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, 10% of RAM for /dev/shm, 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's [Network] section gained a new setting
189 IPv6PDSubnetId= that allows explicit configuration of the preferred
190 subnet that networkd's Prefix Delegation logic assigns to interfaces.
192 * systemd-networkd's [Network] section gained a new setting
193 IPv4AcceptLocal=. If enabled the interface accepts packets with local
196 * systemd-networkd gained support for configuring the HTB queuing
197 discipline in the [HierarchyTokenBucket] and
198 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
199 be configured in the [PFIFO] section, "GRED" in
200 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
201 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
202 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
203 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast], "HHF"
204 in [HeavyHitterFilter], "ETS" in [EnhancedTransmissionSelection] and
205 "QFQ" in [QuickFairQueueing] and [QuickFairQueueingClass].
207 * systemd-networkd gained support for a new Termination= setting in the
208 [CAN] section for configuring the termination resistor. It also
209 gained a new ListenOnly= setting for controlling whether to only
210 listen on CAN interfaces, without interfering with traffic otherwise
211 (which is useful for debugging/monitoring CAN network
212 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
213 been added to configure various CAN-FD aspects.
215 * systemd-networkd's [DHCPv6] section gained a new option WithoutRA=.
216 When enabled, DHCPv6 will be attempted right-away without requiring an
217 Router Advertisement packet suggesting it first (i.e. without the 'M'
218 or 'O' flags set). The [IPv6AcceptRA] section gained a boolean option
219 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
220 the RA packets suggest it.
222 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
223 which may be used to turn off use of the gateway information provided
224 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
225 used to configure how to process leases that lack a lifetime option.
227 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
228 setting SendVendorOption= allowing configuration of additional vendor
229 options to send in the DHCP requests/responses. The [DHCPv6] section
230 gained a new SendOption= setting for sending arbitrary DHCP
231 options. RequestOptions= has been added to request arbitrary options
232 from the server. UserClass= has been added to set the DHCP user class
235 * systemd-networkd's [DHCPServer] section gained a new set of options
236 EmitPOP3=/POP3=, EmitSMTP=/SMTP=, EmitLPR=/LPR= for including server
237 information about these three protocols in the DHCP lease. It also
238 gained support for including "MUD" URLs ("Manufacturer Usage
239 Description"). Support for "MUD" URLs was also added to the LLDP
240 stack, configurable in the [LLDP] section in .network files.
242 * The Mode= settings in [MACVLAN] and [MACVTAP] now support 'source'
243 mode. Also, the sections now support a new setting SourceMACAddress=.
245 * systemd-networkd's .netdev files now support a new setting
246 VLANProtocol= in the [Bridge] section that allows configuration of
247 the VLAN protocol to use.
249 * systemd-networkd supports a new Group= setting in the [Link] section
250 of the .network files, to control the link group.
252 * systemd-networkd's [Network] section gained a new
253 IPv6LinkLocalAddressGenerationMode= setting, which specifies how IPv6
254 link local address is generated.
256 * A new default .network file is now shipped that matches TUN/TAP
257 devices that begin with "vt-" in their name. Such interfaces will
258 have IP routing onto the host links set up automatically. This is
259 supposed to be used by VM managers to trivially acquire a network
260 interface which is fully set up for host communication, simply by
261 carefully picking an interface name to use.
263 * A new boolean option AssignAcquiredDelegatedPrefixAddress= has been
264 added to the [DHCPv6] section of .network files. If enabled (which is
265 the default) an address from any acquired delegated prefix is
266 automatically chosen and assigned to the interface.
268 * systemd-networkd's [DHCPv6] section gained a new setting RouteMetric=
269 which sets the route priority for routes specified by the DHCP server.
271 * systemd-networkd's [DHCPv6] section gained a new setting VendorClass=
272 which configures the vendor class information sent to DHCP server.
274 * The BlackList= settings in .network files' [DHCPv4] and
275 [IPv6AcceptRA] sections have been renamed DenyList=. The old names
276 are still understood to provide compatibility.
278 * networkctl gained the new "forcerenew" command for forcing all DHCP
279 server clients to renew their lease. The interface "status" output
280 will now show numerous additional fields of information about an
281 interface. There are new "up" and "down" commands to bring specific
282 interfaces up or down.
284 * systemd-resolved's DNS= configuration option now optionally accepts a
285 port number (after ":") and a host name (after "#"). When the host
286 name is specified, the DNS-over-TLS certificate is validated to match
287 the specified hostname. Additionally, in case of IPv6 addresses, an
288 interface may be specified (after "%").
290 * systemd-resolved may be configured to forward single-label DNS names.
291 This is not standard-conformant, but may make sense in setups where
292 public DNS servers are not used.
294 * systemd-resolved's DNS-over-TLS support gained SNI validation.
296 * systemd-nspawn's --resolv-conf= switch gained a number of new
297 supported values. Specifically, options starting with "replace-" are
298 like those prefixed "copy-" but replace any existing resolv.conf
299 file. And options ending in "-uplink" and "-stub" can now be used to
300 propagate other flavours of resolv.conf into the container (as
301 defined by systemd-resolved).
303 * The various programs included in systemd can now optionally output
304 their log messages on stderr prefixed with a timestamp, controlled by
305 the $SYSTEMD_LOG_TIME environment variable.
307 * systemctl gained a new "-P" switch that is a shortcut for "--value
310 * "systemctl list-units" and "systemctl list-machines" no longer hide
311 their first output column with --no-legend. To hide the first column,
314 * "systemctl reboot" takes the option "--reboot-argument=".
315 The optional positional argument to "systemctl reboot" is now
316 being deprecated in favor of this option.
318 * systemd-run gained a new switch --slice-inherit. If specified the
319 unit it generates is placed in the same slice as the systemd-run
322 * systemd-journald gained support for zstd compression of large fields
323 in journal files. The hash tables in journal files have been hardened
324 against hash collisions. This is an incompatible change and means
325 that journal files created with new systemd versions are not readable
326 with old versions. If the $SYSTEMD_JOURNAL_KEYED_HASH boolean
327 environment variable for systemd-journald.service is set to 0 this
328 new hardening functionality may be turned off, so that generated
329 journal files remain compatible with older journalctl
332 * journalctl will now include a clickable link in the default output for
333 each log message for which an URL with further documentation is
334 known. This is only supported on terminal emulators that support
335 clickable hyperlinks, and is turned off if a pager is used (since
336 "less" still doesn't support hyperlinks,
337 unfortunately). Documentation URLs may be included in log messages
338 either by including a DOCUMENTATION= journal field in it, or by
339 associating a journal message catalog entry with the log message's
340 MESSAGE_ID, which then carries a "Documentation:" tag.
342 * journald.conf gained a new boolean setting Audit= that may be used to
343 control whether systemd-journald will enable audit during
346 * when systemd-journald's log stream is broken up into multiple lines
347 because the PID of the sender changed this is indicated in the
348 generated log records via the _LINE_BREAK=pid-change field.
350 * journalctl's "-o cat" output mode will now show one or more journal
351 fields specified with --output-fields= instead of unconditionally
352 MESSAGE=. This is useful to retrieve a very specific set of fields
353 without any decoration.
355 * The sd-journal.h API gained two new functions:
356 sd_journal_enumerate_available_unique() and
357 sd_journal_enumerate_available_data() that operate like their
358 counterparts that lack the _available_ in the name, but skip items
359 that cannot be read and processed by the local implementation
360 (i.e. are compressed in an unsupported format or such),
362 * coredumpctl gained a new --file= switch, matching the same one in
363 journalctl: a specific journal file may be specified to read the
366 * coredumps collected by systemd-coredump may now be compressed using
369 * systemd-binfmt gained a new switch --unregister for unregistering all
370 registered entries at once. This is now invoked automatically at
371 shutdown, so that binary formats registered with the "F" flag will
372 not block clean file system unmounting.
374 * systemd-notify's --pid= switch gained new values: "parent", "self",
375 "auto" for controlling which PID to send to the service manager: the
376 systemd-notify process' PID, or the one of the process invoking it.
378 * systemd-logind's Session bus object learnt a new method call
379 SetType() for temporarily updating the session type of an already
380 allocated session. This is useful for upgrading tty sessions to
381 graphical ones once a compositor is invoked.
383 * systemd-socket-proxy gained a new switch --exit-idle-time= for
384 configuring an exit-on-idle time.
386 * systemd-repart's --empty= setting gained a new value "create". If
387 specified a new empty regular disk image file is created under the
388 specified name. Its size may be specified with the new --size=
389 option. The latter is also supported without the "create" mode, in
390 order to grow existing disk image files to the specified size. These
391 two new options are useful when creating or manipulating disk images
392 instead of operating on actual block devices.
394 * systemd-repart drop-ins now support a new UUID= setting to control
395 the UUID to assign to a newly created partition.
397 * systemd-repart's SizeMin= per-partition parameter now defaults to 10M
400 * systemd-repart's Label= setting now support the usual, simple
403 * systemd-homed's LUKS backend gained the ability to discard empty file
404 system blocks automatically when the user logs out. This is enabled
405 by default to ensure that home directories take minimal space when
406 logged out but get full size guarantees when logged in. This may be
407 controlled with the new --luks-offline-discard= switch to homectl.
409 * If systemd-homed detects that /home/ is encrypted as a whole it will
410 now default to the directory or subvolume backends instead of the
411 LUKS backend, in order to avoid double encryption. The default
412 storage and file system may now be configured explicitly, too, via
413 the new /etc/systemd/homed.conf configuration file.
415 * systemd-homed now supports unlocking home directories with FIDO2
416 security tokens that support the 'hmac-secret' extension, in addition
417 to the existing support for PKCS#11 security token unlocking
418 support. Note that many recent hardware security tokens support both
419 interfaces. The FIDO2 support is accessible via homectl's
420 --fido2-device= option.
422 * homectl's --pkcs11-uri= setting now accepts two special parameters:
423 if "auto" is specified and only one suitable PKCS#11 security token
424 is plugged in, its URL is automatically determined and enrolled for
425 unlocking the home directory. If "list" is specified a brief table of
426 suitable PKCS#11 security tokens is shown. Similar, the new
427 --fido2-device= option also supports these two special values, for
428 automatically selecting and listing suitable FIDO2 devices.
430 * The /etc/crypttab tmp option now optionally takes an argument
431 selecting the file system to use. Moreover, the default is now
432 changed from ext2 to ext4.
434 * There's a new /etc/crypttab option "keyfile-erase". If specified the
435 key file listed in the same line is removed after use, regardless if
436 volume activation was successful or not. This is useful if the key
437 file is only acquired transiently at runtime and shall be erased
438 before the system continues to boot.
440 * There's also a new /etc/crypttab option "try-empty-password". If
441 specified, before asking the user for a password it is attempted to
442 unlock the volume with an empty password. This is useful for
443 installing encrypted images whose password shall be set on first boot
444 instead of at installation time.
446 * systemd-cryptsetup will now attempt to load the keys to unlock
447 volumes with automatically from files in
448 /etc/cryptsetup-keys.d/<volume>.key and
449 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
451 * systemd-cryptsetup may now activate Microsoft BitLocker volumes via
452 /etc/crypttab, during boot.
454 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
455 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
458 * A new generator systemd-xdg-autostart-generator has been added. It
459 generates systemd unit files from XDG autostart .desktop files, and
460 may be used to let the systemd user instance manage services that are
461 started automatically as part of the desktop session.
463 * "bootctl" gained a new verb "reboot-to-firmware" that may be used
464 to query and change the firmware's 'reboot into firmware' setup flag.
466 * systemd-firstboot gained a new switch --kernel-command-line= that may
467 be used to initialize the /etc/kernel/cmdline file of the image. It
468 also gained a new switch --root-password-hashed= which is like
469 --root-password= but accepts a pre-hashed UNIX password as
470 argument. The new option --delete-root-password may be used to unset
471 any password for the root user (dangerous!). The --root-shell= switch
472 may be used to control the shell to use for the root account. A new
473 --force option may be used to override any already set settings with
474 the parameters specified on the command line (by default, the tool
475 will not override what has already been set before, i.e. is purely
478 * systemd-firstboot gained support for a new --image= switch, which is
479 similar to --root= but accepts the path to a disk image file, on
480 which it then operates.
482 * A new sd-path.h API has been added to libsystemd. It provides a
483 simple API for retrieving various search paths and primary
484 directories for various resources.
486 * A new call sd_notify_barrier() has been added to the sd-daemon.h
487 API. The call will block until all previously sent sd_notify()
488 messages have been processed by the service manager. This is useful
489 to remove races caused by a process already having disappeared at the
490 time a notification message is processed by the service manager,
491 making correct attribution impossible. The systemd-notify tool will
492 now make use of this call implicitly, but this can be turned off again
493 via the new --no-block switch.
495 * When sending a file descriptor (fd) to the service manager to keep
496 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
497 may be specified. If passed the service manager will refrain from
498 poll()ing on the file descriptor. Traditionally (and when the
499 parameter is not specified), the service manager will poll it for
500 POLLHUP or POLLERR events, and immediately close the fds in that
503 * The service manager (PID1) gained a new D-Bus method call
504 SetShowStatus() which may be used to control whether it shall show
505 boot-time status output on the console. This method has a similar
506 effect to sending SIGRTMIN+20/SIGRTMIN+21 to PID 1.
508 * The sd-bus API gained a number of convenience functions that take
509 va_list arguments rather than "...". For example, there's now
510 sd_bus_call_methodv() to match sd_bus_call_method(). Those calls make
511 it easier to build wrappers that accept variadic arguments and want
512 to pass a ready va_list structure to sd-bus.
514 * sd-bus vtable entries can have a new SD_BUS_VTABLE_ABSOLUTE_OFFSET
515 flag which alters how the userdata pointer to pass to the callbacks
516 is determined. When the flag is set, the offset field is converted
517 as-is into a pointer, without adding it to the object pointer the
518 vtable is associated with.
520 * sd-bus now exposes four new functions:
521 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
522 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
523 validate strings to check if they qualify as various D-Bus concepts.
525 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
526 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
527 that simplify adding argument names to D-Bus methods and signals.
529 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
531 * Various D-Bus APIs of systemd daemons now have man pages that
532 document the methods, signals and properties.
534 * The expectations on user/group name syntax are now documented in
535 detail; documentation on how classic home directories may be
536 converted into home directories managed by homed has been added;
537 documentation regarding integration of homed/userdb functionality in
538 desktops has been added:
540 https://systemd.io/USER_NAMES
541 https://systemd.io/CONVERTING_TO_HOMED
542 https://systemd.io/USERDB_AND_DESKTOPS
544 * Documentation for the on-disk Journal file format has been updated
545 and has now moved to:
547 https://systemd.io/JOURNAL_FILE_FORMAT
549 * The interface for containers (https://systemd.io/CONTAINER_INTERFACE)
550 has been extended by a set of environment variables that expose
551 select fields from the host's os-release file to the container
552 payload. Similarly, host's os-release files can be mounted into the
553 container underneath /run/host. Together, those mechanisms provide a
554 standardized way to expose information about the host to the
555 container payload. Both interfaces are implemented in systemd-nspawn.
557 * All D-Bus services shipped in systemd now implement the generic
558 LogControl1 D-Bus API which allows clients to change log level +
559 target of the service during runtime.
561 * Only relevant for developers: the mkosi.default symlink has been
562 dropped from version control. Please create a symlink to one of the
563 distribution-specific defaults in .mkosi/ based on your preference.
565 Contributions from: 24bisquitz, Adam Nielsen, Alan Perry, Alexander
566 Malafeev, Alin Popa, Alvin Šipraga, Amos Bird, Andreas Rammhold,
567 AndreRH, Andrew Doran, Anita Zhang, Ankit Jain, antznin, Arnaud
568 Ferraris, Arthur Moraes do Lago, Arusekk, Balaji Punnuru, Balint
569 Reczey, Bastien Nocera, bemarek, Benjamin Berg, Benjamin Dahlhoff,
570 Benjamin Robin, Chris Down, Chris Kerr, Christian Göttsche, Christian
571 Hesse, Christian Oder, Ciprian Hacman, Clinton Roy, codicodi, Corey
572 Hinshaw, Daan De Meyer, Dana Olson, Dan Callaghan, Daniel Fullmer,
573 Daniel Rusek, Dan Streetman, Dave Reisner, David Edmundson, David Wood,
574 Denis Pronin, Diego Escalante Urrelo, Dimitri John Ledkov,
575 dolphrundgren, duguxy, Einsler Lee, Elisei Roca, Emmanuel Garette, Eric
576 Anderson, Eric DeVolder, Evgeny Vereshchagin, ExtinctFire, fangxiuning,
577 Ferran Pallarès Roca, Filipe Brandenburger, Filippo Falezza, Finn,
578 Florian Klink, Florian Mayer, Franck Bui, Frantisek Sumsal, gaurav,
579 Georg Müller, Gergely Polonkai, Giedrius Statkevičius, Gigadoc2,
580 gogogogi, gzjsgdsb, Hans de Goede, Haochen Tong, ianhi, ignapk, Jakov
581 Smolic, James T. Lee, Jan Janssen, Jan Klötzke, Jan Palus, Jay Burger,
582 Jeremy Cline, Jérémy Rosen, Jian-Hong Pan, Jiri Slaby, Joel Shapiro,
583 Joerg Behrmann, Jörg Thalheim, Jouke Witteveen, Kai-Heng Feng, Kenny
584 Levinsen, Kevin Kuehler, Kumar Kartikeya Dwivedi, layderv, laydervus,
585 Lénaïc Huard, Lennart Poettering, Lidong Zhong, Luca Boccassi, Luca
586 BRUNO, Lucas Werkmeister, Lukas Klingsbo, Lukáš Nykrýn, Łukasz
587 Stelmach, Maciej S. Szmigiero, MadMcCrow, Marc-André Lureau, Marcel
588 Holtmann, Marc Kleine-Budde, Martin Hundebøll, Matthew Leeds, Matt
589 Ranostay, Maxim Fomin, MaxVerevkin, Michael Biebl, Michael Chapman,
590 Michael Gubbels, Michael Marley, Michał Bartoszkiewicz, Michal Koutný,
591 Michal Sekletár, Mike Gilbert, Mike Kazantsev, Mikhail Novosyolov, ml,
592 Motiejus Jakštys, nabijaczleweli, nerdopolis, Niccolò Maggioni, Niklas
593 Hambüchen, Norbert Lange, Paul Cercueil, pelzvieh, Peter Hutterer,
594 Piero La Terza, Pieter Lexis, Piotr Drąg, Rafael Fontenelle, Richard
595 Petri, Ronan Pigott, Ross Lagerwall, Rubens Figueiredo, satmandu,
596 Sean-StarLabs, Sebastian Jennen, sterlinghughes, Surhud More, Susant
597 Sahani, szb512, Thomas Haller, Tobias Hunger, Tom, Tomáš Pospíšek,
598 Tomer Shechner, Tom Hughes, Topi Miettinen, Tudor Roman, Uwe
599 Kleine-König, Valery0xff, Vito Caputo, Vladimir Panteleev, Vladyslav
600 Tronko, Wen Yang, Yegor Vialov, Yigal Korman, Yi Gao, YmrDtnJu, Yuri
601 Chornoivan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Zhu Li, Дамјан
608 * A new tool "systemd-repart" has been added, that operates as an
609 idempotent declarative repartitioner for GPT partition tables.
610 Specifically, a set of partitions that must or may exist can be
611 configured via drop-in files, and during every boot the partition
612 table on disk is compared with these files, creating missing
613 partitions or growing existing ones based on configurable relative
614 and absolute size constraints. The tool is strictly incremental,
615 i.e. does not delete, shrink or move partitions, but only adds and
616 grows them. The primary use-case is OS images that ship in minimized
617 form, that on first boot are grown to the size of the underlying
618 block device or augmented with additional partitions. For example,
619 the root partition could be extended to cover the whole disk, or a
620 swap or /home partitions could be added on first boot. It can also be
621 used for systems that use an A/B update scheme but ship images with
622 just the A partition, with B added on first boot. The tool is
623 primarily intended to be run in the initrd, shortly before
624 transitioning into the host OS, but can also be run after the
625 transition took place. It automatically discovers the disk backing
626 the root file system, and should hence not require any additional
627 configuration besides the partition definition drop-ins. If no
628 configuration drop-ins are present, no action is taken.
630 * A new component "userdb" has been added, along with a small daemon
631 "systemd-userdb.service" and a client tool "userdbctl". The framework
632 allows defining rich user and group records in a JSON format,
633 extending on the classic "struct passwd" and "struct group"
634 structures. Various components in systemd have been updated to
635 process records in this format, including systemd-logind and
636 pam-systemd. The user records are intended to be extensible, and
637 allow setting various resource management, security and runtime
638 parameters that shall be applied to processes and sessions of the
639 user as they log in. This facility is intended to allow associating
640 such metadata directly with user/group records so that they can be
641 produced, extended and consumed in unified form. We hope that
642 eventually frameworks such as sssd will generate records this way, so
643 that for the first time resource management and various other
644 per-user settings can be configured in LDAP directories and then
645 provided to systemd (specifically to systemd-logind and pam-system)
646 to apply on login. For further details see:
648 https://systemd.io/USER_RECORD
649 https://systemd.io/GROUP_RECORD
650 https://systemd.io/USER_GROUP_API
652 * A small new service systemd-homed.service has been added, that may be
653 used to securely manage home directories with built-in encryption.
654 The complete user record data is unified with the home directory,
655 thus making home directories naturally migratable. Its primary
656 back-end is based on LUKS volumes, but fscrypt, plain directories,
657 and other storage schemes are also supported. This solves a couple of
658 problems we saw with traditional ways to manage home directories, in
659 particular when it comes to encryption. For further discussion of
660 this, see the video of Lennart's talk at AllSystemsGo! 2019:
662 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
664 For further details about the format and expectations on home
665 directories this new daemon makes, see:
667 https://systemd.io/HOME_DIRECTORY
669 * systemd-journald is now multi-instantiable. In addition to the main
670 instance systemd-journald.service there's now a template unit
671 systemd-journald@.service, with each instance defining a new named
672 log 'namespace' (whose name is specified via the instance part of the
673 unit name). A new unit file setting LogNamespace= has been added,
674 taking such a namespace name, that assigns services to the specified
675 log namespaces. As each log namespace is serviced by its own
676 independent journal daemon, this functionality may be used to improve
677 performance and increase isolation of applications, at the price of
678 losing global message ordering. Each instance of journald has a
679 separate set of configuration files, with possibly different disk
680 usage limitations and other settings.
682 journalctl now takes a new option --namespace= to show logs from a
683 specific log namespace. The sd-journal.h API gained
684 sd_journal_open_namespace() for opening the log stream of a specific
685 log namespace. systemd-journald also gained the ability to exit on
686 idle, which is useful in the context of log namespaces, as this means
687 log daemons for log namespaces can be activated automatically on
688 demand and will stop automatically when no longer used, minimizing
691 * When systemd-tmpfiles copies a file tree using the 'C' line type it
692 will now label every copied file according to the SELinux database.
694 * When systemd/PID 1 detects it is used in the initrd it will now boot
695 into initrd.target rather than default.target by default. This should
696 make it simpler to build initrds with systemd as for many cases the
697 only difference between a host OS image and an initrd image now is
698 the presence of the /etc/initrd-release file.
700 * A new kernel command line option systemd.cpu_affinity= is now
701 understood. It's equivalent to the CPUAffinity= option in
702 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
703 itself and the default for all other processes.
705 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
706 equivalent), the SELinux database is now reloaded, ensuring that
707 sockets and other file system objects are generated taking the new
708 database into account.
710 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
711 "quiet" has been changed to imply that instead of
712 "systemd.show-status=auto". In this mode, only messages about errors
713 and significant delays in boot are shown on the console.
715 * The sd-event.h API gained native support for the new Linux "pidfd"
716 concept. This permits watching processes using file descriptors
717 instead of PID numbers, which fixes a number of races and makes
718 process supervision more robust and efficient. All of systemd's
719 components will now use pidfds if the kernel supports it for process
720 watching, with the exception of PID 1 itself, unfortunately. We hope
721 to move PID 1 to exclusively using pidfds too eventually, but this
722 requires some more kernel work first. (Background: PID 1 watches
723 processes using waitid() with the P_ALL flag, and that does not play
724 together nicely with pidfds yet.)
726 * Closely related to this, the sd-event.h API gained two new calls
727 sd_event_source_send_child_signal() (for sending a signal to a
728 watched process) and sd_event_source_get_child_process_own() (for
729 marking a process so that it is killed automatically whenever the
730 event source watching it is freed).
732 * systemd-networkd gained support for configuring Token Bucket Filter
733 (TBF) parameters in its qdisc configuration support. Similarly,
734 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
735 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
737 * systemd-networkd gained support for Intermediate Functional Block
738 (IFB) network devices.
740 * systemd-networkd gained support for configuring multi-path IP routes,
741 using the new MultiPathRoute= setting in the [Route] section.
743 * systemd-networkd's DHCPv4 client has been updated to support a new
744 SendDecline= option. If enabled, duplicate address detection is done
745 after a DHCP offer is received from the server. If a conflict is
746 detected, the address is declined. The DHCPv4 client also gained
747 support for a new RouteMTUBytes= setting that allows to configure the
748 MTU size to be used for routes generated from DHCPv4 leases.
750 * The PrefixRoute= setting in systemd-networkd's [Address] section of
751 .network files has been deprecated, and replaced by AddPrefixRoute=,
752 with its sense inverted.
754 * The Gateway= setting of [Route] sections of .network files gained
755 support for a special new value "_dhcp". If set, the configured
756 static route uses the gateway host configured via DHCP.
758 * New User= and SuppressPrefixLength= settings have been implemented
759 for the [RoutingPolicyRule] section of .network files to configure
760 source routing based on UID ranges and prefix length, respectively.
762 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
763 D-Bus message object as "sensitive". Those objects are erased from
764 memory when they are freed. This concept is intended to be used for
765 messages that contain security sensitive data. A new flag
766 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
767 in sd-bus vtables, causing any incoming and outgoing messages of
768 those methods to be implicitly marked as "sensitive".
770 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
771 contents of a message (or parts thereof) to standard output for
774 * systemd-sysusers gained support for creating users with the primary
775 group named differently than the user.
777 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
778 gained support for growing XFS partitions. Previously it supported
779 only ext4 and btrfs partitions.
781 * The support for /etc/crypttab gained a new x-initrd.attach option. If
782 set, the specified encrypted volume is unlocked already in the
783 initrd. This concept corresponds to the x-initrd.mount option in
786 * systemd-cryptsetup gained native support for unlocking encrypted
787 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
788 encryption of volumes to YubiKeys. This is exposed in the new
789 pkcs11-uri= option in /etc/crypttab.
791 * The /etc/fstab support in systemd now supports two new mount options
792 x-systemd.{required,wanted}-by=, for explicitly configuring the units
793 that the specified mount shall be pulled in by, in place of
794 the usual local-fs.target/remote-fs.target.
796 * The https://systemd.io/ web site has been relaunched, directly
797 populated with most of the documentation included in the systemd
798 repository. systemd also acquired a new logo, thanks to Tobias
801 * systemd-udevd gained support for managing "alternative" network
802 interface names, as supported by new Linux kernels. For the first
803 time this permits assigning multiple (and longer!) names to a network
804 interface. systemd-udevd will now by default assign the names
805 generated via all supported naming schemes to each interface. This
806 may be further tweaked with .link files and the AlternativeName= and
807 AlternativeNamesPolicy= settings. Other components of systemd have
808 been updated to support the new alternative names wherever
809 appropriate. For example, systemd-nspawn will now generate
810 alternative interface names for the host-facing side of container
811 veth links based on the full container name without truncation.
813 * systemd-nspawn interface naming logic has been updated in another way
814 too: if the main interface name (i.e. as opposed to new-style
815 "alternative" names) based on the container name is truncated, a
816 simple hashing scheme is used to give different interface names to
817 multiple containers whose names all begin with the same prefix. Since
818 this changes the primary interface names pointing to containers if
819 truncation happens, the old scheme may still be requested by
820 selecting an older naming scheme, via the net.naming-scheme= kernel
823 * PrivateUsers= in service files now works in services run by the
824 systemd --user per-user instance of the service manager.
826 * A new per-service sandboxing option ProtectClock= has been added that
827 locks down write access to the system clock. It takes away device
828 node access to /dev/rtc as well as the system calls that set the
829 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
830 Note that this option does not affect access to auxiliary services
831 that allow changing the clock, for example access to
834 * The systemd-id128 tool gained a new "show" verb for listing or
835 resolving a number of well-known UUIDs/128bit IDs, currently mostly
836 GPT partition table types.
838 * The Discoverable Partitions Specification has been updated to support
839 /var and /var/tmp partition discovery. Support for this has been
840 added to systemd-gpt-auto-generator. For details see:
842 https://systemd.io/DISCOVERABLE_PARTITIONS
844 * "systemctl list-unit-files" has been updated to show a new column
845 with the suggested enablement state based on the vendor preset files
846 for the respective units.
848 * "systemctl" gained a new option "--with-dependencies". If specified
849 commands such as "systemctl status" or "systemctl cat" will now show
850 all specified units along with all units they depend on.
852 * networkctl gained support for showing per-interface logs in its
855 * systemd-networkd-wait-online gained support for specifying the maximum
856 operational state to wait for, and to wait for interfaces to
859 * The [Match] section of .link and .network files now supports a new
860 option PermanentMACAddress= which may be used to check against the
861 permanent MAC address of a network device even if a randomized MAC
864 * The [TrafficControlQueueingDiscipline] section in .network files has
865 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
866 dropped from the individual setting names.
868 * Any .link and .network files that have an empty [Match] section (this
869 also includes empty and commented-out files) will now be
870 rejected. systemd-udev and systemd-networkd started warning about
871 such files in version 243.
873 * systemd-logind will now validate access to the operation of changing
874 the virtual terminal via a polkit action. By default, only users
875 with at least one session on a local VT are granted permission.
877 * When systemd sets up PAM sessions that invoked service processes
878 shall run in, the pam_setcred() API is now invoked, thus permitting
879 PAM modules to set additional credentials for the processes.
881 * portablectl attach/detach verbs now accept --now and --enable options
882 to combine attachment with enablement and invocation, or detachment
883 with stopping and disablement.
885 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
886 fixed, which in turn exposed bugs in unit configuration of services
887 which have Type=oneshot and should only run once, but do not have
888 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
889 service may be started again after exiting successfully, for example
890 as a dependency in another transaction. Affected services included
891 some internal systemd services (most notably
892 systemd-vconsole-setup.service, which was updated to have
893 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
894 plymouth has been suitably updated or patched before upgrading to
895 this systemd release. See
896 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
897 additional discussion.
899 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
900 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
901 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
902 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
903 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
904 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
905 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
906 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
907 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
908 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
909 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
910 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
911 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
912 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
913 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
914 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
915 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
916 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
917 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
918 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
919 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
920 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
921 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
922 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
929 * Support for the cpuset cgroups v2 controller has been added.
930 Processes may be restricted to specific CPUs using the new
931 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
932 AllowedMemoryNodes= setting.
934 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
935 now be configured using a new RestartKillSignal= setting. This
936 allows units which signals to request termination to implement
937 different behaviour when stopping in preparation for a restart.
939 * "systemctl clean" may now be used also for socket, mount, and swap
942 * systemd will also read configuration options from the EFI variable
943 SystemdOptions. This may be used to configure systemd behaviour when
944 modifying the kernel command line is inconvenient, but configuration
945 on disk is read too late, for example for the options related to
946 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
947 set the EFI variable.
949 * systemd will now disable printk ratelimits in early boot. This should
950 allow us to capture more logs from the early boot phase where normal
951 storage is not available and the kernel ring buffer is used for
952 logging. Configuration on the kernel command line has higher priority
953 and overrides the systemd setting.
955 systemd programs which log to /dev/kmsg directly use internal
956 ratelimits to prevent runaway logging. (Normally this is only used
957 during early boot, so in practice this change has very little
960 * Unit files now support top level dropin directories of the form
961 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
962 that affects all corresponding unit files.
964 * systemctl gained support for 'stop --job-mode=triggering' which will
965 stop the specified unit and any units which could trigger it.
967 * Unit status display now includes units triggering and triggered by
968 the unit being shown.
970 * The RuntimeMaxSec= setting is now supported by scopes, not just
971 .service units. This is particularly useful for PAM sessions which
972 create a scope unit for the user login. systemd.runtime_max_sec=
973 setting may used with the pam_systemd module to limit the duration
974 of the PAM session, for example for time-limited logins.
976 * A new @pkey system call group is now defined to make it easier to
977 allow-list memory protection syscalls for containers and services
978 which need to use them.
980 * systemd-udevd: removed the 30s timeout for killing stale workers on
981 exit. systemd-udevd now waits for workers to finish. The hard-coded
982 exit timeout of 30s was too short for some large installations, where
983 driver initialization could be prematurely interrupted during initrd
984 processing if the root file system had been mounted and init was
985 preparing to switch root. If udevd is run without systemd and workers
986 are hanging while udevd receives an exit signal, udevd will now exit
987 when udev.event_timeout is reached for the last hanging worker. With
988 systemd, the exit timeout can additionally be configured using
989 TimeoutStopSec= in systemd-udevd.service.
991 * udev now provides a program (fido_id) that identifies FIDO CTAP1
992 ("U2F")/CTAP2 security tokens based on the usage declared in their
993 report and descriptor and outputs suitable environment variables.
994 This replaces the externally maintained allow lists of all known
995 security tokens that were used previously.
997 * Automatically generated autosuspend udev rules for allow-listed
998 devices have been imported from the Chromium OS project. This should
999 improve power saving with many more devices.
1001 * udev gained a new "CONST{key}=value" setting that allows matching
1002 against system-wide constants without forking a helper binary.
1003 Currently "arch" and "virt" keys are supported.
1005 * udev now opens CDROMs in non-exclusive mode when querying their
1006 capabilities. This should fix issues where other programs trying to
1007 use the CDROM cannot gain access to it, but carries a risk of
1008 interfering with programs writing to the disk, if they did not open
1009 the device in exclusive mode as they should.
1011 * systemd-networkd does not create a default route for IPv4 link local
1012 addressing anymore. The creation of the route was unexpected and was
1013 breaking routing in various cases, but people who rely on it being
1014 created implicitly will need to adjust. Such a route may be requested
1015 with DefaultRouteOnDevice=yes.
1017 Similarly, systemd-networkd will not assign a link-local IPv6 address
1018 when IPv6 link-local routing is not enabled.
1020 * Receive and transmit buffers may now be configured on links with
1021 the new RxBufferSize= and TxBufferSize= settings.
1023 * systemd-networkd may now advertise additional IPv6 routes. A new
1024 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
1027 * systemd-networkd may now configure "next hop" routes using the
1028 [NextHop] section and Gateway= and Id= settings.
1030 * systemd-networkd will now retain DHCP config on restarts by default
1031 (but this may be overridden using the KeepConfiguration= setting).
1032 The default for SendRelease= has been changed to true.
1034 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
1035 received from the server.
1037 The client will use the received SIP server list if UseSIP=yes is
1040 The client may be configured to request specific options from the
1041 server using a new RequestOptions= setting.
1043 The client may be configured to send arbitrary options to the server
1044 using a new SendOption= setting.
1046 A new IPServiceType= setting has been added to configure the "IP
1047 service type" value used by the client.
1049 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
1050 request prefix hints in the DHCPv6 solicitation.
1052 * The DHCPv4 server may be configured to send arbitrary options using
1053 a new SendOption= setting.
1055 * The DHCPv4 server may now be configured to emit SIP server list using
1056 the new EmitSIP= and SIP= settings.
1058 * systemd-networkd and networkctl may now renew DHCP leases on demand.
1059 networkctl has a new 'networkctl renew' verb.
1061 * systemd-networkd may now reconfigure links on demand. networkctl
1062 gained two new verbs: "reload" will reload the configuration, and
1063 "reconfigure DEVICE…" will reconfigure one or more devices.
1065 * .network files may now match on SSID and BSSID of a wireless network,
1066 i.e. the access point name and hardware address using the new SSID=
1067 and BSSID= options. networkctl will display the current SSID and
1068 BSSID for wireless links.
1070 .network files may also match on the wireless network type using the
1071 new WLANInterfaceType= option.
1073 * systemd-networkd now includes default configuration that enables
1074 link-local addressing when connected to an ad-hoc wireless network.
1076 * systemd-networkd may configure the Traffic Control queueing
1077 disciplines in the kernel using the new
1078 [TrafficControlQueueingDiscipline] section and Parent=,
1079 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
1080 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
1081 NetworkEmulatorDuplicateRate= settings.
1083 * systemd-tmpfiles gained a new w+ setting to append to files.
1085 * systemd-analyze dump will now report when the memory configuration in
1086 the kernel does not match what systemd has configured (usually,
1087 because some external program has modified the kernel configuration
1090 * systemd-analyze gained a new --base-time= switch instructs the
1091 'calendar' verb to resolve times relative to that timestamp instead
1092 of the present time.
1094 * journalctl --update-catalog now produces deterministic output (making
1095 reproducible image builds easier).
1097 * A new devicetree-overlay setting is now documented in the Boot Loader
1100 * The default value of the WatchdogSec= setting used in systemd
1101 services (the ones bundled with the project itself) may be set at
1102 configuration time using the -Dservice-watchdog= setting. If set to
1103 empty, the watchdogs will be disabled.
1105 * systemd-resolved validates IP addresses in certificates now when GnuTLS
1108 * libcryptsetup >= 2.0.1 is now required.
1110 * A configuration option -Duser-path= may be used to override the $PATH
1111 used by the user service manager. The default is again to use the same
1112 path as the system manager.
1114 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
1115 outputting the 128bit IDs in UUID format (i.e. in the "canonical
1118 * Service units gained a new sandboxing option ProtectKernelLogs= which
1119 makes sure the program cannot get direct access to the kernel log
1120 buffer anymore, i.e. the syslog() system call (not to be confused
1121 with the API of the same name in libc, which is not affected), the
1122 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
1123 inaccessible to the service. It's recommended to enable this setting
1124 for all services that should not be able to read from or write to the
1125 kernel log buffer, which are probably almost all.
1127 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
1128 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
1129 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
1130 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
1131 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
1132 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
1133 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
1134 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
1135 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
1136 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
1137 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
1138 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
1139 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
1140 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
1141 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
1142 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
1143 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
1144 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
1145 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
1146 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
1147 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
1148 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
1149 Zach Smith, Zbigniew Jędrzejewski-Szmek
1151 – Warsaw, 2019-11-29
1155 * This release enables unprivileged programs (i.e. requiring neither
1156 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
1157 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
1158 kernel for the whole UNIX group range, i.e. all processes. This
1159 change should be reasonably safe, as the kernel support for it was
1160 specifically implemented to allow safe access to ICMP Echo for
1161 processes lacking any privileges. If this is not desirable, it can be
1162 disabled again by setting the parameter to "1 0".
1164 * Previously, filters defined with SystemCallFilter= would have the
1165 effect that any calling of an offending system call would terminate
1166 the calling thread. This behaviour never made much sense, since
1167 killing individual threads of unsuspecting processes is likely to
1168 create more problems than it solves. With this release the default
1169 action changed from killing the thread to killing the whole
1170 process. For this to work correctly both a kernel version (>= 4.14)
1171 and a libseccomp version (>= 2.4.0) supporting this new seccomp
1172 action is required. If an older kernel or libseccomp is used the old
1173 behaviour continues to be used. This change does not affect any
1174 services that have no system call filters defined, or that use
1175 SystemCallErrorNumber= (and thus see EPERM or another error instead
1176 of being killed when calling an offending system call). Note that
1177 systemd documentation always claimed that the whole process is
1178 killed. With this change behaviour is thus adjusted to match the
1181 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
1182 4194304 by default, i.e. the full 22bit range the kernel allows, up
1183 from the old 16bit range. This should improve security and
1184 robustness, as PID collisions are made less likely (though certainly
1185 still possible). There are rumours this might create compatibility
1186 problems, though at this moment no practical ones are known to
1187 us. Downstream distributions are hence advised to undo this change in
1188 their builds if they are concerned about maximum compatibility, but
1189 for everybody else we recommend leaving the value bumped. Besides
1190 improving security and robustness this should also simplify things as
1191 the maximum number of allowed concurrent tasks was previously bounded
1192 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
1193 only a single knob is left ("kernel.threads-max"). There have been
1194 concerns that usability is affected by this change because larger PID
1195 numbers are harder to type, but we believe the change from 5 digits
1196 to 7 digits doesn't hamper usability.
1198 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
1199 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
1200 hierarchically set default memory protection values for a particular
1201 subtree of the unit hierarchy.
1203 * Memory protection directives can now take a value of zero, allowing
1204 explicit opting out of a default value propagated by an ancestor.
1206 * systemd now defaults to the "unified" cgroup hierarchy setup during
1207 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
1208 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
1209 change reflects the fact that cgroupsv2 support has matured
1210 substantially in both systemd and in the kernel, and is clearly the
1211 way forward. Downstream production distributions might want to
1212 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
1213 their builds as unfortunately the popular container managers have not
1214 caught up with the kernel API changes.
1216 * Man pages are not built by default anymore (html pages were already
1217 disabled by default), to make development builds quicker. When
1218 building systemd for a full installation with documentation, meson
1219 should be called with -Dman=true and/or -Dhtml=true as appropriate.
1220 The default was changed based on the assumption that quick one-off or
1221 repeated development builds are much more common than full optimized
1222 builds for installation, and people need to pass various other
1223 options to when doing "proper" builds anyway, so the gain from making
1224 development builds quicker is bigger than the one time disruption for
1227 Two scripts are created in the *build* directory to generate and
1228 preview man and html pages on demand, e.g.:
1230 build/man/man systemctl
1231 build/man/html systemd.index
1233 * libidn2 is used by default if both libidn2 and libidn are installed.
1234 Please use -Dlibidn=true if libidn is preferred.
1236 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
1237 big-endian machines. Before, bytes were written and read in native
1238 machine order as exposed by the native libc __cpu_mask interface.
1239 Now, little-endian order is always used (CPUs 0–7 are described by
1240 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
1241 This change fixes D-Bus calls that cross endianness boundary.
1243 The presentation format used for CPUAffinity= by "systemctl show" and
1244 "systemd-analyze dump" is changed to present CPU indices instead of
1245 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
1246 shown as CPUAffinity=03000000000000000000000000000… (on
1247 little-endian) or CPUAffinity=00000000000000300000000000000… (on
1248 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
1249 input format. The maximum integer that will be printed in the new
1250 format is 8191 (four digits), while the old format always used a very
1251 long number (with the length varying by architecture), so they can be
1252 unambiguously distinguished.
1254 * /usr/sbin/halt.local is no longer supported. Implementation in
1255 distributions was inconsistent and it seems this functionality was
1258 To replace this functionality, users should:
1259 - either define a new unit and make it a dependency of final.target
1260 (systemctl add-wants final.target my-halt-local.service)
1261 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
1262 and ensure that it accepts "halt", "poweroff", "reboot", and
1263 "kexec" as an argument, see the description in systemd-shutdown(8).
1265 * When a [Match] section in .link or .network file is empty (contains
1266 no match patterns), a warning will be emitted. Please add any "match
1267 all" pattern instead, e.g. OriginalName=* or Name=* in case all
1268 interfaces should really be matched.
1270 * A new setting NUMAPolicy= may be used to set process memory
1271 allocation policy. This setting can be specified in
1272 /etc/systemd/system.conf and hence will set the default policy for
1273 PID1. The default policy can be overridden on a per-service
1274 basis. The related setting NUMAMask= is used to specify NUMA node
1275 mask that should be associated with the selected policy.
1277 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
1278 generates when processes it manages are reaching their memory limits,
1279 and will place their units in a special state, and optionally kill or
1280 stop the whole unit.
1282 * The service manager will now expose bus properties for the IO
1283 resources used by units. This information is also shown in "systemctl
1284 status" now (for services that have IOAccounting=yes set). Moreover,
1285 the IO accounting data is included in the resource log message
1286 generated whenever a unit stops.
1288 * Units may now configure an explicit timeout to wait for when killed
1289 with SIGABRT, for example when a service watchdog is hit. Previously,
1290 the regular TimeoutStopSec= timeout was applied in this case too —
1291 now a separate timeout may be set using TimeoutAbortSec=.
1293 * Services may now send a special WATCHDOG=trigger message with
1294 sd_notify() to trigger an immediate "watchdog missed" event, and thus
1295 trigger service termination. This is useful both for testing watchdog
1296 handling, but also for defining error paths in services, that shall
1297 be handled the same way as watchdog events.
1299 * There are two new per-unit settings IPIngressFilterPath= and
1300 IPEgressFilterPath= which allow configuration of a BPF program
1301 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
1302 to apply to the IP packet ingress/egress path of all processes of a
1303 unit. This is useful to allow running systemd services with BPF
1304 programs set up externally.
1306 * systemctl gained a new "clean" verb for removing the state, cache,
1307 runtime or logs directories of a service while it is terminated. The
1308 new verb may also be used to remove the state maintained on disk for
1309 timer units that have Persistent= configured.
1311 * During the last phase of shutdown systemd will now automatically
1312 increase the log level configured in the "kernel.printk" sysctl so
1313 that any relevant loggable events happening during late shutdown are
1314 made visible. Previously, loggable events happening so late during
1315 shutdown were generally lost if the "kernel.printk" sysctl was set to
1316 high thresholds, as regular logging daemons are terminated at that
1317 time and thus nothing is written to disk.
1319 * If processes terminated during the last phase of shutdown do not exit
1320 quickly systemd will now show their names after a short time, to make
1321 debugging easier. After a longer timeout they are forcibly killed,
1324 * journalctl (and the other tools that display logs) will now highlight
1325 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
1326 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
1327 are now shown in blue color, to separate them visually from regular
1328 logs. References to configuration files are now turned into clickable
1329 links on terminals that support that.
1331 * systemd-journald will now stop logging to /var/log/journal during
1332 shutdown when /var/ is on a separate mount, so that it can be
1333 unmounted safely during shutdown.
1335 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
1337 * systemd-resolved "Cache=" configuration option in resolved.conf has
1338 been extended to also accept the 'no-negative' value. Previously,
1339 only a boolean option was allowed (yes/no), having yes as the
1340 default. If this option is set to 'no-negative', negative answers are
1341 not cached while the old cache heuristics are used positive answers.
1342 The default remains unchanged.
1344 * The predictable naming scheme for network devices now supports
1345 generating predictable names for "netdevsim" devices.
1347 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
1350 Those two changes form a new net.naming-policy-scheme= entry.
1351 Distributions which want to preserve naming stability may want to set
1352 the -Ddefault-net-naming-scheme= configuration option.
1354 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
1355 interfaces natively.
1357 * systemd-networkd's bridge FDB support now allows configuration of a
1358 destination address for each entry (Destination=), as well as the
1359 VXLAN VNI (VNI=), as well as an option to declare what an entry is
1360 associated with (AssociatedWith=).
1362 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
1363 option for configuring the maximum number of DHCP lease requests. It
1364 also learnt a new BlackList= option for deny-listing DHCP servers (a
1365 similar setting has also been added to the IPv6 RA client), as well
1366 as a SendRelease= option for configuring whether to send a DHCP
1367 RELEASE message when terminating.
1369 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
1370 separately in the [DHCPv4] and [DHCPv6] sections.
1372 * systemd-networkd's DHCP support will now optionally create an
1373 implicit host route to the DNS server specified in the DHCP lease, in
1374 addition to the routes listed explicitly in the lease. This should
1375 ensure that in multi-homed systems DNS traffic leaves the systems on
1376 the interface that acquired the DNS server information even if other
1377 routes such as default routes exist. This behaviour may be turned on
1378 with the new RoutesToDNS= option.
1380 * systemd-networkd's VXLAN support gained a new option
1381 GenericProtocolExtension= for enabling VXLAN Generic Protocol
1382 Extension support, as well as IPDoNotFragment= for setting the IP
1383 "Don't fragment" bit on outgoing packets. A similar option has been
1384 added to the GENEVE support.
1386 * In systemd-networkd's [Route] section you may now configure
1387 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
1388 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
1389 propagation. The Type= setting now supports local, broadcast,
1390 anycast, multicast, any, xresolve routes, too.
1392 * systemd-networkd's [Network] section learnt a new option
1393 DefaultRouteOnDevice= for automatically configuring a default route
1394 onto the network device.
1396 * systemd-networkd's bridging support gained two new options ProxyARP=
1397 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
1398 MulticastRouter= for configuring multicast routing behaviour. A new
1399 option MulticastIGMPVersion= may be used to change bridge's multicast
1400 Internet Group Management Protocol (IGMP) version.
1402 * systemd-networkd's FooOverUDP support gained the ability to configure
1403 local and peer IP addresses via Local= and Peer=. A new option
1404 PeerPort= may be used to configure the peer's IP port.
1406 * systemd-networkd's TUN support gained a new setting VnetHeader= for
1407 tweaking Generic Segment Offload support.
1409 * The address family for policy rules may be specified using the new
1410 Family= option in the [RoutingPolicyRule] section.
1412 * networkctl gained a new "delete" command for removing virtual network
1413 devices, as well as a new "--stats" switch for showing device
1416 * networkd.conf gained a new setting SpeedMeter= and
1417 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
1418 measured speed may be shown by 'networkctl status'.
1420 * "networkctl status" now displays MTU and queue lengths, and more
1421 detailed information about VXLAN and bridge devices.
1423 * systemd-networkd's .network and .link files gained a new Property=
1424 setting in the [Match] section, to match against devices with
1425 specific udev properties.
1427 * systemd-networkd's tunnel support gained a new option
1428 AssignToLoopback= for selecting whether to use the loopback device
1429 "lo" as underlying device.
1431 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
1432 been renamed to LinkLayerAddress=, and it now allows configuration of
1435 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
1436 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
1437 IPv6 configuration (static or DHCPv6) was found for a given
1438 interface. It will not touch the sysctl otherwise.
1440 * The order of entries is $PATH used by the user manager instance was
1441 changed to put bin/ entries before the corresponding sbin/ entries.
1442 It is recommended to not rely on this order, and only ever have one
1443 binary with a given name in the system paths under /usr.
1445 * A new tool systemd-network-generator has been added that may generate
1446 .network, .netdev and .link files from IP configuration specified on
1447 the kernel command line in the format used by Dracut.
1449 * The CriticalConnection= setting in .network files is now deprecated,
1450 and replaced by a new KeepConfiguration= setting which allows more
1451 detailed configuration of the IP configuration to keep in place.
1453 * systemd-analyze gained a few new verbs:
1455 - "systemd-analyze timestamp" parses and converts timestamps. This is
1456 similar to the existing "systemd-analyze calendar" command which
1457 does the same for recurring calendar events.
1459 - "systemd-analyze timespan" parses and converts timespans (i.e.
1460 durations as opposed to points in time).
1462 - "systemd-analyze condition" will parse and test ConditionXYZ=
1465 - "systemd-analyze exit-status" will parse and convert exit status
1466 codes to their names and back.
1468 - "systemd-analyze unit-files" will print a list of all unit
1469 file paths and unit aliases.
1471 * SuccessExitStatus=, RestartPreventExitStatus=, and
1472 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
1473 is equivalent to "65"). Those exit status name mappings may be
1474 displayed with the systemd-analyze exit-status verb describe above.
1476 * systemd-logind now exposes a per-session SetBrightness() bus call,
1477 which may be used to securely change the brightness of a kernel
1478 brightness device, if it belongs to the session's seat. By using this
1479 call unprivileged clients can make changes to "backlight" and "leds"
1480 devices securely with strict requirements on session membership.
1481 Desktop environments may use this to generically make brightness
1482 changes to such devices without shipping private SUID binaries or
1483 udev rules for that purpose.
1485 * "udevadm info" gained a --wait-for-initialization switch to wait for
1486 a device to be initialized.
1488 * systemd-hibernate-resume-generator will now look for resumeflags= on
1489 the kernel command line, which is similar to rootflags= and may be
1490 used to configure device timeout for the hibernation device.
1492 * sd-event learnt a new API call sd_event_source_disable_unref() for
1493 disabling and unref'ing an event source in a single function. A
1494 related call sd_event_source_disable_unrefp() has been added for use
1495 with gcc's cleanup extension.
1497 * The sd-id128.h public API gained a new definition
1498 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
1501 * "busctl introspect" gained a new switch --xml-interface for dumping
1502 XML introspection data unmodified.
1504 * PID 1 may now show the unit name instead of the unit description
1505 string in its status output during boot. This may be configured in
1506 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
1507 kernel command line option systemd.status_unit_format=.
1509 * PID 1 now understands a new option KExecWatchdogSec= in
1510 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
1511 Previously watchdog functionality was only available for regular
1512 reboots. The new setting defaults to off, because we don't know in
1513 the general case if the watchdog will be reset after kexec (some
1514 drivers do reset it, but not all), and the new userspace might not be
1515 configured to handle the watchdog.
1517 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
1518 RebootWatchdogSec= to more clearly communicate what it is about. The
1519 old name is still accepted for compatibility.
1521 * The systemd.debug_shell kernel command line option now optionally
1522 takes a tty name to spawn the debug shell on, which allows a
1523 different tty to be selected than the built-in default.
1525 * Service units gained a new ExecCondition= setting which will run
1526 before ExecStartPre= and either continue execution of the unit (for
1527 clean exit codes), stop execution without marking the unit failed
1528 (for exit codes 1 through 254), or stop execution and fail the unit
1529 (for exit code 255 or abnormal termination).
1531 * A new service systemd-pstore.service has been added that pulls data
1532 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
1535 * timedatectl gained new verbs for configuring per-interface NTP
1536 service configuration for systemd-timesyncd.
1538 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
1539 2019. (You can set non-UTF-8 locales though, if you know their name.)
1541 * If variable assignments in sysctl.d/ files are prefixed with "-" any
1542 failures to apply them are now ignored.
1544 * systemd-random-seed.service now optionally credits entropy when
1545 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
1546 true for the service to enable this behaviour, but please consult the
1547 documentation first, since this comes with a couple of caveats.
1549 * systemd-random-seed.service is now a synchronization point for full
1550 initialization of the kernel's entropy pool. Services that require
1551 /dev/urandom to be correctly initialized should be ordered after this
1554 * The systemd-boot boot loader has been updated to optionally maintain
1555 a random seed file in the EFI System Partition (ESP). During the boot
1556 phase, this random seed is read and updated with a new seed
1557 cryptographically derived from it. Another derived seed is passed to
1558 the OS. The latter seed is then credited to the kernel's entropy pool
1559 very early during userspace initialization (from PID 1). This allows
1560 systems to boot up with a fully initialized kernel entropy pool from
1561 earliest boot on, and thus entirely removes all entropy pool
1562 initialization delays from systems using systemd-boot. Special care
1563 is taken to ensure different seeds are derived on system images
1564 replicated to multiple systems. "bootctl status" will show whether
1565 a seed was received from the boot loader.
1567 * bootctl gained two new verbs:
1569 - "bootctl random-seed" will generate the file in ESP and an EFI
1570 variable to allow a random seed to be passed to the OS as described
1573 - "bootctl is-installed" checks whether systemd-boot is currently
1576 * bootctl will warn if it detects that boot entries are misconfigured
1577 (for example if the kernel image was removed without purging the
1580 * A new document has been added describing systemd's use and support
1581 for the kernel's entropy pool subsystem:
1583 https://systemd.io/RANDOM_SEEDS
1585 * When the system is hibernated the swap device to write the
1586 hibernation image to is now automatically picked from all available
1587 swap devices, preferring the swap device with the highest configured
1588 priority over all others, and picking the device with the most free
1589 space if there are multiple devices with the highest priority.
1591 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
1592 option that permits selecting the timeout how long to wait for a
1593 device with an encryption key before asking for the password.
1595 * IOWeight= has learnt to properly set the IO weight when using the
1596 BFQ scheduler officially found in kernels 5.0+.
1598 * A new mailing list has been created for reporting of security issues:
1599 systemd-security@redhat.com. For mode details, see
1600 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
1602 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
1603 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
1604 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
1605 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
1606 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
1607 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
1608 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
1609 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
1610 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
1611 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
1612 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
1613 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
1614 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
1615 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
1616 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
1617 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
1618 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
1619 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
1620 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
1621 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
1622 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
1623 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
1624 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
1625 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
1626 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
1627 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
1628 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
1629 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
1630 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
1631 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
1633 – Camerino, 2019-09-03
1637 * In .link files, MACAddressPolicy=persistent (the default) is changed
1638 to cover more devices. For devices like bridges, tun, tap, bond, and
1639 similar interfaces that do not have other identifying information,
1640 the interface name is used as the basis for persistent seed for MAC
1641 and IPv4LL addresses. The way that devices that were handled
1642 previously is not changed, and this change is about covering more
1643 devices then previously by the "persistent" policy.
1645 MACAddressPolicy=random may be used to force randomized MACs and
1646 IPv4LL addresses for a device if desired.
1648 Hint: the log output from udev (at debug level) was enhanced to
1649 clarify what policy is followed and which attributes are used.
1650 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
1651 may be used to view this.
1653 Hint: if a bridge interface is created without any slaves, and gains
1654 a slave later, then now the bridge does not inherit slave's MAC.
1655 To inherit slave's MAC, for example, create the following file:
1657 # /etc/systemd/network/98-bridge-inherit-mac.link
1662 MACAddressPolicy=none
1665 * The .device units generated by systemd-fstab-generator and other
1666 generators do not automatically pull in the corresponding .mount unit
1667 as a Wants= dependency. This means that simply plugging in the device
1668 will not cause the mount unit to be started automatically. But please
1669 note that the mount unit may be started for other reasons, in
1670 particular if it is part of local-fs.target, and any unit which
1671 (transitively) depends on local-fs.target is started.
1673 * networkctl list/status/lldp now accept globbing wildcards for network
1674 interface names to match against all existing interfaces.
1676 * The $PIDFILE environment variable is set to point the absolute path
1677 configured with PIDFile= for processes of that service.
1679 * The fallback DNS server list was augmented with Cloudflare public DNS
1680 servers. Use `-Ddns-servers=` to set a different fallback.
1682 * A new special target usb-gadget.target will be started automatically
1683 when a USB Device Controller is detected (which means that the system
1684 is a USB peripheral).
1686 * A new unit setting CPUQuotaPeriodSec= assigns the time period
1687 relatively to which the CPU time quota specified by CPUQuota= is
1690 * A new unit setting ProtectHostname= may be used to prevent services
1691 from modifying hostname information (even if they otherwise would
1692 have privileges to do so).
1694 * A new unit setting NetworkNamespacePath= may be used to specify a
1695 namespace for service or socket units through a path referring to a
1696 Linux network namespace pseudo-file.
1698 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
1699 have an effect on .socket units: when used the listening socket is
1700 created within the configured network namespace instead of the host
1703 * ExecStart= command lines in unit files may now be prefixed with ':'
1704 in which case environment variable substitution is
1705 disabled. (Supported for the other ExecXYZ= settings, too.)
1707 * .timer units gained two new boolean settings OnClockChange= and
1708 OnTimezoneChange= which may be used to also trigger a unit when the
1709 system clock is changed or the local timezone is
1710 modified. systemd-run has been updated to make these options easily
1711 accessible from the command line for transient timers.
1713 * Two new conditions for units have been added: ConditionMemory= may be
1714 used to conditionalize a unit based on installed system
1715 RAM. ConditionCPUs= may be used to conditionalize a unit based on
1716 installed CPU cores.
1718 * The @default system call filter group understood by SystemCallFilter=
1719 has been updated to include the new rseq() system call introduced in
1722 * A new time-set.target has been added that indicates that the system
1723 time has been set from a local source (possibly imprecise). The
1724 existing time-sync.target is stronger and indicates that the time has
1725 been synchronized with a precise external source. Services where
1726 approximate time is sufficient should use the new target.
1728 * "systemctl start" (and related commands) learnt a new
1729 --show-transaction option. If specified brief information about all
1730 jobs queued because of the requested operation is shown.
1732 * systemd-networkd recognizes a new operation state 'enslaved', used
1733 (instead of 'degraded' or 'carrier') for interfaces which form a
1734 bridge, bond, or similar, and an new 'degraded-carrier' operational
1735 state used for the bond or bridge master interface when one of the
1736 enslaved devices is not operational.
1738 * .network files learnt the new IgnoreCarrierLoss= option for leaving
1739 networks configured even if the carrier is lost.
1741 * The RequiredForOnline= setting in .network files may now specify a
1742 minimum operational state required for the interface to be considered
1743 "online" by systemd-networkd-wait-online. Related to this
1744 systemd-networkd-wait-online gained a new option --operational-state=
1745 to configure the same, and its --interface= option was updated to
1746 optionally also take an operational state specific for an interface.
1748 * systemd-networkd-wait-online gained a new setting --any for waiting
1749 for only one of the requested interfaces instead of all of them.
1751 * systemd-networkd now implements L2TP tunnels.
1753 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
1754 may be used to cause autonomous and onlink prefixes received in IPv6
1755 Router Advertisements to be ignored.
1757 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
1758 file settings may be used to tweak bridge behaviour.
1760 * The new TripleSampling= option in .network files may be used to
1761 configure CAN triple sampling.
1763 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
1764 used to point to private or preshared key for a WireGuard interface.
1766 * /etc/crypttab now supports the same-cpu-crypt and
1767 submit-from-crypt-cpus options to tweak encryption work scheduling
1770 * systemd-tmpfiles will now take a BSD file lock before operating on a
1771 contents of directory. This may be used to temporarily exclude
1772 directories from aging by taking the same lock (useful for example
1773 when extracting a tarball into /tmp or /var/tmp as a privileged user,
1774 which might create files with really old timestamps, which
1775 nevertheless should not be deleted). For further details, see:
1777 https://systemd.io/TEMPORARY_DIRECTORIES
1779 * systemd-tmpfiles' h line type gained support for the
1780 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
1781 controlling project quota inheritance.
1783 * sd-boot and bootctl now implement support for an Extended Boot Loader
1784 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
1785 addition to the ESP partition mounted to /efi or /boot/efi.
1786 Configuration file fragments, kernels, initrds and other EFI images
1787 to boot will be loaded from both the ESP and XBOOTLDR partitions.
1788 The XBOOTLDR partition was previously described by the Boot Loader
1789 Specification, but implementation was missing in sd-boot. Support for
1790 this concept allows using the sd-boot boot loader in more
1791 conservative scenarios where the boot loader itself is placed in the
1792 ESP but the kernels to boot (and their metadata) in a separate
1795 * A system may now be booted with systemd.volatile=overlay on the
1796 kernel command line, which causes the root file system to be set up
1797 an overlayfs mount combining the root-only root directory with a
1798 writable tmpfs. In this setup, the underlying root device is not
1799 modified, and any changes are lost at reboot.
1801 * Similar, systemd-nspawn can now boot containers with a volatile
1802 overlayfs root with the new --volatile=overlay switch.
1804 * systemd-nspawn can now consume OCI runtime bundles using a new
1805 --oci-bundle= option. This implementation is fully usable, with most
1806 features in the specification implemented, but since this a lot of
1807 new code and functionality, this feature should most likely not
1808 be used in production yet.
1810 * systemd-nspawn now supports various options described by the OCI
1811 runtime specification on the command-line and in .nspawn files:
1812 --inaccessible=/Inaccessible= may be used to mask parts of the file
1813 system tree, --console=/--pipe may be used to configure how standard
1814 input, output, and error are set up.
1816 * busctl learned the `emit` verb to generate D-Bus signals.
1818 * systemd-analyze cat-config may be used to gather and display
1819 configuration spread over multiple files, for example system and user
1820 presets, tmpfiles.d, sysusers.d, udev rules, etc.
1822 * systemd-analyze calendar now takes an optional new parameter
1823 --iterations= which may be used to show a maximum number of iterations
1824 the specified expression will elapse next.
1826 * The sd-bus C API gained support for naming method parameters in the
1829 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
1830 the reboot() system call expects.
1832 * journalctl learnt a new --cursor-file= option that points to a file
1833 from which a cursor should be loaded in the beginning and to which
1834 the updated cursor should be stored at the end.
1836 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
1837 detected by systemd-detect-virt (and may also be used in
1838 ConditionVirtualization=).
1840 * The behaviour of systemd-logind may now be modified with environment
1841 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
1842 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
1843 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
1844 skip the relevant operation completely (when set to false), or to
1845 create a flag file in /run/systemd (when set to true), instead of
1846 actually commencing the real operation when requested. The presence
1847 of /run/systemd/reboot-to-firmware-setup,
1848 /run/systemd/reboot-to-boot-loader-menu, and
1849 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
1850 boot loader implementations to replace some steps logind performs
1851 during reboot with their own operations.
1853 * systemctl can be used to request a reboot into the boot loader menu
1854 or a specific boot loader entry with the new --boot-load-menu= and
1855 --boot-loader-entry= options to a reboot command. (This requires a
1856 boot loader that supports this, for example sd-boot.)
1858 * kernel-install will no longer unconditionally create the output
1859 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
1860 snippets, but will do only if the machine-specific parent directory
1861 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
1862 to create this parent directory during sd-boot installation.
1864 This makes it easier to use kernel-install with plugins which support
1865 a different layout of the bootloader partitions (for example grub2).
1867 * During package installation (with `ninja install`), we would create
1868 symlinks for getty@tty1.service, systemd-networkd.service,
1869 systemd-networkd.socket, systemd-resolved.service,
1870 remote-cryptsetup.target, remote-fs.target,
1871 systemd-networkd-wait-online.service, and systemd-timesyncd.service
1872 in /etc, as if `systemctl enable` was called for those units, to make
1873 the system usable immediately after installation. Now this is not
1874 done anymore, and instead calling `systemctl preset-all` is
1875 recommended after the first installation of systemd.
1877 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
1878 is built on seccomp. When turned on creation of SUID/SGID files is
1881 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
1882 implied if DynamicUser= is turned on for a service. This hardens
1883 these services, so that they neither can benefit from nor create
1884 SUID/SGID executables. This is a minor compatibility breakage, given
1885 that when DynamicUser= was first introduced SUID/SGID behaviour was
1886 unaffected. However, the security benefit of these two options is
1887 substantial, and the setting is still relatively new, hence we opted
1888 to make it mandatory for services with dynamic users.
1890 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
1891 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
1892 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
1893 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
1894 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
1895 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
1896 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
1897 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
1898 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
1899 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
1900 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
1901 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
1902 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
1903 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
1904 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
1905 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
1906 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
1907 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1909 — Warsaw, 2019-04-11
1913 * The default locale can now be configured at compile time. Otherwise,
1914 a suitable default will be selected automatically (one of C.UTF-8,
1915 en_US.UTF-8, and C).
1917 * The version string shown by systemd and other tools now includes the
1918 git commit hash when built from git. An override may be specified
1919 during compilation, which is intended to be used by distributions to
1920 include the package release information.
1922 * systemd-cat can now filter standard input and standard error streams
1923 for different syslog priorities using the new --stderr-priority=
1926 * systemd-journald and systemd-journal-remote reject entries which
1927 contain too many fields (CVE-2018-16865) and set limits on the
1928 process' command line length (CVE-2018-16864).
1930 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
1933 * A new network device NamePolicy "keep" is implemented for link files,
1934 and used by default in 99-default.link (the fallback configuration
1935 provided by systemd). With this policy, if the network device name
1936 was already set by userspace, the device will not be renamed again.
1937 This matches the naming scheme that was implemented before
1938 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
1939 is also enabled by default, even if not specified. Effectively, this
1940 means that if naming-scheme >= 240 is specified, network devices will
1941 be renamed according to the configuration, even if they have been
1942 renamed already, if "keep" is not specified as the naming policy in
1943 the .link file. The 99-default.link file provided by systemd includes
1944 "keep" for backwards compatibility, but it is recommended for user
1945 installed .link files to *not* include it.
1947 The "kernel" policy, which keeps kernel names declared to be
1948 "persistent", now works again as documented.
1950 * kernel-install script now optionally takes the paths to one or more
1951 initrd files, and passes them to all plugins.
1953 * The mincore() system call has been dropped from the @system-service
1954 system call filter group, as it is pretty exotic and may potentially
1955 used for side-channel attacks.
1957 * -fPIE is dropped from compiler and linker options. Please specify
1958 -Db_pie=true option to meson to build position-independent
1959 executables. Note that the meson option is supported since meson-0.49.
1961 * The fs.protected_regular and fs.protected_fifos sysctls, which were
1962 added in Linux 4.19 to make some data spoofing attacks harder, are
1963 now enabled by default. While this will hopefully improve the
1964 security of most installations, it is technically a backwards
1965 incompatible change; to disable these sysctls again, place the
1966 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
1968 fs.protected_regular = 0
1969 fs.protected_fifos = 0
1971 Note that the similar hardlink and symlink protection has been
1972 enabled since v199, and may be disabled likewise.
1974 * The files read from the EnvironmentFile= setting in unit files now
1975 parse backslashes inside quotes literally, matching the behaviour of
1978 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
1979 now automatically become NOPs when run in a chroot() environment.
1981 * The tmpfiles.d/ "C" line type will now copy directory trees not only
1982 when the destination is so far missing, but also if it already exists
1983 as a directory and is empty. This is useful to cater for systems
1984 where directory trees are put together from multiple separate mount
1985 points but otherwise empty.
1987 * A new function sd_bus_close_unref() (and the associated
1988 sd_bus_close_unrefp()) has been added to libsystemd, that combines
1989 sd_bus_close() and sd_bus_unref() in one.
1991 * udevadm control learnt a new option for --ping for testing whether a
1992 systemd-udevd instance is running and reacting.
1994 * udevadm trigger learnt a new option for --wait-daemon for waiting
1995 systemd-udevd daemon to be initialized.
1997 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
1998 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
1999 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
2000 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
2001 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
2002 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
2003 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
2004 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
2005 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
2006 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
2007 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
2008 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
2009 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
2010 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
2011 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
2012 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
2013 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
2015 — Berlin, 2019-02-14
2019 * NoNewPrivileges=yes has been set for all long-running services
2020 implemented by systemd. Previously, this was problematic due to
2021 SELinux (as this would also prohibit the transition from PID1's label
2022 to the service's label). This restriction has since been lifted, but
2023 an SELinux policy update is required.
2024 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
2026 * DynamicUser=yes is dropped from systemd-networkd.service,
2027 systemd-resolved.service and systemd-timesyncd.service, which was
2028 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
2029 and since v236 for systemd-timesyncd.service. The users and groups
2030 systemd-network, systemd-resolve and systemd-timesync are created
2031 by systemd-sysusers again. Distributors or system administrators
2032 may need to create these users and groups if they not exist (or need
2033 to re-enable DynamicUser= for those units) while upgrading systemd.
2034 Also, the clock file for systemd-timesyncd may need to move from
2035 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
2037 * When unit files are loaded from disk, previously systemd would
2038 sometimes (depending on the unit loading order) load units from the
2039 target path of symlinks in .wants/ or .requires/ directories of other
2040 units. This meant that unit could be loaded from different paths
2041 depending on whether the unit was requested explicitly or as a
2042 dependency of another unit, not honouring the priority of directories
2043 in search path. It also meant that it was possible to successfully
2044 load and start units which are not found in the unit search path, as
2045 long as they were requested as a dependency and linked to from
2046 .wants/ or .requires/. The target paths of those symlinks are not
2047 used for loading units anymore and the unit file must be found in
2050 * A new service type has been added: Type=exec. It's very similar to
2051 Type=simple but ensures the service manager will wait for both fork()
2052 and execve() of the main service binary to complete before proceeding
2053 with follow-up units. This is primarily useful so that the manager
2054 propagates any errors in the preparation phase of service execution
2055 back to the job that requested the unit to be started. For example,
2056 consider a service that has ExecStart= set to a file system binary
2057 that doesn't exist. With Type=simple starting the unit would be
2058 considered instantly successful, as only fork() has to complete
2059 successfully and the manager does not wait for execve(), and hence
2060 its failure is seen "too late". With the new Type=exec service type
2061 starting the unit will fail, as the manager will wait for the
2062 execve() and notice its failure, which is then propagated back to the
2065 NOTE: with the next release 241 of systemd we intend to change the
2066 systemd-run tool to default to Type=exec for transient services
2067 started by it. This should be mostly safe, but in specific corner
2068 cases might result in problems, as the systemd-run tool will then
2069 block on NSS calls (such as user name look-ups due to User=) done
2070 between the fork() and execve(), which under specific circumstances
2071 might cause problems. It is recommended to specify "-p Type=simple"
2072 explicitly in the few cases where this applies. For regular,
2073 non-transient services (i.e. those defined with unit files on disk)
2074 we will continue to default to Type=simple.
2076 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
2077 userspace processes is set to 1024 (soft) and 4096
2078 (hard). Previously, systemd passed this on unmodified to all
2079 processes it forked off. With this systemd release the hard limit
2080 systemd passes on is increased to 512K, overriding the kernel's
2081 defaults and substantially increasing the number of simultaneous file
2082 descriptors unprivileged userspace processes can allocate. Note that
2083 the soft limit remains at 1024 for compatibility reasons: the
2084 traditional UNIX select() call cannot deal with file descriptors >=
2085 1024 and increasing the soft limit globally might thus result in
2086 programs unexpectedly allocating a high file descriptor and thus
2087 failing abnormally when attempting to use it with select() (of
2088 course, programs shouldn't use select() anymore, and prefer
2089 poll()/epoll, but the call unfortunately remains undeservedly popular
2090 at this time). This change reflects the fact that file descriptor
2091 handling in the Linux kernel has been optimized in more recent
2092 kernels and allocating large numbers of them should be much cheaper
2093 both in memory and in performance than it used to be. Programs that
2094 want to take benefit of the increased limit have to "opt-in" into
2095 high file descriptors explicitly by raising their soft limit. Of
2096 course, when they do that they must acknowledge that they cannot use
2097 select() anymore (and neither can any shared library they use — or
2098 any shared library used by any shared library they use and so on).
2099 Which default hard limit is most appropriate is of course hard to
2100 decide. However, given reports that ~300K file descriptors are used
2101 in real-life applications we believe 512K is sufficiently high as new
2102 default for now. Note that there are also reports that using very
2103 high hard limits (e.g. 1G) is problematic: some software allocates
2104 large arrays with one element for each potential file descriptor
2105 (Java, …) — a high hard limit thus triggers excessively large memory
2106 allocations in these applications. Hopefully, the new default of 512K
2107 is a good middle ground: higher than what real-life applications
2108 currently need, and low enough for avoid triggering excessively large
2109 allocations in problematic software. (And yes, somebody should fix
2112 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
2113 to the highest possible values, as separate accounting of file
2114 descriptors is no longer necessary, as memcg tracks them correctly as
2115 part of the memory accounting anyway. Thus, from the four limits on
2116 file descriptors currently enforced (fs.file-max, fs.nr_open,
2117 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
2118 and keep only the latter two. A set of build-time options
2119 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
2120 has been added to revert this change in behaviour, which might be
2121 an option for systems that turn off memcg in the kernel.
2123 * When no /etc/locale.conf file exists (and hence no locale settings
2124 are in place), systemd will now use the "C.UTF-8" locale by default,
2125 and set LANG= to it. This locale is supported by various
2126 distributions including Fedora, with clear indications that upstream
2127 glibc is going to make it available too. This locale enables UTF-8
2128 mode by default, which appears appropriate for 2018.
2130 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
2131 default. This effectively switches the RFC3704 Reverse Path filtering
2132 from Strict mode to Loose mode. This is more appropriate for hosts
2133 that have multiple links with routes to the same networks (e.g.
2134 a client with a Wi-Fi and Ethernet both connected to the internet).
2136 Consult the kernel documentation for details on this sysctl:
2137 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
2139 * CPUAccounting=yes no longer enables the CPU controller when using
2140 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
2141 statistics are now provided independently from the CPU controller.
2143 * Support for disabling a particular cgroup controller within a sub-tree
2144 has been added through the DisableControllers= directive.
2146 * cgroup_no_v1=all on the kernel command line now also implies
2147 using the unified cgroup hierarchy, unless one explicitly passes
2148 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
2150 * The new "MemoryMin=" unit file property may now be used to set the
2151 memory usage protection limit of processes invoked by the unit. This
2152 controls the cgroup v2 memory.min attribute. Similarly, the new
2153 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
2154 cgroup v2 io.latency cgroup property for configuring per-service I/O
2157 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
2158 to the cgroup v1 "devices" cgroup controller.
2160 * systemd-escape now is able to combine --unescape with --template. It
2161 also learnt a new option --instance for extracting and unescaping the
2162 instance part of a unit name.
2164 * sd-bus now provides the sd_bus_message_readv() which is similar to
2165 sd_bus_message_read() but takes a va_list object. The pair
2166 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
2167 has been added for configuring the default method call timeout to
2168 use. sd_bus_error_move() may be used to efficiently move the contents
2169 from one sd_bus_error structure to another, invalidating the
2170 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
2171 be used to control whether a bus connection object is automatically
2172 flushed when an sd-event loop is exited.
2174 * When processing classic BSD syslog log messages, journald will now
2175 save the original time-stamp string supplied in the new
2176 SYSLOG_TIMESTAMP= journal field. This permits consumers to
2177 reconstruct the original BSD syslog message more correctly.
2179 * StandardOutput=/StandardError= in service files gained support for
2180 new "append:…" parameters, for connecting STDOUT/STDERR of a service
2181 to a file, and appending to it.
2183 * The signal to use as last step of killing of unit processes is now
2184 configurable. Previously it was hard-coded to SIGKILL, which may now
2185 be overridden with the new KillSignal= setting. Note that this is the
2186 signal used when regular termination (i.e. SIGTERM) does not suffice.
2187 Similarly, the signal used when aborting a program in case of a
2188 watchdog timeout may now be configured too (WatchdogSignal=).
2190 * The XDG_SESSION_DESKTOP environment variable may now be configured in
2191 the pam_systemd argument line, using the new desktop= switch. This is
2192 useful to initialize it properly from a display manager without
2193 having to touch C code.
2195 * Most configuration options that previously accepted percentage values
2196 now also accept permille values with the '‰' suffix (instead of '%').
2198 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
2201 * systemd-resolved's configuration file resolved.conf gained a new
2202 option ReadEtcHosts= which may be used to turn off processing and
2203 honoring /etc/hosts entries.
2205 * The "--wait" switch may now be passed to "systemctl
2206 is-system-running", in which case the tool will synchronously wait
2207 until the system finished start-up.
2209 * hostnamed gained a new bus call to determine the DMI product UUID.
2211 * On x86-64 systemd will now prefer using the RDRAND processor
2212 instruction over /dev/urandom whenever it requires randomness that
2213 neither has to be crypto-grade nor should be reproducible. This
2214 should substantially reduce the amount of entropy systemd requests
2215 from the kernel during initialization on such systems, though not
2216 reduce it to zero. (Why not zero? systemd still needs to allocate
2217 UUIDs and such uniquely, which require high-quality randomness.)
2219 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
2220 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
2221 for forcing the "Other Information" bit in IPv6 RA messages. The
2222 bonding logic gained four new options AdActorSystemPriority=,
2223 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
2224 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
2225 shuffling of flows. The tunnel logic gained a new
2226 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
2227 Deployment. The policy rule logic gained four new options IPProtocol=,
2228 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
2229 support for the MulticastToUnicast= option. networkd also gained
2230 support for configuring static IPv4 ARP or IPv6 neighbor entries.
2232 * .preset files (as read by 'systemctl preset') may now be used to
2233 instantiate services.
2235 * /etc/crypttab now understands the sector-size= option to configure
2236 the sector size for an encrypted partition.
2238 * Key material for encrypted disks may now be placed on a formatted
2239 medium, and referenced from /etc/crypttab by the UUID of the file
2240 system, followed by "=" suffixed by the path to the key file.
2242 * The "collect" udev component has been removed without replacement, as
2243 it is neither used nor maintained.
2245 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
2246 LogsDirectory=, ConfigurationDirectory= settings are used in a
2247 service the executed processes will now receive a set of environment
2248 variables containing the full paths of these directories.
2249 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
2250 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
2251 are used. Note that these options may be used multiple times per
2252 service in which case the resulting paths will be concatenated and
2253 separated by colons.
2255 * Predictable interface naming has been extended to cover InfiniBand
2256 NICs. They will be exposed with an "ib" prefix.
2258 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
2259 which case the respective line failing is ignored.
2261 * .link files may now be used to configure the equivalent to the
2262 "ethtool advertise" commands.
2264 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
2265 alternative to libudev.h. Previously, the latter was just an internal
2266 wrapper around the former, but now these two APIs are exposed
2269 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
2270 which calculates an app-specific boot ID similar to how
2271 sd_id128_get_machine_app_specific() generates an app-specific machine
2274 * A new tool systemd-id128 has been added that can be used to determine
2275 and generate various 128bit IDs.
2277 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
2280 * systemd-hibernate-resume-generator will now honor the "noresume"
2281 kernel command line option, in which case it will bypass resuming
2282 from any hibernated image.
2284 * The systemd-sleep.conf configuration file gained new options
2285 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
2286 AllowHybridSleep= for prohibiting specific sleep modes even if the
2287 kernel exports them.
2289 * portablectl is now officially supported and has thus moved to
2292 * bootctl learnt the two new commands "set-default" and "set-oneshot"
2293 for setting the default boot loader item to boot to (either
2294 persistently or only for the next boot). This is currently only
2295 compatible with sd-boot, but may be implemented on other boot loaders
2296 too, that follow the boot loader interface. The updated interface is
2297 now documented here:
2299 https://systemd.io/BOOT_LOADER_INTERFACE
2301 * A new kernel command line option systemd.early_core_pattern= is now
2302 understood which may be used to influence the core_pattern PID 1
2303 installs during early boot.
2305 * busctl learnt two new options -j and --json= for outputting method
2306 call replies, properties and monitoring output in JSON.
2308 * journalctl's JSON output now supports simple ANSI coloring as well as
2309 a new "json-seq" mode for generating RFC7464 output.
2311 * Unit files now support the %g/%G specifiers that resolve to the UNIX
2312 group/GID of the service manager runs as, similar to the existing
2313 %u/%U specifiers that resolve to the UNIX user/UID.
2315 * systemd-logind learnt a new global configuration option
2316 UserStopDelaySec= that may be set in logind.conf. It specifies how
2317 long the systemd --user instance shall remain started after a user
2318 logs out. This is useful to speed up repetitive re-connections of the
2319 same user, as it means the user's service manager doesn't have to be
2320 stopped/restarted on each iteration, but can be reused between
2321 subsequent options. This setting defaults to 10s. systemd-logind also
2322 exports two new properties on its Manager D-Bus objects indicating
2323 whether the system's lid is currently closed, and whether the system
2326 * systemd gained support for a generic boot counting logic, which
2327 generically permits automatic reverting to older boot loader entries
2328 if newer updated ones don't work. The boot loader side is implemented
2329 in sd-boot, but is kept open for other boot loaders too. For details
2332 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
2334 * The SuccessAction=/FailureAction= unit file settings now learnt two
2335 new parameters: "exit" and "exit-force", which result in immediate
2336 exiting of the service manager, and are only useful in systemd --user
2337 and container environments.
2339 * Unit files gained support for a pair of options
2340 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
2341 exit status to use as service manager exit status when
2342 SuccessAction=/FailureAction= is set to exit or exit-force.
2344 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
2345 options may now be used to configure the log rate limiting applied by
2346 journald per-service.
2348 * systemd-analyze gained a new verb "timespan" for parsing and
2349 normalizing time span values (i.e. strings like "5min 7s 8us").
2351 * systemd-analyze also gained a new verb "security" for analyzing the
2352 security and sand-boxing settings of services in order to determine an
2353 "exposure level" for them, indicating whether a service would benefit
2354 from more sand-boxing options turned on for them.
2356 * "systemd-analyze syscall-filter" will now also show system calls
2357 supported by the local kernel but not included in any of the defined
2360 * .nspawn files now understand the Ephemeral= setting, matching the
2361 --ephemeral command line switch.
2363 * sd-event gained the new APIs sd_event_source_get_floating() and
2364 sd_event_source_set_floating() for controlling whether a specific
2365 event source is "floating", i.e. destroyed along with the even loop
2368 * Unit objects on D-Bus gained a new "Refs" property that lists all
2369 clients that currently have a reference on the unit (to ensure it is
2372 * The JoinControllers= option in system.conf is no longer supported, as
2373 it didn't work correctly, is hard to support properly, is legacy (as
2374 the concept only exists on cgroup v1) and apparently wasn't used.
2376 * Journal messages that are generated whenever a unit enters the failed
2377 state are now tagged with a unique MESSAGE_ID. Similarly, messages
2378 generated whenever a service process exits are now made recognizable,
2379 too. A tagged message is also emitted whenever a unit enters the
2380 "dead" state on success.
2382 * systemd-run gained a new switch --working-directory= for configuring
2383 the working directory of the service to start. A shortcut -d is
2384 equivalent, setting the working directory of the service to the
2385 current working directory of the invoking program. The new --shell
2386 (or just -S) option has been added for invoking the $SHELL of the
2387 caller as a service, and implies --pty --same-dir --wait --collect
2388 --service-type=exec. Or in other words, "systemd-run -S" is now the
2389 quickest way to quickly get an interactive in a fully clean and
2390 well-defined system service context.
2392 * machinectl gained a new verb "import-fs" for importing an OS tree
2393 from a directory. Moreover, when a directory or tarball is imported
2394 and single top-level directory found with the OS itself below the OS
2395 tree is automatically mangled and moved one level up.
2397 * systemd-importd will no longer set up an implicit btrfs loop-back
2398 file system on /var/lib/machines. If one is already set up, it will
2399 continue to be used.
2401 * A new generator "systemd-run-generator" has been added. It will
2402 synthesize a unit from one or more program command lines included in
2403 the kernel command line. This is very useful in container managers
2406 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
2408 This will run "systemd-nspawn" on an image, invoke the specified
2409 command line and immediately shut down the container again, returning
2410 the command line's exit code.
2412 * The block device locking logic is now documented:
2414 https://systemd.io/BLOCK_DEVICE_LOCKING
2416 * loginctl and machinectl now optionally output the various tables in
2417 JSON using the --output= switch. It is our intention to add similar
2418 support to systemctl and all other commands.
2420 * udevadm's query and trigger verb now optionally take a .device unit
2423 * systemd-udevd's network naming logic now understands a new
2424 net.naming-scheme= kernel command line switch, which may be used to
2425 pick a specific version of the naming scheme. This helps stabilizing
2426 interface names even as systemd/udev are updated and the naming logic
2429 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
2430 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
2431 initialize one to all 0xFF.
2433 * After loading the SELinux policy systemd will now recursively relabel
2434 all files and directories listed in
2435 /run/systemd/relabel-extra.d/*.relabel (which should be simple
2436 newline separated lists of paths) in addition to the ones it already
2437 implicitly relabels in /run, /dev and /sys. After the relabelling is
2438 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
2439 removed. This is useful to permit initrds (i.e. code running before
2440 the SELinux policy is in effect) to generate files in the host
2441 filesystem safely and ensure that the correct label is applied during
2442 the transition to the host OS.
2444 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
2445 mknod() handling in user namespaces. Previously mknod() would always
2446 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
2447 but device nodes generated that way cannot be opened, and attempts to
2448 open them result in EPERM. This breaks the "graceful fallback" logic
2449 in systemd's PrivateDevices= sand-boxing option. This option is
2450 implemented defensively, so that when systemd detects it runs in a
2451 restricted environment (such as a user namespace, or an environment
2452 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
2453 where device nodes cannot be created the effect of PrivateDevices= is
2454 bypassed (following the logic that 2nd-level sand-boxing is not
2455 essential if the system systemd runs in is itself already sand-boxed
2456 as a whole). This logic breaks with 4.18 in container managers where
2457 user namespacing is used: suddenly PrivateDevices= succeeds setting
2458 up a private /dev/ file system containing devices nodes — but when
2459 these are opened they don't work.
2461 At this point it is recommended that container managers utilizing
2462 user namespaces that intend to run systemd in the payload explicitly
2463 block mknod() with seccomp or similar, so that the graceful fallback
2466 We are very sorry for the breakage and the requirement to change
2467 container configurations for newer kernels. It's purely caused by an
2468 incompatible kernel change. The relevant kernel developers have been
2469 notified about this userspace breakage quickly, but they chose to
2472 * PermissionsStartOnly= setting is deprecated (but is still supported
2473 for backwards compatibility). The same functionality is provided by
2474 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
2477 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
2478 pam_systemd anymore.
2480 * The naming scheme for network devices was changed to always rename
2481 devices, even if they were already renamed by userspace. The "kernel"
2482 policy was changed to only apply as a fallback, if no other naming
2485 * The requirements to build systemd is bumped to meson-0.46 and
2488 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
2489 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
2490 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
2491 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
2492 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
2493 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
2494 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
2495 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
2496 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
2497 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
2498 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
2499 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
2500 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
2501 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
2502 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
2503 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
2504 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
2505 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
2506 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
2507 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
2508 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
2509 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
2510 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
2511 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
2512 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
2513 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
2514 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
2515 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
2516 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
2517 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
2518 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
2519 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
2520 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
2521 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
2522 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
2523 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
2524 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
2525 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
2526 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
2527 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
2528 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
2529 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
2530 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
2531 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
2532 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
2534 — Warsaw, 2018-12-21
2538 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
2539 builtin will name network interfaces differently than in previous
2540 versions for virtual network interfaces created with SR-IOV and NPAR
2541 and for devices where the PCI network controller device does not have
2542 a slot number associated.
2544 SR-IOV virtual devices are now named based on the name of the parent
2545 interface, with a suffix of "v<N>", where <N> is the virtual device
2546 number. Previously those virtual devices were named as if completely
2549 The ninth and later NPAR virtual devices will be named following the
2550 scheme used for the first eight NPAR partitions. Previously those
2551 devices were not renamed and the kernel default (eth<n>) was used.
2553 "net_id" will also generate names for PCI devices where the PCI
2554 network controller device does not have an associated slot number
2555 itself, but one of its parents does. Previously those devices were
2556 not renamed and the kernel default (eth<n>) was used.
2558 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
2559 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
2560 the unit. So, it is expected that the default behavior of
2561 systemd-logind is not changed. However, if distribution packagers or
2562 administrators disabled or modified IPAddressDeny= setting by a
2563 drop-in config file, then it may be necessary to update the file to
2564 re-enable AF_INET and AF_INET6 to support network user name services,
2567 * When the RestrictNamespaces= unit property is specified multiple
2568 times, then the specified types are merged now. Previously, only the
2569 last assignment was used. So, if distribution packagers or
2570 administrators modified the setting by a drop-in config file, then it
2571 may be necessary to update the file.
2573 * When OnFailure= is used in combination with Restart= on a service
2574 unit, then the specified units will no longer be triggered on
2575 failures that result in restarting. Previously, the specified units
2576 would be activated each time the unit failed, even when the unit was
2577 going to be restarted automatically. This behaviour contradicted the
2578 documentation. With this release the code is adjusted to match the
2581 * systemd-tmpfiles will now print a notice whenever it encounters
2582 tmpfiles.d/ lines referencing the /var/run/ directory. It will
2583 recommend reworking them to use the /run/ directory instead (for
2584 which /var/run/ is simply a symlinked compatibility alias). This way
2585 systemd-tmpfiles can properly detect line conflicts and merge lines
2586 referencing the same file by two paths, without having to access
2589 * systemctl disable/unmask/preset/preset-all cannot be used with
2590 --runtime. Previously this was allowed, but resulted in unintuitive
2591 behaviour that wasn't useful. systemctl disable/unmask will now undo
2592 both runtime and persistent enablement/masking, i.e. it will remove
2593 any relevant symlinks both in /run and /etc.
2595 * Note that all long-running system services shipped with systemd will
2596 now default to a system call allow list (rather than a deny list, as
2597 before). In particular, systemd-udevd will now enforce one too. For
2598 most cases this should be safe, however downstream distributions
2599 which disabled sandboxing of systemd-udevd (specifically the
2600 MountFlags= setting), might want to disable this security feature
2601 too, as the default allow-listing will prohibit all mount, swap,
2602 reboot and clock changing operations from udev rules.
2604 * sd-boot acquired new loader configuration settings to optionally turn
2605 off Windows and MacOS boot partition discovery as well as
2606 reboot-into-firmware menu items. It is also able to pick a better
2607 screen resolution for HiDPI systems, and now provides loader
2608 configuration settings to change the resolution explicitly.
2610 * systemd-resolved now supports DNS-over-TLS. It's still
2611 turned off by default, use DNSOverTLS=opportunistic to turn it on in
2612 resolved.conf. We intend to make this the default as soon as couple
2613 of additional techniques for optimizing the initial latency caused by
2614 establishing a TLS/TCP connection are implemented.
2616 * systemd-resolved.service and systemd-networkd.service now set
2617 DynamicUser=yes. The users systemd-resolve and systemd-network are
2618 not created by systemd-sysusers anymore.
2620 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
2621 that embed a network facing module into any process using getpwuid()
2622 or related call: the dynamic allocation of the user ID for
2623 systemd-resolved.service means the service manager has to check NSS
2624 if the user name is already taken when forking off the service. Since
2625 the user in the common case won't be defined in /etc/passwd the
2626 lookup is likely to trigger nss-ldap which in turn might use NSS to
2627 ask systemd-resolved for hostname lookups. This will hence result in
2628 a deadlock: a user name lookup in order to start
2629 systemd-resolved.service will result in a hostname lookup for which
2630 systemd-resolved.service needs to be started already. There are
2631 multiple ways to work around this problem: pre-allocate the
2632 "systemd-resolve" user on such systems, so that nss-ldap won't be
2633 triggered; or use a different NSS package that doesn't do networking
2634 in-process but provides a local asynchronous name cache; or configure
2635 the NSS package to avoid lookups for UIDs in the range `pkg-config
2636 systemd --variable=dynamicuidmin` … `pkg-config systemd
2637 --variable=dynamicuidmax`, so that it does not consider itself
2638 authoritative for the same UID range systemd allocates dynamic users
2641 * The systemd-resolve tool has been renamed to resolvectl (it also
2642 remains available under the old name, for compatibility), and its
2643 interface is now verb-based, similar in style to the other <xyz>ctl
2644 tools, such as systemctl or loginctl.
2646 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
2647 compatibility. It may be symlinked under the 'resolvconf' name, in
2648 which case it will take arguments and input compatible with the
2649 Debian and FreeBSD resolvconf tool.
2651 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
2652 where the system initially suspends, and after a timeout resumes and
2655 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
2656 set the client will only send a DUID as client identifier.
2658 * The nss-systemd glibc NSS module will now enumerate dynamic users and
2659 groups in effect. Previously, it could resolve UIDs/GIDs to user
2660 names/groups and vice versa, but did not support enumeration.
2662 * journald's Compress= configuration setting now optionally accepts a
2663 byte threshold value. All journal objects larger than this threshold
2664 will be compressed, smaller ones will not. Previously this threshold
2665 was not configurable and set to 512.
2667 * A new system.conf setting NoNewPrivileges= is now available which may
2668 be used to turn off acquisition of new privileges system-wide
2669 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
2670 for all its children). Note that turning this option on means setuid
2671 binaries and file system capabilities lose their special powers.
2672 While turning on this option is a big step towards a more secure
2673 system, doing so is likely to break numerous pre-existing UNIX tools,
2674 in particular su and sudo.
2676 * A new service systemd-time-sync-wait.service has been added. If
2677 enabled it will delay the time-sync.target unit at boot until time
2678 synchronization has been received from the network. This
2679 functionality is useful on systems lacking a local RTC or where it is
2680 acceptable that the boot process shall be delayed by external network
2683 * When hibernating, systemd will now inform the kernel of the image
2684 write offset, on kernels new enough to support this. This means swap
2685 files should work for hibernation now.
2687 * When loading unit files, systemd will now look for drop-in unit files
2688 extensions in additional places. Previously, for a unit file name
2689 "foo-bar-baz.service" it would look for dropin files in
2690 "foo-bar-baz.service.d/*.conf". Now, it will also look in
2691 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
2692 service name truncated after all inner dashes. This scheme allows
2693 writing drop-ins easily that apply to a whole set of unit files at
2694 once. It's particularly useful for mount and slice units (as their
2695 naming is prefix based), but is also useful for service and other
2696 units, for packages that install multiple unit files at once,
2697 following a strict naming regime of beginning the unit file name with
2698 the package's name. Two new specifiers are now supported in unit
2699 files to match this: %j and %J are replaced by the part of the unit
2700 name following the last dash.
2702 * Unit files and other configuration files that support specifier
2703 expansion now understand another three new specifiers: %T and %V will
2704 resolve to /tmp and /var/tmp respectively, or whatever temporary
2705 directory has been set for the calling user. %E will expand to either
2706 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
2708 * The ExecStart= lines of unit files are no longer required to
2709 reference absolute paths. If non-absolute paths are specified the
2710 specified binary name is searched within the service manager's
2711 built-in $PATH, which may be queried with 'systemd-path
2712 search-binaries-default'. It's generally recommended to continue to
2713 use absolute paths for all binaries specified in unit files.
2715 * Units gained a new load state "bad-setting", which is used when a
2716 unit file was loaded, but contained fatal errors which prevent it
2717 from being started (for example, a service unit has been defined
2718 lacking both ExecStart= and ExecStop= lines).
2720 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
2721 support alternative debuggers, for example lldb. The old name
2722 continues to be available however, for compatibility reasons. Use the
2723 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
2724 to pick an alternative debugger instead of the default gdb.
2726 * systemctl and the other tools will now output escape sequences that
2727 generate proper clickable hyperlinks in various terminal emulators
2728 where useful (for example, in the "systemctl status" output you can
2729 now click on the unit file name to quickly open it in the
2730 editor/viewer of your choice). Note that not all terminal emulators
2731 support this functionality yet, but many do. Unfortunately, the
2732 "less" pager doesn't support this yet, hence this functionality is
2733 currently automatically turned off when a pager is started (which
2734 happens quite often due to auto-paging). We hope to remove this
2735 limitation as soon as "less" learns these escape sequences. This new
2736 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
2737 environment variable. For details on these escape sequences see:
2738 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
2740 * networkd's .network files now support a new IPv6MTUBytes= option for
2741 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
2742 option in the [Route] section to configure the MTU to use for
2743 specific routes. It also gained support for configuration of the DHCP
2744 "UserClass" option through the new UserClass= setting. It gained
2745 three new options in the new [CAN] section for configuring CAN
2746 networks. The MULTICAST and ALLMULTI interface flags may now be
2747 controlled explicitly with the new Multicast= and AllMulticast=
2750 * networkd will now automatically make use of the kernel's route
2751 expiration feature, if it is available.
2753 * udevd's .link files now support setting the number of receive and
2754 transmit channels, using the RxChannels=, TxChannels=,
2755 OtherChannels=, CombinedChannels= settings.
2757 * Support for UDPSegmentationOffload= has been removed, given its
2758 limited support in hardware, and waning software support.
2760 * networkd's .netdev files now support creating "netdevsim" interfaces.
2762 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
2763 to query the unit belonging to a specific kernel control group.
2765 * systemd-analyze gained a new verb "cat-config", which may be used to
2766 dump the contents of any configuration file, with all its matching
2767 drop-in files added in, and honouring the usual search and masking
2768 logic applied to systemd configuration files. For example use
2769 "systemd-analyze cat-config systemd/system.conf" to get the complete
2770 system configuration file of systemd how it would be loaded by PID 1
2771 itself. Similar to this, various tools such as systemd-tmpfiles or
2772 systemd-sysusers, gained a new option "--cat-config", which does the
2773 corresponding operation for their own configuration settings. For
2774 example, "systemd-tmpfiles --cat-config" will now output the full
2775 list of tmpfiles.d/ lines in place.
2777 * timedatectl gained three new verbs: "show" shows bus properties of
2778 systemd-timedated, "timesync-status" shows the current NTP
2779 synchronization state of systemd-timesyncd, and "show-timesync"
2780 shows bus properties of systemd-timesyncd.
2782 * systemd-timesyncd gained a bus interface on which it exposes details
2785 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
2786 understood by systemd-timedated. It takes a colon-separated list of
2787 unit names of NTP client services. The list is used by
2788 "timedatectl set-ntp".
2790 * systemd-nspawn gained a new --rlimit= switch for setting initial
2791 resource limits for the container payload. There's a new switch
2792 --hostname= to explicitly override the container's hostname. A new
2793 --no-new-privileges= switch may be used to control the
2794 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
2795 --oom-score-adjust= switch controls the OOM scoring adjustment value
2796 for the payload. The new --cpu-affinity= switch controls the CPU
2797 affinity of the container payload. The new --resolv-conf= switch
2798 allows more detailed control of /etc/resolv.conf handling of the
2799 container. Similarly, the new --timezone= switch allows more detailed
2800 control of /etc/localtime handling of the container.
2802 * systemd-detect-virt gained a new --list switch, which will print a
2803 list of all currently known VM and container environments.
2805 * Support for "Portable Services" has been added, see
2806 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
2807 experimental, but this is expected to change soon. Reflecting this
2808 experimental state, the "portablectl" binary is not installed into
2809 /usr/bin yet. The binary has to be called with the full path
2810 /usr/lib/systemd/portablectl instead.
2812 * journalctl's and systemctl's -o switch now knows a new log output
2813 mode "with-unit". The output it generates is very similar to the
2814 regular "short" mode, but displays the unit name instead of the
2815 syslog tag for each log line. Also, the date is shown with timezone
2816 information. This mode is probably more useful than the classic
2817 "short" output mode for most purposes, except where pixel-perfect
2818 compatibility with classic /var/log/messages formatting is required.
2820 * A new --dump-bus-properties switch has been added to the systemd
2821 binary, which may be used to dump all supported D-Bus properties.
2822 (Options which are still supported, but are deprecated, are *not*
2825 * sd-bus gained a set of new calls:
2826 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
2827 enable/disable the "floating" state of a bus slot object,
2828 i.e. whether the slot object pins the bus it is allocated for into
2829 memory or if the bus slot object gets disconnected when the bus goes
2830 away. sd_bus_open_with_description(),
2831 sd_bus_open_user_with_description(),
2832 sd_bus_open_system_with_description() may be used to allocate bus
2833 objects and set their description string already during allocation.
2835 * sd-event gained support for watching inotify events from the event
2836 loop, in an efficient way, sharing inotify handles between multiple
2837 users. For this a new function sd_event_add_inotify() has been added.
2839 * sd-event and sd-bus gained support for calling special user-supplied
2840 destructor functions for userdata pointers associated with
2841 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
2842 functions sd_bus_slot_set_destroy_callback,
2843 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
2844 sd_bus_track_get_destroy_callback,
2845 sd_event_source_set_destroy_callback,
2846 sd_event_source_get_destroy_callback have been added.
2848 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
2850 * PID 1 will now automatically reschedule .timer units whenever the
2851 local timezone changes. (They previously got rescheduled
2852 automatically when the system clock changed.)
2854 * New documentation has been added to document cgroups delegation,
2855 portable services and the various code quality tools we have set up:
2857 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
2858 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
2859 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
2861 * The Boot Loader Specification has been added to the source tree.
2863 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
2865 While moving it into our source tree we have updated it and further
2866 changes are now accepted through the usual github PR workflow.
2868 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
2869 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
2870 earlier PAM modules. The data in these fields is used to initialize
2871 the session scope's resource properties. Thus external PAM modules
2872 may now configure per-session limits, for example sourced from
2873 external user databases.
2875 * socket units with Accept=yes will now maintain a "refused" counter in
2876 addition to the existing "accepted" counter, counting connections
2877 refused due to the enforced limits.
2879 * The "systemd-path search-binaries-default" command may now be use to
2880 query the default, built-in $PATH PID 1 will pass to the services it
2883 * A new unit file setting PrivateMounts= has been added. It's a boolean
2884 option. If enabled the unit's processes are invoked in their own file
2885 system namespace. Note that this behaviour is also implied if any
2886 other file system namespacing options (such as PrivateTmp=,
2887 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
2888 primarily useful for services that do not use any of the other file
2889 system namespacing options. One such service is systemd-udevd.service
2890 where this is now used by default.
2892 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
2893 when the system is booted in UEFI "secure mode".
2895 * A new unit "system-update-pre.target" is added, which defines an
2896 optional synchronization point for offline system updates, as
2897 implemented by the pre-existing "system-update.target" unit. It
2898 allows ordering services before the service that executes the actual
2899 update process in a generic way.
2901 * Systemd now emits warnings whenever .include syntax is used.
2903 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
2904 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
2905 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
2906 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
2907 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
2908 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
2909 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
2910 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
2911 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
2912 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
2913 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
2914 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
2915 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
2916 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
2917 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
2918 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
2919 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
2920 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
2921 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
2922 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
2923 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
2924 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
2925 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
2926 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
2927 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
2928 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
2929 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
2930 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
2931 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2933 — Berlin, 2018-06-22
2937 * The MemoryAccounting= unit property now defaults to on. After
2938 discussions with the upstream control group maintainers we learnt
2939 that the negative impact of cgroup memory accounting on current
2940 kernels is finally relatively minimal, so that it should be safe to
2941 enable this by default without affecting system performance. Besides
2942 memory accounting only task accounting is turned on by default, all
2943 other forms of resource accounting (CPU, IO, IP) remain off for now,
2944 because it's not clear yet that their impact is small enough to move
2945 from opt-in to opt-out. We recommend downstreams to leave memory
2946 accounting on by default if kernel 4.14 or higher is primarily
2947 used. On very resource constrained systems or when support for old
2948 kernels is a necessity, -Dmemory-accounting-default=false can be used
2949 to revert this change.
2951 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
2952 %udev_rules_update) and the journal catalog (%journal_catalog_update)
2953 from the upgrade scriptlets of individual packages now do nothing.
2954 Transfiletriggers have been added which will perform those updates
2955 once at the end of the transaction.
2957 Similar transfiletriggers have been added to execute any sysctl.d
2958 and binfmt.d rules. Thus, it should be unnecessary to provide any
2959 scriptlets to execute this configuration from package installation
2962 * systemd-sysusers gained a mode where the configuration to execute is
2963 specified on the command line, but this configuration is not executed
2964 directly, but instead it is merged with the configuration on disk,
2965 and the result is executed. This is useful for package installation
2966 scripts which want to create the user before installing any files on
2967 disk (in case some of those files are owned by that user), while
2968 still allowing local admin overrides.
2970 This functionality is exposed to rpm scriptlets through a new
2971 %sysusers_create_package macro. Old %sysusers_create and
2972 %sysusers_create_inline macros are deprecated.
2974 A transfiletrigger for sysusers.d configuration is now installed,
2975 which means that it should be unnecessary to call systemd-sysusers from
2976 package installation scripts, unless the package installs any files
2977 owned by those newly-created users, in which case
2978 %sysusers_create_package should be used.
2980 * Analogous change has been done for systemd-tmpfiles: it gained a mode
2981 where the command-line configuration is merged with the configuration
2982 on disk. This is exposed as the new %tmpfiles_create_package macro,
2983 and %tmpfiles_create is deprecated. A transfiletrigger is installed
2984 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
2985 from package installation scripts.
2987 * sysusers.d configuration for a user may now also specify the group
2988 number, in addition to the user number ("u username 123:456"), or
2989 without the user number ("u username -:456").
2991 * Configution items for systemd-sysusers can now be specified as
2992 positional arguments when the new --inline switch is used.
2994 * The login shell of users created through sysusers.d may now be
2995 specified (previously, it was always /bin/sh for root and
2996 /sbin/nologin for other users).
2998 * systemd-analyze gained a new --global switch to look at global user
2999 configuration. It also gained a unit-paths verb to list the unit load
3000 paths that are compiled into systemd (which can be used with
3001 --systemd, --user, or --global).
3003 * udevadm trigger gained a new --settle/-w option to wait for any
3004 triggered events to finish (but just those, and not any other events
3005 which are triggered meanwhile).
3007 * The action that systemd-logind takes when the lid is closed and the
3008 machine is connected to external power can now be configured using
3009 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
3010 was determined by HandleLidSwitch=, and, for backwards compatibility,
3011 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
3013 * journalctl will periodically call sd_journal_process() to make it
3014 resilient against inotify queue overruns when journal files are
3015 rotated very quickly.
3017 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
3018 sd_bus_get_n_queued_write — may be used to check the number of
3019 pending bus messages.
3021 * systemd gained a new
3022 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
3023 which can be used to migrate foreign processes to scope and service
3024 units. The primary user for this new API is systemd itself: the
3025 systemd --user instance uses this call of the systemd --system
3026 instance to migrate processes if it itself gets the request to
3027 migrate processes and the kernel refuses this due to access
3028 restrictions. Thanks to this "systemd-run --scope --user …" works
3029 again in pure cgroup v2 environments when invoked from the user
3032 * A new TemporaryFileSystem= setting can be used to mask out part of
3033 the real file system tree with tmpfs mounts. This may be combined
3034 with BindPaths= and BindReadOnlyPaths= to hide files or directories
3035 not relevant to the unit, while still allowing some paths lower in
3036 the tree to be accessed.
3038 ProtectHome=tmpfs may now be used to hide user home and runtime
3039 directories from units, in a way that is mostly equivalent to
3040 "TemporaryFileSystem=/home /run/user /root".
3042 * Non-service units are now started with KeyringMode=shared by default.
3043 This means that mount and swapon and other mount tools have access
3044 to keys in the main keyring.
3046 * /sys/fs/bpf is now mounted automatically.
3048 * QNX virtualization is now detected by systemd-detect-virt and may
3049 be used in ConditionVirtualization=.
3051 * IPAccounting= may now be enabled also for slice units.
3053 * A new -Dsplit-bin= build configuration switch may be used to specify
3054 whether bin and sbin directories are merged, or if they should be
3055 included separately in $PATH and various listings of executable
3056 directories. The build configuration scripts will try to autodetect
3057 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
3058 system, but distributions are encouraged to configure this
3061 * A new -Dok-color= build configuration switch may be used to change
3062 the colour of "OK" status messages.
3064 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
3065 PrivateNetwork=yes was buggy in previous versions of systemd. This
3066 means that after the upgrade and daemon-reexec, any such units must
3069 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
3070 will not exclude read-only files owned by root from cleanup.
3072 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
3073 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
3074 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
3075 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
3076 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
3077 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
3078 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
3079 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
3080 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
3081 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
3082 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
3083 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
3084 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
3085 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
3086 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
3087 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
3089 — Warsaw, 2018-03-05
3093 * Some keyboards come with a zoom see-saw or rocker which until now got
3094 mapped to the Linux "zoomin/out" keys in hwdb. However, these
3095 keycodes are not recognized by any major desktop. They now produce
3096 Up/Down key events so that they can be used for scrolling.
3098 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
3099 slightly: previously, if an argument was specified for lines of this
3100 type (i.e. the right-most column was set) this string was appended to
3101 existing files each time systemd-tmpfiles was run. This behaviour was
3102 different from what the documentation said, and not particularly
3103 useful, as repeated systemd-tmpfiles invocations would not be
3104 idempotent and grow such files without bounds. With this release
3105 behaviour has been altered to match what the documentation says:
3106 lines of this type only have an effect if the indicated files don't
3107 exist yet, and only then the argument string is written to the file.
3109 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
3110 systemd-tmpfiles behaviour: previously, read-only files owned by root
3111 were always excluded from the file "aging" algorithm (i.e. the
3112 automatic clean-up of directories like /tmp based on
3113 atime/mtime/ctime). We intend to drop this restriction, and age files
3114 by default even when owned by root and read-only. This behaviour was
3115 inherited from older tools, but there have been requests to remove
3116 it, and it's not obvious why this restriction was made in the first
3117 place. Please speak up now, if you are aware of software that reqires
3118 this behaviour, otherwise we'll remove the restriction in v238.
3120 * A new environment variable $SYSTEMD_OFFLINE is now understood by
3121 systemctl. It takes a boolean argument. If on, systemctl assumes it
3122 operates on an "offline" OS tree, and will not attempt to talk to the
3123 service manager. Previously, this mode was implicitly enabled if a
3124 chroot() environment was detected, and this new environment variable
3125 now provides explicit control.
3127 * .path and .socket units may now be created transiently, too.
3128 Previously only service, mount, automount and timer units were
3129 supported as transient units. The systemd-run tool has been updated
3130 to expose this new functionality, you may hence use it now to bind
3131 arbitrary commands to path or socket activation on-the-fly from the
3132 command line. Moreover, almost all properties are now exposed for the
3133 unit types that already supported transient operation.
3135 * The systemd-mount command gained support for a new --owner= parameter
3136 which takes a user name, which is then resolved and included in uid=
3137 and gid= mount options string of the file system to mount.
3139 * A new unit condition ConditionControlGroupController= has been added
3140 that checks whether a specific cgroup controller is available.
3142 * Unit files, udev's .link files, and systemd-networkd's .netdev and
3143 .network files all gained support for a new condition
3144 ConditionKernelVersion= for checking against specific kernel
3147 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
3148 support for configuring device flags in the Flags= setting. In the
3149 same files, the [Tunnel] section gained support for configuring
3150 AllowLocalRemote=. The [Route] section in .network files gained
3151 support for configuring InitialCongestionWindow=,
3152 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
3153 understands RapidCommit=.
3155 * systemd-networkd's DHCPv6 support gained support for Prefix
3158 * sd-bus gained support for a new "watch-bind" feature. When this
3159 feature is enabled, an sd_bus connection may be set up to connect to
3160 an AF_UNIX socket in the file system as soon as it is created. This
3161 functionality is useful for writing early-boot services that
3162 automatically connect to the system bus as soon as it is started,
3163 without ugly time-based polling. systemd-networkd and
3164 systemd-resolved have been updated to make use of this
3165 functionality. busctl exposes this functionality in a new
3166 --watch-bind= command line switch.
3168 * sd-bus will now optionally synthesize a local "Connected" signal as
3169 soon as a D-Bus connection is set up fully. This message mirrors the
3170 already existing "Disconnected" signal which is synthesized when the
3171 connection is terminated. This signal is generally useful but
3172 particularly handy in combination with the "watch-bind" feature
3173 described above. Synthesizing of this message has to be requested
3174 explicitly through the new API call sd_bus_set_connected_signal(). In
3175 addition a new call sd_bus_is_ready() has been added that checks
3176 whether a connection is fully set up (i.e. between the "Connected" and
3177 "Disconnected" signals).
3179 * sd-bus gained two new calls sd_bus_request_name_async() and
3180 sd_bus_release_name_async() for asynchronously registering bus
3181 names. Similar, there is now sd_bus_add_match_async() for installing
3182 a signal match asynchronously. All of systemd's own services have
3183 been updated to make use of these calls. Doing these operations
3184 asynchronously has two benefits: it reduces the risk of deadlocks in
3185 case of cyclic dependencies between bus services, and it speeds up
3186 service initialization since synchronization points for bus
3187 round-trips are removed.
3189 * sd-bus gained two new calls sd_bus_match_signal() and
3190 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
3191 and sd_bus_add_match_async() but instead of taking a D-Bus match
3192 string take match fields as normal function parameters.
3194 * sd-bus gained two new calls sd_bus_set_sender() and
3195 sd_bus_message_set_sender() for setting the sender name of outgoing
3196 messages (either for all outgoing messages or for just one specific
3197 one). These calls are only useful in direct connections as on
3198 brokered connections the broker fills in the sender anyway,
3199 overwriting whatever the client filled in.
3201 * sd-event gained a new pseudo-handle that may be specified on all API
3202 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
3203 used this refers to the default event loop object of the calling
3204 thread. Note however that this does not implicitly allocate one —
3205 which has to be done prior by using sd_event_default(). Similarly
3206 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
3207 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
3208 to the default bus of the specified type of the calling thread. Here
3209 too this does not implicitly allocate bus connection objects, this
3210 has to be done prior with sd_bus_default() and friends.
3212 * sd-event gained a new call pair
3213 sd_event_source_{get|set}_io_fd_own(). This may be used to request
3214 automatic closure of the file descriptor an IO event source watches
3215 when the event source is destroyed.
3217 * systemd-networkd gained support for natively configuring WireGuard
3220 * In previous versions systemd synthesized user records both for the
3221 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
3222 internally. In order to simplify distribution-wide renames of the
3223 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
3224 new transitional flag file has been added: if
3225 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
3226 user and group record within the systemd codebase is disabled.
3228 * systemd-notify gained a new --uid= option for selecting the source
3229 user/UID to use for notification messages sent to the service
3232 * journalctl gained a new --grep= option to list only entries in which
3233 the message matches a certain pattern. By default matching is case
3234 insensitive if the pattern is lowercase, and case sensitive
3235 otherwise. Option --case-sensitive=yes|no can be used to override
3236 this an specify case sensitivity or case insensitivity.
3238 * There's now a "systemd-analyze service-watchdogs" command for printing
3239 the current state of the service runtime watchdog, and optionally
3240 enabling or disabling the per-service watchdogs system-wide if given a
3241 boolean argument (i.e. the concept you configure in WatchdogSec=), for
3242 debugging purposes. There's also a kernel command line option
3243 systemd.service_watchdogs= for controlling the same.
3245 * Two new "log-level" and "log-target" options for systemd-analyze were
3246 added that merge the now deprecated get-log-level, set-log-level and
3247 get-log-target, set-log-target pairs. The deprecated options are still
3248 understood for backwards compatibility. The two new options print the
3249 current value when no arguments are given, and set them when a
3250 level/target is given as an argument.
3252 * sysusers.d's "u" lines now optionally accept both a UID and a GID
3253 specification, separated by a ":" character, in order to create users
3254 where UID and GID do not match.
3256 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
3257 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
3258 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
3259 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
3260 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
3261 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
3262 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
3263 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
3264 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
3265 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
3266 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
3267 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
3268 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
3269 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
3270 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
3271 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
3272 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
3273 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
3274 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
3275 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
3282 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
3283 in v235 has been extended to also set the dummy.ko module option
3284 numdummies=0, preventing the kernel from automatically creating
3285 dummy0. All dummy interfaces must now be explicitly created.
3287 * Unknown '%' specifiers in configuration files are now rejected. This
3288 applies to units and tmpfiles.d configuration. Any percent characters
3289 that are followed by a letter or digit that are not supposed to be
3290 interpreted as the beginning of a specifier should be escaped by
3291 doubling ("%%"). (So "size=5%" is still accepted, as well as
3292 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
3293 valid specifiers today.)
3295 * systemd-resolved now maintains a new dynamic
3296 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
3297 recommended to make /etc/resolv.conf a symlink to it. This file
3298 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
3299 includes dynamically acquired search domains, achieving more correct
3300 DNS resolution by software that bypasses local DNS APIs such as NSS.
3302 * The "uaccess" udev tag has been dropped from /dev/kvm and
3303 /dev/dri/renderD*. These devices now have the 0666 permissions by
3304 default (but this may be changed at build-time). /dev/dri/renderD*
3305 will now be owned by the "render" group along with /dev/kfd.
3307 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
3308 systemd-journal-gatewayd.service and
3309 systemd-journal-upload.service. This means "nss-systemd" must be
3310 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
3311 services are resolved properly.
3313 * In /etc/fstab two new mount options are now understood:
3314 x-systemd.makefs and x-systemd.growfs. The former has the effect that
3315 the configured file system is formatted before it is mounted, the
3316 latter that the file system is resized to the full block device size
3317 after it is mounted (i.e. if the file system is smaller than the
3318 partition it resides on, it's grown). This is similar to the fsck
3319 logic in /etc/fstab, and pulls in systemd-makefs@.service and
3320 systemd-growfs@.service as necessary, similar to
3321 systemd-fsck@.service. Resizing is currently only supported on ext4
3324 * In systemd-networkd, the IPv6 RA logic now optionally may announce
3325 DNS server and domain information.
3327 * Support for the LUKS2 on-disk format for encrypted partitions has
3328 been added. This requires libcryptsetup2 during compilation and
3331 * The systemd --user instance will now signal "readiness" when its
3332 basic.target unit has been reached, instead of when the run queue ran
3333 empty for the first time.
3335 * Tmpfiles.d with user configuration are now also supported.
3336 systemd-tmpfiles gained a new --user switch, and snippets placed in
3337 ~/.config/user-tmpfiles.d/ and corresponding directories will be
3338 executed by systemd-tmpfiles --user running in the new
3339 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
3340 running in the user session.
3342 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
3343 %S resolves to the top-level state directory (/var/lib for the system
3344 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
3345 top-level cache directory (/var/cache for the system instance,
3346 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
3347 logs directory (/var/log for the system instance,
3348 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
3349 existing %t specifier, that resolves to the top-level runtime
3350 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
3353 * journalctl learnt a new parameter --output-fields= for limiting the
3354 set of journal fields to output in verbose and JSON output modes.
3356 * systemd-timesyncd's configuration file gained a new option
3357 RootDistanceMaxSec= for setting the maximum root distance of servers
3358 it'll use, as well as the new options PollIntervalMinSec= and
3359 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
3361 * bootctl gained a new command "list" for listing all available boot
3362 menu items on systems that follow the boot loader specification.
3364 * systemctl gained a new --dry-run switch that shows what would be done
3365 instead of doing it, and is currently supported by the shutdown and
3368 * ConditionSecurity= can now detect the TOMOYO security module.
3370 * Unit file [Install] sections are now also respected in unit drop-in
3371 files. This is intended to be used by drop-ins under /usr/lib/.
3373 * systemd-firstboot may now also set the initial keyboard mapping.
3375 * Udev "changed" events for devices which are exposed as systemd
3376 .device units are now propagated to units specified in
3377 ReloadPropagatedFrom= as reload requests.
3379 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
3380 unit template name (i.e. a name in the form of 'foobar@.service',
3381 without the instance component between the '@' and - the '.'), then
3382 the escaped sysfs path of the device is automatically used as the
3385 * SystemCallFilter= in unit files has been extended so that an "errno"
3386 can be specified individually for each system call. Example:
3387 SystemCallFilter=~uname:EILSEQ.
3389 * The cgroup delegation logic has been substantially updated. Delegate=
3390 now optionally takes a list of controllers (instead of a boolean, as
3391 before), which lists the controllers to delegate at least.
3393 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
3395 * A new LogLevelMax= setting configures the maximum log level any
3396 process of the service may log at (i.e. anything with a lesser
3397 priority than what is specified is automatically dropped). A new
3398 LogExtraFields= setting allows configuration of additional journal
3399 fields to attach to all log records generated by any of the unit's
3402 * New StandardInputData= and StandardInputText= settings along with the
3403 new option StandardInput=data may be used to configure textual or
3404 binary data that shall be passed to the executed service process via
3405 standard input, encoded in-line in the unit file.
3407 * StandardInput=, StandardOutput= and StandardError= may now be used to
3408 connect stdin/stdout/stderr of executed processes directly with a
3409 file or AF_UNIX socket in the file system, using the new "file:" option.
3411 * A new unit file option CollectMode= has been added, that allows
3412 tweaking the garbage collection logic for units. It may be used to
3413 tell systemd to garbage collect units that have failed automatically
3414 (normally it only GCs units that exited successfully). systemd-run
3415 and systemd-mount expose this new functionality with a new -G option.
3417 * "machinectl bind" may now be used to bind mount non-directories
3418 (i.e. regularfiles, devices, fifos, sockets).
3420 * systemd-analyze gained a new verb "calendar" for validating and
3421 testing calendar time specifications to use for OnCalendar= in timer
3422 units. Besides validating the expression it will calculate the next
3423 time the specified expression would elapse.
3425 * In addition to the pre-existing FailureAction= unit file setting
3426 there's now SuccessAction=, for configuring a shutdown action to
3427 execute when a unit completes successfully. This is useful in
3428 particular inside containers that shall terminate after some workload
3429 has been completed. Also, both options are now supported for all unit
3430 types, not just services.
3432 * networkds's IP rule support gained two new options
3433 IncomingInterface= and OutgoingInterface= for configuring the incoming
3434 and outgoing interfaces of configured rules. systemd-networkd also
3435 gained support for "vxcan" network devices.
3437 * networkd gained a new setting RequiredForOnline=, taking a
3438 boolean. If set, systemd-wait-online will take it into consideration
3439 when determining that the system is up, otherwise it will ignore the
3440 interface for this purpose.
3442 * The sd_notify() protocol gained support for a new operation: with
3443 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
3444 store again, ahead of POLLHUP or POLLERR when they are removed
3447 * A new document doc/UIDS-GIDS.md has been added to the source tree,
3448 that documents the UID/GID range and assignment assumptions and
3449 requirements of systemd.
3451 * The watchdog device PID 1 will ping may now be configured through the
3452 WatchdogDevice= configuration file setting, or by setting the
3453 systemd.watchdog_service= kernel commandline option.
3455 * systemd-resolved's gained support for registering DNS-SD services on
3456 the local network using MulticastDNS. Services may either be
3457 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
3458 the same dir below /run, /usr/lib), or through its D-Bus API.
3460 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
3461 extend the effective start, runtime, and stop time. The service must
3462 continue to send EXTEND_TIMEOUT_USEC within the period specified to
3463 prevent the service manager from making the service as timedout.
3465 * systemd-resolved's DNSSEC support gained support for RFC 8080
3466 (Ed25519 keys and signatures).
3468 * The systemd-resolve command line tool gained a new set of options
3469 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
3470 --set-nta= and --revert to configure per-interface DNS configuration
3471 dynamically during runtime. It's useful for pushing DNS information
3472 into systemd-resolved from DNS hook scripts that various interface
3473 managing software supports (such as pppd).
3475 * systemd-nspawn gained a new --network-namespace-path= command line
3476 option, which may be used to make a container join an existing
3477 network namespace, by specifying a path to a "netns" file.
3479 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
3480 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
3481 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
3482 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
3483 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
3484 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
3485 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
3486 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
3487 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
3488 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
3489 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
3490 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
3491 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
3492 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
3493 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
3494 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
3495 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
3496 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
3497 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
3498 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
3499 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
3500 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
3501 Jędrzejewski-Szmek, Zeal Jagannatha
3503 — Berlin, 2017-12-14
3507 * INCOMPATIBILITY: systemd-logind.service and other long-running
3508 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
3509 communication with the outside. This generally improves security of
3510 the system, and is in almost all cases a safe and good choice, as
3511 these services do not and should not provide any network-facing
3512 functionality. However, systemd-logind uses the glibc NSS API to
3513 query the user database. This creates problems on systems where NSS
3514 is set up to directly consult network services for user database
3515 lookups. In particular, this creates incompatibilities with the
3516 "nss-nis" module, which attempts to directly contact the NIS/YP
3517 network servers it is configured for, and will now consistently
3518 fail. In such cases, it is possible to turn off IP sandboxing for
3519 systemd-logind.service (set IPAddressDeny= in its [Service] section
3520 to the empty string, via a .d/ unit file drop-in). Downstream
3521 distributions might want to update their nss-nis packaging to include
3522 such a drop-in snippet, accordingly, to hide this incompatibility
3523 from the user. Another option is to make use of glibc's nscd service
3524 to proxy such network requests through a privilege-separated, minimal
3525 local caching daemon, or to switch to more modern technologies such
3526 sssd, whose NSS hook-ups generally do not involve direct network
3527 access. In general, we think it's definitely time to question the
3528 implementation choices of nss-nis, i.e. whether it's a good idea
3529 today to embed a network-facing loadable module into all local
3530 processes that need to query the user database, including the most
3531 trivial and benign ones, such as "ls". For more details about
3532 IPAddressDeny= see below.
3534 * A new modprobe.d drop-in is now shipped by default that sets the
3535 bonding module option max_bonds=0. This overrides the kernel default,
3536 to avoid conflicts and ambiguity as to whether or not bond0 should be
3537 managed by systemd-networkd or not. This resolves multiple issues
3538 with bond0 properties not being applied, when bond0 is configured
3539 with systemd-networkd. Distributors may choose to not package this,
3540 however in that case users will be prevented from correctly managing
3541 bond0 interface using systemd-networkd.
3543 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
3544 which print the logging level and target of the system manager. They
3545 complement the existing "set-log-level" and "set-log-target" verbs
3546 used to change those values.
3548 * journald.conf gained a new boolean setting ReadKMsg= which defaults
3549 to on. If turned off kernel log messages will not be read by
3550 systemd-journald or included in the logs. It also gained a new
3551 setting LineMax= for configuring the maximum line length in
3552 STDOUT/STDERR log streams. The new default for this value is 48K, up
3553 from the previous hardcoded 2048.
3555 * A new unit setting RuntimeDirectoryPreserve= has been added, which
3556 allows more detailed control of what to do with a runtime directory
3557 configured with RuntimeDirectory= (i.e. a directory below /run or
3558 $XDG_RUNTIME_DIR) after a unit is stopped.
3560 * The RuntimeDirectory= setting for units gained support for creating
3561 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
3562 one top-level directory.
3564 * Units gained new options StateDirectory=, CacheDirectory=,
3565 LogsDirectory= and ConfigurationDirectory= which are closely related
3566 to RuntimeDirectory= but manage per-service directories below
3567 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
3568 possible to write unit files which when activated automatically gain
3569 properly owned service specific directories in these locations, thus
3570 making unit files self-contained and increasing compatibility with
3571 stateless systems and factory reset where /etc or /var are
3572 unpopulated at boot. Matching these new settings there's also
3573 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
3574 ConfigurationDirectoryMode= for configuring the access mode of these
3575 directories. These settings are particularly useful in combination
3576 with DynamicUser=yes as they provide secure, properly-owned,
3577 writable, and stateful locations for storage, excluded from the
3578 sandbox that such services live in otherwise.
3580 * Automake support has been removed from this release. systemd is now
3583 * systemd-journald will now aggressively cache client metadata during
3584 runtime, speeding up log write performance under pressure. This comes
3585 at a small price though: as much of the metadata is read
3586 asynchronously from /proc/ (and isn't implicitly attached to log
3587 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
3588 metadata stored alongside a log entry might be slightly
3589 out-of-date. Previously it could only be slightly newer than the log
3590 message. The time window is small however, and given that the kernel
3591 is unlikely to be improved anytime soon in this regard, this appears
3594 * nss-myhostname/systemd-resolved will now by default synthesize an
3595 A/AAAA resource record for the "_gateway" hostname, pointing to the
3596 current default IP gateway. Previously it did that for the "gateway"
3597 name, hampering adoption, as some distributions wanted to leave that
3598 hostname open for local use. The old behaviour may still be
3599 requested at build time.
3601 * systemd-networkd's [Address] section in .network files gained a new
3602 Scope= setting for configuring the IP address scope. The [Network]
3603 section gained a new boolean setting ConfigureWithoutCarrier= that
3604 tells systemd-networkd to ignore link sensing when configuring the
3605 device. The [DHCP] section gained a new Anonymize= boolean option for
3606 turning on a number of options suggested in RFC 7844. A new
3607 [RoutingPolicyRule] section has been added for configuring the IP
3608 routing policy. The [Route] section has gained support for a new
3609 Type= setting which permits configuring
3610 blackhole/unreachable/prohibit routes.
3612 * The [VRF] section in .netdev files gained a new Table= setting for
3613 configuring the routing table to use. The [Tunnel] section gained a
3614 new Independent= boolean field for configuring tunnels independent of
3615 an underlying network interface. The [Bridge] section gained a new
3616 GroupForwardMask= option for configuration of propagation of link
3617 local frames between bridge ports.
3619 * The WakeOnLan= setting in .link files gained support for a number of
3620 new modes. A new TCP6SegmentationOffload= setting has been added for
3621 configuring TCP/IPv6 hardware segmentation offload.
3623 * The IPv6 RA sender implementation may now optionally send out RDNSS
3624 and RDNSSL records to supply DNS configuration to peers.
3626 * systemd-nspawn gained support for a new --system-call-filter= command
3627 line option for adding and removing entries in the default system
3628 call filter it applies. Moreover systemd-nspawn has been changed to
3629 implement a system call allow list instead of a deny list.
3631 * systemd-run gained support for a new --pipe command line option. If
3632 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
3633 are directly passed on to the activated transient service
3634 executable. This allows invoking arbitrary processes as systemd
3635 services (for example to take benefit of dependency management,
3636 accounting management, resource management or log management that is
3637 done automatically for services) — while still allowing them to be
3638 integrated in a classic UNIX shell pipeline.
3640 * When a service sends RELOAD=1 via sd_notify() and reload propagation
3641 using ReloadPropagationTo= is configured, a reload is now propagated
3642 to configured units. (Previously this was only done on explicitly
3643 requested reloads, using "systemctl reload" or an equivalent
3646 * For each service unit a restart counter is now kept: it is increased
3647 each time the service is restarted due to Restart=, and may be
3648 queried using "systemctl show -p NRestarts …".
3650 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
3651 @signal and @timer have been added, for usage with SystemCallFilter=
3652 in unit files and the new --system-call-filter= command line option
3653 of systemd-nspawn (see above).
3655 * ExecStart= lines in unit files gained two new modifiers: when a
3656 command line is prefixed with "!" the command will be executed as
3657 configured, except for the credentials applied by
3658 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
3659 "+", but does still apply namespacing options unlike "+". There's
3660 also "!!" now, which is mostly identical, but becomes a NOP on
3661 systems that support ambient capabilities. This is useful to write
3662 unit files that work with ambient capabilities where possible but
3663 automatically fall back to traditional privilege dropping mechanisms
3664 on systems where this is not supported.
3666 * ListenNetlink= settings in socket units now support RDMA netlink
3669 * A new unit file setting LockPersonality= has been added which permits
3670 locking down the chosen execution domain ("personality") of a service
3673 * A new special target "getty-pre.target" has been added, which is
3674 ordered before all text logins, and may be used to order services
3675 before textual logins acquire access to the console.
3677 * systemd will now attempt to load the virtio-rng.ko kernel module very
3678 early on if a VM environment supporting this is detected. This should
3679 improve entropy during early boot in virtualized environments.
3681 * A _netdev option is now supported in /etc/crypttab that operates in a
3682 similar way as the same option in /etc/fstab: it permits configuring
3683 encrypted devices that need to be ordered after the network is up.
3684 Following this logic, two new special targets
3685 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
3686 added that are to cryptsetup.target what remote-fs.target and
3687 remote-fs-pre.target are to local-fs.target.
3689 * Service units gained a new UnsetEnvironment= setting which permits
3690 unsetting specific environment variables for services that are
3691 normally passed to it (for example in order to mask out locale
3692 settings for specific services that can't deal with it).
3694 * Units acquired a new boolean option IPAccounting=. When turned on, IP
3695 traffic accounting (packet count as well as byte count) is done for
3696 the service, and shown as part of "systemctl status" or "systemd-run
3699 * Service units acquired two new options IPAddressAllow= and
3700 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
3701 for configuring a simple IP access control list for all sockets of
3702 the unit. These options are available also on .slice and .socket
3703 units, permitting flexible access list configuration for individual
3704 services as well as groups of services (as defined by a slice unit),
3705 including system-wide. Note that IP ACLs configured this way are
3706 enforced on every single IPv4 and IPv6 socket created by any process
3707 of the service unit, and apply to ingress as well as egress traffic.
3709 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
3710 structured log message is generated each time the unit is stopped,
3711 containing information about the consumed resources of this
3714 * A new setting KeyringMode= has been added to unit files, which may be
3715 used to control how the kernel keyring is set up for executed
3718 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
3719 "systemctl kexec" and "systemctl exit" are now always asynchronous in
3720 behaviour (that is: these commands return immediately after the
3721 operation was enqueued instead of waiting for the operation to
3722 complete). Previously, "systemctl poweroff" and "systemctl reboot"
3723 were asynchronous on systems using systemd-logind (i.e. almost
3724 always, and like they were on sysvinit), and the other three commands
3725 were unconditionally synchronous. With this release this is cleaned
3726 up, and callers will see the same asynchronous behaviour on all
3727 systems for all five operations.
3729 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
3732 * .timer units now accept calendar specifications in other timezones
3733 than UTC or the local timezone.
3735 * The tmpfiles snippet var.conf has been changed to create
3736 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
3737 the "utmp" group already, and it appears to be generally understood
3738 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
3739 databases. Previously this was implemented correctly for all these
3740 databases excepts btmp, which has been opened up like this now
3741 too. Note that while the other databases are world-readable
3742 (i.e. 0644), btmp is not and remains more restrictive.
3744 * The systemd-resolve tool gained a new --reset-server-features
3745 switch. When invoked like this systemd-resolved will forget
3746 everything it learnt about the features supported by the configured
3747 upstream DNS servers, and restarts the feature probing logic on the
3748 next resolver look-up for them at the highest feature level
3751 * The status dump systemd-resolved sends to the logs upon receiving
3752 SIGUSR1 now also includes information about all DNS servers it is
3753 configured to use, and the features levels it probed for them.
3755 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
3756 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
3757 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
3758 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
3759 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
3760 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
3761 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
3762 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
3763 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
3764 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
3765 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
3766 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
3767 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
3768 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
3769 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
3770 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
3771 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
3772 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
3773 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
3774 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
3776 — Berlin, 2017-10-06
3780 * Meson is now supported as build system in addition to Automake. It is
3781 our plan to remove Automake in one of our next releases, so that
3782 Meson becomes our exclusive build system. Hence, please start using
3783 the Meson build system in your downstream packaging. There's plenty
3784 of documentation around how to use Meson, the extremely brief
3787 ./autogen.sh && ./configure && make && sudo make install
3791 meson build && ninja -C build && sudo ninja -C build install
3793 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
3794 which permits configuring a timeout on the time a job is
3795 running. This is particularly useful for setting timeouts on jobs for
3798 * Unit files gained two new options ConditionUser= and ConditionGroup=
3799 for conditionalizing units based on the identity of the user/group
3800 running a systemd user instance.
3802 * systemd-networkd now understands a new FlowLabel= setting in the
3803 [VXLAN] section of .network files, as well as a Priority= in
3804 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
3805 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
3806 gained support for configuration of GENEVE links, and IPv6 address
3807 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
3809 * .link files now understand a new Port= setting.
3811 * systemd-networkd's DHCP support gained support for DHCP option 119
3812 (domain search list).
3814 * systemd-networkd gained support for serving IPv6 address ranges using
3815 the Router Advertisement protocol. The new .network configuration
3816 section [IPv6Prefix] may be used to configure the ranges to
3817 serve. This is implemented based on a new, minimal, native server
3818 implementation of RA.
3820 * journalctl's --output= switch gained support for a new parameter
3821 "short-iso-precise" for a mode where timestamps are shown as precise
3824 * systemd-udevd's "net_id" builtin may now generate stable network
3825 interface names from IBM PowerVM VIO devices as well as ACPI platform
3828 * MulticastDNS support in systemd-resolved may now be explicitly
3829 enabled/disabled using the new MulticastDNS= configuration file
3832 * systemd-resolved may now optionally use libidn2 instead of the libidn
3833 for processing internationalized domain names. Support for libidn2
3834 should be considered experimental and should not be enabled by
3837 * "machinectl pull-tar" and related call may now do verification of
3838 downloaded images using SUSE-style .sha256 checksum files in addition
3839 to the already existing support for validating using Ubuntu-style
3842 * sd-bus gained support for a new sd_bus_message_appendv() call which
3843 is va_list equivalent of sd_bus_message_append().
3845 * sd-boot gained support for validating images using SHIM/MOK.
3847 * The SMACK code learnt support for "onlycap".
3849 * systemd-mount --umount is now much smarter in figuring out how to
3850 properly unmount a device given its mount or device path.
3852 * The code to call libnss_dns as a fallback from libnss_resolve when
3853 the communication with systemd-resolved fails was removed. This
3854 fallback was redundant and interfered with the [!UNAVAIL=return]
3855 suffix. See nss-resolve(8) for the recommended configuration.
3857 * systemd-logind may now be restarted without losing state. It stores
3858 the file descriptors for devices it manages in the system manager
3859 using the FDSTORE= mechanism. Please note that further changes in
3860 other components may be required to make use of this (for example
3861 Xorg has code to listen for stops of systemd-logind and terminate
3862 itself when logind is stopped or restarted, in order to avoid using
3863 stale file descriptors for graphical devices, which is now
3864 counterproductive and must be reverted in order for restarts of
3865 systemd-logind to be safe. See
3866 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
3868 * All kernel-install plugins are called with the environment variable
3869 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
3870 /etc/machine-id. If the machine ID could not be determined,
3871 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
3872 anything in the entry directory (passed as the second argument) if
3873 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
3874 temporary directory is passed as the entry directory and removed
3875 after all the plugins exit.
3877 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
3878 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
3879 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
3880 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
3881 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
3882 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
3883 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
3884 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
3885 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
3886 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
3887 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
3888 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
3889 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
3890 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
3891 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
3892 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
3893 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
3894 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
3895 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
3896 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
3897 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
3898 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
3899 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
3900 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
3901 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
3902 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
3903 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
3904 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
3905 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
3908 — Berlin, 2017-07-12
3912 * The "hybrid" control group mode has been modified to improve
3913 compatibility with "legacy" cgroups-v1 setups. Specifically, the
3914 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
3915 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
3916 cgroups-v1 hierarchy), the only externally visible change being that
3917 the cgroups-v2 hierarchy is also mounted, to
3918 /sys/fs/cgroup/unified. This should provide a large degree of
3919 compatibility with "legacy" cgroups-v1, while taking benefit of the
3920 better management capabilities of cgroups-v2.
3922 * The default control group setup mode may be selected both a boot-time
3923 via a set of kernel command line parameters (specifically:
3924 systemd.unified_cgroup_hierarchy= and
3925 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
3926 default selected on the configure command line
3927 (--with-default-hierarchy=). The upstream default is "hybrid"
3928 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
3929 this will change in a future systemd version to be "unified" (pure
3930 cgroups-v2 mode). The third option for the compile time option is
3931 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
3932 distributions to default to "hybrid" mode for release distributions,
3933 starting with v233. We recommend "unified" for development
3934 distributions (specifically: distributions such as Fedora's rawhide)
3935 as that's where things are headed in the long run. Use "legacy" for
3936 greatest stability and compatibility only.
3938 * Note one current limitation of "unified" and "hybrid" control group
3939 setup modes: the kernel currently does not permit the systemd --user
3940 instance (i.e. unprivileged code) to migrate processes between two
3941 disconnected cgroup subtrees, even if both are managed and owned by
3942 the user. This effectively means "systemd-run --user --scope" doesn't
3943 work when invoked from outside of any "systemd --user" service or
3944 scope. Specifically, it is not supported from session scopes. We are
3945 working on fixing this in a future systemd version. (See #3388 for
3946 further details about this.)
3948 * DBus policy files are now installed into /usr rather than /etc. Make
3949 sure your system has dbus >= 1.9.18 running before upgrading to this
3950 version, or override the install path with --with-dbuspolicydir= .
3952 * All python scripts shipped with systemd (specifically: the various
3953 tests written in Python) now require Python 3.
3955 * systemd unit tests can now run standalone (without the source or
3956 build directories), and can be installed into /usr/lib/systemd/tests/
3957 with 'make install-tests'.
3959 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
3960 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
3963 * Support for the %c, %r, %R specifiers in unit files has been
3964 removed. Specifiers are not supposed to be dependent on configuration
3965 in the unit file itself (so that they resolve the same regardless
3966 where used in the unit files), but these specifiers were influenced
3967 by the Slice= option.
3969 * The shell invoked by debug-shell.service now defaults to /bin/sh in
3970 all cases. If distributions want to use a different shell for this
3971 purpose (for example Fedora's /sbin/sushell) they need to specify
3972 this explicitly at configure time using --with-debug-shell=.
3974 * The confirmation spawn prompt has been reworked to offer the
3977 (c)ontinue, proceed without asking anymore
3978 (D)ump, show the state of the unit
3979 (f)ail, don't execute the command and pretend it failed
3981 (i)nfo, show a short summary of the unit
3982 (j)obs, show jobs that are in progress
3983 (s)kip, don't execute the command and pretend it succeeded
3984 (y)es, execute the command
3986 The 'n' choice for the confirmation spawn prompt has been removed,
3987 because its meaning was confusing.
3989 The prompt may now also be redirected to an alternative console by
3990 specifying the console as parameter to systemd.confirm_spawn=.
3992 * Services of Type=notify require a READY=1 notification to be sent
3993 during startup. If no such message is sent, the service now fails,
3994 even if the main process exited with a successful exit code.
3996 * Services that fail to start up correctly now always have their
3997 ExecStopPost= commands executed. Previously, they'd enter "failed"
3998 state directly, without executing these commands.
4000 * The option MulticastDNS= of network configuration files has acquired
4001 an actual implementation. With MulticastDNS=yes a host can resolve
4002 names of remote hosts and reply to mDNS A and AAAA requests.
4004 * When units are about to be started an additional check is now done to
4005 ensure that all dependencies of type BindsTo= (when used in
4006 combination with After=) have been started.
4008 * systemd-analyze gained a new verb "syscall-filter" which shows which
4009 system call groups are defined for the SystemCallFilter= unit file
4010 setting, and which system calls they contain.
4012 * A new system call filter group "@filesystem" has been added,
4013 consisting of various file system related system calls. Group
4014 "@reboot" has been added, covering reboot, kexec and shutdown related
4015 calls. Finally, group "@swap" has been added covering swap
4016 configuration related calls.
4018 * A new unit file option RestrictNamespaces= has been added that may be
4019 used to restrict access to the various process namespace types the
4020 Linux kernel provides. Specifically, it may be used to take away the
4021 right for a service unit to create additional file system, network,
4022 user, and other namespaces. This sandboxing option is particularly
4023 relevant due to the high amount of recently discovered namespacing
4024 related vulnerabilities in the kernel.
4026 * systemd-udev's .link files gained support for a new AutoNegotiation=
4027 setting for configuring Ethernet auto-negotiation.
4029 * systemd-networkd's .network files gained support for a new
4030 ListenPort= setting in the [DHCP] section to explicitly configure the
4031 UDP client port the DHCP client shall listen on.
4033 * .network files gained a new Unmanaged= boolean setting for explicitly
4034 excluding one or more interfaces from management by systemd-networkd.
4036 * The systemd-networkd ProxyARP= option has been renamed to
4037 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
4038 renamed to ReduceARPProxy=. The old names continue to be available
4041 * systemd-networkd gained support for configuring IPv6 Proxy NDP
4042 addresses via the new IPv6ProxyNDPAddress= .network file setting.
4044 * systemd-networkd's bonding device support gained support for two new
4045 configuration options ActiveSlave= and PrimarySlave=.
4047 * The various options in the [Match] section of .network files gained
4048 support for negative matching.
4050 * New systemd-specific mount options are now understood in /etc/fstab:
4052 x-systemd.mount-timeout= may be used to configure the maximum
4053 permitted runtime of the mount command.
4055 x-systemd.device-bound may be set to bind a mount point to its
4056 backing device unit, in order to automatically remove a mount point
4057 if its backing device is unplugged. This option may also be
4058 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
4059 on the block device, which is now automatically set for all CDROM
4060 drives, so that mounted CDs are automatically unmounted when they are
4061 removed from the drive.
4063 x-systemd.after= and x-systemd.before= may be used to explicitly
4064 order a mount after or before another unit or mount point.
4066 * Enqueued start jobs for device units are now automatically garbage
4067 collected if there are no jobs waiting for them anymore.
4069 * systemctl list-jobs gained two new switches: with --after, for every
4070 queued job the jobs it's waiting for are shown; with --before the
4071 jobs which it's blocking are shown.
4073 * systemd-nspawn gained support for ephemeral boots from disk images
4074 (or in other words: --ephemeral and --image= may now be
4075 combined). Moreover, ephemeral boots are now supported for normal
4076 directories, even if the backing file system is not btrfs. Of course,
4077 if the file system does not support file system snapshots or
4078 reflinks, the initial copy operation will be relatively expensive, but
4079 this should still be suitable for many use cases.
4081 * Calendar time specifications in .timer units now support
4082 specifications relative to the end of a month by using "~" instead of
4083 "-" as separator between month and day. For example, "*-02~03" means
4084 "the third last day in February". In addition a new syntax for
4085 repeated events has been added using the "/" character. For example,
4086 "9..17/2:00" means "every two hours from 9am to 5pm".
4088 * systemd-socket-proxyd gained a new parameter --connections-max= for
4089 configuring the maximum number of concurrent connections.
4091 * sd-id128 gained a new API for generating unique IDs for the host in a
4092 way that does not leak the machine ID. Specifically,
4093 sd_id128_get_machine_app_specific() derives an ID based on the
4094 machine ID a in well-defined, non-reversible, stable way. This is
4095 useful whenever an identifier for the host is needed but where the
4096 identifier shall not be useful to identify the system beyond the
4097 scope of the application itself. (Internally this uses HMAC-SHA256 as
4098 keyed hash function using the machine ID as input.)
4100 * NotifyAccess= gained a new supported value "exec". When set
4101 notifications are accepted from all processes systemd itself invoked,
4102 including all control processes.
4104 * .nspawn files gained support for defining overlay mounts using the
4105 Overlay= and OverlayReadOnly= options. Previously this functionality
4106 was only available on the systemd-nspawn command line.
4108 * systemd-nspawn's --bind= and --overlay= options gained support for
4109 bind/overlay mounts whose source lies within the container tree by
4110 prefixing the source path with "+".
4112 * systemd-nspawn's --bind= and --overlay= options gained support for
4113 automatically allocating a temporary source directory in /var/tmp
4114 that is removed when the container dies. Specifically, if the source
4115 directory is specified as empty string this mechanism is selected. An
4116 example usage is --overlay=+/var::/var, which creates an overlay
4117 mount based on the original /var contained in the image, overlaid
4118 with a temporary directory in the host's /var/tmp. This way changes
4119 to /var are automatically flushed when the container shuts down.
4121 * systemd-nspawn --image= option does now permit raw file system block
4122 devices (in addition to images containing partition tables, as
4125 * The disk image dissection logic in systemd-nspawn gained support for
4126 automatically setting up LUKS encrypted as well as Verity protected
4127 partitions. When a container is booted from an encrypted image the
4128 passphrase is queried at start-up time. When a container with Verity
4129 data is started, the root hash is search in a ".roothash" file
4130 accompanying the disk image (alternatively, pass the root hash via
4131 the new --root-hash= command line option).
4133 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
4134 be used to dissect disk images the same way as systemd-nspawn does
4135 it, following the Bootable Partition Specification. It may even be
4136 used to mount disk images with complex partition setups (including
4137 LUKS and Verity partitions) to a local host directory, in order to
4138 inspect them. This tool is not considered public API (yet), and is
4139 thus not installed into /usr/bin. Please do not rely on its
4140 existence, since it might go away or be changed in later systemd
4143 * A new generator "systemd-verity-generator" has been added, similar in
4144 style to "systemd-cryptsetup-generator", permitting automatic setup of
4145 Verity root partitions when systemd boots up. In order to make use of
4146 this your partition setup should follow the Discoverable Partitions
4147 Specification, and the GPT partition ID of the root file system
4148 partition should be identical to the upper 128bit of the Verity root
4149 hash. The GPT partition ID of the Verity partition protecting it
4150 should be the lower 128bit of the Verity root hash. If the partition
4151 image follows this model it is sufficient to specify a single
4152 "roothash=" kernel command line argument to both configure which root
4153 image and verity partition to use as well as the root hash for
4154 it. Note that systemd-nspawn's Verity support follows the same
4155 semantics, meaning that disk images with proper Verity data in place
4156 may be booted in containers with systemd-nspawn as well as on
4157 physical systems via the verity generator. Also note that the "mkosi"
4158 tool available at https://github.com/systemd/mkosi has been updated
4159 to generate Verity protected disk images following this scheme. In
4160 fact, it has been updated to generate disk images that optionally
4161 implement a complete UEFI SecureBoot trust chain, involving a signed
4162 kernel and initrd image that incorporates such a root hash as well as
4163 a Verity-enabled root partition.
4165 * The hardware database (hwdb) udev supports has been updated to carry
4166 accelerometer quirks.
4168 * All system services are now run with a fresh kernel keyring set up
4169 for them. The invocation ID is stored by default in it, thus
4170 providing a safe, non-overridable way to determine the invocation
4173 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
4174 options for bind mounting arbitrary paths in a service-specific
4175 way. When these options are used, arbitrary host or service files and
4176 directories may be mounted to arbitrary locations in the service's
4179 * Documentation has been added that lists all of systemd's low-level
4180 environment variables:
4182 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
4184 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
4185 whether a specific socket file descriptor matches a specified socket
4188 * systemd-firstboot has been updated to check for the
4189 systemd.firstboot= kernel command line option. It accepts a boolean
4190 and when set to false the first boot questions are skipped.
4192 * systemd-fstab-generator has been updated to check for the
4193 systemd.volatile= kernel command line option, which either takes an
4194 optional boolean parameter or the special value "state". If used the
4195 system may be booted in a "volatile" boot mode. Specifically,
4196 "systemd.volatile" is used, the root directory will be mounted as
4197 tmpfs, and only /usr is mounted from the actual root file system. If
4198 "systemd.volatile=state" is used, the root directory will be mounted
4199 as usual, but /var is mounted as tmpfs. This concept provides similar
4200 functionality as systemd-nspawn's --volatile= option, but provides it
4201 on physical boots. Use this option for implementing stateless
4202 systems, or testing systems with all state and/or configuration reset
4203 to the defaults. (Note though that many distributions are not
4204 prepared to boot up without a populated /etc or /var, though.)
4206 * systemd-gpt-auto-generator gained support for LUKS encrypted root
4207 partitions. Previously it only supported LUKS encrypted partitions
4208 for all other uses, except for the root partition itself.
4210 * Socket units gained support for listening on AF_VSOCK sockets for
4211 communication in virtualized QEMU environments.
4213 * The "configure" script gained a new option --with-fallback-hostname=
4214 for specifying the fallback hostname to use if none is configured in
4215 /etc/hostname. For example, by specifying
4216 --with-fallback-hostname=fedora it is possible to default to a
4217 hostname of "fedora" on pristine installations.
4219 * systemd-cgls gained support for a new --unit= switch for listing only
4220 the control groups of a specific unit. Similar --user-unit= has been
4221 added for listing only the control groups of a specific user unit.
4223 * systemd-mount gained a new --umount switch for unmounting a mount or
4224 automount point (and all mount/automount points below it).
4226 * systemd will now refuse full configuration reloads (via systemctl
4227 daemon-reload and related calls) unless at least 16MiB of free space
4228 are available in /run. This is a safety precaution in order to ensure
4229 that generators can safely operate after the reload completed.
4231 * A new unit file option RootImage= has been added, which has a similar
4232 effect as RootDirectory= but mounts the service's root directory from
4233 a disk image instead of plain directory. This logic reuses the same
4234 image dissection and mount logic that systemd-nspawn already uses,
4235 and hence supports any disk images systemd-nspawn supports, including
4236 those following the Discoverable Partition Specification, as well as
4237 Verity enabled images. This option enables systemd to run system
4238 services directly off disk images acting as resource bundles,
4239 possibly even including full integrity data.
4241 * A new MountAPIVFS= unit file option has been added, taking a boolean
4242 argument. If enabled /proc, /sys and /dev (collectively called the
4243 "API VFS") will be mounted for the service. This is only relevant if
4244 RootDirectory= or RootImage= is used for the service, as these mounts
4245 are of course in place in the host mount namespace anyway.
4247 * systemd-nspawn gained support for a new --pivot-root= switch. If
4248 specified the root directory within the container image is pivoted to
4249 the specified mount point, while the original root disk is moved to a
4250 different place. This option enables booting of ostree images
4251 directly with systemd-nspawn.
4253 * The systemd build scripts will no longer complain if the NTP server
4254 addresses are not changed from the defaults. Google now supports
4255 these NTP servers officially. We still recommend downstreams to
4256 properly register an NTP pool with the NTP pool project though.
4258 * coredumpctl gained a new "--reverse" option for printing the list
4259 of coredumps in reverse order.
4261 * coredumpctl will now show additional information about truncated and
4262 inaccessible coredumps, as well as coredumps that are still being
4263 processed. It also gained a new --quiet switch for suppressing
4264 additional informational message in its output.
4266 * coredumpctl gained support for only showing coredumps newer and/or
4267 older than specific timestamps, using the new --since= and --until=
4268 options, reminiscent of journalctl's options by the same name.
4270 * The systemd-coredump logic has been improved so that it may be reused
4271 to collect backtraces in non-compiled languages, for example in
4272 scripting languages such as Python.
4274 * machinectl will now show the UID shift of local containers, if user
4275 namespacing is enabled for them.
4277 * systemd will now optionally run "environment generator" binaries at
4278 configuration load time. They may be used to add environment
4279 variables to the environment block passed to services invoked. One
4280 user environment generator is shipped by default that sets up
4281 environment variables based on files dropped into /etc/environment.d
4282 and ~/.config/environment.d/.
4284 * systemd-resolved now includes the new, recently published 2017 DNSSEC
4287 * hostnamed has been updated to report a new chassis type of
4288 "convertible" to cover "foldable" laptops that can both act as a
4289 tablet and as a laptop, such as various Lenovo Yoga devices.
4291 Contributions from: Adrián López, Alexander Galanin, Alexander
4292 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
4293 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
4294 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
4295 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
4296 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
4297 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
4298 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
4299 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
4300 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
4301 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
4302 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
4303 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
4304 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
4305 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
4306 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
4307 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
4308 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
4309 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
4310 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
4311 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
4312 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
4313 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
4314 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
4315 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
4316 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
4317 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
4318 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
4321 — Berlin, 2017-03-01
4325 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
4326 RestrictAddressFamilies= enabled. These sandboxing options should
4327 generally be compatible with the various external udev call-out
4328 binaries we are aware of, however there may be exceptions, in
4329 particular when exotic languages for these call-outs are used. In
4330 this case, consider turning off these settings locally.
4332 * The new RemoveIPC= option can be used to remove IPC objects owned by
4333 the user or group of a service when that service exits.
4335 * The new ProtectKernelModules= option can be used to disable explicit
4336 load and unload operations of kernel modules by a service. In
4337 addition access to /usr/lib/modules is removed if this option is set.
4339 * ProtectSystem= option gained a new value "strict", which causes the
4340 whole file system tree with the exception of /dev, /proc, and /sys,
4341 to be remounted read-only for a service.
4343 * The new ProtectKernelTunables= option can be used to disable
4344 modification of configuration files in /sys and /proc by a service.
4345 Various directories and files are remounted read-only, so access is
4346 restricted even if the file permissions would allow it.
4348 * The new ProtectControlGroups= option can be used to disable write
4349 access by a service to /sys/fs/cgroup.
4351 * Various systemd services have been hardened with
4352 ProtectKernelTunables=yes, ProtectControlGroups=yes,
4353 RestrictAddressFamilies=.
4355 * Support for dynamically creating users for the lifetime of a service
4356 has been added. If DynamicUser=yes is specified, user and group IDs
4357 will be allocated from the range 61184..65519 for the lifetime of the
4358 service. They can be resolved using the new nss-systemd.so NSS
4359 module. The module must be enabled in /etc/nsswitch.conf. Services
4360 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
4361 any resources allocated by the service will be cleaned up when the
4362 service exits. They also have ProtectHome=read-only and
4363 ProtectSystem=strict enabled, so they are not able to make any
4364 permanent modifications to the system.
4366 * The nss-systemd module also always resolves root and nobody, making
4367 it possible to have no /etc/passwd or /etc/group files in minimal
4368 container or chroot environments.
4370 * Services may be started with their own user namespace using the new
4371 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
4372 under which the service is running are mapped. All other users are
4375 * Support for the cgroup namespace has been added to systemd-nspawn. If
4376 supported by kernel, the container system started by systemd-nspawn
4377 will have its own view of the cgroup hierarchy. This new behaviour
4378 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
4380 * The new MemorySwapMax= option can be used to limit the maximum swap
4381 usage under the unified cgroup hierarchy.
4383 * Support for the CPU controller in the unified cgroup hierarchy has
4384 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
4385 options. This controller requires out-of-tree patches for the kernel
4386 and the support is provisional.
4388 * Mount and automount units may now be created transiently
4389 (i.e. dynamically at runtime via the bus API, instead of requiring
4390 unit files in the file system).
4392 * systemd-mount is a new tool which may mount file systems – much like
4393 mount(8), optionally pulling in additional dependencies through
4394 transient .mount and .automount units. For example, this tool
4395 automatically runs fsck on a backing block device before mounting,
4396 and allows the automount logic to be used dynamically from the
4397 command line for establishing mount points. This tool is particularly
4398 useful when dealing with removable media, as it will ensure fsck is
4399 run – if necessary – before the first access and that the file system
4400 is quickly unmounted after each access by utilizing the automount
4401 logic. This maximizes the chance that the file system on the
4402 removable media stays in a clean state, and if it isn't in a clean
4403 state is fixed automatically.
4405 * LazyUnmount=yes option for mount units has been added to expose the
4406 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
4409 * /efi will be used as the mount point of the EFI boot partition, if
4410 the directory is present, and the mount point was not configured
4411 through other means (e.g. fstab). If /efi directory does not exist,
4412 /boot will be used as before. This makes it easier to automatically
4413 mount the EFI partition on systems where /boot is used for something
4416 * When operating on GPT disk images for containers, systemd-nspawn will
4417 now mount the ESP to /boot or /efi according to the same rules as PID
4418 1 running on a host. This allows tools like "bootctl" to operate
4419 correctly within such containers, in order to make container images
4420 bootable on physical systems.
4422 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
4424 * Two new user session targets have been added to support running
4425 graphical sessions under the systemd --user instance:
4426 graphical-session.target and graphical-session-pre.target. See
4427 systemd.special(7) for a description of how those targets should be
4430 * The vconsole initialization code has been significantly reworked to
4431 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
4432 support unicode keymaps. Font and keymap configuration will now be
4433 copied to all allocated virtual consoles.
4435 * FreeBSD's bhyve virtualization is now detected.
4437 * Information recorded in the journal for core dumps now includes the
4438 contents of /proc/mountinfo and the command line of the process at
4439 the top of the process hierarchy (which is usually the init process
4442 * systemd-journal-gatewayd learned the --directory= option to serve
4443 files from the specified location.
4445 * journalctl --root=… can be used to peruse the journal in the
4446 /var/log/ directories inside of a container tree. This is similar to
4447 the existing --machine= option, but does not require the container to
4450 * The hardware database has been extended to support
4451 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
4454 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
4455 specify the click rate for mice which include a horizontal wheel with
4456 a click rate that is different than the one for the vertical wheel.
4458 * systemd-run gained a new --wait option that makes service execution
4459 synchronous. (Specifically, the command will not return until the
4460 specified service binary exited.)
4462 * systemctl gained a new --wait option that causes the start command to
4463 wait until the units being started have terminated again.
4465 * A new journal output mode "short-full" has been added which displays
4466 timestamps with abbreviated English day names and adds a timezone
4467 suffix. Those timestamps include more information than the default
4468 "short" output mode, and can be passed directly to journalctl's
4469 --since= and --until= options.
4471 * /etc/resolv.conf will be bind-mounted into containers started by
4472 systemd-nspawn, if possible, so any changes to resolv.conf contents
4473 are automatically propagated to the container.
4475 * The number of instances for socket-activated services originating
4476 from a single IP address can be limited with
4477 MaxConnectionsPerSource=, extending the existing setting of
4480 * systemd-networkd gained support for vcan ("Virtual CAN") interface
4483 * .netdev and .network configuration can now be extended through
4486 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
4487 Segmentation Offload, Generic Receive Offload, Large Receive Offload
4488 can be enabled and disabled using the new UDPSegmentationOffload=,
4489 TCPSegmentationOffload=, GenericSegmentationOffload=,
4490 GenericReceiveOffload=, LargeReceiveOffload= options in the
4491 [Link] section of .link files.
4493 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
4494 Port VLAN ID can be configured for bridge devices using the new STP=,
4495 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
4496 section of .netdev files.
4498 * The route table to which routes received over DHCP or RA should be
4499 added can be configured with the new RouteTable= option in the [DHCP]
4500 and [IPv6AcceptRA] sections of .network files.
4502 * The Address Resolution Protocol can be disabled on links managed by
4503 systemd-networkd using the ARP=no setting in the [Link] section of
4506 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
4507 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
4508 encode information about the result and exit codes of the current
4509 service runtime cycle.
4511 * systemd-sysctl will now configure kernel parameters in the order
4512 they occur in the configuration files. This matches what sysctl
4513 has been traditionally doing.
4515 * kernel-install "plugins" that are executed to perform various
4516 tasks after a new kernel is added and before an old one is removed
4517 can now return a special value to terminate the procedure and
4518 prevent any later plugins from running.
4520 * Journald's SplitMode=login setting has been deprecated. It has been
4521 removed from documentation, and its use is discouraged. In a future
4522 release it will be completely removed, and made equivalent to current
4523 default of SplitMode=uid.
4525 * Storage=both option setting in /etc/systemd/coredump.conf has been
4526 removed. With fast LZ4 compression storing the core dump twice is not
4529 * The --share-system systemd-nspawn option has been replaced with an
4530 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
4531 this functionality is discouraged. In addition the variables
4532 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
4533 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
4534 individual namespaces.
4536 * "machinectl list" now shows the IP address of running containers in
4537 the output, as well as OS release information.
4539 * "loginctl list" now shows the TTY of each session in the output.
4541 * sd-bus gained new API calls sd_bus_track_set_recursive(),
4542 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
4543 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
4544 tracking objects in a "recursive" mode, where a single client can be
4545 counted multiple times, if it takes multiple references.
4547 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
4548 sd_bus_get_exit_on_disconnect(). They may be used to make a
4549 process using sd-bus automatically exit if the bus connection is
4552 * Bus clients of the service manager may now "pin" loaded units into
4553 memory, by taking an explicit reference on them. This is useful to
4554 ensure the client can retrieve runtime data about the service even
4555 after the service completed execution. Taking such a reference is
4556 available only for privileged clients and should be helpful to watch
4557 running services in a race-free manner, and in particular collect
4558 information about exit statuses and results.
4560 * The nss-resolve module has been changed to strictly return UNAVAIL
4561 when communication via D-Bus with resolved failed, and NOTFOUND when
4562 a lookup completed but was negative. This means it is now possible to
4563 neatly configure fallbacks using nsswitch.conf result checking
4564 expressions. Taking benefit of this, the new recommended
4565 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
4567 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
4569 * A new setting CtrlAltDelBurstAction= has been added to
4570 /etc/systemd/system.conf which may be used to configure the precise
4571 behaviour if the user on the console presses Ctrl-Alt-Del more often
4572 than 7 times in 2s. Previously this would unconditionally result in
4573 an expedited, immediate reboot. With this new setting the precise
4574 operation may be configured in more detail, and also turned off
4577 * In .netdev files two new settings RemoteChecksumTx= and
4578 RemoteChecksumRx= are now understood that permit configuring the
4579 remote checksumming logic for VXLAN networks.
4581 * The service manager learnt a new "invocation ID" concept for invoked
4582 services. Each runtime cycle of a service will get a new invocation
4583 ID (a 128bit random UUID) assigned that identifies the current
4584 run of the service uniquely and globally. A new invocation ID
4585 is generated each time a service starts up. The journal will store
4586 the invocation ID of a service along with any logged messages, thus
4587 making the invocation ID useful for matching the online runtime of a
4588 service with the offline log data it generated in a safe way without
4589 relying on synchronized timestamps. In many ways this new service
4590 invocation ID concept is similar to the kernel's boot ID concept that
4591 uniquely and globally identifies the runtime of each boot. The
4592 invocation ID of a service is passed to the service itself via an
4593 environment variable ($INVOCATION_ID). A new bus call
4594 GetUnitByInvocationID() has been added that is similar to GetUnit()
4595 but instead of retrieving the bus path for a unit by its name
4596 retrieves it by its invocation ID. The returned path is valid only as
4597 long as the passed invocation ID is current.
4599 * systemd-resolved gained a new "DNSStubListener" setting in
4600 resolved.conf. It either takes a boolean value or the special values
4601 "udp" and "tcp", and configures whether to enable the stub DNS
4602 listener on 127.0.0.53:53.
4604 * IP addresses configured via networkd may now carry additional
4605 configuration settings supported by the kernel. New options include:
4606 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
4607 PrefixRoute=, AutoJoin=.
4609 * The PAM configuration fragment file for "user@.service" shipped with
4610 systemd (i.e. the --user instance of systemd) has been stripped to
4611 the minimum necessary to make the system boot. Previously, it
4612 contained Fedora-specific stanzas that did not apply to other
4613 distributions. It is expected that downstream distributions add
4614 additional configuration lines, matching their needs to this file,
4615 using it only as rough template of what systemd itself needs. Note
4616 that this reduced fragment does not even include an invocation of
4617 pam_limits which most distributions probably want to add, even though
4618 systemd itself does not need it. (There's also the new build time
4619 option --with-pamconfdir=no to disable installation of the PAM
4622 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
4623 capability is now also dropped from its set (in addition to
4624 CAP_SYS_MKNOD as before).
4626 * In service unit files it is now possible to connect a specific named
4627 file descriptor with stdin/stdout/stdout of an executed service. The
4628 name may be specified in matching .socket units using the
4629 FileDescriptorName= setting.
4631 * A number of journal settings may now be configured on the kernel
4632 command line. Specifically, the following options are now understood:
4633 systemd.journald.max_level_console=,
4634 systemd.journald.max_level_store=,
4635 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
4636 systemd.journald.max_level_wall=.
4638 * "systemctl is-enabled --full" will now show by which symlinks a unit
4639 file is enabled in the unit dependency tree.
4641 * Support for VeraCrypt encrypted partitions has been added to the
4642 "cryptsetup" logic and /etc/crypttab.
4644 * systemd-detect-virt gained support for a new --private-users switch
4645 that checks whether the invoking processes are running inside a user
4646 namespace. Similar, a new special value "private-users" for the
4647 existing ConditionVirtualization= setting has been added, permitting
4648 skipping of specific units in user namespace environments.
4650 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
4651 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
4652 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
4653 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
4654 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
4655 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
4656 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
4657 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
4658 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
4659 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
4660 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
4661 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
4662 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
4663 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
4664 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
4665 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
4666 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
4667 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
4668 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
4669 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
4670 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
4671 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
4672 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
4673 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
4674 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
4675 Jędrzejewski-Szmek, Zeal Jagannatha
4677 — Santa Fe, 2016-11-03
4681 * In service units the various ExecXYZ= settings have been extended
4682 with an additional special character as first argument of the
4683 assigned value: if the character '+' is used the specified command
4684 line it will be run with full privileges, regardless of User=,
4685 Group=, CapabilityBoundingSet= and similar options. The effect is
4686 similar to the existing PermissionsStartOnly= option, but allows
4687 configuration of this concept for each executed command line
4690 * Services may now alter the service watchdog timeout at runtime by
4691 sending a WATCHDOG_USEC= message via sd_notify().
4693 * MemoryLimit= and related unit settings now optionally take percentage
4694 specifications. The percentage is taken relative to the amount of
4695 physical memory in the system (or in case of containers, the assigned
4696 amount of memory). This allows scaling service resources neatly with
4697 the amount of RAM available on the system. Similarly, systemd-logind's
4698 RuntimeDirectorySize= option now also optionally takes percentage
4701 * In similar fashion TasksMax= takes percentage values now, too. The
4702 value is taken relative to the configured maximum number of processes
4703 on the system. The per-service task maximum has been changed to 15%
4704 using this functionality. (Effectively this is an increase of 512 →
4705 4915 for service units, given the kernel's default pid_max setting.)
4707 * Calendar time specifications in .timer units now understand a ".."
4708 syntax for time ranges. Example: "4..7:10" may now be used for
4709 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
4712 * The InaccessableDirectories=, ReadOnlyDirectories= and
4713 ReadWriteDirectories= unit file settings have been renamed to
4714 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
4715 applied to all kinds of file nodes, and not just directories, with
4716 the exception of symlinks. Specifically these settings may now be
4717 used on block and character device nodes, UNIX sockets and FIFOS as
4718 well as regular files. The old names of these settings remain
4719 available for compatibility.
4721 * systemd will now log about all service processes it kills forcibly
4722 (using SIGKILL) because they remained after the clean shutdown phase
4723 of the service completed. This should help identifying services that
4724 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
4725 systemd-logind's configuration a similar log message is generated for
4726 processes killed at the end of each session due to this setting.
4728 * systemd will now set the $JOURNAL_STREAM environment variable for all
4729 services whose stdout/stderr are connected to the Journal (which
4730 effectively means by default: all services). The variable contains
4731 the device and inode number of the file descriptor used for
4732 stdout/stderr. This may be used by invoked programs to detect whether
4733 their stdout/stderr is connected to the Journal, in which case they
4734 can switch over to direct Journal communication, thus being able to
4735 pass extended, structured metadata along with their log messages. As
4736 one example, this is now used by glib's logging primitives.
4738 * When using systemd's default tmp.mount unit for /tmp, the mount point
4739 will now be established with the "nosuid" and "nodev" options. This
4740 avoids privilege escalation attacks that put traps and exploits into
4741 /tmp. However, this might cause problems if you e. g. put container
4742 images or overlays into /tmp; if you need this, override tmp.mount's
4743 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
4746 * systemd now supports the "memory" cgroup controller also on
4749 * The systemd-cgtop tool now optionally takes a control group path as
4750 command line argument. If specified, the control group list shown is
4751 limited to subgroups of that group.
4753 * The SystemCallFilter= unit file setting gained support for
4754 pre-defined, named system call filter sets. For example
4755 SystemCallFilter=@clock is now an effective way to make all clock
4756 changing-related system calls unavailable to a service. A number of
4757 similar pre-defined groups are defined. Writing system call filters
4758 for system services is simplified substantially with this new
4759 concept. Accordingly, all of systemd's own, long-running services now
4760 enable system call filtering based on this, by default.
4762 * A new service setting MemoryDenyWriteExecute= has been added, taking
4763 a boolean value. If turned on, a service may no longer create memory
4764 mappings that are writable and executable at the same time. This
4765 enhances security for services where this is enabled as it becomes
4766 harder to dynamically write and then execute memory in exploited
4767 service processes. This option has been enabled for all of systemd's
4768 own long-running services.
4770 * A new RestrictRealtime= service setting has been added, taking a
4771 boolean argument. If set the service's processes may no longer
4772 acquire realtime scheduling. This improves security as realtime
4773 scheduling may otherwise be used to easily freeze the system.
4775 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
4776 value. This may be used for requesting that the system manager inside
4777 of the container reports start-up completion to nspawn which then
4778 propagates this notification further to the service manager
4779 supervising nspawn itself. A related option NotifyReady= in .nspawn
4780 files has been added too. This functionality allows ordering of the
4781 start-up of multiple containers using the usual systemd ordering
4784 * machinectl gained a new command "stop" that is an alias for
4787 * systemd-resolved gained support for contacting DNS servers on
4788 link-local IPv6 addresses.
4790 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
4791 its caches. A method call for requesting the same operation has been
4792 added to the bus API too, and is made available via "systemd-resolve
4795 * systemd-resolve gained a new --status switch. If passed a brief
4796 summary of the used DNS configuration with per-interface information
4799 * resolved.conf gained a new Cache= boolean option, defaulting to
4800 on. If turned off local DNS caching is disabled. This comes with a
4801 performance penalty in particular when DNSSEC is enabled. Note that
4802 resolved disables its internal caching implicitly anyway, when the
4803 configured DNS server is on a host-local IP address such as ::1 or
4804 127.0.0.1, thus automatically avoiding double local caching.
4806 * systemd-resolved now listens on the local IP address 127.0.0.53:53
4807 for DNS requests. This improves compatibility with local programs
4808 that do not use the libc NSS or systemd-resolved's bus APIs for name
4809 resolution. This minimal DNS service is only available to local
4810 programs and does not implement the full DNS protocol, but enough to
4811 cover local DNS clients. A new, static resolv.conf file, listing just
4812 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
4813 now recommended to make /etc/resolv.conf a symlink to this file in
4814 order to route all DNS lookups to systemd-resolved, regardless if
4815 done via NSS, the bus API or raw DNS packets. Note that this local
4816 DNS service is not as fully featured as the libc NSS or
4817 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
4818 used to deliver link-local address information (as this implies
4819 sending a local interface index along), LLMNR/mDNS support via this
4820 interface is severely restricted. It is thus strongly recommended for
4821 all applications to use the libc NSS API or native systemd-resolved
4824 * systemd-networkd's bridge support learned a new setting
4825 VLANFiltering= for controlling VLAN filtering. Moreover a new section
4826 in .network files has been added for configuring VLAN bridging in
4827 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
4829 * systemd-networkd's IPv6 Router Advertisement code now makes use of
4830 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
4831 now be acquired without relying on DHCPv6. Two new options
4832 UseDomains= and UseDNS= have been added to configure this behaviour.
4834 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
4835 renamed IPv6AcceptRA=, without altering its behaviour. The old
4836 setting name remains available for compatibility reasons.
4838 * The systemd-networkd VTI/VTI6 tunneling support gained new options
4839 Key=, InputKey= and OutputKey=.
4841 * systemd-networkd gained support for VRF ("Virtual Routing Function")
4842 interface configuration.
4844 * "systemctl edit" may now be used to create new unit files by
4845 specifying the --force switch.
4847 * sd-event gained a new function sd_event_get_iteration() for
4848 requesting the current iteration counter of the event loop. It starts
4849 at zero and is increased by one with each event loop iteration.
4851 * A new rpm macro %systemd_ordering is provided by the macros.systemd
4852 file. It can be used in lieu of %systemd_requires in packages which
4853 don't use any systemd functionality and are intended to be installed
4854 in minimal containers without systemd present. This macro provides
4855 ordering dependencies to ensure that if the package is installed in
4856 the same rpm transaction as systemd, systemd will be installed before
4857 the scriptlets for the package are executed, allowing unit presets
4860 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
4861 been added to simplify packaging of generators.
4863 * The os-release file gained VERSION_CODENAME field for the
4864 distribution nickname (e.g. VERSION_CODENAME=woody).
4866 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
4867 can be set to disable parsing of metadata and the creation
4868 of persistent symlinks for that device.
4870 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
4871 to make them available to logged-in users has been reverted.
4873 * Much of the common code of the various systemd components is now
4874 built into an internal shared library libsystemd-shared-231.so
4875 (incorporating the systemd version number in the name, to be updated
4876 with future releases) that the components link to. This should
4877 decrease systemd footprint both in memory during runtime and on
4878 disk. Note that the shared library is not for public use, and is
4879 neither API nor ABI stable, but is likely to change with every new
4880 released update. Packagers need to make sure that binaries
4881 linking to libsystemd-shared.so are updated in step with the
4884 * Configuration for "mkosi" is now part of the systemd
4885 repository. mkosi is a tool to easily build legacy-free OS images,
4886 and is available on github: https://github.com/systemd/mkosi. If
4887 "mkosi" is invoked in the build tree a new raw OS image is generated
4888 incorporating the systemd sources currently being worked on and a
4889 clean, fresh distribution installation. The generated OS image may be
4890 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
4891 UEFI PC. This functionality is particularly useful to easily test
4892 local changes made to systemd in a pristine, defined environment. See
4893 doc/HACKING for details.
4895 * configure learned the --with-support-url= option to specify the
4896 distribution's bugtracker.
4898 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
4899 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
4900 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
4901 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
4902 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
4903 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
4904 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
4905 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
4906 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
4907 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
4908 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
4909 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
4910 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
4911 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
4912 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
4913 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
4914 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
4915 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
4916 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
4918 — Berlin, 2016-07-25
4922 * DNSSEC is now turned on by default in systemd-resolved (in
4923 "allow-downgrade" mode), but may be turned off during compile time by
4924 passing "--with-default-dnssec=no" to "configure" (and of course,
4925 during runtime with DNSSEC= in resolved.conf). We recommend
4926 downstreams to leave this on at least during development cycles and
4927 report any issues with the DNSSEC logic upstream. We are very
4928 interested in collecting feedback about the DNSSEC validator and its
4929 limitations in the wild. Note however, that DNSSEC support is
4930 probably nothing downstreams should turn on in stable distros just
4931 yet, as it might create incompatibilities with a few DNS servers and
4932 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
4933 automatically whenever we detect such incompatible setups, but there
4934 might be systems we do not cover yet. Hence: please help us testing
4935 the DNSSEC code, leave this on where you can, report back, but then
4936 again don't consider turning this on in your stable, LTS or
4937 production release just yet. (Note that you have to enable
4938 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
4939 and its DNSSEC mode for hostname resolution from local
4942 * systemd-resolve conveniently resolves DANE records with the --tlsa
4943 option and OPENPGPKEY records with the --openpgp option. It also
4944 supports dumping raw DNS record data via the new --raw= switch.
4946 * systemd-logind will now by default terminate user processes that are
4947 part of the user session scope unit (session-XX.scope) when the user
4948 logs out. This behavior is controlled by the KillUserProcesses=
4949 setting in logind.conf, and the previous default of "no" is now
4950 changed to "yes". This means that user sessions will be properly
4951 cleaned up after, but additional steps are necessary to allow
4952 intentionally long-running processes to survive logout.
4954 While the user is logged in at least once, user@.service is running,
4955 and any service that should survive the end of any individual login
4956 session can be started at a user service or scope using systemd-run.
4957 systemd-run(1) man page has been extended with an example which shows
4958 how to run screen in a scope unit underneath user@.service. The same
4959 command works for tmux.
4961 After the user logs out of all sessions, user@.service will be
4962 terminated too, by default, unless the user has "lingering" enabled.
4963 To effectively allow users to run long-term tasks even if they are
4964 logged out, lingering must be enabled for them. See loginctl(1) for
4965 details. The default polkit policy was modified to allow users to
4966 set lingering for themselves without authentication.
4968 Previous defaults can be restored at compile time by the
4969 --without-kill-user-processes option to "configure".
4971 * systemd-logind gained new configuration settings SessionsMax= and
4972 InhibitorsMax=, both with a default of 8192. It will not register new
4973 user sessions or inhibitors above this limit.
4975 * systemd-logind will now reload configuration on SIGHUP.
4977 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
4978 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
4979 enable. Also, support for the "io" cgroup controller in the unified
4980 hierarchy has been added, so that the "memory", "pids" and "io" are
4981 now the controllers that are supported on the unified hierarchy.
4983 WARNING: it is not possible to use previous systemd versions with
4984 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
4985 is necessary to also update systemd in the initramfs if using the
4986 unified hierarchy. An updated SELinux policy is also required.
4988 * LLDP support has been extended, and both passive (receive-only) and
4989 active (sender) modes are supported. Passive mode ("routers-only") is
4990 enabled by default in systemd-networkd. Active LLDP mode is enabled
4991 by default for containers on the internal network. The "networkctl
4992 lldp" command may be used to list information gathered. "networkctl
4993 status" will also show basic LLDP information on connected peers now.
4995 * The IAID and DUID unique identifier sent in DHCP requests may now be
4996 configured for the system and each .network file managed by
4997 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
4999 * systemd-networkd gained support for configuring proxy ARP support for
5000 each interface, via the ProxyArp= setting in .network files. It also
5001 gained support for configuring the multicast querier feature of
5002 bridge devices, via the new MulticastQuerier= setting in .netdev
5003 files. Similarly, snooping on the IGMP traffic can be controlled
5004 via the new setting MulticastSnooping=.
5006 A new setting PreferredLifetime= has been added for addresses
5007 configured in .network file to configure the lifetime intended for an
5010 The systemd-networkd DHCP server gained the option EmitRouter=, which
5011 defaults to yes, to configure whether the DHCP Option 3 (Router)
5014 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
5015 systemd-socket-activate and installed into /usr/bin. It is now fully
5018 * systemd-journald now uses separate threads to flush changes to disk
5019 when closing journal files, thus reducing impact of slow disk I/O on
5020 logging performance.
5022 * The sd-journal API gained two new calls
5023 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
5024 can be used to open journal files using file descriptors instead of
5025 file or directory paths. sd_journal_open_container() has been
5026 deprecated, sd_journal_open_directory_fd() should be used instead
5027 with the flag SD_JOURNAL_OS_ROOT.
5029 * journalctl learned a new output mode "-o short-unix" that outputs log
5030 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
5031 UTC). It also gained support for a new --no-hostname setting to
5032 suppress the hostname column in the family of "short" output modes.
5034 * systemd-ask-password now optionally skips printing of the password to
5035 stdout with --no-output which can be useful in scripts.
5037 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
5038 (devices tagged with ID_MAKER_TOOL) are now tagged with
5039 "uaccess" and are available to logged in users.
5041 * The DeviceAllow= unit setting now supports specifiers (with "%").
5043 * "systemctl show" gained a new --value switch, which allows print a
5044 only the contents of a specific unit property, without also printing
5045 the property's name. Similar support was added to "show*" verbs
5046 of loginctl and machinectl that output "key=value" lists.
5048 * A new unit type "generated" was added for files dynamically generated
5049 by generator tools. Similarly, a new unit type "transient" is used
5050 for unit files created using the runtime API. "systemctl enable" will
5051 refuse to operate on such files.
5053 * A new command "systemctl revert" has been added that may be used to
5054 revert to the vendor version of a unit file, in case local changes
5055 have been made by adding drop-ins or overriding the unit file.
5057 * "machinectl clean" gained a new verb to automatically remove all or
5058 just hidden container images.
5060 * systemd-tmpfiles gained support for a new line type "e" for emptying
5061 directories, if they exist, without creating them if they don't.
5063 * systemd-nspawn gained support for automatically patching the UID/GIDs
5064 of the owners and the ACLs of all files and directories in a
5065 container tree to match the UID/GID user namespacing range selected
5066 for the container invocation. This mode is enabled via the new
5067 --private-users-chown switch. It also gained support for
5068 automatically choosing a free, previously unused UID/GID range when
5069 starting a container, via the new --private-users=pick setting (which
5070 implies --private-users-chown). Together, these options for the first
5071 time make user namespacing for nspawn containers fully automatic and
5072 thus deployable. The systemd-nspawn@.service template unit file has
5073 been changed to use this functionality by default.
5075 * systemd-nspawn gained a new --network-zone= switch, that allows
5076 creating ad-hoc virtual Ethernet links between multiple containers,
5077 that only exist as long as at least one container referencing them is
5078 running. This allows easy connecting of multiple containers with a
5079 common link that implements an Ethernet broadcast domain. Each of
5080 these network "zones" may be named relatively freely by the user, and
5081 may be referenced by any number of containers, but each container may
5082 only reference one of these "zones". On the lower level, this is
5083 implemented by an automatically managed bridge network interface for
5084 each zone, that is created when the first container referencing its
5085 zone is created and removed when the last one referencing its zone
5088 * The default start timeout may now be configured on the kernel command
5089 line via systemd.default_timeout_start_sec=. It was already
5090 configurable via the DefaultTimeoutStartSec= option in
5091 /etc/systemd/system.conf.
5093 * Socket units gained a new TriggerLimitIntervalSec= and
5094 TriggerLimitBurst= setting to configure a limit on the activation
5095 rate of the socket unit.
5097 * The LimitNICE= setting now optionally takes normal UNIX nice values
5098 in addition to the raw integer limit value. If the specified
5099 parameter is prefixed with "+" or "-" and is in the range -20..19 the
5100 value is understood as UNIX nice value. If not prefixed like this it
5101 is understood as raw RLIMIT_NICE limit.
5103 * Note that the effect of the PrivateDevices= unit file setting changed
5104 slightly with this release: the per-device /dev file system will be
5105 mounted read-only from this version on, and will have "noexec"
5106 set. This (minor) change of behavior might cause some (exceptional)
5107 legacy software to break, when PrivateDevices=yes is set for its
5108 service. Please leave PrivateDevices= off if you run into problems
5111 * systemd-bootchart has been split out to a separate repository:
5112 https://github.com/systemd/systemd-bootchart
5114 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
5115 merged into the kernel in its current form.
5117 * The compatibility libraries libsystemd-daemon.so,
5118 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
5119 which have been deprecated since systemd-209 have been removed along
5120 with the corresponding pkg-config files. All symbols provided by
5121 those libraries are provided by libsystemd.so.
5123 * The Capabilities= unit file setting has been removed (it is ignored
5124 for backwards compatibility). AmbientCapabilities= and
5125 CapabilityBoundingSet= should be used instead.
5127 * A new special target has been added, initrd-root-device.target,
5128 which creates a synchronization point for dependencies of the root
5129 device in early userspace. Initramfs builders must ensure that this
5130 target is now included in early userspace.
5132 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
5133 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
5134 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
5135 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
5136 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
5137 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
5138 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
5139 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
5140 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
5141 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
5142 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
5143 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
5144 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
5145 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
5146 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
5147 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
5148 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
5149 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
5150 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
5151 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
5152 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
5153 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
5154 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
5155 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
5156 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
5159 — Fairfax, 2016-05-21
5163 * The systemd-resolved DNS resolver service has gained a substantial
5164 set of new features, most prominently it may now act as a DNSSEC
5165 validating stub resolver. DNSSEC mode is currently turned off by
5166 default, but is expected to be turned on by default in one of the
5167 next releases. For now, we invite everybody to test the DNSSEC logic
5168 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
5169 service also gained a full set of D-Bus interfaces, including calls
5170 to configure DNS and DNSSEC settings per link (for use by external
5171 network management software). systemd-resolved and systemd-networkd
5172 now distinguish between "search" and "routing" domains. The former
5173 are used to qualify single-label names, the latter are used purely
5174 for routing lookups within certain domains to specific links.
5175 resolved now also synthesizes RRs for all entries from /etc/hosts.
5177 * The systemd-resolve tool (which is a client utility for
5178 systemd-resolved) has been improved considerably and is now fully
5179 supported and documented. Hence it has moved from /usr/lib/systemd to
5182 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
5185 * The coredump collection logic has been reworked: when a coredump is
5186 collected it is now written to disk, compressed and processed
5187 (including stacktrace extraction) from a new instantiated service
5188 systemd-coredump@.service, instead of directly from the
5189 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
5190 processing large coredumps can take up a substantial amount of
5191 resources and time, and this previously happened entirely outside of
5192 systemd's service supervision. With the new logic the core_pattern
5193 hook only does minimal metadata collection before passing off control
5194 to the new instantiated service, which is configured with a time
5195 limit, a nice level and other settings to minimize negative impact on
5196 the rest of the system. Also note that the new logic will honour the
5197 RLIMIT_CORE setting of the crashed process, which now allows users
5198 and processes to turn off coredumping for their processes by setting
5201 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
5202 and all forked processes by default. Previously, PID 1 would leave
5203 the setting at "0" for all processes, as set by the kernel. Note that
5204 the resource limit traditionally has no effect on the generated
5205 coredumps on the system if the /proc/sys/kernel/core_pattern hook
5206 logic is used. Since the limit is now honoured (see above) its
5207 default has been changed so that the coredumping logic is enabled by
5208 default for all processes, while allowing specific opt-out.
5210 * When the stacktrace is extracted from processes of system users, this
5211 is now done as "systemd-coredump" user, in order to sandbox this
5212 potentially security sensitive parsing operation. (Note that when
5213 processing coredumps of normal users this is done under the user ID
5214 of process that crashed, as before.) Packagers should take notice
5215 that it is now necessary to create the "systemd-coredump" system user
5216 and group at package installation time.
5218 * The systemd-activate socket activation testing tool gained support
5219 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
5220 and --seqpacket switches. It also has been extended to support both
5221 new-style and inetd-style file descriptor passing. Use the new
5222 --inetd switch to request inetd-style file descriptor passing.
5224 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
5225 variable, which takes a boolean value. If set to false, ANSI color
5226 output is disabled in the tools even when run on a terminal that
5229 * The VXLAN support in networkd now supports two new settings
5230 DestinationPort= and PortRange=.
5232 * A new systemd.machine_id= kernel command line switch has been added,
5233 that may be used to set the machine ID in /etc/machine-id if it is
5234 not initialized yet. This command line option has no effect if the
5235 file is already initialized.
5237 * systemd-nspawn gained a new --as-pid2 switch that invokes any
5238 specified command line as PID 2 rather than PID 1 in the
5239 container. In this mode PID 1 is a minimal stub init process that
5240 implements the special POSIX and Linux semantics of PID 1 regarding
5241 signal and child process management. Note that this stub init process
5242 is implemented in nspawn itself and requires no support from the
5243 container image. This new logic is useful to support running
5244 arbitrary commands in the container, as normal processes are
5245 generally not prepared to run as PID 1.
5247 * systemd-nspawn gained a new --chdir= switch for setting the current
5248 working directory for the process started in the container.
5250 * "journalctl /dev/sda" will now output all kernel log messages for
5251 specified device from the current boot, in addition to all devices
5252 that are parents of it. This should make log output about devices
5253 pretty useful, as long as kernel drivers attach enough metadata to
5254 the log messages. (The usual SATA drivers do.)
5256 * The sd-journal API gained two new calls
5257 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
5258 that report whether log data from /run or /var has been found.
5260 * journalctl gained a new switch "--fields" that prints all journal
5261 record field names currently in use in the journal. This is backed
5262 by two new sd-journal API calls sd_journal_enumerate_fields() and
5263 sd_journal_restart_fields().
5265 * Most configurable timeouts in systemd now expect an argument of
5266 "infinity" to turn them off, instead of "0" as before. The semantics
5267 from now on is that a timeout of "0" means "now", and "infinity"
5268 means "never". To maintain backwards compatibility, "0" continues to
5269 turn off previously existing timeout settings.
5271 * "systemctl reload-or-try-restart" has been renamed to "systemctl
5272 try-reload-or-restart" to clarify what it actually does: the "try"
5273 logic applies to both reloading and restarting, not just restarting.
5274 The old name continues to be accepted for compatibility.
5276 * On boot-up, when PID 1 detects that the system clock is behind the
5277 release date of the systemd version in use, the clock is now set
5278 to the latter. Previously, this was already done in timesyncd, in order
5279 to avoid running with clocks set to the various clock epochs such as
5280 1902, 1938 or 1970. With this change the logic is now done in PID 1
5281 in addition to timesyncd during early boot-up, so that it is enforced
5282 before the first process is spawned by systemd. Note that the logic
5283 in timesyncd remains, as it is more comprehensive and ensures
5284 clock monotonicity by maintaining a persistent timestamp file in
5285 /var. Since /var is generally not available in earliest boot or the
5286 initrd, this part of the logic remains in timesyncd, and is not done
5289 * Support for tweaking details in net_cls.class_id through the
5290 NetClass= configuration directive has been removed, as the kernel
5291 people have decided to deprecate that controller in cgroup v2.
5292 Userspace tools such as nftables are moving over to setting rules
5293 that are specific to the full cgroup path of a task, which obsoletes
5294 these controllers anyway. The NetClass= directive is kept around for
5295 legacy compatibility reasons. For a more in-depth description of the
5296 kernel change, please refer to the respective upstream commit:
5298 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
5300 * A new service setting RuntimeMaxSec= has been added that may be used
5301 to specify a maximum runtime for a service. If the timeout is hit, the
5302 service is terminated and put into a failure state.
5304 * A new service setting AmbientCapabilities= has been added. It allows
5305 configuration of additional Linux process capabilities that are
5306 passed to the activated processes. This is only available on very
5309 * The process resource limit settings in service units may now be used
5310 to configure hard and soft limits individually.
5312 * The various libsystemd APIs such as sd-bus or sd-event now publicly
5313 expose support for gcc's __attribute__((cleanup())) C extension.
5314 Specifically, for many object destructor functions alternative
5315 versions have been added that have names suffixed with "p" and take a
5316 pointer to a pointer to the object to destroy, instead of just a
5317 pointer to the object itself. This is useful because these destructor
5318 functions may be used directly as parameters to the cleanup
5319 construct. Internally, systemd has been a heavy user of this GCC
5320 extension for a long time, and with this change similar support is
5321 now available to consumers of the library outside of systemd. Note
5322 that by using this extension in your sources compatibility with old
5323 and strictly ANSI compatible C compilers is lost. However, all gcc or
5324 LLVM versions of recent years support this extension.
5326 * Timer units gained support for a new setting RandomizedDelaySec= that
5327 allows configuring some additional randomized delay to the configured
5328 time. This is useful to spread out timer events to avoid load peaks in
5329 clusters or larger setups.
5331 * Calendar time specifications now support sub-second accuracy.
5333 * Socket units now support listening on SCTP and UDP-lite protocol
5336 * The sd-event API now comes with a full set of man pages.
5338 * Older versions of systemd contained experimental support for
5339 compressing journal files and coredumps with the LZ4 compressor that
5340 was not compatible with the lz4 binary (due to API limitations of the
5341 lz4 library). This support has been removed; only support for files
5342 compatible with the lz4 binary remains. This LZ4 logic is now
5343 officially supported and no longer considered experimental.
5345 * The dkr image import logic has been removed again from importd. dkr's
5346 micro-services focus doesn't fit into the machine image focus of
5347 importd, and quickly got out of date with the upstream dkr API.
5349 * Creation of the /run/lock/lockdev/ directory was dropped from
5350 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
5351 been available for many years. If you still need this, you need to
5352 create your own tmpfiles.d config file with:
5354 d /run/lock/lockdev 0775 root lock -
5356 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
5357 and RebootArgument= have been moved from the [Service] section of
5358 unit files to [Unit], and they are now supported on all unit types,
5359 not just service units. Of course, systemd will continue to
5360 understand these settings also at the old location, in order to
5361 maintain compatibility.
5363 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
5364 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
5365 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
5366 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
5367 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
5368 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
5369 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
5370 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
5371 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
5372 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
5373 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
5374 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
5375 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
5376 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
5377 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
5378 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
5379 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
5380 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
5381 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
5383 — Berlin, 2016-02-11
5387 * A number of properties previously only settable in unit
5388 files are now also available as properties to set when
5389 creating transient units programmatically via the bus, as it
5390 is exposed with systemd-run's --property=
5391 setting. Specifically, these are: SyslogIdentifier=,
5392 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
5393 EnvironmentFile=, ReadWriteDirectories=,
5394 ReadOnlyDirectories=, InaccessibleDirectories=,
5395 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
5397 * When creating transient services via the bus API it is now
5398 possible to pass in a set of file descriptors to use as
5399 STDIN/STDOUT/STDERR for the invoked process.
5401 * Slice units may now be created transiently via the bus APIs,
5402 similar to the way service and scope units may already be
5403 created transiently.
5405 * Wherever systemd expects a calendar timestamp specification
5406 (like in journalctl's --since= and --until= switches) UTC
5407 timestamps are now supported. Timestamps suffixed with "UTC"
5408 are now considered to be in Universal Time Coordinated
5409 instead of the local timezone. Also, timestamps may now
5410 optionally be specified with sub-second accuracy. Both of
5411 these additions also apply to recurring calendar event
5412 specification, such as OnCalendar= in timer units.
5414 * journalctl gained a new "--sync" switch that asks the
5415 journal daemon to write all so far unwritten log messages to
5416 disk and sync the files, before returning.
5418 * systemd-tmpfiles learned two new line types "q" and "Q" that
5419 operate like "v", but also set up a basic btrfs quota
5420 hierarchy when used on a btrfs file system with quota
5423 * tmpfiles' "v", "q" and "Q" will now create a plain directory
5424 instead of a subvolume (even on a btrfs file system) if the
5425 root directory is a plain directory, and not a
5426 subvolume. This should simplify things with certain chroot()
5427 environments which are not aware of the concept of btrfs
5430 * systemd-detect-virt gained a new --chroot switch to detect
5431 whether execution takes place in a chroot() environment.
5433 * CPUAffinity= now takes CPU index ranges in addition to
5436 * The various memory-related resource limit settings (such as
5437 LimitAS=) now understand the usual K, M, G, ... suffixes to
5438 the base of 1024 (IEC). Similar, the time-related resource
5439 limit settings understand the usual min, h, day, ...
5442 * There's a new system.conf setting DefaultTasksMax= to
5443 control the default TasksMax= setting for services and
5444 scopes running on the system. (TasksMax= is the primary
5445 setting that exposes the "pids" cgroup controller on systemd
5446 and was introduced in the previous systemd release.) The
5447 setting now defaults to 512, which means services that are
5448 not explicitly configured otherwise will only be able to
5449 create 512 processes or threads at maximum, from this
5450 version on. Note that this means that thread- or
5451 process-heavy services might need to be reconfigured to set
5452 TasksMax= to a higher value. It is sufficient to set
5453 TasksMax= in these specific unit files to a higher value, or
5454 even "infinity". Similar, there's now a logind.conf setting
5455 UserTasksMax= that defaults to 4096 and limits the total
5456 number of processes or tasks each user may own
5457 concurrently. nspawn containers also have the TasksMax=
5458 value set by default now, to 8192. Note that all of this
5459 only has an effect if the "pids" cgroup controller is
5460 enabled in the kernel. The general benefit of these changes
5461 should be a more robust and safer system, that provides a
5462 certain amount of per-service fork() bomb protection.
5464 * systemd-nspawn gained the new --network-veth-extra= switch
5465 to define additional and arbitrarily-named virtual Ethernet
5466 links between the host and the container.
5468 * A new service execution setting PassEnvironment= has been
5469 added that allows importing select environment variables
5470 from PID1's environment block into the environment block of
5473 * Timer units gained support for a new RemainAfterElapse=
5474 setting which takes a boolean argument. It defaults to on,
5475 exposing behaviour unchanged to previous releases. If set to
5476 off, timer units are unloaded after they elapsed if they
5477 cannot elapse again. This is particularly useful for
5478 transient timer units, which shall not stay around longer
5479 than until they first elapse.
5481 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
5482 default now (the kernel default is 16). This is beneficial
5483 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
5484 allows substantially larger numbers of queued
5485 datagrams. This should increase the capability of systemd to
5486 parallelize boot-up, as logging and sd_notify() are unlikely
5487 to stall execution anymore. If you need to change the value
5488 from the new defaults, use the usual sysctl.d/ snippets.
5490 * The compression framing format used by the journal or
5491 coredump processing has changed to be in line with what the
5492 official LZ4 tools generate. LZ4 compression support in
5493 systemd was considered unsupported previously, as the format
5494 was not compatible with the normal tools. With this release
5495 this has changed now, and it is hence safe for downstream
5496 distributions to turn it on. While not compressing as well
5497 as the XZ, LZ4 is substantially faster, which makes
5498 it a good default choice for the compression logic in the
5499 journal and in coredump handling.
5501 * Any reference to /etc/mtab has been dropped from
5502 systemd. The file has been obsolete since a while, but
5503 systemd refused to work on systems where it was incorrectly
5504 set up (it should be a symlink or non-existent). Please make
5505 sure to update to util-linux 2.27.1 or newer in conjunction
5506 with this systemd release, which also drops any reference to
5507 /etc/mtab. If you maintain a distribution make sure that no
5508 software you package still references it, as this is a
5509 likely source of bugs. There's also a glibc bug pending,
5510 asking for removal of any reference to this obsolete file:
5512 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
5514 Note that only util-linux versions built with
5515 --enable-libmount-force-mountinfo are supported.
5517 * Support for the ".snapshot" unit type has been removed. This
5518 feature turned out to be little useful and little used, and
5519 has now been removed from the core and from systemctl.
5521 * The dependency types RequiresOverridable= and
5522 RequisiteOverridable= have been removed from systemd. They
5523 have been used only very sparingly to our knowledge and
5524 other options that provide a similar effect (such as
5525 systemctl --mode=ignore-dependencies) are much more useful
5526 and commonly used. Moreover, they were only half-way
5527 implemented as the option to control behaviour regarding
5528 these dependencies was never added to systemctl. By removing
5529 these dependency types the execution engine becomes a bit
5530 simpler. Unit files that use these dependencies should be
5531 changed to use the non-Overridable dependency types
5532 instead. In fact, when parsing unit files with these
5533 options, that's what systemd will automatically convert them
5534 too, but it will also warn, asking users to fix the unit
5535 files accordingly. Removal of these dependency types should
5536 only affect a negligible number of unit files in the wild.
5538 * Behaviour of networkd's IPForward= option changed
5539 (again). It will no longer maintain a per-interface setting,
5540 but propagate one way from interfaces where this is enabled
5541 to the global kernel setting. The global setting will be
5542 enabled when requested by a network that is set up, but
5543 never be disabled again. This change was made to make sure
5544 IPv4 and IPv6 behaviour regarding packet forwarding is
5545 similar (as the Linux IPv6 stack does not support
5546 per-interface control of this setting) and to minimize
5549 * In unit files the behaviour of %u, %U, %h, %s has
5550 changed. These specifiers will now unconditionally resolve
5551 to the various user database fields of the user that the
5552 systemd instance is running as, instead of the user
5553 configured in the specific unit via User=. Note that this
5554 effectively doesn't change much, as resolving of these
5555 specifiers was already turned off in the --system instance
5556 of systemd, as we cannot do NSS lookups from PID 1. In the
5557 --user instance of systemd these specifiers where correctly
5558 resolved, but hardly made any sense, since the user instance
5559 lacks privileges to do user switches anyway, and User= is
5560 hence useless. Moreover, even in the --user instance of
5561 systemd behaviour was awkward as it would only take settings
5562 from User= assignment placed before the specifier into
5563 account. In order to unify and simplify the logic around
5564 this the specifiers will now always resolve to the
5565 credentials of the user invoking the manager (which in case
5566 of PID 1 is the root user).
5568 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
5569 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
5570 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
5571 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
5572 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
5573 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
5574 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
5575 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
5576 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
5577 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
5578 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
5579 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
5580 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
5581 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
5584 — Berlin, 2015-11-18
5588 * systemd now depends on util-linux v2.27. More specifically,
5589 the newly added mount monitor feature in libmount now
5590 replaces systemd's former own implementation.
5592 * libmount mandates /etc/mtab not to be regular file, and
5593 systemd now enforces this condition at early boot.
5594 /etc/mtab has been deprecated and warned about for a very
5595 long time, so systems running systemd should already have
5596 stopped having this file around as anything else than a
5597 symlink to /proc/self/mounts.
5599 * Support for the "pids" cgroup controller has been added. It
5600 allows accounting the number of tasks in a cgroup and
5601 enforcing limits on it. This adds two new setting
5602 TasksAccounting= and TasksMax= to each unit, as well as a
5603 global option DefaultTasksAccounting=.
5605 * Support for the "net_cls" cgroup controller has been added.
5606 It allows assigning a net class ID to each task in the
5607 cgroup, which can then be used in firewall rules and traffic
5608 shaping configurations. Note that the kernel netfilter net
5609 class code does not currently work reliably for ingress
5610 packets on unestablished sockets.
5612 This adds a new config directive called NetClass= to CGroup
5613 enabled units. Allowed values are positive numbers for fixed
5614 assignments and "auto" for picking a free value
5617 * 'systemctl is-system-running' now returns 'offline' if the
5618 system is not booted with systemd. This command can now be
5619 used as a substitute for 'systemd-notify --booted'.
5621 * Watchdog timeouts have been increased to 3 minutes for all
5622 in-tree service files. Apparently, disk IO issues are more
5623 frequent than we hoped, and user reported >1 minute waiting
5626 * 'machine-id-commit' functionality has been merged into
5627 'machine-id-setup --commit'. The separate binary has been
5630 * The WorkingDirectory= directive in unit files may now be set
5631 to the special value '~'. In this case, the working
5632 directory is set to the home directory of the user
5633 configured in User=.
5635 * "machinectl shell" will now open the shell in the home
5636 directory of the selected user by default.
5638 * The CrashChVT= configuration file setting is renamed to
5639 CrashChangeVT=, following our usual logic of not
5640 abbreviating unnecessarily. The old directive is still
5641 supported for compat reasons. Also, this directive now takes
5642 an integer value between 1 and 63, or a boolean value. The
5643 formerly supported '-1' value for disabling stays around for
5646 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
5647 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
5648 RootDirectory= properties can now be set for transient
5651 * The systemd-analyze tool gained a new "set-log-target" verb
5652 to change the logging target the system manager logs to
5653 dynamically during runtime. This is similar to how
5654 "systemd-analyze set-log-level" already changes the log
5657 * In nspawn /sys is now mounted as tmpfs, with only a selected
5658 set of subdirectories mounted in from the real sysfs. This
5659 enhances security slightly, and is useful for ensuring user
5660 namespaces work correctly.
5662 * Support for USB FunctionFS activation has been added. This
5663 allows implementation of USB gadget services that are
5664 activated as soon as they are requested, so that they don't
5665 have to run continuously, similar to classic socket
5668 * The "systemctl exit" command now optionally takes an
5669 additional parameter that sets the exit code to return from
5670 the systemd manager when exiting. This is only relevant when
5671 running the systemd user instance, or when running the
5672 system instance in a container.
5674 * sd-bus gained the new API calls sd_bus_path_encode_many()
5675 and sd_bus_path_decode_many() that allow easy encoding and
5676 decoding of multiple identifier strings inside a D-Bus
5677 object path. Another new call sd_bus_default_flush_close()
5678 has been added to flush and close per-thread default
5681 * systemd-cgtop gained support for a -M/--machine= switch to
5682 show the control groups within a certain container only.
5684 * "systemctl kill" gained support for an optional --fail
5685 switch. If specified the requested operation will fail of no
5686 processes have been killed, because the unit had no
5687 processes attached, or similar.
5689 * A new systemd.crash_reboot=1 kernel command line option has
5690 been added that triggers a reboot after crashing. This can
5691 also be set through CrashReboot= in systemd.conf.
5693 * The RuntimeDirectory= setting now understands unit
5694 specifiers like %i or %f.
5696 * A new (still internal) library API sd-ipv4acd has been added,
5697 that implements address conflict detection for IPv4. It's
5698 based on code from sd-ipv4ll, and will be useful for
5699 detecting DHCP address conflicts.
5701 * File descriptors passed during socket activation may now be
5702 named. A new API sd_listen_fds_with_names() is added to
5703 access the names. The default names may be overridden,
5704 either in the .socket file using the FileDescriptorName=
5705 parameter, or by passing FDNAME= when storing the file
5706 descriptors using sd_notify().
5708 * systemd-networkd gained support for:
5710 - Setting the IPv6 Router Advertisement settings via
5711 IPv6AcceptRouterAdvertisements= in .network files.
5713 - Configuring the HelloTimeSec=, MaxAgeSec= and
5714 ForwardDelaySec= bridge parameters in .netdev files.
5716 - Configuring PreferredSource= for static routes in
5719 * The "ask-password" framework used to query for LUKS harddisk
5720 passwords or SSL passwords during boot gained support for
5721 caching passwords in the kernel keyring, if it is
5722 available. This makes sure that the user only has to type in
5723 a passphrase once if there are multiple objects to unlock
5724 with the same one. Previously, such password caching was
5725 available only when Plymouth was used; this moves the
5726 caching logic into the systemd codebase itself. The
5727 "systemd-ask-password" utility gained a new --keyname=
5728 switch to control which kernel keyring key to use for
5729 caching a password in. This functionality is also useful for
5730 enabling display managers such as gdm to automatically
5731 unlock the user's GNOME keyring if its passphrase, the
5732 user's password and the harddisk password are the same, if
5733 gdm-autologin is used.
5735 * When downloading tar or raw images using "machinectl
5736 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
5737 file is now also downloaded, if it is available and stored
5738 next to the image file.
5740 * Units of type ".socket" gained a new boolean setting
5741 Writable= which is only useful in conjunction with
5742 ListenSpecial=. If true, enables opening the specified
5743 special file in O_RDWR mode rather than O_RDONLY mode.
5745 * systemd-rfkill has been reworked to become a singleton
5746 service that is activated through /dev/rfkill on each rfkill
5747 state change and saves the settings to disk. This way,
5748 systemd-rfkill is now compatible with devices that exist
5749 only intermittendly, and even restores state if the previous
5750 system shutdown was abrupt rather than clean.
5752 * The journal daemon gained support for vacuuming old journal
5753 files controlled by the number of files that shall remain,
5754 in addition to the already existing control by size and by
5755 date. This is useful as journal interleaving performance
5756 degrades with too many separate journal files, and allows
5757 putting an effective limit on them. The new setting defaults
5758 to 100, but this may be changed by setting SystemMaxFiles=
5759 and RuntimeMaxFiles= in journald.conf. Also, the
5760 "journalctl" tool gained the new --vacuum-files= switch to
5761 manually vacuum journal files to leave only the specified
5762 number of files in place.
5764 * udev will now create /dev/disk/by-path links for ATA devices
5765 on kernels where that is supported.
5767 * Galician, Serbian, Turkish and Korean translations were added.
5769 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
5770 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
5771 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
5772 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
5773 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
5774 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
5775 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
5776 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
5777 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
5778 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
5779 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
5780 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
5781 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
5782 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
5783 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
5784 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
5785 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
5786 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
5788 — Berlin, 2015-10-07
5792 * The DHCP implementation of systemd-networkd gained a set of
5795 - The DHCP server now supports emitting DNS and NTP
5796 information. It may be enabled and configured via
5797 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
5798 and NTP information is enabled, but no servers are
5799 configured, the corresponding uplink information (if there
5800 is any) is propagated.
5802 - Server and client now support transmission and reception
5803 of timezone information. It can be configured via the
5804 newly introduced network options UseTimezone=,
5805 EmitTimezone=, and Timezone=. Transmission of timezone
5806 information is enabled between host and containers by
5807 default now: the container will change its local timezone
5808 to what the host has set.
5810 - Lease timeouts can now be configured via
5811 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
5813 - The DHCP server improved on the stability of
5814 leases. Clients are more likely to get the same lease
5815 information back, even if the server loses state.
5817 - The DHCP server supports two new configuration options to
5818 control the lease address pool metrics, PoolOffset= and
5821 * The encapsulation limit of tunnels in systemd-networkd may
5822 now be configured via 'EncapsulationLimit='. It allows
5823 modifying the maximum additional levels of encapsulation
5824 that are permitted to be prepended to a packet.
5826 * systemd now supports the concept of user buses replacing
5827 session buses, if used with dbus-1.10 (and enabled via dbus
5828 --enable-user-session). It previously only supported this on
5829 kdbus-enabled systems, and this release expands this to
5830 'dbus-daemon' systems.
5832 * systemd-networkd now supports predictable interface names
5835 * systemd now optionally supports the new Linux kernel
5836 "unified" control group hierarchy. If enabled via the kernel
5837 command-line option 'systemd.unified_cgroup_hierarchy=1',
5838 systemd will try to mount the unified cgroup hierarchy
5839 directly on /sys/fs/cgroup. If not enabled, or not
5840 available, systemd will fall back to the legacy cgroup
5841 hierarchy setup, as before. Host system and containers can
5842 mix and match legacy and unified hierarchies as they
5843 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
5844 environment variable to individually select the hierarchy to
5845 use for executed containers. By default, nspawn will use the
5846 unified hierarchy for the containers if the host uses the
5847 unified hierarchy, and the legacy hierarchy otherwise.
5848 Please note that at this point the unified hierarchy is an
5849 experimental kernel feature and is likely to change in one
5850 of the next kernel releases. Therefore, it should not be
5851 enabled by default in downstream distributions yet. The
5852 minimum required kernel version for the unified hierarchy to
5853 work is 4.2. Note that when the unified hierarchy is used
5854 for the first time delegated access to controllers is
5855 safe. Because of this systemd-nspawn containers will get
5856 access to controllers now, as will systemd user
5857 sessions. This means containers and user sessions may now
5858 manage their own resources, partitioning up what the system
5861 * A new special scope unit "init.scope" has been introduced
5862 that encapsulates PID 1 of the system. It may be used to
5863 determine resource usage and enforce resource limits on PID
5864 1 itself. PID 1 hence moved out of the root of the control
5867 * The cgtop tool gained support for filtering out kernel
5868 threads when counting tasks in a control group. Also, the
5869 count of processes is now recursively summed up by
5870 default. Two options -k and --recursive= have been added to
5871 revert to old behaviour. The tool has also been updated to
5872 work correctly in containers now.
5874 * systemd-nspawn's --bind= and --bind-ro= options have been
5875 extended to allow creation of non-recursive bind mounts.
5877 * libsystemd gained two new calls sd_pid_get_cgroup() and
5878 sd_peer_get_cgroup() which return the control group path of
5879 a process or peer of a connected AF_UNIX socket. This
5880 function call is particularly useful when implementing
5881 delegated subtrees support in the control group hierarchy.
5883 * The "sd-event" event loop API of libsystemd now supports
5884 correct dequeuing of real-time signals, without losing
5887 * When systemd requests a polkit decision when managing units it
5888 will now add additional fields to the request, including unit
5889 name and desired operation. This enables more powerful polkit
5890 policies, that make decisions depending on these parameters.
5892 * nspawn learnt support for .nspawn settings files, that may
5893 accompany the image files or directories of containers, and
5894 may contain additional settings for the container. This is
5895 an alternative to configuring container parameters via the
5896 nspawn command line.
5898 Contributions from: Cristian Rodríguez, Daniel Mack, David
5899 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
5900 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
5901 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
5902 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
5903 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
5904 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
5905 Andersen, Tom Gundersen, Torstein Husebø
5907 — Berlin, 2015-09-08
5911 * machinectl gained a new verb 'shell' which opens a fresh
5912 shell on the target container or the host. It is similar to
5913 the existing 'login' command of machinectl, but spawns the
5914 shell directly without prompting for username or
5915 password. The pseudo machine '.host' now refers to the local
5916 host and is used by default. Hence, 'machinectl shell' can
5917 be used as replacement for 'su -' which spawns a session as
5918 a fresh systemd unit in a way that is fully isolated from
5919 the originating session.
5921 * systemd-networkd learned to cope with private-zone DHCP
5922 options and allows other programs to query the values.
5924 * SELinux access control when enabling/disabling units is no
5925 longer enforced with this release. The previous implementation
5926 was incorrect, and a new corrected implementation is not yet
5927 available. As unit file operations are still protected via
5928 polkit and D-Bus policy this is not a security problem. Yet,
5929 distributions which care about optimal SELinux support should
5930 probably not stabilize on this release.
5932 * sd-bus gained support for matches of type "arg0has=", that
5933 test for membership of strings in string arrays sent in bus
5936 * systemd-resolved now dumps the contents of its DNS and LLMNR
5937 caches to the logs on reception of the SIGUSR1 signal. This
5938 is useful to debug DNS behaviour.
5940 * The coredumpctl tool gained a new --directory= option to
5941 operate on journal files in a specific directory.
5943 * "systemctl reboot" and related commands gained a new
5944 "--message=" option which may be used to set a free-text
5945 wall message when shutting down or rebooting the
5946 system. This message is also logged, which is useful for
5947 figuring out the reason for a reboot or shutdown a
5950 * The "systemd-resolve-host" tool's -i switch now takes
5951 network interface numbers as alternative to interface names.
5953 * A new unit file setting for services has been introduced:
5954 UtmpMode= allows configuration of how precisely systemd
5955 handles utmp and wtmp entries for the service if this is
5956 enabled. This allows writing services that appear similar to
5957 user sessions in the output of the "w", "who", "last" and
5960 * systemd-resolved will now locally synthesize DNS resource
5961 records for the "localhost" and "gateway" domains as well as
5962 the local hostname. This should ensure that clients querying
5963 RRs via resolved will get similar results as those going via
5964 NSS, if nss-myhostname is enabled.
5966 Contributions from: Alastair Hughes, Alex Crawford, Daniel
5967 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
5968 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
5969 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
5970 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
5971 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
5972 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
5973 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
5974 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
5975 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
5976 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
5977 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
5979 — Berlin, 2015-08-27
5983 * The systemd-efi-boot-generator functionality was merged into
5984 systemd-gpt-auto-generator.
5986 * systemd-networkd now supports Group Policy for vxlan
5987 devices. It can be enabled via the new boolean configuration
5988 option called 'GroupPolicyExtension='.
5990 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
5991 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
5992 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
5994 — Berlin, 2015-07-31
5998 * The python-systemd code has been removed from the systemd repository.
5999 A new repository has been created which accommodates the code from
6000 now on, and we kindly ask distributions to create a separate package
6001 for this: https://github.com/systemd/python-systemd
6003 * The systemd daemon will now reload its main configuration
6004 (/etc/systemd/system.conf) on daemon-reload.
6006 * sd-dhcp now exposes vendor specific extensions via
6007 sd_dhcp_lease_get_vendor_specific().
6009 * systemd-networkd gained a number of new configuration options.
6011 - A new boolean configuration option for TAP devices called
6012 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
6013 device, thus allowing to send and receive GSO packets.
6015 - A new tunnel configuration option called 'CopyDSCP='.
6016 If enabled, the DSCP field of ip6 tunnels is copied into the
6017 decapsulated packet.
6019 - A set of boolean bridge configuration options were added.
6020 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
6021 and 'UnicastFlood=' are now parsed by networkd and applied to the
6022 respective bridge link device via the respective IFLA_BRPORT_*
6025 - A new string configuration option to override the hostname sent
6026 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
6027 is true, networkd will use the configured hostname instead of the
6028 system hostname when sending DHCP requests.
6030 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
6031 networkd will configure the IPv6 flow-label of the tunnel device
6032 according to RFC2460.
6034 - The 'macvtap' virtual network devices are now supported, similar to
6035 the already supported 'macvlan' devices.
6037 * systemd-resolved now implements RFC5452 to improve resilience against
6038 cache poisoning. Additionally, source port randomization is enabled
6039 by default to further protect against DNS spoofing attacks.
6041 * nss-mymachines now supports translating UIDs and GIDs of running
6042 containers with user-namespaces enabled. If a container 'foo'
6043 translates a host uid 'UID' to the container uid 'TUID', then
6044 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
6045 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
6046 mapped as 'vg-foo-TGID'.
6048 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
6049 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
6050 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
6051 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
6052 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
6053 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
6054 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
6055 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
6056 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
6057 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
6059 — Berlin, 2015-07-29
6063 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
6064 There are no known issues with current sysfs, and udev does not need
6065 or should be used to work around such bugs.
6067 * udev does no longer enable USB HID power management. Several reports
6068 indicate, that some devices cannot handle that setting.
6070 * The udev accelerometer helper was removed. The functionality
6071 is now fully included in iio-sensor-proxy. But this means,
6072 older iio-sensor-proxy versions will no longer provide
6073 accelerometer/orientation data with this systemd version.
6074 Please upgrade iio-sensor-proxy to version 1.0.
6076 * networkd gained a new configuration option IPv6PrivacyExtensions=
6077 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
6078 for Stateless Address") on selected networks.
6080 * For the sake of fewer build-time dependencies and less code in the
6081 main repository, the python bindings are about to be removed in the
6082 next release. A new repository has been created which accommodates
6083 the code from now on, and we kindly ask distributions to create a
6084 separate package for this. The removal will take place in v223.
6086 https://github.com/systemd/python-systemd
6088 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
6089 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
6090 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
6091 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
6092 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
6093 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
6094 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
6095 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
6096 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
6097 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
6099 — Berlin, 2015-07-07
6103 * The sd-bus.h and sd-event.h APIs have now been declared
6104 stable and have been added to the official interface of
6105 libsystemd.so. sd-bus implements an alternative D-Bus client
6106 library, that is relatively easy to use, very efficient and
6107 supports both classic D-Bus as well as kdbus as transport
6108 backend. sd-event is a generic event loop abstraction that
6109 is built around Linux epoll, but adds features such as event
6110 prioritization or efficient timer handling. Both APIs are good
6111 choices for C programs looking for a bus and/or event loop
6112 implementation that is minimal and does not have to be
6113 portable to other kernels.
6115 * kdbus support is no longer compile-time optional. It is now
6116 always built-in. However, it can still be disabled at
6117 runtime using the kdbus=0 kernel command line setting, and
6118 that setting may be changed to default to off, by specifying
6119 --disable-kdbus at build-time. Note though that the kernel
6120 command line setting has no effect if the kdbus.ko kernel
6121 module is not installed, in which case kdbus is (obviously)
6122 also disabled. We encourage all downstream distributions to
6123 begin testing kdbus by adding it to the kernel images in the
6124 development distributions, and leaving kdbus support in
6127 * The minimal required util-linux version has been bumped to
6130 * Support for chkconfig (--enable-chkconfig) was removed in
6131 favor of calling an abstraction tool
6132 /lib/systemd/systemd-sysv-install. This needs to be
6133 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
6134 in README for details.
6136 * If there's a systemd unit and a SysV init script for the
6137 same service name, and the user executes "systemctl enable"
6138 for it (or a related call), then this will now enable both
6139 (or execute the related operation on both), not just the
6142 * The libudev API documentation has been converted from gtkdoc
6145 * gudev has been removed from the systemd tree, it is now an
6148 * The systemd-cgtop tool learnt a new --raw switch to generate
6149 "raw" (machine parsable) output.
6151 * networkd's IPForwarding= .network file setting learnt the
6152 new setting "kernel", which ensures that networkd does not
6153 change the IP forwarding sysctl from the default kernel
6156 * The systemd-logind bus API now exposes a new boolean
6157 property "Docked" that reports whether logind considers the
6158 system "docked", i.e. connected to a docking station or not.
6160 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
6161 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
6162 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
6163 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
6164 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
6165 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
6166 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
6167 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
6168 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
6169 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
6170 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
6171 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
6172 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
6173 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
6174 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
6175 Fink, Zbigniew Jędrzejewski-Szmek
6177 — Berlin, 2015-06-19
6181 * The gudev library has been extracted into a separate repository
6182 available at: https://git.gnome.org/browse/libgudev/
6183 It is now managed as part of the Gnome project. Distributions
6184 are recommended to pass --disable-gudev to systemd and use
6185 gudev from the Gnome project instead. gudev is still included
6186 in systemd, for now. It will be removed soon, though. Please
6187 also see the announcement-thread on systemd-devel:
6188 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
6190 * systemd now exposes a CPUUsageNSec= property for each
6191 service unit on the bus, that contains the overall consumed
6192 CPU time of a service (the sum of what each process of the
6193 service consumed). This value is only available if
6194 CPUAccounting= is turned on for a service, and is then shown
6195 in the "systemctl status" output.
6197 * Support for configuring alternative mappings of the old SysV
6198 runlevels to systemd targets has been removed. They are now
6199 hardcoded in a way that runlevels 2, 3, 4 all map to
6200 multi-user.target and 5 to graphical.target (which
6201 previously was already the default behaviour).
6203 * The auto-mounter logic gained support for mount point
6204 expiry, using a new TimeoutIdleSec= setting in .automount
6205 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
6207 * The EFI System Partition (ESP) as mounted to /boot by
6208 systemd-efi-boot-generator will now be unmounted
6209 automatically after 2 minutes of not being used. This should
6210 minimize the risk of ESP corruptions.
6212 * New /etc/fstab options x-systemd.requires= and
6213 x-systemd.requires-mounts-for= are now supported to express
6214 additional dependencies for mounts. This is useful for
6215 journalling file systems that support external journal
6216 devices or overlay file systems that require underlying file
6217 systems to be mounted.
6219 * systemd does not support direct live-upgrades (via systemctl
6220 daemon-reexec) from versions older than v44 anymore. As no
6221 distribution we are aware of shipped such old versions in a
6222 stable release this should not be problematic.
6224 * When systemd forks off a new per-connection service instance
6225 it will now set the $REMOTE_ADDR environment variable to the
6226 remote IP address, and $REMOTE_PORT environment variable to
6227 the remote IP port. This behaviour is similar to the
6228 corresponding environment variables defined by CGI.
6230 * systemd-networkd gained support for uplink failure
6231 detection. The BindCarrier= option allows binding interface
6232 configuration dynamically to the link sense of other
6233 interfaces. This is useful to achieve behaviour like in
6236 * systemd-networkd gained support for configuring the DHCP
6237 client identifier to use when requesting leases.
6239 * systemd-networkd now has a per-network UseNTP= option to
6240 configure whether NTP server information acquired via DHCP
6241 is passed on to services like systemd-timesyncd.
6243 * systemd-networkd gained support for vti6 tunnels.
6245 * Note that systemd-networkd manages the sysctl variable
6246 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
6247 it is configured for since v219. The variable controls IP
6248 forwarding, and is a per-interface alternative to the global
6249 /proc/sys/net/ipv[46]/ip_forward. This setting is
6250 configurable in the IPForward= option, which defaults to
6251 "no". This means if networkd is used for an interface it is
6252 no longer sufficient to set the global sysctl option to turn
6253 on IP forwarding! Instead, the .network file option
6254 IPForward= needs to be turned on! Note that the
6255 implementation of this behaviour was broken in v219 and has
6258 * Many bonding and vxlan options are now configurable in
6261 * systemd-nspawn gained a new --property= setting to set unit
6262 properties for the container scope. This is useful for
6263 setting resource parameters (e.g. "CPUShares=500") on
6264 containers started from the command line.
6266 * systemd-nspawn gained a new --private-users= switch to make
6267 use of user namespacing available on recent Linux kernels.
6269 * systemd-nspawn may now be called as part of a shell pipeline
6270 in which case the pipes used for stdin and stdout are passed
6271 directly to the process invoked in the container, without
6272 indirection via a pseudo tty.
6274 * systemd-nspawn gained a new switch to control the UNIX
6275 signal to use when killing the init process of the container
6278 * systemd-nspawn gained a new --overlay= switch for mounting
6279 overlay file systems into the container using the new kernel
6282 * When a container image is imported via systemd-importd and
6283 the host file system is not btrfs, a loopback block device
6284 file is created in /var/lib/machines.raw with a btrfs file
6285 system inside. It is then mounted to /var/lib/machines to
6286 enable btrfs features for container management. The loopback
6287 file and btrfs file system is grown as needed when container
6288 images are imported via systemd-importd.
6290 * systemd-machined/systemd-importd gained support for btrfs
6291 quota, to enforce container disk space limits on disk. This
6292 is exposed in "machinectl set-limit".
6294 * systemd-importd now can import containers from local .tar,
6295 .raw and .qcow2 images, and export them to .tar and .raw. It
6296 can also import dkr v2 images now from the network (on top
6299 * systemd-importd gained support for verifying downloaded
6300 images with gpg2 (previously only gpg1 was supported).
6302 * systemd-machined, systemd-logind, systemd: most bus calls are
6303 now accessible to unprivileged processes via polkit. Also,
6304 systemd-logind will now allow users to kill their own sessions
6305 without further privileges or authorization.
6307 * systemd-shutdownd has been removed. This service was
6308 previously responsible for implementing scheduled shutdowns
6309 as exposed in /usr/bin/shutdown's time parameter. This
6310 functionality has now been moved into systemd-logind and is
6311 accessible via a bus interface.
6313 * "systemctl reboot" gained a new switch --firmware-setup that
6314 can be used to reboot into the EFI firmware setup, if that
6315 is available. systemd-logind now exposes an API on the bus
6316 to trigger such reboots, in case graphical desktop UIs want
6317 to cover this functionality.
6319 * "systemctl enable", "systemctl disable" and "systemctl mask"
6320 now support a new "--now" switch. If specified the units
6321 that are enabled will also be started, and the ones
6322 disabled/masked also stopped.
6324 * The Gummiboot EFI boot loader tool has been merged into
6325 systemd, and renamed to "systemd-boot". The bootctl tool has been
6326 updated to support systemd-boot.
6328 * An EFI kernel stub has been added that may be used to create
6329 kernel EFI binaries that contain not only the actual kernel,
6330 but also an initrd, boot splash, command line and OS release
6331 information. This combined binary can then be signed as a
6332 single image, so that the firmware can verify it all in one
6333 step. systemd-boot has special support for EFI binaries created
6334 like this and can extract OS release information from them
6335 and show them in the boot menu. This functionality is useful
6336 to implement cryptographically verified boot schemes.
6338 * Optional support has been added to systemd-fsck to pass
6339 fsck's progress report to an AF_UNIX socket in the file
6342 * udev will no longer create device symlinks for all block devices by
6343 default. A deny list for excluding special block devices from this
6344 logic has been turned into a allow list that requires picking block
6345 devices explicitly that require device symlinks.
6347 * A new (currently still internal) API sd-device.h has been
6348 added to libsystemd. This modernized API is supposed to
6349 replace libudev eventually. In fact, already much of libudev
6350 is now just a wrapper around sd-device.h.
6352 * A new hwdb database for storing metadata about pointing
6353 stick devices has been added.
6355 * systemd-tmpfiles gained support for setting file attributes
6356 similar to the "chattr" tool with new 'h' and 'H' lines.
6358 * systemd-journald will no longer unconditionally set the
6359 btrfs NOCOW flag on new journal files. This is instead done
6360 with tmpfiles snippet using the new 'h' line type. This
6361 allows easy disabling of this logic, by masking the
6362 journal-nocow.conf tmpfiles file.
6364 * systemd-journald will now translate audit message types to
6365 human readable identifiers when writing them to the
6366 journal. This should improve readability of audit messages.
6368 * The LUKS logic gained support for the offset= and skip=
6369 options in /etc/crypttab, as previously implemented by
6372 * /usr/lib/os-release gained a new optional field VARIANT= for
6373 distributions that support multiple variants (such as a
6374 desktop edition, a server edition, ...)
6376 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
6377 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
6378 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
6379 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
6380 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
6381 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
6382 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
6383 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
6384 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
6385 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
6386 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
6387 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
6388 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
6389 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
6390 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
6391 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
6392 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
6393 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
6394 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
6395 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
6396 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
6397 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
6398 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
6399 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
6400 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
6401 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
6402 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
6404 — Berlin, 2015-05-22
6408 * Introduce a new API "sd-hwdb.h" for querying the hardware
6409 metadata database. With this minimal interface one can query
6410 and enumerate the udev hwdb, decoupled from the old libudev
6411 library. libudev's interface for this is now only a wrapper
6412 around sd-hwdb. A new tool systemd-hwdb has been added to
6413 interface with and update the database.
6415 * When any of systemd's tools copies files (for example due to
6416 tmpfiles' C lines) a btrfs reflink will attempted first,
6417 before bytewise copying is done.
6419 * systemd-nspawn gained a new --ephemeral switch. When
6420 specified a btrfs snapshot is taken of the container's root
6421 directory, and immediately removed when the container
6422 terminates again. Thus, a container can be started whose
6423 changes never alter the container's root directory, and are
6424 lost on container termination. This switch can also be used
6425 for starting a container off the root file system of the
6426 host without affecting the host OS. This switch is only
6427 available on btrfs file systems.
6429 * systemd-nspawn gained a new --template= switch. It takes the
6430 path to a container tree to use as template for the tree
6431 specified via --directory=, should that directory be
6432 missing. This allows instantiating containers dynamically,
6433 on first run. This switch is only available on btrfs file
6436 * When a .mount unit refers to a mount point on which multiple
6437 mounts are stacked, and the .mount unit is stopped all of
6438 the stacked mount points will now be unmounted until no
6439 mount point remains.
6441 * systemd now has an explicit notion of supported and
6442 unsupported unit types. Jobs enqueued for unsupported unit
6443 types will now fail with an "unsupported" error code. More
6444 specifically .swap, .automount and .device units are not
6445 supported in containers, .busname units are not supported on
6446 non-kdbus systems. .swap and .automount are also not
6447 supported if their respective kernel compile time options
6450 * machinectl gained support for two new "copy-from" and
6451 "copy-to" commands for copying files from a running
6452 container to the host or vice versa.
6454 * machinectl gained support for a new "bind" command to bind
6455 mount host directories into local containers. This is
6456 currently only supported for nspawn containers.
6458 * networkd gained support for configuring bridge forwarding
6459 database entries (fdb) from .network files.
6461 * A new tiny daemon "systemd-importd" has been added that can
6462 download container images in tar, raw, qcow2 or dkr formats,
6463 and make them available locally in /var/lib/machines, so
6464 that they can run as nspawn containers. The daemon can GPG
6465 verify the downloads (not supported for dkr, since it has no
6466 provisions for verifying downloads). It will transparently
6467 decompress bz2, xz, gzip compressed downloads if necessary,
6468 and restore sparse files on disk. The daemon uses privilege
6469 separation to ensure the actual download logic runs with
6470 fewer privileges than the daemon itself. machinectl has
6471 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
6472 make the functionality of importd available to the
6473 user. With this in place the Fedora and Ubuntu "Cloud"
6474 images can be downloaded and booted as containers unmodified
6475 (the Fedora images lack the appropriate GPG signature files
6476 currently, so they cannot be verified, but this will change
6477 soon, hopefully). Note that downloading images is currently
6478 only fully supported on btrfs.
6480 * machinectl is now able to list container images found in
6481 /var/lib/machines, along with some metadata about sizes of
6482 disk and similar. If the directory is located on btrfs and
6483 quota is enabled, this includes quota display. A new command
6484 "image-status" has been added that shows additional
6485 information about images.
6487 * machinectl is now able to clone container images
6488 efficiently, if the underlying file system (btrfs) supports
6489 it, with the new "machinectl clone" command. It also
6490 gained commands for renaming and removing images, as well as
6491 marking them read-only or read-write (supported also on
6492 legacy file systems).
6494 * networkd gained support for collecting LLDP network
6495 announcements, from hardware that supports this. This is
6496 shown in networkctl output.
6498 * systemd-run gained support for a new -t (--pty) switch for
6499 invoking a binary on a pty whose input and output is
6500 connected to the invoking terminal. This allows executing
6501 processes as system services while interactively
6502 communicating with them via the terminal. Most interestingly
6503 this is supported across container boundaries. Invoking
6504 "systemd-run -t /bin/bash" is an alternative to running a
6505 full login session, the difference being that the former
6506 will not register a session, nor go through the PAM session
6509 * tmpfiles gained support for a new "v" line type for creating
6510 btrfs subvolumes. If the underlying file system is a legacy
6511 file system, this automatically degrades to creating a
6512 normal directory. Among others /var/lib/machines is now
6513 created like this at boot, should it be missing.
6515 * The directory /var/lib/containers/ has been deprecated and
6516 been replaced by /var/lib/machines. The term "machines" has
6517 been used in the systemd context as generic term for both
6518 VMs and containers, and hence appears more appropriate for
6519 this, as the directory can also contain raw images bootable
6522 * systemd-nspawn when invoked with -M but without --directory=
6523 or --image= is now capable of searching for the container
6524 root directory, subvolume or disk image automatically, in
6525 /var/lib/machines. systemd-nspawn@.service has been updated
6526 to make use of this, thus allowing it to be used for raw
6529 * A new machines.target unit has been introduced that is
6530 supposed to group all containers/VMs invoked as services on
6531 the system. systemd-nspawn@.service has been updated to
6532 integrate with that.
6534 * machinectl gained a new "start" command, for invoking a
6535 container as a service. "machinectl start foo" is mostly
6536 equivalent to "systemctl start systemd-nspawn@foo.service",
6537 but handles escaping in a nicer way.
6539 * systemd-nspawn will now mount most of the cgroupfs tree
6540 read-only into each container, with the exception of the
6541 container's own subtree in the name=systemd hierarchy.
6543 * journald now sets the special FS_NOCOW file flag for its
6544 journal files. This should improve performance on btrfs, by
6545 avoiding heavy fragmentation when journald's write-pattern
6546 is used on COW file systems. It degrades btrfs' data
6547 integrity guarantees for the files to the same levels as for
6548 ext3/ext4 however. This should be OK though as journald does
6549 its own data integrity checks and all its objects are
6550 checksummed on disk. Also, journald should handle btrfs disk
6551 full events a lot more gracefully now, by processing SIGBUS
6552 errors, and not relying on fallocate() anymore.
6554 * When journald detects that journal files it is writing to
6555 have been deleted it will immediately start new journal
6558 * systemd now provides a way to store file descriptors
6559 per-service in PID 1. This is useful for daemons to ensure
6560 that fds they require are not lost during a daemon
6561 restart. The fds are passed to the daemon on the next
6562 invocation in the same way socket activation fds are
6563 passed. This is now used by journald to ensure that the
6564 various sockets connected to all the system's stdout/stderr
6565 are not lost when journald is restarted. File descriptors
6566 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
6567 an extension to sd_notify(). Note that a limit is enforced
6568 on the number of fds a service can store in PID 1, and it
6569 defaults to 0, so that no fds may be stored, unless this is
6570 explicitly turned on.
6572 * The default TERM variable to use for units connected to a
6573 terminal, when no other value is explicitly is set is now
6574 vt220 rather than vt102. This should be fairly safe still,
6575 but allows PgUp/PgDn work.
6577 * The /etc/crypttab option header= as known from Debian is now
6580 * "loginctl user-status" and "loginctl session-status" will
6581 now show the last 10 lines of log messages of the
6582 user/session following the status output. Similar,
6583 "machinectl status" will show the last 10 log lines
6584 associated with a virtual machine or container
6585 service. (Note that this is usually not the log messages
6586 done in the VM/container itself, but simply what the
6587 container manager logs. For nspawn this includes all console
6590 * "loginctl session-status" without further argument will now
6591 show the status of the session of the caller. Similar,
6592 "lock-session", "unlock-session", "activate",
6593 "enable-linger", "disable-linger" may now be called without
6594 session/user parameter in which case they apply to the
6595 caller's session/user.
6597 * An X11 session scriptlet is now shipped that uploads
6598 $DISPLAY and $XAUTHORITY into the environment of the systemd
6599 --user daemon if a session begins. This should improve
6600 compatibility with X11 enabled applications run as systemd
6603 * Generators are now subject to masking via /etc and /run, the
6604 same way as unit files.
6606 * networkd .network files gained support for configuring
6607 per-link IPv4/IPv6 packet forwarding as well as IPv4
6608 masquerading. This is by default turned on for veth links to
6609 containers, as registered by systemd-nspawn. This means that
6610 nspawn containers run with --network-veth will now get
6611 automatic routed access to the host's networks without any
6612 further configuration or setup, as long as networkd runs on
6615 * systemd-nspawn gained the --port= (-p) switch to expose TCP
6616 or UDP posts of a container on the host. With this in place
6617 it is possible to run containers with private veth links
6618 (--network-veth), and have their functionality exposed on
6619 the host as if their services were running directly on the
6622 * systemd-nspawn's --network-veth switch now gained a short
6623 version "-n", since with the changes above it is now truly
6624 useful out-of-the-box. The systemd-nspawn@.service has been
6625 updated to make use of it too by default.
6627 * systemd-nspawn will now maintain a per-image R/W lock, to
6628 ensure that the same image is not started more than once
6629 writable. (It's OK to run an image multiple times
6630 simultaneously in read-only mode.)
6632 * systemd-nspawn's --image= option is now capable of
6633 dissecting and booting MBR and GPT disk images that contain
6634 only a single active Linux partition. Previously it
6635 supported only GPT disk images with proper GPT type
6636 IDs. This allows running cloud images from major
6637 distributions directly with systemd-nspawn, without
6640 * In addition to collecting mouse dpi data in the udev
6641 hardware database, there's now support for collecting angle
6642 information for mouse scroll wheels. The database is
6643 supposed to guarantee similar scrolling behavior on mice
6644 that it knows about. There's also support for collecting
6645 information about Touchpad types.
6647 * udev's input_id built-in will now also collect touch screen
6648 dimension data and attach it to probed devices.
6650 * /etc/os-release gained support for a Distribution Privacy
6653 * networkd gained support for creating "ipvlan", "gretap",
6654 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
6656 * systemd-tmpfiles gained support for "a" lines for setting
6659 * systemd-nspawn will now mount /tmp in the container to
6660 tmpfs, automatically.
6662 * systemd now exposes the memory.usage_in_bytes cgroup
6663 attribute and shows it for each service in the "systemctl
6664 status" output, if available.
6666 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
6667 immediate reboot is triggered. This useful if shutdown is
6668 hung and is unable to complete, to expedite the
6669 operation. Note that this kind of reboot will still unmount
6670 all file systems, and hence should not result in fsck being
6673 * A .device unit for an optical block device will now be
6674 considered active only when a medium is in the drive. Also,
6675 mount units are now bound to their backing devices thus
6676 triggering automatic unmounting when devices become
6677 unavailable. With this in place systemd will now
6678 automatically unmount left-over mounts when a CD-ROM is
6679 ejected or an USB stick is yanked from the system.
6681 * networkd-wait-online now has support for waiting for
6682 specific interfaces only (with globbing), and for giving up
6683 after a configurable timeout.
6685 * networkd now exits when idle. It will be automatically
6686 restarted as soon as interfaces show up, are removed or
6687 change state. networkd will stay around as long as there is
6688 at least one DHCP state machine or similar around, that keep
6691 * networkd may now configure IPv6 link-local addressing in
6692 addition to IPv4 link-local addressing.
6694 * The IPv6 "token" for use in SLAAC may now be configured for
6695 each .network interface in networkd.
6697 * Routes configured with networkd may now be assigned a scope
6700 * networkd's [Match] sections now support globbing and lists
6701 of multiple space-separated matches per item.
6703 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
6704 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
6705 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
6706 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
6707 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
6708 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
6709 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
6710 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
6711 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
6712 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
6713 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
6714 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
6715 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
6716 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
6717 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
6718 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
6719 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
6720 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
6721 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
6722 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
6723 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
6724 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
6725 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
6726 Hoffmann, Zbigniew Jędrzejewski-Szmek
6728 — Berlin, 2015-02-16
6732 * When querying unit file enablement status (for example via
6733 "systemctl is-enabled"), a new state "indirect" is now known
6734 which indicates that a unit might not be enabled itself, but
6735 another unit listed in its Also= setting might be.
6737 * Similar to the various existing ConditionXYZ= settings for
6738 units, there are now matching AssertXYZ= settings. While
6739 failing conditions cause a unit to be skipped, but its job
6740 to succeed, failing assertions declared like this will cause
6741 a unit start operation and its job to fail.
6743 * hostnamed now knows a new chassis type "embedded".
6745 * systemctl gained a new "edit" command. When used on a unit
6746 file, this allows extending unit files with .d/ drop-in
6747 configuration snippets or editing the full file (after
6748 copying it from /usr/lib to /etc). This will invoke the
6749 user's editor (as configured with $EDITOR), and reload the
6750 modified configuration after editing.
6752 * "systemctl status" now shows the suggested enablement state
6753 for a unit, as declared in the (usually vendor-supplied)
6754 system preset files.
6756 * nss-myhostname will now resolve the single-label hostname
6757 "gateway" to the locally configured default IP routing
6758 gateways, ordered by their metrics. This assigns a stable
6759 name to the used gateways, regardless which ones are
6760 currently configured. Note that the name will only be
6761 resolved after all other name sources (if nss-myhostname is
6762 configured properly) and should hence not negatively impact
6763 systems that use the single-label hostname "gateway" in
6766 * systemd-inhibit now allows filtering by mode when listing
6769 * Scope and service units gained a new "Delegate" boolean
6770 property, which, when set, allows processes running inside the
6771 unit to further partition resources. This is primarily
6772 useful for systemd user instances as well as container
6775 * journald will now pick up audit messages directly from
6776 the kernel, and log them like any other log message. The
6777 audit fields are split up and fully indexed. This means that
6778 journalctl in many ways is now a (nicer!) alternative to
6779 ausearch, the traditional audit client. Note that this
6780 implements only a minimal audit client. If you want the
6781 special audit modes like reboot-on-log-overflow, please use
6782 the traditional auditd instead, which can be used in
6783 parallel to journald.
6785 * The ConditionSecurity= unit file option now understands the
6786 special string "audit" to check whether auditing is
6789 * journalctl gained two new commands --vacuum-size= and
6790 --vacuum-time= to delete old journal files until the
6791 remaining ones take up no more than the specified size on disk,
6792 or are not older than the specified time.
6794 * A new, native PPPoE library has been added to sd-network,
6795 systemd's library of light-weight networking protocols. This
6796 library will be used in a future version of networkd to
6797 enable PPPoE communication without an external pppd daemon.
6799 * The busctl tool now understands a new "capture" verb that
6800 works similar to "monitor", but writes a packet capture
6801 trace to STDOUT that can be redirected to a file which is
6802 compatible with libcap's capture file format. This can then
6803 be loaded in Wireshark and similar tools to inspect bus
6806 * The busctl tool now understands a new "tree" verb that shows
6807 the object trees of a specific service on the bus, or of all
6810 * The busctl tool now understands a new "introspect" verb that
6811 shows all interfaces and members of objects on the bus,
6812 including their signature and values. This is particularly
6813 useful to get more information about bus objects shown by
6814 the new "busctl tree" command.
6816 * The busctl tool now understands new verbs "call",
6817 "set-property" and "get-property" for invoking bus method
6818 calls, setting and getting bus object properties in a
6821 * busctl gained a new --augment-creds= argument that controls
6822 whether the tool shall augment credential information it
6823 gets from the bus with data from /proc, in a possibly
6826 * nspawn's --link-journal= switch gained two new values
6827 "try-guest" and "try-host" that work like "guest" and
6828 "host", but do not fail if the host has no persistent
6829 journalling enabled. -j is now equivalent to
6830 --link-journal=try-guest.
6832 * macvlan network devices created by nspawn will now have
6833 stable MAC addresses.
6835 * A new SmackProcessLabel= unit setting has been added, which
6836 controls the SMACK security label processes forked off by
6837 the respective unit shall use.
6839 * If compiled with --enable-xkbcommon, systemd-localed will
6840 verify x11 keymap settings by compiling the given keymap. It
6841 will spew out warnings if the compilation fails. This
6842 requires libxkbcommon to be installed.
6844 * When a coredump is collected, a larger number of metadata
6845 fields is now collected and included in the journal records
6846 created for it. More specifically, control group membership,
6847 environment variables, memory maps, working directory,
6848 chroot directory, /proc/$PID/status, and a list of open file
6849 descriptors is now stored in the log entry.
6851 * The udev hwdb now contains DPI information for mice. For
6854 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
6856 * All systemd programs that read standalone configuration
6857 files in /etc now also support a corresponding series of
6858 .conf.d configuration directories in /etc/, /run/,
6859 /usr/local/lib/, /usr/lib/, and (if configured with
6860 --enable-split-usr) /lib/. In particular, the following
6861 configuration files now have corresponding configuration
6862 directories: system.conf user.conf, logind.conf,
6863 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
6864 resolved.conf, timesyncd.conf, journal-remote.conf, and
6865 journal-upload.conf. Note that distributions should use the
6866 configuration directories in /usr/lib/; the directories in
6867 /etc/ are reserved for the system administrator.
6869 * systemd-rfkill will no longer take the rfkill device name
6870 into account when storing rfkill state on disk, as the name
6871 might be dynamically assigned and not stable. Instead, the
6872 ID_PATH udev variable combined with the rfkill type (wlan,
6873 bluetooth, ...) is used.
6875 * A new service systemd-machine-id-commit.service has been
6876 added. When used on systems where /etc is read-only during
6877 boot, and /etc/machine-id is not initialized (but an empty
6878 file), this service will copy the temporary machine ID
6879 created as replacement into /etc after the system is fully
6880 booted up. This is useful for systems that are freshly
6881 installed with a non-initialized machine ID, but should get
6882 a fixed machine ID for subsequent boots.
6884 * networkd's .netdev files now provide a large set of
6885 configuration parameters for VXLAN devices. Similarly, the
6886 bridge port cost parameter is now configurable in .network
6887 files. There's also new support for configuring IP source
6888 routing. networkd .link files gained support for a new
6889 OriginalName= match that is useful to match against the
6890 original interface name the kernel assigned. .network files
6891 may include MTU= and MACAddress= fields for altering the MTU
6892 and MAC address while being connected to a specific network
6895 * The LUKS logic gained supported for configuring
6896 UUID-specific key files. There's also new support for naming
6897 LUKS device from the kernel command line, using the new
6898 luks.name= argument.
6900 * Timer units may now be transiently created via the bus API
6901 (this was previously already available for scope and service
6902 units). In addition it is now possible to create multiple
6903 transient units at the same time with a single bus call. The
6904 "systemd-run" tool has been updated to make use of this for
6905 running commands on a specified time, in at(1)-style.
6907 * tmpfiles gained support for "t" lines, for assigning
6908 extended attributes to files. Among other uses this may be
6909 used to assign SMACK labels to files.
6911 Contributions from: Alin Rauta, Alison Chaiken, Andrej
6912 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
6913 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
6914 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
6915 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
6916 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
6917 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
6918 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
6919 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
6920 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
6921 Michael Biebl, Michael Chapman, Michael Marineau, Michal
6922 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
6923 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
6924 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
6925 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
6926 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6927 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
6928 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
6930 — Berlin, 2014-12-10
6934 * journalctl gained the new options -t/--identifier= to match
6935 on the syslog identifier (aka "tag"), as well as --utc to
6936 show log timestamps in the UTC timezone. journalctl now also
6937 accepts -n/--lines=all to disable line capping in a pager.
6939 * journalctl gained a new switch, --flush, that synchronously
6940 flushes logs from /run/log/journal to /var/log/journal if
6941 persistent storage is enabled. systemd-journal-flush.service
6942 now waits until the operation is complete.
6944 * Services can notify the manager before they start a reload
6945 (by sending RELOADING=1) or shutdown (by sending
6946 STOPPING=1). This allows the manager to track and show the
6947 internal state of daemons and closes a race condition when
6948 the process is still running but has closed its D-Bus
6951 * Services with Type=oneshot do not have to have any ExecStart
6954 * User units are now loaded also from
6955 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
6956 /run/systemd/user directory that was already previously
6957 supported, but is under the control of the user.
6959 * Job timeouts (i.e. timeouts on the time a job that is
6960 queued stays in the run queue) can now optionally result in
6961 immediate reboot or power-off actions (JobTimeoutAction= and
6962 JobTimeoutRebootArgument=). This is useful on ".target"
6963 units, to limit the maximum time a target remains
6964 undispatched in the run queue, and to trigger an emergency
6965 operation in such a case. This is now used by default to
6966 turn off the system if boot-up (as defined by everything in
6967 basic.target) hangs and does not complete for at least
6968 15min. Also, if power-off or reboot hang for at least 30min
6969 an immediate power-off/reboot operation is triggered. This
6970 functionality is particularly useful to increase reliability
6971 on embedded devices, but also on laptops which might
6972 accidentally get powered on when carried in a backpack and
6973 whose boot stays stuck in a hard disk encryption passphrase
6976 * systemd-logind can be configured to also handle lid switch
6977 events even when the machine is docked or multiple displays
6978 are attached (HandleLidSwitchDocked= option).
6980 * A helper binary and a service have been added which can be
6981 used to resume from hibernation in the initramfs. A
6982 generator will parse the resume= option on the kernel
6983 command line to trigger resume.
6985 * A user console daemon systemd-consoled has been
6986 added. Currently, it is a preview, and will so far open a
6987 single terminal on each session of the user marked as
6988 Desktop=systemd-console.
6990 * Route metrics can be specified for DHCP routes added by
6993 * The SELinux context of socket-activated services can be set
6994 from the information provided by the networking stack
6995 (SELinuxContextFromNet= option).
6997 * Userspace firmware loading support has been removed and
6998 the minimum supported kernel version is thus bumped to 3.7.
7000 * Timeout for udev workers has been increased from 1 to 3
7001 minutes, but a warning will be printed after 1 minute to
7002 help diagnose kernel modules that take a long time to load.
7004 * Udev rules can now remove tags on devices with TAG-="foobar".
7006 * systemd's readahead implementation has been removed. In many
7007 circumstances it didn't give expected benefits even for
7008 rotational disk drives and was becoming less relevant in the
7009 age of SSDs. As none of the developers has been using
7010 rotating media anymore, and nobody stepped up to actively
7011 maintain this component of systemd it has now been removed.
7013 * Swap units can use Options= to specify discard options.
7014 Discard options specified for swaps in /etc/fstab are now
7017 * Docker containers are now detected as a separate type of
7020 * The Password Agent protocol gained support for queries where
7021 the user input is shown, useful e.g. for user names.
7022 systemd-ask-password gained a new --echo option to turn that
7025 * The default sysctl.d/ snippets will now set:
7027 net.core.default_qdisc = fq_codel
7029 This selects Fair Queuing Controlled Delay as the default
7030 queuing discipline for network interfaces. fq_codel helps
7031 fight the network bufferbloat problem. It is believed to be
7032 a good default with no tuning required for most workloads.
7033 Downstream distributions may override this choice. On 10Gbit
7034 servers that do not do forwarding, "fq" may perform better.
7035 Systems without a good clocksource should use "pfifo_fast".
7037 * If kdbus is enabled during build a new option BusPolicy= is
7038 available for service units, that allows locking all service
7039 processes into a stricter bus policy, in order to limit
7040 access to various bus services, or even hide most of them
7041 from the service's view entirely.
7043 * networkctl will now show the .network and .link file
7044 networkd has applied to a specific interface.
7046 * sd-login gained a new API call sd_session_get_desktop() to
7047 query which desktop environment has been selected for a
7050 * UNIX utmp support is now compile-time optional to support
7051 legacy-free systems.
7053 * systemctl gained two new commands "add-wants" and
7054 "add-requires" for pulling in units from specific targets
7057 * If the word "rescue" is specified on the kernel command line
7058 the system will now boot into rescue mode (aka
7059 rescue.target), which was previously available only by
7060 specifying "1" or "systemd.unit=rescue.target" on the kernel
7061 command line. This new kernel command line option nicely
7062 mirrors the already existing "emergency" kernel command line
7065 * New kernel command line options mount.usr=, mount.usrflags=,
7066 mount.usrfstype= have been added that match root=, rootflags=,
7067 rootfstype= but allow mounting a specific file system to
7070 * The $NOTIFY_SOCKET is now also passed to control processes of
7071 services, not only the main process.
7073 * This version reenables support for fsck's -l switch. This
7074 means at least version v2.25 of util-linux is required for
7075 operation, otherwise dead-locks on device nodes may
7076 occur. Again: you need to update util-linux to at least
7077 v2.25 when updating systemd to v217.
7079 * The "multi-seat-x" tool has been removed from systemd, as
7080 its functionality has been integrated into X servers 1.16,
7081 and the tool is hence redundant. It is recommended to update
7082 display managers invoking this tool to simply invoke X
7083 directly from now on, again.
7085 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
7086 message flag has been added for all of systemd's polkit
7087 authenticated method calls has been added. In particular this
7088 now allows optional interactive authorization via polkit for
7089 many of PID1's privileged operations such as unit file
7090 enabling and disabling.
7092 * "udevadm hwdb --update" learnt a new switch "--usr" for
7093 placing the rebuilt hardware database in /usr instead of
7094 /etc. When used only hardware database entries stored in
7095 /usr will be used, and any user database entries in /etc are
7096 ignored. This functionality is useful for vendors to ship a
7097 pre-built database on systems where local configuration is
7098 unnecessary or unlikely.
7100 * Calendar time specifications in .timer units now also
7101 understand the strings "semi-annually", "quarterly" and
7102 "minutely" as shortcuts (in addition to the preexisting
7103 "annually", "hourly", ...).
7105 * systemd-tmpfiles will now correctly create files in /dev
7106 at boot which are marked for creation only at boot. It is
7107 recommended to always create static device nodes with 'c!'
7108 and 'b!', so that they are created only at boot and not
7109 overwritten at runtime.
7111 * When the watchdog logic is used for a service (WatchdogSec=)
7112 and the watchdog timeout is hit the service will now be
7113 terminated with SIGABRT (instead of just SIGTERM), in order
7114 to make sure a proper coredump and backtrace is
7115 generated. This ensures that hanging services will result in
7116 similar coredump/backtrace behaviour as services that hit a
7119 Contributions from: Andreas Henriksson, Andrei Borzenkov,
7120 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
7121 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
7122 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
7123 Herrmann, David Sommerseth, David Strauss, Emil Renner
7124 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
7125 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
7126 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
7127 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
7128 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
7129 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
7130 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
7131 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
7132 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
7133 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
7134 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
7135 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
7136 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
7137 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
7138 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
7139 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
7142 — Berlin, 2014-10-28
7146 * timedated no longer reads NTP implementation unit names from
7147 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
7148 implementations should add a
7150 Conflicts=systemd-timesyncd.service
7152 to their unit files to take over and replace systemd's NTP
7153 default functionality.
7155 * systemd-sysusers gained a new line type "r" for configuring
7156 which UID/GID ranges to allocate system users/groups
7157 from. Lines of type "u" may now add an additional column
7158 that specifies the home directory for the system user to be
7159 created. Also, systemd-sysusers may now optionally read user
7160 information from STDIN instead of a file. This is useful for
7161 invoking it from RPM preinst scriptlets that need to create
7162 users before the first RPM file is installed since these
7163 files might need to be owned by them. A new
7164 %sysusers_create_inline RPM macro has been introduced to do
7165 just that. systemd-sysusers now updates the shadow files as
7166 well as the user/group databases, which should enhance
7167 compatibility with certain tools like grpck.
7169 * A number of bus APIs of PID 1 now optionally consult polkit to
7170 permit access for otherwise unprivileged clients under certain
7171 conditions. Note that this currently doesn't support
7172 interactive authentication yet, but this is expected to be
7173 added eventually, too.
7175 * /etc/machine-info now has new fields for configuring the
7176 deployment environment of the machine, as well as the
7177 location of the machine. hostnamectl has been updated with
7178 new command to update these fields.
7180 * systemd-timesyncd has been updated to automatically acquire
7181 NTP server information from systemd-networkd, which might
7182 have been discovered via DHCP.
7184 * systemd-resolved now includes a caching DNS stub resolver
7185 and a complete LLMNR name resolution implementation. A new
7186 NSS module "nss-resolve" has been added which can be used
7187 instead of glibc's own "nss-dns" to resolve hostnames via
7188 systemd-resolved. Hostnames, addresses and arbitrary RRs may
7189 be resolved via systemd-resolved D-Bus APIs. In contrast to
7190 the glibc internal resolver systemd-resolved is aware of
7191 multi-homed system, and keeps DNS server and caches separate
7192 and per-interface. Queries are sent simultaneously on all
7193 interfaces that have DNS servers configured, in order to
7194 properly handle VPNs and local LANs which might resolve
7195 separate sets of domain names. systemd-resolved may acquire
7196 DNS server information from systemd-networkd automatically,
7197 which in turn might have discovered them via DHCP. A tool
7198 "systemd-resolve-host" has been added that may be used to
7199 query the DNS logic in resolved. systemd-resolved implements
7200 IDNA and automatically uses IDNA or UTF-8 encoding depending
7201 on whether classic DNS or LLMNR is used as transport. In the
7202 next releases we intend to add a DNSSEC and mDNS/DNS-SD
7203 implementation to systemd-resolved.
7205 * A new NSS module nss-mymachines has been added, that
7206 automatically resolves the names of all local registered
7207 containers to their respective IP addresses.
7209 * A new client tool "networkctl" for systemd-networkd has been
7210 added. It currently is entirely passive and will query
7211 networking configuration from udev, rtnetlink and networkd,
7212 and present it to the user in a very friendly
7213 way. Eventually, we hope to extend it to become a full
7214 control utility for networkd.
7216 * .socket units gained a new DeferAcceptSec= setting that
7217 controls the kernels' TCP_DEFER_ACCEPT sockopt for
7218 TCP. Similarly, support for controlling TCP keep-alive
7219 settings has been added (KeepAliveTimeSec=,
7220 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
7221 turning off Nagle's algorithm on TCP has been added
7224 * logind learned a new session type "web", for use in projects
7225 like Cockpit which register web clients as PAM sessions.
7227 * timer units with at least one OnCalendar= setting will now
7228 be started only after time-sync.target has been
7229 reached. This way they will not elapse before the system
7230 clock has been corrected by a local NTP client or
7231 similar. This is particular useful on RTC-less embedded
7232 machines, that come up with an invalid system clock.
7234 * systemd-nspawn's --network-veth= switch should now result in
7235 stable MAC addresses for both the outer and the inner side
7238 * systemd-nspawn gained a new --volatile= switch for running
7239 container instances with /etc or /var unpopulated.
7241 * The kdbus client code has been updated to use the new Linux
7242 3.17 memfd subsystem instead of the old kdbus-specific one.
7244 * systemd-networkd's DHCP client and server now support
7245 FORCERENEW. There are also new configuration options to
7246 configure the vendor client identifier and broadcast mode
7249 * systemd will no longer inform the kernel about the current
7250 timezone, as this is necessarily incorrect and racy as the
7251 kernel has no understanding of DST and similar
7252 concepts. This hence means FAT timestamps will be always
7253 considered UTC, similar to what Android is already
7254 doing. Also, when the RTC is configured to the local time
7255 (rather than UTC) systemd will never synchronize back to it,
7256 as this might confuse Windows at a later boot.
7258 * systemd-analyze gained a new command "verify" for offline
7259 validation of unit files.
7261 * systemd-networkd gained support for a couple of additional
7262 settings for bonding networking setups. Also, the metric for
7263 statically configured routes may now be configured. For
7264 network interfaces where this is appropriate the peer IP
7265 address may now be configured.
7267 * systemd-networkd's DHCP client will no longer request
7268 broadcasting by default, as this tripped up some networks.
7269 For hardware where broadcast is required the feature should
7270 be switched back on using RequestBroadcast=yes.
7272 * systemd-networkd will now set up IPv4LL addresses (when
7273 enabled) even if DHCP is configured successfully.
7275 * udev will now default to respect network device names given
7276 by the kernel when the kernel indicates that these are
7277 predictable. This behavior can be tweaked by changing
7278 NamePolicy= in the relevant .link file.
7280 * A new library systemd-terminal has been added that
7281 implements full TTY stream parsing and rendering. This
7282 library is supposed to be used later on for implementing a
7283 full userspace VT subsystem, replacing the current kernel
7286 * A new tool systemd-journal-upload has been added to push
7287 journal data to a remote system running
7288 systemd-journal-remote.
7290 * journald will no longer forward all local data to another
7291 running syslog daemon. This change has been made because
7292 rsyslog (which appears to be the most commonly used syslog
7293 implementation these days) no longer makes use of this, and
7294 instead pulls the data out of the journal on its own. Since
7295 forwarding the messages to a non-existent syslog server is
7296 more expensive than we assumed we have now turned this
7297 off. If you run a syslog server that is not a recent rsyslog
7298 version, you have to turn this option on again
7299 (ForwardToSyslog= in journald.conf).
7301 * journald now optionally supports the LZ4 compressor for
7302 larger journal fields. This compressor should perform much
7303 better than XZ which was the previous default.
7305 * machinectl now shows the IP addresses of local containers,
7306 if it knows them, plus the interface name of the container.
7308 * A new tool "systemd-escape" has been added that makes it
7309 easy to escape strings to build unit names and similar.
7311 * sd_notify() messages may now include a new ERRNO= field
7312 which is parsed and collected by systemd and shown among the
7313 "systemctl status" output for a service.
7315 * A new component "systemd-firstboot" has been added that
7316 queries the most basic systemd information (timezone,
7317 hostname, root password) interactively on first
7318 boot. Alternatively it may also be used to provision these
7319 things offline on OS images installed into directories.
7321 * The default sysctl.d/ snippets will now set
7323 net.ipv4.conf.default.promote_secondaries=1
7325 This has the benefit of no flushing secondary IP addresses
7326 when primary addresses are removed.
7328 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
7329 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
7330 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
7331 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
7332 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
7333 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
7334 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
7335 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
7336 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
7337 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
7338 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
7339 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
7340 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
7341 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
7342 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
7344 — Berlin, 2014-08-19
7348 * A new tool systemd-sysusers has been added. This tool
7349 creates system users and groups in /etc/passwd and
7350 /etc/group, based on static declarative system user/group
7351 definitions in /usr/lib/sysusers.d/. This is useful to
7352 enable factory resets and volatile systems that boot up with
7353 an empty /etc directory, and thus need system users and
7354 groups created during early boot. systemd now also ships
7355 with two default sysusers.d/ files for the most basic
7356 users and groups systemd and the core operating system
7359 * A new tmpfiles snippet has been added that rebuilds the
7360 essential files in /etc on boot, should they be missing.
7362 * A directive for ensuring automatic clean-up of
7363 /var/cache/man/ has been removed from the default
7364 configuration. This line should now be shipped by the man
7365 implementation. The necessary change has been made to the
7366 man-db implementation. Note that you need to update your man
7367 implementation to one that ships this line, otherwise no
7368 automatic clean-up of /var/cache/man will take place.
7370 * A new condition ConditionNeedsUpdate= has been added that
7371 may conditionalize services to only run when /etc or /var
7372 are "older" than the vendor operating system resources in
7373 /usr. This is useful for reconstructing or updating /etc
7374 after an offline update of /usr or a factory reset, on the
7375 next reboot. Services that want to run once after such an
7376 update or reset should use this condition and order
7377 themselves before the new systemd-update-done.service, which
7378 will mark the two directories as fully updated. A number of
7379 service files have been added making use of this, to rebuild
7380 the udev hardware database, the journald message catalog and
7381 dynamic loader cache (ldconfig). The systemd-sysusers tool
7382 described above also makes use of this now. With this in
7383 place it is now possible to start up a minimal operating
7384 system with /etc empty cleanly. For more information on the
7385 concepts involved see this recent blog story:
7387 http://0pointer.de/blog/projects/stateless.html
7389 * A new system group "input" has been introduced, and all
7390 input device nodes get this group assigned. This is useful
7391 for system-level software to get access to input devices. It
7392 complements what is already done for "audio" and "video".
7394 * systemd-networkd learnt minimal DHCPv4 server support in
7395 addition to the existing DHCPv4 client support. It also
7396 learnt DHCPv6 client and IPv6 Router Solicitation client
7397 support. The DHCPv4 client gained support for static routes
7398 passed in from the server. Note that the [DHCPv4] section
7399 known in older systemd-networkd versions has been renamed to
7400 [DHCP] and is now also used by the DHCPv6 client. Existing
7401 .network files using settings of this section should be
7402 updated, though compatibility is maintained. Optionally, the
7403 client hostname may now be sent to the DHCP server.
7405 * networkd gained support for vxlan virtual networks as well
7406 as tun/tap and dummy devices.
7408 * networkd gained support for automatic allocation of address
7409 ranges for interfaces from a system-wide pool of
7410 addresses. This is useful for dynamically managing a large
7411 number of interfaces with a single network configuration
7412 file. In particular this is useful to easily assign
7413 appropriate IP addresses to the veth links of a large number
7414 of nspawn instances.
7416 * RPM macros for processing sysusers, sysctl and binfmt
7417 drop-in snippets at package installation time have been
7420 * The /etc/os-release file should now be placed in
7421 /usr/lib/os-release. The old location is automatically
7422 created as symlink. /usr/lib is the more appropriate
7423 location of this file, since it shall actually describe the
7424 vendor operating system shipped in /usr, and not the
7425 configuration stored in /etc.
7427 * .mount units gained a new boolean SloppyOptions= setting
7428 that maps to mount(8)'s -s option which enables permissive
7429 parsing of unknown mount options.
7431 * tmpfiles learnt a new "L+" directive which creates a symlink
7432 but (unlike "L") deletes a pre-existing file first, should
7433 it already exist and not already be the correct
7434 symlink. Similarly, "b+", "c+" and "p+" directives have been
7435 added as well, which create block and character devices, as
7436 well as fifos in the filesystem, possibly removing any
7437 pre-existing files of different types.
7439 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
7440 'argument' field (which so far specified the source to
7441 symlink/copy the files from) is now optional. If omitted the
7442 same file os copied from /usr/share/factory/ suffixed by the
7443 full destination path. This is useful for populating /etc
7444 with essential files, by copying them from vendor defaults
7445 shipped in /usr/share/factory/etc.
7447 * A new command "systemctl preset-all" has been added that
7448 applies the service preset settings to all installed unit
7449 files. A new switch --preset-mode= has been added that
7450 controls whether only enable or only disable operations
7453 * A new command "systemctl is-system-running" has been added
7454 that allows checking the overall state of the system, for
7455 example whether it is fully up and running.
7457 * When the system boots up with an empty /etc, the equivalent
7458 to "systemctl preset-all" is executed during early boot, to
7459 make sure all default services are enabled after a factory
7462 * systemd now contains a minimal preset file that enables the
7463 most basic services systemd ships by default.
7465 * Unit files' [Install] section gained a new DefaultInstance=
7466 field for defining the default instance to create if a
7467 template unit is enabled with no instance specified.
7469 * A new passive target cryptsetup-pre.target has been added
7470 that may be used by services that need to make they run and
7471 finish before the first LUKS cryptographic device is set up.
7473 * The /dev/loop-control and /dev/btrfs-control device nodes
7474 are now owned by the "disk" group by default, opening up
7475 access to this group.
7477 * systemd-coredump will now automatically generate a
7478 stack trace of all core dumps taking place on the system,
7479 based on elfutils' libdw library. This stack trace is logged
7482 * systemd-coredump may now optionally store coredumps directly
7483 on disk (in /var/lib/systemd/coredump, possibly compressed),
7484 instead of storing them unconditionally in the journal. This
7485 mode is the new default. A new configuration file
7486 /etc/systemd/coredump.conf has been added to configure this
7487 and other parameters of systemd-coredump.
7489 * coredumpctl gained a new "info" verb to show details about a
7490 specific coredump. A new switch "-1" has also been added
7491 that makes sure to only show information about the most
7492 recent entry instead of all entries. Also, as the tool is
7493 generally useful now the "systemd-" prefix of the binary
7494 name has been removed. Distributions that want to maintain
7495 compatibility with the old name should add a symlink from
7496 the old name to the new name.
7498 * journald's SplitMode= now defaults to "uid". This makes sure
7499 that unprivileged users can access their own coredumps with
7500 coredumpctl without restrictions.
7502 * New kernel command line options "systemd.wants=" (for
7503 pulling an additional unit during boot), "systemd.mask="
7504 (for masking a specific unit for the boot), and
7505 "systemd.debug-shell" (for enabling the debug shell on tty9)
7506 have been added. This is implemented in the new generator
7507 "systemd-debug-generator".
7509 * systemd-nspawn will now by default filter a couple of
7510 syscalls for containers, among them those required for
7511 kernel module loading, direct x86 IO port access, swap
7512 management, and kexec. Most importantly though
7513 open_by_handle_at() is now prohibited for containers,
7514 closing a hole similar to a recently discussed vulnerability
7515 in docker regarding access to files on file hierarchies the
7516 container should normally not have access to. Note that, for
7517 nspawn, we generally make no security claims anyway (and
7518 this is explicitly documented in the man page), so this is
7519 just a fix for one of the most obvious problems.
7521 * A new man page file-hierarchy(7) has been added that
7522 contains a minimized, modernized version of the file system
7523 layout systemd expects, similar in style to the FHS
7524 specification or hier(5). A new tool systemd-path(1) has
7525 been added to query many of these paths for the local
7528 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
7529 longer done. Since the directory now has a per-user size
7530 limit, and is cleaned on logout this appears unnecessary,
7531 in particular since this now brings the lifecycle of this
7532 directory closer in line with how IPC objects are handled.
7534 * systemd.pc now exports a number of additional directories,
7535 including $libdir (which is useful to identify the library
7536 path for the primary architecture of the system), and a
7537 couple of drop-in directories.
7539 * udev's predictable network interface names now use the dev_port
7540 sysfs attribute, introduced in linux 3.15 instead of dev_id to
7541 distinguish between ports of the same PCI function. dev_id should
7542 only be used for ports using the same HW address, hence the need
7545 * machined has been updated to export the OS version of a
7546 container (read from /etc/os-release and
7547 /usr/lib/os-release) on the bus. This is now shown in
7548 "machinectl status" for a machine.
7550 * A new service setting RestartForceExitStatus= has been
7551 added. If configured to a set of exit signals or process
7552 return values, the service will be restarted when the main
7553 daemon process exits with any of them, regardless of the
7556 * systemctl's -H switch for connecting to remote systemd
7557 machines has been extended so that it may be used to
7558 directly connect to a specific container on the
7559 host. "systemctl -H root@foobar:waldi" will now connect as
7560 user "root" to host "foobar", and then proceed directly to
7561 the container named "waldi". Note that currently you have to
7562 authenticate as user "root" for this to work, as entering
7563 containers is a privileged operation.
7565 Contributions from: Andreas Henriksson, Benjamin Steinwender,
7566 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
7567 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
7568 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
7569 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
7570 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
7571 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
7572 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
7573 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
7574 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
7575 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
7576 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
7578 — Berlin, 2014-07-03
7582 * As an experimental feature, udev now tries to lock the
7583 disk device node (flock(LOCK_SH|LOCK_NB)) while it
7584 executes events for the disk or any of its partitions.
7585 Applications like partitioning programs can lock the
7586 disk device node (flock(LOCK_EX)) and claim temporary
7587 device ownership that way; udev will entirely skip all event
7588 handling for this disk and its partitions. If the disk
7589 was opened for writing, the close will trigger a partition
7590 table rescan in udev's "watch" facility, and if needed
7591 synthesize "change" events for the disk and all its partitions.
7592 This is now unconditionally enabled, and if it turns out to
7593 cause major problems, we might turn it on only for specific
7594 devices, or might need to disable it entirely. Device Mapper
7595 devices are excluded from this logic.
7597 * We temporarily dropped the "-l" switch for fsck invocations,
7598 since they collide with the flock() logic above. util-linux
7599 upstream has been changed already to avoid this conflict,
7600 and we will re-add "-l" as soon as util-linux with this
7601 change has been released.
7603 * The dependency on libattr has been removed. Since a long
7604 time, the extended attribute calls have moved to glibc, and
7605 libattr is thus unnecessary.
7607 * Virtualization detection works without privileges now. This
7608 means the systemd-detect-virt binary no longer requires
7609 CAP_SYS_PTRACE file capabilities, and our daemons can run
7610 with fewer privileges.
7612 * systemd-networkd now runs under its own "systemd-network"
7613 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
7614 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
7615 loses the ability to write to files owned by root this way.
7617 * Similarly, systemd-resolved now runs under its own
7618 "systemd-resolve" user with no capabilities remaining.
7620 * Similarly, systemd-bus-proxyd now runs under its own
7621 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
7623 * systemd-networkd gained support for setting up "veth"
7624 virtual Ethernet devices for container connectivity, as well
7625 as GRE and VTI tunnels.
7627 * systemd-networkd will no longer automatically attempt to
7628 manually load kernel modules necessary for certain tunnel
7629 transports. Instead, it is assumed the kernel loads them
7630 automatically when required. This only works correctly on
7631 very new kernels. On older kernels, please consider adding
7632 the kernel modules to /etc/modules-load.d/ as a work-around.
7634 * The resolv.conf file systemd-resolved generates has been
7635 moved to /run/systemd/resolve/. If you have a symlink from
7636 /etc/resolv.conf, it might be necessary to correct it.
7638 * Two new service settings, ProtectHome= and ProtectSystem=,
7639 have been added. When enabled, they will make the user data
7640 (such as /home) inaccessible or read-only and the system
7641 (such as /usr) read-only, for specific services. This allows
7642 very light-weight per-service sandboxing to avoid
7643 modifications of user data or system files from
7644 services. These two new switches have been enabled for all
7645 of systemd's long-running services, where appropriate.
7647 * Socket units gained new SocketUser= and SocketGroup=
7648 settings to set the owner user and group of AF_UNIX sockets
7649 and FIFOs in the file system.
7651 * Socket units gained a new RemoveOnStop= setting. If enabled,
7652 all FIFOS and sockets in the file system will be removed
7653 when the specific socket unit is stopped.
7655 * Socket units gained a new Symlinks= setting. It takes a list
7656 of symlinks to create to file system sockets or FIFOs
7657 created by the specific Unix sockets. This is useful to
7658 manage symlinks to socket nodes with the same lifecycle as
7661 * The /dev/log socket and /dev/initctl FIFO have been moved to
7662 /run, and have been replaced by symlinks. This allows
7663 connecting to these facilities even if PrivateDevices=yes is
7664 used for a service (which makes /dev/log itself unavailable,
7665 but /run is left). This also has the benefit of ensuring
7666 that /dev only contains device nodes, directories and
7667 symlinks, and nothing else.
7669 * sd-daemon gained two new calls sd_pid_notify() and
7670 sd_pid_notifyf(). They are similar to sd_notify() and
7671 sd_notifyf(), but allow overriding of the source PID of
7672 notification messages if permissions permit this. This is
7673 useful to send notify messages on behalf of a different
7674 process (for example, the parent process). The
7675 systemd-notify tool has been updated to make use of this
7676 when sending messages (so that notification messages now
7677 originate from the shell script invoking systemd-notify and
7678 not the systemd-notify process itself. This should minimize
7679 a race where systemd fails to associate notification
7680 messages to services when the originating process already
7683 * A new "on-abnormal" setting for Restart= has been added. If
7684 set, it will result in automatic restarts on all "abnormal"
7685 reasons for a process to exit, which includes unclean
7686 signals, core dumps, timeouts and watchdog timeouts, but
7687 does not include clean and unclean exit codes or clean
7688 signals. Restart=on-abnormal is an alternative for
7689 Restart=on-failure for services that shall be able to
7690 terminate and avoid restarts on certain errors, by
7691 indicating so with an unclean exit code. Restart=on-failure
7692 or Restart=on-abnormal is now the recommended setting for
7693 all long-running services.
7695 * If the InaccessibleDirectories= service setting points to a
7696 mount point (or if there are any submounts contained within
7697 it), it is now attempted to completely unmount it, to make
7698 the file systems truly unavailable for the respective
7701 * The ReadOnlyDirectories= service setting and
7702 systemd-nspawn's --read-only parameter are now recursively
7703 applied to all submounts, too.
7705 * Mount units may now be created transiently via the bus APIs.
7707 * The support for SysV and LSB init scripts has been removed
7708 from the systemd daemon itself. Instead, it is now
7709 implemented as a generator that creates native systemd units
7710 from these scripts when needed. This enables us to remove a
7711 substantial amount of legacy code from PID 1, following the
7712 fact that many distributions only ship a very small number
7713 of LSB/SysV init scripts nowadays.
7715 * Privileged Xen (dom0) domains are not considered
7716 virtualization anymore by the virtualization detection
7717 logic. After all, they generally have unrestricted access to
7718 the hardware and usually are used to manage the unprivileged
7721 * systemd-tmpfiles gained a new "C" line type, for copying
7722 files or entire directories.
7724 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
7725 lines. So far, they have been non-globbing versions of the
7726 latter, and have thus been redundant. In future, it is
7727 recommended to only use "z". "m" has hence been removed
7728 from the documentation, even though it stays supported.
7730 * A tmpfiles snippet to recreate the most basic structure in
7731 /var has been added. This is enough to create the /var/run →
7732 /run symlink and create a couple of structural
7733 directories. This allows systems to boot up with an empty or
7734 volatile /var. Of course, while with this change, the core OS
7735 now is capable with dealing with a volatile /var, not all
7736 user services are ready for it. However, we hope that sooner
7737 or later, many service daemons will be changed upstream so
7738 that they are able to automatically create their necessary
7739 directories in /var at boot, should they be missing. This is
7740 the first step to allow state-less systems that only require
7741 the vendor image for /usr to boot.
7743 * systemd-nspawn has gained a new --tmpfs= switch to mount an
7744 empty tmpfs instance to a specific directory. This is
7745 particularly useful for making use of the automatic
7746 reconstruction of /var (see above), by passing --tmpfs=/var.
7748 * Access modes specified in tmpfiles snippets may now be
7749 prefixed with "~", which indicates that they shall be masked
7750 by whether the existing file or directory is currently
7751 writable, readable or executable at all. Also, if specified,
7752 the sgid/suid/sticky bits will be masked for all
7755 * A new passive target unit "network-pre.target" has been
7756 added which is useful for services that shall run before any
7757 network is configured, for example firewall scripts.
7759 * The "floppy" group that previously owned the /dev/fd*
7760 devices is no longer used. The "disk" group is now used
7761 instead. Distributions should probably deprecate usage of
7764 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
7765 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
7766 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
7767 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
7768 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
7769 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
7772 — Berlin, 2014-06-11
7776 * A new "systemd-timesyncd" daemon has been added for
7777 synchronizing the system clock across the network. It
7778 implements an SNTP client. In contrast to NTP
7779 implementations such as chrony or the NTP reference server,
7780 this only implements a client side, and does not bother with
7781 the full NTP complexity, focusing only on querying time from
7782 one remote server and synchronizing the local clock to
7783 it. Unless you intend to serve NTP to networked clients or
7784 want to connect to local hardware clocks, this simple NTP
7785 client should be more than appropriate for most
7786 installations. The daemon runs with minimal privileges, and
7787 has been hooked up with networkd to only operate when
7788 network connectivity is available. The daemon saves the
7789 current clock to disk every time a new NTP sync has been
7790 acquired, and uses this to possibly correct the system clock
7791 early at bootup, in order to accommodate for systems that
7792 lack an RTC such as the Raspberry Pi and embedded devices,
7793 and to make sure that time monotonically progresses on these
7794 systems, even if it is not always correct. To make use of
7795 this daemon, a new system user and group "systemd-timesync"
7796 needs to be created on installation of systemd.
7798 * The queue "seqnum" interface of libudev has been disabled, as
7799 it was generally incompatible with device namespacing as
7800 sequence numbers of devices go "missing" if the devices are
7801 part of a different namespace.
7803 * "systemctl list-timers" and "systemctl list-sockets" gained
7804 a --recursive switch for showing units of these types also
7805 for all local containers, similar in style to the already
7806 supported --recursive switch for "systemctl list-units".
7808 * A new RebootArgument= setting has been added for service
7809 units, which may be used to specify a kernel reboot argument
7810 to use when triggering reboots with StartLimitAction=.
7812 * A new FailureAction= setting has been added for service
7813 units which may be used to specify an operation to trigger
7814 when a service fails. This works similarly to
7815 StartLimitAction=, but unlike it, controls what is done
7816 immediately rather than only after several attempts to
7817 restart the service in question.
7819 * hostnamed got updated to also expose the kernel name,
7820 release, and version on the bus. This is useful for
7821 executing commands like hostnamectl with the -H switch.
7822 systemd-analyze makes use of this to properly display
7823 details when running non-locally.
7825 * The bootchart tool can now show cgroup information in the
7826 graphs it generates.
7828 * The CFS CPU quota cgroup attribute is now exposed for
7829 services. The new CPUQuota= switch has been added for this
7830 which takes a percentage value. Setting this will have the
7831 result that a service may never get more CPU time than the
7832 specified percentage, even if the machine is otherwise idle.
7834 * systemd-networkd learned IPIP and SIT tunnel support.
7836 * LSB init scripts exposing a dependency on $network will now
7837 get a dependency on network-online.target rather than simply
7838 network.target. This should bring LSB handling closer to
7839 what it was on SysV systems.
7841 * A new fsck.repair= kernel option has been added to control
7842 how fsck shall deal with unclean file systems at boot.
7844 * The (.ini) configuration file parser will now silently ignore
7845 sections whose names begin with "X-". This may be used to maintain
7846 application-specific extension sections in unit files.
7848 * machined gained a new API to query the IP addresses of
7849 registered containers. "machinectl status" has been updated
7850 to show these addresses in its output.
7852 * A new call sd_uid_get_display() has been added to the
7853 sd-login APIs for querying the "primary" session of a
7854 user. The "primary" session of the user is elected from the
7855 user's sessions and generally a graphical session is
7856 preferred over a text one.
7858 * A minimal systemd-resolved daemon has been added. It
7859 currently simply acts as a companion to systemd-networkd and
7860 manages resolv.conf based on per-interface DNS
7861 configuration, possibly supplied via DHCP. In the long run
7862 we hope to extend this into a local DNSSEC enabled DNS and
7865 * The systemd-networkd-wait-online tool is now enabled by
7866 default. It will delay network-online.target until a network
7867 connection has been configured. The tool primarily integrates
7868 with networkd, but will also make a best effort to make sense
7869 of network configuration performed in some other way.
7871 * Two new service options StartupCPUShares= and
7872 StartupBlockIOWeight= have been added that work similarly to
7873 CPUShares= and BlockIOWeight= however only apply during
7874 system startup. This is useful to prioritize certain services
7875 differently during bootup than during normal runtime.
7877 * hostnamed has been changed to prefer the statically
7878 configured hostname in /etc/hostname (unless set to
7879 'localhost' or empty) over any dynamic one supplied by
7880 dhcp. With this change, the rules for picking the hostname
7881 match more closely the rules of other configuration settings
7882 where the local administrator's configuration in /etc always
7883 overrides any other settings.
7885 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
7886 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
7887 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
7888 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
7889 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
7890 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
7891 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
7892 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
7893 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
7894 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
7895 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
7896 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
7897 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
7898 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
7899 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
7900 Lindskog, WaLyong Cho, Will Woods, Zbigniew
7903 — Beijing, 2014-05-28
7907 * When restoring the screen brightness at boot, stay away from
7908 the darkest setting or from the lowest 5% of the available
7909 range, depending on which is the larger value of both. This
7910 should effectively protect the user from rebooting into a
7911 black screen, should the brightness have been set to minimum
7914 * sd-login gained a new sd_machine_get_class() call to
7915 determine the class ("vm" or "container") of a machine
7916 registered with machined.
7918 * sd-login gained new calls
7919 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
7920 to query the identity of the peer of a local AF_UNIX
7921 connection. They operate similarly to their sd_pid_get_xyz()
7924 * PID 1 will now maintain a system-wide system state engine
7925 with the states "starting", "running", "degraded",
7926 "maintenance", "stopping". These states are bound to system
7927 startup, normal runtime, runtime with at least one failed
7928 service, rescue/emergency mode and system shutdown. This
7929 state is shown in the "systemctl status" output when no unit
7930 name is passed. It is useful to determine system state, in
7931 particularly when doing so for many systems or containers at
7934 * A new command "list-machines" has been added to "systemctl"
7935 that lists all local OS containers and shows their system
7936 state (see above), if systemd runs inside of them.
7938 * systemctl gained a new "-r" switch to recursively enumerate
7939 units on all local containers, when used with the
7940 "list-unit" command (which is the default one that is
7941 executed when no parameters are specified).
7943 * The GPT automatic partition discovery logic will now honour
7944 two GPT partition flags: one may be set on a partition to
7945 cause it to be mounted read-only, and the other may be set
7946 on a partition to ignore it during automatic discovery.
7948 * Two new GPT type UUIDs have been added for automatic root
7949 partition discovery, for 32-bit and 64-bit ARM. This is not
7950 particularly useful for discovering the root directory on
7951 these architectures during bare-metal boots (since UEFI is
7952 not common there), but still very useful to allow booting of
7953 ARM disk images in nspawn with the -i option.
7955 * MAC addresses of interfaces created with nspawn's
7956 --network-interface= switch will now be generated from the
7957 machine name, and thus be stable between multiple invocations
7960 * logind will now automatically remove all IPC objects owned
7961 by a user if she or he fully logs out. This makes sure that
7962 users who are logged out cannot continue to consume IPC
7963 resources. This covers SysV memory, semaphores and message
7964 queues as well as POSIX shared memory and message
7965 queues. Traditionally, SysV and POSIX IPC had no lifecycle
7966 limits. With this functionality, that is corrected. This may
7967 be turned off by using the RemoveIPC= switch of logind.conf.
7969 * The systemd-machine-id-setup and tmpfiles tools gained a
7970 --root= switch to operate on a specific root directory,
7973 * journald can now forward logged messages to the TTYs of all
7974 logged in users ("wall"). This is the default for all
7975 emergency messages now.
7977 * A new tool systemd-journal-remote has been added to stream
7978 journal log messages across the network.
7980 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
7981 controller trees are mounted into it. Note that the
7982 directories mounted beneath it are not read-only. This is a
7983 security measure and is particularly useful because glibc
7984 actually includes a search logic to pick any tmpfs it can
7985 find to implement shm_open() if /dev/shm is not available
7986 (which it might very well be in namespaced setups).
7988 * machinectl gained a new "poweroff" command to cleanly power
7989 down a local OS container.
7991 * The PrivateDevices= unit file setting will now also drop the
7992 CAP_MKNOD capability from the capability bound set, and
7993 imply DevicePolicy=closed.
7995 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
7996 comprehensively on all long-running systemd services where
7997 this is appropriate.
7999 * systemd-udevd will now run in a disassociated mount
8000 namespace. To mount directories from udev rules, make sure to
8001 pull in mount units via SYSTEMD_WANTS properties.
8003 * The kdbus support gained support for uploading policy into
8004 the kernel. sd-bus gained support for creating "monitoring"
8005 connections that can eavesdrop into all bus communication
8006 for debugging purposes.
8008 * Timestamps may now be specified in seconds since the UNIX
8009 epoch Jan 1st, 1970 by specifying "@" followed by the value
8012 * Native tcpwrap support in systemd has been removed. tcpwrap
8013 is old code, not really maintained anymore and has serious
8014 shortcomings, and better options such as firewalls
8015 exist. For setups that require tcpwrap usage, please
8016 consider invoking your socket-activated service via tcpd,
8017 like on traditional inetd.
8019 * A new system.conf configuration option
8020 DefaultTimerAccuracySec= has been added that controls the
8021 default AccuracySec= setting of .timer units.
8023 * Timer units gained a new WakeSystem= switch. If enabled,
8024 timers configured this way will cause the system to resume
8025 from system suspend (if the system supports that, which most
8028 * Timer units gained a new Persistent= switch. If enabled,
8029 timers configured this way will save to disk when they have
8030 been last triggered. This information is then used on next
8031 reboot to possible execute overdue timer events, that
8032 could not take place because the system was powered off.
8033 This enables simple anacron-like behaviour for timer units.
8035 * systemctl's "list-timers" will now also list the time a
8036 timer unit was last triggered in addition to the next time
8037 it will be triggered.
8039 * systemd-networkd will now assign predictable IPv4LL
8040 addresses to its local interfaces.
8042 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
8043 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
8044 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
8045 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
8046 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
8047 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
8048 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
8049 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
8052 — Berlin, 2014-03-25
8056 * A new unit file setting RestrictAddressFamilies= has been
8057 added to restrict which socket address families unit
8058 processes gain access to. This takes address family names
8059 like "AF_INET" or "AF_UNIX", and is useful to minimize the
8060 attack surface of services via exotic protocol stacks. This
8061 is built on seccomp system call filters.
8063 * Two new unit file settings RuntimeDirectory= and
8064 RuntimeDirectoryMode= have been added that may be used to
8065 manage a per-daemon runtime directories below /run. This is
8066 an alternative for setting up directory permissions with
8067 tmpfiles snippets, and has the advantage that the runtime
8068 directory's lifetime is bound to the daemon runtime and that
8069 the daemon starts up with an empty directory each time. This
8070 is particularly useful when writing services that drop
8071 privileges using the User= or Group= setting.
8073 * The DeviceAllow= unit setting now supports globbing for
8074 matching against device group names.
8076 * The systemd configuration file system.conf gained new
8077 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
8078 DefaultMemoryAccounting= to globally turn on/off accounting
8079 for specific resources (cgroups) for all units. These
8080 settings may still be overridden individually in each unit
8083 * systemd-gpt-auto-generator is now able to discover /srv and
8084 root partitions in addition to /home and swap partitions. It
8085 also supports LUKS-encrypted partitions now. With this in
8086 place, automatic discovery of partitions to mount following
8087 the Discoverable Partitions Specification
8088 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
8089 is now a lot more complete. This allows booting without
8090 /etc/fstab and without root= on the kernel command line on
8091 systems prepared appropriately.
8093 * systemd-nspawn gained a new --image= switch which allows
8094 booting up disk images and Linux installations on any block
8095 device that follow the Discoverable Partitions Specification
8096 (see above). This means that installations made with
8097 appropriately updated installers may now be started and
8098 deployed using container managers, completely
8099 unmodified. (We hope that libvirt-lxc will add support for
8100 this feature soon, too.)
8102 * systemd-nspawn gained a new --network-macvlan= setting to
8103 set up a private macvlan interface for the
8104 container. Similarly, systemd-networkd gained a new
8105 Kind=macvlan setting in .netdev files.
8107 * systemd-networkd now supports configuring local addresses
8110 * A new tool systemd-network-wait-online has been added to
8111 synchronously wait for network connectivity using
8114 * The sd-bus.h bus API gained a new sd_bus_track object for
8115 tracking the lifecycle of bus peers. Note that sd-bus.h is
8116 still not a public API though (unless you specify
8117 --enable-kdbus on the configure command line, which however
8118 voids your warranty and you get no API stability guarantee).
8120 * The $XDG_RUNTIME_DIR runtime directories for each user are
8121 now individual tmpfs instances, which has the benefit of
8122 introducing separate pools for each user, with individual
8123 size limits, and thus making sure that unprivileged clients
8124 can no longer negatively impact the system or other users by
8125 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
8126 RuntimeDirectorySize= has been introduced that allows
8127 controlling the default size limit for all users. It
8128 defaults to 10% of the available physical memory. This is no
8129 replacement for quotas on tmpfs though (which the kernel
8130 still does not support), as /dev/shm and /tmp are still
8131 shared resources used by both the system and unprivileged
8134 * logind will now automatically turn off automatic suspending
8135 on laptop lid close when more than one display is
8136 connected. This was previously expected to be implemented
8137 individually in desktop environments (such as GNOME),
8138 however has been added to logind now, in order to fix a
8139 boot-time race where a desktop environment might not have
8140 been started yet and thus not been able to take an inhibitor
8141 lock at the time where logind already suspends the system
8142 due to a closed lid.
8144 * logind will now wait at least 30s after each system
8145 suspend/resume cycle, and 3min after system boot before
8146 suspending the system due to a closed laptop lid. This
8147 should give USB docking stations and similar enough time to
8148 be probed and configured after system resume and boot in
8149 order to then act as suspend blocker.
8151 * systemd-run gained a new --property= setting which allows
8152 initialization of resource control properties (and others)
8153 for the created scope or service unit. Example: "systemd-run
8154 --property=BlockIOWeight=10 updatedb" may be used to run
8155 updatedb at a low block IO scheduling weight.
8157 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
8158 now also work in --scope mode.
8160 * When systemd is compiled with kdbus support, basic support
8161 for enforced policies is now in place. (Note that enabling
8162 kdbus still voids your warranty and no API compatibility
8165 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
8166 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
8167 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
8168 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
8169 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
8170 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
8171 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
8172 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
8173 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
8174 Zbigniew Jędrzejewski-Szmek
8176 — Berlin, 2014-03-12
8180 * systemd will now relabel /dev after loading the SMACK policy
8181 according to SMACK rules.
8183 * A new unit file option AppArmorProfile= has been added to
8184 set the AppArmor profile for the processes of a unit.
8186 * A new condition check ConditionArchitecture= has been added
8187 to conditionalize units based on the system architecture, as
8188 reported by uname()'s "machine" field.
8190 * systemd-networkd now supports matching on the system
8191 virtualization, architecture, kernel command line, hostname
8194 * logind is now a lot more aggressive when suspending the
8195 machine due to a closed laptop lid. Instead of acting only
8196 on the lid close action, it will continuously watch the lid
8197 status and act on it. This is useful for laptops where the
8198 power button is on the outside of the chassis so that it can
8199 be reached without opening the lid (such as the Lenovo
8200 Yoga). On those machines, logind will now immediately
8201 re-suspend the machine if the power button has been
8202 accidentally pressed while the laptop was suspended and in a
8203 backpack or similar.
8205 * logind will now watch SW_DOCK switches and inhibit reaction
8206 to the lid switch if it is pressed. This means that logind
8207 will not suspend the machine anymore if the lid is closed
8208 and the system is docked, if the laptop supports SW_DOCK
8209 notifications via the input layer. Note that ACPI docking
8210 stations do not generate this currently. Also note that this
8211 logic is usually not fully sufficient and Desktop
8212 Environments should take a lid switch inhibitor lock when an
8213 external display is connected, as systemd will not watch
8216 * nspawn will now make use of the devices cgroup controller by
8217 default, and only permit creation of and access to the usual
8218 API device nodes like /dev/null or /dev/random, as well as
8219 access to (but not creation of) the pty devices.
8221 * We will now ship a default .network file for
8222 systemd-networkd that automatically configures DHCP for
8223 network interfaces created by nspawn's --network-veth or
8224 --network-bridge= switches.
8226 * systemd will now understand the usual M, K, G, T suffixes
8227 according to SI conventions (i.e. to the base 1000) when
8228 referring to throughput and hardware metrics. It will stay
8229 with IEC conventions (i.e. to the base 1024) for software
8230 metrics, according to what is customary according to
8231 Wikipedia. We explicitly document which base applies for
8232 each configuration option.
8234 * The DeviceAllow= setting in unit files now supports a syntax to
8235 allow-list an entire group of devices node majors at once, based on
8236 the /proc/devices listing. For example, with the string "char-pts",
8237 it is now possible to allow-list all current and future pseudo-TTYs
8240 * sd-event learned a new "post" event source. Event sources of
8241 this type are triggered by the dispatching of any event
8242 source of a type that is not "post". This is useful for
8243 implementing clean-up and check event sources that are
8244 triggered by other work being done in the program.
8246 * systemd-networkd is no longer statically enabled, but uses
8247 the usual [Install] sections so that it can be
8248 enabled/disabled using systemctl. It still is enabled by
8251 * When creating a veth interface pair with systemd-nspawn, the
8252 host side will now be prefixed with "vb-" if
8253 --network-bridge= is used, and with "ve-" if --network-veth
8254 is used. This way, it is easy to distinguish these cases on
8255 the host, for example to apply different configuration to
8256 them with systemd-networkd.
8258 * The compatibility libraries for libsystemd-journal.so,
8259 libsystem-id128.so, libsystemd-login.so and
8260 libsystemd-daemon.so do not make use of IFUNC
8261 anymore. Instead, we now build libsystemd.so multiple times
8262 under these alternative names. This means that the footprint
8263 is drastically increased, but given that these are
8264 transitional compatibility libraries, this should not matter
8265 much. This change has been made necessary to support the ARM
8266 platform for these compatibility libraries, as the ARM
8267 toolchain is not really at the same level as the toolchain
8268 for other architectures like x86 and does not support
8269 IFUNC. Please make sure to use --enable-compat-libs only
8270 during a transitional period!
8272 * The .include syntax has been deprecated and is not documented
8273 anymore. Drop-in files in .d directories should be used instead.
8275 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
8276 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
8277 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
8278 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
8279 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
8280 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
8281 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
8282 Zbigniew Jędrzejewski-Szmek
8284 — Berlin, 2014-02-24
8288 * A new component "systemd-networkd" has been added that can
8289 be used to configure local network interfaces statically or
8290 via DHCP. It is capable of bringing up bridges, VLANs, and
8291 bonding. Currently, no hook-ups for interactive network
8292 configuration are provided. Use this for your initrd,
8293 container, embedded, or server setup if you need a simple,
8294 yet powerful, network configuration solution. This
8295 configuration subsystem is quite nifty, as it allows wildcard
8296 hotplug matching in interfaces. For example, with a single
8297 configuration snippet, you can configure that all Ethernet
8298 interfaces showing up are automatically added to a bridge,
8299 or similar. It supports link-sensing and more.
8301 * A new tool "systemd-socket-proxyd" has been added which can
8302 act as a bidirectional proxy for TCP sockets. This is
8303 useful for adding socket activation support to services that
8304 do not actually support socket activation, including virtual
8305 machines and the like.
8307 * Add a new tool to save/restore rfkill state on
8310 * Save/restore state of keyboard backlights in addition to
8311 display backlights on shutdown/boot.
8313 * udev learned a new SECLABEL{} construct to label device
8314 nodes with a specific security label when they appear. For
8315 now, only SECLABEL{selinux} is supported, but the syntax is
8316 prepared for additional security frameworks.
8318 * udev gained a new scheme to configure link-level attributes
8319 from files in /etc/systemd/network/*.link. These files can
8320 match against MAC address, device path, driver name and type,
8321 and will apply attributes like the naming policy, link speed,
8322 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
8323 address assignment policy (randomized, ...).
8325 * The configuration of network interface naming rules for
8326 "permanent interface names" has changed: a new NamePolicy=
8327 setting in the [Link] section of .link files determines the
8328 priority of possible naming schemes (onboard, slot, MAC,
8329 path). The default value of this setting is determined by
8330 /usr/lib/net/links/99-default.link. Old
8331 80-net-name-slot.rules udev configuration file has been
8332 removed, so local configuration overriding this file should
8333 be adapted to override 99-default.link instead.
8335 * When the User= switch is used in a unit file, also
8336 initialize $SHELL= based on the user database entry.
8338 * systemd no longer depends on libdbus. All communication is
8339 now done with sd-bus, systemd's low-level bus library
8342 * kdbus support has been added to PID 1 itself. When kdbus is
8343 enabled, this causes PID 1 to set up the system bus and
8344 enable support for a new ".busname" unit type that
8345 encapsulates bus name activation on kdbus. It works a little
8346 bit like ".socket" units, except for bus names. A new
8347 generator has been added that converts classic dbus1 service
8348 activation files automatically into native systemd .busname
8351 * sd-bus: add a light-weight vtable implementation that allows
8352 defining objects on the bus with a simple static const
8353 vtable array of its methods, signals and properties.
8355 * systemd will not generate or install static dbus
8356 introspection data anymore to /usr/share/dbus-1/interfaces,
8357 as the precise format of these files is unclear, and
8358 nothing makes use of it.
8360 * A proxy daemon is now provided to proxy clients connecting
8361 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
8362 compatibility with classic D-Bus.
8364 * A bus driver implementation has been added that supports the
8365 classic D-Bus bus driver calls on kdbus, also for
8366 compatibility purposes.
8368 * A new API "sd-event.h" has been added that implements a
8369 minimal event loop API built around epoll. It provides a
8370 couple of features that direct epoll usage is lacking:
8371 prioritization of events, scales to large numbers of timer
8372 events, per-event timer slack (accuracy), system-wide
8373 coalescing of timer events, exit handlers, watchdog
8374 supervision support using systemd's sd_notify() API, child
8377 * A new API "sd-rntl.h" has been added that provides an API
8378 around the route netlink interface of the kernel, similar in
8379 style to "sd-bus.h".
8381 * A new API "sd-dhcp-client.h" has been added that provides a
8382 small DHCPv4 client-side implementation. This is used by
8385 * There is a new kernel command line option
8386 "systemd.restore_state=0|1". When set to "0", none of the
8387 systemd tools will restore saved runtime state to hardware
8388 devices. More specifically, the rfkill and backlight states
8391 * The FsckPassNo= compatibility option in mount/service units
8392 has been removed. The fstab generator will now add the
8393 necessary dependencies automatically, and does not require
8394 PID1's support for that anymore.
8396 * journalctl gained a new switch, --list-boots, that lists
8397 recent boots with their times and boot IDs.
8399 * The various tools like systemctl, loginctl, timedatectl,
8400 busctl, systemd-run, ... have gained a new switch "-M" to
8401 connect to a specific, local OS container (as direct
8402 connection, without requiring SSH). This works on any
8403 container that is registered with machined, such as those
8404 created by libvirt-lxc or nspawn.
8406 * systemd-run and systemd-analyze also gained support for "-H"
8407 to connect to remote hosts via SSH. This is particularly
8408 useful for systemd-run because it enables queuing of jobs
8409 onto remote systems.
8411 * machinectl gained a new command "login" to open a getty
8412 login in any local container. This works with any container
8413 that is registered with machined (such as those created by
8414 libvirt-lxc or nspawn), and which runs systemd inside.
8416 * machinectl gained a new "reboot" command that may be used to
8417 trigger a reboot on a specific container that is registered
8418 with machined. This works on any container that runs an init
8419 system of some kind.
8421 * systemctl gained a new "list-timers" command to print a nice
8422 listing of installed timer units with the times they elapse
8425 * Alternative reboot() parameters may now be specified on the
8426 "systemctl reboot" command line and are passed to the
8427 reboot() system call.
8429 * systemctl gained a new --job-mode= switch to configure the
8430 mode to queue a job with. This is a more generic version of
8431 --fail, --irreversible, and --ignore-dependencies, which are
8432 still available but not advertised anymore.
8434 * /etc/systemd/system.conf gained new settings to configure
8435 various default timeouts of units, as well as the default
8436 start limit interval and burst. These may still be overridden
8439 * PID1 will now export on the bus profile data of the security
8440 policy upload process (such as the SELinux policy upload to
8443 * journald: when forwarding logs to the console, include
8444 timestamps (following the setting in
8445 /sys/module/printk/parameters/time).
8447 * OnCalendar= in timer units now understands the special
8448 strings "yearly" and "annually". (Both are equivalent)
8450 * The accuracy of timer units is now configurable with the new
8451 AccuracySec= setting. It defaults to 1min.
8453 * A new dependency type JoinsNamespaceOf= has been added that
8454 allows running two services within the same /tmp and network
8455 namespace, if PrivateNetwork= or PrivateTmp= are used.
8457 * A new command "cat" has been added to systemctl. It outputs
8458 the original unit file of a unit, and concatenates the
8459 contents of additional "drop-in" unit file snippets, so that
8460 the full configuration is shown.
8462 * systemctl now supports globbing on the various "list-xyz"
8463 commands, like "list-units" or "list-sockets", as well as on
8464 those commands which take multiple unit names.
8466 * journalctl's --unit= switch gained support for globbing.
8468 * All systemd daemons now make use of the watchdog logic so
8469 that systemd automatically notices when they hang.
8471 * If the $container_ttys environment variable is set,
8472 getty-generator will automatically spawn a getty for each
8473 listed tty. This is useful for container managers to request
8474 login gettys to be spawned on as many ttys as needed.
8476 * %h, %s, %U specifier support is not available anymore when
8477 used in unit files for PID 1. This is because NSS calls are
8478 not safe from PID 1. They stay available for --user
8479 instances of systemd, and as special case for the root user.
8481 * loginctl gained a new "--no-legend" switch to turn off output
8484 * The "sd-login.h" API gained three new calls:
8485 sd_session_is_remote(), sd_session_get_remote_user(),
8486 sd_session_get_remote_host() to query information about
8489 * The udev hardware database now also carries vendor/product
8490 information of SDIO devices.
8492 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
8493 determine whether watchdog notifications are requested by
8496 * Socket-activated per-connection services now include a
8497 short description of the connection parameters in the
8500 * tmpfiles gained a new "--boot" option. When this is not used,
8501 only lines where the command character is not suffixed with
8502 "!" are executed. When this option is specified, those
8503 options are executed too. This partitions tmpfiles
8504 directives into those that can be safely executed at any
8505 time, and those which should be run only at boot (for
8506 example, a line that creates /run/nologin).
8508 * A new API "sd-resolve.h" has been added which provides a simple
8509 asynchronous wrapper around glibc NSS hostname resolution
8510 calls, such as getaddrinfo(). In contrast to glibc's
8511 getaddrinfo_a(), it does not use signals. In contrast to most
8512 other asynchronous name resolution libraries, this one does
8513 not reimplement DNS, but reuses NSS, so that alternate
8514 hostname resolution systems continue to work, such as mDNS,
8515 LDAP, etc. This API is based on libasyncns, but it has been
8516 cleaned up for inclusion in systemd.
8518 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
8519 "sd-daemon.h" are no longer found in individual libraries
8520 libsystemd-journal.so, libsystemd-login.so,
8521 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
8522 merged them into a single library, libsystemd.so, which
8523 provides all symbols. The reason for this is cyclic
8524 dependencies, as these libraries tend to use each other's
8525 symbols. So far, we have managed to workaround that by linking
8526 a copy of a good part of our code into each of these
8527 libraries again and again, which, however, makes certain
8528 things hard to do, like sharing static variables. Also, it
8529 substantially increases footprint. With this change, there
8530 is only one library for the basic APIs systemd
8531 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
8532 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
8533 library as well, however are subject to the --enable-kdbus
8534 switch (see below). Note that "sd-dhcp-client.h" is not part
8535 of this library (this is because it only consumes, never
8536 provides, services of/to other APIs). To make the transition
8537 easy from the separate libraries to the unified one, we
8538 provide the --enable-compat-libs compile-time switch which
8539 will generate stub libraries that are compatible with the
8540 old ones but redirect all calls to the new one.
8542 * All of the kdbus logic and the new APIs "sd-bus.h",
8543 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
8544 and "sd-utf8.h" are compile-time optional via the
8545 "--enable-kdbus" switch, and they are not compiled in by
8546 default. To make use of kdbus, you have to explicitly enable
8547 the switch. Note however, that neither the kernel nor the
8548 userspace API for all of this is considered stable yet. We
8549 want to maintain the freedom to still change the APIs for
8550 now. By specifying this build-time switch, you acknowledge
8551 that you are aware of the instability of the current
8554 * Also, note that while kdbus is pretty much complete,
8555 it lacks one thing: proper policy support. This means you
8556 can build a fully working system with all features; however,
8557 it will be highly insecure. Policy support will be added in
8558 one of the next releases, at the same time that we will
8559 declare the APIs stable.
8561 * When the kernel command line argument "kdbus" is specified,
8562 systemd will automatically load the kdbus.ko kernel module. At
8563 this stage of development, it is only useful for testing kdbus
8564 and should not be used in production. Note: if "--enable-kdbus"
8565 is specified, and the kdbus.ko kernel module is available, and
8566 "kdbus" is added to the kernel command line, the entire system
8567 runs with kdbus instead of dbus-daemon, with the above mentioned
8568 problem of missing the system policy enforcement. Also a future
8569 version of kdbus.ko or a newer systemd will not be compatible with
8570 each other, and will unlikely be able to boot the machine if only
8571 one of them is updated.
8573 * systemctl gained a new "import-environment" command which
8574 uploads the caller's environment (or parts thereof) into the
8575 service manager so that it is inherited by services started
8576 by the manager. This is useful to upload variables like
8577 $DISPLAY into the user service manager.
8579 * A new PrivateDevices= switch has been added to service units
8580 which allows running a service with a namespaced /dev
8581 directory that does not contain any device nodes for
8582 physical devices. More specifically, it only includes devices
8583 such as /dev/null, /dev/urandom, and /dev/zero which are API
8586 * logind has been extended to support behaviour like VT
8587 switching on seats that do not support a VT. This makes
8588 multi-session available on seats that are not the first seat
8589 (seat0), and on systems where kernel support for VTs has
8590 been disabled at compile-time.
8592 * If a process holds a delay lock for system sleep or shutdown
8593 and fails to release it in time, we will now log its
8594 identity. This makes it easier to identify processes that
8595 cause slow suspends or power-offs.
8597 * When parsing /etc/crypttab, support for a new key-slot=
8598 option as supported by Debian is added. It allows indicating
8599 which LUKS slot to use on disk, speeding up key loading.
8601 * The sd_journal_sendv() API call has been checked and
8602 officially declared to be async-signal-safe so that it may
8603 be invoked from signal handlers for logging purposes.
8605 * Boot-time status output is now enabled automatically after a
8606 short timeout if boot does not progress, in order to give
8607 the user an indication what she or he is waiting for.
8609 * The boot-time output has been improved to show how much time
8610 remains until jobs expire.
8612 * The KillMode= switch in service units gained a new possible
8613 value "mixed". If set, and the unit is shut down, then the
8614 initial SIGTERM signal is sent only to the main daemon
8615 process, while the following SIGKILL signal is sent to
8616 all remaining processes of the service.
8618 * When a scope unit is registered, a new property "Controller"
8619 may be set. If set to a valid bus name, systemd will send a
8620 RequestStop() signal to this name when it would like to shut
8621 down the scope. This may be used to hook manager logic into
8622 the shutdown logic of scope units. Also, scope units may now
8623 be put in a special "abandoned" state, in which case the
8624 manager process which created them takes no further
8625 responsibilities for it.
8627 * When reading unit files, systemd will now verify
8628 the access mode of these files, and warn about certain
8629 suspicious combinations. This has been added to make it
8630 easier to track down packaging bugs where unit files are
8631 marked executable or world-writable.
8633 * systemd-nspawn gained a new "--setenv=" switch to set
8634 container-wide environment variables. The similar option in
8635 systemd-activate was renamed from "--environment=" to
8636 "--setenv=" for consistency.
8638 * systemd-nspawn has been updated to create a new kdbus domain
8639 for each container that is invoked, thus allowing each
8640 container to have its own set of system and user buses,
8641 independent of the host.
8643 * systemd-nspawn gained a new --drop-capability= switch to run
8644 the container with less capabilities than the default. Both
8645 --drop-capability= and --capability= now take the special
8646 string "all" for dropping or keeping all capabilities.
8648 * systemd-nspawn gained new switches for executing containers
8649 with specific SELinux labels set.
8651 * systemd-nspawn gained a new --quiet switch to not generate
8652 any additional output but the container's own console
8655 * systemd-nspawn gained a new --share-system switch to run a
8656 container without PID namespacing enabled.
8658 * systemd-nspawn gained a new --register= switch to control
8659 whether the container is registered with systemd-machined or
8660 not. This is useful for containers that do not run full
8661 OS images, but only specific apps.
8663 * systemd-nspawn gained a new --keep-unit which may be used
8664 when invoked as the only program from a service unit, and
8665 results in registration of the unit service itself in
8666 systemd-machined, instead of a newly opened scope unit.
8668 * systemd-nspawn gained a new --network-interface= switch for
8669 moving arbitrary interfaces to the container. The new
8670 --network-veth switch creates a virtual Ethernet connection
8671 between host and container. The new --network-bridge=
8672 switch then allows assigning the host side of this virtual
8673 Ethernet connection to a bridge device.
8675 * systemd-nspawn gained a new --personality= switch for
8676 setting the kernel personality for the container. This is
8677 useful when running a 32-bit container on a 64-bit host. A
8678 similar option Personality= is now also available for service
8681 * logind will now also track a "Desktop" identifier for each
8682 session which encodes the desktop environment of it. This is
8683 useful for desktop environments that want to identify
8684 multiple running sessions of itself easily.
8686 * A new SELinuxContext= setting for service units has been
8687 added that allows setting a specific SELinux execution
8688 context for a service.
8690 * Most systemd client tools will now honour $SYSTEMD_LESS for
8691 settings of the "less" pager. By default, these tools will
8692 override $LESS to allow certain operations to work, such as
8693 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
8694 influence this logic.
8696 * systemd's "seccomp" hook-up has been changed to make use of
8697 the libseccomp library instead of using its own
8698 implementation. This has benefits for portability among
8701 * For usage together with SystemCallFilter=, a new
8702 SystemCallErrorNumber= setting has been introduced that
8703 allows configuration of a system error number to be returned
8704 on filtered system calls, instead of immediately killing the
8705 process. Also, SystemCallArchitectures= has been added to
8706 limit access to system calls of a particular architecture
8707 (in order to turn off support for unused secondary
8708 architectures). There is also a global
8709 SystemCallArchitectures= setting in system.conf now to turn
8710 off support for non-native system calls system-wide.
8712 * systemd requires a kernel with a working name_to_handle_at(),
8713 please see the kernel config requirements in the README file.
8715 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
8716 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
8717 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
8718 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
8719 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
8720 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
8721 Elia Pinto, Florian Weimer, George McCollister, Goffredo
8722 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
8723 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
8724 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
8725 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
8726 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
8727 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
8728 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
8729 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
8730 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
8731 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
8732 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
8733 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
8734 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
8735 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
8736 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
8737 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
8738 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
8740 — Berlin, 2014-02-20
8744 * logind has gained support for facilitating privileged input
8745 and drm device access for unprivileged clients. This work is
8746 useful to allow Wayland display servers (and similar
8747 programs, such as kmscon) to run under the user's ID and
8748 access input and drm devices which are normally
8749 protected. When this is used (and the kernel is new enough)
8750 logind will "mute" IO on the file descriptors passed to
8751 Wayland as long as it is in the background and "unmute" it
8752 if it returns into the foreground. This allows secure
8753 session switching without allowing background sessions to
8754 eavesdrop on input and display data. This also introduces
8755 session switching support if VT support is turned off in the
8756 kernel, and on seats that are not seat0.
8758 * A new kernel command line option luks.options= is understood
8759 now which allows specifying LUKS options for usage for LUKS
8760 encrypted partitions specified with luks.uuid=.
8762 * tmpfiles.d(5) snippets may now use specifier expansion in
8763 path names. More specifically %m, %b, %H, %v, are now
8764 replaced by the local machine id, boot id, hostname, and
8765 kernel version number.
8767 * A new tmpfiles.d(5) command "m" has been introduced which
8768 may be used to change the owner/group/access mode of a file
8769 or directory if it exists, but do nothing if it does not.
8771 * This release removes high-level support for the
8772 MemorySoftLimit= cgroup setting. The underlying kernel
8773 cgroup attribute memory.soft_limit= is currently badly
8774 designed and likely to be removed from the kernel API in its
8775 current form, hence we should not expose it for now.
8777 * The memory.use_hierarchy cgroup attribute is now enabled for
8778 all cgroups systemd creates in the memory cgroup
8779 hierarchy. This option is likely to be come the built-in
8780 default in the kernel anyway, and the non-hierarchical mode
8781 never made much sense in the intrinsically hierarchical
8784 * A new field _SYSTEMD_SLICE= is logged along with all journal
8785 messages containing the slice a message was generated
8786 from. This is useful to allow easy per-customer filtering of
8787 logs among other things.
8789 * systemd-journald will no longer adjust the group of journal
8790 files it creates to the "systemd-journal" group. Instead we
8791 rely on the journal directory to be owned by the
8792 "systemd-journal" group, and its setgid bit set, so that the
8793 kernel file system layer will automatically enforce that
8794 journal files inherit this group assignment. The reason for
8795 this change is that we cannot allow NSS look-ups from
8796 journald which would be necessary to resolve
8797 "systemd-journal" to a numeric GID, because this might
8798 create deadlocks if NSS involves synchronous queries to
8799 other daemons (such as nscd, or sssd) which in turn are
8800 logging clients of journald and might block on it, which
8801 would then dead lock. A tmpfiles.d(5) snippet included in
8802 systemd will make sure the setgid bit and group are
8803 properly set on the journal directory if it exists on every
8804 boot. However, we recommend adjusting it manually after
8805 upgrades too (or from RPM scriptlets), so that the change is
8806 not delayed until next reboot.
8808 * Backlight and random seed files in /var/lib/ have moved into
8809 the /var/lib/systemd/ directory, in order to centralize all
8810 systemd generated files in one directory.
8812 * Boot time performance measurements (as displayed by
8813 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
8814 performance information if that's available to determine how
8815 much time BIOS and boot loader initialization required. With
8816 a sufficiently new BIOS you hence no longer need to boot
8817 with Gummiboot to get access to such information.
8819 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
8820 Cristian Rodríguez, Dave Reisner, David Herrmann, David
8821 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
8822 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
8823 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
8824 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
8825 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8827 — Berlin, 2013-10-02
8831 * The Restart= option for services now understands a new
8832 on-watchdog setting, which will restart the service
8833 automatically if the service stops sending out watchdog keep
8834 alive messages (as configured with WatchdogSec=).
8836 * The getty generator (which is responsible for bringing up a
8837 getty on configured serial consoles) will no longer only
8838 start a getty on the primary kernel console but on all
8839 others, too. This makes the order in which console= is
8840 specified on the kernel command line less important.
8842 * libsystemd-logind gained a new sd_session_get_vt() call to
8843 retrieve the VT number of a session.
8845 * If the option "tries=0" is set for an entry of /etc/crypttab
8846 its passphrase is queried indefinitely instead of any
8847 maximum number of tries.
8849 * If a service with a configure PID file terminates its PID
8850 file will now be removed automatically if it still exists
8851 afterwards. This should put an end to stale PID files.
8853 * systemd-run will now also take relative binary path names
8854 for execution and no longer insists on absolute paths.
8856 * InaccessibleDirectories= and ReadOnlyDirectories= now take
8857 paths that are optionally prefixed with "-" to indicate that
8858 it should not be considered a failure if they do not exist.
8860 * journalctl -o (and similar commands) now understands a new
8861 output mode "short-precise", it is similar to "short" but
8862 shows timestamps with usec accuracy.
8864 * The option "discard" (as known from Debian) is now
8865 synonymous to "allow-discards" in /etc/crypttab. In fact,
8866 "discard" is preferred now (since it is easier to remember
8869 * Some licensing clean-ups were made, so that more code is now
8870 LGPL-2.1 licensed than before.
8872 * A minimal tool to save/restore the display backlight
8873 brightness across reboots has been added. It will store the
8874 backlight setting as late as possible at shutdown, and
8875 restore it as early as possible during reboot.
8877 * A logic to automatically discover and enable home and swap
8878 partitions on GPT disks has been added. With this in place
8879 /etc/fstab becomes optional for many setups as systemd can
8880 discover certain partitions located on the root disk
8881 automatically. Home partitions are recognized under their
8882 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
8883 partitions are recognized under their GPT type ID
8884 0657fd6da4ab43c484e50933c84b4f4f.
8886 * systemd will no longer pass any environment from the kernel
8887 or initrd to system services. If you want to set an
8888 environment for all services, do so via the kernel command
8889 line systemd.setenv= assignment.
8891 * The systemd-sysctl tool no longer natively reads the file
8892 /etc/sysctl.conf. If desired, the file should be symlinked
8893 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
8894 legacy support by a symlink rather than built-in code, it
8895 also makes the otherwise hidden order of application of the
8896 different files visible. (Note that this partly reverts to a
8897 pre-198 application order of sysctl knobs!)
8899 * The "systemctl set-log-level" and "systemctl dump" commands
8900 have been moved to systemd-analyze.
8902 * systemd-run learned the new --remain-after-exit switch,
8903 which causes the scope unit not to be cleaned up
8904 automatically after the process terminated.
8906 * tmpfiles learned a new --exclude-prefix= switch to exclude
8907 certain paths from operation.
8909 * journald will now automatically flush all messages to disk
8910 as soon as a message at the log level CRIT, ALERT or EMERG
8913 Contributions from: Andrew Cook, Brandon Philips, Christian
8914 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
8915 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
8916 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
8917 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
8918 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
8919 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
8920 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
8921 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
8922 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
8923 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
8924 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
8925 William Giokas, Zbigniew Jędrzejewski-Szmek
8927 — Berlin, 2013-09-13
8931 * The documentation has been updated to cover the various new
8932 concepts introduced with 205.
8934 * Unit files now understand the new %v specifier which
8935 resolves to the kernel version string as returned by "uname
8938 * systemctl now supports filtering the unit list output by
8939 load state, active state and sub state, using the new
8942 * "systemctl status" will now show the results of the
8943 condition checks (like ConditionPathExists= and similar) of
8944 the last start attempts of the unit. They are also logged to
8947 * "journalctl -b" may now be used to look for boot output of a
8948 specific boot. Try "journalctl -b -1" for the previous boot,
8949 but the syntax is substantially more powerful.
8951 * "journalctl --show-cursor" has been added which prints the
8952 cursor string the last shown log line. This may then be used
8953 with the new "journalctl --after-cursor=" switch to continue
8954 browsing logs from that point on.
8956 * "journalctl --force" may now be used to force regeneration
8959 * Creation of "dead" device nodes has been moved from udev
8960 into kmod and tmpfiles. Previously, udev would read the kmod
8961 databases to pre-generate dead device nodes based on meta
8962 information contained in kernel modules, so that these would
8963 be auto-loaded on access rather then at boot. As this
8964 does not really have much to do with the exposing actual
8965 kernel devices to userspace this has always been slightly
8966 alien in the udev codebase. Following the new scheme kmod
8967 will now generate a runtime snippet for tmpfiles from the
8968 module meta information and it now is tmpfiles' job to the
8969 create the nodes. This also allows overriding access and
8970 other parameters for the nodes using the usual tmpfiles
8971 facilities. As side effect this allows us to remove the
8972 CAP_SYS_MKNOD capability bit from udevd entirely.
8974 * logind's device ACLs may now be applied to these "dead"
8975 devices nodes too, thus finally allowing managed access to
8976 devices such as /dev/snd/sequencer without loading the
8977 backing module right-away.
8979 * A new RPM macro has been added that may be used to apply
8980 tmpfiles configuration during package installation.
8982 * systemd-detect-virt and ConditionVirtualization= now can
8983 detect User-Mode-Linux machines (UML).
8985 * journald will now implicitly log the effective capabilities
8986 set of processes in the message metadata.
8988 * systemd-cryptsetup has gained support for TrueCrypt volumes.
8990 * The initrd interface has been simplified (more specifically,
8991 support for passing performance data via environment
8992 variables and fsck results via files in /run has been
8993 removed). These features were non-essential, and are
8994 nowadays available in a much nicer way by having systemd in
8995 the initrd serialize its state and have the hosts systemd
8996 deserialize it again.
8998 * The udev "keymap" data files and tools to apply keyboard
8999 specific mappings of scan to key codes, and force-release
9000 scan code lists have been entirely replaced by a udev
9001 "keyboard" builtin and a hwdb data file.
9003 * systemd will now honour the kernel's "quiet" command line
9004 argument also during late shutdown, resulting in a
9005 completely silent shutdown when used.
9007 * There's now an option to control the SO_REUSEPORT socket
9008 option in .socket units.
9010 * Instance units will now automatically get a per-template
9011 subslice of system.slice unless something else is explicitly
9012 configured. For example, instances of sshd@.service will now
9013 implicitly be placed in system-sshd.slice rather than
9014 system.slice as before.
9016 * Test coverage support may now be enabled at build time.
9018 Contributions from: Dave Reisner, Frederic Crozat, Harald
9019 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
9020 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
9021 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
9022 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
9023 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
9024 Giokas, Zbigniew Jędrzejewski-Szmek
9026 — Berlin, 2013-07-23
9030 * Two new unit types have been introduced:
9032 Scope units are very similar to service units, however, are
9033 created out of pre-existing processes — instead of PID 1
9034 forking off the processes. By using scope units it is
9035 possible for system services and applications to group their
9036 own child processes (worker processes) in a powerful way
9037 which then maybe used to organize them, or kill them
9038 together, or apply resource limits on them.
9040 Slice units may be used to partition system resources in an
9041 hierarchical fashion and then assign other units to them. By
9042 default there are now three slices: system.slice (for all
9043 system services), user.slice (for all user sessions),
9044 machine.slice (for VMs and containers).
9046 Slices and scopes have been introduced primarily in
9047 context of the work to move cgroup handling to a
9048 single-writer scheme, where only PID 1
9049 creates/removes/manages cgroups.
9051 * There's a new concept of "transient" units. In contrast to
9052 normal units these units are created via an API at runtime,
9053 not from configuration from disk. More specifically this
9054 means it is now possible to run arbitrary programs as
9055 independent services, with all execution parameters passed
9056 in via bus APIs rather than read from disk. Transient units
9057 make systemd substantially more dynamic then it ever was,
9058 and useful as a general batch manager.
9060 * logind has been updated to make use of scope and slice units
9061 for managing user sessions. As a user logs in he will get
9062 his own private slice unit, to which all sessions are added
9063 as scope units. We also added support for automatically
9064 adding an instance of user@.service for the user into the
9065 slice. Effectively logind will no longer create cgroup
9066 hierarchies on its own now, it will defer entirely to PID 1
9067 for this by means of scope, service and slice units. Since
9068 user sessions this way become entities managed by PID 1
9069 the output of "systemctl" is now a lot more comprehensive.
9071 * A new mini-daemon "systemd-machined" has been added which
9072 may be used by virtualization managers to register local
9073 VMs/containers. nspawn has been updated accordingly, and
9074 libvirt will be updated shortly. machined will collect a bit
9075 of meta information about the VMs/containers, and assign
9076 them their own scope unit (see above). The collected
9077 meta-data is then made available via the "machinectl" tool,
9078 and exposed in "ps" and similar tools. machined/machinectl
9079 is compile-time optional.
9081 * As discussed earlier, the low-level cgroup configuration
9082 options ControlGroup=, ControlGroupModify=,
9083 ControlGroupPersistent=, ControlGroupAttribute= have been
9084 removed. Please use high-level attribute settings instead as
9085 well as slice units.
9087 * A new bus call SetUnitProperties() has been added to alter
9088 various runtime parameters of a unit. This is primarily
9089 useful to alter cgroup parameters dynamically in a nice way,
9090 but will be extended later on to make more properties
9091 modifiable at runtime. systemctl gained a new set-properties
9092 command that wraps this call.
9094 * A new tool "systemd-run" has been added which can be used to
9095 run arbitrary command lines as transient services or scopes,
9096 while configuring a number of settings via the command
9097 line. This tool is currently very basic, however already
9098 very useful. We plan to extend this tool to even allow
9099 queuing of execution jobs with time triggers from the
9100 command line, similar in fashion to "at".
9102 * nspawn will now inform the user explicitly that kernels with
9103 audit enabled break containers, and suggest the user to turn
9106 * Support for detecting the IMA and AppArmor security
9107 frameworks with ConditionSecurity= has been added.
9109 * journalctl gained a new "-k" switch for showing only kernel
9110 messages, mimicking dmesg output; in addition to "--user"
9111 and "--system" switches for showing only user's own logs
9114 * systemd-delta can now show information about drop-in
9115 snippets extending unit files.
9117 * libsystemd-bus has been substantially updated but is still
9118 not available as public API.
9120 * systemd will now look for the "debug" argument on the kernel
9121 command line and enable debug logging, similar to what
9122 "systemd.log_level=debug" already did before.
9124 * "systemctl set-default", "systemctl get-default" has been
9125 added to configure the default.target symlink, which
9126 controls what to boot into by default.
9128 * "systemctl set-log-level" has been added as a convenient
9129 way to raise and lower systemd logging threshold.
9131 * "systemd-analyze plot" will now show the time the various
9132 generators needed for execution, as well as information
9133 about the unit file loading.
9135 * libsystemd-journal gained a new sd_journal_open_files() call
9136 for opening specific journal files. journactl also gained a
9137 new switch to expose this new functionality. Previously we
9138 only supported opening all files from a directory, or all
9139 files from the system, as opening individual files only is
9140 racy due to journal file rotation.
9142 * systemd gained the new DefaultEnvironment= setting in
9143 /etc/systemd/system.conf to set environment variables for
9146 * If a privileged process logs a journal message with the
9147 OBJECT_PID= field set, then journald will automatically
9148 augment this with additional OBJECT_UID=, OBJECT_GID=,
9149 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
9150 system services want to log events about specific client
9151 processes. journactl/systemctl has been updated to make use
9152 of this information if all log messages regarding a specific
9155 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
9156 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
9157 Reisner, David Coppa, David King, David Strauss, Eelco
9158 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
9159 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
9160 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
9161 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
9162 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
9163 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
9164 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
9165 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
9166 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
9167 Łukasz Stelmach, 장동준
9171 * The Python bindings gained some minimal support for the APIs
9172 exposed by libsystemd-logind.
9174 * ConditionSecurity= gained support for detecting SMACK. Since
9175 this condition already supports SELinux and AppArmor we only
9176 miss IMA for this. Patches welcome!
9178 Contributions from: Karol Lewandowski, Lennart Poettering,
9179 Zbigniew Jędrzejewski-Szmek
9183 * systemd-nspawn will now create /etc/resolv.conf if
9184 necessary, before bind-mounting the host's file onto it.
9186 * systemd-nspawn will now store meta information about a
9187 container on the container's cgroup as extended attribute
9188 fields, including the root directory.
9190 * The cgroup hierarchy has been reworked in many ways. All
9191 objects any of the components systemd creates in the cgroup
9192 tree are now suffixed. More specifically, user sessions are
9193 now placed in cgroups suffixed with ".session", users in
9194 cgroups suffixed with ".user", and nspawn containers in
9195 cgroups suffixed with ".nspawn". Furthermore, all cgroup
9196 names are now escaped in a simple scheme to avoid collision
9197 of userspace object names with kernel filenames. This work
9198 is preparation for making these objects relocatable in the
9199 cgroup tree, in order to allow easy resource partitioning of
9200 these objects without causing naming conflicts.
9202 * systemctl list-dependencies gained the new switches
9203 --plain, --reverse, --after and --before.
9205 * systemd-inhibit now shows the process name of processes that
9206 have taken an inhibitor lock.
9208 * nss-myhostname will now also resolve "localhost"
9209 implicitly. This makes /etc/hosts an optional file and
9210 nicely handles that on IPv6 ::1 maps to both "localhost" and
9213 * libsystemd-logind.so gained a new call
9214 sd_get_machine_names() to enumerate running containers and
9215 VMs (currently only supported by very new libvirt and
9216 nspawn). sd_login_monitor can now be used to watch
9217 VMs/containers coming and going.
9219 * .include is not allowed recursively anymore, and only in
9220 unit files. Usually it is better to use drop-in snippets in
9221 .d/*.conf anyway, as introduced with systemd 198.
9223 * systemd-analyze gained a new "critical-chain" command that
9224 determines the slowest chain of units run during system
9225 boot-up. It is very useful for tracking down where
9226 optimizing boot time is the most beneficial.
9228 * systemd will no longer allow manipulating service paths in
9229 the name=systemd:/system cgroup tree using ControlGroup= in
9230 units. (But is still fine with it in all other dirs.)
9232 * There's a new systemd-nspawn@.service service file that may
9233 be used to easily run nspawn containers as system
9234 services. With the container's root directory in
9235 /var/lib/container/foobar it is now sufficient to run
9236 "systemctl start systemd-nspawn@foobar.service" to boot it.
9238 * systemd-cgls gained a new parameter "--machine" to list only
9239 the processes within a certain container.
9241 * ConditionSecurity= now can check for "apparmor". We still
9242 are lacking checks for SMACK and IMA for this condition
9243 check though. Patches welcome!
9245 * A new configuration file /etc/systemd/sleep.conf has been
9246 added that may be used to configure which kernel operation
9247 systemd is supposed to execute when "suspend", "hibernate"
9248 or "hybrid-sleep" is requested. This makes the new kernel
9249 "freeze" state accessible to the user.
9251 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
9252 the passed argument if applicable.
9254 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
9255 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
9256 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
9257 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
9258 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
9259 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
9260 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
9265 * The output of 'systemctl list-jobs' got some polishing. The
9266 '--type=' argument may now be passed more than once. A new
9267 command 'systemctl list-sockets' has been added which shows
9268 a list of kernel sockets systemd is listening on with the
9269 socket units they belong to, plus the units these socket
9272 * The experimental libsystemd-bus library got substantial
9273 updates to work in conjunction with the (also experimental)
9274 kdbus kernel project. It works well enough to exchange
9275 messages with some sophistication. Note that kdbus is not
9276 ready yet, and the library is mostly an elaborate test case
9277 for now, and not installable.
9279 * systemd gained a new unit 'systemd-static-nodes.service'
9280 that generates static device nodes earlier during boot, and
9281 can run in conjunction with udev.
9283 * libsystemd-login gained a new call sd_pid_get_user_unit()
9284 to retrieve the user systemd unit a process is running
9285 in. This is useful for systems where systemd is used as
9288 * systemd-nspawn now places all containers in the new /machine
9289 top-level cgroup directory in the name=systemd
9290 hierarchy. libvirt will soon do the same, so that we get a
9291 uniform separation of /system, /user and /machine for system
9292 services, user processes and containers/virtual
9293 machines. This new cgroup hierarchy is also useful to stick
9294 stable names to specific container instances, which can be
9295 recognized later this way (this name may be controlled
9296 via systemd-nspawn's new -M switch). libsystemd-login also
9297 gained a new call sd_pid_get_machine_name() to retrieve the
9298 name of the container/VM a specific process belongs to.
9300 * bootchart can now store its data in the journal.
9302 * libsystemd-journal gained a new call
9303 sd_journal_add_conjunction() for AND expressions to the
9304 matching logic. This can be used to express more complex
9305 logical expressions.
9307 * journactl can now take multiple --unit= and --user-unit=
9310 * The cryptsetup logic now understands the "luks.key=" kernel
9311 command line switch for specifying a file to read the
9312 decryption key from. Also, if a configured key file is not
9313 found the tool will now automatically fall back to prompting
9316 * Python systemd.journal module was updated to wrap recently
9317 added functions from libsystemd-journal. The interface was
9318 changed to bring the low level interface in s.j._Reader
9319 closer to the C API, and the high level interface in
9320 s.j.Reader was updated to wrap and convert all data about
9323 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
9324 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
9325 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
9326 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
9327 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
9328 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9332 * journalctl --update-catalog now understands a new --root=
9333 option to operate on catalogs found in a different root
9336 * During shutdown after systemd has terminated all running
9337 services a final killing loop kills all remaining left-over
9338 processes. We will now print the name of these processes
9339 when we send SIGKILL to them, since this usually indicates a
9342 * If /etc/crypttab refers to password files stored on
9343 configured mount points automatic dependencies will now be
9344 generated to ensure the specific mount is established first
9345 before the key file is attempted to be read.
9347 * 'systemctl status' will now show information about the
9348 network sockets a socket unit is listening on.
9350 * 'systemctl status' will also shown information about any
9351 drop-in configuration file for units. (Drop-In configuration
9352 files in this context are files such as
9353 /etc/systemd/systemd/foobar.service.d/*.conf)
9355 * systemd-cgtop now optionally shows summed up CPU times of
9356 cgroups. Press '%' while running cgtop to switch between
9357 percentage and absolute mode. This is useful to determine
9358 which cgroups use up the most CPU time over the entire
9359 runtime of the system. systemd-cgtop has also been updated
9360 to be 'pipeable' for processing with further shell tools.
9362 * 'hostnamectl set-hostname' will now allow setting of FQDN
9365 * The formatting and parsing of time span values has been
9366 changed. The parser now understands fractional expressions
9367 such as "5.5h". The formatter will now output fractional
9368 expressions for all time spans under 1min, i.e. "5.123456s"
9369 rather than "5s 123ms 456us". For time spans under 1s
9370 millisecond values are shown, for those under 1ms
9371 microsecond values are shown. This should greatly improve
9372 all time-related output of systemd.
9374 * libsystemd-login and libsystemd-journal gained new
9375 functions for querying the poll() events mask and poll()
9376 timeout value for integration into arbitrary event
9379 * localectl gained the ability to list available X11 keymaps
9380 (models, layouts, variants, options).
9382 * 'systemd-analyze dot' gained the ability to filter for
9383 specific units via shell-style globs, to create smaller,
9384 more useful graphs. I.e. it is now possible to create simple
9385 graphs of all the dependencies between only target units, or
9386 of all units that Avahi has dependencies with.
9388 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
9389 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
9390 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
9391 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
9392 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
9393 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
9394 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
9398 * The boot-time readahead implementation for rotating media
9399 will now read the read-ahead data in multiple passes which
9400 consist of all read requests made in equidistant time
9401 intervals. This means instead of strictly reading read-ahead
9402 data in its physical order on disk we now try to find a
9403 middle ground between physical and access time order.
9405 * /etc/os-release files gained a new BUILD_ID= field for usage
9406 on operating systems that provide continuous builds of OS
9409 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
9410 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
9411 William Douglas, Zbigniew Jędrzejewski-Szmek
9415 * systemd-python gained an API exposing libsystemd-daemon.
9417 * The SMACK setup logic gained support for uploading CIPSO
9420 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
9421 ReadOnlyDirectories= and InaccessibleDirectories= has
9422 changed. The private /tmp and /var/tmp directories are now
9423 shared by all processes of a service (which means
9424 ExecStartPre= may now leave data in /tmp that ExecStart= of
9425 the same service can still access). When a service is
9426 stopped its temporary directories are immediately deleted
9427 (normal clean-up with tmpfiles is still done in addition to
9430 * By default, systemd will now set a couple of sysctl
9431 variables in the kernel: the safe sysrq options are turned
9432 on, IP route verification is turned on, and source routing
9433 disabled. The recently added hardlink and softlink
9434 protection of the kernel is turned on. These settings should
9435 be reasonably safe, and good defaults for all new systems.
9437 * The predictable network naming logic may now be turned off
9438 with a new kernel command line switch: net.ifnames=0.
9440 * A new libsystemd-bus module has been added that implements a
9441 pretty complete D-Bus client library. For details see:
9443 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
9445 * journald will now explicitly flush the journal files to disk
9446 at the latest 5min after each write. The file will then also
9447 be marked offline until the next write. This should increase
9448 reliability in case of a crash. The synchronization delay
9449 can be configured via SyncIntervalSec= in journald.conf.
9451 * There's a new remote-fs-setup.target unit that can be used
9452 to pull in specific services when at least one remote file
9453 system is to be mounted.
9455 * There are new targets timers.target and paths.target as
9456 canonical targets to pull user timer and path units in
9457 from. This complements sockets.target with a similar
9458 purpose for socket units.
9460 * libudev gained a new call udev_device_set_attribute_value()
9461 to set sysfs attributes of a device.
9463 * The udev daemon now sets the default number of worker
9464 processes executed in parallel based on the number of available
9465 CPUs instead of the amount of available RAM. This is supposed
9466 to provide a more reliable default and limit a too aggressive
9467 parallelism for setups with 1000s of devices connected.
9469 Contributions from: Auke Kok, Colin Walters, Cristian
9470 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
9471 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
9472 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
9473 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
9474 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
9475 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
9476 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
9477 Zbigniew Jędrzejewski-Szmek
9481 * Configuration of unit files may now be extended via drop-in
9482 files without having to edit/override the unit files
9483 themselves. More specifically, if the administrator wants to
9484 change one value for a service file foobar.service he can
9485 now do so by dropping in a configuration snippet into
9486 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
9487 will load all these snippets and apply them on top of the
9488 main unit configuration file, possibly extending or
9489 overriding its settings. Using these drop-in snippets is
9490 generally nicer than the two earlier options for changing
9491 unit files locally: copying the files from
9492 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
9493 them there; or creating a new file in /etc/systemd/system/
9494 that incorporates the original one via ".include". Drop-in
9495 snippets into these .d/ directories can be placed in any
9496 directory systemd looks for units in, and the usual
9497 overriding semantics between /usr/lib, /etc and /run apply
9500 * Most unit file settings which take lists of items can now be
9501 reset by assigning the empty string to them. For example,
9502 normally, settings such as Environment=FOO=BAR append a new
9503 environment variable assignment to the environment block,
9504 each time they are used. By assigning Environment= the empty
9505 string the environment block can be reset to empty. This is
9506 particularly useful with the .d/*.conf drop-in snippets
9507 mentioned above, since this adds the ability to reset list
9508 settings from vendor unit files via these drop-ins.
9510 * systemctl gained a new "list-dependencies" command for
9511 listing the dependencies of a unit recursively.
9513 * Inhibitors are now honored and listed by "systemctl
9514 suspend", "systemctl poweroff" (and similar) too, not only
9515 GNOME. These commands will also list active sessions by
9518 * Resource limits (as exposed by the various control group
9519 controllers) can now be controlled dynamically at runtime
9520 for all units. More specifically, you can now use a command
9521 like "systemctl set-cgroup-attr foobar.service cpu.shares
9522 2000" to alter the CPU shares a specific service gets. These
9523 settings are stored persistently on disk, and thus allow the
9524 administrator to easily adjust the resource usage of
9525 services with a few simple commands. This dynamic resource
9526 management logic is also available to other programs via the
9527 bus. Almost any kernel cgroup attribute and controller is
9530 * systemd-vconsole-setup will now copy all font settings to
9531 all allocated VTs, where it previously applied them only to
9534 * libsystemd-login gained the new sd_session_get_tty() API
9537 * This release drops support for a few legacy or
9538 distribution-specific LSB facility names when parsing init
9539 scripts: $x-display-manager, $mail-transfer-agent,
9540 $mail-transport-agent, $mail-transfer-agent, $smtp,
9541 $null. Also, the mail-transfer-agent.target unit backing
9542 this has been removed. Distributions which want to retain
9543 compatibility with this should carry the burden for
9544 supporting this themselves and patch support for these back
9545 in, if they really need to. Also, the facilities $syslog and
9546 $local_fs are now ignored, since systemd does not support
9547 early-boot LSB init scripts anymore, and these facilities
9548 are implied anyway for normal services. syslog.target has
9551 * There are new bus calls on PID1's Manager object for
9552 cancelling jobs, and removing snapshot units. Previously,
9553 both calls were only available on the Job and Snapshot
9556 * systemd-journal-gatewayd gained SSL support.
9558 * The various "environment" files, such as /etc/locale.conf
9559 now support continuation lines with a backslash ("\") as
9560 last character in the line, similarly in style (but different)
9561 to how this is supported in shells.
9563 * For normal user processes the _SYSTEMD_USER_UNIT= field is
9564 now implicitly appended to every log entry logged. systemctl
9565 has been updated to filter by this field when operating on a
9566 user systemd instance.
9568 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
9569 CAP_AUDIT_CONTROL capabilities to the capabilities set for
9570 the container. This makes it easier to boot unmodified
9571 Fedora systems in a container, which however still requires
9572 audit=0 to be passed on the kernel command line. Auditing in
9573 kernel and userspace is unfortunately still too broken in
9574 context of containers, hence we recommend compiling it out
9575 of the kernel or using audit=0. Hopefully this will be fixed
9576 one day for good in the kernel.
9578 * nspawn gained the new --bind= and --bind-ro= parameters to
9579 bind mount specific directories from the host into the
9582 * nspawn will now mount its own devpts file system instance
9583 into the container, in order not to leak pty devices from
9584 the host into the container.
9586 * systemd will now read the firmware boot time performance
9587 information from the EFI variables, if the used boot loader
9588 supports this, and takes it into account for boot performance
9589 analysis via "systemd-analyze". This is currently supported
9590 only in conjunction with Gummiboot, but could be supported
9591 by other boot loaders too. For details see:
9593 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
9595 * A new generator has been added that automatically mounts the
9596 EFI System Partition (ESP) to /boot, if that directory
9597 exists, is empty, and no other file system has been
9598 configured to be mounted there.
9600 * logind will now send out PrepareForSleep(false) out
9601 unconditionally, after coming back from suspend. This may be
9602 used by applications as asynchronous notification for
9603 system resume events.
9605 * "systemctl unlock-sessions" has been added, that allows
9606 unlocking the screens of all user sessions at once, similar
9607 to how "systemctl lock-sessions" already locked all users
9608 sessions. This is backed by a new D-Bus call UnlockSessions().
9610 * "loginctl seat-status" will now show the master device of a
9611 seat. (i.e. the device of a seat that needs to be around for
9612 the seat to be considered available, usually the graphics
9615 * tmpfiles gained a new "X" line type, that allows
9616 configuration of files and directories (with wildcards) that
9617 shall be excluded from automatic cleanup ("aging").
9619 * udev default rules set the device node permissions now only
9620 at "add" events, and do not change them any longer with a
9621 later "change" event.
9623 * The log messages for lid events and power/sleep keypresses
9624 now carry a message ID.
9626 * We now have a substantially larger unit test suite, but this
9627 continues to be work in progress.
9629 * udevadm hwdb gained a new --root= parameter to change the
9630 root directory to operate relative to.
9632 * logind will now issue a background sync() request to the kernel
9633 early at shutdown, so that dirty buffers are flushed to disk early
9634 instead of at the last moment, in order to optimize shutdown
9637 * A new bootctl tool has been added that is an interface for
9638 certain boot loader operations. This is currently a preview
9639 and is likely to be extended into a small mechanism daemon
9640 like timedated, localed, hostnamed, and can be used by
9641 graphical UIs to enumerate available boot options, and
9642 request boot into firmware operations.
9644 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
9645 the rest of the package. It also has been updated to work
9646 correctly in initrds.
9648 * polkit previously has been runtime optional, and is now also
9649 compile time optional via a configure switch.
9651 * systemd-analyze has been reimplemented in C. Also "systemctl
9652 dot" has moved into systemd-analyze.
9654 * "systemctl status" with no further parameters will now print
9655 the status of all active or failed units.
9657 * Operations such as "systemctl start" can now be executed
9658 with a new mode "--irreversible" which may be used to queue
9659 operations that cannot accidentally be reversed by a later
9660 job queuing. This is by default used to make shutdown
9661 requests more robust.
9663 * The Python API of systemd now gained a new module for
9664 reading journal files.
9666 * A new tool kernel-install has been added that can install
9667 kernel images according to the Boot Loader Specification:
9669 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
9671 * Boot time console output has been improved to provide
9672 animated boot time output for hanging jobs.
9674 * A new tool systemd-activate has been added which can be used
9675 to test socket activation with, directly from the command
9676 line. This should make it much easier to test and debug
9677 socket activation in daemons.
9679 * journalctl gained a new "--reverse" (or -r) option to show
9680 journal output in reverse order (i.e. newest line first).
9682 * journalctl gained a new "--pager-end" (or -e) option to jump
9683 to immediately jump to the end of the journal in the
9684 pager. This is only supported in conjunction with "less".
9686 * journalctl gained a new "--user-unit=" option, that works
9687 similarly to "--unit=" but filters for user units rather than
9690 * A number of unit files to ease adoption of systemd in
9691 initrds has been added. This moves some minimal logic from
9692 the various initrd implementations into systemd proper.
9694 * The journal files are now owned by a new group
9695 "systemd-journal", which exists specifically to allow access
9696 to the journal, and nothing else. Previously, we used the
9697 "adm" group for that, which however possibly covers more
9698 than just journal/log file access. This new group is now
9699 already used by systemd-journal-gatewayd to ensure this
9700 daemon gets access to the journal files and as little else
9701 as possible. Note that "make install" will also set FS ACLs
9702 up for /var/log/journal to give "adm" and "wheel" read
9703 access to it, in addition to "systemd-journal" which owns
9704 the journal files. We recommend that packaging scripts also
9705 add read access to "adm" + "wheel" to /var/log/journal, and
9706 all existing/future journal files. To normal users and
9707 administrators little changes, however packagers need to
9708 ensure to create the "systemd-journal" system group at
9709 package installation time.
9711 * The systemd-journal-gatewayd now runs as unprivileged user
9712 systemd-journal-gateway:systemd-journal-gateway. Packaging
9713 scripts need to create these system user/group at
9716 * timedated now exposes a new boolean property CanNTP that
9717 indicates whether a local NTP service is available or not.
9719 * systemd-detect-virt will now also detect xen PVs
9721 * The pstore file system is now mounted by default, if it is
9724 * In addition to the SELinux and IMA policies we will now also
9725 load SMACK policies at early boot.
9727 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
9728 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
9729 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
9730 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
9731 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
9732 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
9733 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
9734 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
9735 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
9736 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
9737 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
9738 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
9739 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
9740 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
9744 * Timer units now support calendar time events in addition to
9745 monotonic time events. That means you can now trigger a unit
9746 based on a calendar time specification such as "Thu,Fri
9747 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
9748 or fifth day of any month of the year 2013, given that it is
9749 a thursday or friday. This brings timer event support
9750 considerably closer to cron's capabilities. For details on
9751 the supported calendar time specification language see
9754 * udev now supports a number of different naming policies for
9755 network interfaces for predictable names, and a combination
9756 of these policies is now the default. Please see this wiki
9757 document for details:
9759 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
9761 * Auke Kok's bootchart implementation has been added to the
9762 systemd tree. It is an optional component that can graph the
9763 boot in quite some detail. It is one of the best bootchart
9764 implementations around and minimal in its code and
9767 * nss-myhostname has been integrated into the systemd source
9768 tree. nss-myhostname guarantees that the local hostname
9769 always stays resolvable via NSS. It has been a weak
9770 requirement of systemd-hostnamed since a long time, and
9771 since its code is actually trivial we decided to just
9772 include it in systemd's source tree. It can be turned off
9773 with a configure switch.
9775 * The read-ahead logic is now capable of properly detecting
9776 whether a btrfs file system is on SSD or rotating media, in
9777 order to optimize the read-ahead scheme. Previously, it was
9778 only capable of detecting this on traditional file systems
9781 * In udev, additional device properties are now read from the
9782 IAB in addition to the OUI database. Also, Bluetooth company
9783 identities are attached to the devices as well.
9785 * In service files %U may be used as specifier that is
9786 replaced by the configured user name of the service.
9788 * nspawn may now be invoked without a controlling TTY. This
9789 makes it suitable for invocation as its own service. This
9790 may be used to set up a simple containerized server system
9791 using only core OS tools.
9793 * systemd and nspawn can now accept socket file descriptors
9794 when they are started for socket activation. This enables
9795 implementation of socket activated nspawn
9796 containers. i.e. think about autospawning an entire OS image
9797 when the first SSH or HTTP connection is received. We expect
9798 that similar functionality will also be added to libvirt-lxc
9801 * journalctl will now suppress ANSI color codes when
9802 presenting log data.
9804 * systemctl will no longer show control group information for
9805 a unit if the control group is empty anyway.
9807 * logind can now automatically suspend/hibernate/shutdown the
9810 * /etc/machine-info and hostnamed now also expose the chassis
9811 type of the system. This can be used to determine whether
9812 the local system is a laptop, desktop, handset or
9813 tablet. This information may either be configured by the
9814 user/vendor or is automatically determined from ACPI and DMI
9815 information if possible.
9817 * A number of polkit actions are now bound together with "imply"
9818 rules. This should simplify creating UIs because many actions
9819 will now authenticate similar ones as well.
9821 * Unit files learnt a new condition ConditionACPower= which
9822 may be used to conditionalize a unit depending on whether an
9823 AC power source is connected or not, of whether the system
9824 is running on battery power.
9826 * systemctl gained a new "is-failed" verb that may be used in
9827 shell scripts and suchlike to check whether a specific unit
9828 is in the "failed" state.
9830 * The EnvironmentFile= setting in unit files now supports file
9831 globbing, and can hence be used to easily read a number of
9832 environment files at once.
9834 * systemd will no longer detect and recognize specific
9835 distributions. All distribution-specific #ifdeffery has been
9836 removed, systemd is now fully generic and
9837 distribution-agnostic. Effectively, not too much is lost as
9838 a lot of the code is still accessible via explicit configure
9839 switches. However, support for some distribution specific
9840 legacy configuration file formats has been dropped. We
9841 recommend distributions to simply adopt the configuration
9842 files everybody else uses now and convert the old
9843 configuration from packaging scripts. Most distributions
9844 already did that. If that's not possible or desirable,
9845 distributions are welcome to forward port the specific
9846 pieces of code locally from the git history.
9848 * When logging a message about a unit systemd will now always
9849 log the unit name in the message meta data.
9851 * localectl will now also discover system locale data that is
9852 not stored in locale archives, but directly unpacked.
9854 * logind will no longer unconditionally use framebuffer
9855 devices as seat masters, i.e. as devices that are required
9856 to be existing before a seat is considered preset. Instead,
9857 it will now look for all devices that are tagged as
9858 "seat-master" in udev. By default, framebuffer devices will
9859 be marked as such, but depending on local systems, other
9860 devices might be marked as well. This may be used to
9861 integrate graphics cards using closed source drivers (such
9862 as NVidia ones) more nicely into logind. Note however, that
9863 we recommend using the open source NVidia drivers instead,
9864 and no udev rules for the closed-source drivers will be
9865 shipped from us upstream.
9867 Contributions from: Adam Williamson, Alessandro Crismani, Auke
9868 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
9869 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
9870 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
9871 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
9872 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
9873 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
9874 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
9875 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
9876 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
9877 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
9878 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
9883 * udev gained support for loading additional device properties
9884 from an indexed database that is keyed by vendor/product IDs
9885 and similar device identifiers. For the beginning this
9886 "hwdb" is populated with data from the well-known PCI and
9887 USB database, but also includes PNP, ACPI and OID data. In
9888 the longer run this indexed database shall grow into
9889 becoming the one central database for non-essential
9890 userspace device metadata. Previously, data from the PCI/USB
9891 database was only attached to select devices, since the
9892 lookup was a relatively expensive operation due to O(n) time
9893 complexity (with n being the number of entries in the
9894 database). Since this is now O(1), we decided to add in this
9895 data for all devices where this is available, by
9896 default. Note that the indexed database needs to be rebuilt
9897 when new data files are installed. To achieve this you need
9898 to update your packaging scripts to invoke "udevadm hwdb
9899 --update" after installation of hwdb data files. For
9900 RPM-based distributions we introduced the new
9901 %udev_hwdb_update macro for this purpose.
9903 * The Journal gained support for the "Message Catalog", an
9904 indexed database to link up additional information with
9905 journal entries. For further details please check:
9907 https://www.freedesktop.org/wiki/Software/systemd/catalog
9909 The indexed message catalog database also needs to be
9910 rebuilt after installation of message catalog files. Use
9911 "journalctl --update-catalog" for this. For RPM-based
9912 distributions we introduced the %journal_catalog_update
9913 macro for this purpose.
9915 * The Python Journal bindings gained support for the standard
9916 Python logging framework.
9918 * The Journal API gained new functions for checking whether
9919 the underlying file system of a journal file is capable of
9920 properly reporting file change notifications, or whether
9921 applications that want to reflect journal changes "live"
9922 need to recheck journal files continuously in appropriate
9925 * It is now possible to set the "age" field for tmpfiles
9926 entries to 0, indicating that files matching this entry
9927 shall always be removed when the directories are cleaned up.
9929 * coredumpctl gained a new "gdb" verb which invokes gdb
9930 right-away on the selected coredump.
9932 * There's now support for "hybrid sleep" on kernels that
9933 support this, in addition to "suspend" and "hibernate". Use
9934 "systemctl hybrid-sleep" to make use of this.
9936 * logind's HandleSuspendKey= setting (and related settings)
9937 now gained support for a new "lock" setting to simply
9938 request the screen lock on all local sessions, instead of
9939 actually executing a suspend or hibernation.
9941 * systemd will now mount the EFI variables file system by
9944 * Socket units now gained support for configuration of the
9945 SMACK security label.
9947 * timedatectl will now output the time of the last and next
9948 daylight saving change.
9950 * We dropped support for various legacy and distro-specific
9951 concepts, such as insserv, early-boot SysV services
9952 (i.e. those for non-standard runlevels such as 'b' or 'S')
9953 or ArchLinux /etc/rc.conf support. We recommend the
9954 distributions who still need support this to either continue
9955 to maintain the necessary patches downstream, or find a
9956 different solution. (Talk to us if you have questions!)
9958 * Various systemd components will now bypass polkit checks for
9959 root and otherwise handle properly if polkit is not found to
9960 be around. This should fix most issues for polkit-less
9961 systems. Quite frankly this should have been this way since
9962 day one. It is absolutely our intention to make systemd work
9963 fine on polkit-less systems, and we consider it a bug if
9964 something does not work as it should if polkit is not around.
9966 * For embedded systems it is now possible to build udev and
9967 systemd without blkid and/or kmod support.
9969 * "systemctl switch-root" is now capable of switching root
9970 more than once. I.e. in addition to transitions from the
9971 initrd to the host OS it is now possible to transition to
9972 further OS images from the host. This is useful to implement
9973 offline updating tools.
9975 * Various other additions have been made to the RPM macros
9976 shipped with systemd. Use %udev_rules_update() after
9977 installing new udev rules files. %_udevhwdbdir,
9978 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
9979 %_sysctldir are now available which resolve to the right
9980 directories for packages to place various data files in.
9982 * journalctl gained the new --full switch (in addition to
9983 --all, to disable ellipsation for long messages.
9985 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
9986 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
9987 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
9988 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
9989 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
9990 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
9991 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
9992 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
9993 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
9997 * journalctl gained new --since= and --until= switches to
9998 filter by time. It also now supports nice filtering for
9999 units via --unit=/-u.
10001 * Type=oneshot services may use ExecReload= and do the
10004 * The journal daemon now supports time-based rotation and
10005 vacuuming, in addition to the usual disk-space based
10008 * The journal will now index the available field values for
10009 each field name. This enables clients to show pretty drop
10010 downs of available match values when filtering. The bash
10011 completion of journalctl has been updated
10012 accordingly. journalctl gained a new switch -F to list all
10013 values a certain field takes in the journal database.
10015 * More service events are now written as structured messages
10016 to the journal, and made recognizable via message IDs.
10018 * The timedated, localed and hostnamed mini-services which
10019 previously only provided support for changing time, locale
10020 and hostname settings from graphical DEs such as GNOME now
10021 also have a minimal (but very useful) text-based client
10022 utility each. This is probably the nicest way to changing
10023 these settings from the command line now, especially since
10024 it lists available options and is fully integrated with bash
10027 * There's now a new tool "systemd-coredumpctl" to list and
10028 extract coredumps from the journal.
10030 * We now install a README each in /var/log/ and
10031 /etc/rc.d/init.d explaining where the system logs and init
10032 scripts went. This hopefully should help folks who go to
10033 that dirs and look into the otherwise now empty void and
10034 scratch their heads.
10036 * When user-services are invoked (by systemd --user) the
10037 $MANAGERPID env var is set to the PID of systemd.
10039 * SIGRTMIN+24 when sent to a --user instance will now result
10040 in immediate termination of systemd.
10042 * gatewayd received numerous feature additions such as a
10043 "follow" mode, for live syncing and filtering.
10045 * browse.html now allows filtering and showing detailed
10046 information on specific entries. Keyboard navigation and
10047 mouse screen support has been added.
10049 * gatewayd/journalctl now supports HTML5/JSON
10050 Server-Sent-Events as output.
10052 * The SysV init script compatibility logic will now
10053 heuristically determine whether a script supports the
10054 "reload" verb, and only then make this available as
10055 "systemctl reload".
10057 * "systemctl status --follow" has been removed, use "journalctl
10060 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
10061 have been removed since they are hardly useful to be
10064 * And I'd like to take the opportunity to specifically mention
10065 Zbigniew for his great contributions. Zbigniew, you rock!
10067 Contributions from: Andrew Eikum, Christian Hesse, Colin
10068 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
10069 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
10070 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
10071 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
10072 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
10073 Jędrzejewski-Szmek, Сковорода Никита Андреевич
10077 * If /etc/vconsole.conf is non-existent or empty we will no
10078 longer load any console font or key map at boot by
10079 default. Instead the kernel defaults will be left
10080 intact. This is definitely the right thing to do, as no
10081 configuration should mean no configuration, and hard-coding
10082 font names that are different on all archs is probably a bad
10083 idea. Also, the kernel default key map and font should be
10084 good enough for most cases anyway, and mostly identical to
10085 the userspace fonts/key maps we previously overloaded them
10086 with. If distributions want to continue to default to a
10087 non-kernel font or key map they should ship a default
10088 /etc/vconsole.conf with the appropriate contents.
10090 Contributions from: Colin Walters, Daniel J Walsh, Dave
10091 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
10092 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
10096 * journalctl gained a new --cursor= switch to show entries
10097 starting from the specified location in the journal.
10099 * We now enforce a size limit on journal entry fields exported
10100 with "-o json" in journalctl. Fields larger than 4K will be
10101 assigned null. This can be turned off with --all.
10103 * An (optional) journal gateway daemon is now available as
10104 "systemd-journal-gatewayd.service". This service provides
10105 access to the journal via HTTP and JSON. This functionality
10106 will be used to implement live log synchronization in both
10107 pull and push modes, but has various other users too, such
10108 as easy log access for debugging of embedded devices. Right
10109 now it is already useful to retrieve the journal via HTTP:
10111 # systemctl start systemd-journal-gatewayd.service
10112 # wget http://localhost:19531/entries
10114 This will download the journal contents in a
10115 /var/log/messages compatible format. The same as JSON:
10117 # curl -H"Accept: application/json" http://localhost:19531/entries
10119 This service is also accessible via a web browser where a
10120 single static HTML5 app is served that uses the JSON logic
10121 to enable the user to do some basic browsing of the
10122 journal. This will be extended later on. Here's an example
10123 screenshot of this app in its current state:
10125 http://0pointer.de/public/journal-gatewayd
10127 Contributions from: Kay Sievers, Lennart Poettering, Robert
10128 Milasan, Tom Gundersen
10132 * The bash completion logic is now available for journalctl
10135 * We do not mount the "cpuset" controller anymore together with
10136 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
10137 started if no parameters are assigned to it. "cpuset" hence
10138 broke code that assumed it could create "cpu" groups and
10141 * journalctl -f will now subscribe to terminal size changes,
10142 and line break accordingly.
10144 Contributions from: Dave Reisner, Kay Sievers, Lennart
10145 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
10149 * nspawn will now create a symlink /etc/localtime in the
10150 container environment, copying the host's timezone
10151 setting. Previously this has been done via a bind mount, but
10152 since symlinks cannot be bind mounted this has now been
10153 changed to create/update the appropriate symlink.
10155 * journalctl -n's line number argument is now optional, and
10156 will default to 10 if omitted.
10158 * journald will now log the maximum size the journal files may
10159 take up on disk. This is particularly useful if the default
10160 built-in logic of determining this parameter from the file
10161 system size is used. Use "systemctl status
10162 systemd-journald.service" to see this information.
10164 * The multi-seat X wrapper tool has been stripped down. As X
10165 is now capable of enumerating graphics devices via udev in a
10166 seat-aware way the wrapper is not strictly necessary
10167 anymore. A stripped down temporary stop-gap is still shipped
10168 until the upstream display managers have been updated to
10169 fully support the new X logic. Expect this wrapper to be
10170 removed entirely in one of the next releases.
10172 * HandleSleepKey= in logind.conf has been split up into
10173 HandleSuspendKey= and HandleHibernateKey=. The old setting
10174 is not available anymore. X11 and the kernel are
10175 distinguishing between these keys and we should too. This
10176 also means the inhibition lock for these keys has been split
10179 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
10180 Poettering, Lukas Nykryn, Václav Pavlín
10184 * Whenever a unit changes state we will now log this to the
10185 journal and show along the unit's own log output in
10186 "systemctl status".
10188 * ConditionPathIsMountPoint= can now properly detect bind
10189 mount points too. (Previously, a bind mount of one file
10190 system to another place in the same file system could not be
10191 detected as mount, since they shared struct stat's st_dev
10194 * We will now mount the cgroup controllers cpu, cpuacct,
10195 cpuset and the controllers net_cls, net_prio together by
10198 * nspawn containers will now have a virtualized boot
10199 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
10200 over with a randomized ID at container initialization). This
10201 has the effect of making "journalctl -b" do the right thing
10204 * The JSON output journal serialization has been updated not
10205 to generate "endless" list objects anymore, but rather one
10206 JSON object per line. This is more in line how most JSON
10207 parsers expect JSON objects. The new output mode
10208 "json-pretty" has been added to provide similar output, but
10209 neatly aligned for readability by humans.
10211 * We dropped all explicit sync() invocations in the shutdown
10212 code. The kernel does this implicitly anyway in the kernel
10213 reboot() syscall. halt(8)'s -n option is now a compatibility
10216 * We now support virtualized reboot() in containers, as
10217 supported by newer kernels. We will fall back to exit() if
10218 CAP_SYS_REBOOT is not available to the container. Also,
10219 nspawn makes use of this now and will actually reboot the
10220 container if the containerized OS asks for that.
10222 * journalctl will only show local log output by default
10223 now. Use --merge (-m) to show remote log output, too.
10225 * libsystemd-journal gained the new sd_journal_get_usage()
10226 call to determine the current disk usage of all journal
10227 files. This is exposed in the new "journalctl --disk-usage"
10230 * journald gained a new configuration setting SplitMode= in
10231 journald.conf which may be used to control how user journals
10232 are split off. See journald.conf(5) for details.
10234 * A new condition type ConditionFileNotEmpty= has been added.
10236 * tmpfiles' "w" lines now support file globbing, to write
10237 multiple files at once.
10239 * We added Python bindings for the journal submission
10240 APIs. More Python APIs for a number of selected APIs will
10241 likely follow. Note that we intend to add native bindings
10242 only for the Python language, as we consider it common
10243 enough to deserve bindings shipped within systemd. There are
10244 various projects outside of systemd that provide bindings
10245 for languages such as PHP or Lua.
10247 * Many conditions will now resolve specifiers such as %i. In
10248 addition, PathChanged= and related directives of .path units
10249 now support specifiers as well.
10251 * There's now a new RPM macro definition for the system preset
10254 * journald will now warn if it ca not forward a message to the
10255 syslog daemon because its socket is full.
10257 * timedated will no longer write or process /etc/timezone,
10258 except on Debian. As we do not support late mounted /usr
10259 anymore /etc/localtime always being a symlink is now safe,
10260 and hence the information in /etc/timezone is not necessary
10263 * logind will now always reserve one VT for a text getty (VT6
10264 by default). Previously if more than 6 X sessions where
10265 started they took up all the VTs with auto-spawned gettys,
10266 so that no text gettys were available anymore.
10268 * udev will now automatically inform the btrfs kernel logic
10269 about btrfs RAID components showing up. This should make
10270 simple hotplug based btrfs RAID assembly work.
10272 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
10273 (but not for its children which will stay at the kernel
10274 default). This should allow setups with a lot more listening
10277 * systemd will now always pass the configured timezone to the
10278 kernel at boot. timedated will do the same when the timezone
10281 * logind's inhibition logic has been updated. By default,
10282 logind will now handle the lid switch, the power and sleep
10283 keys all the time, even in graphical sessions. If DEs want
10284 to handle these events on their own they should take the new
10285 handle-power-key, handle-sleep-key and handle-lid-switch
10286 inhibitors during their runtime. A simple way to achieve
10287 that is to invoke the DE wrapped in an invocation of:
10289 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
10291 * Access to unit operations is now checked via SELinux taking
10292 the unit file label and client process label into account.
10294 * systemd will now notify the administrator in the journal
10295 when he over-mounts a non-empty directory.
10297 * There are new specifiers that are resolved in unit files,
10298 for the hostname (%H), the machine ID (%m) and the boot ID
10301 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
10302 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
10303 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
10304 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
10305 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
10306 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
10307 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
10311 * Support for reading structured kernel messages from
10312 /dev/kmsg has now been added and is enabled by default.
10314 * Support for reading kernel messages from /proc/kmsg has now
10315 been removed. If you want kernel messages in the journal
10316 make sure to run a recent kernel (>= 3.5) that supports
10317 reading structured messages from /dev/kmsg (see
10318 above). /proc/kmsg is now exclusive property of classic
10319 syslog daemons again.
10321 * The libudev API gained the new
10322 udev_device_new_from_device_id() call.
10324 * The logic for file system namespace (ReadOnlyDirectory=,
10325 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
10326 require pivot_root() anymore. This means fewer temporary
10327 directories are created below /tmp for this feature.
10329 * nspawn containers will now see and receive all submounts
10330 made on the host OS below the root file system of the
10333 * Forward Secure Sealing is now supported for Journal files,
10334 which provide cryptographical sealing of journal files so
10335 that attackers cannot alter log history anymore without this
10336 being detectable. Lennart will soon post a blog story about
10337 this explaining it in more detail.
10339 * There are two new service settings RestartPreventExitStatus=
10340 and SuccessExitStatus= which allow configuration of exit
10341 status (exit code or signal) which will be excepted from the
10342 restart logic, resp. consider successful.
10344 * journalctl gained the new --verify switch that can be used
10345 to check the integrity of the structure of journal files and
10346 (if Forward Secure Sealing is enabled) the contents of
10349 * nspawn containers will now be run with /dev/stdin, /dev/fd/
10350 and similar symlinks pre-created. This makes running shells
10351 as container init process a lot more fun.
10353 * The fstab support can now handle PARTUUID= and PARTLABEL=
10356 * A new ConditionHost= condition has been added to match
10357 against the hostname (with globs) and machine ID. This is
10358 useful for clusters where a single OS image is used to
10359 provision a large number of hosts which shall run slightly
10360 different sets of services.
10362 * Services which hit the restart limit will now be placed in a
10365 Contributions from: Bertram Poettering, Dave Reisner, Huang
10366 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
10367 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
10371 * When running in --user mode systemd will now become a
10372 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
10373 tree a lot more organized.
10375 * A new PartOf= unit dependency type has been introduced that
10376 may be used to group services in a natural way.
10378 * "systemctl enable" may now be used to enable instances of
10381 * journalctl now prints error log levels in red, and
10382 warning/notice log levels in bright white. It also supports
10383 filtering by log level now.
10385 * cgtop gained a new -n switch (similar to top), to configure
10386 the maximum number of iterations to run for. It also gained
10387 -b, to run in batch mode (accepting no input).
10389 * The suffix ".service" may now be omitted on most systemctl
10390 command lines involving service unit names.
10392 * There's a new bus call in logind to lock all sessions, as
10393 well as a loginctl verb for it "lock-sessions".
10395 * libsystemd-logind.so gained a new call sd_journal_perror()
10396 that works similar to libc perror() but logs to the journal
10397 and encodes structured information about the error number.
10399 * /etc/crypttab entries now understand the new keyfile-size=
10402 * shutdown(8) now can send a (configurable) wall message when
10403 a shutdown is cancelled.
10405 * The mount propagation mode for the root file system will now
10406 default to "shared", which is useful to make containers work
10407 nicely out-of-the-box so that they receive new mounts from
10408 the host. This can be undone locally by running "mount
10409 --make-rprivate /" if needed.
10411 * The prefdm.service file has been removed. Distributions
10412 should maintain this unit downstream if they intend to keep
10413 it around. However, we recommend writing normal unit files
10414 for display managers instead.
10416 * Since systemd is a crucial part of the OS we will now
10417 default to a number of compiler switches that improve
10418 security (hardening) such as read-only relocations, stack
10419 protection, and suchlike.
10421 * The TimeoutSec= setting for services is now split into
10422 TimeoutStartSec= and TimeoutStopSec= to allow configuration
10423 of individual time outs for the start and the stop phase of
10426 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
10427 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
10428 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
10429 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
10430 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
10431 Gundersen, Zbigniew Jędrzejewski-Szmek
10435 * The journal and id128 C APIs are now fully documented as man
10438 * Extra safety checks have been added when transitioning from
10439 the initial RAM disk to the main system to avoid accidental
10442 * /etc/crypttab entries now understand the new keyfile-offset=
10445 * systemctl -t can now be used to filter by unit load state.
10447 * The journal C API gained the new sd_journal_wait() call to
10448 make writing synchronous journal clients easier.
10450 * journalctl gained the new -D switch to show journals from a
10451 specific directory.
10453 * journalctl now displays a special marker between log
10454 messages of two different boots.
10456 * The journal is now explicitly flushed to /var via a service
10457 systemd-journal-flush.service, rather than implicitly simply
10458 by seeing /var/log/journal to be writable.
10460 * journalctl (and the journal C APIs) can now match for much
10461 more complex expressions, with alternatives and
10464 * When transitioning from the initial RAM disk to the main
10465 system we will now kill all processes in a killing spree to
10466 ensure no processes stay around by accident.
10468 * Three new specifiers may be used in unit files: %u, %h, %s
10469 resolve to the user name, user home directory resp. user
10470 shell. This is useful for running systemd user instances.
10472 * We now automatically rotate journal files if their data
10473 object hash table gets a fill level > 75%. We also size the
10474 hash table based on the configured maximum file size. This
10475 together should lower hash collisions drastically and thus
10476 speed things up a bit.
10478 * journalctl gained the new "--header" switch to introspect
10479 header data of journal files.
10481 * A new setting SystemCallFilters= has been added to services which may
10482 be used to apply deny lists or allow lists to system calls. This is
10483 based on SECCOMP Mode 2 of Linux 3.5.
10485 * nspawn gained a new --link-journal= switch (and quicker: -j)
10486 to link the container journal with the host. This makes it
10487 very easy to centralize log viewing on the host for all
10488 guests while still keeping the journal files separated.
10490 * Many bugfixes and optimizations
10492 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
10493 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
10494 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
10499 * Several tools now understand kernel command line arguments,
10500 which are only read when run in an initial RAM disk. They
10501 usually follow closely their normal counterparts, but are
10504 * There's a new tool to analyze the readahead files that are
10505 automatically generated at boot. Use:
10507 /usr/lib/systemd/systemd-readahead analyze /.readahead
10509 * We now provide an early debug shell on tty9 if this enabled. Use:
10511 systemctl enable debug-shell.service
10513 * All plymouth related units have been moved into the Plymouth
10514 package. Please make sure to upgrade your Plymouth version
10517 * systemd-tmpfiles now supports getting passed the basename of
10518 a configuration file only, in which case it will look for it
10519 in all appropriate directories automatically.
10521 * udevadm info now takes a /dev or /sys path as argument, and
10522 does the right thing. Example:
10524 udevadm info /dev/sda
10525 udevadm info /sys/class/block/sda
10527 * systemctl now prints a warning if a unit is stopped but a
10528 unit that might trigger it continues to run. Example: a
10529 service is stopped but the socket that activates it is left
10532 * "systemctl status" will now mention if the log output was
10533 shortened due to rotation since a service has been started.
10535 * The journal API now exposes functions to determine the
10536 "cutoff" times due to rotation.
10538 * journald now understands SIGUSR1 and SIGUSR2 for triggering
10539 immediately flushing of runtime logs to /var if possible,
10540 resp. for triggering immediate rotation of the journal
10543 * It is now considered an error if a service is attempted to
10544 be stopped that is not loaded.
10546 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
10548 * systemd-analyze now supports Python 3
10550 * tmpfiles now supports cleaning up directories via aging
10551 where the first level dirs are always kept around but
10552 directories beneath it automatically aged. This is enabled
10553 by prefixing the age field with '~'.
10555 * Seat objects now expose CanGraphical, CanTTY properties
10556 which is required to deal with very fast bootups where the
10557 display manager might be running before the graphics drivers
10558 completed initialization.
10560 * Seat objects now expose a State property.
10562 * We now include RPM macros for service enabling/disabling
10563 based on the preset logic. We recommend RPM based
10564 distributions to make use of these macros if possible. This
10565 makes it simpler to reuse RPM spec files across
10568 * We now make sure that the collected systemd unit name is
10569 always valid when services log to the journal via
10572 * There's a new man page kernel-command-line(7) detailing all
10573 command line options we understand.
10575 * The fstab generator may now be disabled at boot by passing
10576 fstab=0 on the kernel command line.
10578 * A new kernel command line option modules-load= is now understood
10579 to load a specific kernel module statically, early at boot.
10581 * Unit names specified on the systemctl command line are now
10582 automatically escaped as needed. Also, if file system or
10583 device paths are specified they are automatically turned
10584 into the appropriate mount or device unit names. Example:
10586 systemctl status /home
10587 systemctl status /dev/sda
10589 * The SysVConsole= configuration option has been removed from
10590 system.conf parsing.
10592 * The SysV search path is no longer exported on the D-Bus
10595 * The Names= option has been removed from unit file parsing.
10597 * There's a new man page bootup(7) detailing the boot process.
10599 * Every unit and every generator we ship with systemd now
10600 comes with full documentation. The self-explanatory boot is
10603 * A couple of services gained "systemd-" prefixes in their
10604 name if they wrap systemd code, rather than only external
10605 code. Among them fsck@.service which is now
10606 systemd-fsck@.service.
10608 * The HaveWatchdog property has been removed from the D-Bus
10611 * systemd.confirm_spawn= on the kernel command line should now
10614 * There's a new man page crypttab(5) which details all options
10615 we actually understand.
10617 * systemd-nspawn gained a new --capability= switch to pass
10618 additional capabilities to the container.
10620 * timedated will now read known NTP implementation unit names
10621 from /usr/lib/systemd/ntp-units.d/*.list,
10622 systemd-timedated-ntp.target has been removed.
10624 * journalctl gained a new switch "-b" that lists log data of
10625 the current boot only.
10627 * The notify socket is in the abstract namespace again, in
10628 order to support daemons which chroot() at start-up.
10630 * There is a new Storage= configuration option for journald
10631 which allows configuration of where log data should go. This
10632 also provides a way to disable journal logging entirely, so
10633 that data collected is only forwarded to the console, the
10634 kernel log buffer or another syslog implementation.
10636 * Many bugfixes and optimizations
10638 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
10639 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
10640 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
10641 Shawn Landden, Tom Gundersen
10645 * "systemctl help <unit>" now shows the man page if one is
10648 * Several new man pages have been added.
10650 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
10651 MaxLevelConsole= can now be specified in
10652 journald.conf. These options allow reducing the amount of
10653 data stored on disk or forwarded by the log level.
10655 * TimerSlackNSec= can now be specified in system.conf for
10656 PID1. This allows system-wide power savings.
10658 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
10659 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
10664 * logind is now capable of (optionally) handling power and
10665 sleep keys as well as the lid switch.
10667 * journalctl now understands the syntax "journalctl
10668 /usr/bin/avahi-daemon" to get all log output of a specific
10671 * CapabilityBoundingSet= in system.conf now also influences
10672 the capability bound set of usermode helpers of the kernel.
10674 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
10675 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
10676 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
10677 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
10681 * Note that we skipped 139 releases here in order to set the
10682 new version to something that is greater than both udev's
10683 and systemd's most recent version number.
10685 * udev: all udev sources are merged into the systemd source tree now.
10686 All future udev development will happen in the systemd tree. It
10687 is still fully supported to use the udev daemon and tools without
10688 systemd running, like in initramfs or other init systems. Building
10689 udev though, will require the *build* of the systemd tree, but
10690 udev can be properly *run* without systemd.
10692 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
10693 should be used to create dead device nodes as workarounds for broken
10696 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
10697 no longer supported. udev_monitor_new_from_netlink() needs to be
10698 used to subscribe to events.
10700 * udev: when udevd is started by systemd, processes which are left
10701 behind by forking them off of udev rules, are unconditionally cleaned
10702 up and killed now after the event handling has finished. Services or
10703 daemons must be started as systemd services. Services can be
10704 pulled-in by udev to get started, but they can no longer be directly
10705 forked by udev rules.
10707 * udev: the daemon binary is called systemd-udevd now and installed
10708 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
10709 to adapt to that, create symlink, or rename the binary after building
10712 * libudev no longer provides these symbols:
10713 udev_monitor_from_socket()
10714 udev_queue_get_failed_list_entry()
10715 udev_get_{dev,sys,run}_path()
10716 The versions number was bumped and symbol versioning introduced.
10718 * systemd-loginctl and systemd-journalctl have been renamed
10719 to loginctl and journalctl to match systemctl.
10721 * The config files: /etc/systemd/systemd-logind.conf and
10722 /etc/systemd/systemd-journald.conf have been renamed to
10723 logind.conf and journald.conf. Package updates should rename
10724 the files to the new names on upgrade.
10726 * For almost all files the license is now LGPL2.1+, changed
10727 from the previous GPL2.0+. Exceptions are some minor stuff
10728 of udev (which will be changed to LGPL2.1 eventually, too),
10729 and the MIT licensed sd-daemon.[ch] library that is suitable
10730 to be used as drop-in files.
10732 * systemd and logind now handle system sleep states, in
10733 particular suspending and hibernating.
10735 * logind now implements a sleep/shutdown/idle inhibiting logic
10736 suitable for a variety of uses. Soonishly Lennart will blog
10737 about this in more detail.
10739 * var-run.mount and var-lock.mount are no longer provided
10740 (which previously bind mounted these directories to their new
10741 places). Distributions which have not converted these
10742 directories to symlinks should consider stealing these files
10743 from git history and add them downstream.
10745 * We introduced the Documentation= field for units and added
10746 this to all our shipped units. This is useful to make it
10747 easier to explore the boot and the purpose of the various
10750 * All smaller setup units (such as
10751 systemd-vconsole-setup.service) now detect properly if they
10752 are run in a container and are skipped when
10753 appropriate. This guarantees an entirely noise-free boot in
10754 Linux container environments such as systemd-nspawn.
10756 * A framework for implementing offline system updates is now
10757 integrated, for details see:
10758 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
10760 * A new service type Type=idle is available now which helps us
10761 avoiding ugly interleaving of getty output and boot status
10764 * There's now a system-wide CapabilityBoundingSet= option to
10765 globally reduce the set of capabilities for the
10766 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
10767 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
10768 even CAP_NET_ADMIN system-wide for secure systems.
10770 * There are now system-wide DefaultLimitXXX= options to
10771 globally change the defaults of the various resource limits
10772 for all units started by PID 1.
10774 * Harald Hoyer's systemd test suite has been integrated into
10775 systemd which allows easy testing of systemd builds in qemu
10776 and nspawn. (This is really awesome! Ask us for details!)
10778 * The fstab parser is now implemented as generator, not inside
10781 * systemctl will now warn you if .mount units generated from
10782 /etc/fstab are out of date due to changes in fstab that
10783 have not been read by systemd yet.
10785 * systemd is now suitable for usage in initrds. Dracut has
10786 already been updated to make use of this. With this in place
10787 initrds get a slight bit faster but primarily are much
10788 easier to introspect and debug since "systemctl status" in
10789 the host system can be used to introspect initrd services,
10790 and the journal from the initrd is kept around too.
10792 * systemd-delta has been added, a tool to explore differences
10793 between user/admin configuration and vendor defaults.
10795 * PrivateTmp= now affects both /tmp and /var/tmp.
10797 * Boot time status messages are now much prettier and feature
10798 proper english language. Booting up systemd has never been
10801 * Read-ahead pack files now include the inode number of all
10802 files to pre-cache. When the inode changes the pre-caching
10803 is not attempted. This should be nicer to deal with updated
10804 packages which might result in changes of read-ahead
10807 * We now temporaritly lower the kernel's read_ahead_kb variable
10808 when collecting read-ahead data to ensure the kernel's
10809 built-in read-ahead does not add noise to our measurements
10810 of necessary blocks to pre-cache.
10812 * There's now RequiresMountsFor= to add automatic dependencies
10813 for all mounts necessary for a specific file system path.
10815 * MountAuto= and SwapAuto= have been removed from
10816 system.conf. Mounting file systems at boot has to take place
10819 * nspawn now learned a new switch --uuid= to set the machine
10820 ID on the command line.
10822 * nspawn now learned the -b switch to automatically search
10823 for an init system.
10825 * vt102 is now the default TERM for serial TTYs, upgraded from
10828 * systemd-logind now works on VT-less systems.
10830 * The build tree has been reorganized. The individual
10831 components now have directories of their own.
10833 * A new condition type ConditionPathIsReadWrite= is now available.
10835 * nspawn learned the new -C switch to create cgroups for the
10836 container in other hierarchies.
10838 * We now have support for hardware watchdogs, configurable in
10841 * The scheduled shutdown logic now has a public API.
10843 * We now mount /tmp as tmpfs by default, but this can be
10844 masked and /etc/fstab can override it.
10846 * Since udisks does not make use of /media anymore we are not
10847 mounting a tmpfs on it anymore.
10849 * journalctl gained a new --local switch to only interleave
10850 locally generated journal files.
10852 * We can now load the IMA policy at boot automatically.
10854 * The GTK tools have been split off into a systemd-ui.
10856 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
10857 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
10858 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
10859 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
10860 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
10861 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
10862 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
10863 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
10864 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
10869 * This is mostly a bugfix release
10871 * Support optional initialization of the machine ID from the
10872 KVM or container configured UUID.
10874 * Support immediate reboots with "systemctl reboot -ff"
10876 * Show /etc/os-release data in systemd-analyze output
10878 * Many bugfixes for the journal, including endianness fixes and
10879 ensuring that disk space enforcement works
10881 * sd-login.h is C++ compatible again
10883 * Extend the /etc/os-release format on request of the Debian
10886 * We now refuse non-UTF8 strings used in various configuration
10887 and unit files. This is done to ensure we do not pass invalid
10888 data over D-Bus or expose it elsewhere.
10890 * Register Mimo USB Screens as suitable for automatic seat
10893 * Read SELinux client context from journal clients in a race
10896 * Reorder configuration file lookup order. /etc now always
10897 overrides /run in order to allow the administrator to always
10898 and unconditionally override vendor-supplied or
10899 automatically generated data.
10901 * The various user visible bits of the journal now have man
10902 pages. We still lack man pages for the journal API calls
10905 * We now ship all man pages in HTML format again in the
10908 Contributions from: Dave Reisner, Dirk Eibach, Frederic
10909 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
10910 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
10915 * This is mostly a bugfix release
10917 * systems lacking /etc/os-release are no longer supported.
10919 * Various functionality updates to libsystemd-login.so
10921 * Track class of PAM logins to distinguish greeters from
10922 normal user logins.
10924 Contributions from: Kay Sievers, Lennart Poettering, Michael
10929 * This is an important bugfix release for v41.
10931 * Building man pages is now optional which should be useful
10932 for those building systemd from git but unwilling to install
10935 * Watchdog support for supervising services is now usable. In
10936 a future release support for hardware watchdogs
10937 (i.e. /dev/watchdog) will be added building on this.
10939 * Service start rate limiting is now configurable and can be
10940 turned off per service. When a start rate limit is hit a
10941 reboot can automatically be triggered.
10943 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
10945 Contributions from: Benjamin Franzke, Bill Nottingham,
10946 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
10947 Schmidt, Michał Górny, Piotr Drąg
10951 * The systemd binary is installed /usr/lib/systemd/systemd now;
10952 An existing /sbin/init symlink needs to be adapted with the
10955 * The code that loads kernel modules has been ported to invoke
10956 libkmod directly, instead of modprobe. This means we do not
10957 support systems with module-init-tools anymore.
10959 * Watchdog support is now already useful, but still not
10962 * A new kernel command line option systemd.setenv= is
10963 understood to set system wide environment variables
10964 dynamically at boot.
10966 * We now limit the set of capabilities of systemd-journald.
10968 * We now set SIGPIPE to ignore by default, since it only is
10969 useful in shell pipelines, and has little use in general
10970 code. This can be disabled with IgnoreSIPIPE=no in unit
10973 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
10974 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
10979 * This is mostly a bugfix release
10981 * We now expose the reason why a service failed in the
10982 "Result" D-Bus property.
10984 * Rudimentary service watchdog support (will be completed over
10985 the next few releases.)
10987 * When systemd forks off in order execute some service we will
10988 now immediately changes its argv[0] to reflect which process
10989 it will execute. This is useful to minimize the time window
10990 with a generic argv[0], which makes bootcharts more useful
10992 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
10993 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
10994 Mike Kazantsev, Ray Strode
10998 * This is mostly a test release, but incorporates many
11001 * New systemd-cgtop tool to show control groups by their
11004 * Linking against libacl for ACLs is optional again. If
11005 disabled, support tracking device access for active logins
11006 goes becomes unavailable, and so does access to the user
11007 journals by the respective users.
11009 * If a group "adm" exists, journal files are automatically
11010 owned by them, thus allow members of this group full access
11011 to the system journal as well as all user journals.
11013 * The journal now stores the SELinux context of the logging
11014 client for all entries.
11016 * Add C++ inclusion guards to all public headers
11018 * New output mode "cat" in the journal to print only text
11019 messages, without any meta data like date or time.
11021 * Include tiny X server wrapper as a temporary stop-gap to
11022 teach XOrg udev display enumeration. This is used by display
11023 managers such as gdm, and will go away as soon as XOrg
11024 learned native udev hotplugging for display devices.
11026 * Add new systemd-cat tool for executing arbitrary programs
11027 with STDERR/STDOUT connected to the journal. Can also act as
11028 BSD logger replacement, and does so by default.
11030 * Optionally store all locally generated coredumps in the
11031 journal along with meta data.
11033 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
11034 writing short strings to files (for usage for /sys), and for
11035 creating symlinks, character and block device nodes.
11037 * New unit file option ControlGroupPersistent= to make cgroups
11038 persistent, following the mechanisms outlined in
11039 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
11041 * Support multiple local RTCs in a sane way
11043 * No longer monopolize IO when replaying readahead data on
11044 rotating disks, since we might starve non-file-system IO to
11045 death, since fanotify() will not see accesses done by blkid,
11048 * Do not show kernel threads in systemd-cgls anymore, unless
11049 requested with new -k switch.
11051 Contributions from: Dan Horák, Kay Sievers, Lennart
11052 Poettering, Michal Schmidt
11056 * This is mostly a test release, but incorporates many
11059 * The git repository moved to:
11060 git://anongit.freedesktop.org/systemd/systemd
11061 ssh://git.freedesktop.org/git/systemd/systemd
11063 * First release with the journal
11064 http://0pointer.de/blog/projects/the-journal.html
11066 * The journal replaces both systemd-kmsg-syslogd and
11067 systemd-stdout-bridge.
11069 * New sd_pid_get_unit() API call in libsystemd-logind
11071 * Many systemadm clean-ups
11073 * Introduce remote-fs-pre.target which is ordered before all
11074 remote mounts and may be used to start services before all
11077 * Added Mageia support
11079 * Add bash completion for systemd-loginctl
11081 * Actively monitor PID file creation for daemons which exit in
11082 the parent process before having finished writing the PID
11083 file in the daemon process. Daemons which do this need to be
11084 fixed (i.e. PID file creation must have finished before the
11085 parent exits), but we now react a bit more gracefully to them.
11087 * Add colourful boot output, mimicking the well-known output
11088 of existing distributions.
11090 * New option PassCredentials= for socket units, for
11091 compatibility with a recent kernel ABI breakage.
11093 * /etc/rc.local is now hooked in via a generator binary, and
11094 thus will no longer act as synchronization point during
11097 * systemctl list-unit-files now supports --root=.
11099 * systemd-tmpfiles now understands two new commands: z, Z for
11100 relabelling files according to the SELinux database. This is
11101 useful to apply SELinux labels to specific files in /sys,
11102 among other things.
11104 * Output of SysV services is now forwarded to both the console
11105 and the journal by default, not only just the console.
11107 * New man pages for all APIs from libsystemd-login.
11109 * The build tree got reorganized and the build system is a
11110 lot more modular allowing embedded setups to specifically
11111 select the components of systemd they are interested in.
11113 * Support for Linux systems lacking the kernel VT subsystem is
11116 * configure's --with-rootdir= got renamed to
11117 --with-rootprefix= to follow the naming used by udev and
11120 * Unless specified otherwise we will now install to /usr instead
11121 of /usr/local by default.
11123 * Processes with '@' in argv[0][0] are now excluded from the
11124 final shut-down killing spree, following the logic explained
11126 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
11128 * All processes remaining in a service cgroup when we enter
11129 the START or START_PRE states are now killed with
11130 SIGKILL. That means it is no longer possible to spawn
11131 background processes from ExecStart= lines (which was never
11132 supported anyway, and bad style).
11134 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
11135 reloading of units together.
11137 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
11138 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
11139 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
11140 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
11141 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek