sd-bus gained two new calls sd_bus_match_signal() and
3191 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
3192 and sd_bus_add_match_async() but instead of taking a D-Bus match
3193 string take match fields as normal function parameters.
3195 * sd-bus gained two new calls sd_bus_set_sender() and
3196 sd_bus_message_set_sender() for setting the sender name of outgoing
3197 messages (either for all outgoing messages or for just one specific
3198 one). These calls are only useful in direct connections as on
3199 brokered connections the broker fills in the sender anyway,
3200 overwriting whatever the client filled in.
3202 * sd-event gained a new pseudo-handle that may be specified on all API
3203 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
3204 used this refers to the default event loop object of the calling
3205 thread. Note however that this does not implicitly allocate one —
3206 which has to be done prior by using sd_event_default(). Similarly
3207 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
3208 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
3209 to the default bus of the specified type of the calling thread. Here
3210 too this does not implicitly allocate bus connection objects, this
3211 has to be done prior with sd_bus_default() and friends.
3213 * sd-event gained a new call pair
3214 sd_event_source_{get|set}_io_fd_own(). This may be used to request
3215 automatic closure of the file descriptor an IO event source watches
3216 when the event source is destroyed.
3218 * systemd-networkd gained support for natively configuring WireGuard
3221 * In previous versions systemd synthesized user records both for the
3222 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
3223 internally. In order to simplify distribution-wide renames of the
3224 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
3225 new transitional flag file has been added: if
3226 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
3227 user and group record within the systemd codebase is disabled.
3229 * systemd-notify gained a new --uid= option for selecting the source
3230 user/UID to use for notification messages sent to the service
3233 * journalctl gained a new --grep= option to list only entries in which
3234 the message matches a certain pattern. By default matching is case
3235 insensitive if the pattern is lowercase, and case sensitive
3236 otherwise. Option --case-sensitive=yes|no can be used to override
3237 this an specify case sensitivity or case insensitivity.
3239 * There's now a "systemd-analyze service-watchdogs" command for printing
3240 the current state of the service runtime watchdog, and optionally
3241 enabling or disabling the per-service watchdogs system-wide if given a
3242 boolean argument (i.e. the concept you configure in WatchdogSec=), for
3243 debugging purposes. There's also a kernel command line option
3244 systemd.service_watchdogs= for controlling the same.
3246 * Two new "log-level" and "log-target" options for systemd-analyze were
3247 added that merge the now deprecated get-log-level, set-log-level and
3248 get-log-target, set-log-target pairs. The deprecated options are still
3249 understood for backwards compatibility. The two new options print the
3250 current value when no arguments are given, and set them when a
3251 level/target is given as an argument.
3253 * sysusers.d's "u" lines now optionally accept both a UID and a GID
3254 specification, separated by a ":" character, in order to create users
3255 where UID and GID do not match.
3257 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
3258 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
3259 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
3260 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
3261 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
3262 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
3263 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
3264 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
3265 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
3266 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
3267 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
3268 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
3269 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
3270 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
3271 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
3272 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
3273 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
3274 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
3275 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
3276 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
3283 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
3284 in v235 has been extended to also set the dummy.ko module option
3285 numdummies=0, preventing the kernel from automatically creating
3286 dummy0. All dummy interfaces must now be explicitly created.
3288 * Unknown '%' specifiers in configuration files are now rejected. This
3289 applies to units and tmpfiles.d configuration. Any percent characters
3290 that are followed by a letter or digit that are not supposed to be
3291 interpreted as the beginning of a specifier should be escaped by
3292 doubling ("%%"). (So "size=5%" is still accepted, as well as
3293 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
3294 valid specifiers today.)
3296 * systemd-resolved now maintains a new dynamic
3297 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
3298 recommended to make /etc/resolv.conf a symlink to it. This file
3299 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
3300 includes dynamically acquired search domains, achieving more correct
3301 DNS resolution by software that bypasses local DNS APIs such as NSS.
3303 * The "uaccess" udev tag has been dropped from /dev/kvm and
3304 /dev/dri/renderD*. These devices now have the 0666 permissions by
3305 default (but this may be changed at build-time). /dev/dri/renderD*
3306 will now be owned by the "render" group along with /dev/kfd.
3308 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
3309 systemd-journal-gatewayd.service and
3310 systemd-journal-upload.service. This means "nss-systemd" must be
3311 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
3312 services are resolved properly.
3314 * In /etc/fstab two new mount options are now understood:
3315 x-systemd.makefs and x-systemd.growfs. The former has the effect that
3316 the configured file system is formatted before it is mounted, the
3317 latter that the file system is resized to the full block device size
3318 after it is mounted (i.e. if the file system is smaller than the
3319 partition it resides on, it's grown). This is similar to the fsck
3320 logic in /etc/fstab, and pulls in systemd-makefs@.service and
3321 systemd-growfs@.service as necessary, similar to
3322 systemd-fsck@.service. Resizing is currently only supported on ext4
3325 * In systemd-networkd, the IPv6 RA logic now optionally may announce
3326 DNS server and domain information.
3328 * Support for the LUKS2 on-disk format for encrypted partitions has
3329 been added. This requires libcryptsetup2 during compilation and
3332 * The systemd --user instance will now signal "readiness" when its
3333 basic.target unit has been reached, instead of when the run queue ran
3334 empty for the first time.
3336 * Tmpfiles.d with user configuration are now also supported.
3337 systemd-tmpfiles gained a new --user switch, and snippets placed in
3338 ~/.config/user-tmpfiles.d/ and corresponding directories will be
3339 executed by systemd-tmpfiles --user running in the new
3340 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
3341 running in the user session.
3343 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
3344 %S resolves to the top-level state directory (/var/lib for the system
3345 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
3346 top-level cache directory (/var/cache for the system instance,
3347 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
3348 logs directory (/var/log for the system instance,
3349 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
3350 existing %t specifier, that resolves to the top-level runtime
3351 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
3354 * journalctl learnt a new parameter --output-fields= for limiting the
3355 set of journal fields to output in verbose and JSON output modes.
3357 * systemd-timesyncd's configuration file gained a new option
3358 RootDistanceMaxSec= for setting the maximum root distance of servers
3359 it'll use, as well as the new options PollIntervalMinSec= and
3360 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
3362 * bootctl gained a new command "list" for listing all available boot
3363 menu items on systems that follow the boot loader specification.
3365 * systemctl gained a new --dry-run switch that shows what would be done
3366 instead of doing it, and is currently supported by the shutdown and
3369 * ConditionSecurity= can now detect the TOMOYO security module.
3371 * Unit file [Install] sections are now also respected in unit drop-in
3372 files. This is intended to be used by drop-ins under /usr/lib/.
3374 * systemd-firstboot may now also set the initial keyboard mapping.
3376 * Udev "changed" events for devices which are exposed as systemd
3377 .device units are now propagated to units specified in
3378 ReloadPropagatedFrom= as reload requests.
3380 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
3381 unit template name (i.e. a name in the form of 'foobar@.service',
3382 without the instance component between the '@' and - the '.'), then
3383 the escaped sysfs path of the device is automatically used as the
3386 * SystemCallFilter= in unit files has been extended so that an "errno"
3387 can be specified individually for each system call. Example:
3388 SystemCallFilter=~uname:EILSEQ.
3390 * The cgroup delegation logic has been substantially updated. Delegate=
3391 now optionally takes a list of controllers (instead of a boolean, as
3392 before), which lists the controllers to delegate at least.
3394 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
3396 * A new LogLevelMax= setting configures the maximum log level any
3397 process of the service may log at (i.e. anything with a lesser
3398 priority than what is specified is automatically dropped). A new
3399 LogExtraFields= setting allows configuration of additional journal
3400 fields to attach to all log records generated by any of the unit's
3403 * New StandardInputData= and StandardInputText= settings along with the
3404 new option StandardInput=data may be used to configure textual or
3405 binary data that shall be passed to the executed service process via
3406 standard input, encoded in-line in the unit file.
3408 * StandardInput=, StandardOutput= and StandardError= may now be used to
3409 connect stdin/stdout/stderr of executed processes directly with a
3410 file or AF_UNIX socket in the file system, using the new "file:" option.
3412 * A new unit file option CollectMode= has been added, that allows
3413 tweaking the garbage collection logic for units. It may be used to
3414 tell systemd to garbage collect units that have failed automatically
3415 (normally it only GCs units that exited successfully). systemd-run
3416 and systemd-mount expose this new functionality with a new -G option.
3418 * "machinectl bind" may now be used to bind mount non-directories
3419 (i.e. regularfiles, devices, fifos, sockets).
3421 * systemd-analyze gained a new verb "calendar" for validating and
3422 testing calendar time specifications to use for OnCalendar= in timer
3423 units. Besides validating the expression it will calculate the next
3424 time the specified expression would elapse.
3426 * In addition to the pre-existing FailureAction= unit file setting
3427 there's now SuccessAction=, for configuring a shutdown action to
3428 execute when a unit completes successfully. This is useful in
3429 particular inside containers that shall terminate after some workload
3430 has been completed. Also, both options are now supported for all unit
3431 types, not just services.
3433 * networkds's IP rule support gained two new options
3434 IncomingInterface= and OutgoingInterface= for configuring the incoming
3435 and outgoing interfaces of configured rules. systemd-networkd also
3436 gained support for "vxcan" network devices.
3438 * networkd gained a new setting RequiredForOnline=, taking a
3439 boolean. If set, systemd-wait-online will take it into consideration
3440 when determining that the system is up, otherwise it will ignore the
3441 interface for this purpose.
3443 * The sd_notify() protocol gained support for a new operation: with
3444 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
3445 store again, ahead of POLLHUP or POLLERR when they are removed
3448 * A new document doc/UIDS-GIDS.md has been added to the source tree,
3449 that documents the UID/GID range and assignment assumptions and
3450 requirements of systemd.
3452 * The watchdog device PID 1 will ping may now be configured through the
3453 WatchdogDevice= configuration file setting, or by setting the
3454 systemd.watchdog_service= kernel commandline option.
3456 * systemd-resolved's gained support for registering DNS-SD services on
3457 the local network using MulticastDNS. Services may either be
3458 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
3459 the same dir below /run, /usr/lib), or through its D-Bus API.
3461 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
3462 extend the effective start, runtime, and stop time. The service must
3463 continue to send EXTEND_TIMEOUT_USEC within the period specified to
3464 prevent the service manager from making the service as timedout.
3466 * systemd-resolved's DNSSEC support gained support for RFC 8080
3467 (Ed25519 keys and signatures).
3469 * The systemd-resolve command line tool gained a new set of options
3470 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
3471 --set-nta= and --revert to configure per-interface DNS configuration
3472 dynamically during runtime. It's useful for pushing DNS information
3473 into systemd-resolved from DNS hook scripts that various interface
3474 managing software supports (such as pppd).
3476 * systemd-nspawn gained a new --network-namespace-path= command line
3477 option, which may be used to make a container join an existing
3478 network namespace, by specifying a path to a "netns" file.
3480 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
3481 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
3482 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
3483 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
3484 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
3485 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
3486 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
3487 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
3488 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
3489 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
3490 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
3491 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
3492 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
3493 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
3494 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
3495 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
3496 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
3497 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
3498 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
3499 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
3500 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
3501 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
3502 Jędrzejewski-Szmek, Zeal Jagannatha
3504 — Berlin, 2017-12-14
3508 * INCOMPATIBILITY: systemd-logind.service and other long-running
3509 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
3510 communication with the outside. This generally improves security of
3511 the system, and is in almost all cases a safe and good choice, as
3512 these services do not and should not provide any network-facing
3513 functionality. However, systemd-logind uses the glibc NSS API to
3514 query the user database. This creates problems on systems where NSS
3515 is set up to directly consult network services for user database
3516 lookups. In particular, this creates incompatibilities with the
3517 "nss-nis" module, which attempts to directly contact the NIS/YP
3518 network servers it is configured for, and will now consistently
3519 fail. In such cases, it is possible to turn off IP sandboxing for
3520 systemd-logind.service (set IPAddressDeny= in its [Service] section
3521 to the empty string, via a .d/ unit file drop-in). Downstream
3522 distributions might want to update their nss-nis packaging to include
3523 such a drop-in snippet, accordingly, to hide this incompatibility
3524 from the user. Another option is to make use of glibc's nscd service
3525 to proxy such network requests through a privilege-separated, minimal
3526 local caching daemon, or to switch to more modern technologies such
3527 sssd, whose NSS hook-ups generally do not involve direct network
3528 access. In general, we think it's definitely time to question the
3529 implementation choices of nss-nis, i.e. whether it's a good idea
3530 today to embed a network-facing loadable module into all local
3531 processes that need to query the user database, including the most
3532 trivial and benign ones, such as "ls". For more details about
3533 IPAddressDeny= see below.
3535 * A new modprobe.d drop-in is now shipped by default that sets the
3536 bonding module option max_bonds=0. This overrides the kernel default,
3537 to avoid conflicts and ambiguity as to whether or not bond0 should be
3538 managed by systemd-networkd or not. This resolves multiple issues
3539 with bond0 properties not being applied, when bond0 is configured
3540 with systemd-networkd. Distributors may choose to not package this,
3541 however in that case users will be prevented from correctly managing
3542 bond0 interface using systemd-networkd.
3544 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
3545 which print the logging level and target of the system manager. They
3546 complement the existing "set-log-level" and "set-log-target" verbs
3547 used to change those values.
3549 * journald.conf gained a new boolean setting ReadKMsg= which defaults
3550 to on. If turned off kernel log messages will not be read by
3551 systemd-journald or included in the logs. It also gained a new
3552 setting LineMax= for configuring the maximum line length in
3553 STDOUT/STDERR log streams. The new default for this value is 48K, up
3554 from the previous hardcoded 2048.
3556 * A new unit setting RuntimeDirectoryPreserve= has been added, which
3557 allows more detailed control of what to do with a runtime directory
3558 configured with RuntimeDirectory= (i.e. a directory below /run or
3559 $XDG_RUNTIME_DIR) after a unit is stopped.
3561 * The RuntimeDirectory= setting for units gained support for creating
3562 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
3563 one top-level directory.
3565 * Units gained new options StateDirectory=, CacheDirectory=,
3566 LogsDirectory= and ConfigurationDirectory= which are closely related
3567 to RuntimeDirectory= but manage per-service directories below
3568 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
3569 possible to write unit files which when activated automatically gain
3570 properly owned service specific directories in these locations, thus
3571 making unit files self-contained and increasing compatibility with
3572 stateless systems and factory reset where /etc or /var are
3573 unpopulated at boot. Matching these new settings there's also
3574 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
3575 ConfigurationDirectoryMode= for configuring the access mode of these
3576 directories. These settings are particularly useful in combination
3577 with DynamicUser=yes as they provide secure, properly-owned,
3578 writable, and stateful locations for storage, excluded from the
3579 sandbox that such services live in otherwise.
3581 * Automake support has been removed from this release. systemd is now
3584 * systemd-journald will now aggressively cache client metadata during
3585 runtime, speeding up log write performance under pressure. This comes
3586 at a small price though: as much of the metadata is read
3587 asynchronously from /proc/ (and isn't implicitly attached to log
3588 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
3589 metadata stored alongside a log entry might be slightly
3590 out-of-date. Previously it could only be slightly newer than the log
3591 message. The time window is small however, and given that the kernel
3592 is unlikely to be improved anytime soon in this regard, this appears
3595 * nss-myhostname/systemd-resolved will now by default synthesize an
3596 A/AAAA resource record for the "_gateway" hostname, pointing to the
3597 current default IP gateway. Previously it did that for the "gateway"
3598 name, hampering adoption, as some distributions wanted to leave that
3599 hostname open for local use. The old behaviour may still be
3600 requested at build time.
3602 * systemd-networkd's [Address] section in .network files gained a new
3603 Scope= setting for configuring the IP address scope. The [Network]
3604 section gained a new boolean setting ConfigureWithoutCarrier= that
3605 tells systemd-networkd to ignore link sensing when configuring the
3606 device. The [DHCP] section gained a new Anonymize= boolean option for
3607 turning on a number of options suggested in RFC 7844. A new
3608 [RoutingPolicyRule] section has been added for configuring the IP
3609 routing policy. The [Route] section has gained support for a new
3610 Type= setting which permits configuring
3611 blackhole/unreachable/prohibit routes.
3613 * The [VRF] section in .netdev files gained a new Table= setting for
3614 configuring the routing table to use. The [Tunnel] section gained a
3615 new Independent= boolean field for configuring tunnels independent of
3616 an underlying network interface. The [Bridge] section gained a new
3617 GroupForwardMask= option for configuration of propagation of link
3618 local frames between bridge ports.
3620 * The WakeOnLan= setting in .link files gained support for a number of
3621 new modes. A new TCP6SegmentationOffload= setting has been added for
3622 configuring TCP/IPv6 hardware segmentation offload.
3624 * The IPv6 RA sender implementation may now optionally send out RDNSS
3625 and RDNSSL records to supply DNS configuration to peers.
3627 * systemd-nspawn gained support for a new --system-call-filter= command
3628 line option for adding and removing entries in the default system
3629 call filter it applies. Moreover systemd-nspawn has been changed to
3630 implement a system call allow list instead of a deny list.
3632 * systemd-run gained support for a new --pipe command line option. If
3633 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
3634 are directly passed on to the activated transient service
3635 executable. This allows invoking arbitrary processes as systemd
3636 services (for example to take benefit of dependency management,
3637 accounting management, resource management or log management that is
3638 done automatically for services) — while still allowing them to be
3639 integrated in a classic UNIX shell pipeline.
3641 * When a service sends RELOAD=1 via sd_notify() and reload propagation
3642 using ReloadPropagationTo= is configured, a reload is now propagated
3643 to configured units. (Previously this was only done on explicitly
3644 requested reloads, using "systemctl reload" or an equivalent
3647 * For each service unit a restart counter is now kept: it is increased
3648 each time the service is restarted due to Restart=, and may be
3649 queried using "systemctl show -p NRestarts …".
3651 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
3652 @signal and @timer have been added, for usage with SystemCallFilter=
3653 in unit files and the new --system-call-filter= command line option
3654 of systemd-nspawn (see above).
3656 * ExecStart= lines in unit files gained two new modifiers: when a
3657 command line is prefixed with "!" the command will be executed as
3658 configured, except for the credentials applied by
3659 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
3660 "+", but does still apply namespacing options unlike "+". There's
3661 also "!!" now, which is mostly identical, but becomes a NOP on
3662 systems that support ambient capabilities. This is useful to write
3663 unit files that work with ambient capabilities where possible but
3664 automatically fall back to traditional privilege dropping mechanisms
3665 on systems where this is not supported.
3667 * ListenNetlink= settings in socket units now support RDMA netlink
3670 * A new unit file setting LockPersonality= has been added which permits
3671 locking down the chosen execution domain ("personality") of a service
3674 * A new special target "getty-pre.target" has been added, which is
3675 ordered before all text logins, and may be used to order services
3676 before textual logins acquire access to the console.
3678 * systemd will now attempt to load the virtio-rng.ko kernel module very
3679 early on if a VM environment supporting this is detected. This should
3680 improve entropy during early boot in virtualized environments.
3682 * A _netdev option is now supported in /etc/crypttab that operates in a
3683 similar way as the same option in /etc/fstab: it permits configuring
3684 encrypted devices that need to be ordered after the network is up.
3685 Following this logic, two new special targets
3686 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
3687 added that are to cryptsetup.target what remote-fs.target and
3688 remote-fs-pre.target are to local-fs.target.
3690 * Service units gained a new UnsetEnvironment= setting which permits
3691 unsetting specific environment variables for services that are
3692 normally passed to it (for example in order to mask out locale
3693 settings for specific services that can't deal with it).
3695 * Units acquired a new boolean option IPAccounting=. When turned on, IP
3696 traffic accounting (packet count as well as byte count) is done for
3697 the service, and shown as part of "systemctl status" or "systemd-run
3700 * Service units acquired two new options IPAddressAllow= and
3701 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
3702 for configuring a simple IP access control list for all sockets of
3703 the unit. These options are available also on .slice and .socket
3704 units, permitting flexible access list configuration for individual
3705 services as well as groups of services (as defined by a slice unit),
3706 including system-wide. Note that IP ACLs configured this way are
3707 enforced on every single IPv4 and IPv6 socket created by any process
3708 of the service unit, and apply to ingress as well as egress traffic.
3710 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
3711 structured log message is generated each time the unit is stopped,
3712 containing information about the consumed resources of this
3715 * A new setting KeyringMode= has been added to unit files, which may be
3716 used to control how the kernel keyring is set up for executed
3719 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
3720 "systemctl kexec" and "systemctl exit" are now always asynchronous in
3721 behaviour (that is: these commands return immediately after the
3722 operation was enqueued instead of waiting for the operation to
3723 complete). Previously, "systemctl poweroff" and "systemctl reboot"
3724 were asynchronous on systems using systemd-logind (i.e. almost
3725 always, and like they were on sysvinit), and the other three commands
3726 were unconditionally synchronous. With this release this is cleaned
3727 up, and callers will see the same asynchronous behaviour on all
3728 systems for all five operations.
3730 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
3733 * .timer units now accept calendar specifications in other timezones
3734 than UTC or the local timezone.
3736 * The tmpfiles snippet var.conf has been changed to create
3737 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
3738 the "utmp" group already, and it appears to be generally understood
3739 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
3740 databases. Previously this was implemented correctly for all these
3741 databases excepts btmp, which has been opened up like this now
3742 too. Note that while the other databases are world-readable
3743 (i.e. 0644), btmp is not and remains more restrictive.
3745 * The systemd-resolve tool gained a new --reset-server-features
3746 switch. When invoked like this systemd-resolved will forget
3747 everything it learnt about the features supported by the configured
3748 upstream DNS servers, and restarts the feature probing logic on the
3749 next resolver look-up for them at the highest feature level
3752 * The status dump systemd-resolved sends to the logs upon receiving
3753 SIGUSR1 now also includes information about all DNS servers it is
3754 configured to use, and the features levels it probed for them.
3756 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
3757 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
3758 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
3759 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
3760 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
3761 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
3762 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
3763 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
3764 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
3765 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
3766 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
3767 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
3768 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
3769 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
3770 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
3771 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
3772 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
3773 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
3774 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
3775 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
3777 — Berlin, 2017-10-06
3781 * Meson is now supported as build system in addition to Automake. It is
3782 our plan to remove Automake in one of our next releases, so that
3783 Meson becomes our exclusive build system. Hence, please start using
3784 the Meson build system in your downstream packaging. There's plenty
3785 of documentation around how to use Meson, the extremely brief
3788 ./autogen.sh && ./configure && make && sudo make install
3792 meson build && ninja -C build && sudo ninja -C build install
3794 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
3795 which permits configuring a timeout on the time a job is
3796 running. This is particularly useful for setting timeouts on jobs for
3799 * Unit files gained two new options ConditionUser= and ConditionGroup=
3800 for conditionalizing units based on the identity of the user/group
3801 running a systemd user instance.
3803 * systemd-networkd now understands a new FlowLabel= setting in the
3804 [VXLAN] section of .network files, as well as a Priority= in
3805 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
3806 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
3807 gained support for configuration of GENEVE links, and IPv6 address
3808 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
3810 * .link files now understand a new Port= setting.
3812 * systemd-networkd's DHCP support gained support for DHCP option 119
3813 (domain search list).
3815 * systemd-networkd gained support for serving IPv6 address ranges using
3816 the Router Advertisement protocol. The new .network configuration
3817 section [IPv6Prefix] may be used to configure the ranges to
3818 serve. This is implemented based on a new, minimal, native server
3819 implementation of RA.
3821 * journalctl's --output= switch gained support for a new parameter
3822 "short-iso-precise" for a mode where timestamps are shown as precise
3825 * systemd-udevd's "net_id" builtin may now generate stable network
3826 interface names from IBM PowerVM VIO devices as well as ACPI platform
3829 * MulticastDNS support in systemd-resolved may now be explicitly
3830 enabled/disabled using the new MulticastDNS= configuration file
3833 * systemd-resolved may now optionally use libidn2 instead of the libidn
3834 for processing internationalized domain names. Support for libidn2
3835 should be considered experimental and should not be enabled by
3838 * "machinectl pull-tar" and related call may now do verification of
3839 downloaded images using SUSE-style .sha256 checksum files in addition
3840 to the already existing support for validating using Ubuntu-style
3843 * sd-bus gained support for a new sd_bus_message_appendv() call which
3844 is va_list equivalent of sd_bus_message_append().
3846 * sd-boot gained support for validating images using SHIM/MOK.
3848 * The SMACK code learnt support for "onlycap".
3850 * systemd-mount --umount is now much smarter in figuring out how to
3851 properly unmount a device given its mount or device path.
3853 * The code to call libnss_dns as a fallback from libnss_resolve when
3854 the communication with systemd-resolved fails was removed. This
3855 fallback was redundant and interfered with the [!UNAVAIL=return]
3856 suffix. See nss-resolve(8) for the recommended configuration.
3858 * systemd-logind may now be restarted without losing state. It stores
3859 the file descriptors for devices it manages in the system manager
3860 using the FDSTORE= mechanism. Please note that further changes in
3861 other components may be required to make use of this (for example
3862 Xorg has code to listen for stops of systemd-logind and terminate
3863 itself when logind is stopped or restarted, in order to avoid using
3864 stale file descriptors for graphical devices, which is now
3865 counterproductive and must be reverted in order for restarts of
3866 systemd-logind to be safe. See
3867 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
3869 * All kernel-install plugins are called with the environment variable
3870 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
3871 /etc/machine-id. If the machine ID could not be determined,
3872 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
3873 anything in the entry directory (passed as the second argument) if
3874 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
3875 temporary directory is passed as the entry directory and removed
3876 after all the plugins exit.
3878 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
3879 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
3880 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
3881 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
3882 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
3883 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
3884 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
3885 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
3886 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
3887 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
3888 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
3889 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
3890 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
3891 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
3892 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
3893 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
3894 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
3895 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
3896 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
3897 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
3898 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
3899 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
3900 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
3901 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
3902 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
3903 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
3904 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
3905 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
3906 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
3909 — Berlin, 2017-07-12
3913 * The "hybrid" control group mode has been modified to improve
3914 compatibility with "legacy" cgroups-v1 setups. Specifically, the
3915 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
3916 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
3917 cgroups-v1 hierarchy), the only externally visible change being that
3918 the cgroups-v2 hierarchy is also mounted, to
3919 /sys/fs/cgroup/unified. This should provide a large degree of
3920 compatibility with "legacy" cgroups-v1, while taking benefit of the
3921 better management capabilities of cgroups-v2.
3923 * The default control group setup mode may be selected both a boot-time
3924 via a set of kernel command line parameters (specifically:
3925 systemd.unified_cgroup_hierarchy= and
3926 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
3927 default selected on the configure command line
3928 (--with-default-hierarchy=). The upstream default is "hybrid"
3929 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
3930 this will change in a future systemd version to be "unified" (pure
3931 cgroups-v2 mode). The third option for the compile time option is
3932 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
3933 distributions to default to "hybrid" mode for release distributions,
3934 starting with v233. We recommend "unified" for development
3935 distributions (specifically: distributions such as Fedora's rawhide)
3936 as that's where things are headed in the long run. Use "legacy" for
3937 greatest stability and compatibility only.
3939 * Note one current limitation of "unified" and "hybrid" control group
3940 setup modes: the kernel currently does not permit the systemd --user
3941 instance (i.e. unprivileged code) to migrate processes between two
3942 disconnected cgroup subtrees, even if both are managed and owned by
3943 the user. This effectively means "systemd-run --user --scope" doesn't
3944 work when invoked from outside of any "systemd --user" service or
3945 scope. Specifically, it is not supported from session scopes. We are
3946 working on fixing this in a future systemd version. (See #3388 for
3947 further details about this.)
3949 * DBus policy files are now installed into /usr rather than /etc. Make
3950 sure your system has dbus >= 1.9.18 running before upgrading to this
3951 version, or override the install path with --with-dbuspolicydir= .
3953 * All python scripts shipped with systemd (specifically: the various
3954 tests written in Python) now require Python 3.
3956 * systemd unit tests can now run standalone (without the source or
3957 build directories), and can be installed into /usr/lib/systemd/tests/
3958 with 'make install-tests'.
3960 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
3961 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
3964 * Support for the %c, %r, %R specifiers in unit files has been
3965 removed. Specifiers are not supposed to be dependent on configuration
3966 in the unit file itself (so that they resolve the same regardless
3967 where used in the unit files), but these specifiers were influenced
3968 by the Slice= option.
3970 * The shell invoked by debug-shell.service now defaults to /bin/sh in
3971 all cases. If distributions want to use a different shell for this
3972 purpose (for example Fedora's /sbin/sushell) they need to specify
3973 this explicitly at configure time using --with-debug-shell=.
3975 * The confirmation spawn prompt has been reworked to offer the
3978 (c)ontinue, proceed without asking anymore
3979 (D)ump, show the state of the unit
3980 (f)ail, don't execute the command and pretend it failed
3982 (i)nfo, show a short summary of the unit
3983 (j)obs, show jobs that are in progress
3984 (s)kip, don't execute the command and pretend it succeeded
3985 (y)es, execute the command
3987 The 'n' choice for the confirmation spawn prompt has been removed,
3988 because its meaning was confusing.
3990 The prompt may now also be redirected to an alternative console by
3991 specifying the console as parameter to systemd.confirm_spawn=.
3993 * Services of Type=notify require a READY=1 notification to be sent
3994 during startup. If no such message is sent, the service now fails,
3995 even if the main process exited with a successful exit code.
3997 * Services that fail to start up correctly now always have their
3998 ExecStopPost= commands executed. Previously, they'd enter "failed"
3999 state directly, without executing these commands.
4001 * The option MulticastDNS= of network configuration files has acquired
4002 an actual implementation. With MulticastDNS=yes a host can resolve
4003 names of remote hosts and reply to mDNS A and AAAA requests.
4005 * When units are about to be started an additional check is now done to
4006 ensure that all dependencies of type BindsTo= (when used in
4007 combination with After=) have been started.
4009 * systemd-analyze gained a new verb "syscall-filter" which shows which
4010 system call groups are defined for the SystemCallFilter= unit file
4011 setting, and which system calls they contain.
4013 * A new system call filter group "@filesystem" has been added,
4014 consisting of various file system related system calls. Group
4015 "@reboot" has been added, covering reboot, kexec and shutdown related
4016 calls. Finally, group "@swap" has been added covering swap
4017 configuration related calls.
4019 * A new unit file option RestrictNamespaces= has been added that may be
4020 used to restrict access to the various process namespace types the
4021 Linux kernel provides. Specifically, it may be used to take away the
4022 right for a service unit to create additional file system, network,
4023 user, and other namespaces. This sandboxing option is particularly
4024 relevant due to the high amount of recently discovered namespacing
4025 related vulnerabilities in the kernel.
4027 * systemd-udev's .link files gained support for a new AutoNegotiation=
4028 setting for configuring Ethernet auto-negotiation.
4030 * systemd-networkd's .network files gained support for a new
4031 ListenPort= setting in the [DHCP] section to explicitly configure the
4032 UDP client port the DHCP client shall listen on.
4034 * .network files gained a new Unmanaged= boolean setting for explicitly
4035 excluding one or more interfaces from management by systemd-networkd.
4037 * The systemd-networkd ProxyARP= option has been renamed to
4038 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
4039 renamed to ReduceARPProxy=. The old names continue to be available
4042 * systemd-networkd gained support for configuring IPv6 Proxy NDP
4043 addresses via the new IPv6ProxyNDPAddress= .network file setting.
4045 * systemd-networkd's bonding device support gained support for two new
4046 configuration options ActiveSlave= and PrimarySlave=.
4048 * The various options in the [Match] section of .network files gained
4049 support for negative matching.
4051 * New systemd-specific mount options are now understood in /etc/fstab:
4053 x-systemd.mount-timeout= may be used to configure the maximum
4054 permitted runtime of the mount command.
4056 x-systemd.device-bound may be set to bind a mount point to its
4057 backing device unit, in order to automatically remove a mount point
4058 if its backing device is unplugged. This option may also be
4059 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
4060 on the block device, which is now automatically set for all CDROM
4061 drives, so that mounted CDs are automatically unmounted when they are
4062 removed from the drive.
4064 x-systemd.after= and x-systemd.before= may be used to explicitly
4065 order a mount after or before another unit or mount point.
4067 * Enqueued start jobs for device units are now automatically garbage
4068 collected if there are no jobs waiting for them anymore.
4070 * systemctl list-jobs gained two new switches: with --after, for every
4071 queued job the jobs it's waiting for are shown; with --before the
4072 jobs which it's blocking are shown.
4074 * systemd-nspawn gained support for ephemeral boots from disk images
4075 (or in other words: --ephemeral and --image= may now be
4076 combined). Moreover, ephemeral boots are now supported for normal
4077 directories, even if the backing file system is not btrfs. Of course,
4078 if the file system does not support file system snapshots or
4079 reflinks, the initial copy operation will be relatively expensive, but
4080 this should still be suitable for many use cases.
4082 * Calendar time specifications in .timer units now support
4083 specifications relative to the end of a month by using "~" instead of
4084 "-" as separator between month and day. For example, "*-02~03" means
4085 "the third last day in February". In addition a new syntax for
4086 repeated events has been added using the "/" character. For example,
4087 "9..17/2:00" means "every two hours from 9am to 5pm".
4089 * systemd-socket-proxyd gained a new parameter --connections-max= for
4090 configuring the maximum number of concurrent connections.
4092 * sd-id128 gained a new API for generating unique IDs for the host in a
4093 way that does not leak the machine ID. Specifically,
4094 sd_id128_get_machine_app_specific() derives an ID based on the
4095 machine ID a in well-defined, non-reversible, stable way. This is
4096 useful whenever an identifier for the host is needed but where the
4097 identifier shall not be useful to identify the system beyond the
4098 scope of the application itself. (Internally this uses HMAC-SHA256 as
4099 keyed hash function using the machine ID as input.)
4101 * NotifyAccess= gained a new supported value "exec". When set
4102 notifications are accepted from all processes systemd itself invoked,
4103 including all control processes.
4105 * .nspawn files gained support for defining overlay mounts using the
4106 Overlay= and OverlayReadOnly= options. Previously this functionality
4107 was only available on the systemd-nspawn command line.
4109 * systemd-nspawn's --bind= and --overlay= options gained support for
4110 bind/overlay mounts whose source lies within the container tree by
4111 prefixing the source path with "+".
4113 * systemd-nspawn's --bind= and --overlay= options gained support for
4114 automatically allocating a temporary source directory in /var/tmp
4115 that is removed when the container dies. Specifically, if the source
4116 directory is specified as empty string this mechanism is selected. An
4117 example usage is --overlay=+/var::/var, which creates an overlay
4118 mount based on the original /var contained in the image, overlaid
4119 with a temporary directory in the host's /var/tmp. This way changes
4120 to /var are automatically flushed when the container shuts down.
4122 * systemd-nspawn --image= option does now permit raw file system block
4123 devices (in addition to images containing partition tables, as
4126 * The disk image dissection logic in systemd-nspawn gained support for
4127 automatically setting up LUKS encrypted as well as Verity protected
4128 partitions. When a container is booted from an encrypted image the
4129 passphrase is queried at start-up time. When a container with Verity
4130 data is started, the root hash is search in a ".roothash" file
4131 accompanying the disk image (alternatively, pass the root hash via
4132 the new --root-hash= command line option).
4134 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
4135 be used to dissect disk images the same way as systemd-nspawn does
4136 it, following the Bootable Partition Specification. It may even be
4137 used to mount disk images with complex partition setups (including
4138 LUKS and Verity partitions) to a local host directory, in order to
4139 inspect them. This tool is not considered public API (yet), and is
4140 thus not installed into /usr/bin. Please do not rely on its
4141 existence, since it might go away or be changed in later systemd
4144 * A new generator "systemd-verity-generator" has been added, similar in
4145 style to "systemd-cryptsetup-generator", permitting automatic setup of
4146 Verity root partitions when systemd boots up. In order to make use of
4147 this your partition setup should follow the Discoverable Partitions
4148 Specification, and the GPT partition ID of the root file system
4149 partition should be identical to the upper 128bit of the Verity root
4150 hash. The GPT partition ID of the Verity partition protecting it
4151 should be the lower 128bit of the Verity root hash. If the partition
4152 image follows this model it is sufficient to specify a single
4153 "roothash=" kernel command line argument to both configure which root
4154 image and verity partition to use as well as the root hash for
4155 it. Note that systemd-nspawn's Verity support follows the same
4156 semantics, meaning that disk images with proper Verity data in place
4157 may be booted in containers with systemd-nspawn as well as on
4158 physical systems via the verity generator. Also note that the "mkosi"
4159 tool available at https://github.com/systemd/mkosi has been updated
4160 to generate Verity protected disk images following this scheme. In
4161 fact, it has been updated to generate disk images that optionally
4162 implement a complete UEFI SecureBoot trust chain, involving a signed
4163 kernel and initrd image that incorporates such a root hash as well as
4164 a Verity-enabled root partition.
4166 * The hardware database (hwdb) udev supports has been updated to carry
4167 accelerometer quirks.
4169 * All system services are now run with a fresh kernel keyring set up
4170 for them. The invocation ID is stored by default in it, thus
4171 providing a safe, non-overridable way to determine the invocation
4174 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
4175 options for bind mounting arbitrary paths in a service-specific
4176 way. When these options are used, arbitrary host or service files and
4177 directories may be mounted to arbitrary locations in the service's
4180 * Documentation has been added that lists all of systemd's low-level
4181 environment variables:
4183 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
4185 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
4186 whether a specific socket file descriptor matches a specified socket
4189 * systemd-firstboot has been updated to check for the
4190 systemd.firstboot= kernel command line option. It accepts a boolean
4191 and when set to false the first boot questions are skipped.
4193 * systemd-fstab-generator has been updated to check for the
4194 systemd.volatile= kernel command line option, which either takes an
4195 optional boolean parameter or the special value "state". If used the
4196 system may be booted in a "volatile" boot mode. Specifically,
4197 "systemd.volatile" is used, the root directory will be mounted as
4198 tmpfs, and only /usr is mounted from the actual root file system. If
4199 "systemd.volatile=state" is used, the root directory will be mounted
4200 as usual, but /var is mounted as tmpfs. This concept provides similar
4201 functionality as systemd-nspawn's --volatile= option, but provides it
4202 on physical boots. Use this option for implementing stateless
4203 systems, or testing systems with all state and/or configuration reset
4204 to the defaults. (Note though that many distributions are not
4205 prepared to boot up without a populated /etc or /var, though.)
4207 * systemd-gpt-auto-generator gained support for LUKS encrypted root
4208 partitions. Previously it only supported LUKS encrypted partitions
4209 for all other uses, except for the root partition itself.
4211 * Socket units gained support for listening on AF_VSOCK sockets for
4212 communication in virtualized QEMU environments.
4214 * The "configure" script gained a new option --with-fallback-hostname=
4215 for specifying the fallback hostname to use if none is configured in
4216 /etc/hostname. For example, by specifying
4217 --with-fallback-hostname=fedora it is possible to default to a
4218 hostname of "fedora" on pristine installations.
4220 * systemd-cgls gained support for a new --unit= switch for listing only
4221 the control groups of a specific unit. Similar --user-unit= has been
4222 added for listing only the control groups of a specific user unit.
4224 * systemd-mount gained a new --umount switch for unmounting a mount or
4225 automount point (and all mount/automount points below it).
4227 * systemd will now refuse full configuration reloads (via systemctl
4228 daemon-reload and related calls) unless at least 16MiB of free space
4229 are available in /run. This is a safety precaution in order to ensure
4230 that generators can safely operate after the reload completed.
4232 * A new unit file option RootImage= has been added, which has a similar
4233 effect as RootDirectory= but mounts the service's root directory from
4234 a disk image instead of plain directory. This logic reuses the same
4235 image dissection and mount logic that systemd-nspawn already uses,
4236 and hence supports any disk images systemd-nspawn supports, including
4237 those following the Discoverable Partition Specification, as well as
4238 Verity enabled images. This option enables systemd to run system
4239 services directly off disk images acting as resource bundles,
4240 possibly even including full integrity data.
4242 * A new MountAPIVFS= unit file option has been added, taking a boolean
4243 argument. If enabled /proc, /sys and /dev (collectively called the
4244 "API VFS") will be mounted for the service. This is only relevant if
4245 RootDirectory= or RootImage= is used for the service, as these mounts
4246 are of course in place in the host mount namespace anyway.
4248 * systemd-nspawn gained support for a new --pivot-root= switch. If
4249 specified the root directory within the container image is pivoted to
4250 the specified mount point, while the original root disk is moved to a
4251 different place. This option enables booting of ostree images
4252 directly with systemd-nspawn.
4254 * The systemd build scripts will no longer complain if the NTP server
4255 addresses are not changed from the defaults. Google now supports
4256 these NTP servers officially. We still recommend downstreams to
4257 properly register an NTP pool with the NTP pool project though.
4259 * coredumpctl gained a new "--reverse" option for printing the list
4260 of coredumps in reverse order.
4262 * coredumpctl will now show additional information about truncated and
4263 inaccessible coredumps, as well as coredumps that are still being
4264 processed. It also gained a new --quiet switch for suppressing
4265 additional informational message in its output.
4267 * coredumpctl gained support for only showing coredumps newer and/or
4268 older than specific timestamps, using the new --since= and --until=
4269 options, reminiscent of journalctl's options by the same name.
4271 * The systemd-coredump logic has been improved so that it may be reused
4272 to collect backtraces in non-compiled languages, for example in
4273 scripting languages such as Python.
4275 * machinectl will now show the UID shift of local containers, if user
4276 namespacing is enabled for them.
4278 * systemd will now optionally run "environment generator" binaries at
4279 configuration load time. They may be used to add environment
4280 variables to the environment block passed to services invoked. One
4281 user environment generator is shipped by default that sets up
4282 environment variables based on files dropped into /etc/environment.d
4283 and ~/.config/environment.d/.
4285 * systemd-resolved now includes the new, recently published 2017 DNSSEC
4288 * hostnamed has been updated to report a new chassis type of
4289 "convertible" to cover "foldable" laptops that can both act as a
4290 tablet and as a laptop, such as various Lenovo Yoga devices.
4292 Contributions from: Adrián López, Alexander Galanin, Alexander
4293 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
4294 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
4295 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
4296 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
4297 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
4298 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
4299 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
4300 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
4301 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
4302 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
4303 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
4304 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
4305 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
4306 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
4307 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
4308 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
4309 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
4310 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
4311 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
4312 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
4313 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
4314 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
4315 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
4316 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
4317 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
4318 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
4319 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
4322 — Berlin, 2017-03-01
4326 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
4327 RestrictAddressFamilies= enabled. These sandboxing options should
4328 generally be compatible with the various external udev call-out
4329 binaries we are aware of, however there may be exceptions, in
4330 particular when exotic languages for these call-outs are used. In
4331 this case, consider turning off these settings locally.
4333 * The new RemoveIPC= option can be used to remove IPC objects owned by
4334 the user or group of a service when that service exits.
4336 * The new ProtectKernelModules= option can be used to disable explicit
4337 load and unload operations of kernel modules by a service. In
4338 addition access to /usr/lib/modules is removed if this option is set.
4340 * ProtectSystem= option gained a new value "strict", which causes the
4341 whole file system tree with the exception of /dev, /proc, and /sys,
4342 to be remounted read-only for a service.
4344 * The new ProtectKernelTunables= option can be used to disable
4345 modification of configuration files in /sys and /proc by a service.
4346 Various directories and files are remounted read-only, so access is
4347 restricted even if the file permissions would allow it.
4349 * The new ProtectControlGroups= option can be used to disable write
4350 access by a service to /sys/fs/cgroup.
4352 * Various systemd services have been hardened with
4353 ProtectKernelTunables=yes, ProtectControlGroups=yes,
4354 RestrictAddressFamilies=.
4356 * Support for dynamically creating users for the lifetime of a service
4357 has been added. If DynamicUser=yes is specified, user and group IDs
4358 will be allocated from the range 61184..65519 for the lifetime of the
4359 service. They can be resolved using the new nss-systemd.so NSS
4360 module. The module must be enabled in /etc/nsswitch.conf. Services
4361 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
4362 any resources allocated by the service will be cleaned up when the
4363 service exits. They also have ProtectHome=read-only and
4364 ProtectSystem=strict enabled, so they are not able to make any
4365 permanent modifications to the system.
4367 * The nss-systemd module also always resolves root and nobody, making
4368 it possible to have no /etc/passwd or /etc/group files in minimal
4369 container or chroot environments.
4371 * Services may be started with their own user namespace using the new
4372 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
4373 under which the service is running are mapped. All other users are
4376 * Support for the cgroup namespace has been added to systemd-nspawn. If
4377 supported by kernel, the container system started by systemd-nspawn
4378 will have its own view of the cgroup hierarchy. This new behaviour
4379 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
4381 * The new MemorySwapMax= option can be used to limit the maximum swap
4382 usage under the unified cgroup hierarchy.
4384 * Support for the CPU controller in the unified cgroup hierarchy has
4385 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
4386 options. This controller requires out-of-tree patches for the kernel
4387 and the support is provisional.
4389 * Mount and automount units may now be created transiently
4390 (i.e. dynamically at runtime via the bus API, instead of requiring
4391 unit files in the file system).
4393 * systemd-mount is a new tool which may mount file systems – much like
4394 mount(8), optionally pulling in additional dependencies through
4395 transient .mount and .automount units. For example, this tool
4396 automatically runs fsck on a backing block device before mounting,
4397 and allows the automount logic to be used dynamically from the
4398 command line for establishing mount points. This tool is particularly
4399 useful when dealing with removable media, as it will ensure fsck is
4400 run – if necessary – before the first access and that the file system
4401 is quickly unmounted after each access by utilizing the automount
4402 logic. This maximizes the chance that the file system on the
4403 removable media stays in a clean state, and if it isn't in a clean
4404 state is fixed automatically.
4406 * LazyUnmount=yes option for mount units has been added to expose the
4407 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
4410 * /efi will be used as the mount point of the EFI boot partition, if
4411 the directory is present, and the mount point was not configured
4412 through other means (e.g. fstab). If /efi directory does not exist,
4413 /boot will be used as before. This makes it easier to automatically
4414 mount the EFI partition on systems where /boot is used for something
4417 * When operating on GPT disk images for containers, systemd-nspawn will
4418 now mount the ESP to /boot or /efi according to the same rules as PID
4419 1 running on a host. This allows tools like "bootctl" to operate
4420 correctly within such containers, in order to make container images
4421 bootable on physical systems.
4423 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
4425 * Two new user session targets have been added to support running
4426 graphical sessions under the systemd --user instance:
4427 graphical-session.target and graphical-session-pre.target. See
4428 systemd.special(7) for a description of how those targets should be
4431 * The vconsole initialization code has been significantly reworked to
4432 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
4433 support unicode keymaps. Font and keymap configuration will now be
4434 copied to all allocated virtual consoles.
4436 * FreeBSD's bhyve virtualization is now detected.
4438 * Information recorded in the journal for core dumps now includes the
4439 contents of /proc/mountinfo and the command line of the process at
4440 the top of the process hierarchy (which is usually the init process
4443 * systemd-journal-gatewayd learned the --directory= option to serve
4444 files from the specified location.
4446 * journalctl --root=… can be used to peruse the journal in the
4447 /var/log/ directories inside of a container tree. This is similar to
4448 the existing --machine= option, but does not require the container to
4451 * The hardware database has been extended to support
4452 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
4455 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
4456 specify the click rate for mice which include a horizontal wheel with
4457 a click rate that is different than the one for the vertical wheel.
4459 * systemd-run gained a new --wait option that makes service execution
4460 synchronous. (Specifically, the command will not return until the
4461 specified service binary exited.)
4463 * systemctl gained a new --wait option that causes the start command to
4464 wait until the units being started have terminated again.
4466 * A new journal output mode "short-full" has been added which displays
4467 timestamps with abbreviated English day names and adds a timezone
4468 suffix. Those timestamps include more information than the default
4469 "short" output mode, and can be passed directly to journalctl's
4470 --since= and --until= options.
4472 * /etc/resolv.conf will be bind-mounted into containers started by
4473 systemd-nspawn, if possible, so any changes to resolv.conf contents
4474 are automatically propagated to the container.
4476 * The number of instances for socket-activated services originating
4477 from a single IP address can be limited with
4478 MaxConnectionsPerSource=, extending the existing setting of
4481 * systemd-networkd gained support for vcan ("Virtual CAN") interface
4484 * .netdev and .network configuration can now be extended through
4487 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
4488 Segmentation Offload, Generic Receive Offload, Large Receive Offload
4489 can be enabled and disabled using the new UDPSegmentationOffload=,
4490 TCPSegmentationOffload=, GenericSegmentationOffload=,
4491 GenericReceiveOffload=, LargeReceiveOffload= options in the
4492 [Link] section of .link files.
4494 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
4495 Port VLAN ID can be configured for bridge devices using the new STP=,
4496 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
4497 section of .netdev files.
4499 * The route table to which routes received over DHCP or RA should be
4500 added can be configured with the new RouteTable= option in the [DHCP]
4501 and [IPv6AcceptRA] sections of .network files.
4503 * The Address Resolution Protocol can be disabled on links managed by
4504 systemd-networkd using the ARP=no setting in the [Link] section of
4507 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
4508 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
4509 encode information about the result and exit codes of the current
4510 service runtime cycle.
4512 * systemd-sysctl will now configure kernel parameters in the order
4513 they occur in the configuration files. This matches what sysctl
4514 has been traditionally doing.
4516 * kernel-install "plugins" that are executed to perform various
4517 tasks after a new kernel is added and before an old one is removed
4518 can now return a special value to terminate the procedure and
4519 prevent any later plugins from running.
4521 * Journald's SplitMode=login setting has been deprecated. It has been
4522 removed from documentation, and its use is discouraged. In a future
4523 release it will be completely removed, and made equivalent to current
4524 default of SplitMode=uid.
4526 * Storage=both option setting in /etc/systemd/coredump.conf has been
4527 removed. With fast LZ4 compression storing the core dump twice is not
4530 * The --share-system systemd-nspawn option has been replaced with an
4531 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
4532 this functionality is discouraged. In addition the variables
4533 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
4534 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
4535 individual namespaces.
4537 * "machinectl list" now shows the IP address of running containers in
4538 the output, as well as OS release information.
4540 * "loginctl list" now shows the TTY of each session in the output.
4542 * sd-bus gained new API calls sd_bus_track_set_recursive(),
4543 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
4544 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
4545 tracking objects in a "recursive" mode, where a single client can be
4546 counted multiple times, if it takes multiple references.
4548 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
4549 sd_bus_get_exit_on_disconnect(). They may be used to make a
4550 process using sd-bus automatically exit if the bus connection is
4553 * Bus clients of the service manager may now "pin" loaded units into
4554 memory, by taking an explicit reference on them. This is useful to
4555 ensure the client can retrieve runtime data about the service even
4556 after the service completed execution. Taking such a reference is
4557 available only for privileged clients and should be helpful to watch
4558 running services in a race-free manner, and in particular collect
4559 information about exit statuses and results.
4561 * The nss-resolve module has been changed to strictly return UNAVAIL
4562 when communication via D-Bus with resolved failed, and NOTFOUND when
4563 a lookup completed but was negative. This means it is now possible to
4564 neatly configure fallbacks using nsswitch.conf result checking
4565 expressions. Taking benefit of this, the new recommended
4566 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
4568 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
4570 * A new setting CtrlAltDelBurstAction= has been added to
4571 /etc/systemd/system.conf which may be used to configure the precise
4572 behaviour if the user on the console presses Ctrl-Alt-Del more often
4573 than 7 times in 2s. Previously this would unconditionally result in
4574 an expedited, immediate reboot. With this new setting the precise
4575 operation may be configured in more detail, and also turned off
4578 * In .netdev files two new settings RemoteChecksumTx= and
4579 RemoteChecksumRx= are now understood that permit configuring the
4580 remote checksumming logic for VXLAN networks.
4582 * The service manager learnt a new "invocation ID" concept for invoked
4583 services. Each runtime cycle of a service will get a new invocation
4584 ID (a 128bit random UUID) assigned that identifies the current
4585 run of the service uniquely and globally. A new invocation ID
4586 is generated each time a service starts up. The journal will store
4587 the invocation ID of a service along with any logged messages, thus
4588 making the invocation ID useful for matching the online runtime of a
4589 service with the offline log data it generated in a safe way without
4590 relying on synchronized timestamps. In many ways this new service
4591 invocation ID concept is similar to the kernel's boot ID concept that
4592 uniquely and globally identifies the runtime of each boot. The
4593 invocation ID of a service is passed to the service itself via an
4594 environment variable ($INVOCATION_ID). A new bus call
4595 GetUnitByInvocationID() has been added that is similar to GetUnit()
4596 but instead of retrieving the bus path for a unit by its name
4597 retrieves it by its invocation ID. The returned path is valid only as
4598 long as the passed invocation ID is current.
4600 * systemd-resolved gained a new "DNSStubListener" setting in
4601 resolved.conf. It either takes a boolean value or the special values
4602 "udp" and "tcp", and configures whether to enable the stub DNS
4603 listener on 127.0.0.53:53.
4605 * IP addresses configured via networkd may now carry additional
4606 configuration settings supported by the kernel. New options include:
4607 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
4608 PrefixRoute=, AutoJoin=.
4610 * The PAM configuration fragment file for "user@.service" shipped with
4611 systemd (i.e. the --user instance of systemd) has been stripped to
4612 the minimum necessary to make the system boot. Previously, it
4613 contained Fedora-specific stanzas that did not apply to other
4614 distributions. It is expected that downstream distributions add
4615 additional configuration lines, matching their needs to this file,
4616 using it only as rough template of what systemd itself needs. Note
4617 that this reduced fragment does not even include an invocation of
4618 pam_limits which most distributions probably want to add, even though
4619 systemd itself does not need it. (There's also the new build time
4620 option --with-pamconfdir=no to disable installation of the PAM
4623 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
4624 capability is now also dropped from its set (in addition to
4625 CAP_SYS_MKNOD as before).
4627 * In service unit files it is now possible to connect a specific named
4628 file descriptor with stdin/stdout/stdout of an executed service. The
4629 name may be specified in matching .socket units using the
4630 FileDescriptorName= setting.
4632 * A number of journal settings may now be configured on the kernel
4633 command line. Specifically, the following options are now understood:
4634 systemd.journald.max_level_console=,
4635 systemd.journald.max_level_store=,
4636 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
4637 systemd.journald.max_level_wall=.
4639 * "systemctl is-enabled --full" will now show by which symlinks a unit
4640 file is enabled in the unit dependency tree.
4642 * Support for VeraCrypt encrypted partitions has been added to the
4643 "cryptsetup" logic and /etc/crypttab.
4645 * systemd-detect-virt gained support for a new --private-users switch
4646 that checks whether the invoking processes are running inside a user
4647 namespace. Similar, a new special value "private-users" for the
4648 existing ConditionVirtualization= setting has been added, permitting
4649 skipping of specific units in user namespace environments.
4651 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
4652 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
4653 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
4654 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
4655 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
4656 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
4657 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
4658 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
4659 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
4660 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
4661 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
4662 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
4663 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
4664 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
4665 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
4666 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
4667 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
4668 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
4669 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
4670 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
4671 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
4672 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
4673 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
4674 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
4675 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
4676 Jędrzejewski-Szmek, Zeal Jagannatha
4678 — Santa Fe, 2016-11-03
4682 * In service units the various ExecXYZ= settings have been extended
4683 with an additional special character as first argument of the
4684 assigned value: if the character '+' is used the specified command
4685 line it will be run with full privileges, regardless of User=,
4686 Group=, CapabilityBoundingSet= and similar options. The effect is
4687 similar to the existing PermissionsStartOnly= option, but allows
4688 configuration of this concept for each executed command line
4691 * Services may now alter the service watchdog timeout at runtime by
4692 sending a WATCHDOG_USEC= message via sd_notify().
4694 * MemoryLimit= and related unit settings now optionally take percentage
4695 specifications. The percentage is taken relative to the amount of
4696 physical memory in the system (or in case of containers, the assigned
4697 amount of memory). This allows scaling service resources neatly with
4698 the amount of RAM available on the system. Similarly, systemd-logind's
4699 RuntimeDirectorySize= option now also optionally takes percentage
4702 * In similar fashion TasksMax= takes percentage values now, too. The
4703 value is taken relative to the configured maximum number of processes
4704 on the system. The per-service task maximum has been changed to 15%
4705 using this functionality. (Effectively this is an increase of 512 →
4706 4915 for service units, given the kernel's default pid_max setting.)
4708 * Calendar time specifications in .timer units now understand a ".."
4709 syntax for time ranges. Example: "4..7:10" may now be used for
4710 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
4713 * The InaccessableDirectories=, ReadOnlyDirectories= and
4714 ReadWriteDirectories= unit file settings have been renamed to
4715 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
4716 applied to all kinds of file nodes, and not just directories, with
4717 the exception of symlinks. Specifically these settings may now be
4718 used on block and character device nodes, UNIX sockets and FIFOS as
4719 well as regular files. The old names of these settings remain
4720 available for compatibility.
4722 * systemd will now log about all service processes it kills forcibly
4723 (using SIGKILL) because they remained after the clean shutdown phase
4724 of the service completed. This should help identifying services that
4725 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
4726 systemd-logind's configuration a similar log message is generated for
4727 processes killed at the end of each session due to this setting.
4729 * systemd will now set the $JOURNAL_STREAM environment variable for all
4730 services whose stdout/stderr are connected to the Journal (which
4731 effectively means by default: all services). The variable contains
4732 the device and inode number of the file descriptor used for
4733 stdout/stderr. This may be used by invoked programs to detect whether
4734 their stdout/stderr is connected to the Journal, in which case they
4735 can switch over to direct Journal communication, thus being able to
4736 pass extended, structured metadata along with their log messages. As
4737 one example, this is now used by glib's logging primitives.
4739 * When using systemd's default tmp.mount unit for /tmp, the mount point
4740 will now be established with the "nosuid" and "nodev" options. This
4741 avoids privilege escalation attacks that put traps and exploits into
4742 /tmp. However, this might cause problems if you e. g. put container
4743 images or overlays into /tmp; if you need this, override tmp.mount's
4744 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
4747 * systemd now supports the "memory" cgroup controller also on
4750 * The systemd-cgtop tool now optionally takes a control group path as
4751 command line argument. If specified, the control group list shown is
4752 limited to subgroups of that group.
4754 * The SystemCallFilter= unit file setting gained support for
4755 pre-defined, named system call filter sets. For example
4756 SystemCallFilter=@clock is now an effective way to make all clock
4757 changing-related system calls unavailable to a service. A number of
4758 similar pre-defined groups are defined. Writing system call filters
4759 for system services is simplified substantially with this new
4760 concept. Accordingly, all of systemd's own, long-running services now
4761 enable system call filtering based on this, by default.
4763 * A new service setting MemoryDenyWriteExecute= has been added, taking
4764 a boolean value. If turned on, a service may no longer create memory
4765 mappings that are writable and executable at the same time. This
4766 enhances security for services where this is enabled as it becomes
4767 harder to dynamically write and then execute memory in exploited
4768 service processes. This option has been enabled for all of systemd's
4769 own long-running services.
4771 * A new RestrictRealtime= service setting has been added, taking a
4772 boolean argument. If set the service's processes may no longer
4773 acquire realtime scheduling. This improves security as realtime
4774 scheduling may otherwise be used to easily freeze the system.
4776 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
4777 value. This may be used for requesting that the system manager inside
4778 of the container reports start-up completion to nspawn which then
4779 propagates this notification further to the service manager
4780 supervising nspawn itself. A related option NotifyReady= in .nspawn
4781 files has been added too. This functionality allows ordering of the
4782 start-up of multiple containers using the usual systemd ordering
4785 * machinectl gained a new command "stop" that is an alias for
4788 * systemd-resolved gained support for contacting DNS servers on
4789 link-local IPv6 addresses.
4791 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
4792 its caches. A method call for requesting the same operation has been
4793 added to the bus API too, and is made available via "systemd-resolve
4796 * systemd-resolve gained a new --status switch. If passed a brief
4797 summary of the used DNS configuration with per-interface information
4800 * resolved.conf gained a new Cache= boolean option, defaulting to
4801 on. If turned off local DNS caching is disabled. This comes with a
4802 performance penalty in particular when DNSSEC is enabled. Note that
4803 resolved disables its internal caching implicitly anyway, when the
4804 configured DNS server is on a host-local IP address such as ::1 or
4805 127.0.0.1, thus automatically avoiding double local caching.
4807 * systemd-resolved now listens on the local IP address 127.0.0.53:53
4808 for DNS requests. This improves compatibility with local programs
4809 that do not use the libc NSS or systemd-resolved's bus APIs for name
4810 resolution. This minimal DNS service is only available to local
4811 programs and does not implement the full DNS protocol, but enough to
4812 cover local DNS clients. A new, static resolv.conf file, listing just
4813 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
4814 now recommended to make /etc/resolv.conf a symlink to this file in
4815 order to route all DNS lookups to systemd-resolved, regardless if
4816 done via NSS, the bus API or raw DNS packets. Note that this local
4817 DNS service is not as fully featured as the libc NSS or
4818 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
4819 used to deliver link-local address information (as this implies
4820 sending a local interface index along), LLMNR/mDNS support via this
4821 interface is severely restricted. It is thus strongly recommended for
4822 all applications to use the libc NSS API or native systemd-resolved
4825 * systemd-networkd's bridge support learned a new setting
4826 VLANFiltering= for controlling VLAN filtering. Moreover a new section
4827 in .network files has been added for configuring VLAN bridging in
4828 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
4830 * systemd-networkd's IPv6 Router Advertisement code now makes use of
4831 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
4832 now be acquired without relying on DHCPv6. Two new options
4833 UseDomains= and UseDNS= have been added to configure this behaviour.
4835 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
4836 renamed IPv6AcceptRA=, without altering its behaviour. The old
4837 setting name remains available for compatibility reasons.
4839 * The systemd-networkd VTI/VTI6 tunneling support gained new options
4840 Key=, InputKey= and OutputKey=.
4842 * systemd-networkd gained support for VRF ("Virtual Routing Function")
4843 interface configuration.
4845 * "systemctl edit" may now be used to create new unit files by
4846 specifying the --force switch.
4848 * sd-event gained a new function sd_event_get_iteration() for
4849 requesting the current iteration counter of the event loop. It starts
4850 at zero and is increased by one with each event loop iteration.
4852 * A new rpm macro %systemd_ordering is provided by the macros.systemd
4853 file. It can be used in lieu of %systemd_requires in packages which
4854 don't use any systemd functionality and are intended to be installed
4855 in minimal containers without systemd present. This macro provides
4856 ordering dependencies to ensure that if the package is installed in
4857 the same rpm transaction as systemd, systemd will be installed before
4858 the scriptlets for the package are executed, allowing unit presets
4861 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
4862 been added to simplify packaging of generators.
4864 * The os-release file gained VERSION_CODENAME field for the
4865 distribution nickname (e.g. VERSION_CODENAME=woody).
4867 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
4868 can be set to disable parsing of metadata and the creation
4869 of persistent symlinks for that device.
4871 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
4872 to make them available to logged-in users has been reverted.
4874 * Much of the common code of the various systemd components is now
4875 built into an internal shared library libsystemd-shared-231.so
4876 (incorporating the systemd version number in the name, to be updated
4877 with future releases) that the components link to. This should
4878 decrease systemd footprint both in memory during runtime and on
4879 disk. Note that the shared library is not for public use, and is
4880 neither API nor ABI stable, but is likely to change with every new
4881 released update. Packagers need to make sure that binaries
4882 linking to libsystemd-shared.so are updated in step with the
4885 * Configuration for "mkosi" is now part of the systemd
4886 repository. mkosi is a tool to easily build legacy-free OS images,
4887 and is available on github: https://github.com/systemd/mkosi. If
4888 "mkosi" is invoked in the build tree a new raw OS image is generated
4889 incorporating the systemd sources currently being worked on and a
4890 clean, fresh distribution installation. The generated OS image may be
4891 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
4892 UEFI PC. This functionality is particularly useful to easily test
4893 local changes made to systemd in a pristine, defined environment. See
4894 doc/HACKING for details.
4896 * configure learned the --with-support-url= option to specify the
4897 distribution's bugtracker.
4899 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
4900 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
4901 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
4902 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
4903 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
4904 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
4905 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
4906 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
4907 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
4908 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
4909 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
4910 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
4911 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
4912 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
4913 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
4914 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
4915 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
4916 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
4917 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
4919 — Berlin, 2016-07-25
4923 * DNSSEC is now turned on by default in systemd-resolved (in
4924 "allow-downgrade" mode), but may be turned off during compile time by
4925 passing "--with-default-dnssec=no" to "configure" (and of course,
4926 during runtime with DNSSEC= in resolved.conf). We recommend
4927 downstreams to leave this on at least during development cycles and
4928 report any issues with the DNSSEC logic upstream. We are very
4929 interested in collecting feedback about the DNSSEC validator and its
4930 limitations in the wild. Note however, that DNSSEC support is
4931 probably nothing downstreams should turn on in stable distros just
4932 yet, as it might create incompatibilities with a few DNS servers and
4933 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
4934 automatically whenever we detect such incompatible setups, but there
4935 might be systems we do not cover yet. Hence: please help us testing
4936 the DNSSEC code, leave this on where you can, report back, but then
4937 again don't consider turning this on in your stable, LTS or
4938 production release just yet. (Note that you have to enable
4939 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
4940 and its DNSSEC mode for hostname resolution from local
4943 * systemd-resolve conveniently resolves DANE records with the --tlsa
4944 option and OPENPGPKEY records with the --openpgp option. It also
4945 supports dumping raw DNS record data via the new --raw= switch.
4947 * systemd-logind will now by default terminate user processes that are
4948 part of the user session scope unit (session-XX.scope) when the user
4949 logs out. This behavior is controlled by the KillUserProcesses=
4950 setting in logind.conf, and the previous default of "no" is now
4951 changed to "yes". This means that user sessions will be properly
4952 cleaned up after, but additional steps are necessary to allow
4953 intentionally long-running processes to survive logout.
4955 While the user is logged in at least once, user@.service is running,
4956 and any service that should survive the end of any individual login
4957 session can be started at a user service or scope using systemd-run.
4958 systemd-run(1) man page has been extended with an example which shows
4959 how to run screen in a scope unit underneath user@.service. The same
4960 command works for tmux.
4962 After the user logs out of all sessions, user@.service will be
4963 terminated too, by default, unless the user has "lingering" enabled.
4964 To effectively allow users to run long-term tasks even if they are
4965 logged out, lingering must be enabled for them. See loginctl(1) for
4966 details. The default polkit policy was modified to allow users to
4967 set lingering for themselves without authentication.
4969 Previous defaults can be restored at compile time by the
4970 --without-kill-user-processes option to "configure".
4972 * systemd-logind gained new configuration settings SessionsMax= and
4973 InhibitorsMax=, both with a default of 8192. It will not register new
4974 user sessions or inhibitors above this limit.
4976 * systemd-logind will now reload configuration on SIGHUP.
4978 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
4979 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
4980 enable. Also, support for the "io" cgroup controller in the unified
4981 hierarchy has been added, so that the "memory", "pids" and "io" are
4982 now the controllers that are supported on the unified hierarchy.
4984 WARNING: it is not possible to use previous systemd versions with
4985 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
4986 is necessary to also update systemd in the initramfs if using the
4987 unified hierarchy. An updated SELinux policy is also required.
4989 * LLDP support has been extended, and both passive (receive-only) and
4990 active (sender) modes are supported. Passive mode ("routers-only") is
4991 enabled by default in systemd-networkd. Active LLDP mode is enabled
4992 by default for containers on the internal network. The "networkctl
4993 lldp" command may be used to list information gathered. "networkctl
4994 status" will also show basic LLDP information on connected peers now.
4996 * The IAID and DUID unique identifier sent in DHCP requests may now be
4997 configured for the system and each .network file managed by
4998 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
5000 * systemd-networkd gained support for configuring proxy ARP support for
5001 each interface, via the ProxyArp= setting in .network files. It also
5002 gained support for configuring the multicast querier feature of
5003 bridge devices, via the new MulticastQuerier= setting in .netdev
5004 files. Similarly, snooping on the IGMP traffic can be controlled
5005 via the new setting MulticastSnooping=.
5007 A new setting PreferredLifetime= has been added for addresses
5008 configured in .network file to configure the lifetime intended for an
5011 The systemd-networkd DHCP server gained the option EmitRouter=, which
5012 defaults to yes, to configure whether the DHCP Option 3 (Router)
5015 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
5016 systemd-socket-activate and installed into /usr/bin. It is now fully
5019 * systemd-journald now uses separate threads to flush changes to disk
5020 when closing journal files, thus reducing impact of slow disk I/O on
5021 logging performance.
5023 * The sd-journal API gained two new calls
5024 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
5025 can be used to open journal files using file descriptors instead of
5026 file or directory paths. sd_journal_open_container() has been
5027 deprecated, sd_journal_open_directory_fd() should be used instead
5028 with the flag SD_JOURNAL_OS_ROOT.
5030 * journalctl learned a new output mode "-o short-unix" that outputs log
5031 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
5032 UTC). It also gained support for a new --no-hostname setting to
5033 suppress the hostname column in the family of "short" output modes.
5035 * systemd-ask-password now optionally skips printing of the password to
5036 stdout with --no-output which can be useful in scripts.
5038 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
5039 (devices tagged with ID_MAKER_TOOL) are now tagged with
5040 "uaccess" and are available to logged in users.
5042 * The DeviceAllow= unit setting now supports specifiers (with "%").
5044 * "systemctl show" gained a new --value switch, which allows print a
5045 only the contents of a specific unit property, without also printing
5046 the property's name. Similar support was added to "show*" verbs
5047 of loginctl and machinectl that output "key=value" lists.
5049 * A new unit type "generated" was added for files dynamically generated
5050 by generator tools. Similarly, a new unit type "transient" is used
5051 for unit files created using the runtime API. "systemctl enable" will
5052 refuse to operate on such files.
5054 * A new command "systemctl revert" has been added that may be used to
5055 revert to the vendor version of a unit file, in case local changes
5056 have been made by adding drop-ins or overriding the unit file.
5058 * "machinectl clean" gained a new verb to automatically remove all or
5059 just hidden container images.
5061 * systemd-tmpfiles gained support for a new line type "e" for emptying
5062 directories, if they exist, without creating them if they don't.
5064 * systemd-nspawn gained support for automatically patching the UID/GIDs
5065 of the owners and the ACLs of all files and directories in a
5066 container tree to match the UID/GID user namespacing range selected
5067 for the container invocation. This mode is enabled via the new
5068 --private-users-chown switch. It also gained support for
5069 automatically choosing a free, previously unused UID/GID range when
5070 starting a container, via the new --private-users=pick setting (which
5071 implies --private-users-chown). Together, these options for the first
5072 time make user namespacing for nspawn containers fully automatic and
5073 thus deployable. The systemd-nspawn@.service template unit file has
5074 been changed to use this functionality by default.
5076 * systemd-nspawn gained a new --network-zone= switch, that allows
5077 creating ad-hoc virtual Ethernet links between multiple containers,
5078 that only exist as long as at least one container referencing them is
5079 running. This allows easy connecting of multiple containers with a
5080 common link that implements an Ethernet broadcast domain. Each of
5081 these network "zones" may be named relatively freely by the user, and
5082 may be referenced by any number of containers, but each container may
5083 only reference one of these "zones". On the lower level, this is
5084 implemented by an automatically managed bridge network interface for
5085 each zone, that is created when the first container referencing its
5086 zone is created and removed when the last one referencing its zone
5089 * The default start timeout may now be configured on the kernel command
5090 line via systemd.default_timeout_start_sec=. It was already
5091 configurable via the DefaultTimeoutStartSec= option in
5092 /etc/systemd/system.conf.
5094 * Socket units gained a new TriggerLimitIntervalSec= and
5095 TriggerLimitBurst= setting to configure a limit on the activation
5096 rate of the socket unit.
5098 * The LimitNICE= setting now optionally takes normal UNIX nice values
5099 in addition to the raw integer limit value. If the specified
5100 parameter is prefixed with "+" or "-" and is in the range -20..19 the
5101 value is understood as UNIX nice value. If not prefixed like this it
5102 is understood as raw RLIMIT_NICE limit.
5104 * Note that the effect of the PrivateDevices= unit file setting changed
5105 slightly with this release: the per-device /dev file system will be
5106 mounted read-only from this version on, and will have "noexec"
5107 set. This (minor) change of behavior might cause some (exceptional)
5108 legacy software to break, when PrivateDevices=yes is set for its
5109 service. Please leave PrivateDevices= off if you run into problems
5112 * systemd-bootchart has been split out to a separate repository:
5113 https://github.com/systemd/systemd-bootchart
5115 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
5116 merged into the kernel in its current form.
5118 * The compatibility libraries libsystemd-daemon.so,
5119 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
5120 which have been deprecated since systemd-209 have been removed along
5121 with the corresponding pkg-config files. All symbols provided by
5122 those libraries are provided by libsystemd.so.
5124 * The Capabilities= unit file setting has been removed (it is ignored
5125 for backwards compatibility). AmbientCapabilities= and
5126 CapabilityBoundingSet= should be used instead.
5128 * A new special target has been added, initrd-root-device.target,
5129 which creates a synchronization point for dependencies of the root
5130 device in early userspace. Initramfs builders must ensure that this
5131 target is now included in early userspace.
5133 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
5134 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
5135 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
5136 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
5137 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
5138 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
5139 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
5140 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
5141 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
5142 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
5143 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
5144 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
5145 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
5146 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
5147 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
5148 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
5149 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
5150 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
5151 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
5152 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
5153 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
5154 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
5155 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
5156 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
5157 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
5160 — Fairfax, 2016-05-21
5164 * The systemd-resolved DNS resolver service has gained a substantial
5165 set of new features, most prominently it may now act as a DNSSEC
5166 validating stub resolver. DNSSEC mode is currently turned off by
5167 default, but is expected to be turned on by default in one of the
5168 next releases. For now, we invite everybody to test the DNSSEC logic
5169 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
5170 service also gained a full set of D-Bus interfaces, including calls
5171 to configure DNS and DNSSEC settings per link (for use by external
5172 network management software). systemd-resolved and systemd-networkd
5173 now distinguish between "search" and "routing" domains. The former
5174 are used to qualify single-label names, the latter are used purely
5175 for routing lookups within certain domains to specific links.
5176 resolved now also synthesizes RRs for all entries from /etc/hosts.
5178 * The systemd-resolve tool (which is a client utility for
5179 systemd-resolved) has been improved considerably and is now fully
5180 supported and documented. Hence it has moved from /usr/lib/systemd to
5183 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
5186 * The coredump collection logic has been reworked: when a coredump is
5187 collected it is now written to disk, compressed and processed
5188 (including stacktrace extraction) from a new instantiated service
5189 systemd-coredump@.service, instead of directly from the
5190 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
5191 processing large coredumps can take up a substantial amount of
5192 resources and time, and this previously happened entirely outside of
5193 systemd's service supervision. With the new logic the core_pattern
5194 hook only does minimal metadata collection before passing off control
5195 to the new instantiated service, which is configured with a time
5196 limit, a nice level and other settings to minimize negative impact on
5197 the rest of the system. Also note that the new logic will honour the
5198 RLIMIT_CORE setting of the crashed process, which now allows users
5199 and processes to turn off coredumping for their processes by setting
5202 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
5203 and all forked processes by default. Previously, PID 1 would leave
5204 the setting at "0" for all processes, as set by the kernel. Note that
5205 the resource limit traditionally has no effect on the generated
5206 coredumps on the system if the /proc/sys/kernel/core_pattern hook
5207 logic is used. Since the limit is now honoured (see above) its
5208 default has been changed so that the coredumping logic is enabled by
5209 default for all processes, while allowing specific opt-out.
5211 * When the stacktrace is extracted from processes of system users, this
5212 is now done as "systemd-coredump" user, in order to sandbox this
5213 potentially security sensitive parsing operation. (Note that when
5214 processing coredumps of normal users this is done under the user ID
5215 of process that crashed, as before.) Packagers should take notice
5216 that it is now necessary to create the "systemd-coredump" system user
5217 and group at package installation time.
5219 * The systemd-activate socket activation testing tool gained support
5220 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
5221 and --seqpacket switches. It also has been extended to support both
5222 new-style and inetd-style file descriptor passing. Use the new
5223 --inetd switch to request inetd-style file descriptor passing.
5225 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
5226 variable, which takes a boolean value. If set to false, ANSI color
5227 output is disabled in the tools even when run on a terminal that
5230 * The VXLAN support in networkd now supports two new settings
5231 DestinationPort= and PortRange=.
5233 * A new systemd.machine_id= kernel command line switch has been added,
5234 that may be used to set the machine ID in /etc/machine-id if it is
5235 not initialized yet. This command line option has no effect if the
5236 file is already initialized.
5238 * systemd-nspawn gained a new --as-pid2 switch that invokes any
5239 specified command line as PID 2 rather than PID 1 in the
5240 container. In this mode PID 1 is a minimal stub init process that
5241 implements the special POSIX and Linux semantics of PID 1 regarding
5242 signal and child process management. Note that this stub init process
5243 is implemented in nspawn itself and requires no support from the
5244 container image. This new logic is useful to support running
5245 arbitrary commands in the container, as normal processes are
5246 generally not prepared to run as PID 1.
5248 * systemd-nspawn gained a new --chdir= switch for setting the current
5249 working directory for the process started in the container.
5251 * "journalctl /dev/sda" will now output all kernel log messages for
5252 specified device from the current boot, in addition to all devices
5253 that are parents of it. This should make log output about devices
5254 pretty useful, as long as kernel drivers attach enough metadata to
5255 the log messages. (The usual SATA drivers do.)
5257 * The sd-journal API gained two new calls
5258 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
5259 that report whether log data from /run or /var has been found.
5261 * journalctl gained a new switch "--fields" that prints all journal
5262 record field names currently in use in the journal. This is backed
5263 by two new sd-journal API calls sd_journal_enumerate_fields() and
5264 sd_journal_restart_fields().
5266 * Most configurable timeouts in systemd now expect an argument of
5267 "infinity" to turn them off, instead of "0" as before. The semantics
5268 from now on is that a timeout of "0" means "now", and "infinity"
5269 means "never". To maintain backwards compatibility, "0" continues to
5270 turn off previously existing timeout settings.
5272 * "systemctl reload-or-try-restart" has been renamed to "systemctl
5273 try-reload-or-restart" to clarify what it actually does: the "try"
5274 logic applies to both reloading and restarting, not just restarting.
5275 The old name continues to be accepted for compatibility.
5277 * On boot-up, when PID 1 detects that the system clock is behind the
5278 release date of the systemd version in use, the clock is now set
5279 to the latter. Previously, this was already done in timesyncd, in order
5280 to avoid running with clocks set to the various clock epochs such as
5281 1902, 1938 or 1970. With this change the logic is now done in PID 1
5282 in addition to timesyncd during early boot-up, so that it is enforced
5283 before the first process is spawned by systemd. Note that the logic
5284 in timesyncd remains, as it is more comprehensive and ensures
5285 clock monotonicity by maintaining a persistent timestamp file in
5286 /var. Since /var is generally not available in earliest boot or the
5287 initrd, this part of the logic remains in timesyncd, and is not done
5290 * Support for tweaking details in net_cls.class_id through the
5291 NetClass= configuration directive has been removed, as the kernel
5292 people have decided to deprecate that controller in cgroup v2.
5293 Userspace tools such as nftables are moving over to setting rules
5294 that are specific to the full cgroup path of a task, which obsoletes
5295 these controllers anyway. The NetClass= directive is kept around for
5296 legacy compatibility reasons. For a more in-depth description of the
5297 kernel change, please refer to the respective upstream commit:
5299 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
5301 * A new service setting RuntimeMaxSec= has been added that may be used
5302 to specify a maximum runtime for a service. If the timeout is hit, the
5303 service is terminated and put into a failure state.
5305 * A new service setting AmbientCapabilities= has been added. It allows
5306 configuration of additional Linux process capabilities that are
5307 passed to the activated processes. This is only available on very
5310 * The process resource limit settings in service units may now be used
5311 to configure hard and soft limits individually.
5313 * The various libsystemd APIs such as sd-bus or sd-event now publicly
5314 expose support for gcc's __attribute__((cleanup())) C extension.
5315 Specifically, for many object destructor functions alternative
5316 versions have been added that have names suffixed with "p" and take a
5317 pointer to a pointer to the object to destroy, instead of just a
5318 pointer to the object itself. This is useful because these destructor
5319 functions may be used directly as parameters to the cleanup
5320 construct. Internally, systemd has been a heavy user of this GCC
5321 extension for a long time, and with this change similar support is
5322 now available to consumers of the library outside of systemd. Note
5323 that by using this extension in your sources compatibility with old
5324 and strictly ANSI compatible C compilers is lost. However, all gcc or
5325 LLVM versions of recent years support this extension.
5327 * Timer units gained support for a new setting RandomizedDelaySec= that
5328 allows configuring some additional randomized delay to the configured
5329 time. This is useful to spread out timer events to avoid load peaks in
5330 clusters or larger setups.
5332 * Calendar time specifications now support sub-second accuracy.
5334 * Socket units now support listening on SCTP and UDP-lite protocol
5337 * The sd-event API now comes with a full set of man pages.
5339 * Older versions of systemd contained experimental support for
5340 compressing journal files and coredumps with the LZ4 compressor that
5341 was not compatible with the lz4 binary (due to API limitations of the
5342 lz4 library). This support has been removed; only support for files
5343 compatible with the lz4 binary remains. This LZ4 logic is now
5344 officially supported and no longer considered experimental.
5346 * The dkr image import logic has been removed again from importd. dkr's
5347 micro-services focus doesn't fit into the machine image focus of
5348 importd, and quickly got out of date with the upstream dkr API.
5350 * Creation of the /run/lock/lockdev/ directory was dropped from
5351 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
5352 been available for many years. If you still need this, you need to
5353 create your own tmpfiles.d config file with:
5355 d /run/lock/lockdev 0775 root lock -
5357 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
5358 and RebootArgument= have been moved from the [Service] section of
5359 unit files to [Unit], and they are now supported on all unit types,
5360 not just service units. Of course, systemd will continue to
5361 understand these settings also at the old location, in order to
5362 maintain compatibility.
5364 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
5365 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
5366 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
5367 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
5368 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
5369 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
5370 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
5371 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
5372 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
5373 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
5374 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
5375 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
5376 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
5377 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
5378 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
5379 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
5380 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
5381 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
5382 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
5384 — Berlin, 2016-02-11
5388 * A number of properties previously only settable in unit
5389 files are now also available as properties to set when
5390 creating transient units programmatically via the bus, as it
5391 is exposed with systemd-run's --property=
5392 setting. Specifically, these are: SyslogIdentifier=,
5393 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
5394 EnvironmentFile=, ReadWriteDirectories=,
5395 ReadOnlyDirectories=, InaccessibleDirectories=,
5396 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
5398 * When creating transient services via the bus API it is now
5399 possible to pass in a set of file descriptors to use as
5400 STDIN/STDOUT/STDERR for the invoked process.
5402 * Slice units may now be created transiently via the bus APIs,
5403 similar to the way service and scope units may already be
5404 created transiently.
5406 * Wherever systemd expects a calendar timestamp specification
5407 (like in journalctl's --since= and --until= switches) UTC
5408 timestamps are now supported. Timestamps suffixed with "UTC"
5409 are now considered to be in Universal Time Coordinated
5410 instead of the local timezone. Also, timestamps may now
5411 optionally be specified with sub-second accuracy. Both of
5412 these additions also apply to recurring calendar event
5413 specification, such as OnCalendar= in timer units.
5415 * journalctl gained a new "--sync" switch that asks the
5416 journal daemon to write all so far unwritten log messages to
5417 disk and sync the files, before returning.
5419 * systemd-tmpfiles learned two new line types "q" and "Q" that
5420 operate like "v", but also set up a basic btrfs quota
5421 hierarchy when used on a btrfs file system with quota
5424 * tmpfiles' "v", "q" and "Q" will now create a plain directory
5425 instead of a subvolume (even on a btrfs file system) if the
5426 root directory is a plain directory, and not a
5427 subvolume. This should simplify things with certain chroot()
5428 environments which are not aware of the concept of btrfs
5431 * systemd-detect-virt gained a new --chroot switch to detect
5432 whether execution takes place in a chroot() environment.
5434 * CPUAffinity= now takes CPU index ranges in addition to
5437 * The various memory-related resource limit settings (such as
5438 LimitAS=) now understand the usual K, M, G, ... suffixes to
5439 the base of 1024 (IEC). Similar, the time-related resource
5440 limit settings understand the usual min, h, day, ...
5443 * There's a new system.conf setting DefaultTasksMax= to
5444 control the default TasksMax= setting for services and
5445 scopes running on the system. (TasksMax= is the primary
5446 setting that exposes the "pids" cgroup controller on systemd
5447 and was introduced in the previous systemd release.) The
5448 setting now defaults to 512, which means services that are
5449 not explicitly configured otherwise will only be able to
5450 create 512 processes or threads at maximum, from this
5451 version on. Note that this means that thread- or
5452 process-heavy services might need to be reconfigured to set
5453 TasksMax= to a higher value. It is sufficient to set
5454 TasksMax= in these specific unit files to a higher value, or
5455 even "infinity". Similar, there's now a logind.conf setting
5456 UserTasksMax= that defaults to 4096 and limits the total
5457 number of processes or tasks each user may own
5458 concurrently. nspawn containers also have the TasksMax=
5459 value set by default now, to 8192. Note that all of this
5460 only has an effect if the "pids" cgroup controller is
5461 enabled in the kernel. The general benefit of these changes
5462 should be a more robust and safer system, that provides a
5463 certain amount of per-service fork() bomb protection.
5465 * systemd-nspawn gained the new --network-veth-extra= switch
5466 to define additional and arbitrarily-named virtual Ethernet
5467 links between the host and the container.
5469 * A new service execution setting PassEnvironment= has been
5470 added that allows importing select environment variables
5471 from PID1's environment block into the environment block of
5474 * Timer units gained support for a new RemainAfterElapse=
5475 setting which takes a boolean argument. It defaults to on,
5476 exposing behaviour unchanged to previous releases. If set to
5477 off, timer units are unloaded after they elapsed if they
5478 cannot elapse again. This is particularly useful for
5479 transient timer units, which shall not stay around longer
5480 than until they first elapse.
5482 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
5483 default now (the kernel default is 16). This is beneficial
5484 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
5485 allows substantially larger numbers of queued
5486 datagrams. This should increase the capability of systemd to
5487 parallelize boot-up, as logging and sd_notify() are unlikely
5488 to stall execution anymore. If you need to change the value
5489 from the new defaults, use the usual sysctl.d/ snippets.
5491 * The compression framing format used by the journal or
5492 coredump processing has changed to be in line with what the
5493 official LZ4 tools generate. LZ4 compression support in
5494 systemd was considered unsupported previously, as the format
5495 was not compatible with the normal tools. With this release
5496 this has changed now, and it is hence safe for downstream
5497 distributions to turn it on. While not compressing as well
5498 as the XZ, LZ4 is substantially faster, which makes
5499 it a good default choice for the compression logic in the
5500 journal and in coredump handling.
5502 * Any reference to /etc/mtab has been dropped from
5503 systemd. The file has been obsolete since a while, but
5504 systemd refused to work on systems where it was incorrectly
5505 set up (it should be a symlink or non-existent). Please make
5506 sure to update to util-linux 2.27.1 or newer in conjunction
5507 with this systemd release, which also drops any reference to
5508 /etc/mtab. If you maintain a distribution make sure that no
5509 software you package still references it, as this is a
5510 likely source of bugs. There's also a glibc bug pending,
5511 asking for removal of any reference to this obsolete file:
5513 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
5515 Note that only util-linux versions built with
5516 --enable-libmount-force-mountinfo are supported.
5518 * Support for the ".snapshot" unit type has been removed. This
5519 feature turned out to be little useful and little used, and
5520 has now been removed from the core and from systemctl.
5522 * The dependency types RequiresOverridable= and
5523 RequisiteOverridable= have been removed from systemd. They
5524 have been used only very sparingly to our knowledge and
5525 other options that provide a similar effect (such as
5526 systemctl --mode=ignore-dependencies) are much more useful
5527 and commonly used. Moreover, they were only half-way
5528 implemented as the option to control behaviour regarding
5529 these dependencies was never added to systemctl. By removing
5530 these dependency types the execution engine becomes a bit
5531 simpler. Unit files that use these dependencies should be
5532 changed to use the non-Overridable dependency types
5533 instead. In fact, when parsing unit files with these
5534 options, that's what systemd will automatically convert them
5535 too, but it will also warn, asking users to fix the unit
5536 files accordingly. Removal of these dependency types should
5537 only affect a negligible number of unit files in the wild.
5539 * Behaviour of networkd's IPForward= option changed
5540 (again). It will no longer maintain a per-interface setting,
5541 but propagate one way from interfaces where this is enabled
5542 to the global kernel setting. The global setting will be
5543 enabled when requested by a network that is set up, but
5544 never be disabled again. This change was made to make sure
5545 IPv4 and IPv6 behaviour regarding packet forwarding is
5546 similar (as the Linux IPv6 stack does not support
5547 per-interface control of this setting) and to minimize
5550 * In unit files the behaviour of %u, %U, %h, %s has
5551 changed. These specifiers will now unconditionally resolve
5552 to the various user database fields of the user that the
5553 systemd instance is running as, instead of the user
5554 configured in the specific unit via User=. Note that this
5555 effectively doesn't change much, as resolving of these
5556 specifiers was already turned off in the --system instance
5557 of systemd, as we cannot do NSS lookups from PID 1. In the
5558 --user instance of systemd these specifiers where correctly
5559 resolved, but hardly made any sense, since the user instance
5560 lacks privileges to do user switches anyway, and User= is
5561 hence useless. Moreover, even in the --user instance of
5562 systemd behaviour was awkward as it would only take settings
5563 from User= assignment placed before the specifier into
5564 account. In order to unify and simplify the logic around
5565 this the specifiers will now always resolve to the
5566 credentials of the user invoking the manager (which in case
5567 of PID 1 is the root user).
5569 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
5570 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
5571 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
5572 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
5573 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
5574 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
5575 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
5576 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
5577 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
5578 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
5579 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
5580 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
5581 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
5582 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
5585 — Berlin, 2015-11-18
5589 * systemd now depends on util-linux v2.27. More specifically,
5590 the newly added mount monitor feature in libmount now
5591 replaces systemd's former own implementation.
5593 * libmount mandates /etc/mtab not to be regular file, and
5594 systemd now enforces this condition at early boot.
5595 /etc/mtab has been deprecated and warned about for a very
5596 long time, so systems running systemd should already have
5597 stopped having this file around as anything else than a
5598 symlink to /proc/self/mounts.
5600 * Support for the "pids" cgroup controller has been added. It
5601 allows accounting the number of tasks in a cgroup and
5602 enforcing limits on it. This adds two new setting
5603 TasksAccounting= and TasksMax= to each unit, as well as a
5604 global option DefaultTasksAccounting=.
5606 * Support for the "net_cls" cgroup controller has been added.
5607 It allows assigning a net class ID to each task in the
5608 cgroup, which can then be used in firewall rules and traffic
5609 shaping configurations. Note that the kernel netfilter net
5610 class code does not currently work reliably for ingress
5611 packets on unestablished sockets.
5613 This adds a new config directive called NetClass= to CGroup
5614 enabled units. Allowed values are positive numbers for fixed
5615 assignments and "auto" for picking a free value
5618 * 'systemctl is-system-running' now returns 'offline' if the
5619 system is not booted with systemd. This command can now be
5620 used as a substitute for 'systemd-notify --booted'.
5622 * Watchdog timeouts have been increased to 3 minutes for all
5623 in-tree service files. Apparently, disk IO issues are more
5624 frequent than we hoped, and user reported >1 minute waiting
5627 * 'machine-id-commit' functionality has been merged into
5628 'machine-id-setup --commit'. The separate binary has been
5631 * The WorkingDirectory= directive in unit files may now be set
5632 to the special value '~'. In this case, the working
5633 directory is set to the home directory of the user
5634 configured in User=.
5636 * "machinectl shell" will now open the shell in the home
5637 directory of the selected user by default.
5639 * The CrashChVT= configuration file setting is renamed to
5640 CrashChangeVT=, following our usual logic of not
5641 abbreviating unnecessarily. The old directive is still
5642 supported for compat reasons. Also, this directive now takes
5643 an integer value between 1 and 63, or a boolean value. The
5644 formerly supported '-1' value for disabling stays around for
5647 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
5648 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
5649 RootDirectory= properties can now be set for transient
5652 * The systemd-analyze tool gained a new "set-log-target" verb
5653 to change the logging target the system manager logs to
5654 dynamically during runtime. This is similar to how
5655 "systemd-analyze set-log-level" already changes the log
5658 * In nspawn /sys is now mounted as tmpfs, with only a selected
5659 set of subdirectories mounted in from the real sysfs. This
5660 enhances security slightly, and is useful for ensuring user
5661 namespaces work correctly.
5663 * Support for USB FunctionFS activation has been added. This
5664 allows implementation of USB gadget services that are
5665 activated as soon as they are requested, so that they don't
5666 have to run continuously, similar to classic socket
5669 * The "systemctl exit" command now optionally takes an
5670 additional parameter that sets the exit code to return from
5671 the systemd manager when exiting. This is only relevant when
5672 running the systemd user instance, or when running the
5673 system instance in a container.
5675 * sd-bus gained the new API calls sd_bus_path_encode_many()
5676 and sd_bus_path_decode_many() that allow easy encoding and
5677 decoding of multiple identifier strings inside a D-Bus
5678 object path. Another new call sd_bus_default_flush_close()
5679 has been added to flush and close per-thread default
5682 * systemd-cgtop gained support for a -M/--machine= switch to
5683 show the control groups within a certain container only.
5685 * "systemctl kill" gained support for an optional --fail
5686 switch. If specified the requested operation will fail of no
5687 processes have been killed, because the unit had no
5688 processes attached, or similar.
5690 * A new systemd.crash_reboot=1 kernel command line option has
5691 been added that triggers a reboot after crashing. This can
5692 also be set through CrashReboot= in systemd.conf.
5694 * The RuntimeDirectory= setting now understands unit
5695 specifiers like %i or %f.
5697 * A new (still internal) library API sd-ipv4acd has been added,
5698 that implements address conflict detection for IPv4. It's
5699 based on code from sd-ipv4ll, and will be useful for
5700 detecting DHCP address conflicts.
5702 * File descriptors passed during socket activation may now be
5703 named. A new API sd_listen_fds_with_names() is added to
5704 access the names. The default names may be overridden,
5705 either in the .socket file using the FileDescriptorName=
5706 parameter, or by passing FDNAME= when storing the file
5707 descriptors using sd_notify().
5709 * systemd-networkd gained support for:
5711 - Setting the IPv6 Router Advertisement settings via
5712 IPv6AcceptRouterAdvertisements= in .network files.
5714 - Configuring the HelloTimeSec=, MaxAgeSec= and
5715 ForwardDelaySec= bridge parameters in .netdev files.
5717 - Configuring PreferredSource= for static routes in
5720 * The "ask-password" framework used to query for LUKS harddisk
5721 passwords or SSL passwords during boot gained support for
5722 caching passwords in the kernel keyring, if it is
5723 available. This makes sure that the user only has to type in
5724 a passphrase once if there are multiple objects to unlock
5725 with the same one. Previously, such password caching was
5726 available only when Plymouth was used; this moves the
5727 caching logic into the systemd codebase itself. The
5728 "systemd-ask-password" utility gained a new --keyname=
5729 switch to control which kernel keyring key to use for
5730 caching a password in. This functionality is also useful for
5731 enabling display managers such as gdm to automatically
5732 unlock the user's GNOME keyring if its passphrase, the
5733 user's password and the harddisk password are the same, if
5734 gdm-autologin is used.
5736 * When downloading tar or raw images using "machinectl
5737 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
5738 file is now also downloaded, if it is available and stored
5739 next to the image file.
5741 * Units of type ".socket" gained a new boolean setting
5742 Writable= which is only useful in conjunction with
5743 ListenSpecial=. If true, enables opening the specified
5744 special file in O_RDWR mode rather than O_RDONLY mode.
5746 * systemd-rfkill has been reworked to become a singleton
5747 service that is activated through /dev/rfkill on each rfkill
5748 state change and saves the settings to disk. This way,
5749 systemd-rfkill is now compatible with devices that exist
5750 only intermittendly, and even restores state if the previous
5751 system shutdown was abrupt rather than clean.
5753 * The journal daemon gained support for vacuuming old journal
5754 files controlled by the number of files that shall remain,
5755 in addition to the already existing control by size and by
5756 date. This is useful as journal interleaving performance
5757 degrades with too many separate journal files, and allows
5758 putting an effective limit on them. The new setting defaults
5759 to 100, but this may be changed by setting SystemMaxFiles=
5760 and RuntimeMaxFiles= in journald.conf. Also, the
5761 "journalctl" tool gained the new --vacuum-files= switch to
5762 manually vacuum journal files to leave only the specified
5763 number of files in place.
5765 * udev will now create /dev/disk/by-path links for ATA devices
5766 on kernels where that is supported.
5768 * Galician, Serbian, Turkish and Korean translations were added.
5770 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
5771 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
5772 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
5773 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
5774 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
5775 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
5776 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
5777 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
5778 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
5779 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
5780 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
5781 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
5782 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
5783 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
5784 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
5785 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
5786 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
5787 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
5789 — Berlin, 2015-10-07
5793 * The DHCP implementation of systemd-networkd gained a set of
5796 - The DHCP server now supports emitting DNS and NTP
5797 information. It may be enabled and configured via
5798 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
5799 and NTP information is enabled, but no servers are
5800 configured, the corresponding uplink information (if there
5801 is any) is propagated.
5803 - Server and client now support transmission and reception
5804 of timezone information. It can be configured via the
5805 newly introduced network options UseTimezone=,
5806 EmitTimezone=, and Timezone=. Transmission of timezone
5807 information is enabled between host and containers by
5808 default now: the container will change its local timezone
5809 to what the host has set.
5811 - Lease timeouts can now be configured via
5812 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
5814 - The DHCP server improved on the stability of
5815 leases. Clients are more likely to get the same lease
5816 information back, even if the server loses state.
5818 - The DHCP server supports two new configuration options to
5819 control the lease address pool metrics, PoolOffset= and
5822 * The encapsulation limit of tunnels in systemd-networkd may
5823 now be configured via 'EncapsulationLimit='. It allows
5824 modifying the maximum additional levels of encapsulation
5825 that are permitted to be prepended to a packet.
5827 * systemd now supports the concept of user buses replacing
5828 session buses, if used with dbus-1.10 (and enabled via dbus
5829 --enable-user-session). It previously only supported this on
5830 kdbus-enabled systems, and this release expands this to
5831 'dbus-daemon' systems.
5833 * systemd-networkd now supports predictable interface names
5836 * systemd now optionally supports the new Linux kernel
5837 "unified" control group hierarchy. If enabled via the kernel
5838 command-line option 'systemd.unified_cgroup_hierarchy=1',
5839 systemd will try to mount the unified cgroup hierarchy
5840 directly on /sys/fs/cgroup. If not enabled, or not
5841 available, systemd will fall back to the legacy cgroup
5842 hierarchy setup, as before. Host system and containers can
5843 mix and match legacy and unified hierarchies as they
5844 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
5845 environment variable to individually select the hierarchy to
5846 use for executed containers. By default, nspawn will use the
5847 unified hierarchy for the containers if the host uses the
5848 unified hierarchy, and the legacy hierarchy otherwise.
5849 Please note that at this point the unified hierarchy is an
5850 experimental kernel feature and is likely to change in one
5851 of the next kernel releases. Therefore, it should not be
5852 enabled by default in downstream distributions yet. The
5853 minimum required kernel version for the unified hierarchy to
5854 work is 4.2. Note that when the unified hierarchy is used
5855 for the first time delegated access to controllers is
5856 safe. Because of this systemd-nspawn containers will get
5857 access to controllers now, as will systemd user
5858 sessions. This means containers and user sessions may now
5859 manage their own resources, partitioning up what the system
5862 * A new special scope unit "init.scope" has been introduced
5863 that encapsulates PID 1 of the system. It may be used to
5864 determine resource usage and enforce resource limits on PID
5865 1 itself. PID 1 hence moved out of the root of the control
5868 * The cgtop tool gained support for filtering out kernel
5869 threads when counting tasks in a control group. Also, the
5870 count of processes is now recursively summed up by
5871 default. Two options -k and --recursive= have been added to
5872 revert to old behaviour. The tool has also been updated to
5873 work correctly in containers now.
5875 * systemd-nspawn's --bind= and --bind-ro= options have been
5876 extended to allow creation of non-recursive bind mounts.
5878 * libsystemd gained two new calls sd_pid_get_cgroup() and
5879 sd_peer_get_cgroup() which return the control group path of
5880 a process or peer of a connected AF_UNIX socket. This
5881 function call is particularly useful when implementing
5882 delegated subtrees support in the control group hierarchy.
5884 * The "sd-event" event loop API of libsystemd now supports
5885 correct dequeuing of real-time signals, without losing
5888 * When systemd requests a polkit decision when managing units it
5889 will now add additional fields to the request, including unit
5890 name and desired operation. This enables more powerful polkit
5891 policies, that make decisions depending on these parameters.
5893 * nspawn learnt support for .nspawn settings files, that may
5894 accompany the image files or directories of containers, and
5895 may contain additional settings for the container. This is
5896 an alternative to configuring container parameters via the
5897 nspawn command line.
5899 Contributions from: Cristian Rodríguez, Daniel Mack, David
5900 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
5901 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
5902 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
5903 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
5904 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
5905 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
5906 Andersen, Tom Gundersen, Torstein Husebø
5908 — Berlin, 2015-09-08
5912 * machinectl gained a new verb 'shell' which opens a fresh
5913 shell on the target container or the host. It is similar to
5914 the existing 'login' command of machinectl, but spawns the
5915 shell directly without prompting for username or
5916 password. The pseudo machine '.host' now refers to the local
5917 host and is used by default. Hence, 'machinectl shell' can
5918 be used as replacement for 'su -' which spawns a session as
5919 a fresh systemd unit in a way that is fully isolated from
5920 the originating session.
5922 * systemd-networkd learned to cope with private-zone DHCP
5923 options and allows other programs to query the values.
5925 * SELinux access control when enabling/disabling units is no
5926 longer enforced with this release. The previous implementation
5927 was incorrect, and a new corrected implementation is not yet
5928 available. As unit file operations are still protected via
5929 polkit and D-Bus policy this is not a security problem. Yet,
5930 distributions which care about optimal SELinux support should
5931 probably not stabilize on this release.
5933 * sd-bus gained support for matches of type "arg0has=", that
5934 test for membership of strings in string arrays sent in bus
5937 * systemd-resolved now dumps the contents of its DNS and LLMNR
5938 caches to the logs on reception of the SIGUSR1 signal. This
5939 is useful to debug DNS behaviour.
5941 * The coredumpctl tool gained a new --directory= option to
5942 operate on journal files in a specific directory.
5944 * "systemctl reboot" and related commands gained a new
5945 "--message=" option which may be used to set a free-text
5946 wall message when shutting down or rebooting the
5947 system. This message is also logged, which is useful for
5948 figuring out the reason for a reboot or shutdown a
5951 * The "systemd-resolve-host" tool's -i switch now takes
5952 network interface numbers as alternative to interface names.
5954 * A new unit file setting for services has been introduced:
5955 UtmpMode= allows configuration of how precisely systemd
5956 handles utmp and wtmp entries for the service if this is
5957 enabled. This allows writing services that appear similar to
5958 user sessions in the output of the "w", "who", "last" and
5961 * systemd-resolved will now locally synthesize DNS resource
5962 records for the "localhost" and "gateway" domains as well as
5963 the local hostname. This should ensure that clients querying
5964 RRs via resolved will get similar results as those going via
5965 NSS, if nss-myhostname is enabled.
5967 Contributions from: Alastair Hughes, Alex Crawford, Daniel
5968 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
5969 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
5970 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
5971 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
5972 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
5973 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
5974 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
5975 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
5976 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
5977 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
5978 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
5980 — Berlin, 2015-08-27
5984 * The systemd-efi-boot-generator functionality was merged into
5985 systemd-gpt-auto-generator.
5987 * systemd-networkd now supports Group Policy for vxlan
5988 devices. It can be enabled via the new boolean configuration
5989 option called 'GroupPolicyExtension='.
5991 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
5992 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
5993 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
5995 — Berlin, 2015-07-31
5999 * The python-systemd code has been removed from the systemd repository.
6000 A new repository has been created which accommodates the code from
6001 now on, and we kindly ask distributions to create a separate package
6002 for this: https://github.com/systemd/python-systemd
6004 * The systemd daemon will now reload its main configuration
6005 (/etc/systemd/system.conf) on daemon-reload.
6007 * sd-dhcp now exposes vendor specific extensions via
6008 sd_dhcp_lease_get_vendor_specific().
6010 * systemd-networkd gained a number of new configuration options.
6012 - A new boolean configuration option for TAP devices called
6013 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
6014 device, thus allowing to send and receive GSO packets.
6016 - A new tunnel configuration option called 'CopyDSCP='.
6017 If enabled, the DSCP field of ip6 tunnels is copied into the
6018 decapsulated packet.
6020 - A set of boolean bridge configuration options were added.
6021 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
6022 and 'UnicastFlood=' are now parsed by networkd and applied to the
6023 respective bridge link device via the respective IFLA_BRPORT_*
6026 - A new string configuration option to override the hostname sent
6027 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
6028 is true, networkd will use the configured hostname instead of the
6029 system hostname when sending DHCP requests.
6031 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
6032 networkd will configure the IPv6 flow-label of the tunnel device
6033 according to RFC2460.
6035 - The 'macvtap' virtual network devices are now supported, similar to
6036 the already supported 'macvlan' devices.
6038 * systemd-resolved now implements RFC5452 to improve resilience against
6039 cache poisoning. Additionally, source port randomization is enabled
6040 by default to further protect against DNS spoofing attacks.
6042 * nss-mymachines now supports translating UIDs and GIDs of running
6043 containers with user-namespaces enabled. If a container 'foo'
6044 translates a host uid 'UID' to the container uid 'TUID', then
6045 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
6046 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
6047 mapped as 'vg-foo-TGID'.
6049 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
6050 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
6051 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
6052 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
6053 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
6054 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
6055 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
6056 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
6057 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
6058 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
6060 — Berlin, 2015-07-29
6064 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
6065 There are no known issues with current sysfs, and udev does not need
6066 or should be used to work around such bugs.
6068 * udev does no longer enable USB HID power management. Several reports
6069 indicate, that some devices cannot handle that setting.
6071 * The udev accelerometer helper was removed. The functionality
6072 is now fully included in iio-sensor-proxy. But this means,
6073 older iio-sensor-proxy versions will no longer provide
6074 accelerometer/orientation data with this systemd version.
6075 Please upgrade iio-sensor-proxy to version 1.0.
6077 * networkd gained a new configuration option IPv6PrivacyExtensions=
6078 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
6079 for Stateless Address") on selected networks.
6081 * For the sake of fewer build-time dependencies and less code in the
6082 main repository, the python bindings are about to be removed in the
6083 next release. A new repository has been created which accommodates
6084 the code from now on, and we kindly ask distributions to create a
6085 separate package for this. The removal will take place in v223.
6087 https://github.com/systemd/python-systemd
6089 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
6090 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
6091 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
6092 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
6093 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
6094 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
6095 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
6096 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
6097 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
6098 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
6100 — Berlin, 2015-07-07
6104 * The sd-bus.h and sd-event.h APIs have now been declared
6105 stable and have been added to the official interface of
6106 libsystemd.so. sd-bus implements an alternative D-Bus client
6107 library, that is relatively easy to use, very efficient and
6108 supports both classic D-Bus as well as kdbus as transport
6109 backend. sd-event is a generic event loop abstraction that
6110 is built around Linux epoll, but adds features such as event
6111 prioritization or efficient timer handling. Both APIs are good
6112 choices for C programs looking for a bus and/or event loop
6113 implementation that is minimal and does not have to be
6114 portable to other kernels.
6116 * kdbus support is no longer compile-time optional. It is now
6117 always built-in. However, it can still be disabled at
6118 runtime using the kdbus=0 kernel command line setting, and
6119 that setting may be changed to default to off, by specifying
6120 --disable-kdbus at build-time. Note though that the kernel
6121 command line setting has no effect if the kdbus.ko kernel
6122 module is not installed, in which case kdbus is (obviously)
6123 also disabled. We encourage all downstream distributions to
6124 begin testing kdbus by adding it to the kernel images in the
6125 development distributions, and leaving kdbus support in
6128 * The minimal required util-linux version has been bumped to
6131 * Support for chkconfig (--enable-chkconfig) was removed in
6132 favor of calling an abstraction tool
6133 /lib/systemd/systemd-sysv-install. This needs to be
6134 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
6135 in README for details.
6137 * If there's a systemd unit and a SysV init script for the
6138 same service name, and the user executes "systemctl enable"
6139 for it (or a related call), then this will now enable both
6140 (or execute the related operation on both), not just the
6143 * The libudev API documentation has been converted from gtkdoc
6146 * gudev has been removed from the systemd tree, it is now an
6149 * The systemd-cgtop tool learnt a new --raw switch to generate
6150 "raw" (machine parsable) output.
6152 * networkd's IPForwarding= .network file setting learnt the
6153 new setting "kernel", which ensures that networkd does not
6154 change the IP forwarding sysctl from the default kernel
6157 * The systemd-logind bus API now exposes a new boolean
6158 property "Docked" that reports whether logind considers the
6159 system "docked", i.e. connected to a docking station or not.
6161 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
6162 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
6163 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
6164 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
6165 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
6166 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
6167 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
6168 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
6169 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
6170 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
6171 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
6172 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
6173 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
6174 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
6175 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
6176 Fink, Zbigniew Jędrzejewski-Szmek
6178 — Berlin, 2015-06-19
6182 * The gudev library has been extracted into a separate repository
6183 available at: https://git.gnome.org/browse/libgudev/
6184 It is now managed as part of the Gnome project. Distributions
6185 are recommended to pass --disable-gudev to systemd and use
6186 gudev from the Gnome project instead. gudev is still included
6187 in systemd, for now. It will be removed soon, though. Please
6188 also see the announcement-thread on systemd-devel:
6189 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
6191 * systemd now exposes a CPUUsageNSec= property for each
6192 service unit on the bus, that contains the overall consumed
6193 CPU time of a service (the sum of what each process of the
6194 service consumed). This value is only available if
6195 CPUAccounting= is turned on for a service, and is then shown
6196 in the "systemctl status" output.
6198 * Support for configuring alternative mappings of the old SysV
6199 runlevels to systemd targets has been removed. They are now
6200 hardcoded in a way that runlevels 2, 3, 4 all map to
6201 multi-user.target and 5 to graphical.target (which
6202 previously was already the default behaviour).
6204 * The auto-mounter logic gained support for mount point
6205 expiry, using a new TimeoutIdleSec= setting in .automount
6206 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
6208 * The EFI System Partition (ESP) as mounted to /boot by
6209 systemd-efi-boot-generator will now be unmounted
6210 automatically after 2 minutes of not being used. This should
6211 minimize the risk of ESP corruptions.
6213 * New /etc/fstab options x-systemd.requires= and
6214 x-systemd.requires-mounts-for= are now supported to express
6215 additional dependencies for mounts. This is useful for
6216 journalling file systems that support external journal
6217 devices or overlay file systems that require underlying file
6218 systems to be mounted.
6220 * systemd does not support direct live-upgrades (via systemctl
6221 daemon-reexec) from versions older than v44 anymore. As no
6222 distribution we are aware of shipped such old versions in a
6223 stable release this should not be problematic.
6225 * When systemd forks off a new per-connection service instance
6226 it will now set the $REMOTE_ADDR environment variable to the
6227 remote IP address, and $REMOTE_PORT environment variable to
6228 the remote IP port. This behaviour is similar to the
6229 corresponding environment variables defined by CGI.
6231 * systemd-networkd gained support for uplink failure
6232 detection. The BindCarrier= option allows binding interface
6233 configuration dynamically to the link sense of other
6234 interfaces. This is useful to achieve behaviour like in
6237 * systemd-networkd gained support for configuring the DHCP
6238 client identifier to use when requesting leases.
6240 * systemd-networkd now has a per-network UseNTP= option to
6241 configure whether NTP server information acquired via DHCP
6242 is passed on to services like systemd-timesyncd.
6244 * systemd-networkd gained support for vti6 tunnels.
6246 * Note that systemd-networkd manages the sysctl variable
6247 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
6248 it is configured for since v219. The variable controls IP
6249 forwarding, and is a per-interface alternative to the global
6250 /proc/sys/net/ipv[46]/ip_forward. This setting is
6251 configurable in the IPForward= option, which defaults to
6252 "no". This means if networkd is used for an interface it is
6253 no longer sufficient to set the global sysctl option to turn
6254 on IP forwarding! Instead, the .network file option
6255 IPForward= needs to be turned on! Note that the
6256 implementation of this behaviour was broken in v219 and has
6259 * Many bonding and vxlan options are now configurable in
6262 * systemd-nspawn gained a new --property= setting to set unit
6263 properties for the container scope. This is useful for
6264 setting resource parameters (e.g. "CPUShares=500") on
6265 containers started from the command line.
6267 * systemd-nspawn gained a new --private-users= switch to make
6268 use of user namespacing available on recent Linux kernels.
6270 * systemd-nspawn may now be called as part of a shell pipeline
6271 in which case the pipes used for stdin and stdout are passed
6272 directly to the process invoked in the container, without
6273 indirection via a pseudo tty.
6275 * systemd-nspawn gained a new switch to control the UNIX
6276 signal to use when killing the init process of the container
6279 * systemd-nspawn gained a new --overlay= switch for mounting
6280 overlay file systems into the container using the new kernel
6283 * When a container image is imported via systemd-importd and
6284 the host file system is not btrfs, a loopback block device
6285 file is created in /var/lib/machines.raw with a btrfs file
6286 system inside. It is then mounted to /var/lib/machines to
6287 enable btrfs features for container management. The loopback
6288 file and btrfs file system is grown as needed when container
6289 images are imported via systemd-importd.
6291 * systemd-machined/systemd-importd gained support for btrfs
6292 quota, to enforce container disk space limits on disk. This
6293 is exposed in "machinectl set-limit".
6295 * systemd-importd now can import containers from local .tar,
6296 .raw and .qcow2 images, and export them to .tar and .raw. It
6297 can also import dkr v2 images now from the network (on top
6300 * systemd-importd gained support for verifying downloaded
6301 images with gpg2 (previously only gpg1 was supported).
6303 * systemd-machined, systemd-logind, systemd: most bus calls are
6304 now accessible to unprivileged processes via polkit. Also,
6305 systemd-logind will now allow users to kill their own sessions
6306 without further privileges or authorization.
6308 * systemd-shutdownd has been removed. This service was
6309 previously responsible for implementing scheduled shutdowns
6310 as exposed in /usr/bin/shutdown's time parameter. This
6311 functionality has now been moved into systemd-logind and is
6312 accessible via a bus interface.
6314 * "systemctl reboot" gained a new switch --firmware-setup that
6315 can be used to reboot into the EFI firmware setup, if that
6316 is available. systemd-logind now exposes an API on the bus
6317 to trigger such reboots, in case graphical desktop UIs want
6318 to cover this functionality.
6320 * "systemctl enable", "systemctl disable" and "systemctl mask"
6321 now support a new "--now" switch. If specified the units
6322 that are enabled will also be started, and the ones
6323 disabled/masked also stopped.
6325 * The Gummiboot EFI boot loader tool has been merged into
6326 systemd, and renamed to "systemd-boot". The bootctl tool has been
6327 updated to support systemd-boot.
6329 * An EFI kernel stub has been added that may be used to create
6330 kernel EFI binaries that contain not only the actual kernel,
6331 but also an initrd, boot splash, command line and OS release
6332 information. This combined binary can then be signed as a
6333 single image, so that the firmware can verify it all in one
6334 step. systemd-boot has special support for EFI binaries created
6335 like this and can extract OS release information from them
6336 and show them in the boot menu. This functionality is useful
6337 to implement cryptographically verified boot schemes.
6339 * Optional support has been added to systemd-fsck to pass
6340 fsck's progress report to an AF_UNIX socket in the file
6343 * udev will no longer create device symlinks for all block devices by
6344 default. A deny list for excluding special block devices from this
6345 logic has been turned into a allow list that requires picking block
6346 devices explicitly that require device symlinks.
6348 * A new (currently still internal) API sd-device.h has been
6349 added to libsystemd. This modernized API is supposed to
6350 replace libudev eventually. In fact, already much of libudev
6351 is now just a wrapper around sd-device.h.
6353 * A new hwdb database for storing metadata about pointing
6354 stick devices has been added.
6356 * systemd-tmpfiles gained support for setting file attributes
6357 similar to the "chattr" tool with new 'h' and 'H' lines.
6359 * systemd-journald will no longer unconditionally set the
6360 btrfs NOCOW flag on new journal files. This is instead done
6361 with tmpfiles snippet using the new 'h' line type. This
6362 allows easy disabling of this logic, by masking the
6363 journal-nocow.conf tmpfiles file.
6365 * systemd-journald will now translate audit message types to
6366 human readable identifiers when writing them to the
6367 journal. This should improve readability of audit messages.
6369 * The LUKS logic gained support for the offset= and skip=
6370 options in /etc/crypttab, as previously implemented by
6373 * /usr/lib/os-release gained a new optional field VARIANT= for
6374 distributions that support multiple variants (such as a
6375 desktop edition, a server edition, ...)
6377 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
6378 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
6379 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
6380 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
6381 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
6382 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
6383 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
6384 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
6385 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
6386 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
6387 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
6388 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
6389 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
6390 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
6391 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
6392 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
6393 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
6394 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
6395 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
6396 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
6397 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
6398 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
6399 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
6400 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
6401 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
6402 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
6403 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
6405 — Berlin, 2015-05-22
6409 * Introduce a new API "sd-hwdb.h" for querying the hardware
6410 metadata database. With this minimal interface one can query
6411 and enumerate the udev hwdb, decoupled from the old libudev
6412 library. libudev's interface for this is now only a wrapper
6413 around sd-hwdb. A new tool systemd-hwdb has been added to
6414 interface with and update the database.
6416 * When any of systemd's tools copies files (for example due to
6417 tmpfiles' C lines) a btrfs reflink will attempted first,
6418 before bytewise copying is done.
6420 * systemd-nspawn gained a new --ephemeral switch. When
6421 specified a btrfs snapshot is taken of the container's root
6422 directory, and immediately removed when the container
6423 terminates again. Thus, a container can be started whose
6424 changes never alter the container's root directory, and are
6425 lost on container termination. This switch can also be used
6426 for starting a container off the root file system of the
6427 host without affecting the host OS. This switch is only
6428 available on btrfs file systems.
6430 * systemd-nspawn gained a new --template= switch. It takes the
6431 path to a container tree to use as template for the tree
6432 specified via --directory=, should that directory be
6433 missing. This allows instantiating containers dynamically,
6434 on first run. This switch is only available on btrfs file
6437 * When a .mount unit refers to a mount point on which multiple
6438 mounts are stacked, and the .mount unit is stopped all of
6439 the stacked mount points will now be unmounted until no
6440 mount point remains.
6442 * systemd now has an explicit notion of supported and
6443 unsupported unit types. Jobs enqueued for unsupported unit
6444 types will now fail with an "unsupported" error code. More
6445 specifically .swap, .automount and .device units are not
6446 supported in containers, .busname units are not supported on
6447 non-kdbus systems. .swap and .automount are also not
6448 supported if their respective kernel compile time options
6451 * machinectl gained support for two new "copy-from" and
6452 "copy-to" commands for copying files from a running
6453 container to the host or vice versa.
6455 * machinectl gained support for a new "bind" command to bind
6456 mount host directories into local containers. This is
6457 currently only supported for nspawn containers.
6459 * networkd gained support for configuring bridge forwarding
6460 database entries (fdb) from .network files.
6462 * A new tiny daemon "systemd-importd" has been added that can
6463 download container images in tar, raw, qcow2 or dkr formats,
6464 and make them available locally in /var/lib/machines, so
6465 that they can run as nspawn containers. The daemon can GPG
6466 verify the downloads (not supported for dkr, since it has no
6467 provisions for verifying downloads). It will transparently
6468 decompress bz2, xz, gzip compressed downloads if necessary,
6469 and restore sparse files on disk. The daemon uses privilege
6470 separation to ensure the actual download logic runs with
6471 fewer privileges than the daemon itself. machinectl has
6472 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
6473 make the functionality of importd available to the
6474 user. With this in place the Fedora and Ubuntu "Cloud"
6475 images can be downloaded and booted as containers unmodified
6476 (the Fedora images lack the appropriate GPG signature files
6477 currently, so they cannot be verified, but this will change
6478 soon, hopefully). Note that downloading images is currently
6479 only fully supported on btrfs.
6481 * machinectl is now able to list container images found in
6482 /var/lib/machines, along with some metadata about sizes of
6483 disk and similar. If the directory is located on btrfs and
6484 quota is enabled, this includes quota display. A new command
6485 "image-status" has been added that shows additional
6486 information about images.
6488 * machinectl is now able to clone container images
6489 efficiently, if the underlying file system (btrfs) supports
6490 it, with the new "machinectl clone" command. It also
6491 gained commands for renaming and removing images, as well as
6492 marking them read-only or read-write (supported also on
6493 legacy file systems).
6495 * networkd gained support for collecting LLDP network
6496 announcements, from hardware that supports this. This is
6497 shown in networkctl output.
6499 * systemd-run gained support for a new -t (--pty) switch for
6500 invoking a binary on a pty whose input and output is
6501 connected to the invoking terminal. This allows executing
6502 processes as system services while interactively
6503 communicating with them via the terminal. Most interestingly
6504 this is supported across container boundaries. Invoking
6505 "systemd-run -t /bin/bash" is an alternative to running a
6506 full login session, the difference being that the former
6507 will not register a session, nor go through the PAM session
6510 * tmpfiles gained support for a new "v" line type for creating
6511 btrfs subvolumes. If the underlying file system is a legacy
6512 file system, this automatically degrades to creating a
6513 normal directory. Among others /var/lib/machines is now
6514 created like this at boot, should it be missing.
6516 * The directory /var/lib/containers/ has been deprecated and
6517 been replaced by /var/lib/machines. The term "machines" has
6518 been used in the systemd context as generic term for both
6519 VMs and containers, and hence appears more appropriate for
6520 this, as the directory can also contain raw images bootable
6523 * systemd-nspawn when invoked with -M but without --directory=
6524 or --image= is now capable of searching for the container
6525 root directory, subvolume or disk image automatically, in
6526 /var/lib/machines. systemd-nspawn@.service has been updated
6527 to make use of this, thus allowing it to be used for raw
6530 * A new machines.target unit has been introduced that is
6531 supposed to group all containers/VMs invoked as services on
6532 the system. systemd-nspawn@.service has been updated to
6533 integrate with that.
6535 * machinectl gained a new "start" command, for invoking a
6536 container as a service. "machinectl start foo" is mostly
6537 equivalent to "systemctl start systemd-nspawn@foo.service",
6538 but handles escaping in a nicer way.
6540 * systemd-nspawn will now mount most of the cgroupfs tree
6541 read-only into each container, with the exception of the
6542 container's own subtree in the name=systemd hierarchy.
6544 * journald now sets the special FS_NOCOW file flag for its
6545 journal files. This should improve performance on btrfs, by
6546 avoiding heavy fragmentation when journald's write-pattern
6547 is used on COW file systems. It degrades btrfs' data
6548 integrity guarantees for the files to the same levels as for
6549 ext3/ext4 however. This should be OK though as journald does
6550 its own data integrity checks and all its objects are
6551 checksummed on disk. Also, journald should handle btrfs disk
6552 full events a lot more gracefully now, by processing SIGBUS
6553 errors, and not relying on fallocate() anymore.
6555 * When journald detects that journal files it is writing to
6556 have been deleted it will immediately start new journal
6559 * systemd now provides a way to store file descriptors
6560 per-service in PID 1. This is useful for daemons to ensure
6561 that fds they require are not lost during a daemon
6562 restart. The fds are passed to the daemon on the next
6563 invocation in the same way socket activation fds are
6564 passed. This is now used by journald to ensure that the
6565 various sockets connected to all the system's stdout/stderr
6566 are not lost when journald is restarted. File descriptors
6567 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
6568 an extension to sd_notify(). Note that a limit is enforced
6569 on the number of fds a service can store in PID 1, and it
6570 defaults to 0, so that no fds may be stored, unless this is
6571 explicitly turned on.
6573 * The default TERM variable to use for units connected to a
6574 terminal, when no other value is explicitly is set is now
6575 vt220 rather than vt102. This should be fairly safe still,
6576 but allows PgUp/PgDn work.
6578 * The /etc/crypttab option header= as known from Debian is now
6581 * "loginctl user-status" and "loginctl session-status" will
6582 now show the last 10 lines of log messages of the
6583 user/session following the status output. Similar,
6584 "machinectl status" will show the last 10 log lines
6585 associated with a virtual machine or container
6586 service. (Note that this is usually not the log messages
6587 done in the VM/container itself, but simply what the
6588 container manager logs. For nspawn this includes all console
6591 * "loginctl session-status" without further argument will now
6592 show the status of the session of the caller. Similar,
6593 "lock-session", "unlock-session", "activate",
6594 "enable-linger", "disable-linger" may now be called without
6595 session/user parameter in which case they apply to the
6596 caller's session/user.
6598 * An X11 session scriptlet is now shipped that uploads
6599 $DISPLAY and $XAUTHORITY into the environment of the systemd
6600 --user daemon if a session begins. This should improve
6601 compatibility with X11 enabled applications run as systemd
6604 * Generators are now subject to masking via /etc and /run, the
6605 same way as unit files.
6607 * networkd .network files gained support for configuring
6608 per-link IPv4/IPv6 packet forwarding as well as IPv4
6609 masquerading. This is by default turned on for veth links to
6610 containers, as registered by systemd-nspawn. This means that
6611 nspawn containers run with --network-veth will now get
6612 automatic routed access to the host's networks without any
6613 further configuration or setup, as long as networkd runs on
6616 * systemd-nspawn gained the --port= (-p) switch to expose TCP
6617 or UDP posts of a container on the host. With this in place
6618 it is possible to run containers with private veth links
6619 (--network-veth), and have their functionality exposed on
6620 the host as if their services were running directly on the
6623 * systemd-nspawn's --network-veth switch now gained a short
6624 version "-n", since with the changes above it is now truly
6625 useful out-of-the-box. The systemd-nspawn@.service has been
6626 updated to make use of it too by default.
6628 * systemd-nspawn will now maintain a per-image R/W lock, to
6629 ensure that the same image is not started more than once
6630 writable. (It's OK to run an image multiple times
6631 simultaneously in read-only mode.)
6633 * systemd-nspawn's --image= option is now capable of
6634 dissecting and booting MBR and GPT disk images that contain
6635 only a single active Linux partition. Previously it
6636 supported only GPT disk images with proper GPT type
6637 IDs. This allows running cloud images from major
6638 distributions directly with systemd-nspawn, without
6641 * In addition to collecting mouse dpi data in the udev
6642 hardware database, there's now support for collecting angle
6643 information for mouse scroll wheels. The database is
6644 supposed to guarantee similar scrolling behavior on mice
6645 that it knows about. There's also support for collecting
6646 information about Touchpad types.
6648 * udev's input_id built-in will now also collect touch screen
6649 dimension data and attach it to probed devices.
6651 * /etc/os-release gained support for a Distribution Privacy
6654 * networkd gained support for creating "ipvlan", "gretap",
6655 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
6657 * systemd-tmpfiles gained support for "a" lines for setting
6660 * systemd-nspawn will now mount /tmp in the container to
6661 tmpfs, automatically.
6663 * systemd now exposes the memory.usage_in_bytes cgroup
6664 attribute and shows it for each service in the "systemctl
6665 status" output, if available.
6667 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
6668 immediate reboot is triggered. This useful if shutdown is
6669 hung and is unable to complete, to expedite the
6670 operation. Note that this kind of reboot will still unmount
6671 all file systems, and hence should not result in fsck being
6674 * A .device unit for an optical block device will now be
6675 considered active only when a medium is in the drive. Also,
6676 mount units are now bound to their backing devices thus
6677 triggering automatic unmounting when devices become
6678 unavailable. With this in place systemd will now
6679 automatically unmount left-over mounts when a CD-ROM is
6680 ejected or an USB stick is yanked from the system.
6682 * networkd-wait-online now has support for waiting for
6683 specific interfaces only (with globbing), and for giving up
6684 after a configurable timeout.
6686 * networkd now exits when idle. It will be automatically
6687 restarted as soon as interfaces show up, are removed or
6688 change state. networkd will stay around as long as there is
6689 at least one DHCP state machine or similar around, that keep
6692 * networkd may now configure IPv6 link-local addressing in
6693 addition to IPv4 link-local addressing.
6695 * The IPv6 "token" for use in SLAAC may now be configured for
6696 each .network interface in networkd.
6698 * Routes configured with networkd may now be assigned a scope
6701 * networkd's [Match] sections now support globbing and lists
6702 of multiple space-separated matches per item.
6704 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
6705 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
6706 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
6707 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
6708 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
6709 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
6710 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
6711 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
6712 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
6713 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
6714 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
6715 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
6716 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
6717 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
6718 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
6719 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
6720 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
6721 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
6722 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
6723 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
6724 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
6725 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
6726 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
6727 Hoffmann, Zbigniew Jędrzejewski-Szmek
6729 — Berlin, 2015-02-16
6733 * When querying unit file enablement status (for example via
6734 "systemctl is-enabled"), a new state "indirect" is now known
6735 which indicates that a unit might not be enabled itself, but
6736 another unit listed in its Also= setting might be.
6738 * Similar to the various existing ConditionXYZ= settings for
6739 units, there are now matching AssertXYZ= settings. While
6740 failing conditions cause a unit to be skipped, but its job
6741 to succeed, failing assertions declared like this will cause
6742 a unit start operation and its job to fail.
6744 * hostnamed now knows a new chassis type "embedded".
6746 * systemctl gained a new "edit" command. When used on a unit
6747 file, this allows extending unit files with .d/ drop-in
6748 configuration snippets or editing the full file (after
6749 copying it from /usr/lib to /etc). This will invoke the
6750 user's editor (as configured with $EDITOR), and reload the
6751 modified configuration after editing.
6753 * "systemctl status" now shows the suggested enablement state
6754 for a unit, as declared in the (usually vendor-supplied)
6755 system preset files.
6757 * nss-myhostname will now resolve the single-label hostname
6758 "gateway" to the locally configured default IP routing
6759 gateways, ordered by their metrics. This assigns a stable
6760 name to the used gateways, regardless which ones are
6761 currently configured. Note that the name will only be
6762 resolved after all other name sources (if nss-myhostname is
6763 configured properly) and should hence not negatively impact
6764 systems that use the single-label hostname "gateway" in
6767 * systemd-inhibit now allows filtering by mode when listing
6770 * Scope and service units gained a new "Delegate" boolean
6771 property, which, when set, allows processes running inside the
6772 unit to further partition resources. This is primarily
6773 useful for systemd user instances as well as container
6776 * journald will now pick up audit messages directly from
6777 the kernel, and log them like any other log message. The
6778 audit fields are split up and fully indexed. This means that
6779 journalctl in many ways is now a (nicer!) alternative to
6780 ausearch, the traditional audit client. Note that this
6781 implements only a minimal audit client. If you want the
6782 special audit modes like reboot-on-log-overflow, please use
6783 the traditional auditd instead, which can be used in
6784 parallel to journald.
6786 * The ConditionSecurity= unit file option now understands the
6787 special string "audit" to check whether auditing is
6790 * journalctl gained two new commands --vacuum-size= and
6791 --vacuum-time= to delete old journal files until the
6792 remaining ones take up no more than the specified size on disk,
6793 or are not older than the specified time.
6795 * A new, native PPPoE library has been added to sd-network,
6796 systemd's library of light-weight networking protocols. This
6797 library will be used in a future version of networkd to
6798 enable PPPoE communication without an external pppd daemon.
6800 * The busctl tool now understands a new "capture" verb that
6801 works similar to "monitor", but writes a packet capture
6802 trace to STDOUT that can be redirected to a file which is
6803 compatible with libcap's capture file format. This can then
6804 be loaded in Wireshark and similar tools to inspect bus
6807 * The busctl tool now understands a new "tree" verb that shows
6808 the object trees of a specific service on the bus, or of all
6811 * The busctl tool now understands a new "introspect" verb that
6812 shows all interfaces and members of objects on the bus,
6813 including their signature and values. This is particularly
6814 useful to get more information about bus objects shown by
6815 the new "busctl tree" command.
6817 * The busctl tool now understands new verbs "call",
6818 "set-property" and "get-property" for invoking bus method
6819 calls, setting and getting bus object properties in a
6822 * busctl gained a new --augment-creds= argument that controls
6823 whether the tool shall augment credential information it
6824 gets from the bus with data from /proc, in a possibly
6827 * nspawn's --link-journal= switch gained two new values
6828 "try-guest" and "try-host" that work like "guest" and
6829 "host", but do not fail if the host has no persistent
6830 journalling enabled. -j is now equivalent to
6831 --link-journal=try-guest.
6833 * macvlan network devices created by nspawn will now have
6834 stable MAC addresses.
6836 * A new SmackProcessLabel= unit setting has been added, which
6837 controls the SMACK security label processes forked off by
6838 the respective unit shall use.
6840 * If compiled with --enable-xkbcommon, systemd-localed will
6841 verify x11 keymap settings by compiling the given keymap. It
6842 will spew out warnings if the compilation fails. This
6843 requires libxkbcommon to be installed.
6845 * When a coredump is collected, a larger number of metadata
6846 fields is now collected and included in the journal records
6847 created for it. More specifically, control group membership,
6848 environment variables, memory maps, working directory,
6849 chroot directory, /proc/$PID/status, and a list of open file
6850 descriptors is now stored in the log entry.
6852 * The udev hwdb now contains DPI information for mice. For
6855 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
6857 * All systemd programs that read standalone configuration
6858 files in /etc now also support a corresponding series of
6859 .conf.d configuration directories in /etc/, /run/,
6860 /usr/local/lib/, /usr/lib/, and (if configured with
6861 --enable-split-usr) /lib/. In particular, the following
6862 configuration files now have corresponding configuration
6863 directories: system.conf user.conf, logind.conf,
6864 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
6865 resolved.conf, timesyncd.conf, journal-remote.conf, and
6866 journal-upload.conf. Note that distributions should use the
6867 configuration directories in /usr/lib/; the directories in
6868 /etc/ are reserved for the system administrator.
6870 * systemd-rfkill will no longer take the rfkill device name
6871 into account when storing rfkill state on disk, as the name
6872 might be dynamically assigned and not stable. Instead, the
6873 ID_PATH udev variable combined with the rfkill type (wlan,
6874 bluetooth, ...) is used.
6876 * A new service systemd-machine-id-commit.service has been
6877 added. When used on systems where /etc is read-only during
6878 boot, and /etc/machine-id is not initialized (but an empty
6879 file), this service will copy the temporary machine ID
6880 created as replacement into /etc after the system is fully
6881 booted up. This is useful for systems that are freshly
6882 installed with a non-initialized machine ID, but should get
6883 a fixed machine ID for subsequent boots.
6885 * networkd's .netdev files now provide a large set of
6886 configuration parameters for VXLAN devices. Similarly, the
6887 bridge port cost parameter is now configurable in .network
6888 files. There's also new support for configuring IP source
6889 routing. networkd .link files gained support for a new
6890 OriginalName= match that is useful to match against the
6891 original interface name the kernel assigned. .network files
6892 may include MTU= and MACAddress= fields for altering the MTU
6893 and MAC address while being connected to a specific network
6896 * The LUKS logic gained supported for configuring
6897 UUID-specific key files. There's also new support for naming
6898 LUKS device from the kernel command line, using the new
6899 luks.name= argument.
6901 * Timer units may now be transiently created via the bus API
6902 (this was previously already available for scope and service
6903 units). In addition it is now possible to create multiple
6904 transient units at the same time with a single bus call. The
6905 "systemd-run" tool has been updated to make use of this for
6906 running commands on a specified time, in at(1)-style.
6908 * tmpfiles gained support for "t" lines, for assigning
6909 extended attributes to files. Among other uses this may be
6910 used to assign SMACK labels to files.
6912 Contributions from: Alin Rauta, Alison Chaiken, Andrej
6913 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
6914 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
6915 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
6916 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
6917 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
6918 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
6919 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
6920 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
6921 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
6922 Michael Biebl, Michael Chapman, Michael Marineau, Michal
6923 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
6924 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
6925 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
6926 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
6927 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6928 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
6929 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
6931 — Berlin, 2014-12-10
6935 * journalctl gained the new options -t/--identifier= to match
6936 on the syslog identifier (aka "tag"), as well as --utc to
6937 show log timestamps in the UTC timezone. journalctl now also
6938 accepts -n/--lines=all to disable line capping in a pager.
6940 * journalctl gained a new switch, --flush, that synchronously
6941 flushes logs from /run/log/journal to /var/log/journal if
6942 persistent storage is enabled. systemd-journal-flush.service
6943 now waits until the operation is complete.
6945 * Services can notify the manager before they start a reload
6946 (by sending RELOADING=1) or shutdown (by sending
6947 STOPPING=1). This allows the manager to track and show the
6948 internal state of daemons and closes a race condition when
6949 the process is still running but has closed its D-Bus
6952 * Services with Type=oneshot do not have to have any ExecStart
6955 * User units are now loaded also from
6956 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
6957 /run/systemd/user directory that was already previously
6958 supported, but is under the control of the user.
6960 * Job timeouts (i.e. timeouts on the time a job that is
6961 queued stays in the run queue) can now optionally result in
6962 immediate reboot or power-off actions (JobTimeoutAction= and
6963 JobTimeoutRebootArgument=). This is useful on ".target"
6964 units, to limit the maximum time a target remains
6965 undispatched in the run queue, and to trigger an emergency
6966 operation in such a case. This is now used by default to
6967 turn off the system if boot-up (as defined by everything in
6968 basic.target) hangs and does not complete for at least
6969 15min. Also, if power-off or reboot hang for at least 30min
6970 an immediate power-off/reboot operation is triggered. This
6971 functionality is particularly useful to increase reliability
6972 on embedded devices, but also on laptops which might
6973 accidentally get powered on when carried in a backpack and
6974 whose boot stays stuck in a hard disk encryption passphrase
6977 * systemd-logind can be configured to also handle lid switch
6978 events even when the machine is docked or multiple displays
6979 are attached (HandleLidSwitchDocked= option).
6981 * A helper binary and a service have been added which can be
6982 used to resume from hibernation in the initramfs. A
6983 generator will parse the resume= option on the kernel
6984 command line to trigger resume.
6986 * A user console daemon systemd-consoled has been
6987 added. Currently, it is a preview, and will so far open a
6988 single terminal on each session of the user marked as
6989 Desktop=systemd-console.
6991 * Route metrics can be specified for DHCP routes added by
6994 * The SELinux context of socket-activated services can be set
6995 from the information provided by the networking stack
6996 (SELinuxContextFromNet= option).
6998 * Userspace firmware loading support has been removed and
6999 the minimum supported kernel version is thus bumped to 3.7.
7001 * Timeout for udev workers has been increased from 1 to 3
7002 minutes, but a warning will be printed after 1 minute to
7003 help diagnose kernel modules that take a long time to load.
7005 * Udev rules can now remove tags on devices with TAG-="foobar".
7007 * systemd's readahead implementation has been removed. In many
7008 circumstances it didn't give expected benefits even for
7009 rotational disk drives and was becoming less relevant in the
7010 age of SSDs. As none of the developers has been using
7011 rotating media anymore, and nobody stepped up to actively
7012 maintain this component of systemd it has now been removed.
7014 * Swap units can use Options= to specify discard options.
7015 Discard options specified for swaps in /etc/fstab are now
7018 * Docker containers are now detected as a separate type of
7021 * The Password Agent protocol gained support for queries where
7022 the user input is shown, useful e.g. for user names.
7023 systemd-ask-password gained a new --echo option to turn that
7026 * The default sysctl.d/ snippets will now set:
7028 net.core.default_qdisc = fq_codel
7030 This selects Fair Queuing Controlled Delay as the default
7031 queuing discipline for network interfaces. fq_codel helps
7032 fight the network bufferbloat problem. It is believed to be
7033 a good default with no tuning required for most workloads.
7034 Downstream distributions may override this choice. On 10Gbit
7035 servers that do not do forwarding, "fq" may perform better.
7036 Systems without a good clocksource should use "pfifo_fast".
7038 * If kdbus is enabled during build a new option BusPolicy= is
7039 available for service units, that allows locking all service
7040 processes into a stricter bus policy, in order to limit
7041 access to various bus services, or even hide most of them
7042 from the service's view entirely.
7044 * networkctl will now show the .network and .link file
7045 networkd has applied to a specific interface.
7047 * sd-login gained a new API call sd_session_get_desktop() to
7048 query which desktop environment has been selected for a
7051 * UNIX utmp support is now compile-time optional to support
7052 legacy-free systems.
7054 * systemctl gained two new commands "add-wants" and
7055 "add-requires" for pulling in units from specific targets
7058 * If the word "rescue" is specified on the kernel command line
7059 the system will now boot into rescue mode (aka
7060 rescue.target), which was previously available only by
7061 specifying "1" or "systemd.unit=rescue.target" on the kernel
7062 command line. This new kernel command line option nicely
7063 mirrors the already existing "emergency" kernel command line
7066 * New kernel command line options mount.usr=, mount.usrflags=,
7067 mount.usrfstype= have been added that match root=, rootflags=,
7068 rootfstype= but allow mounting a specific file system to
7071 * The $NOTIFY_SOCKET is now also passed to control processes of
7072 services, not only the main process.
7074 * This version reenables support for fsck's -l switch. This
7075 means at least version v2.25 of util-linux is required for
7076 operation, otherwise dead-locks on device nodes may
7077 occur. Again: you need to update util-linux to at least
7078 v2.25 when updating systemd to v217.
7080 * The "multi-seat-x" tool has been removed from systemd, as
7081 its functionality has been integrated into X servers 1.16,
7082 and the tool is hence redundant. It is recommended to update
7083 display managers invoking this tool to simply invoke X
7084 directly from now on, again.
7086 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
7087 message flag has been added for all of systemd's polkit
7088 authenticated method calls has been added. In particular this
7089 now allows optional interactive authorization via polkit for
7090 many of PID1's privileged operations such as unit file
7091 enabling and disabling.
7093 * "udevadm hwdb --update" learnt a new switch "--usr" for
7094 placing the rebuilt hardware database in /usr instead of
7095 /etc. When used only hardware database entries stored in
7096 /usr will be used, and any user database entries in /etc are
7097 ignored. This functionality is useful for vendors to ship a
7098 pre-built database on systems where local configuration is
7099 unnecessary or unlikely.
7101 * Calendar time specifications in .timer units now also
7102 understand the strings "semi-annually", "quarterly" and
7103 "minutely" as shortcuts (in addition to the preexisting
7104 "annually", "hourly", ...).
7106 * systemd-tmpfiles will now correctly create files in /dev
7107 at boot which are marked for creation only at boot. It is
7108 recommended to always create static device nodes with 'c!'
7109 and 'b!', so that they are created only at boot and not
7110 overwritten at runtime.
7112 * When the watchdog logic is used for a service (WatchdogSec=)
7113 and the watchdog timeout is hit the service will now be
7114 terminated with SIGABRT (instead of just SIGTERM), in order
7115 to make sure a proper coredump and backtrace is
7116 generated. This ensures that hanging services will result in
7117 similar coredump/backtrace behaviour as services that hit a
7120 Contributions from: Andreas Henriksson, Andrei Borzenkov,
7121 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
7122 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
7123 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
7124 Herrmann, David Sommerseth, David Strauss, Emil Renner
7125 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
7126 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
7127 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
7128 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
7129 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
7130 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
7131 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
7132 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
7133 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
7134 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
7135 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
7136 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
7137 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
7138 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
7139 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
7140 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
7143 — Berlin, 2014-10-28
7147 * timedated no longer reads NTP implementation unit names from
7148 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
7149 implementations should add a
7151 Conflicts=systemd-timesyncd.service
7153 to their unit files to take over and replace systemd's NTP
7154 default functionality.
7156 * systemd-sysusers gained a new line type "r" for configuring
7157 which UID/GID ranges to allocate system users/groups
7158 from. Lines of type "u" may now add an additional column
7159 that specifies the home directory for the system user to be
7160 created. Also, systemd-sysusers may now optionally read user
7161 information from STDIN instead of a file. This is useful for
7162 invoking it from RPM preinst scriptlets that need to create
7163 users before the first RPM file is installed since these
7164 files might need to be owned by them. A new
7165 %sysusers_create_inline RPM macro has been introduced to do
7166 just that. systemd-sysusers now updates the shadow files as
7167 well as the user/group databases, which should enhance
7168 compatibility with certain tools like grpck.
7170 * A number of bus APIs of PID 1 now optionally consult polkit to
7171 permit access for otherwise unprivileged clients under certain
7172 conditions. Note that this currently doesn't support
7173 interactive authentication yet, but this is expected to be
7174 added eventually, too.
7176 * /etc/machine-info now has new fields for configuring the
7177 deployment environment of the machine, as well as the
7178 location of the machine. hostnamectl has been updated with
7179 new command to update these fields.
7181 * systemd-timesyncd has been updated to automatically acquire
7182 NTP server information from systemd-networkd, which might
7183 have been discovered via DHCP.
7185 * systemd-resolved now includes a caching DNS stub resolver
7186 and a complete LLMNR name resolution implementation. A new
7187 NSS module "nss-resolve" has been added which can be used
7188 instead of glibc's own "nss-dns" to resolve hostnames via
7189 systemd-resolved. Hostnames, addresses and arbitrary RRs may
7190 be resolved via systemd-resolved D-Bus APIs. In contrast to
7191 the glibc internal resolver systemd-resolved is aware of
7192 multi-homed system, and keeps DNS server and caches separate
7193 and per-interface. Queries are sent simultaneously on all
7194 interfaces that have DNS servers configured, in order to
7195 properly handle VPNs and local LANs which might resolve
7196 separate sets of domain names. systemd-resolved may acquire
7197 DNS server information from systemd-networkd automatically,
7198 which in turn might have discovered them via DHCP. A tool
7199 "systemd-resolve-host" has been added that may be used to
7200 query the DNS logic in resolved. systemd-resolved implements
7201 IDNA and automatically uses IDNA or UTF-8 encoding depending
7202 on whether classic DNS or LLMNR is used as transport. In the
7203 next releases we intend to add a DNSSEC and mDNS/DNS-SD
7204 implementation to systemd-resolved.
7206 * A new NSS module nss-mymachines has been added, that
7207 automatically resolves the names of all local registered
7208 containers to their respective IP addresses.
7210 * A new client tool "networkctl" for systemd-networkd has been
7211 added. It currently is entirely passive and will query
7212 networking configuration from udev, rtnetlink and networkd,
7213 and present it to the user in a very friendly
7214 way. Eventually, we hope to extend it to become a full
7215 control utility for networkd.
7217 * .socket units gained a new DeferAcceptSec= setting that
7218 controls the kernels' TCP_DEFER_ACCEPT sockopt for
7219 TCP. Similarly, support for controlling TCP keep-alive
7220 settings has been added (KeepAliveTimeSec=,
7221 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
7222 turning off Nagle's algorithm on TCP has been added
7225 * logind learned a new session type "web", for use in projects
7226 like Cockpit which register web clients as PAM sessions.
7228 * timer units with at least one OnCalendar= setting will now
7229 be started only after time-sync.target has been
7230 reached. This way they will not elapse before the system
7231 clock has been corrected by a local NTP client or
7232 similar. This is particular useful on RTC-less embedded
7233 machines, that come up with an invalid system clock.
7235 * systemd-nspawn's --network-veth= switch should now result in
7236 stable MAC addresses for both the outer and the inner side
7239 * systemd-nspawn gained a new --volatile= switch for running
7240 container instances with /etc or /var unpopulated.
7242 * The kdbus client code has been updated to use the new Linux
7243 3.17 memfd subsystem instead of the old kdbus-specific one.
7245 * systemd-networkd's DHCP client and server now support
7246 FORCERENEW. There are also new configuration options to
7247 configure the vendor client identifier and broadcast mode
7250 * systemd will no longer inform the kernel about the current
7251 timezone, as this is necessarily incorrect and racy as the
7252 kernel has no understanding of DST and similar
7253 concepts. This hence means FAT timestamps will be always
7254 considered UTC, similar to what Android is already
7255 doing. Also, when the RTC is configured to the local time
7256 (rather than UTC) systemd will never synchronize back to it,
7257 as this might confuse Windows at a later boot.
7259 * systemd-analyze gained a new command "verify" for offline
7260 validation of unit files.
7262 * systemd-networkd gained support for a couple of additional
7263 settings for bonding networking setups. Also, the metric for
7264 statically configured routes may now be configured. For
7265 network interfaces where this is appropriate the peer IP
7266 address may now be configured.
7268 * systemd-networkd's DHCP client will no longer request
7269 broadcasting by default, as this tripped up some networks.
7270 For hardware where broadcast is required the feature should
7271 be switched back on using RequestBroadcast=yes.
7273 * systemd-networkd will now set up IPv4LL addresses (when
7274 enabled) even if DHCP is configured successfully.
7276 * udev will now default to respect network device names given
7277 by the kernel when the kernel indicates that these are
7278 predictable. This behavior can be tweaked by changing
7279 NamePolicy= in the relevant .link file.
7281 * A new library systemd-terminal has been added that
7282 implements full TTY stream parsing and rendering. This
7283 library is supposed to be used later on for implementing a
7284 full userspace VT subsystem, replacing the current kernel
7287 * A new tool systemd-journal-upload has been added to push
7288 journal data to a remote system running
7289 systemd-journal-remote.
7291 * journald will no longer forward all local data to another
7292 running syslog daemon. This change has been made because
7293 rsyslog (which appears to be the most commonly used syslog
7294 implementation these days) no longer makes use of this, and
7295 instead pulls the data out of the journal on its own. Since
7296 forwarding the messages to a non-existent syslog server is
7297 more expensive than we assumed we have now turned this
7298 off. If you run a syslog server that is not a recent rsyslog
7299 version, you have to turn this option on again
7300 (ForwardToSyslog= in journald.conf).
7302 * journald now optionally supports the LZ4 compressor for
7303 larger journal fields. This compressor should perform much
7304 better than XZ which was the previous default.
7306 * machinectl now shows the IP addresses of local containers,
7307 if it knows them, plus the interface name of the container.
7309 * A new tool "systemd-escape" has been added that makes it
7310 easy to escape strings to build unit names and similar.
7312 * sd_notify() messages may now include a new ERRNO= field
7313 which is parsed and collected by systemd and shown among the
7314 "systemctl status" output for a service.
7316 * A new component "systemd-firstboot" has been added that
7317 queries the most basic systemd information (timezone,
7318 hostname, root password) interactively on first
7319 boot. Alternatively it may also be used to provision these
7320 things offline on OS images installed into directories.
7322 * The default sysctl.d/ snippets will now set
7324 net.ipv4.conf.default.promote_secondaries=1
7326 This has the benefit of no flushing secondary IP addresses
7327 when primary addresses are removed.
7329 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
7330 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
7331 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
7332 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
7333 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
7334 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
7335 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
7336 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
7337 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
7338 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
7339 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
7340 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
7341 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
7342 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
7343 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
7345 — Berlin, 2014-08-19
7349 * A new tool systemd-sysusers has been added. This tool
7350 creates system users and groups in /etc/passwd and
7351 /etc/group, based on static declarative system user/group
7352 definitions in /usr/lib/sysusers.d/. This is useful to
7353 enable factory resets and volatile systems that boot up with
7354 an empty /etc directory, and thus need system users and
7355 groups created during early boot. systemd now also ships
7356 with two default sysusers.d/ files for the most basic
7357 users and groups systemd and the core operating system
7360 * A new tmpfiles snippet has been added that rebuilds the
7361 essential files in /etc on boot, should they be missing.
7363 * A directive for ensuring automatic clean-up of
7364 /var/cache/man/ has been removed from the default
7365 configuration. This line should now be shipped by the man
7366 implementation. The necessary change has been made to the
7367 man-db implementation. Note that you need to update your man
7368 implementation to one that ships this line, otherwise no
7369 automatic clean-up of /var/cache/man will take place.
7371 * A new condition ConditionNeedsUpdate= has been added that
7372 may conditionalize services to only run when /etc or /var
7373 are "older" than the vendor operating system resources in
7374 /usr. This is useful for reconstructing or updating /etc
7375 after an offline update of /usr or a factory reset, on the
7376 next reboot. Services that want to run once after such an
7377 update or reset should use this condition and order
7378 themselves before the new systemd-update-done.service, which
7379 will mark the two directories as fully updated. A number of
7380 service files have been added making use of this, to rebuild
7381 the udev hardware database, the journald message catalog and
7382 dynamic loader cache (ldconfig). The systemd-sysusers tool
7383 described above also makes use of this now. With this in
7384 place it is now possible to start up a minimal operating
7385 system with /etc empty cleanly. For more information on the
7386 concepts involved see this recent blog story:
7388 http://0pointer.de/blog/projects/stateless.html
7390 * A new system group "input" has been introduced, and all
7391 input device nodes get this group assigned. This is useful
7392 for system-level software to get access to input devices. It
7393 complements what is already done for "audio" and "video".
7395 * systemd-networkd learnt minimal DHCPv4 server support in
7396 addition to the existing DHCPv4 client support. It also
7397 learnt DHCPv6 client and IPv6 Router Solicitation client
7398 support. The DHCPv4 client gained support for static routes
7399 passed in from the server. Note that the [DHCPv4] section
7400 known in older systemd-networkd versions has been renamed to
7401 [DHCP] and is now also used by the DHCPv6 client. Existing
7402 .network files using settings of this section should be
7403 updated, though compatibility is maintained. Optionally, the
7404 client hostname may now be sent to the DHCP server.
7406 * networkd gained support for vxlan virtual networks as well
7407 as tun/tap and dummy devices.
7409 * networkd gained support for automatic allocation of address
7410 ranges for interfaces from a system-wide pool of
7411 addresses. This is useful for dynamically managing a large
7412 number of interfaces with a single network configuration
7413 file. In particular this is useful to easily assign
7414 appropriate IP addresses to the veth links of a large number
7415 of nspawn instances.
7417 * RPM macros for processing sysusers, sysctl and binfmt
7418 drop-in snippets at package installation time have been
7421 * The /etc/os-release file should now be placed in
7422 /usr/lib/os-release. The old location is automatically
7423 created as symlink. /usr/lib is the more appropriate
7424 location of this file, since it shall actually describe the
7425 vendor operating system shipped in /usr, and not the
7426 configuration stored in /etc.
7428 * .mount units gained a new boolean SloppyOptions= setting
7429 that maps to mount(8)'s -s option which enables permissive
7430 parsing of unknown mount options.
7432 * tmpfiles learnt a new "L+" directive which creates a symlink
7433 but (unlike "L") deletes a pre-existing file first, should
7434 it already exist and not already be the correct
7435 symlink. Similarly, "b+", "c+" and "p+" directives have been
7436 added as well, which create block and character devices, as
7437 well as fifos in the filesystem, possibly removing any
7438 pre-existing files of different types.
7440 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
7441 'argument' field (which so far specified the source to
7442 symlink/copy the files from) is now optional. If omitted the
7443 same file os copied from /usr/share/factory/ suffixed by the
7444 full destination path. This is useful for populating /etc
7445 with essential files, by copying them from vendor defaults
7446 shipped in /usr/share/factory/etc.
7448 * A new command "systemctl preset-all" has been added that
7449 applies the service preset settings to all installed unit
7450 files. A new switch --preset-mode= has been added that
7451 controls whether only enable or only disable operations
7454 * A new command "systemctl is-system-running" has been added
7455 that allows checking the overall state of the system, for
7456 example whether it is fully up and running.
7458 * When the system boots up with an empty /etc, the equivalent
7459 to "systemctl preset-all" is executed during early boot, to
7460 make sure all default services are enabled after a factory
7463 * systemd now contains a minimal preset file that enables the
7464 most basic services systemd ships by default.
7466 * Unit files' [Install] section gained a new DefaultInstance=
7467 field for defining the default instance to create if a
7468 template unit is enabled with no instance specified.
7470 * A new passive target cryptsetup-pre.target has been added
7471 that may be used by services that need to make they run and
7472 finish before the first LUKS cryptographic device is set up.
7474 * The /dev/loop-control and /dev/btrfs-control device nodes
7475 are now owned by the "disk" group by default, opening up
7476 access to this group.
7478 * systemd-coredump will now automatically generate a
7479 stack trace of all core dumps taking place on the system,
7480 based on elfutils' libdw library. This stack trace is logged
7483 * systemd-coredump may now optionally store coredumps directly
7484 on disk (in /var/lib/systemd/coredump, possibly compressed),
7485 instead of storing them unconditionally in the journal. This
7486 mode is the new default. A new configuration file
7487 /etc/systemd/coredump.conf has been added to configure this
7488 and other parameters of systemd-coredump.
7490 * coredumpctl gained a new "info" verb to show details about a
7491 specific coredump. A new switch "-1" has also been added
7492 that makes sure to only show information about the most
7493 recent entry instead of all entries. Also, as the tool is
7494 generally useful now the "systemd-" prefix of the binary
7495 name has been removed. Distributions that want to maintain
7496 compatibility with the old name should add a symlink from
7497 the old name to the new name.
7499 * journald's SplitMode= now defaults to "uid". This makes sure
7500 that unprivileged users can access their own coredumps with
7501 coredumpctl without restrictions.
7503 * New kernel command line options "systemd.wants=" (for
7504 pulling an additional unit during boot), "systemd.mask="
7505 (for masking a specific unit for the boot), and
7506 "systemd.debug-shell" (for enabling the debug shell on tty9)
7507 have been added. This is implemented in the new generator
7508 "systemd-debug-generator".
7510 * systemd-nspawn will now by default filter a couple of
7511 syscalls for containers, among them those required for
7512 kernel module loading, direct x86 IO port access, swap
7513 management, and kexec. Most importantly though
7514 open_by_handle_at() is now prohibited for containers,
7515 closing a hole similar to a recently discussed vulnerability
7516 in docker regarding access to files on file hierarchies the
7517 container should normally not have access to. Note that, for
7518 nspawn, we generally make no security claims anyway (and
7519 this is explicitly documented in the man page), so this is
7520 just a fix for one of the most obvious problems.
7522 * A new man page file-hierarchy(7) has been added that
7523 contains a minimized, modernized version of the file system
7524 layout systemd expects, similar in style to the FHS
7525 specification or hier(5). A new tool systemd-path(1) has
7526 been added to query many of these paths for the local
7529 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
7530 longer done. Since the directory now has a per-user size
7531 limit, and is cleaned on logout this appears unnecessary,
7532 in particular since this now brings the lifecycle of this
7533 directory closer in line with how IPC objects are handled.
7535 * systemd.pc now exports a number of additional directories,
7536 including $libdir (which is useful to identify the library
7537 path for the primary architecture of the system), and a
7538 couple of drop-in directories.
7540 * udev's predictable network interface names now use the dev_port
7541 sysfs attribute, introduced in linux 3.15 instead of dev_id to
7542 distinguish between ports of the same PCI function. dev_id should
7543 only be used for ports using the same HW address, hence the need
7546 * machined has been updated to export the OS version of a
7547 container (read from /etc/os-release and
7548 /usr/lib/os-release) on the bus. This is now shown in
7549 "machinectl status" for a machine.
7551 * A new service setting RestartForceExitStatus= has been
7552 added. If configured to a set of exit signals or process
7553 return values, the service will be restarted when the main
7554 daemon process exits with any of them, regardless of the
7557 * systemctl's -H switch for connecting to remote systemd
7558 machines has been extended so that it may be used to
7559 directly connect to a specific container on the
7560 host. "systemctl -H root@foobar:waldi" will now connect as
7561 user "root" to host "foobar", and then proceed directly to
7562 the container named "waldi". Note that currently you have to
7563 authenticate as user "root" for this to work, as entering
7564 containers is a privileged operation.
7566 Contributions from: Andreas Henriksson, Benjamin Steinwender,
7567 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
7568 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
7569 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
7570 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
7571 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
7572 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
7573 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
7574 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
7575 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
7576 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
7577 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
7579 — Berlin, 2014-07-03
7583 * As an experimental feature, udev now tries to lock the
7584 disk device node (flock(LOCK_SH|LOCK_NB)) while it
7585 executes events for the disk or any of its partitions.
7586 Applications like partitioning programs can lock the
7587 disk device node (flock(LOCK_EX)) and claim temporary
7588 device ownership that way; udev will entirely skip all event
7589 handling for this disk and its partitions. If the disk
7590 was opened for writing, the close will trigger a partition
7591 table rescan in udev's "watch" facility, and if needed
7592 synthesize "change" events for the disk and all its partitions.
7593 This is now unconditionally enabled, and if it turns out to
7594 cause major problems, we might turn it on only for specific
7595 devices, or might need to disable it entirely. Device Mapper
7596 devices are excluded from this logic.
7598 * We temporarily dropped the "-l" switch for fsck invocations,
7599 since they collide with the flock() logic above. util-linux
7600 upstream has been changed already to avoid this conflict,
7601 and we will re-add "-l" as soon as util-linux with this
7602 change has been released.
7604 * The dependency on libattr has been removed. Since a long
7605 time, the extended attribute calls have moved to glibc, and
7606 libattr is thus unnecessary.
7608 * Virtualization detection works without privileges now. This
7609 means the systemd-detect-virt binary no longer requires
7610 CAP_SYS_PTRACE file capabilities, and our daemons can run
7611 with fewer privileges.
7613 * systemd-networkd now runs under its own "systemd-network"
7614 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
7615 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
7616 loses the ability to write to files owned by root this way.
7618 * Similarly, systemd-resolved now runs under its own
7619 "systemd-resolve" user with no capabilities remaining.
7621 * Similarly, systemd-bus-proxyd now runs under its own
7622 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
7624 * systemd-networkd gained support for setting up "veth"
7625 virtual Ethernet devices for container connectivity, as well
7626 as GRE and VTI tunnels.
7628 * systemd-networkd will no longer automatically attempt to
7629 manually load kernel modules necessary for certain tunnel
7630 transports. Instead, it is assumed the kernel loads them
7631 automatically when required. This only works correctly on
7632 very new kernels. On older kernels, please consider adding
7633 the kernel modules to /etc/modules-load.d/ as a work-around.
7635 * The resolv.conf file systemd-resolved generates has been
7636 moved to /run/systemd/resolve/. If you have a symlink from
7637 /etc/resolv.conf, it might be necessary to correct it.
7639 * Two new service settings, ProtectHome= and ProtectSystem=,
7640 have been added. When enabled, they will make the user data
7641 (such as /home) inaccessible or read-only and the system
7642 (such as /usr) read-only, for specific services. This allows
7643 very light-weight per-service sandboxing to avoid
7644 modifications of user data or system files from
7645 services. These two new switches have been enabled for all
7646 of systemd's long-running services, where appropriate.
7648 * Socket units gained new SocketUser= and SocketGroup=
7649 settings to set the owner user and group of AF_UNIX sockets
7650 and FIFOs in the file system.
7652 * Socket units gained a new RemoveOnStop= setting. If enabled,
7653 all FIFOS and sockets in the file system will be removed
7654 when the specific socket unit is stopped.
7656 * Socket units gained a new Symlinks= setting. It takes a list
7657 of symlinks to create to file system sockets or FIFOs
7658 created by the specific Unix sockets. This is useful to
7659 manage symlinks to socket nodes with the same lifecycle as
7662 * The /dev/log socket and /dev/initctl FIFO have been moved to
7663 /run, and have been replaced by symlinks. This allows
7664 connecting to these facilities even if PrivateDevices=yes is
7665 used for a service (which makes /dev/log itself unavailable,
7666 but /run is left). This also has the benefit of ensuring
7667 that /dev only contains device nodes, directories and
7668 symlinks, and nothing else.
7670 * sd-daemon gained two new calls sd_pid_notify() and
7671 sd_pid_notifyf(). They are similar to sd_notify() and
7672 sd_notifyf(), but allow overriding of the source PID of
7673 notification messages if permissions permit this. This is
7674 useful to send notify messages on behalf of a different
7675 process (for example, the parent process). The
7676 systemd-notify tool has been updated to make use of this
7677 when sending messages (so that notification messages now
7678 originate from the shell script invoking systemd-notify and
7679 not the systemd-notify process itself. This should minimize
7680 a race where systemd fails to associate notification
7681 messages to services when the originating process already
7684 * A new "on-abnormal" setting for Restart= has been added. If
7685 set, it will result in automatic restarts on all "abnormal"
7686 reasons for a process to exit, which includes unclean
7687 signals, core dumps, timeouts and watchdog timeouts, but
7688 does not include clean and unclean exit codes or clean
7689 signals. Restart=on-abnormal is an alternative for
7690 Restart=on-failure for services that shall be able to
7691 terminate and avoid restarts on certain errors, by
7692 indicating so with an unclean exit code. Restart=on-failure
7693 or Restart=on-abnormal is now the recommended setting for
7694 all long-running services.
7696 * If the InaccessibleDirectories= service setting points to a
7697 mount point (or if there are any submounts contained within
7698 it), it is now attempted to completely unmount it, to make
7699 the file systems truly unavailable for the respective
7702 * The ReadOnlyDirectories= service setting and
7703 systemd-nspawn's --read-only parameter are now recursively
7704 applied to all submounts, too.
7706 * Mount units may now be created transiently via the bus APIs.
7708 * The support for SysV and LSB init scripts has been removed
7709 from the systemd daemon itself. Instead, it is now
7710 implemented as a generator that creates native systemd units
7711 from these scripts when needed. This enables us to remove a
7712 substantial amount of legacy code from PID 1, following the
7713 fact that many distributions only ship a very small number
7714 of LSB/SysV init scripts nowadays.
7716 * Privileged Xen (dom0) domains are not considered
7717 virtualization anymore by the virtualization detection
7718 logic. After all, they generally have unrestricted access to
7719 the hardware and usually are used to manage the unprivileged
7722 * systemd-tmpfiles gained a new "C" line type, for copying
7723 files or entire directories.
7725 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
7726 lines. So far, they have been non-globbing versions of the
7727 latter, and have thus been redundant. In future, it is
7728 recommended to only use "z". "m" has hence been removed
7729 from the documentation, even though it stays supported.
7731 * A tmpfiles snippet to recreate the most basic structure in
7732 /var has been added. This is enough to create the /var/run →
7733 /run symlink and create a couple of structural
7734 directories. This allows systems to boot up with an empty or
7735 volatile /var. Of course, while with this change, the core OS
7736 now is capable with dealing with a volatile /var, not all
7737 user services are ready for it. However, we hope that sooner
7738 or later, many service daemons will be changed upstream so
7739 that they are able to automatically create their necessary
7740 directories in /var at boot, should they be missing. This is
7741 the first step to allow state-less systems that only require
7742 the vendor image for /usr to boot.
7744 * systemd-nspawn has gained a new --tmpfs= switch to mount an
7745 empty tmpfs instance to a specific directory. This is
7746 particularly useful for making use of the automatic
7747 reconstruction of /var (see above), by passing --tmpfs=/var.
7749 * Access modes specified in tmpfiles snippets may now be
7750 prefixed with "~", which indicates that they shall be masked
7751 by whether the existing file or directory is currently
7752 writable, readable or executable at all. Also, if specified,
7753 the sgid/suid/sticky bits will be masked for all
7756 * A new passive target unit "network-pre.target" has been
7757 added which is useful for services that shall run before any
7758 network is configured, for example firewall scripts.
7760 * The "floppy" group that previously owned the /dev/fd*
7761 devices is no longer used. The "disk" group is now used
7762 instead. Distributions should probably deprecate usage of
7765 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
7766 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
7767 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
7768 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
7769 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
7770 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
7773 — Berlin, 2014-06-11
7777 * A new "systemd-timesyncd" daemon has been added for
7778 synchronizing the system clock across the network. It
7779 implements an SNTP client. In contrast to NTP
7780 implementations such as chrony or the NTP reference server,
7781 this only implements a client side, and does not bother with
7782 the full NTP complexity, focusing only on querying time from
7783 one remote server and synchronizing the local clock to
7784 it. Unless you intend to serve NTP to networked clients or
7785 want to connect to local hardware clocks, this simple NTP
7786 client should be more than appropriate for most
7787 installations. The daemon runs with minimal privileges, and
7788 has been hooked up with networkd to only operate when
7789 network connectivity is available. The daemon saves the
7790 current clock to disk every time a new NTP sync has been
7791 acquired, and uses this to possibly correct the system clock
7792 early at bootup, in order to accommodate for systems that
7793 lack an RTC such as the Raspberry Pi and embedded devices,
7794 and to make sure that time monotonically progresses on these
7795 systems, even if it is not always correct. To make use of
7796 this daemon, a new system user and group "systemd-timesync"
7797 needs to be created on installation of systemd.
7799 * The queue "seqnum" interface of libudev has been disabled, as
7800 it was generally incompatible with device namespacing as
7801 sequence numbers of devices go "missing" if the devices are
7802 part of a different namespace.
7804 * "systemctl list-timers" and "systemctl list-sockets" gained
7805 a --recursive switch for showing units of these types also
7806 for all local containers, similar in style to the already
7807 supported --recursive switch for "systemctl list-units".
7809 * A new RebootArgument= setting has been added for service
7810 units, which may be used to specify a kernel reboot argument
7811 to use when triggering reboots with StartLimitAction=.
7813 * A new FailureAction= setting has been added for service
7814 units which may be used to specify an operation to trigger
7815 when a service fails. This works similarly to
7816 StartLimitAction=, but unlike it, controls what is done
7817 immediately rather than only after several attempts to
7818 restart the service in question.
7820 * hostnamed got updated to also expose the kernel name,
7821 release, and version on the bus. This is useful for
7822 executing commands like hostnamectl with the -H switch.
7823 systemd-analyze makes use of this to properly display
7824 details when running non-locally.
7826 * The bootchart tool can now show cgroup information in the
7827 graphs it generates.
7829 * The CFS CPU quota cgroup attribute is now exposed for
7830 services. The new CPUQuota= switch has been added for this
7831 which takes a percentage value. Setting this will have the
7832 result that a service may never get more CPU time than the
7833 specified percentage, even if the machine is otherwise idle.
7835 * systemd-networkd learned IPIP and SIT tunnel support.
7837 * LSB init scripts exposing a dependency on $network will now
7838 get a dependency on network-online.target rather than simply
7839 network.target. This should bring LSB handling closer to
7840 what it was on SysV systems.
7842 * A new fsck.repair= kernel option has been added to control
7843 how fsck shall deal with unclean file systems at boot.
7845 * The (.ini) configuration file parser will now silently ignore
7846 sections whose names begin with "X-". This may be used to maintain
7847 application-specific extension sections in unit files.
7849 * machined gained a new API to query the IP addresses of
7850 registered containers. "machinectl status" has been updated
7851 to show these addresses in its output.
7853 * A new call sd_uid_get_display() has been added to the
7854 sd-login APIs for querying the "primary" session of a
7855 user. The "primary" session of the user is elected from the
7856 user's sessions and generally a graphical session is
7857 preferred over a text one.
7859 * A minimal systemd-resolved daemon has been added. It
7860 currently simply acts as a companion to systemd-networkd and
7861 manages resolv.conf based on per-interface DNS
7862 configuration, possibly supplied via DHCP. In the long run
7863 we hope to extend this into a local DNSSEC enabled DNS and
7866 * The systemd-networkd-wait-online tool is now enabled by
7867 default. It will delay network-online.target until a network
7868 connection has been configured. The tool primarily integrates
7869 with networkd, but will also make a best effort to make sense
7870 of network configuration performed in some other way.
7872 * Two new service options StartupCPUShares= and
7873 StartupBlockIOWeight= have been added that work similarly to
7874 CPUShares= and BlockIOWeight= however only apply during
7875 system startup. This is useful to prioritize certain services
7876 differently during bootup than during normal runtime.
7878 * hostnamed has been changed to prefer the statically
7879 configured hostname in /etc/hostname (unless set to
7880 'localhost' or empty) over any dynamic one supplied by
7881 dhcp. With this change, the rules for picking the hostname
7882 match more closely the rules of other configuration settings
7883 where the local administrator's configuration in /etc always
7884 overrides any other settings.
7886 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
7887 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
7888 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
7889 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
7890 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
7891 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
7892 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
7893 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
7894 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
7895 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
7896 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
7897 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
7898 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
7899 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
7900 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
7901 Lindskog, WaLyong Cho, Will Woods, Zbigniew
7904 — Beijing, 2014-05-28
7908 * When restoring the screen brightness at boot, stay away from
7909 the darkest setting or from the lowest 5% of the available
7910 range, depending on which is the larger value of both. This
7911 should effectively protect the user from rebooting into a
7912 black screen, should the brightness have been set to minimum
7915 * sd-login gained a new sd_machine_get_class() call to
7916 determine the class ("vm" or "container") of a machine
7917 registered with machined.
7919 * sd-login gained new calls
7920 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
7921 to query the identity of the peer of a local AF_UNIX
7922 connection. They operate similarly to their sd_pid_get_xyz()
7925 * PID 1 will now maintain a system-wide system state engine
7926 with the states "starting", "running", "degraded",
7927 "maintenance", "stopping". These states are bound to system
7928 startup, normal runtime, runtime with at least one failed
7929 service, rescue/emergency mode and system shutdown. This
7930 state is shown in the "systemctl status" output when no unit
7931 name is passed. It is useful to determine system state, in
7932 particularly when doing so for many systems or containers at
7935 * A new command "list-machines" has been added to "systemctl"
7936 that lists all local OS containers and shows their system
7937 state (see above), if systemd runs inside of them.
7939 * systemctl gained a new "-r" switch to recursively enumerate
7940 units on all local containers, when used with the
7941 "list-unit" command (which is the default one that is
7942 executed when no parameters are specified).
7944 * The GPT automatic partition discovery logic will now honour
7945 two GPT partition flags: one may be set on a partition to
7946 cause it to be mounted read-only, and the other may be set
7947 on a partition to ignore it during automatic discovery.
7949 * Two new GPT type UUIDs have been added for automatic root
7950 partition discovery, for 32-bit and 64-bit ARM. This is not
7951 particularly useful for discovering the root directory on
7952 these architectures during bare-metal boots (since UEFI is
7953 not common there), but still very useful to allow booting of
7954 ARM disk images in nspawn with the -i option.
7956 * MAC addresses of interfaces created with nspawn's
7957 --network-interface= switch will now be generated from the
7958 machine name, and thus be stable between multiple invocations
7961 * logind will now automatically remove all IPC objects owned
7962 by a user if she or he fully logs out. This makes sure that
7963 users who are logged out cannot continue to consume IPC
7964 resources. This covers SysV memory, semaphores and message
7965 queues as well as POSIX shared memory and message
7966 queues. Traditionally, SysV and POSIX IPC had no lifecycle
7967 limits. With this functionality, that is corrected. This may
7968 be turned off by using the RemoveIPC= switch of logind.conf.
7970 * The systemd-machine-id-setup and tmpfiles tools gained a
7971 --root= switch to operate on a specific root directory,
7974 * journald can now forward logged messages to the TTYs of all
7975 logged in users ("wall"). This is the default for all
7976 emergency messages now.
7978 * A new tool systemd-journal-remote has been added to stream
7979 journal log messages across the network.
7981 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
7982 controller trees are mounted into it. Note that the
7983 directories mounted beneath it are not read-only. This is a
7984 security measure and is particularly useful because glibc
7985 actually includes a search logic to pick any tmpfs it can
7986 find to implement shm_open() if /dev/shm is not available
7987 (which it might very well be in namespaced setups).
7989 * machinectl gained a new "poweroff" command to cleanly power
7990 down a local OS container.
7992 * The PrivateDevices= unit file setting will now also drop the
7993 CAP_MKNOD capability from the capability bound set, and
7994 imply DevicePolicy=closed.
7996 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
7997 comprehensively on all long-running systemd services where
7998 this is appropriate.
8000 * systemd-udevd will now run in a disassociated mount
8001 namespace. To mount directories from udev rules, make sure to
8002 pull in mount units via SYSTEMD_WANTS properties.
8004 * The kdbus support gained support for uploading policy into
8005 the kernel. sd-bus gained support for creating "monitoring"
8006 connections that can eavesdrop into all bus communication
8007 for debugging purposes.
8009 * Timestamps may now be specified in seconds since the UNIX
8010 epoch Jan 1st, 1970 by specifying "@" followed by the value
8013 * Native tcpwrap support in systemd has been removed. tcpwrap
8014 is old code, not really maintained anymore and has serious
8015 shortcomings, and better options such as firewalls
8016 exist. For setups that require tcpwrap usage, please
8017 consider invoking your socket-activated service via tcpd,
8018 like on traditional inetd.
8020 * A new system.conf configuration option
8021 DefaultTimerAccuracySec= has been added that controls the
8022 default AccuracySec= setting of .timer units.
8024 * Timer units gained a new WakeSystem= switch. If enabled,
8025 timers configured this way will cause the system to resume
8026 from system suspend (if the system supports that, which most
8029 * Timer units gained a new Persistent= switch. If enabled,
8030 timers configured this way will save to disk when they have
8031 been last triggered. This information is then used on next
8032 reboot to possible execute overdue timer events, that
8033 could not take place because the system was powered off.
8034 This enables simple anacron-like behaviour for timer units.
8036 * systemctl's "list-timers" will now also list the time a
8037 timer unit was last triggered in addition to the next time
8038 it will be triggered.
8040 * systemd-networkd will now assign predictable IPv4LL
8041 addresses to its local interfaces.
8043 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
8044 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
8045 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
8046 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
8047 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
8048 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
8049 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
8050 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
8053 — Berlin, 2014-03-25
8057 * A new unit file setting RestrictAddressFamilies= has been
8058 added to restrict which socket address families unit
8059 processes gain access to. This takes address family names
8060 like "AF_INET" or "AF_UNIX", and is useful to minimize the
8061 attack surface of services via exotic protocol stacks. This
8062 is built on seccomp system call filters.
8064 * Two new unit file settings RuntimeDirectory= and
8065 RuntimeDirectoryMode= have been added that may be used to
8066 manage a per-daemon runtime directories below /run. This is
8067 an alternative for setting up directory permissions with
8068 tmpfiles snippets, and has the advantage that the runtime
8069 directory's lifetime is bound to the daemon runtime and that
8070 the daemon starts up with an empty directory each time. This
8071 is particularly useful when writing services that drop
8072 privileges using the User= or Group= setting.
8074 * The DeviceAllow= unit setting now supports globbing for
8075 matching against device group names.
8077 * The systemd configuration file system.conf gained new
8078 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
8079 DefaultMemoryAccounting= to globally turn on/off accounting
8080 for specific resources (cgroups) for all units. These
8081 settings may still be overridden individually in each unit
8084 * systemd-gpt-auto-generator is now able to discover /srv and
8085 root partitions in addition to /home and swap partitions. It
8086 also supports LUKS-encrypted partitions now. With this in
8087 place, automatic discovery of partitions to mount following
8088 the Discoverable Partitions Specification
8089 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
8090 is now a lot more complete. This allows booting without
8091 /etc/fstab and without root= on the kernel command line on
8092 systems prepared appropriately.
8094 * systemd-nspawn gained a new --image= switch which allows
8095 booting up disk images and Linux installations on any block
8096 device that follow the Discoverable Partitions Specification
8097 (see above). This means that installations made with
8098 appropriately updated installers may now be started and
8099 deployed using container managers, completely
8100 unmodified. (We hope that libvirt-lxc will add support for
8101 this feature soon, too.)
8103 * systemd-nspawn gained a new --network-macvlan= setting to
8104 set up a private macvlan interface for the
8105 container. Similarly, systemd-networkd gained a new
8106 Kind=macvlan setting in .netdev files.
8108 * systemd-networkd now supports configuring local addresses
8111 * A new tool systemd-network-wait-online has been added to
8112 synchronously wait for network connectivity using
8115 * The sd-bus.h bus API gained a new sd_bus_track object for
8116 tracking the lifecycle of bus peers. Note that sd-bus.h is
8117 still not a public API though (unless you specify
8118 --enable-kdbus on the configure command line, which however
8119 voids your warranty and you get no API stability guarantee).
8121 * The $XDG_RUNTIME_DIR runtime directories for each user are
8122 now individual tmpfs instances, which has the benefit of
8123 introducing separate pools for each user, with individual
8124 size limits, and thus making sure that unprivileged clients
8125 can no longer negatively impact the system or other users by
8126 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
8127 RuntimeDirectorySize= has been introduced that allows
8128 controlling the default size limit for all users. It
8129 defaults to 10% of the available physical memory. This is no
8130 replacement for quotas on tmpfs though (which the kernel
8131 still does not support), as /dev/shm and /tmp are still
8132 shared resources used by both the system and unprivileged
8135 * logind will now automatically turn off automatic suspending
8136 on laptop lid close when more than one display is
8137 connected. This was previously expected to be implemented
8138 individually in desktop environments (such as GNOME),
8139 however has been added to logind now, in order to fix a
8140 boot-time race where a desktop environment might not have
8141 been started yet and thus not been able to take an inhibitor
8142 lock at the time where logind already suspends the system
8143 due to a closed lid.
8145 * logind will now wait at least 30s after each system
8146 suspend/resume cycle, and 3min after system boot before
8147 suspending the system due to a closed laptop lid. This
8148 should give USB docking stations and similar enough time to
8149 be probed and configured after system resume and boot in
8150 order to then act as suspend blocker.
8152 * systemd-run gained a new --property= setting which allows
8153 initialization of resource control properties (and others)
8154 for the created scope or service unit. Example: "systemd-run
8155 --property=BlockIOWeight=10 updatedb" may be used to run
8156 updatedb at a low block IO scheduling weight.
8158 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
8159 now also work in --scope mode.
8161 * When systemd is compiled with kdbus support, basic support
8162 for enforced policies is now in place. (Note that enabling
8163 kdbus still voids your warranty and no API compatibility
8166 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
8167 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
8168 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
8169 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
8170 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
8171 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
8172 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
8173 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
8174 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
8175 Zbigniew Jędrzejewski-Szmek
8177 — Berlin, 2014-03-12
8181 * systemd will now relabel /dev after loading the SMACK policy
8182 according to SMACK rules.
8184 * A new unit file option AppArmorProfile= has been added to
8185 set the AppArmor profile for the processes of a unit.
8187 * A new condition check ConditionArchitecture= has been added
8188 to conditionalize units based on the system architecture, as
8189 reported by uname()'s "machine" field.
8191 * systemd-networkd now supports matching on the system
8192 virtualization, architecture, kernel command line, hostname
8195 * logind is now a lot more aggressive when suspending the
8196 machine due to a closed laptop lid. Instead of acting only
8197 on the lid close action, it will continuously watch the lid
8198 status and act on it. This is useful for laptops where the
8199 power button is on the outside of the chassis so that it can
8200 be reached without opening the lid (such as the Lenovo
8201 Yoga). On those machines, logind will now immediately
8202 re-suspend the machine if the power button has been
8203 accidentally pressed while the laptop was suspended and in a
8204 backpack or similar.
8206 * logind will now watch SW_DOCK switches and inhibit reaction
8207 to the lid switch if it is pressed. This means that logind
8208 will not suspend the machine anymore if the lid is closed
8209 and the system is docked, if the laptop supports SW_DOCK
8210 notifications via the input layer. Note that ACPI docking
8211 stations do not generate this currently. Also note that this
8212 logic is usually not fully sufficient and Desktop
8213 Environments should take a lid switch inhibitor lock when an
8214 external display is connected, as systemd will not watch
8217 * nspawn will now make use of the devices cgroup controller by
8218 default, and only permit creation of and access to the usual
8219 API device nodes like /dev/null or /dev/random, as well as
8220 access to (but not creation of) the pty devices.
8222 * We will now ship a default .network file for
8223 systemd-networkd that automatically configures DHCP for
8224 network interfaces created by nspawn's --network-veth or
8225 --network-bridge= switches.
8227 * systemd will now understand the usual M, K, G, T suffixes
8228 according to SI conventions (i.e. to the base 1000) when
8229 referring to throughput and hardware metrics. It will stay
8230 with IEC conventions (i.e. to the base 1024) for software
8231 metrics, according to what is customary according to
8232 Wikipedia. We explicitly document which base applies for
8233 each configuration option.
8235 * The DeviceAllow= setting in unit files now supports a syntax to
8236 allow-list an entire group of devices node majors at once, based on
8237 the /proc/devices listing. For example, with the string "char-pts",
8238 it is now possible to allow-list all current and future pseudo-TTYs
8241 * sd-event learned a new "post" event source. Event sources of
8242 this type are triggered by the dispatching of any event
8243 source of a type that is not "post". This is useful for
8244 implementing clean-up and check event sources that are
8245 triggered by other work being done in the program.
8247 * systemd-networkd is no longer statically enabled, but uses
8248 the usual [Install] sections so that it can be
8249 enabled/disabled using systemctl. It still is enabled by
8252 * When creating a veth interface pair with systemd-nspawn, the
8253 host side will now be prefixed with "vb-" if
8254 --network-bridge= is used, and with "ve-" if --network-veth
8255 is used. This way, it is easy to distinguish these cases on
8256 the host, for example to apply different configuration to
8257 them with systemd-networkd.
8259 * The compatibility libraries for libsystemd-journal.so,
8260 libsystem-id128.so, libsystemd-login.so and
8261 libsystemd-daemon.so do not make use of IFUNC
8262 anymore. Instead, we now build libsystemd.so multiple times
8263 under these alternative names. This means that the footprint
8264 is drastically increased, but given that these are
8265 transitional compatibility libraries, this should not matter
8266 much. This change has been made necessary to support the ARM
8267 platform for these compatibility libraries, as the ARM
8268 toolchain is not really at the same level as the toolchain
8269 for other architectures like x86 and does not support
8270 IFUNC. Please make sure to use --enable-compat-libs only
8271 during a transitional period!
8273 * The .include syntax has been deprecated and is not documented
8274 anymore. Drop-in files in .d directories should be used instead.
8276 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
8277 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
8278 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
8279 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
8280 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
8281 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
8282 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
8283 Zbigniew Jędrzejewski-Szmek
8285 — Berlin, 2014-02-24
8289 * A new component "systemd-networkd" has been added that can
8290 be used to configure local network interfaces statically or
8291 via DHCP. It is capable of bringing up bridges, VLANs, and
8292 bonding. Currently, no hook-ups for interactive network
8293 configuration are provided. Use this for your initrd,
8294 container, embedded, or server setup if you need a simple,
8295 yet powerful, network configuration solution. This
8296 configuration subsystem is quite nifty, as it allows wildcard
8297 hotplug matching in interfaces. For example, with a single
8298 configuration snippet, you can configure that all Ethernet
8299 interfaces showing up are automatically added to a bridge,
8300 or similar. It supports link-sensing and more.
8302 * A new tool "systemd-socket-proxyd" has been added which can
8303 act as a bidirectional proxy for TCP sockets. This is
8304 useful for adding socket activation support to services that
8305 do not actually support socket activation, including virtual
8306 machines and the like.
8308 * Add a new tool to save/restore rfkill state on
8311 * Save/restore state of keyboard backlights in addition to
8312 display backlights on shutdown/boot.
8314 * udev learned a new SECLABEL{} construct to label device
8315 nodes with a specific security label when they appear. For
8316 now, only SECLABEL{selinux} is supported, but the syntax is
8317 prepared for additional security frameworks.
8319 * udev gained a new scheme to configure link-level attributes
8320 from files in /etc/systemd/network/*.link. These files can
8321 match against MAC address, device path, driver name and type,
8322 and will apply attributes like the naming policy, link speed,
8323 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
8324 address assignment policy (randomized, ...).
8326 * The configuration of network interface naming rules for
8327 "permanent interface names" has changed: a new NamePolicy=
8328 setting in the [Link] section of .link files determines the
8329 priority of possible naming schemes (onboard, slot, MAC,
8330 path). The default value of this setting is determined by
8331 /usr/lib/net/links/99-default.link. Old
8332 80-net-name-slot.rules udev configuration file has been
8333 removed, so local configuration overriding this file should
8334 be adapted to override 99-default.link instead.
8336 * When the User= switch is used in a unit file, also
8337 initialize $SHELL= based on the user database entry.
8339 * systemd no longer depends on libdbus. All communication is
8340 now done with sd-bus, systemd's low-level bus library
8343 * kdbus support has been added to PID 1 itself. When kdbus is
8344 enabled, this causes PID 1 to set up the system bus and
8345 enable support for a new ".busname" unit type that
8346 encapsulates bus name activation on kdbus. It works a little
8347 bit like ".socket" units, except for bus names. A new
8348 generator has been added that converts classic dbus1 service
8349 activation files automatically into native systemd .busname
8352 * sd-bus: add a light-weight vtable implementation that allows
8353 defining objects on the bus with a simple static const
8354 vtable array of its methods, signals and properties.
8356 * systemd will not generate or install static dbus
8357 introspection data anymore to /usr/share/dbus-1/interfaces,
8358 as the precise format of these files is unclear, and
8359 nothing makes use of it.
8361 * A proxy daemon is now provided to proxy clients connecting
8362 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
8363 compatibility with classic D-Bus.
8365 * A bus driver implementation has been added that supports the
8366 classic D-Bus bus driver calls on kdbus, also for
8367 compatibility purposes.
8369 * A new API "sd-event.h" has been added that implements a
8370 minimal event loop API built around epoll. It provides a
8371 couple of features that direct epoll usage is lacking:
8372 prioritization of events, scales to large numbers of timer
8373 events, per-event timer slack (accuracy), system-wide
8374 coalescing of timer events, exit handlers, watchdog
8375 supervision support using systemd's sd_notify() API, child
8378 * A new API "sd-rntl.h" has been added that provides an API
8379 around the route netlink interface of the kernel, similar in
8380 style to "sd-bus.h".
8382 * A new API "sd-dhcp-client.h" has been added that provides a
8383 small DHCPv4 client-side implementation. This is used by
8386 * There is a new kernel command line option
8387 "systemd.restore_state=0|1". When set to "0", none of the
8388 systemd tools will restore saved runtime state to hardware
8389 devices. More specifically, the rfkill and backlight states
8392 * The FsckPassNo= compatibility option in mount/service units
8393 has been removed. The fstab generator will now add the
8394 necessary dependencies automatically, and does not require
8395 PID1's support for that anymore.
8397 * journalctl gained a new switch, --list-boots, that lists
8398 recent boots with their times and boot IDs.
8400 * The various tools like systemctl, loginctl, timedatectl,
8401 busctl, systemd-run, ... have gained a new switch "-M" to
8402 connect to a specific, local OS container (as direct
8403 connection, without requiring SSH). This works on any
8404 container that is registered with machined, such as those
8405 created by libvirt-lxc or nspawn.
8407 * systemd-run and systemd-analyze also gained support for "-H"
8408 to connect to remote hosts via SSH. This is particularly
8409 useful for systemd-run because it enables queuing of jobs
8410 onto remote systems.
8412 * machinectl gained a new command "login" to open a getty
8413 login in any local container. This works with any container
8414 that is registered with machined (such as those created by
8415 libvirt-lxc or nspawn), and which runs systemd inside.
8417 * machinectl gained a new "reboot" command that may be used to
8418 trigger a reboot on a specific container that is registered
8419 with machined. This works on any container that runs an init
8420 system of some kind.
8422 * systemctl gained a new "list-timers" command to print a nice
8423 listing of installed timer units with the times they elapse
8426 * Alternative reboot() parameters may now be specified on the
8427 "systemctl reboot" command line and are passed to the
8428 reboot() system call.
8430 * systemctl gained a new --job-mode= switch to configure the
8431 mode to queue a job with. This is a more generic version of
8432 --fail, --irreversible, and --ignore-dependencies, which are
8433 still available but not advertised anymore.
8435 * /etc/systemd/system.conf gained new settings to configure
8436 various default timeouts of units, as well as the default
8437 start limit interval and burst. These may still be overridden
8440 * PID1 will now export on the bus profile data of the security
8441 policy upload process (such as the SELinux policy upload to
8444 * journald: when forwarding logs to the console, include
8445 timestamps (following the setting in
8446 /sys/module/printk/parameters/time).
8448 * OnCalendar= in timer units now understands the special
8449 strings "yearly" and "annually". (Both are equivalent)
8451 * The accuracy of timer units is now configurable with the new
8452 AccuracySec= setting. It defaults to 1min.
8454 * A new dependency type JoinsNamespaceOf= has been added that
8455 allows running two services within the same /tmp and network
8456 namespace, if PrivateNetwork= or PrivateTmp= are used.
8458 * A new command "cat" has been added to systemctl. It outputs
8459 the original unit file of a unit, and concatenates the
8460 contents of additional "drop-in" unit file snippets, so that
8461 the full configuration is shown.
8463 * systemctl now supports globbing on the various "list-xyz"
8464 commands, like "list-units" or "list-sockets", as well as on
8465 those commands which take multiple unit names.
8467 * journalctl's --unit= switch gained support for globbing.
8469 * All systemd daemons now make use of the watchdog logic so
8470 that systemd automatically notices when they hang.
8472 * If the $container_ttys environment variable is set,
8473 getty-generator will automatically spawn a getty for each
8474 listed tty. This is useful for container managers to request
8475 login gettys to be spawned on as many ttys as needed.
8477 * %h, %s, %U specifier support is not available anymore when
8478 used in unit files for PID 1. This is because NSS calls are
8479 not safe from PID 1. They stay available for --user
8480 instances of systemd, and as special case for the root user.
8482 * loginctl gained a new "--no-legend" switch to turn off output
8485 * The "sd-login.h" API gained three new calls:
8486 sd_session_is_remote(), sd_session_get_remote_user(),
8487 sd_session_get_remote_host() to query information about
8490 * The udev hardware database now also carries vendor/product
8491 information of SDIO devices.
8493 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
8494 determine whether watchdog notifications are requested by
8497 * Socket-activated per-connection services now include a
8498 short description of the connection parameters in the
8501 * tmpfiles gained a new "--boot" option. When this is not used,
8502 only lines where the command character is not suffixed with
8503 "!" are executed. When this option is specified, those
8504 options are executed too. This partitions tmpfiles
8505 directives into those that can be safely executed at any
8506 time, and those which should be run only at boot (for
8507 example, a line that creates /run/nologin).
8509 * A new API "sd-resolve.h" has been added which provides a simple
8510 asynchronous wrapper around glibc NSS hostname resolution
8511 calls, such as getaddrinfo(). In contrast to glibc's
8512 getaddrinfo_a(), it does not use signals. In contrast to most
8513 other asynchronous name resolution libraries, this one does
8514 not reimplement DNS, but reuses NSS, so that alternate
8515 hostname resolution systems continue to work, such as mDNS,
8516 LDAP, etc. This API is based on libasyncns, but it has been
8517 cleaned up for inclusion in systemd.
8519 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
8520 "sd-daemon.h" are no longer found in individual libraries
8521 libsystemd-journal.so, libsystemd-login.so,
8522 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
8523 merged them into a single library, libsystemd.so, which
8524 provides all symbols. The reason for this is cyclic
8525 dependencies, as these libraries tend to use each other's
8526 symbols. So far, we have managed to workaround that by linking
8527 a copy of a good part of our code into each of these
8528 libraries again and again, which, however, makes certain
8529 things hard to do, like sharing static variables. Also, it
8530 substantially increases footprint. With this change, there
8531 is only one library for the basic APIs systemd
8532 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
8533 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
8534 library as well, however are subject to the --enable-kdbus
8535 switch (see below). Note that "sd-dhcp-client.h" is not part
8536 of this library (this is because it only consumes, never
8537 provides, services of/to other APIs). To make the transition
8538 easy from the separate libraries to the unified one, we
8539 provide the --enable-compat-libs compile-time switch which
8540 will generate stub libraries that are compatible with the
8541 old ones but redirect all calls to the new one.
8543 * All of the kdbus logic and the new APIs "sd-bus.h",
8544 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
8545 and "sd-utf8.h" are compile-time optional via the
8546 "--enable-kdbus" switch, and they are not compiled in by
8547 default. To make use of kdbus, you have to explicitly enable
8548 the switch. Note however, that neither the kernel nor the
8549 userspace API for all of this is considered stable yet. We
8550 want to maintain the freedom to still change the APIs for
8551 now. By specifying this build-time switch, you acknowledge
8552 that you are aware of the instability of the current
8555 * Also, note that while kdbus is pretty much complete,
8556 it lacks one thing: proper policy support. This means you
8557 can build a fully working system with all features; however,
8558 it will be highly insecure. Policy support will be added in
8559 one of the next releases, at the same time that we will
8560 declare the APIs stable.
8562 * When the kernel command line argument "kdbus" is specified,
8563 systemd will automatically load the kdbus.ko kernel module. At
8564 this stage of development, it is only useful for testing kdbus
8565 and should not be used in production. Note: if "--enable-kdbus"
8566 is specified, and the kdbus.ko kernel module is available, and
8567 "kdbus" is added to the kernel command line, the entire system
8568 runs with kdbus instead of dbus-daemon, with the above mentioned
8569 problem of missing the system policy enforcement. Also a future
8570 version of kdbus.ko or a newer systemd will not be compatible with
8571 each other, and will unlikely be able to boot the machine if only
8572 one of them is updated.
8574 * systemctl gained a new "import-environment" command which
8575 uploads the caller's environment (or parts thereof) into the
8576 service manager so that it is inherited by services started
8577 by the manager. This is useful to upload variables like
8578 $DISPLAY into the user service manager.
8580 * A new PrivateDevices= switch has been added to service units
8581 which allows running a service with a namespaced /dev
8582 directory that does not contain any device nodes for
8583 physical devices. More specifically, it only includes devices
8584 such as /dev/null, /dev/urandom, and /dev/zero which are API
8587 * logind has been extended to support behaviour like VT
8588 switching on seats that do not support a VT. This makes
8589 multi-session available on seats that are not the first seat
8590 (seat0), and on systems where kernel support for VTs has
8591 been disabled at compile-time.
8593 * If a process holds a delay lock for system sleep or shutdown
8594 and fails to release it in time, we will now log its
8595 identity. This makes it easier to identify processes that
8596 cause slow suspends or power-offs.
8598 * When parsing /etc/crypttab, support for a new key-slot=
8599 option as supported by Debian is added. It allows indicating
8600 which LUKS slot to use on disk, speeding up key loading.
8602 * The sd_journal_sendv() API call has been checked and
8603 officially declared to be async-signal-safe so that it may
8604 be invoked from signal handlers for logging purposes.
8606 * Boot-time status output is now enabled automatically after a
8607 short timeout if boot does not progress, in order to give
8608 the user an indication what she or he is waiting for.
8610 * The boot-time output has been improved to show how much time
8611 remains until jobs expire.
8613 * The KillMode= switch in service units gained a new possible
8614 value "mixed". If set, and the unit is shut down, then the
8615 initial SIGTERM signal is sent only to the main daemon
8616 process, while the following SIGKILL signal is sent to
8617 all remaining processes of the service.
8619 * When a scope unit is registered, a new property "Controller"
8620 may be set. If set to a valid bus name, systemd will send a
8621 RequestStop() signal to this name when it would like to shut
8622 down the scope. This may be used to hook manager logic into
8623 the shutdown logic of scope units. Also, scope units may now
8624 be put in a special "abandoned" state, in which case the
8625 manager process which created them takes no further
8626 responsibilities for it.
8628 * When reading unit files, systemd will now verify
8629 the access mode of these files, and warn about certain
8630 suspicious combinations. This has been added to make it
8631 easier to track down packaging bugs where unit files are
8632 marked executable or world-writable.
8634 * systemd-nspawn gained a new "--setenv=" switch to set
8635 container-wide environment variables. The similar option in
8636 systemd-activate was renamed from "--environment=" to
8637 "--setenv=" for consistency.
8639 * systemd-nspawn has been updated to create a new kdbus domain
8640 for each container that is invoked, thus allowing each
8641 container to have its own set of system and user buses,
8642 independent of the host.
8644 * systemd-nspawn gained a new --drop-capability= switch to run
8645 the container with less capabilities than the default. Both
8646 --drop-capability= and --capability= now take the special
8647 string "all" for dropping or keeping all capabilities.
8649 * systemd-nspawn gained new switches for executing containers
8650 with specific SELinux labels set.
8652 * systemd-nspawn gained a new --quiet switch to not generate
8653 any additional output but the container's own console
8656 * systemd-nspawn gained a new --share-system switch to run a
8657 container without PID namespacing enabled.
8659 * systemd-nspawn gained a new --register= switch to control
8660 whether the container is registered with systemd-machined or
8661 not. This is useful for containers that do not run full
8662 OS images, but only specific apps.
8664 * systemd-nspawn gained a new --keep-unit which may be used
8665 when invoked as the only program from a service unit, and
8666 results in registration of the unit service itself in
8667 systemd-machined, instead of a newly opened scope unit.
8669 * systemd-nspawn gained a new --network-interface= switch for
8670 moving arbitrary interfaces to the container. The new
8671 --network-veth switch creates a virtual Ethernet connection
8672 between host and container. The new --network-bridge=
8673 switch then allows assigning the host side of this virtual
8674 Ethernet connection to a bridge device.
8676 * systemd-nspawn gained a new --personality= switch for
8677 setting the kernel personality for the container. This is
8678 useful when running a 32-bit container on a 64-bit host. A
8679 similar option Personality= is now also available for service
8682 * logind will now also track a "Desktop" identifier for each
8683 session which encodes the desktop environment of it. This is
8684 useful for desktop environments that want to identify
8685 multiple running sessions of itself easily.
8687 * A new SELinuxContext= setting for service units has been
8688 added that allows setting a specific SELinux execution
8689 context for a service.
8691 * Most systemd client tools will now honour $SYSTEMD_LESS for
8692 settings of the "less" pager. By default, these tools will
8693 override $LESS to allow certain operations to work, such as
8694 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
8695 influence this logic.
8697 * systemd's "seccomp" hook-up has been changed to make use of
8698 the libseccomp library instead of using its own
8699 implementation. This has benefits for portability among
8702 * For usage together with SystemCallFilter=, a new
8703 SystemCallErrorNumber= setting has been introduced that
8704 allows configuration of a system error number to be returned
8705 on filtered system calls, instead of immediately killing the
8706 process. Also, SystemCallArchitectures= has been added to
8707 limit access to system calls of a particular architecture
8708 (in order to turn off support for unused secondary
8709 architectures). There is also a global
8710 SystemCallArchitectures= setting in system.conf now to turn
8711 off support for non-native system calls system-wide.
8713 * systemd requires a kernel with a working name_to_handle_at(),
8714 please see the kernel config requirements in the README file.
8716 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
8717 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
8718 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
8719 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
8720 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
8721 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
8722 Elia Pinto, Florian Weimer, George McCollister, Goffredo
8723 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
8724 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
8725 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
8726 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
8727 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
8728 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
8729 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
8730 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
8731 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
8732 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
8733 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
8734 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
8735 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
8736 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
8737 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
8738 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
8739 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
8741 — Berlin, 2014-02-20
8745 * logind has gained support for facilitating privileged input
8746 and drm device access for unprivileged clients. This work is
8747 useful to allow Wayland display servers (and similar
8748 programs, such as kmscon) to run under the user's ID and
8749 access input and drm devices which are normally
8750 protected. When this is used (and the kernel is new enough)
8751 logind will "mute" IO on the file descriptors passed to
8752 Wayland as long as it is in the background and "unmute" it
8753 if it returns into the foreground. This allows secure
8754 session switching without allowing background sessions to
8755 eavesdrop on input and display data. This also introduces
8756 session switching support if VT support is turned off in the
8757 kernel, and on seats that are not seat0.
8759 * A new kernel command line option luks.options= is understood
8760 now which allows specifying LUKS options for usage for LUKS
8761 encrypted partitions specified with luks.uuid=.
8763 * tmpfiles.d(5) snippets may now use specifier expansion in
8764 path names. More specifically %m, %b, %H, %v, are now
8765 replaced by the local machine id, boot id, hostname, and
8766 kernel version number.
8768 * A new tmpfiles.d(5) command "m" has been introduced which
8769 may be used to change the owner/group/access mode of a file
8770 or directory if it exists, but do nothing if it does not.
8772 * This release removes high-level support for the
8773 MemorySoftLimit= cgroup setting. The underlying kernel
8774 cgroup attribute memory.soft_limit= is currently badly
8775 designed and likely to be removed from the kernel API in its
8776 current form, hence we should not expose it for now.
8778 * The memory.use_hierarchy cgroup attribute is now enabled for
8779 all cgroups systemd creates in the memory cgroup
8780 hierarchy. This option is likely to be come the built-in
8781 default in the kernel anyway, and the non-hierarchical mode
8782 never made much sense in the intrinsically hierarchical
8785 * A new field _SYSTEMD_SLICE= is logged along with all journal
8786 messages containing the slice a message was generated
8787 from. This is useful to allow easy per-customer filtering of
8788 logs among other things.
8790 * systemd-journald will no longer adjust the group of journal
8791 files it creates to the "systemd-journal" group. Instead we
8792 rely on the journal directory to be owned by the
8793 "systemd-journal" group, and its setgid bit set, so that the
8794 kernel file system layer will automatically enforce that
8795 journal files inherit this group assignment. The reason for
8796 this change is that we cannot allow NSS look-ups from
8797 journald which would be necessary to resolve
8798 "systemd-journal" to a numeric GID, because this might
8799 create deadlocks if NSS involves synchronous queries to
8800 other daemons (such as nscd, or sssd) which in turn are
8801 logging clients of journald and might block on it, which
8802 would then dead lock. A tmpfiles.d(5) snippet included in
8803 systemd will make sure the setgid bit and group are
8804 properly set on the journal directory if it exists on every
8805 boot. However, we recommend adjusting it manually after
8806 upgrades too (or from RPM scriptlets), so that the change is
8807 not delayed until next reboot.
8809 * Backlight and random seed files in /var/lib/ have moved into
8810 the /var/lib/systemd/ directory, in order to centralize all
8811 systemd generated files in one directory.
8813 * Boot time performance measurements (as displayed by
8814 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
8815 performance information if that's available to determine how
8816 much time BIOS and boot loader initialization required. With
8817 a sufficiently new BIOS you hence no longer need to boot
8818 with Gummiboot to get access to such information.
8820 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
8821 Cristian Rodríguez, Dave Reisner, David Herrmann, David
8822 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
8823 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
8824 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
8825 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
8826 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8828 — Berlin, 2013-10-02
8832 * The Restart= option for services now understands a new
8833 on-watchdog setting, which will restart the service
8834 automatically if the service stops sending out watchdog keep
8835 alive messages (as configured with WatchdogSec=).
8837 * The getty generator (which is responsible for bringing up a
8838 getty on configured serial consoles) will no longer only
8839 start a getty on the primary kernel console but on all
8840 others, too. This makes the order in which console= is
8841 specified on the kernel command line less important.
8843 * libsystemd-logind gained a new sd_session_get_vt() call to
8844 retrieve the VT number of a session.
8846 * If the option "tries=0" is set for an entry of /etc/crypttab
8847 its passphrase is queried indefinitely instead of any
8848 maximum number of tries.
8850 * If a service with a configure PID file terminates its PID
8851 file will now be removed automatically if it still exists
8852 afterwards. This should put an end to stale PID files.
8854 * systemd-run will now also take relative binary path names
8855 for execution and no longer insists on absolute paths.
8857 * InaccessibleDirectories= and ReadOnlyDirectories= now take
8858 paths that are optionally prefixed with "-" to indicate that
8859 it should not be considered a failure if they do not exist.
8861 * journalctl -o (and similar commands) now understands a new
8862 output mode "short-precise", it is similar to "short" but
8863 shows timestamps with usec accuracy.
8865 * The option "discard" (as known from Debian) is now
8866 synonymous to "allow-discards" in /etc/crypttab. In fact,
8867 "discard" is preferred now (since it is easier to remember
8870 * Some licensing clean-ups were made, so that more code is now
8871 LGPL-2.1 licensed than before.
8873 * A minimal tool to save/restore the display backlight
8874 brightness across reboots has been added. It will store the
8875 backlight setting as late as possible at shutdown, and
8876 restore it as early as possible during reboot.
8878 * A logic to automatically discover and enable home and swap
8879 partitions on GPT disks has been added. With this in place
8880 /etc/fstab becomes optional for many setups as systemd can
8881 discover certain partitions located on the root disk
8882 automatically. Home partitions are recognized under their
8883 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
8884 partitions are recognized under their GPT type ID
8885 0657fd6da4ab43c484e50933c84b4f4f.
8887 * systemd will no longer pass any environment from the kernel
8888 or initrd to system services. If you want to set an
8889 environment for all services, do so via the kernel command
8890 line systemd.setenv= assignment.
8892 * The systemd-sysctl tool no longer natively reads the file
8893 /etc/sysctl.conf. If desired, the file should be symlinked
8894 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
8895 legacy support by a symlink rather than built-in code, it
8896 also makes the otherwise hidden order of application of the
8897 different files visible. (Note that this partly reverts to a
8898 pre-198 application order of sysctl knobs!)
8900 * The "systemctl set-log-level" and "systemctl dump" commands
8901 have been moved to systemd-analyze.
8903 * systemd-run learned the new --remain-after-exit switch,
8904 which causes the scope unit not to be cleaned up
8905 automatically after the process terminated.
8907 * tmpfiles learned a new --exclude-prefix= switch to exclude
8908 certain paths from operation.
8910 * journald will now automatically flush all messages to disk
8911 as soon as a message at the log level CRIT, ALERT or EMERG
8914 Contributions from: Andrew Cook, Brandon Philips, Christian
8915 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
8916 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
8917 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
8918 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
8919 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
8920 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
8921 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
8922 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
8923 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
8924 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
8925 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
8926 William Giokas, Zbigniew Jędrzejewski-Szmek
8928 — Berlin, 2013-09-13
8932 * The documentation has been updated to cover the various new
8933 concepts introduced with 205.
8935 * Unit files now understand the new %v specifier which
8936 resolves to the kernel version string as returned by "uname
8939 * systemctl now supports filtering the unit list output by
8940 load state, active state and sub state, using the new
8943 * "systemctl status" will now show the results of the
8944 condition checks (like ConditionPathExists= and similar) of
8945 the last start attempts of the unit. They are also logged to
8948 * "journalctl -b" may now be used to look for boot output of a
8949 specific boot. Try "journalctl -b -1" for the previous boot,
8950 but the syntax is substantially more powerful.
8952 * "journalctl --show-cursor" has been added which prints the
8953 cursor string the last shown log line. This may then be used
8954 with the new "journalctl --after-cursor=" switch to continue
8955 browsing logs from that point on.
8957 * "journalctl --force" may now be used to force regeneration
8960 * Creation of "dead" device nodes has been moved from udev
8961 into kmod and tmpfiles. Previously, udev would read the kmod
8962 databases to pre-generate dead device nodes based on meta
8963 information contained in kernel modules, so that these would
8964 be auto-loaded on access rather then at boot. As this
8965 does not really have much to do with the exposing actual
8966 kernel devices to userspace this has always been slightly
8967 alien in the udev codebase. Following the new scheme kmod
8968 will now generate a runtime snippet for tmpfiles from the
8969 module meta information and it now is tmpfiles' job to the
8970 create the nodes. This also allows overriding access and
8971 other parameters for the nodes using the usual tmpfiles
8972 facilities. As side effect this allows us to remove the
8973 CAP_SYS_MKNOD capability bit from udevd entirely.
8975 * logind's device ACLs may now be applied to these "dead"
8976 devices nodes too, thus finally allowing managed access to
8977 devices such as /dev/snd/sequencer without loading the
8978 backing module right-away.
8980 * A new RPM macro has been added that may be used to apply
8981 tmpfiles configuration during package installation.
8983 * systemd-detect-virt and ConditionVirtualization= now can
8984 detect User-Mode-Linux machines (UML).
8986 * journald will now implicitly log the effective capabilities
8987 set of processes in the message metadata.
8989 * systemd-cryptsetup has gained support for TrueCrypt volumes.
8991 * The initrd interface has been simplified (more specifically,
8992 support for passing performance data via environment
8993 variables and fsck results via files in /run has been
8994 removed). These features were non-essential, and are
8995 nowadays available in a much nicer way by having systemd in
8996 the initrd serialize its state and have the hosts systemd
8997 deserialize it again.
8999 * The udev "keymap" data files and tools to apply keyboard
9000 specific mappings of scan to key codes, and force-release
9001 scan code lists have been entirely replaced by a udev
9002 "keyboard" builtin and a hwdb data file.
9004 * systemd will now honour the kernel's "quiet" command line
9005 argument also during late shutdown, resulting in a
9006 completely silent shutdown when used.
9008 * There's now an option to control the SO_REUSEPORT socket
9009 option in .socket units.
9011 * Instance units will now automatically get a per-template
9012 subslice of system.slice unless something else is explicitly
9013 configured. For example, instances of sshd@.service will now
9014 implicitly be placed in system-sshd.slice rather than
9015 system.slice as before.
9017 * Test coverage support may now be enabled at build time.
9019 Contributions from: Dave Reisner, Frederic Crozat, Harald
9020 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
9021 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
9022 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
9023 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
9024 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
9025 Giokas, Zbigniew Jędrzejewski-Szmek
9027 — Berlin, 2013-07-23
9031 * Two new unit types have been introduced:
9033 Scope units are very similar to service units, however, are
9034 created out of pre-existing processes — instead of PID 1
9035 forking off the processes. By using scope units it is
9036 possible for system services and applications to group their
9037 own child processes (worker processes) in a powerful way
9038 which then maybe used to organize them, or kill them
9039 together, or apply resource limits on them.
9041 Slice units may be used to partition system resources in an
9042 hierarchical fashion and then assign other units to them. By
9043 default there are now three slices: system.slice (for all
9044 system services), user.slice (for all user sessions),
9045 machine.slice (for VMs and containers).
9047 Slices and scopes have been introduced primarily in
9048 context of the work to move cgroup handling to a
9049 single-writer scheme, where only PID 1
9050 creates/removes/manages cgroups.
9052 * There's a new concept of "transient" units. In contrast to
9053 normal units these units are created via an API at runtime,
9054 not from configuration from disk. More specifically this
9055 means it is now possible to run arbitrary programs as
9056 independent services, with all execution parameters passed
9057 in via bus APIs rather than read from disk. Transient units
9058 make systemd substantially more dynamic then it ever was,
9059 and useful as a general batch manager.
9061 * logind has been updated to make use of scope and slice units
9062 for managing user sessions. As a user logs in he will get
9063 his own private slice unit, to which all sessions are added
9064 as scope units. We also added support for automatically
9065 adding an instance of user@.service for the user into the
9066 slice. Effectively logind will no longer create cgroup
9067 hierarchies on its own now, it will defer entirely to PID 1
9068 for this by means of scope, service and slice units. Since
9069 user sessions this way become entities managed by PID 1
9070 the output of "systemctl" is now a lot more comprehensive.
9072 * A new mini-daemon "systemd-machined" has been added which
9073 may be used by virtualization managers to register local
9074 VMs/containers. nspawn has been updated accordingly, and
9075 libvirt will be updated shortly. machined will collect a bit
9076 of meta information about the VMs/containers, and assign
9077 them their own scope unit (see above). The collected
9078 meta-data is then made available via the "machinectl" tool,
9079 and exposed in "ps" and similar tools. machined/machinectl
9080 is compile-time optional.
9082 * As discussed earlier, the low-level cgroup configuration
9083 options ControlGroup=, ControlGroupModify=,
9084 ControlGroupPersistent=, ControlGroupAttribute= have been
9085 removed. Please use high-level attribute settings instead as
9086 well as slice units.
9088 * A new bus call SetUnitProperties() has been added to alter
9089 various runtime parameters of a unit. This is primarily
9090 useful to alter cgroup parameters dynamically in a nice way,
9091 but will be extended later on to make more properties
9092 modifiable at runtime. systemctl gained a new set-properties
9093 command that wraps this call.
9095 * A new tool "systemd-run" has been added which can be used to
9096 run arbitrary command lines as transient services or scopes,
9097 while configuring a number of settings via the command
9098 line. This tool is currently very basic, however already
9099 very useful. We plan to extend this tool to even allow
9100 queuing of execution jobs with time triggers from the
9101 command line, similar in fashion to "at".
9103 * nspawn will now inform the user explicitly that kernels with
9104 audit enabled break containers, and suggest the user to turn
9107 * Support for detecting the IMA and AppArmor security
9108 frameworks with ConditionSecurity= has been added.
9110 * journalctl gained a new "-k" switch for showing only kernel
9111 messages, mimicking dmesg output; in addition to "--user"
9112 and "--system" switches for showing only user's own logs
9115 * systemd-delta can now show information about drop-in
9116 snippets extending unit files.
9118 * libsystemd-bus has been substantially updated but is still
9119 not available as public API.
9121 * systemd will now look for the "debug" argument on the kernel
9122 command line and enable debug logging, similar to what
9123 "systemd.log_level=debug" already did before.
9125 * "systemctl set-default", "systemctl get-default" has been
9126 added to configure the default.target symlink, which
9127 controls what to boot into by default.
9129 * "systemctl set-log-level" has been added as a convenient
9130 way to raise and lower systemd logging threshold.
9132 * "systemd-analyze plot" will now show the time the various
9133 generators needed for execution, as well as information
9134 about the unit file loading.
9136 * libsystemd-journal gained a new sd_journal_open_files() call
9137 for opening specific journal files. journactl also gained a
9138 new switch to expose this new functionality. Previously we
9139 only supported opening all files from a directory, or all
9140 files from the system, as opening individual files only is
9141 racy due to journal file rotation.
9143 * systemd gained the new DefaultEnvironment= setting in
9144 /etc/systemd/system.conf to set environment variables for
9147 * If a privileged process logs a journal message with the
9148 OBJECT_PID= field set, then journald will automatically
9149 augment this with additional OBJECT_UID=, OBJECT_GID=,
9150 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
9151 system services want to log events about specific client
9152 processes. journactl/systemctl has been updated to make use
9153 of this information if all log messages regarding a specific
9156 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
9157 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
9158 Reisner, David Coppa, David King, David Strauss, Eelco
9159 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
9160 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
9161 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
9162 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
9163 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
9164 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
9165 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
9166 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
9167 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
9168 Łukasz Stelmach, 장동준
9172 * The Python bindings gained some minimal support for the APIs
9173 exposed by libsystemd-logind.
9175 * ConditionSecurity= gained support for detecting SMACK. Since
9176 this condition already supports SELinux and AppArmor we only
9177 miss IMA for this. Patches welcome!
9179 Contributions from: Karol Lewandowski, Lennart Poettering,
9180 Zbigniew Jędrzejewski-Szmek
9184 * systemd-nspawn will now create /etc/resolv.conf if
9185 necessary, before bind-mounting the host's file onto it.
9187 * systemd-nspawn will now store meta information about a
9188 container on the container's cgroup as extended attribute
9189 fields, including the root directory.
9191 * The cgroup hierarchy has been reworked in many ways. All
9192 objects any of the components systemd creates in the cgroup
9193 tree are now suffixed. More specifically, user sessions are
9194 now placed in cgroups suffixed with ".session", users in
9195 cgroups suffixed with ".user", and nspawn containers in
9196 cgroups suffixed with ".nspawn". Furthermore, all cgroup
9197 names are now escaped in a simple scheme to avoid collision
9198 of userspace object names with kernel filenames. This work
9199 is preparation for making these objects relocatable in the
9200 cgroup tree, in order to allow easy resource partitioning of
9201 these objects without causing naming conflicts.
9203 * systemctl list-dependencies gained the new switches
9204 --plain, --reverse, --after and --before.
9206 * systemd-inhibit now shows the process name of processes that
9207 have taken an inhibitor lock.
9209 * nss-myhostname will now also resolve "localhost"
9210 implicitly. This makes /etc/hosts an optional file and
9211 nicely handles that on IPv6 ::1 maps to both "localhost" and
9214 * libsystemd-logind.so gained a new call
9215 sd_get_machine_names() to enumerate running containers and
9216 VMs (currently only supported by very new libvirt and
9217 nspawn). sd_login_monitor can now be used to watch
9218 VMs/containers coming and going.
9220 * .include is not allowed recursively anymore, and only in
9221 unit files. Usually it is better to use drop-in snippets in
9222 .d/*.conf anyway, as introduced with systemd 198.
9224 * systemd-analyze gained a new "critical-chain" command that
9225 determines the slowest chain of units run during system
9226 boot-up. It is very useful for tracking down where
9227 optimizing boot time is the most beneficial.
9229 * systemd will no longer allow manipulating service paths in
9230 the name=systemd:/system cgroup tree using ControlGroup= in
9231 units. (But is still fine with it in all other dirs.)
9233 * There's a new systemd-nspawn@.service service file that may
9234 be used to easily run nspawn containers as system
9235 services. With the container's root directory in
9236 /var/lib/container/foobar it is now sufficient to run
9237 "systemctl start systemd-nspawn@foobar.service" to boot it.
9239 * systemd-cgls gained a new parameter "--machine" to list only
9240 the processes within a certain container.
9242 * ConditionSecurity= now can check for "apparmor". We still
9243 are lacking checks for SMACK and IMA for this condition
9244 check though. Patches welcome!
9246 * A new configuration file /etc/systemd/sleep.conf has been
9247 added that may be used to configure which kernel operation
9248 systemd is supposed to execute when "suspend", "hibernate"
9249 or "hybrid-sleep" is requested. This makes the new kernel
9250 "freeze" state accessible to the user.
9252 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
9253 the passed argument if applicable.
9255 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
9256 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
9257 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
9258 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
9259 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
9260 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
9261 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
9266 * The output of 'systemctl list-jobs' got some polishing. The
9267 '--type=' argument may now be passed more than once. A new
9268 command 'systemctl list-sockets' has been added which shows
9269 a list of kernel sockets systemd is listening on with the
9270 socket units they belong to, plus the units these socket
9273 * The experimental libsystemd-bus library got substantial
9274 updates to work in conjunction with the (also experimental)
9275 kdbus kernel project. It works well enough to exchange
9276 messages with some sophistication. Note that kdbus is not
9277 ready yet, and the library is mostly an elaborate test case
9278 for now, and not installable.
9280 * systemd gained a new unit 'systemd-static-nodes.service'
9281 that generates static device nodes earlier during boot, and
9282 can run in conjunction with udev.
9284 * libsystemd-login gained a new call sd_pid_get_user_unit()
9285 to retrieve the user systemd unit a process is running
9286 in. This is useful for systems where systemd is used as
9289 * systemd-nspawn now places all containers in the new /machine
9290 top-level cgroup directory in the name=systemd
9291 hierarchy. libvirt will soon do the same, so that we get a
9292 uniform separation of /system, /user and /machine for system
9293 services, user processes and containers/virtual
9294 machines. This new cgroup hierarchy is also useful to stick
9295 stable names to specific container instances, which can be
9296 recognized later this way (this name may be controlled
9297 via systemd-nspawn's new -M switch). libsystemd-login also
9298 gained a new call sd_pid_get_machine_name() to retrieve the
9299 name of the container/VM a specific process belongs to.
9301 * bootchart can now store its data in the journal.
9303 * libsystemd-journal gained a new call
9304 sd_journal_add_conjunction() for AND expressions to the
9305 matching logic. This can be used to express more complex
9306 logical expressions.
9308 * journactl can now take multiple --unit= and --user-unit=
9311 * The cryptsetup logic now understands the "luks.key=" kernel
9312 command line switch for specifying a file to read the
9313 decryption key from. Also, if a configured key file is not
9314 found the tool will now automatically fall back to prompting
9317 * Python systemd.journal module was updated to wrap recently
9318 added functions from libsystemd-journal. The interface was
9319 changed to bring the low level interface in s.j._Reader
9320 closer to the C API, and the high level interface in
9321 s.j.Reader was updated to wrap and convert all data about
9324 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
9325 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
9326 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
9327 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
9328 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
9329 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9333 * journalctl --update-catalog now understands a new --root=
9334 option to operate on catalogs found in a different root
9337 * During shutdown after systemd has terminated all running
9338 services a final killing loop kills all remaining left-over
9339 processes. We will now print the name of these processes
9340 when we send SIGKILL to them, since this usually indicates a
9343 * If /etc/crypttab refers to password files stored on
9344 configured mount points automatic dependencies will now be
9345 generated to ensure the specific mount is established first
9346 before the key file is attempted to be read.
9348 * 'systemctl status' will now show information about the
9349 network sockets a socket unit is listening on.
9351 * 'systemctl status' will also shown information about any
9352 drop-in configuration file for units. (Drop-In configuration
9353 files in this context are files such as
9354 /etc/systemd/systemd/foobar.service.d/*.conf)
9356 * systemd-cgtop now optionally shows summed up CPU times of
9357 cgroups. Press '%' while running cgtop to switch between
9358 percentage and absolute mode. This is useful to determine
9359 which cgroups use up the most CPU time over the entire
9360 runtime of the system. systemd-cgtop has also been updated
9361 to be 'pipeable' for processing with further shell tools.
9363 * 'hostnamectl set-hostname' will now allow setting of FQDN
9366 * The formatting and parsing of time span values has been
9367 changed. The parser now understands fractional expressions
9368 such as "5.5h". The formatter will now output fractional
9369 expressions for all time spans under 1min, i.e. "5.123456s"
9370 rather than "5s 123ms 456us". For time spans under 1s
9371 millisecond values are shown, for those under 1ms
9372 microsecond values are shown. This should greatly improve
9373 all time-related output of systemd.
9375 * libsystemd-login and libsystemd-journal gained new
9376 functions for querying the poll() events mask and poll()
9377 timeout value for integration into arbitrary event
9380 * localectl gained the ability to list available X11 keymaps
9381 (models, layouts, variants, options).
9383 * 'systemd-analyze dot' gained the ability to filter for
9384 specific units via shell-style globs, to create smaller,
9385 more useful graphs. I.e. it is now possible to create simple
9386 graphs of all the dependencies between only target units, or
9387 of all units that Avahi has dependencies with.
9389 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
9390 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
9391 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
9392 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
9393 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
9394 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
9395 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
9399 * The boot-time readahead implementation for rotating media
9400 will now read the read-ahead data in multiple passes which
9401 consist of all read requests made in equidistant time
9402 intervals. This means instead of strictly reading read-ahead
9403 data in its physical order on disk we now try to find a
9404 middle ground between physical and access time order.
9406 * /etc/os-release files gained a new BUILD_ID= field for usage
9407 on operating systems that provide continuous builds of OS
9410 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
9411 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
9412 William Douglas, Zbigniew Jędrzejewski-Szmek
9416 * systemd-python gained an API exposing libsystemd-daemon.
9418 * The SMACK setup logic gained support for uploading CIPSO
9421 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
9422 ReadOnlyDirectories= and InaccessibleDirectories= has
9423 changed. The private /tmp and /var/tmp directories are now
9424 shared by all processes of a service (which means
9425 ExecStartPre= may now leave data in /tmp that ExecStart= of
9426 the same service can still access). When a service is
9427 stopped its temporary directories are immediately deleted
9428 (normal clean-up with tmpfiles is still done in addition to
9431 * By default, systemd will now set a couple of sysctl
9432 variables in the kernel: the safe sysrq options are turned
9433 on, IP route verification is turned on, and source routing
9434 disabled. The recently added hardlink and softlink
9435 protection of the kernel is turned on. These settings should
9436 be reasonably safe, and good defaults for all new systems.
9438 * The predictable network naming logic may now be turned off
9439 with a new kernel command line switch: net.ifnames=0.
9441 * A new libsystemd-bus module has been added that implements a
9442 pretty complete D-Bus client library. For details see:
9444 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
9446 * journald will now explicitly flush the journal files to disk
9447 at the latest 5min after each write. The file will then also
9448 be marked offline until the next write. This should increase
9449 reliability in case of a crash. The synchronization delay
9450 can be configured via SyncIntervalSec= in journald.conf.
9452 * There's a new remote-fs-setup.target unit that can be used
9453 to pull in specific services when at least one remote file
9454 system is to be mounted.
9456 * There are new targets timers.target and paths.target as
9457 canonical targets to pull user timer and path units in
9458 from. This complements sockets.target with a similar
9459 purpose for socket units.
9461 * libudev gained a new call udev_device_set_attribute_value()
9462 to set sysfs attributes of a device.
9464 * The udev daemon now sets the default number of worker
9465 processes executed in parallel based on the number of available
9466 CPUs instead of the amount of available RAM. This is supposed
9467 to provide a more reliable default and limit a too aggressive
9468 parallelism for setups with 1000s of devices connected.
9470 Contributions from: Auke Kok, Colin Walters, Cristian
9471 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
9472 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
9473 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
9474 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
9475 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
9476 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
9477 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
9478 Zbigniew Jędrzejewski-Szmek
9482 * Configuration of unit files may now be extended via drop-in
9483 files without having to edit/override the unit files
9484 themselves. More specifically, if the administrator wants to
9485 change one value for a service file foobar.service he can
9486 now do so by dropping in a configuration snippet into
9487 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
9488 will load all these snippets and apply them on top of the
9489 main unit configuration file, possibly extending or
9490 overriding its settings. Using these drop-in snippets is
9491 generally nicer than the two earlier options for changing
9492 unit files locally: copying the files from
9493 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
9494 them there; or creating a new file in /etc/systemd/system/
9495 that incorporates the original one via ".include". Drop-in
9496 snippets into these .d/ directories can be placed in any
9497 directory systemd looks for units in, and the usual
9498 overriding semantics between /usr/lib, /etc and /run apply
9501 * Most unit file settings which take lists of items can now be
9502 reset by assigning the empty string to them. For example,
9503 normally, settings such as Environment=FOO=BAR append a new
9504 environment variable assignment to the environment block,
9505 each time they are used. By assigning Environment= the empty
9506 string the environment block can be reset to empty. This is
9507 particularly useful with the .d/*.conf drop-in snippets
9508 mentioned above, since this adds the ability to reset list
9509 settings from vendor unit files via these drop-ins.
9511 * systemctl gained a new "list-dependencies" command for
9512 listing the dependencies of a unit recursively.
9514 * Inhibitors are now honored and listed by "systemctl
9515 suspend", "systemctl poweroff" (and similar) too, not only
9516 GNOME. These commands will also list active sessions by
9519 * Resource limits (as exposed by the various control group
9520 controllers) can now be controlled dynamically at runtime
9521 for all units. More specifically, you can now use a command
9522 like "systemctl set-cgroup-attr foobar.service cpu.shares
9523 2000" to alter the CPU shares a specific service gets. These
9524 settings are stored persistently on disk, and thus allow the
9525 administrator to easily adjust the resource usage of
9526 services with a few simple commands. This dynamic resource
9527 management logic is also available to other programs via the
9528 bus. Almost any kernel cgroup attribute and controller is
9531 * systemd-vconsole-setup will now copy all font settings to
9532 all allocated VTs, where it previously applied them only to
9535 * libsystemd-login gained the new sd_session_get_tty() API
9538 * This release drops support for a few legacy or
9539 distribution-specific LSB facility names when parsing init
9540 scripts: $x-display-manager, $mail-transfer-agent,
9541 $mail-transport-agent, $mail-transfer-agent, $smtp,
9542 $null. Also, the mail-transfer-agent.target unit backing
9543 this has been removed. Distributions which want to retain
9544 compatibility with this should carry the burden for
9545 supporting this themselves and patch support for these back
9546 in, if they really need to. Also, the facilities $syslog and
9547 $local_fs are now ignored, since systemd does not support
9548 early-boot LSB init scripts anymore, and these facilities
9549 are implied anyway for normal services. syslog.target has
9552 * There are new bus calls on PID1's Manager object for
9553 cancelling jobs, and removing snapshot units. Previously,
9554 both calls were only available on the Job and Snapshot
9557 * systemd-journal-gatewayd gained SSL support.
9559 * The various "environment" files, such as /etc/locale.conf
9560 now support continuation lines with a backslash ("\") as
9561 last character in the line, similarly in style (but different)
9562 to how this is supported in shells.
9564 * For normal user processes the _SYSTEMD_USER_UNIT= field is
9565 now implicitly appended to every log entry logged. systemctl
9566 has been updated to filter by this field when operating on a
9567 user systemd instance.
9569 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
9570 CAP_AUDIT_CONTROL capabilities to the capabilities set for
9571 the container. This makes it easier to boot unmodified
9572 Fedora systems in a container, which however still requires
9573 audit=0 to be passed on the kernel command line. Auditing in
9574 kernel and userspace is unfortunately still too broken in
9575 context of containers, hence we recommend compiling it out
9576 of the kernel or using audit=0. Hopefully this will be fixed
9577 one day for good in the kernel.
9579 * nspawn gained the new --bind= and --bind-ro= parameters to
9580 bind mount specific directories from the host into the
9583 * nspawn will now mount its own devpts file system instance
9584 into the container, in order not to leak pty devices from
9585 the host into the container.
9587 * systemd will now read the firmware boot time performance
9588 information from the EFI variables, if the used boot loader
9589 supports this, and takes it into account for boot performance
9590 analysis via "systemd-analyze". This is currently supported
9591 only in conjunction with Gummiboot, but could be supported
9592 by other boot loaders too. For details see:
9594 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
9596 * A new generator has been added that automatically mounts the
9597 EFI System Partition (ESP) to /boot, if that directory
9598 exists, is empty, and no other file system has been
9599 configured to be mounted there.
9601 * logind will now send out PrepareForSleep(false) out
9602 unconditionally, after coming back from suspend. This may be
9603 used by applications as asynchronous notification for
9604 system resume events.
9606 * "systemctl unlock-sessions" has been added, that allows
9607 unlocking the screens of all user sessions at once, similar
9608 to how "systemctl lock-sessions" already locked all users
9609 sessions. This is backed by a new D-Bus call UnlockSessions().
9611 * "loginctl seat-status" will now show the master device of a
9612 seat. (i.e. the device of a seat that needs to be around for
9613 the seat to be considered available, usually the graphics
9616 * tmpfiles gained a new "X" line type, that allows
9617 configuration of files and directories (with wildcards) that
9618 shall be excluded from automatic cleanup ("aging").
9620 * udev default rules set the device node permissions now only
9621 at "add" events, and do not change them any longer with a
9622 later "change" event.
9624 * The log messages for lid events and power/sleep keypresses
9625 now carry a message ID.
9627 * We now have a substantially larger unit test suite, but this
9628 continues to be work in progress.
9630 * udevadm hwdb gained a new --root= parameter to change the
9631 root directory to operate relative to.
9633 * logind will now issue a background sync() request to the kernel
9634 early at shutdown, so that dirty buffers are flushed to disk early
9635 instead of at the last moment, in order to optimize shutdown
9638 * A new bootctl tool has been added that is an interface for
9639 certain boot loader operations. This is currently a preview
9640 and is likely to be extended into a small mechanism daemon
9641 like timedated, localed, hostnamed, and can be used by
9642 graphical UIs to enumerate available boot options, and
9643 request boot into firmware operations.
9645 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
9646 the rest of the package. It also has been updated to work
9647 correctly in initrds.
9649 * polkit previously has been runtime optional, and is now also
9650 compile time optional via a configure switch.
9652 * systemd-analyze has been reimplemented in C. Also "systemctl
9653 dot" has moved into systemd-analyze.
9655 * "systemctl status" with no further parameters will now print
9656 the status of all active or failed units.
9658 * Operations such as "systemctl start" can now be executed
9659 with a new mode "--irreversible" which may be used to queue
9660 operations that cannot accidentally be reversed by a later
9661 job queuing. This is by default used to make shutdown
9662 requests more robust.
9664 * The Python API of systemd now gained a new module for
9665 reading journal files.
9667 * A new tool kernel-install has been added that can install
9668 kernel images according to the Boot Loader Specification:
9670 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
9672 * Boot time console output has been improved to provide
9673 animated boot time output for hanging jobs.
9675 * A new tool systemd-activate has been added which can be used
9676 to test socket activation with, directly from the command
9677 line. This should make it much easier to test and debug
9678 socket activation in daemons.
9680 * journalctl gained a new "--reverse" (or -r) option to show
9681 journal output in reverse order (i.e. newest line first).
9683 * journalctl gained a new "--pager-end" (or -e) option to jump
9684 to immediately jump to the end of the journal in the
9685 pager. This is only supported in conjunction with "less".
9687 * journalctl gained a new "--user-unit=" option, that works
9688 similarly to "--unit=" but filters for user units rather than
9691 * A number of unit files to ease adoption of systemd in
9692 initrds has been added. This moves some minimal logic from
9693 the various initrd implementations into systemd proper.
9695 * The journal files are now owned by a new group
9696 "systemd-journal", which exists specifically to allow access
9697 to the journal, and nothing else. Previously, we used the
9698 "adm" group for that, which however possibly covers more
9699 than just journal/log file access. This new group is now
9700 already used by systemd-journal-gatewayd to ensure this
9701 daemon gets access to the journal files and as little else
9702 as possible. Note that "make install" will also set FS ACLs
9703 up for /var/log/journal to give "adm" and "wheel" read
9704 access to it, in addition to "systemd-journal" which owns
9705 the journal files. We recommend that packaging scripts also
9706 add read access to "adm" + "wheel" to /var/log/journal, and
9707 all existing/future journal files. To normal users and
9708 administrators little changes, however packagers need to
9709 ensure to create the "systemd-journal" system group at
9710 package installation time.
9712 * The systemd-journal-gatewayd now runs as unprivileged user
9713 systemd-journal-gateway:systemd-journal-gateway. Packaging
9714 scripts need to create these system user/group at
9717 * timedated now exposes a new boolean property CanNTP that
9718 indicates whether a local NTP service is available or not.
9720 * systemd-detect-virt will now also detect xen PVs
9722 * The pstore file system is now mounted by default, if it is
9725 * In addition to the SELinux and IMA policies we will now also
9726 load SMACK policies at early boot.
9728 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
9729 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
9730 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
9731 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
9732 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
9733 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
9734 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
9735 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
9736 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
9737 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
9738 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
9739 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
9740 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
9741 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
9745 * Timer units now support calendar time events in addition to
9746 monotonic time events. That means you can now trigger a unit
9747 based on a calendar time specification such as "Thu,Fri
9748 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
9749 or fifth day of any month of the year 2013, given that it is
9750 a thursday or friday. This brings timer event support
9751 considerably closer to cron's capabilities. For details on
9752 the supported calendar time specification language see
9755 * udev now supports a number of different naming policies for
9756 network interfaces for predictable names, and a combination
9757 of these policies is now the default. Please see this wiki
9758 document for details:
9760 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
9762 * Auke Kok's bootchart implementation has been added to the
9763 systemd tree. It is an optional component that can graph the
9764 boot in quite some detail. It is one of the best bootchart
9765 implementations around and minimal in its code and
9768 * nss-myhostname has been integrated into the systemd source
9769 tree. nss-myhostname guarantees that the local hostname
9770 always stays resolvable via NSS. It has been a weak
9771 requirement of systemd-hostnamed since a long time, and
9772 since its code is actually trivial we decided to just
9773 include it in systemd's source tree. It can be turned off
9774 with a configure switch.
9776 * The read-ahead logic is now capable of properly detecting
9777 whether a btrfs file system is on SSD or rotating media, in
9778 order to optimize the read-ahead scheme. Previously, it was
9779 only capable of detecting this on traditional file systems
9782 * In udev, additional device properties are now read from the
9783 IAB in addition to the OUI database. Also, Bluetooth company
9784 identities are attached to the devices as well.
9786 * In service files %U may be used as specifier that is
9787 replaced by the configured user name of the service.
9789 * nspawn may now be invoked without a controlling TTY. This
9790 makes it suitable for invocation as its own service. This
9791 may be used to set up a simple containerized server system
9792 using only core OS tools.
9794 * systemd and nspawn can now accept socket file descriptors
9795 when they are started for socket activation. This enables
9796 implementation of socket activated nspawn
9797 containers. i.e. think about autospawning an entire OS image
9798 when the first SSH or HTTP connection is received. We expect
9799 that similar functionality will also be added to libvirt-lxc
9802 * journalctl will now suppress ANSI color codes when
9803 presenting log data.
9805 * systemctl will no longer show control group information for
9806 a unit if the control group is empty anyway.
9808 * logind can now automatically suspend/hibernate/shutdown the
9811 * /etc/machine-info and hostnamed now also expose the chassis
9812 type of the system. This can be used to determine whether
9813 the local system is a laptop, desktop, handset or
9814 tablet. This information may either be configured by the
9815 user/vendor or is automatically determined from ACPI and DMI
9816 information if possible.
9818 * A number of polkit actions are now bound together with "imply"
9819 rules. This should simplify creating UIs because many actions
9820 will now authenticate similar ones as well.
9822 * Unit files learnt a new condition ConditionACPower= which
9823 may be used to conditionalize a unit depending on whether an
9824 AC power source is connected or not, of whether the system
9825 is running on battery power.
9827 * systemctl gained a new "is-failed" verb that may be used in
9828 shell scripts and suchlike to check whether a specific unit
9829 is in the "failed" state.
9831 * The EnvironmentFile= setting in unit files now supports file
9832 globbing, and can hence be used to easily read a number of
9833 environment files at once.
9835 * systemd will no longer detect and recognize specific
9836 distributions. All distribution-specific #ifdeffery has been
9837 removed, systemd is now fully generic and
9838 distribution-agnostic. Effectively, not too much is lost as
9839 a lot of the code is still accessible via explicit configure
9840 switches. However, support for some distribution specific
9841 legacy configuration file formats has been dropped. We
9842 recommend distributions to simply adopt the configuration
9843 files everybody else uses now and convert the old
9844 configuration from packaging scripts. Most distributions
9845 already did that. If that's not possible or desirable,
9846 distributions are welcome to forward port the specific
9847 pieces of code locally from the git history.
9849 * When logging a message about a unit systemd will now always
9850 log the unit name in the message meta data.
9852 * localectl will now also discover system locale data that is
9853 not stored in locale archives, but directly unpacked.
9855 * logind will no longer unconditionally use framebuffer
9856 devices as seat masters, i.e. as devices that are required
9857 to be existing before a seat is considered preset. Instead,
9858 it will now look for all devices that are tagged as
9859 "seat-master" in udev. By default, framebuffer devices will
9860 be marked as such, but depending on local systems, other
9861 devices might be marked as well. This may be used to
9862 integrate graphics cards using closed source drivers (such
9863 as NVidia ones) more nicely into logind. Note however, that
9864 we recommend using the open source NVidia drivers instead,
9865 and no udev rules for the closed-source drivers will be
9866 shipped from us upstream.
9868 Contributions from: Adam Williamson, Alessandro Crismani, Auke
9869 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
9870 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
9871 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
9872 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
9873 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
9874 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
9875 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
9876 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
9877 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
9878 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
9879 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
9884 * udev gained support for loading additional device properties
9885 from an indexed database that is keyed by vendor/product IDs
9886 and similar device identifiers. For the beginning this
9887 "hwdb" is populated with data from the well-known PCI and
9888 USB database, but also includes PNP, ACPI and OID data. In
9889 the longer run this indexed database shall grow into
9890 becoming the one central database for non-essential
9891 userspace device metadata. Previously, data from the PCI/USB
9892 database was only attached to select devices, since the
9893 lookup was a relatively expensive operation due to O(n) time
9894 complexity (with n being the number of entries in the
9895 database). Since this is now O(1), we decided to add in this
9896 data for all devices where this is available, by
9897 default. Note that the indexed database needs to be rebuilt
9898 when new data files are installed. To achieve this you need
9899 to update your packaging scripts to invoke "udevadm hwdb
9900 --update" after installation of hwdb data files. For
9901 RPM-based distributions we introduced the new
9902 %udev_hwdb_update macro for this purpose.
9904 * The Journal gained support for the "Message Catalog", an
9905 indexed database to link up additional information with
9906 journal entries. For further details please check:
9908 https://www.freedesktop.org/wiki/Software/systemd/catalog
9910 The indexed message catalog database also needs to be
9911 rebuilt after installation of message catalog files. Use
9912 "journalctl --update-catalog" for this. For RPM-based
9913 distributions we introduced the %journal_catalog_update
9914 macro for this purpose.
9916 * The Python Journal bindings gained support for the standard
9917 Python logging framework.
9919 * The Journal API gained new functions for checking whether
9920 the underlying file system of a journal file is capable of
9921 properly reporting file change notifications, or whether
9922 applications that want to reflect journal changes "live"
9923 need to recheck journal files continuously in appropriate
9926 * It is now possible to set the "age" field for tmpfiles
9927 entries to 0, indicating that files matching this entry
9928 shall always be removed when the directories are cleaned up.
9930 * coredumpctl gained a new "gdb" verb which invokes gdb
9931 right-away on the selected coredump.
9933 * There's now support for "hybrid sleep" on kernels that
9934 support this, in addition to "suspend" and "hibernate". Use
9935 "systemctl hybrid-sleep" to make use of this.
9937 * logind's HandleSuspendKey= setting (and related settings)
9938 now gained support for a new "lock" setting to simply
9939 request the screen lock on all local sessions, instead of
9940 actually executing a suspend or hibernation.
9942 * systemd will now mount the EFI variables file system by
9945 * Socket units now gained support for configuration of the
9946 SMACK security label.
9948 * timedatectl will now output the time of the last and next
9949 daylight saving change.
9951 * We dropped support for various legacy and distro-specific
9952 concepts, such as insserv, early-boot SysV services
9953 (i.e. those for non-standard runlevels such as 'b' or 'S')
9954 or ArchLinux /etc/rc.conf support. We recommend the
9955 distributions who still need support this to either continue
9956 to maintain the necessary patches downstream, or find a
9957 different solution. (Talk to us if you have questions!)
9959 * Various systemd components will now bypass polkit checks for
9960 root and otherwise handle properly if polkit is not found to
9961 be around. This should fix most issues for polkit-less
9962 systems. Quite frankly this should have been this way since
9963 day one. It is absolutely our intention to make systemd work
9964 fine on polkit-less systems, and we consider it a bug if
9965 something does not work as it should if polkit is not around.
9967 * For embedded systems it is now possible to build udev and
9968 systemd without blkid and/or kmod support.
9970 * "systemctl switch-root" is now capable of switching root
9971 more than once. I.e. in addition to transitions from the
9972 initrd to the host OS it is now possible to transition to
9973 further OS images from the host. This is useful to implement
9974 offline updating tools.
9976 * Various other additions have been made to the RPM macros
9977 shipped with systemd. Use %udev_rules_update() after
9978 installing new udev rules files. %_udevhwdbdir,
9979 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
9980 %_sysctldir are now available which resolve to the right
9981 directories for packages to place various data files in.
9983 * journalctl gained the new --full switch (in addition to
9984 --all, to disable ellipsation for long messages.
9986 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
9987 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
9988 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
9989 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
9990 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
9991 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
9992 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
9993 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
9994 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
9998 * journalctl gained new --since= and --until= switches to
9999 filter by time. It also now supports nice filtering for
10000 units via --unit=/-u.
10002 * Type=oneshot services may use ExecReload= and do the
10005 * The journal daemon now supports time-based rotation and
10006 vacuuming, in addition to the usual disk-space based
10009 * The journal will now index the available field values for
10010 each field name. This enables clients to show pretty drop
10011 downs of available match values when filtering. The bash
10012 completion of journalctl has been updated
10013 accordingly. journalctl gained a new switch -F to list all
10014 values a certain field takes in the journal database.
10016 * More service events are now written as structured messages
10017 to the journal, and made recognizable via message IDs.
10019 * The timedated, localed and hostnamed mini-services which
10020 previously only provided support for changing time, locale
10021 and hostname settings from graphical DEs such as GNOME now
10022 also have a minimal (but very useful) text-based client
10023 utility each. This is probably the nicest way to changing
10024 these settings from the command line now, especially since
10025 it lists available options and is fully integrated with bash
10028 * There's now a new tool "systemd-coredumpctl" to list and
10029 extract coredumps from the journal.
10031 * We now install a README each in /var/log/ and
10032 /etc/rc.d/init.d explaining where the system logs and init
10033 scripts went. This hopefully should help folks who go to
10034 that dirs and look into the otherwise now empty void and
10035 scratch their heads.
10037 * When user-services are invoked (by systemd --user) the
10038 $MANAGERPID env var is set to the PID of systemd.
10040 * SIGRTMIN+24 when sent to a --user instance will now result
10041 in immediate termination of systemd.
10043 * gatewayd received numerous feature additions such as a
10044 "follow" mode, for live syncing and filtering.
10046 * browse.html now allows filtering and showing detailed
10047 information on specific entries. Keyboard navigation and
10048 mouse screen support has been added.
10050 * gatewayd/journalctl now supports HTML5/JSON
10051 Server-Sent-Events as output.
10053 * The SysV init script compatibility logic will now
10054 heuristically determine whether a script supports the
10055 "reload" verb, and only then make this available as
10056 "systemctl reload".
10058 * "systemctl status --follow" has been removed, use "journalctl
10061 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
10062 have been removed since they are hardly useful to be
10065 * And I'd like to take the opportunity to specifically mention
10066 Zbigniew for his great contributions. Zbigniew, you rock!
10068 Contributions from: Andrew Eikum, Christian Hesse, Colin
10069 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
10070 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
10071 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
10072 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
10073 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
10074 Jędrzejewski-Szmek, Сковорода Никита Андреевич
10078 * If /etc/vconsole.conf is non-existent or empty we will no
10079 longer load any console font or key map at boot by
10080 default. Instead the kernel defaults will be left
10081 intact. This is definitely the right thing to do, as no
10082 configuration should mean no configuration, and hard-coding
10083 font names that are different on all archs is probably a bad
10084 idea. Also, the kernel default key map and font should be
10085 good enough for most cases anyway, and mostly identical to
10086 the userspace fonts/key maps we previously overloaded them
10087 with. If distributions want to continue to default to a
10088 non-kernel font or key map they should ship a default
10089 /etc/vconsole.conf with the appropriate contents.
10091 Contributions from: Colin Walters, Daniel J Walsh, Dave
10092 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
10093 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
10097 * journalctl gained a new --cursor= switch to show entries
10098 starting from the specified location in the journal.
10100 * We now enforce a size limit on journal entry fields exported
10101 with "-o json" in journalctl. Fields larger than 4K will be
10102 assigned null. This can be turned off with --all.
10104 * An (optional) journal gateway daemon is now available as
10105 "systemd-journal-gatewayd.service". This service provides
10106 access to the journal via HTTP and JSON. This functionality
10107 will be used to implement live log synchronization in both
10108 pull and push modes, but has various other users too, such
10109 as easy log access for debugging of embedded devices. Right
10110 now it is already useful to retrieve the journal via HTTP:
10112 # systemctl start systemd-journal-gatewayd.service
10113 # wget http://localhost:19531/entries
10115 This will download the journal contents in a
10116 /var/log/messages compatible format. The same as JSON:
10118 # curl -H"Accept: application/json" http://localhost:19531/entries
10120 This service is also accessible via a web browser where a
10121 single static HTML5 app is served that uses the JSON logic
10122 to enable the user to do some basic browsing of the
10123 journal. This will be extended later on. Here's an example
10124 screenshot of this app in its current state:
10126 http://0pointer.de/public/journal-gatewayd
10128 Contributions from: Kay Sievers, Lennart Poettering, Robert
10129 Milasan, Tom Gundersen
10133 * The bash completion logic is now available for journalctl
10136 * We do not mount the "cpuset" controller anymore together with
10137 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
10138 started if no parameters are assigned to it. "cpuset" hence
10139 broke code that assumed it could create "cpu" groups and
10142 * journalctl -f will now subscribe to terminal size changes,
10143 and line break accordingly.
10145 Contributions from: Dave Reisner, Kay Sievers, Lennart
10146 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
10150 * nspawn will now create a symlink /etc/localtime in the
10151 container environment, copying the host's timezone
10152 setting. Previously this has been done via a bind mount, but
10153 since symlinks cannot be bind mounted this has now been
10154 changed to create/update the appropriate symlink.
10156 * journalctl -n's line number argument is now optional, and
10157 will default to 10 if omitted.
10159 * journald will now log the maximum size the journal files may
10160 take up on disk. This is particularly useful if the default
10161 built-in logic of determining this parameter from the file
10162 system size is used. Use "systemctl status
10163 systemd-journald.service" to see this information.
10165 * The multi-seat X wrapper tool has been stripped down. As X
10166 is now capable of enumerating graphics devices via udev in a
10167 seat-aware way the wrapper is not strictly necessary
10168 anymore. A stripped down temporary stop-gap is still shipped
10169 until the upstream display managers have been updated to
10170 fully support the new X logic. Expect this wrapper to be
10171 removed entirely in one of the next releases.
10173 * HandleSleepKey= in logind.conf has been split up into
10174 HandleSuspendKey= and HandleHibernateKey=. The old setting
10175 is not available anymore. X11 and the kernel are
10176 distinguishing between these keys and we should too. This
10177 also means the inhibition lock for these keys has been split
10180 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
10181 Poettering, Lukas Nykryn, Václav Pavlín
10185 * Whenever a unit changes state we will now log this to the
10186 journal and show along the unit's own log output in
10187 "systemctl status".
10189 * ConditionPathIsMountPoint= can now properly detect bind
10190 mount points too. (Previously, a bind mount of one file
10191 system to another place in the same file system could not be
10192 detected as mount, since they shared struct stat's st_dev
10195 * We will now mount the cgroup controllers cpu, cpuacct,
10196 cpuset and the controllers net_cls, net_prio together by
10199 * nspawn containers will now have a virtualized boot
10200 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
10201 over with a randomized ID at container initialization). This
10202 has the effect of making "journalctl -b" do the right thing
10205 * The JSON output journal serialization has been updated not
10206 to generate "endless" list objects anymore, but rather one
10207 JSON object per line. This is more in line how most JSON
10208 parsers expect JSON objects. The new output mode
10209 "json-pretty" has been added to provide similar output, but
10210 neatly aligned for readability by humans.
10212 * We dropped all explicit sync() invocations in the shutdown
10213 code. The kernel does this implicitly anyway in the kernel
10214 reboot() syscall. halt(8)'s -n option is now a compatibility
10217 * We now support virtualized reboot() in containers, as
10218 supported by newer kernels. We will fall back to exit() if
10219 CAP_SYS_REBOOT is not available to the container. Also,
10220 nspawn makes use of this now and will actually reboot the
10221 container if the containerized OS asks for that.
10223 * journalctl will only show local log output by default
10224 now. Use --merge (-m) to show remote log output, too.
10226 * libsystemd-journal gained the new sd_journal_get_usage()
10227 call to determine the current disk usage of all journal
10228 files. This is exposed in the new "journalctl --disk-usage"
10231 * journald gained a new configuration setting SplitMode= in
10232 journald.conf which may be used to control how user journals
10233 are split off. See journald.conf(5) for details.
10235 * A new condition type ConditionFileNotEmpty= has been added.
10237 * tmpfiles' "w" lines now support file globbing, to write
10238 multiple files at once.
10240 * We added Python bindings for the journal submission
10241 APIs. More Python APIs for a number of selected APIs will
10242 likely follow. Note that we intend to add native bindings
10243 only for the Python language, as we consider it common
10244 enough to deserve bindings shipped within systemd. There are
10245 various projects outside of systemd that provide bindings
10246 for languages such as PHP or Lua.
10248 * Many conditions will now resolve specifiers such as %i. In
10249 addition, PathChanged= and related directives of .path units
10250 now support specifiers as well.
10252 * There's now a new RPM macro definition for the system preset
10255 * journald will now warn if it ca not forward a message to the
10256 syslog daemon because its socket is full.
10258 * timedated will no longer write or process /etc/timezone,
10259 except on Debian. As we do not support late mounted /usr
10260 anymore /etc/localtime always being a symlink is now safe,
10261 and hence the information in /etc/timezone is not necessary
10264 * logind will now always reserve one VT for a text getty (VT6
10265 by default). Previously if more than 6 X sessions where
10266 started they took up all the VTs with auto-spawned gettys,
10267 so that no text gettys were available anymore.
10269 * udev will now automatically inform the btrfs kernel logic
10270 about btrfs RAID components showing up. This should make
10271 simple hotplug based btrfs RAID assembly work.
10273 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
10274 (but not for its children which will stay at the kernel
10275 default). This should allow setups with a lot more listening
10278 * systemd will now always pass the configured timezone to the
10279 kernel at boot. timedated will do the same when the timezone
10282 * logind's inhibition logic has been updated. By default,
10283 logind will now handle the lid switch, the power and sleep
10284 keys all the time, even in graphical sessions. If DEs want
10285 to handle these events on their own they should take the new
10286 handle-power-key, handle-sleep-key and handle-lid-switch
10287 inhibitors during their runtime. A simple way to achieve
10288 that is to invoke the DE wrapped in an invocation of:
10290 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
10292 * Access to unit operations is now checked via SELinux taking
10293 the unit file label and client process label into account.
10295 * systemd will now notify the administrator in the journal
10296 when he over-mounts a non-empty directory.
10298 * There are new specifiers that are resolved in unit files,
10299 for the hostname (%H), the machine ID (%m) and the boot ID
10302 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
10303 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
10304 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
10305 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
10306 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
10307 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
10308 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
10312 * Support for reading structured kernel messages from
10313 /dev/kmsg has now been added and is enabled by default.
10315 * Support for reading kernel messages from /proc/kmsg has now
10316 been removed. If you want kernel messages in the journal
10317 make sure to run a recent kernel (>= 3.5) that supports
10318 reading structured messages from /dev/kmsg (see
10319 above). /proc/kmsg is now exclusive property of classic
10320 syslog daemons again.
10322 * The libudev API gained the new
10323 udev_device_new_from_device_id() call.
10325 * The logic for file system namespace (ReadOnlyDirectory=,
10326 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
10327 require pivot_root() anymore. This means fewer temporary
10328 directories are created below /tmp for this feature.
10330 * nspawn containers will now see and receive all submounts
10331 made on the host OS below the root file system of the
10334 * Forward Secure Sealing is now supported for Journal files,
10335 which provide cryptographical sealing of journal files so
10336 that attackers cannot alter log history anymore without this
10337 being detectable. Lennart will soon post a blog story about
10338 this explaining it in more detail.
10340 * There are two new service settings RestartPreventExitStatus=
10341 and SuccessExitStatus= which allow configuration of exit
10342 status (exit code or signal) which will be excepted from the
10343 restart logic, resp. consider successful.
10345 * journalctl gained the new --verify switch that can be used
10346 to check the integrity of the structure of journal files and
10347 (if Forward Secure Sealing is enabled) the contents of
10350 * nspawn containers will now be run with /dev/stdin, /dev/fd/
10351 and similar symlinks pre-created. This makes running shells
10352 as container init process a lot more fun.
10354 * The fstab support can now handle PARTUUID= and PARTLABEL=
10357 * A new ConditionHost= condition has been added to match
10358 against the hostname (with globs) and machine ID. This is
10359 useful for clusters where a single OS image is used to
10360 provision a large number of hosts which shall run slightly
10361 different sets of services.
10363 * Services which hit the restart limit will now be placed in a
10366 Contributions from: Bertram Poettering, Dave Reisner, Huang
10367 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
10368 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
10372 * When running in --user mode systemd will now become a
10373 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
10374 tree a lot more organized.
10376 * A new PartOf= unit dependency type has been introduced that
10377 may be used to group services in a natural way.
10379 * "systemctl enable" may now be used to enable instances of
10382 * journalctl now prints error log levels in red, and
10383 warning/notice log levels in bright white. It also supports
10384 filtering by log level now.
10386 * cgtop gained a new -n switch (similar to top), to configure
10387 the maximum number of iterations to run for. It also gained
10388 -b, to run in batch mode (accepting no input).
10390 * The suffix ".service" may now be omitted on most systemctl
10391 command lines involving service unit names.
10393 * There's a new bus call in logind to lock all sessions, as
10394 well as a loginctl verb for it "lock-sessions".
10396 * libsystemd-logind.so gained a new call sd_journal_perror()
10397 that works similar to libc perror() but logs to the journal
10398 and encodes structured information about the error number.
10400 * /etc/crypttab entries now understand the new keyfile-size=
10403 * shutdown(8) now can send a (configurable) wall message when
10404 a shutdown is cancelled.
10406 * The mount propagation mode for the root file system will now
10407 default to "shared", which is useful to make containers work
10408 nicely out-of-the-box so that they receive new mounts from
10409 the host. This can be undone locally by running "mount
10410 --make-rprivate /" if needed.
10412 * The prefdm.service file has been removed. Distributions
10413 should maintain this unit downstream if they intend to keep
10414 it around. However, we recommend writing normal unit files
10415 for display managers instead.
10417 * Since systemd is a crucial part of the OS we will now
10418 default to a number of compiler switches that improve
10419 security (hardening) such as read-only relocations, stack
10420 protection, and suchlike.
10422 * The TimeoutSec= setting for services is now split into
10423 TimeoutStartSec= and TimeoutStopSec= to allow configuration
10424 of individual time outs for the start and the stop phase of
10427 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
10428 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
10429 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
10430 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
10431 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
10432 Gundersen, Zbigniew Jędrzejewski-Szmek
10436 * The journal and id128 C APIs are now fully documented as man
10439 * Extra safety checks have been added when transitioning from
10440 the initial RAM disk to the main system to avoid accidental
10443 * /etc/crypttab entries now understand the new keyfile-offset=
10446 * systemctl -t can now be used to filter by unit load state.
10448 * The journal C API gained the new sd_journal_wait() call to
10449 make writing synchronous journal clients easier.
10451 * journalctl gained the new -D switch to show journals from a
10452 specific directory.
10454 * journalctl now displays a special marker between log
10455 messages of two different boots.
10457 * The journal is now explicitly flushed to /var via a service
10458 systemd-journal-flush.service, rather than implicitly simply
10459 by seeing /var/log/journal to be writable.
10461 * journalctl (and the journal C APIs) can now match for much
10462 more complex expressions, with alternatives and
10465 * When transitioning from the initial RAM disk to the main
10466 system we will now kill all processes in a killing spree to
10467 ensure no processes stay around by accident.
10469 * Three new specifiers may be used in unit files: %u, %h, %s
10470 resolve to the user name, user home directory resp. user
10471 shell. This is useful for running systemd user instances.
10473 * We now automatically rotate journal files if their data
10474 object hash table gets a fill level > 75%. We also size the
10475 hash table based on the configured maximum file size. This
10476 together should lower hash collisions drastically and thus
10477 speed things up a bit.
10479 * journalctl gained the new "--header" switch to introspect
10480 header data of journal files.
10482 * A new setting SystemCallFilters= has been added to services which may
10483 be used to apply deny lists or allow lists to system calls. This is
10484 based on SECCOMP Mode 2 of Linux 3.5.
10486 * nspawn gained a new --link-journal= switch (and quicker: -j)
10487 to link the container journal with the host. This makes it
10488 very easy to centralize log viewing on the host for all
10489 guests while still keeping the journal files separated.
10491 * Many bugfixes and optimizations
10493 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
10494 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
10495 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
10500 * Several tools now understand kernel command line arguments,
10501 which are only read when run in an initial RAM disk. They
10502 usually follow closely their normal counterparts, but are
10505 * There's a new tool to analyze the readahead files that are
10506 automatically generated at boot. Use:
10508 /usr/lib/systemd/systemd-readahead analyze /.readahead
10510 * We now provide an early debug shell on tty9 if this enabled. Use:
10512 systemctl enable debug-shell.service
10514 * All plymouth related units have been moved into the Plymouth
10515 package. Please make sure to upgrade your Plymouth version
10518 * systemd-tmpfiles now supports getting passed the basename of
10519 a configuration file only, in which case it will look for it
10520 in all appropriate directories automatically.
10522 * udevadm info now takes a /dev or /sys path as argument, and
10523 does the right thing. Example:
10525 udevadm info /dev/sda
10526 udevadm info /sys/class/block/sda
10528 * systemctl now prints a warning if a unit is stopped but a
10529 unit that might trigger it continues to run. Example: a
10530 service is stopped but the socket that activates it is left
10533 * "systemctl status" will now mention if the log output was
10534 shortened due to rotation since a service has been started.
10536 * The journal API now exposes functions to determine the
10537 "cutoff" times due to rotation.
10539 * journald now understands SIGUSR1 and SIGUSR2 for triggering
10540 immediately flushing of runtime logs to /var if possible,
10541 resp. for triggering immediate rotation of the journal
10544 * It is now considered an error if a service is attempted to
10545 be stopped that is not loaded.
10547 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
10549 * systemd-analyze now supports Python 3
10551 * tmpfiles now supports cleaning up directories via aging
10552 where the first level dirs are always kept around but
10553 directories beneath it automatically aged. This is enabled
10554 by prefixing the age field with '~'.
10556 * Seat objects now expose CanGraphical, CanTTY properties
10557 which is required to deal with very fast bootups where the
10558 display manager might be running before the graphics drivers
10559 completed initialization.
10561 * Seat objects now expose a State property.
10563 * We now include RPM macros for service enabling/disabling
10564 based on the preset logic. We recommend RPM based
10565 distributions to make use of these macros if possible. This
10566 makes it simpler to reuse RPM spec files across
10569 * We now make sure that the collected systemd unit name is
10570 always valid when services log to the journal via
10573 * There's a new man page kernel-command-line(7) detailing all
10574 command line options we understand.
10576 * The fstab generator may now be disabled at boot by passing
10577 fstab=0 on the kernel command line.
10579 * A new kernel command line option modules-load= is now understood
10580 to load a specific kernel module statically, early at boot.
10582 * Unit names specified on the systemctl command line are now
10583 automatically escaped as needed. Also, if file system or
10584 device paths are specified they are automatically turned
10585 into the appropriate mount or device unit names. Example:
10587 systemctl status /home
10588 systemctl status /dev/sda
10590 * The SysVConsole= configuration option has been removed from
10591 system.conf parsing.
10593 * The SysV search path is no longer exported on the D-Bus
10596 * The Names= option has been removed from unit file parsing.
10598 * There's a new man page bootup(7) detailing the boot process.
10600 * Every unit and every generator we ship with systemd now
10601 comes with full documentation. The self-explanatory boot is
10604 * A couple of services gained "systemd-" prefixes in their
10605 name if they wrap systemd code, rather than only external
10606 code. Among them fsck@.service which is now
10607 systemd-fsck@.service.
10609 * The HaveWatchdog property has been removed from the D-Bus
10612 * systemd.confirm_spawn= on the kernel command line should now
10615 * There's a new man page crypttab(5) which details all options
10616 we actually understand.
10618 * systemd-nspawn gained a new --capability= switch to pass
10619 additional capabilities to the container.
10621 * timedated will now read known NTP implementation unit names
10622 from /usr/lib/systemd/ntp-units.d/*.list,
10623 systemd-timedated-ntp.target has been removed.
10625 * journalctl gained a new switch "-b" that lists log data of
10626 the current boot only.
10628 * The notify socket is in the abstract namespace again, in
10629 order to support daemons which chroot() at start-up.
10631 * There is a new Storage= configuration option for journald
10632 which allows configuration of where log data should go. This
10633 also provides a way to disable journal logging entirely, so
10634 that data collected is only forwarded to the console, the
10635 kernel log buffer or another syslog implementation.
10637 * Many bugfixes and optimizations
10639 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
10640 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
10641 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
10642 Shawn Landden, Tom Gundersen
10646 * "systemctl help <unit>" now shows the man page if one is
10649 * Several new man pages have been added.
10651 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
10652 MaxLevelConsole= can now be specified in
10653 journald.conf. These options allow reducing the amount of
10654 data stored on disk or forwarded by the log level.
10656 * TimerSlackNSec= can now be specified in system.conf for
10657 PID1. This allows system-wide power savings.
10659 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
10660 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
10665 * logind is now capable of (optionally) handling power and
10666 sleep keys as well as the lid switch.
10668 * journalctl now understands the syntax "journalctl
10669 /usr/bin/avahi-daemon" to get all log output of a specific
10672 * CapabilityBoundingSet= in system.conf now also influences
10673 the capability bound set of usermode helpers of the kernel.
10675 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
10676 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
10677 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
10678 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
10682 * Note that we skipped 139 releases here in order to set the
10683 new version to something that is greater than both udev's
10684 and systemd's most recent version number.
10686 * udev: all udev sources are merged into the systemd source tree now.
10687 All future udev development will happen in the systemd tree. It
10688 is still fully supported to use the udev daemon and tools without
10689 systemd running, like in initramfs or other init systems. Building
10690 udev though, will require the *build* of the systemd tree, but
10691 udev can be properly *run* without systemd.
10693 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
10694 should be used to create dead device nodes as workarounds for broken
10697 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
10698 no longer supported. udev_monitor_new_from_netlink() needs to be
10699 used to subscribe to events.
10701 * udev: when udevd is started by systemd, processes which are left
10702 behind by forking them off of udev rules, are unconditionally cleaned
10703 up and killed now after the event handling has finished. Services or
10704 daemons must be started as systemd services. Services can be
10705 pulled-in by udev to get started, but they can no longer be directly
10706 forked by udev rules.
10708 * udev: the daemon binary is called systemd-udevd now and installed
10709 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
10710 to adapt to that, create symlink, or rename the binary after building
10713 * libudev no longer provides these symbols:
10714 udev_monitor_from_socket()
10715 udev_queue_get_failed_list_entry()
10716 udev_get_{dev,sys,run}_path()
10717 The versions number was bumped and symbol versioning introduced.
10719 * systemd-loginctl and systemd-journalctl have been renamed
10720 to loginctl and journalctl to match systemctl.
10722 * The config files: /etc/systemd/systemd-logind.conf and
10723 /etc/systemd/systemd-journald.conf have been renamed to
10724 logind.conf and journald.conf. Package updates should rename
10725 the files to the new names on upgrade.
10727 * For almost all files the license is now LGPL2.1+, changed
10728 from the previous GPL2.0+. Exceptions are some minor stuff
10729 of udev (which will be changed to LGPL2.1 eventually, too),
10730 and the MIT licensed sd-daemon.[ch] library that is suitable
10731 to be used as drop-in files.
10733 * systemd and logind now handle system sleep states, in
10734 particular suspending and hibernating.
10736 * logind now implements a sleep/shutdown/idle inhibiting logic
10737 suitable for a variety of uses. Soonishly Lennart will blog
10738 about this in more detail.
10740 * var-run.mount and var-lock.mount are no longer provided
10741 (which previously bind mounted these directories to their new
10742 places). Distributions which have not converted these
10743 directories to symlinks should consider stealing these files
10744 from git history and add them downstream.
10746 * We introduced the Documentation= field for units and added
10747 this to all our shipped units. This is useful to make it
10748 easier to explore the boot and the purpose of the various
10751 * All smaller setup units (such as
10752 systemd-vconsole-setup.service) now detect properly if they
10753 are run in a container and are skipped when
10754 appropriate. This guarantees an entirely noise-free boot in
10755 Linux container environments such as systemd-nspawn.
10757 * A framework for implementing offline system updates is now
10758 integrated, for details see:
10759 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
10761 * A new service type Type=idle is available now which helps us
10762 avoiding ugly interleaving of getty output and boot status
10765 * There's now a system-wide CapabilityBoundingSet= option to
10766 globally reduce the set of capabilities for the
10767 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
10768 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
10769 even CAP_NET_ADMIN system-wide for secure systems.
10771 * There are now system-wide DefaultLimitXXX= options to
10772 globally change the defaults of the various resource limits
10773 for all units started by PID 1.
10775 * Harald Hoyer's systemd test suite has been integrated into
10776 systemd which allows easy testing of systemd builds in qemu
10777 and nspawn. (This is really awesome! Ask us for details!)
10779 * The fstab parser is now implemented as generator, not inside
10782 * systemctl will now warn you if .mount units generated from
10783 /etc/fstab are out of date due to changes in fstab that
10784 have not been read by systemd yet.
10786 * systemd is now suitable for usage in initrds. Dracut has
10787 already been updated to make use of this. With this in place
10788 initrds get a slight bit faster but primarily are much
10789 easier to introspect and debug since "systemctl status" in
10790 the host system can be used to introspect initrd services,
10791 and the journal from the initrd is kept around too.
10793 * systemd-delta has been added, a tool to explore differences
10794 between user/admin configuration and vendor defaults.
10796 * PrivateTmp= now affects both /tmp and /var/tmp.
10798 * Boot time status messages are now much prettier and feature
10799 proper english language. Booting up systemd has never been
10802 * Read-ahead pack files now include the inode number of all
10803 files to pre-cache. When the inode changes the pre-caching
10804 is not attempted. This should be nicer to deal with updated
10805 packages which might result in changes of read-ahead
10808 * We now temporaritly lower the kernel's read_ahead_kb variable
10809 when collecting read-ahead data to ensure the kernel's
10810 built-in read-ahead does not add noise to our measurements
10811 of necessary blocks to pre-cache.
10813 * There's now RequiresMountsFor= to add automatic dependencies
10814 for all mounts necessary for a specific file system path.
10816 * MountAuto= and SwapAuto= have been removed from
10817 system.conf. Mounting file systems at boot has to take place
10820 * nspawn now learned a new switch --uuid= to set the machine
10821 ID on the command line.
10823 * nspawn now learned the -b switch to automatically search
10824 for an init system.
10826 * vt102 is now the default TERM for serial TTYs, upgraded from
10829 * systemd-logind now works on VT-less systems.
10831 * The build tree has been reorganized. The individual
10832 components now have directories of their own.
10834 * A new condition type ConditionPathIsReadWrite= is now available.
10836 * nspawn learned the new -C switch to create cgroups for the
10837 container in other hierarchies.
10839 * We now have support for hardware watchdogs, configurable in
10842 * The scheduled shutdown logic now has a public API.
10844 * We now mount /tmp as tmpfs by default, but this can be
10845 masked and /etc/fstab can override it.
10847 * Since udisks does not make use of /media anymore we are not
10848 mounting a tmpfs on it anymore.
10850 * journalctl gained a new --local switch to only interleave
10851 locally generated journal files.
10853 * We can now load the IMA policy at boot automatically.
10855 * The GTK tools have been split off into a systemd-ui.
10857 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
10858 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
10859 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
10860 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
10861 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
10862 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
10863 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
10864 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
10865 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
10870 * This is mostly a bugfix release
10872 * Support optional initialization of the machine ID from the
10873 KVM or container configured UUID.
10875 * Support immediate reboots with "systemctl reboot -ff"
10877 * Show /etc/os-release data in systemd-analyze output
10879 * Many bugfixes for the journal, including endianness fixes and
10880 ensuring that disk space enforcement works
10882 * sd-login.h is C++ compatible again
10884 * Extend the /etc/os-release format on request of the Debian
10887 * We now refuse non-UTF8 strings used in various configuration
10888 and unit files. This is done to ensure we do not pass invalid
10889 data over D-Bus or expose it elsewhere.
10891 * Register Mimo USB Screens as suitable for automatic seat
10894 * Read SELinux client context from journal clients in a race
10897 * Reorder configuration file lookup order. /etc now always
10898 overrides /run in order to allow the administrator to always
10899 and unconditionally override vendor-supplied or
10900 automatically generated data.
10902 * The various user visible bits of the journal now have man
10903 pages. We still lack man pages for the journal API calls
10906 * We now ship all man pages in HTML format again in the
10909 Contributions from: Dave Reisner, Dirk Eibach, Frederic
10910 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
10911 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
10916 * This is mostly a bugfix release
10918 * systems lacking /etc/os-release are no longer supported.
10920 * Various functionality updates to libsystemd-login.so
10922 * Track class of PAM logins to distinguish greeters from
10923 normal user logins.
10925 Contributions from: Kay Sievers, Lennart Poettering, Michael
10930 * This is an important bugfix release for v41.
10932 * Building man pages is now optional which should be useful
10933 for those building systemd from git but unwilling to install
10936 * Watchdog support for supervising services is now usable. In
10937 a future release support for hardware watchdogs
10938 (i.e. /dev/watchdog) will be added building on this.
10940 * Service start rate limiting is now configurable and can be
10941 turned off per service. When a start rate limit is hit a
10942 reboot can automatically be triggered.
10944 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
10946 Contributions from: Benjamin Franzke, Bill Nottingham,
10947 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
10948 Schmidt, Michał Górny, Piotr Drąg
10952 * The systemd binary is installed /usr/lib/systemd/systemd now;
10953 An existing /sbin/init symlink needs to be adapted with the
10956 * The code that loads kernel modules has been ported to invoke
10957 libkmod directly, instead of modprobe. This means we do not
10958 support systems with module-init-tools anymore.
10960 * Watchdog support is now already useful, but still not
10963 * A new kernel command line option systemd.setenv= is
10964 understood to set system wide environment variables
10965 dynamically at boot.
10967 * We now limit the set of capabilities of systemd-journald.
10969 * We now set SIGPIPE to ignore by default, since it only is
10970 useful in shell pipelines, and has little use in general
10971 code. This can be disabled with IgnoreSIPIPE=no in unit
10974 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
10975 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
10980 * This is mostly a bugfix release
10982 * We now expose the reason why a service failed in the
10983 "Result" D-Bus property.
10985 * Rudimentary service watchdog support (will be completed over
10986 the next few releases.)
10988 * When systemd forks off in order execute some service we will
10989 now immediately changes its argv[0] to reflect which process
10990 it will execute. This is useful to minimize the time window
10991 with a generic argv[0], which makes bootcharts more useful
10993 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
10994 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
10995 Mike Kazantsev, Ray Strode
10999 * This is mostly a test release, but incorporates many
11002 * New systemd-cgtop tool to show control groups by their
11005 * Linking against libacl for ACLs is optional again. If
11006 disabled, support tracking device access for active logins
11007 goes becomes unavailable, and so does access to the user
11008 journals by the respective users.
11010 * If a group "adm" exists, journal files are automatically
11011 owned by them, thus allow members of this group full access
11012 to the system journal as well as all user journals.
11014 * The journal now stores the SELinux context of the logging
11015 client for all entries.
11017 * Add C++ inclusion guards to all public headers
11019 * New output mode "cat" in the journal to print only text
11020 messages, without any meta data like date or time.
11022 * Include tiny X server wrapper as a temporary stop-gap to
11023 teach XOrg udev display enumeration. This is used by display
11024 managers such as gdm, and will go away as soon as XOrg
11025 learned native udev hotplugging for display devices.
11027 * Add new systemd-cat tool for executing arbitrary programs
11028 with STDERR/STDOUT connected to the journal. Can also act as
11029 BSD logger replacement, and does so by default.
11031 * Optionally store all locally generated coredumps in the
11032 journal along with meta data.
11034 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
11035 writing short strings to files (for usage for /sys), and for
11036 creating symlinks, character and block device nodes.
11038 * New unit file option ControlGroupPersistent= to make cgroups
11039 persistent, following the mechanisms outlined in
11040 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
11042 * Support multiple local RTCs in a sane way
11044 * No longer monopolize IO when replaying readahead data on
11045 rotating disks, since we might starve non-file-system IO to
11046 death, since fanotify() will not see accesses done by blkid,
11049 * Do not show kernel threads in systemd-cgls anymore, unless
11050 requested with new -k switch.
11052 Contributions from: Dan Horák, Kay Sievers, Lennart
11053 Poettering, Michal Schmidt
11057 * This is mostly a test release, but incorporates many
11060 * The git repository moved to:
11061 git://anongit.freedesktop.org/systemd/systemd
11062 ssh://git.freedesktop.org/git/systemd/systemd
11064 * First release with the journal
11065 http://0pointer.de/blog/projects/the-journal.html
11067 * The journal replaces both systemd-kmsg-syslogd and
11068 systemd-stdout-bridge.
11070 * New sd_pid_get_unit() API call in libsystemd-logind
11072 * Many systemadm clean-ups
11074 * Introduce remote-fs-pre.target which is ordered before all
11075 remote mounts and may be used to start services before all
11078 * Added Mageia support
11080 * Add bash completion for systemd-loginctl
11082 * Actively monitor PID file creation for daemons which exit in
11083 the parent process before having finished writing the PID
11084 file in the daemon process. Daemons which do this need to be
11085 fixed (i.e. PID file creation must have finished before the
11086 parent exits), but we now react a bit more gracefully to them.
11088 * Add colourful boot output, mimicking the well-known output
11089 of existing distributions.
11091 * New option PassCredentials= for socket units, for
11092 compatibility with a recent kernel ABI breakage.
11094 * /etc/rc.local is now hooked in via a generator binary, and
11095 thus will no longer act as synchronization point during
11098 * systemctl list-unit-files now supports --root=.
11100 * systemd-tmpfiles now understands two new commands: z, Z for
11101 relabelling files according to the SELinux database. This is
11102 useful to apply SELinux labels to specific files in /sys,
11103 among other things.
11105 * Output of SysV services is now forwarded to both the console
11106 and the journal by default, not only just the console.
11108 * New man pages for all APIs from libsystemd-login.
11110 * The build tree got reorganized and the build system is a
11111 lot more modular allowing embedded setups to specifically
11112 select the components of systemd they are interested in.
11114 * Support for Linux systems lacking the kernel VT subsystem is
11117 * configure's --with-rootdir= got renamed to
11118 --with-rootprefix= to follow the naming used by udev and
11121 * Unless specified otherwise we will now install to /usr instead
11122 of /usr/local by default.
11124 * Processes with '@' in argv[0][0] are now excluded from the
11125 final shut-down killing spree, following the logic explained
11127 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
11129 * All processes remaining in a service cgroup when we enter
11130 the START or START_PRE states are now killed with
11131 SIGKILL. That means it is no longer possible to spawn
11132 background processes from ExecStart= lines (which was never
11133 supported anyway, and bad style).
11135 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
11136 reloading of units together.
11138 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
11139 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
11140 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
11141 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
11142 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek