]> git.ipfire.org Git - thirdparty/systemd.git/blame - NEWS
Merge pull request #16061 from filbranden/standalone1
[thirdparty/systemd.git] / NEWS
CommitLineData
d657c51f 1systemd System and Service Manager
220a21d3 2
6635f57d 3CHANGES WITH 246 in spe:
b0d0e0ef
LP
4
5 * The various programs included in systemd can now optionally output
6 their log messages on stderr prefixed with a timestamp, controlled by
7 the $SYSTEMD_LOG_TIME environment variable.
8
9 * A new boolean kernel command line option systemd.swap= has been
10 added, which may be used to turn off automatic activation of swap
11 devices, as listed in /etc/fstab.
12
13 * The CPUAffinity= setting in service unit files now supports a new
14 special value "numa". If used, the NUMA mask is copied into the CPU
15 affinity mask.
16
17 * The man pages for the sd-bus and sd-hwdb APIs have been completed.
18
19 * networkctl gained the new "forcerenew" command for forcing all DHCP
20 server clients to renew their lease. The interface "status" output
21 will now show numerous additional fields of information about an
22 interface. There are new "up" and "down" commands to bring specific
23 interfaces up or down.
24
25 * systemd-networkd's [IPv6Prefix] section in .network files gained a
26 new boolean setting Assign=. If enabled an address from the prefix is
27 automatically assigned to the interface.
28
29 * systemd-networkd's [Network] section gained a new setting
30 IPv6PDSubnetId= that allows explicit configuration of the preferred
31 subnet that networkd's Prefix Delegation logic assigns to an
32 interfaces.
33
34 * systemd-networkd gained support for configuring the HTB queuing
35 discipline in the [HierarchyTokenBucket] and
36 [HierarchyTokenBucketClass] sections. Similar the "pfifo" qdisc may
37 be configured in the [PFIFO] section, "GRED" in
38 [GenericRandomEarlyDetection], "SFB" in [StochasticFairBlue], "cake"
39 in [CAKE], "PIE" in [PIE], "DRR" in [DeficitRoundRobinScheduler] and
40 [DeficitRoundRobinSchedulerClass], "BFIFO" in [BFIFO],
41 "PFIFOHeadDrop" in [PFIFOHeadDrop], "PFIFOFast" in [PFIFOFast] and
42 "HHF" in [HeavyHitterFilter].
43
44 * systemd-networkd gained support for a new Termination= setting in the
45 [CAN] section for configuring the termination resistor. It also
46 gained a new ListenOnly= setting for controlling whether to only
47 listen on CAN interfaces, without interfering with traffic otherwise
48 (which is useful for debugging/monitoring CAN network
49 traffic). DataBitRate=, DataSamplePoint=, FDMode=, FDNonISO= have
50 been added to configure various CAN-FD aspects.
51
52 * .link files managed by systemd-udevd gained options RxFlowControl=,
53 TxFlowControl=, AutoNegotiationFlowControl= in the [Link] section, in
54 order to configure various flow control parameters. They also gained
55 RxMiniBufferSize= and RxJumboBufferSize= in order to configure jumbo
56 frame ring buffer sizes.
57
58 * systemd-networkd's [DHCPv6] section gained a new WithoutRA= boolean
59 setting. If enabled, DHCPv6 will be attempted right-away without
60 requiring an Router Advertisement packet suggesting it
61 first. Conversely, the [IPv6AcceptRA] gained a boolean option
62 DHCPv6Client= that may be used to turn off the DHCPv6 client even if
63 the RA packets suggest it.
64
65 * systemd-networkd's [DHCPv4] section gained a new setting UseGateway=
66 which may be used to turn off use of the gateway information provided
67 by the DHCP lease. A new FallbackLeaseLifetimeSec= setting may be
68 used to configure how to process leases that lack a lifetime option.
69
70 * systemd-networkd's [DHCPv4] and [DHCPServer] sections gained a new
71 setting SendVendorOption= allowing configuration of additional vendor
72 options to send in the DHCP requests/responses. The [DHCPv6] section
73 gained a new SendOption= setting for sending arbitrary DHCP
74 options. RequestOptions= has been added to request arbitrary options
75 from the server. UserClass= has been added to set the DHCP user class
76 field.
77
78 * systemd-networkd's [DHCPServer] section gained a new set of options
79 POP3Servers=, SMTPServers=, LPRServers= for including server
80 information about these three protocols in the DHCP lease. It also
81 gained support for including "MUD" URLs ("Manufacturer Usage
82 Description"). Support for "MUD" URLs was also added to the LLDP
83 stack, configurable in the [LLDP] section in .network files.
84
85 * systemd-resolved's DNS= configuration option now optionally accepts
86 DNS server addresses suffixed by "#" followed by a host name. If
87 used, the DNS-over-TLS certificate is validated to match the
88 specified hostname.
89
6635f57d
ZJS
90 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
91 systemd-coredump to save core files for suid processes. When saving
92 the core file, systemd-coredump will use the effective uid and gid of
93 the process that faulted.
94
1cabd2d0
JW
95 * "systemctl list-units" and "systemctl list-machines" no longer hide
96 their first output column with --no-legend. To hide the first column,
97 use --plain.
98
d9e45bc3
MS
99 * The service manager gained basic support for cgroup v2 freezer. Units
100 can now be suspended or resumed either using new systemctl verbs,
101 freeze and thaw respectively, or via D-Bus.
102
e2099267
MS
103 * systemd-udevd gained new configuration option timeout_signal= as well
104 as coresponding kernel command line option udev.timeout_signal.
105 The option can be used to configure the UNIX signal that the main
106 daemon sends to the worker processes on timeout.
107
b0d0e0ef
LP
108 * A new sd-path.h API has been added to libsystemd. It provides a
109 simple API for retrieving various search paths and primary
110 directories for various resources.
111
112 * The sd-bus API gained a number of convenience functions that take
113 va_list arguments rather than "...". For example, there's now
114 sd_bus_call_methodv() to match sd_bus_call_method(). Previously,
115 these were missing since the calls are convenience calls only and
116 could be put together from the more low-level functions they build
117 on.
118
119 * sd-bus vtable entries learnt a new flag SD_BUS_VTABLE_ABSOLUTE_OFFSET
120 which alters how the userdata pointer to pass to the callbacks is
121 determined. If the flag is set the offset field is converted as-is
122 into a pointer, without adding it to the object pointer the vtable is
123 associated with.
124
1d16f661 125 * sd-bus now exposes four new functions:
b0d0e0ef
LP
126 sd_bus_interface_name_is_valid() + sd_bus_service_name_is_valid() +
127 sd_bus_member_name_is_valid() + sd_bus_object_path_is_valid() will
128 validate strings to check if they qualify as various D-Bus concepts.
129
1d16f661 130 * The sd-bus API gained the SD_BUS_METHOD_WITH_ARGS(),
366d0772
DDM
131 SD_BUS_METHOD_WITH_ARGS_OFFSET() and SD_BUS_SIGNAL_WITH_ARGS() macros
132 that simplify adding argument names to D-Bus methods and signals.
133
b0d0e0ef
LP
134 * systemctl gained a new "-P" switch that is a shortcut for "--value
135 --property=…".
136
137 * The expectations on user/group name syntax are now documented in
138 detail; documentation how classic home directories may be converted
139 into home directories managed by homed has been added; documentation
140 regarding integration of homed/userdb functionality in desktops has
141 been added:
142
143 https://systemd.io/USER_NAMES
144 https://systemd.io/CONVERTING_TO_HOMED
145 https://systemd.io/USERDB_AND_DESKTOPS
146
147 * systemd-run gained a new switch --slice-inherit. If specified the
148 unit it generates is placed in the same slice as the systemd-run
149 process itself.
150
151 * service unit files now accept a new setting CoredumpFilter= which
152 allows configuration of the memory sections coredumps of the
153 service's processes shall include.
154
155 * coredumpctl gained a new --file= switch, matching the same one in
156 journalctl: a specific journal file may be specified to read the
157 coredump data from.
158
159 * Various D-Bus APIs of systemd daemons now have man pages that
160 document the methods, signals and properties.
161
162 * journald.conf gained a new boolean setting Audit= that may be used to
163 control whether systemd-journald will enable audit during
164 initialization.
165
166 * A new default .network file is now shipped that matches TUN/TAP
167 devices that begin with "vt-" in their name. Such interfaces will
168 have IP routing onto the host links set up automatically. This is
169 supposed to be used by VM managers to trivially acquire a network
170 interface which is fully set up for host communication, simply by
171 carefully picking an interface name to use.
172
173 * All D-Bus services shipped in systemd now implement the generic
174 LogControl1 D-Bus API which allows clients to change log level +
175 target of the service during runtime.
176
177 * systemd-nspawn's --resolv-conf= switch gained a number of new
178 supported values. Specifically, options starting with "replace-" are
179 like those prefixed "copy-" but replace any existing resolv.conf
180 file. And options ending in "-uplink" and "-stub" can now be used to
181 propagate other flavours of resolv.conf into the container (as
182 defined by systemd-resolved).
183
184 * systemd-binfmt gained a new switch --unregister for unregistering all
185 registered entries at once. This is now invoked automatically at
186 shutdown, so that binary formats registered with the "F" flag will
187 not block clean file system unmounting.
188
189 * Unit files, tmpfiles.d/ snippets, sysusers.d/ snippets and other
190 configuration files that support specifier expansion learnt six new
191 specifiers: %a resolves to the current architecture, %o/%w/%B/%W
192 resolve to the various ID fields from /etc/os-release, %l resolves to
193 the "short" hostname of the system, i.e. the kernel configured
194 hostname, truncated at the first dot.
195
196 * systemd-notify's --pid= switch gained new values: "parent", "self",
1d16f661 197 "auto" for controlling which PID to send to the service manager: the
b0d0e0ef
LP
198 systemd-notify process' PID, or the one of the process invoking it.
199
200 * When sending a file descriptor (fd) to the service manager to keep
201 track of, using the sd_notify() mechanism, a new parameter FDPOLL=0
202 may be specified. If passed the service manager will refrain from
203 poll()ing on the file descriptor. Traditionally (and when the
204 parameter is not specified), the service manager will poll it for
205 POLLHUP or POLLERR events, and immediately close the fds in that
206 case.
207
208 * A new call sd_notify_barrier() has been added to the sd-daemon.h
209 API. The call will block until all previously sent sd_notify()
210 messages have been processed by the service manager. This is useful
211 to remove races caused by a process already having disappeared at the
212 time a notification message is processed by the service manager,
213 making correct attribution impossible. The systemd-notify tool will
214 now make use of this call implicitly, but this can be turned off again
215 via the new --no-block switch.
216
217 * systemd-logind's Session bus object learnt a new method call
218 SetType() for temporarily updating the session type of an already
219 allocated session. This is useful for upgrading tty sessions to
220 graphical ones once a compositor is invoked.
221
1d16f661 222 * .mount units gained a new ReadWriteOnly= boolean option. If set
b0d0e0ef 223 it will not be attempted to mount a file system read-only if mounting
1d16f661 224 in read-write mode doesn't succeed. An option x-systemd.rw-only is
b0d0e0ef
LP
225 available in /etc/fstab to control the same.
226
227 * coredumps collected by systemd-coredump may now be compressed using
228 the zstd algorithm.
229
230 * journalctl's "-o cat" output mode will now show one or more journal
231 fields specified with --output-fields= instead of unconditionally
232 MESSAGE=. This is useful to retrieve a very specific set of fields
233 without any decoration.
234
235 * systemd-socket-proxy gained a new switch --exit-idle-time= for
236 configuring an exit-on-idle time.
237
238 * systemd-homed's LUKS backend gained the ability to discard empty file
239 system blocks automatically when the user logs out. This is enabled
240 by default to ensure that home directories take minimal space when
241 logged out but get full size guarantees when logged in. This may be
242 controlled with the new --luks-offline-discard= switch to homectl.
243
244 * If systemd-homed detects that /home/ is encrypted as a whole it will
245 now default to the directory or subvolume backends instead of the
246 LUKS backend, in order to avoid double encryption. The default
247 storage and file system may now be configured explicitly, too, via
248 the new /etc/systemd/homed.conf configuration file.
249
250 * when systemd-journald's log stream is broken up into multiple lines
251 because the PID of the sender changed this is indicated in the
252 generated log records via the _LINE_BREAK=pid-change field.
253
254 * systemd-networkd's .netdev files now support a new setting
255 VLANProtocol= in the [Bridge] section that allows configuration of
256 the VLAN protocol to use.
257
258 * systemd-repart's --empty= setting gained a new value "create". If
259 specified a new empty regular disk image file is created under the
1d16f661 260 specified name. Its size may be specified with the new --size=
b0d0e0ef
LP
261 option. The latter is also supported without the "create" mode, in
262 order to grow existing disk image files to the specified size. These
1d16f661
FS
263 two new options are useful when creating or manipulating disk images
264 instead of operating on actual block devices.
b0d0e0ef
LP
265
266 * systemd-repart drop-ins now support a new UUID= setting to control
267 the UUID to assign to a newly created partition.
268
269 * StandardError= and StandardOutput= in unit files no longer support
270 the "syslog" and "syslog-console" switches. They were long removed
271 from the documentation, but will now result in warnings when used,
272 and be converted to "journal" and "journal+console" automatically.
273
274 * systemd-networkd supports a new Group= setting in the [Link] section
275 of the .network files, to control the link group.
276
277 * Two new unit file settings
278 ConditionPathIsEncrypted=/AssertPathIsEncrypted= have been
279 added. They may be used to check whether a specific file system path
280 resides on a block device that is encrypted on the block level
281 (i.e. using dm-crypt/LUKS).
282
283 * Another pair of new settings ConditionEnvironment=/AssertEnvironment=
284 has been added that may be used for simple environment checks. This
285 is particularly useful when passing in environment variables from a
286 container manager (or from PAM in case of the systemd --user
287 instance).
288
289 * The /sys/module/kernel/parameters/crash_kexec_post_notifiers file is
1d16f661 290 now automatically set to "Y" at boot, in order to enable pstore
b0d0e0ef
LP
291 generation for collection with systemd-pstore.
292
293 * New kernel command line options systemd.condition-needs-update= and
294 systemd.condition-first-boot= have been added, which override the
295 result of the ConditionNeedsUpdate= and ConditionFirstBoot=
296 conditions.
297
298 * A new kernel command line option systemd.clock-usec= has been added
299 that allows setting the system clock to the specified time in µs
300 since Jan 1st, 1970 early during boot. This is in particular useful
301 in order to make test cases more reliable.
302
303 * A new kernel command line option systemd.hostname= has been added
304 that allows controlling the hostname that is initialized early during
305 boot.
306
307 * The /etc/crypttab tmp option now optionally takes an argument
308 selecting the file system to use. Moreover, the default is now
309 changed from ext2 to ext4.
310
311 * There's a new /etc/crypttab option "keyfile-erase". If specified the
312 key file listed in the same line is removed after use, regardless if
313 volume activation was successful or not. This is useful if the key
314 file is only acquired transiently at runtime and shall be erased
315 before the system continues to boot.
316
317 * There's also a new /etc/crypttab option "try-empty-password". If
318 specified, before asking the user for a password it is attempted to
319 unlock the volume with an empty password. This is useful for
320 installing encrypted images whose password shall be set on first boot
321 instead of at installation time.
322
323 * systemd-cryptsetup will now attempt to load the keys to unlock
324 volumes with automatically from files in
325 /etc/cryptsetup-keys.d/<volume>.key and
326 /run/cryptsetup-keys.d/<volume>.key, if any of these files exist.
327
328 * logind.conf gained a new RuntimeDirectoryInodesMax= setting to
329 control the inode limit for the per-user $XDG_RUNTIME_DIR tmpfs
330 instance.
331
332 * systemd-firstboot gained a new --root-password-hashed= parameter for
333 setting the root user's password as UNIX password hash. There's a new
334 --delete-root-password switch which instead of setting a password for
335 the root user, removes it so that log-in without a password is
336 permitted. There's now --force which if specified means any existing
337 configuration is overwritten by the specified settings. It also
338 gained a new --kernel-command-line= parameter which may be used to
339 set the /etc/kernel/cmdline file of an OS image.
340
341 * A new generator systemd-xdg-autostart-generator has been added. It
342 automatically generates systemd unit files from XDG autostart
343 .desktop files, and is useful for allowing systemd to manage services
344 defined that way safely and automatically.
345
346 * systemd will now log about all left-over processes remaining in a
347 unit when the unit is stopped. It will now warn about services using
348 KillMode=none, as this is generally an unsafe thing to make use of.
349
47373451
LP
350 * .socket units gained a new boolean setting PassPacketInfo=. If
351 enabled, the kernel will attach additional per-packet metadata to all
352 packets read from the socket, as ancillary message. This controls the
353 IP_PKTINFO, IPV6_RECVPKTINFO, NETLINK_PKTINFO socket options,
354 depending on socket type.
355
356 * A new boolean option AssignAcquiredDelegatedPrefixAddress= has been
357 added to the [DHCPv6] section of .network files. If enabled (which is
1d16f661 358 the default) an address from any acquired delegated prefix is
47373451
LP
359 automatically chosen and assigned to the interface.
360
dae710be 361 * "systemctl reboot" takes the option "--reboot-argument=".
362 The optional positional argument to "systemctl reboot" is now
363 being deprecated in favor of this option.
364
7ade8982
LP
365 * Support for the .include syntax in unit files has been removed. The
366 concept has been obsolete for 6 years and we started warning about
367 its pending removal 2 years ago (also see NEWS file below). It's
368 finally gone now.
369
901d1ce8 370CHANGES WITH 245:
723822f0 371
68410195 372 * A new tool "systemd-repart" has been added, that operates as an
2ad98889
ZJS
373 idempotent declarative repartitioner for GPT partition tables.
374 Specifically, a set of partitions that must or may exist can be
375 configured via drop-in files, and during every boot the partition
376 table on disk is compared with these files, creating missing
377 partitions or growing existing ones based on configurable relative
378 and absolute size constraints. The tool is strictly incremental,
379 i.e. does not delete, shrink or move partitions, but only adds and
380 grows them. The primary use-case is OS images that ship in minimized
381 form, that on first boot are grown to the size of the underlying
382 block device or augmented with additional partitions. For example,
383 the root partition could be extended to cover the whole disk, or a
384 swap or /home partitions could be added on first boot. It can also be
385 used for systems that use an A/B update scheme but ship images with
386 just the A partition, with B added on first boot. The tool is
387 primarily intended to be run in the initrd, shortly before
388 transitioning into the host OS, but can also be run after the
389 transition took place. It automatically discovers the disk backing
390 the root file system, and should hence not require any additional
391 configuration besides the partition definition drop-ins. If no
392 configuration drop-ins are present, no action is taken.
68410195
LP
393
394 * A new component "userdb" has been added, along with a small daemon
395 "systemd-userdb.service" and a client tool "userdbctl". The framework
396 allows defining rich user and group records in a JSON format,
397 extending on the classic "struct passwd" and "struct group"
398 structures. Various components in systemd have been updated to
399 process records in this format, including systemd-logind and
400 pam-systemd. The user records are intended to be extensible, and
401 allow setting various resource management, security and runtime
402 parameters that shall be applied to processes and sessions of the
403 user as they log in. This facility is intended to allow associating
404 such metadata directly with user/group records so that they can be
405 produced, extended and consumed in unified form. We hope that
406 eventually frameworks such as sssd will generate records this way, so
407 that for the first time resource management and various other
408 per-user settings can be configured in LDAP directories and then
409 provided to systemd (specifically to systemd-logind and pam-system)
2ad98889 410 to apply on login. For further details see:
68410195
LP
411
412 https://systemd.io/USER_RECORD
413 https://systemd.io/GROUP_RECORD
414 https://systemd.io/USER_GROUP_API
415
9a4940bf 416 * A small new service systemd-homed.service has been added, that may be
2ad98889
ZJS
417 used to securely manage home directories with built-in encryption.
418 The complete user record data is unified with the home directory,
419 thus making home directories naturally migratable. Its primary
420 back-end is based on LUKS volumes, but fscrypt, plain directories,
421 and other storage schemes are also supported. This solves a couple of
422 problems we saw with traditional ways to manage home directories, in
423 particular when it comes to encryption. For further discussion of
424 this, see the video of Lennart's talk at AllSystemsGo! 2019:
9a4940bf
LP
425
426 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
427
428 For further details about the format and expectations on home
429 directories this new daemon makes, see:
430
431 https://systemd.io/HOME_DIRECTORY
432
433 * systemd-journald is now multi-instantiable. In addition to the main
434 instance systemd-journald.service there's now a template unit
2ad98889
ZJS
435 systemd-journald@.service, with each instance defining a new named
436 log 'namespace' (whose name is specified via the instance part of the
437 unit name). A new unit file setting LogNamespace= has been added,
438 taking such a namespace name, that assigns services to the specified
439 log namespaces. As each log namespace is serviced by its own
440 independent journal daemon, this functionality may be used to improve
441 performance and increase isolation of applications, at the price of
442 losing global message ordering. Each instance of journald has a
443 separate set of configuration files, with possibly different disk
444 usage limitations and other settings.
445
446 journalctl now takes a new option --namespace= to show logs from a
447 specific log namespace. The sd-journal.h API gained
448 sd_journal_open_namespace() for opening the log stream of a specific
449 log namespace. systemd-journald also gained the ability to exit on
450 idle, which is useful in the context of log namespaces, as this means
451 log daemons for log namespaces can be activated automatically on
452 demand and will stop automatically when no longer used, minimizing
453 resource usage.
9a4940bf 454
723822f0 455 * When systemd-tmpfiles copies a file tree using the 'C' line type it
2ad98889 456 will now label every copied file according to the SELinux database.
723822f0
LP
457
458 * When systemd/PID 1 detects it is used in the initrd it will now boot
459 into initrd.target rather than default.target by default. This should
460 make it simpler to build initrds with systemd as for many cases the
461 only difference between a host OS image and an initrd image now is
2ad98889 462 the presence of the /etc/initrd-release file.
723822f0
LP
463
464 * A new kernel command line option systemd.cpu_affinity= is now
465 understood. It's equivalent to the CPUAffinity= option in
466 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
2ad98889 467 itself and the default for all other processes.
723822f0 468
2ad98889
ZJS
469 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
470 equivalent), the SELinux database is now reloaded, ensuring that
723822f0
LP
471 sockets and other file system objects are generated taking the new
472 database into account.
473
d54bb638
ZJS
474 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
475 "quiet" has been changed to imply that instead of
476 "systemd.show-status=auto". In this mode, only messages about errors
477 and significant delays in boot are shown on the console.
478
2ad98889 479 * The sd-event.h API gained native support for the new Linux "pidfd"
723822f0
LP
480 concept. This permits watching processes using file descriptors
481 instead of PID numbers, which fixes a number of races and makes
2ad98889 482 process supervision more robust and efficient. All of systemd's
723822f0
LP
483 components will now use pidfds if the kernel supports it for process
484 watching, with the exception of PID 1 itself, unfortunately. We hope
485 to move PID 1 to exclusively using pidfds too eventually, but this
486 requires some more kernel work first. (Background: PID 1 watches
487 processes using waitid() with the P_ALL flag, and that does not play
488 together nicely with pidfds yet.)
489
490 * Closely related to this, the sd-event.h API gained two new calls
491 sd_event_source_send_child_signal() (for sending a signal to a
492 watched process) and sd_event_source_get_child_process_own() (for
2ad98889
ZJS
493 marking a process so that it is killed automatically whenever the
494 event source watching it is freed).
723822f0 495
60ed2dcf 496 * systemd-networkd gained support for configuring Token Bucket Filter
2ad98889
ZJS
497 (TBF) parameters in its qdisc configuration support. Similarly,
498 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
69f17347 499 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
723822f0
LP
500
501 * systemd-networkd gained support for Intermediate Functional Block
502 (IFB) network devices.
503
504 * systemd-networkd gained support for configuring multi-path IP routes,
505 using the new MultiPathRoute= setting in the [Route] section.
506
2ad98889
ZJS
507 * systemd-networkd's DHCPv4 client has been updated to support a new
508 SendDecline= option. If enabled, duplicate address detection is done
509 after a DHCP offer is received from the server. If a conflict is
510 detected, the address is declined. The DHCPv4 client also gained
723822f0
LP
511 support for a new RouteMTUBytes= setting that allows to configure the
512 MTU size to be used for routes generated from DHCPv4 leases.
513
514 * The PrefixRoute= setting in systemd-networkd's [Address] section of
515 .network files has been deprecated, and replaced by AddPrefixRoute=,
2ad98889 516 with its sense inverted.
723822f0
LP
517
518 * The Gateway= setting of [Route] sections of .network files gained
427928ca
ZJS
519 support for a special new value "_dhcp". If set, the configured
520 static route uses the gateway host configured via DHCP.
723822f0 521
573e58f6 522 * New User= and SuppressPrefixLength= settings have been implemented
f05c0615
YW
523 for the [RoutingPolicyRule] section of .network files to configure
524 source routing based on UID ranges and prefix length, respectively.
9a4940bf 525
2ad98889
ZJS
526 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
527 D-Bus message object as "sensitive". Those objects are erased from
528 memory when they are freed. This concept is intended to be used for
529 messages that contain security sensitive data. A new flag
530 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
531 in sd-bus vtables, causing any incoming and outgoing messages of
532 those methods to be implicitly marked as "sensitive".
723822f0 533
9a4940bf 534 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
2ad98889 535 contents of a message (or parts thereof) to standard output for
9a4940bf
LP
536 debugging purposes.
537
2ad98889
ZJS
538 * systemd-sysusers gained support for creating users with the primary
539 group named differently than the user.
9a4940bf 540
723822f0
LP
541 * systemd-resolved's DNS-over-TLS support gained SNI validation.
542
543 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
544 gained support for growing XFS partitions. Previously it supported
545 only ext4 and btrfs partitions.
546
547 * The support for /etc/crypttab gained a new x-initrd.attach option. If
2ad98889
ZJS
548 set, the specified encrypted volume is unlocked already in the
549 initrd. This concept corresponds to the x-initrd.mount option in
723822f0
LP
550 /etc/fstab.
551
552 * systemd-cryptsetup gained native support for unlocking encrypted
553 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
2ad98889 554 encryption of volumes to YubiKeys. This is exposed in the new
723822f0
LP
555 pkcs11-uri= option in /etc/crypttab.
556
68410195
LP
557 * The /etc/fstab support in systemd now supports two new mount options
558 x-systemd.{required,wanted}-by=, for explicitly configuring the units
559 that the specified mount shall be pulled in by, in place of
560 the usual local-fs.target/remote-fs.target.
561
723822f0
LP
562 * The https://systemd.io/ web site has been relaunched, directly
563 populated with most of the documentation included in the systemd
2ad98889
ZJS
564 repository. systemd also acquired a new logo, thanks to Tobias
565 Bernard.
723822f0
LP
566
567 * systemd-udevd gained support for managing "alternative" network
568 interface names, as supported by new Linux kernels. For the first
569 time this permits assigning multiple (and longer!) names to a network
570 interface. systemd-udevd will now by default assign the names
2ad98889
ZJS
571 generated via all supported naming schemes to each interface. This
572 may be further tweaked with .link files and the AlternativeName= and
573 AlternativeNamesPolicy= settings. Other components of systemd have
574 been updated to support the new alternative names wherever
575 appropriate. For example, systemd-nspawn will now generate
576 alternative interface names for the host-facing side of container
577 veth links based on the full container name without truncation.
723822f0
LP
578
579 * systemd-nspawn interface naming logic has been updated in another way
580 too: if the main interface name (i.e. as opposed to new-style
2ad98889
ZJS
581 "alternative" names) based on the container name is truncated, a
582 simple hashing scheme is used to give different interface names to
583 multiple containers whose names all begin with the same prefix. Since
584 this changes the primary interface names pointing to containers if
585 truncation happens, the old scheme may still be requested by
586 selecting an older naming scheme, via the net.naming-scheme= kernel
587 command line option.
723822f0 588
68410195
LP
589 * PrivateUsers= in service files now works in services run by the
590 systemd --user per-user instance of the service manager.
591
592 * A new per-service sandboxing option ProtectClock= has been added that
593 locks down write access to the system clock. It takes away device
2ad98889
ZJS
594 node access to /dev/rtc as well as the system calls that set the
595 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
596 Note that this option does not affect access to auxiliary services
597 that allow changing the clock, for example access to
598 systemd-timedated.
68410195
LP
599
600 * The systemd-id128 tool gained a new "show" verb for listing or
601 resolving a number of well-known UUIDs/128bit IDs, currently mostly
602 GPT partition table types.
723822f0
LP
603
604 * The Discoverable Partitions Specification has been updated to support
605 /var and /var/tmp partition discovery. Support for this has been
606 added to systemd-gpt-auto-generator. For details see:
607
608 https://systemd.io/DISCOVERABLE_PARTITIONS
609
610 * "systemctl list-unit-files" has been updated to show a new column
611 with the suggested enablement state based on the vendor preset files
612 for the respective units.
613
68410195
LP
614 * "systemctl" gained a new option "--with-dependencies". If specified
615 commands such as "systemctl status" or "systemctl cat" will now show
616 all specified units along with all units they depend on.
617
723822f0
LP
618 * networkctl gained support for showing per-interface logs in its
619 "status" output.
620
a100fe3c 621 * systemd-networkd-wait-online gained support for specifying the maximum
573e58f6
ZJS
622 operational state to wait for, and to wait for interfaces to
623 disappear.
624
723822f0
LP
625 * The [Match] section of .link and .network files now supports a new
626 option PermanentMACAddress= which may be used to check against the
627 permanent MAC address of a network device even if a randomized MAC
628 address is used.
629
69f17347
ZJS
630 * The [TrafficControlQueueingDiscipline] section in .network files has
631 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
632 dropped from the individual setting names.
68410195 633
823b0352
ZJS
634 * Any .link and .network files that have an empty [Match] section (this
635 also includes empty and commented-out files) will now be
636 rejected. systemd-udev and systemd-networkd started warning about
637 such files in version 243.
638
2ad98889 639 * systemd-logind will now validate access to the operation of changing
98ab0dae 640 the virtual terminal via a polkit action. By default, only users
2ad98889 641 with at least one session on a local VT are granted permission.
68410195 642
2ad98889
ZJS
643 * When systemd sets up PAM sessions that invoked service processes
644 shall run in, the pam_setcred() API is now invoked, thus permitting
645 PAM modules to set additional credentials for the processes.
68410195 646
d54bb638
ZJS
647 * portablectl attach/detach verbs now accept --now and --enable options
648 to combine attachment with enablement and invocation, or detachment
649 with stopping and disablement.
650
2ca17c78
ZJS
651 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
652 fixed, which in turn exposed bugs in unit configuration of services
653 which have Type=oneshot and should only run once, but do not have
654 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
655 service may be started again after exiting successfully, for example
656 as a dependency in another transaction. Affected services included
657 some internal systemd services (most notably
658 systemd-vconsole-setup.service, which was updated to have
659 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
660 plymouth has been suitably updated or patched before upgrading to
661 this systemd release. See
662 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
663 additional discussion.
664
9c4d3d79
ZJS
665 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
666 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
667 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
668 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
669 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
670 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
671 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
672 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
673 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
674 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
675 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
676 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
677 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
678 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
679 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
680 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
681 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
682 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
683 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
684 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
685 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
686 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
687 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
688 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
689 DONG
901d1ce8 690
03985d06 691 – Warsaw, 2020-03-06
723822f0 692
bdf2357c 693CHANGES WITH 244:
ee50dada
ZJS
694
695 * Support for the cpuset cgroups v2 controller has been added.
696 Processes may be restricted to specific CPUs using the new
697 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
698 AllowedMemoryNodes= setting.
699
700 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
1e904320 701 now be configured using a new RestartKillSignal= setting. This
ee50dada
ZJS
702 allows units which signals to request termination to implement
703 different behaviour when stopping in preparation for a restart.
704
705 * "systemctl clean" may now be used also for socket, mount, and swap
706 units.
707
708 * systemd will also read configuration options from the EFI variable
709 SystemdOptions. This may be used to configure systemd behaviour when
710 modifying the kernel command line is inconvenient, but configuration
711 on disk is read too late, for example for the options related to
2536752d 712 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
ee50dada
ZJS
713 set the EFI variable.
714
715 * systemd will now disable printk ratelimits in early boot. This should
716 allow us to capture more logs from the early boot phase where normal
717 storage is not available and the kernel ring buffer is used for
718 logging. Configuration on the kernel command line has higher priority
719 and overrides the systemd setting.
720
721 systemd programs which log to /dev/kmsg directly use internal
722 ratelimits to prevent runaway logging. (Normally this is only used
723 during early boot, so in practice this change has very little
724 effect.)
725
3e1db806
AZ
726 * Unit files now support top level dropin directories of the form
727 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
728 that affects all corresponding unit files.
ee50dada 729
852b7272
ZJS
730 * systemctl gained support for 'stop --job-mode=triggering' which will
731 stop the specified unit and any units which could trigger it.
732
733 * Unit status display now includes units triggering and triggered by
734 the unit being shown.
735
ee50dada
ZJS
736 * The RuntimeMaxSec= setting is now supported by scopes, not just
737 .service units. This is particularly useful for PAM sessions which
738 create a scope unit for the user login. systemd.runtime_max_sec=
739 setting may used with the pam_systemd module to limit the duration
740 of the PAM session, for example for time-limited logins.
741
852b7272
ZJS
742 * A new @pkey system call group is now defined to make it easier to
743 whitelist memory protection syscalls for containers and services
744 which need to use them.
745
746 * systemd-udevd: removed the 30s timeout for killing stale workers on
747 exit. systemd-udevd now waits for workers to finish. The hard-coded
748 exit timeout of 30s was too short for some large installations, where
749 driver initialization could be prematurely interrupted during initrd
750 processing if the root file system had been mounted and init was
751 preparing to switch root. If udevd is run without systemd and workers
752 are hanging while udevd receives an exit signal, udevd will now exit
753 when udev.event_timeout is reached for the last hanging worker. With
754 systemd, the exit timeout can additionally be configured using
755 TimeoutStopSec= in systemd-udevd.service.
756
ee50dada
ZJS
757 * udev now provides a program (fido_id) that identifies FIDO CTAP1
758 ("U2F")/CTAP2 security tokens based on the usage declared in their
759 report and descriptor and outputs suitable environment variables.
760 This replaces the externally maintained whitelists of all known
761 security tokens that were used previously.
762
763 * Automatically generated autosuspend udev rules for whitelisted
764 devices have been imported from the Chromium OS project. This should
b7db8b7b 765 improve power saving with many more devices.
ee50dada
ZJS
766
767 * udev gained a new "CONST{key}=value" setting that allows matching
768 against system-wide constants without forking a helper binary.
769 Currently "arch" and "virt" keys are supported.
770
771 * udev now opens CDROMs in non-exclusive mode when querying their
772 capabilities. This should fix issues where other programs trying to
773 use the CDROM cannot gain access to it, but carries a risk of
774 interfering with programs writing to the disk, if they did not open
775 the device in exclusive mode as they should.
776
777 * systemd-networkd does not create a default route for IPv4 link local
778 addressing anymore. The creation of the route was unexpected and was
779 breaking routing in various cases, but people who rely on it being
780 created implicitly will need to adjust. Such a route may be requested
781 with DefaultRouteOnDevice=yes.
782
783 Similarly, systemd-networkd will not assign a link-local IPv6 address
784 when IPv6 link-local routing is not enabled.
785
786 * Receive and transmit buffers may now be configured on links with
787 the new RxBufferSize= and TxBufferSize= settings.
788
789 * systemd-networkd may now advertise additional IPv6 routes. A new
790 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
791 now supported.
792
793 * systemd-networkd may now configure "next hop" routes using the
794 [NextHop] section and Gateway= and Id= settings.
795
796 * systemd-networkd will now retain DHCP config on restarts by default
797 (but this may be overridden using the KeepConfiguration= setting).
798 The default for SendRelease= has been changed to true.
799
800 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
801 received from the server.
802
803 The client will use the received SIP server list if UseSIP=yes is
804 set.
805
806 The client may be configured to request specific options from the
807 server using a new RequestOptions= setting.
808
852b7272
ZJS
809 The client may be configured to send arbitrary options to the server
810 using a new SendOption= setting.
811
ee50dada
ZJS
812 A new IPServiceType= setting has been added to configure the "IP
813 service type" value used by the client.
814
815 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
816 request prefix hints in the DHCPv6 solicitation.
817
852b7272 818 * The DHCPv4 server may be configured to send arbitrary options using
88b86003 819 a new SendOption= setting.
852b7272
ZJS
820
821 * The DHCPv4 server may now be configured to emit SIP server list using
822 the new EmitSIP= and SIP= settings.
823
ee50dada
ZJS
824 * systemd-networkd and networkctl may now renew DHCP leases on demand.
825 networkctl has a new 'networkctl renew' verb.
826
827 * systemd-networkd may now reconfigure links on demand. networkctl
828 gained two new verbs: "reload" will reload the configuration, and
829 "reconfigure DEVICE…" will reconfigure one or more devices.
830
831 * .network files may now match on SSID and BSSID of a wireless network,
832 i.e. the access point name and hardware address using the new SSID=
833 and BSSID= options. networkctl will display the current SSID and
834 BSSID for wireless links.
835
836 .network files may also match on the wireless network type using the
f36e6a4a 837 new WLANInterfaceType= option.
ee50dada
ZJS
838
839 * systemd-networkd now includes default configuration that enables
840 link-local addressing when connected to an ad-hoc wireless network.
841
ee50dada
ZJS
842 * systemd-networkd may configure the Traffic Control queueing
843 disciplines in the kernel using the new
844 [TrafficControlQueueingDiscipline] section and Parent=,
845 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
6878c022
YW
846 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
847 NetworkEmulatorDuplicateRate= settings.
ee50dada
ZJS
848
849 * systemd-tmpfiles gained a new w+ setting to append to files.
850
851 * systemd-analyze dump will now report when the memory configuration in
852 the kernel does not match what systemd has configured (usually,
853 because some external program has modified the kernel configuration
854 on its own).
855
852b7272
ZJS
856 * systemd-analyze gained a new --base-time= switch instructs the
857 'calendar' verb to resolve times relative to that timestamp instead
858 of the present time.
859
ee50dada
ZJS
860 * journalctl --update-catalog now produces deterministic output (making
861 reproducible image builds easier).
862
863 * A new devicetree-overlay setting is now documented in the Boot Loader
864 Specification.
865
866 * The default value of the WatchdogSec= setting used in systemd
867 services (the ones bundled with the project itself) may be set at
868 configuration time using the -Dservice-watchdog= setting. If set to
869 empty, the watchdogs will be disabled.
870
0e4daba1
MK
871 * systemd-resolved validates IP addresses in certificates now when GnuTLS
872 is being used.
597f905c 873
852b7272
ZJS
874 * libcryptsetup >= 2.0.1 is now required.
875
876 * A configuration option -Duser-path= may be used to override the $PATH
877 used by the user service manager. The default is again to use the same
878 path as the system manager.
879
168e131b
LP
880 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
881 outputting the 128bit IDs in UUID format (i.e. in the "canonical
882 representation").
883
884 * Service units gained a new sandboxing option ProtectKernelLogs= which
885 makes sure the program cannot get direct access to the kernel log
886 buffer anymore, i.e. the syslog() system call (not to be confused
887 with the API of the same name in libc, which is not affected), the
888 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
889 inaccessible to the service. It's recommended to enable this setting
890 for all services that should not be able to read from or write to the
891 kernel log buffer, which are probably almost all.
892
bdf2357c 893 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
8bf2a311
LP
894 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
895 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
896 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
897 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
898 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
899 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
900 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
901 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
902 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
903 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
904 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
905 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
906 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
907 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
908 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
909 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
910 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
911 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
912 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
913 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
914 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
915 Zach Smith, Zbigniew Jędrzejewski-Szmek
916
917 – Warsaw, 2019-11-29
bdf2357c 918
efb536d0 919CHANGES WITH 243:
6af90583 920
0338934f
LP
921 * This release enables unprivileged programs (i.e. requiring neither
922 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
08b59539 923 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
0338934f
LP
924 kernel for the whole UNIX group range, i.e. all processes. This
925 change should be reasonably safe, as the kernel support for it was
926 specifically implemented to allow safe access to ICMP Echo for
927 processes lacking any privileges. If this is not desirable, it can be
928 disabled again by setting the parameter to "1 0".
929
4cd82631 930 * Previously, filters defined with SystemCallFilter= would have the
08b59539
CR
931 effect that any calling of an offending system call would terminate
932 the calling thread. This behaviour never made much sense, since
933 killing individual threads of unsuspecting processes is likely to
934 create more problems than it solves. With this release the default
935 action changed from killing the thread to killing the whole
936 process. For this to work correctly both a kernel version (>= 4.14)
937 and a libseccomp version (>= 2.4.0) supporting this new seccomp
938 action is required. If an older kernel or libseccomp is used the old
939 behaviour continues to be used. This change does not affect any
940 services that have no system call filters defined, or that use
941 SystemCallErrorNumber= (and thus see EPERM or another error instead
942 of being killed when calling an offending system call). Note that
943 systemd documentation always claimed that the whole process is
944 killed. With this change behaviour is thus adjusted to match the
945 documentation.
4cd82631 946
4860f5c2
ZJS
947 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
948 4194304 by default, i.e. the full 22bit range the kernel allows, up
949 from the old 16bit range. This should improve security and
950 robustness, as PID collisions are made less likely (though certainly
951 still possible). There are rumours this might create compatibility
6af90583
LP
952 problems, though at this moment no practical ones are known to
953 us. Downstream distributions are hence advised to undo this change in
954 their builds if they are concerned about maximum compatibility, but
955 for everybody else we recommend leaving the value bumped. Besides
956 improving security and robustness this should also simplify things as
957 the maximum number of allowed concurrent tasks was previously bounded
4860f5c2
ZJS
958 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
959 only a single knob is left ("kernel.threads-max"). There have been
960 concerns that usability is affected by this change because larger PID
961 numbers are harder to type, but we believe the change from 5 digits
962 to 7 digits doesn't hamper usability.
6af90583 963
2875a36b
LP
964 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
965 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
acdb4b52
CD
966 hierarchically set default memory protection values for a particular
967 subtree of the unit hierarchy.
968
22bf131b
CD
969 * Memory protection directives can now take a value of zero, allowing
970 explicit opting out of a default value propagated by an ancestor.
971
87cf1f8f
LP
972 * systemd now defaults to the "unified" cgroup hierarchy setup during
973 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
974 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
975 change reflects the fact that cgroupsv2 support has matured
976 substantially in both systemd and in the kernel, and is clearly the
977 way forward. Downstream production distributions might want to
978 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
979 their builds as unfortunately the popular container managers have not
980 caught up with the kernel API changes.
981
51b568f7
ZJS
982 * Man pages are not built by default anymore (html pages were already
983 disabled by default), to make development builds quicker. When
984 building systemd for a full installation with documentation, meson
4860f5c2
ZJS
985 should be called with -Dman=true and/or -Dhtml=true as appropriate.
986 The default was changed based on the assumption that quick one-off or
987 repeated development builds are much more common than full optimized
988 builds for installation, and people need to pass various other
989 options to when doing "proper" builds anyway, so the gain from making
990 development builds quicker is bigger than the one time disruption for
991 packagers.
51b568f7
ZJS
992
993 Two scripts are created in the *build* directory to generate and
994 preview man and html pages on demand, e.g.:
995
996 build/man/man systemctl
997 build/man/html systemd.index
998
e110599b 999 * libidn2 is used by default if both libidn2 and libidn are installed.
4860f5c2 1000 Please use -Dlibidn=true if libidn is preferred.
e110599b 1001
2875a36b 1002 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
9120e2bf
ZJS
1003 big-endian machines. Before, bytes were written and read in native
1004 machine order as exposed by the native libc __cpu_mask interface.
1005 Now, little-endian order is always used (CPUs 0–7 are described by
1006 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
1007 This change fixes D-Bus calls that cross endianness boundary.
1008
2875a36b
LP
1009 The presentation format used for CPUAffinity= by "systemctl show" and
1010 "systemd-analyze dump" is changed to present CPU indices instead of
1011 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
1012 shown as CPUAffinity=03000000000000000000000000000… (on
1013 little-endian) or CPUAffinity=00000000000000300000000000000… (on
1014 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
1015 input format. The maximum integer that will be printed in the new
1016 format is 8191 (four digits), while the old format always used a very
1017 long number (with the length varying by architecture), so they can be
1018 unambiguously distinguished.
9120e2bf 1019
51b568f7
ZJS
1020 * /usr/sbin/halt.local is no longer supported. Implementation in
1021 distributions was inconsistent and it seems this functionality was
1022 very rarely used.
1023
1024 To replace this functionality, users should:
1025 - either define a new unit and make it a dependency of final.target
1026 (systemctl add-wants final.target my-halt-local.service)
1027 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
1028 and ensure that it accepts "halt", "poweroff", "reboot", and
1029 "kexec" as an argument, see the description in systemd-shutdown(8).
1030
1031 * When a [Match] section in .link or .network file is empty (contains
1032 no match patterns), a warning will be emitted. Please add any "match
9120e2bf 1033 all" pattern instead, e.g. OriginalName=* or Name=* in case all
51b568f7
ZJS
1034 interfaces should really be matched.
1035
b070c7c0 1036 * A new setting NUMAPolicy= may be used to set process memory
08b59539
CR
1037 allocation policy. This setting can be specified in
1038 /etc/systemd/system.conf and hence will set the default policy for
1039 PID1. The default policy can be overridden on a per-service
1040 basis. The related setting NUMAMask= is used to specify NUMA node
1041 mask that should be associated with the selected policy.
2875a36b
LP
1042
1043 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
eebaa724 1044 generates when processes it manages are reaching their memory limits,
2875a36b
LP
1045 and will place their units in a special state, and optionally kill or
1046 stop the whole unit.
1047
1048 * The service manager will now expose bus properties for the IO
1049 resources used by units. This information is also shown in "systemctl
1050 status" now (for services that have IOAccounting=yes set). Moreover,
1051 the IO accounting data is included in the resource log message
1052 generated whenever a unit stops.
1053
201632e3 1054 * Units may now configure an explicit timeout to wait for when killed
2875a36b 1055 with SIGABRT, for example when a service watchdog is hit. Previously,
201632e3
ZJS
1056 the regular TimeoutStopSec= timeout was applied in this case too —
1057 now a separate timeout may be set using TimeoutAbortSec=.
2875a36b
LP
1058
1059 * Services may now send a special WATCHDOG=trigger message with
1060 sd_notify() to trigger an immediate "watchdog missed" event, and thus
4860f5c2 1061 trigger service termination. This is useful both for testing watchdog
2875a36b
LP
1062 handling, but also for defining error paths in services, that shall
1063 be handled the same way as watchdog events.
1064
1065 * There are two new per-unit settings IPIngressFilterPath= and
1066 IPEgressFilterPath= which allow configuration of a BPF program
1067 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
1068 to apply to the IP packet ingress/egress path of all processes of a
1069 unit. This is useful to allow running systemd services with BPF
1070 programs set up externally.
1071
1072 * systemctl gained a new "clean" verb for removing the state, cache,
1073 runtime or logs directories of a service while it is terminated. The
1074 new verb may also be used to remove the state maintained on disk for
1075 timer units that have Persistent= configured.
1076
1077 * During the last phase of shutdown systemd will now automatically
1078 increase the log level configured in the "kernel.printk" sysctl so
1079 that any relevant loggable events happening during late shutdown are
1080 made visible. Previously, loggable events happening so late during
1081 shutdown were generally lost if the "kernel.printk" sysctl was set to
1082 high thresholds, as regular logging daemons are terminated at that
1083 time and thus nothing is written to disk.
1084
1085 * If processes terminated during the last phase of shutdown do not exit
1086 quickly systemd will now show their names after a short time, to make
201632e3 1087 debugging easier. After a longer timeout they are forcibly killed,
2875a36b
LP
1088 as before.
1089
1090 * journalctl (and the other tools that display logs) will now highlight
1091 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
1092 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
1093 are now shown in blue color, to separate them visually from regular
1094 logs. References to configuration files are now turned into clickable
1095 links on terminals that support that.
1096
1097 * systemd-journald will now stop logging to /var/log/journal during
1098 shutdown when /var/ is on a separate mount, so that it can be
1099 unmounted safely during shutdown.
1100
1101 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
1102
29db4c3a
LP
1103 * systemd-resolved "Cache=" configuration option in resolved.conf has
1104 been extended to also accept the 'no-negative' value. Previously,
1105 only a boolean option was allowed (yes/no), having yes as the
4860f5c2
ZJS
1106 default. If this option is set to 'no-negative', negative answers are
1107 not cached while the old cache heuristics are used positive answers.
1108 The default remains unchanged.
37d7a7d9 1109
2875a36b
LP
1110 * The predictable naming scheme for network devices now supports
1111 generating predictable names for "netdevsim" devices.
1112
4860f5c2
ZJS
1113 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
1114 udev property.
1115
1116 Those two changes form a new net.naming-policy-scheme= entry.
1117 Distributions which want to preserve naming stability may want to set
1118 the -Ddefault-net-naming-scheme= configuration option.
1119
2875a36b
LP
1120 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
1121 interfaces natively.
1122
1123 * systemd-networkd's bridge FDB support now allows configuration of a
1124 destination address for each entry (Destination=), as well as the
1125 VXLAN VNI (VNI=), as well as an option to declare what an entry is
1126 associated with (AssociatedWith=).
1127
1128 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
08b59539
CR
1129 option for configuring the maximum number of DHCP lease requests. It
1130 also learnt a new BlackList= option for blacklisting DHCP servers (a
1131 similar setting has also been added to the IPv6 RA client), as well
1132 as a SendRelease= option for configuring whether to send a DHCP
1133 RELEASE message when terminating.
2875a36b
LP
1134
1135 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
29db4c3a
LP
1136 separately in the [DHCPv4] and [DHCPv6] sections.
1137
1138 * systemd-networkd's DHCP support will now optionally create an
1139 implicit host route to the DNS server specified in the DHCP lease, in
1140 addition to the routes listed explicitly in the lease. This should
1141 ensure that in multi-homed systems DNS traffic leaves the systems on
1142 the interface that acquired the DNS server information even if other
1143 routes such as default routes exist. This behaviour may be turned on
1144 with the new RoutesToDNS= option.
2875a36b
LP
1145
1146 * systemd-networkd's VXLAN support gained a new option
eebaa724 1147 GenericProtocolExtension= for enabling VXLAN Generic Protocol
2875a36b
LP
1148 Extension support, as well as IPDoNotFragment= for setting the IP
1149 "Don't fragment" bit on outgoing packets. A similar option has been
1150 added to the GENEVE support.
1151
1152 * In systemd-networkd's [Route] section you may now configure
1153 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
1154 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
1155 propagation. The Type= setting now supports local, broadcast,
1156 anycast, multicast, any, xresolve routes, too.
1157
1158 * systemd-networkd's [Network] section learnt a new option
1159 DefaultRouteOnDevice= for automatically configuring a default route
1160 onto the network device.
1161
1162 * systemd-networkd's bridging support gained two new options ProxyARP=
1163 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
9f0d45e4
YW
1164 MulticastRouter= for configuring multicast routing behaviour. A new
1165 option MulticastIGMPVersion= may be used to change bridge's multicast
1166 Internet Group Management Protocol (IGMP) version.
2875a36b
LP
1167
1168 * systemd-networkd's FooOverUDP support gained the ability to configure
1169 local and peer IP addresses via Local= and Peer=. A new option
1170 PeerPort= may be used to configure the peer's IP port.
1171
1172 * systemd-networkd's TUN support gained a new setting VnetHeader= for
1173 tweaking Generic Segment Offload support.
1174
70fcda85
ZJS
1175 * The address family for policy rules may be specified using the new
1176 Family= option in the [RoutingPolicyRule] section.
1177
2875a36b
LP
1178 * networkctl gained a new "delete" command for removing virtual network
1179 devices, as well as a new "--stats" switch for showing device
1180 statistics.
1181
13cb62af
YW
1182 * networkd.conf gained a new setting SpeedMeter= and
1183 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
1184 measured speed may be shown by 'networkctl status'.
1185
4860f5c2
ZJS
1186 * "networkctl status" now displays MTU and queue lengths, and more
1187 detailed information about VXLAN and bridge devices.
1188
2875a36b
LP
1189 * systemd-networkd's .network and .link files gained a new Property=
1190 setting in the [Match] section, to match against devices with
1191 specific udev properties.
1192
1193 * systemd-networkd's tunnel support gained a new option
1194 AssignToLoopback= for selecting whether to use the loopback device
1195 "lo" as underlying device.
1196
70183735 1197 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
2875a36b
LP
1198 been renamed to LinkLayerAddress=, and it now allows configuration of
1199 IP addresses, too.
1200
efb536d0
ZJS
1201 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
1202 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
1203 IPv6 configuration (static or DHCPv6) was found for a given
1204 interface. It will not touch the sysctl otherwise.
1205
1206 * The order of entries is $PATH used by the user manager instance was
1207 changed to put bin/ entries before the corresponding sbin/ entries.
1208 It is recommended to not rely on this order, and only ever have one
1209 binary with a given name in the system paths under /usr.
1210
29db4c3a
LP
1211 * A new tool systemd-network-generator has been added that may generate
1212 .network, .netdev and .link files from IP configuration specified on
4860f5c2 1213 the kernel command line in the format used by Dracut.
29db4c3a 1214
2875a36b
LP
1215 * The CriticalConnection= setting in .network files is now deprecated,
1216 and replaced by a new KeepConfiguration= setting which allows more
1217 detailed configuration of the IP configuration to keep in place.
1218
4860f5c2
ZJS
1219 * systemd-analyze gained a few new verbs:
1220
1221 - "systemd-analyze timestamp" parses and converts timestamps. This is
1222 similar to the existing "systemd-analyze calendar" command which
1223 does the same for recurring calendar events.
1224
1225 - "systemd-analyze timespan" parses and converts timespans (i.e.
1226 durations as opposed to points in time).
1227
1228 - "systemd-analyze condition" will parse and test ConditionXYZ=
1229 expressions.
1230
1231 - "systemd-analyze exit-status" will parse and convert exit status
1232 codes to their names and back.
1233
1234 - "systemd-analyze unit-files" will print a list of all unit
1235 file paths and unit aliases.
1236
1237 * SuccessExitStatus=, RestartPreventExitStatus=, and
1238 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
1239 is equivalent to "65"). Those exit status name mappings may be
cc560ac0 1240 displayed with the systemd-analyze exit-status verb describe above.
2875a36b
LP
1241
1242 * systemd-logind now exposes a per-session SetBrightness() bus call,
1243 which may be used to securely change the brightness of a kernel
1244 brightness device, if it belongs to the session's seat. By using this
1245 call unprivileged clients can make changes to "backlight" and "leds"
4860f5c2
ZJS
1246 devices securely with strict requirements on session membership.
1247 Desktop environments may use this to generically make brightness
1248 changes to such devices without shipping private SUID binaries or
1249 udev rules for that purpose.
2875a36b
LP
1250
1251 * "udevadm info" gained a --wait-for-initialization switch to wait for
1252 a device to be initialized.
1253
1254 * systemd-hibernate-resume-generator will now look for resumeflags= on
1255 the kernel command line, which is similar to rootflags= and may be
4860f5c2 1256 used to configure device timeout for the hibernation device.
2875a36b
LP
1257
1258 * sd-event learnt a new API call sd_event_source_disable_unref() for
1259 disabling and unref'ing an event source in a single function. A
1260 related call sd_event_source_disable_unrefp() has been added for use
4860f5c2 1261 with gcc's cleanup extension.
2875a36b
LP
1262
1263 * The sd-id128.h public API gained a new definition
1264 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
1265 with printf().
1266
1267 * "busctl introspect" gained a new switch --xml-interface for dumping
1268 XML introspection data unmodified.
1269
1270 * PID 1 may now show the unit name instead of the unit description
1271 string in its status output during boot. This may be configured in
1272 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
1273 kernel command line option systemd.status_unit_format=.
1274
907ddcd3 1275 * PID 1 now understands a new option KExecWatchdogSec= in
4860f5c2
ZJS
1276 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
1277 Previously watchdog functionality was only available for regular
1278 reboots. The new setting defaults to off, because we don't know in
1279 the general case if the watchdog will be reset after kexec (some
1280 drivers do reset it, but not all), and the new userspace might not be
1281 configured to handle the watchdog.
1282
1283 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
1284 RebootWatchdogSec= to more clearly communicate what it is about. The
1285 old name is still accepted for compatibility.
907ddcd3 1286
2875a36b 1287 * The systemd.debug_shell kernel command line option now optionally
4860f5c2
ZJS
1288 takes a tty name to spawn the debug shell on, which allows a
1289 different tty to be selected than the built-in default.
2875a36b 1290
a4d5848a
AZ
1291 * Service units gained a new ExecCondition= setting which will run
1292 before ExecStartPre= and either continue execution of the unit (for
1293 clean exit codes), stop execution without marking the unit failed
1294 (for exit codes 1 through 254), or stop execution and fail the unit
4860f5c2 1295 (for exit code 255 or abnormal termination).
a4d5848a 1296
29db4c3a 1297 * A new service systemd-pstore.service has been added that pulls data
08b59539 1298 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
29db4c3a
LP
1299 review.
1300
1301 * timedatectl gained new verbs for configuring per-interface NTP
1302 service configuration for systemd-timesyncd.
1303
1304 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
a18a3aac 1305 2019. (You can set non-UTF-8 locales though, if you know their name.)
29db4c3a 1306
b64c47c0
LP
1307 * If variable assignments in sysctl.d/ files are prefixed with "-" any
1308 failures to apply them are now ignored.
1309
a18a3aac
LP
1310 * systemd-random-seed.service now optionally credits entropy when
1311 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
1312 true for the service to enable this behaviour, but please consult the
1313 documentation first, since this comes with a couple of caveats.
1314
4860f5c2
ZJS
1315 * systemd-random-seed.service is now a synchronization point for full
1316 initialization of the kernel's entropy pool. Services that require
1317 /dev/urandom to be correctly initialized should be ordered after this
1318 service.
a18a3aac
LP
1319
1320 * The systemd-boot boot loader has been updated to optionally maintain
1321 a random seed file in the EFI System Partition (ESP). During the boot
1322 phase, this random seed is read and updated with a new seed
4860f5c2 1323 cryptographically derived from it. Another derived seed is passed to
a18a3aac
LP
1324 the OS. The latter seed is then credited to the kernel's entropy pool
1325 very early during userspace initialization (from PID 1). This allows
1326 systems to boot up with a fully initialized kernel entropy pool from
1327 earliest boot on, and thus entirely removes all entropy pool
1328 initialization delays from systems using systemd-boot. Special care
1329 is taken to ensure different seeds are derived on system images
4860f5c2
ZJS
1330 replicated to multiple systems. "bootctl status" will show whether
1331 a seed was received from the boot loader.
1332
1333 * bootctl gained two new verbs:
1334
1335 - "bootctl random-seed" will generate the file in ESP and an EFI
1336 variable to allow a random seed to be passed to the OS as described
1337 above.
1338
1339 - "bootctl is-installed" checks whether systemd-boot is currently
1340 installed.
a18a3aac 1341
4860f5c2
ZJS
1342 * bootctl will warn if it detects that boot entries are misconfigured
1343 (for example if the kernel image was removed without purging the
1344 bootloader entry).
a18a3aac
LP
1345
1346 * A new document has been added describing systemd's use and support
1347 for the kernel's entropy pool subsystem:
1348
1349 https://systemd.io/RANDOM_SEEDS
1350
48fd50f7
LP
1351 * When the system is hibernated the swap device to write the
1352 hibernation image to is now automatically picked from all available
1353 swap devices, preferring the swap device with the highest configured
1354 priority over all others, and picking the device with the most free
1355 space if there are multiple devices with the highest priority.
1356
1357 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
86b52a39 1358 option that permits selecting the timeout how long to wait for a
48fd50f7
LP
1359 device with an encryption key before asking for the password.
1360
2dbc45ae
KK
1361 * IOWeight= has learnt to properly set the IO weight when using the
1362 BFQ scheduler officially found in kernels 5.0+.
1363
6d8cf864
ZJS
1364 * A new mailing list has been created for reporting of security issues:
1365 systemd-security@redhat.com. For mode details, see
1366 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
1367
f21e2ecb
ZJS
1368 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
1369 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
1370 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
03985d06
ZJS
1371 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
1372 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
1373 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
1374 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
1375 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
1376 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
1377 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
1378 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
1379 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
1380 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
1381 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
1382 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
1383 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
f21e2ecb
ZJS
1384 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
1385 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
1386 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
1387 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
1388 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
1389 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
03985d06
ZJS
1390 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
1391 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
1392 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
1393 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
1394 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
1395 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
1396 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
1397 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
29db4c3a 1398
efb536d0 1399 – Camerino, 2019-09-03
6af90583 1400
d822bd4e 1401CHANGES WITH 242:
9b89e602
ZJS
1402
1403 * In .link files, MACAddressPolicy=persistent (the default) is changed
1404 to cover more devices. For devices like bridges, tun, tap, bond, and
1405 similar interfaces that do not have other identifying information,
1406 the interface name is used as the basis for persistent seed for MAC
1407 and IPv4LL addresses. The way that devices that were handled
fd9baae8
ZJS
1408 previously is not changed, and this change is about covering more
1409 devices then previously by the "persistent" policy.
9b89e602
ZJS
1410
1411 MACAddressPolicy=random may be used to force randomized MACs and
1412 IPv4LL addresses for a device if desired.
1413
1414 Hint: the log output from udev (at debug level) was enhanced to
1415 clarify what policy is followed and which attributes are used.
1416 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
1417 may be used to view this.
1418
c111cd98
YW
1419 Hint: if a bridge interface is created without any slaves, and gains
1420 a slave later, then now the bridge does not inherit slave's MAC.
1421 To inherit slave's MAC, for example, create the following file:
1422 ```
1423 # /etc/systemd/network/98-bridge-inherit-mac.link
1424 [Match]
1425 Type=bridge
1426
1427 [Link]
1428 MACAddressPolicy=none
1429 ```
1430
9b89e602
ZJS
1431 * The .device units generated by systemd-fstab-generator and other
1432 generators do not automatically pull in the corresponding .mount unit
1433 as a Wants= dependency. This means that simply plugging in the device
1434 will not cause the mount unit to be started automatically. But please
5787c509
LP
1435 note that the mount unit may be started for other reasons, in
1436 particular if it is part of local-fs.target, and any unit which
1437 (transitively) depends on local-fs.target is started.
9b89e602 1438
5787c509
LP
1439 * networkctl list/status/lldp now accept globbing wildcards for network
1440 interface names to match against all existing interfaces.
1441
1442 * The $PIDFILE environment variable is set to point the absolute path
1443 configured with PIDFile= for processes of that service.
9b89e602
ZJS
1444
1445 * The fallback DNS server list was augmented with Cloudflare public DNS
1446 servers. Use `-Ddns-servers=` to set a different fallback.
1447
1448 * A new special target usb-gadget.target will be started automatically
1449 when a USB Device Controller is detected (which means that the system
1450 is a USB peripheral).
1451
5787c509
LP
1452 * A new unit setting CPUQuotaPeriodSec= assigns the time period
1453 relatively to which the CPU time quota specified by CPUQuota= is
1454 measured.
9b89e602 1455
5787c509 1456 * A new unit setting ProtectHostname= may be used to prevent services
9b89e602
ZJS
1457 from modifying hostname information (even if they otherwise would
1458 have privileges to do so).
1459
5787c509 1460 * A new unit setting NetworkNamespacePath= may be used to specify a
9b89e602
ZJS
1461 namespace for service or socket units through a path referring to a
1462 Linux network namespace pseudo-file.
1463
5787c509
LP
1464 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
1465 have an effect on .socket units: when used the listening socket is
1466 created within the configured network namespace instead of the host
1467 namespace.
1468
1469 * ExecStart= command lines in unit files may now be prefixed with ':'
1470 in which case environment variable substitution is
1471 disabled. (Supported for the other ExecXYZ= settings, too.)
1472
2eb466fc
LP
1473 * .timer units gained two new boolean settings OnClockChange= and
1474 OnTimezoneChange= which may be used to also trigger a unit when the
1475 system clock is changed or the local timezone is
1476 modified. systemd-run has been updated to make these options easily
1477 accessible from the command line for transient timers.
1478
1479 * Two new conditions for units have been added: ConditionMemory= may be
1480 used to conditionalize a unit based on installed system
1481 RAM. ConditionCPUs= may be used to conditionalize a unit based on
39e445c9 1482 installed CPU cores.
2eb466fc
LP
1483
1484 * The @default system call filter group understood by SystemCallFilter=
1485 has been updated to include the new rseq() system call introduced in
1486 kernel 4.15.
1487
ab80eca1
ZJS
1488 * A new time-set.target has been added that indicates that the system
1489 time has been set from a local source (possibly imprecise). The
1490 existing time-sync.target is stronger and indicates that the time has
1491 been synchronized with a precise external source. Services where
1492 approximate time is sufficient should use the new target.
1493
2eb466fc
LP
1494 * "systemctl start" (and related commands) learnt a new
1495 --show-transaction option. If specified brief information about all
1496 jobs queued because of the requested operation is shown.
1497
5787c509
LP
1498 * systemd-networkd recognizes a new operation state 'enslaved', used
1499 (instead of 'degraded' or 'carrier') for interfaces which form a
1500 bridge, bond, or similar, and an new 'degraded-carrier' operational
1501 state used for the bond or bridge master interface when one of the
1502 enslaved devices is not operational.
9b89e602 1503
5787c509
LP
1504 * .network files learnt the new IgnoreCarrierLoss= option for leaving
1505 networks configured even if the carrier is lost.
1506
1507 * The RequiredForOnline= setting in .network files may now specify a
9b89e602 1508 minimum operational state required for the interface to be considered
5787c509
LP
1509 "online" by systemd-networkd-wait-online. Related to this
1510 systemd-networkd-wait-online gained a new option --operational-state=
1511 to configure the same, and its --interface= option was updated to
1512 optionally also take an operational state specific for an interface.
9b89e602 1513
2eb466fc
LP
1514 * systemd-networkd-wait-online gained a new setting --any for waiting
1515 for only one of the requested interfaces instead of all of them.
1516
9b89e602
ZJS
1517 * systemd-networkd now implements L2TP tunnels.
1518
5787c509
LP
1519 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
1520 may be used to cause autonomous and onlink prefixes received in IPv6
9b89e602
ZJS
1521 Router Advertisements to be ignored.
1522
5787c509
LP
1523 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
1524 file settings may be used to tweak bridge behaviour.
1525
1526 * The new TripleSampling= option in .network files may be used to
1527 configure CAN triple sampling.
9b89e602 1528
78bb2866
YW
1529 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
1530 used to point to private or preshared key for a WireGuard interface.
9b89e602 1531
5787c509
LP
1532 * /etc/crypttab now supports the same-cpu-crypt and
1533 submit-from-crypt-cpus options to tweak encryption work scheduling
1534 details.
9b89e602
ZJS
1535
1536 * systemd-tmpfiles will now take a BSD file lock before operating on a
1537 contents of directory. This may be used to temporarily exclude
1538 directories from aging by taking the same lock (useful for example
1539 when extracting a tarball into /tmp or /var/tmp as a privileged user,
1540 which might create files with really old timestamps, which
5787c509
LP
1541 nevertheless should not be deleted). For further details, see:
1542
1543 https://systemd.io/TEMPORARY_DIRECTORIES
9b89e602 1544
70d8401d
LP
1545 * systemd-tmpfiles' h line type gained support for the
1546 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
1547 controlling project quota inheritance.
1548
9b89e602
ZJS
1549 * sd-boot and bootctl now implement support for an Extended Boot Loader
1550 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
1551 addition to the ESP partition mounted to /efi or /boot/efi.
1552 Configuration file fragments, kernels, initrds and other EFI images
1553 to boot will be loaded from both the ESP and XBOOTLDR partitions.
1554 The XBOOTLDR partition was previously described by the Boot Loader
5787c509
LP
1555 Specification, but implementation was missing in sd-boot. Support for
1556 this concept allows using the sd-boot boot loader in more
1557 conservative scenarios where the boot loader itself is placed in the
1558 ESP but the kernels to boot (and their metadata) in a separate
1559 partition.
9b89e602 1560
5787c509
LP
1561 * A system may now be booted with systemd.volatile=overlay on the
1562 kernel command line, which causes the root file system to be set up
1563 an overlayfs mount combining the root-only root directory with a
1564 writable tmpfs. In this setup, the underlying root device is not
1565 modified, and any changes are lost at reboot.
9b89e602 1566
5787c509
LP
1567 * Similar, systemd-nspawn can now boot containers with a volatile
1568 overlayfs root with the new --volatile=overlay switch.
9b89e602
ZJS
1569
1570 * systemd-nspawn can now consume OCI runtime bundles using a new
1571 --oci-bundle= option. This implementation is fully usable, with most
1572 features in the specification implemented, but since this a lot of
1573 new code and functionality, this feature should most likely not
1574 be used in production yet.
1575
5787c509
LP
1576 * systemd-nspawn now supports various options described by the OCI
1577 runtime specification on the command-line and in .nspawn files:
9b89e602 1578 --inaccessible=/Inaccessible= may be used to mask parts of the file
5787c509 1579 system tree, --console=/--pipe may be used to configure how standard
9b89e602
ZJS
1580 input, output, and error are set up.
1581
1582 * busctl learned the `emit` verb to generate D-Bus signals.
1583
1584 * systemd-analyze cat-config may be used to gather and display
1585 configuration spread over multiple files, for example system and user
1586 presets, tmpfiles.d, sysusers.d, udev rules, etc.
1587
5787c509
LP
1588 * systemd-analyze calendar now takes an optional new parameter
1589 --iterations= which may be used to show a maximum number of iterations
1590 the specified expression will elapse next.
1591
1592 * The sd-bus C API gained support for naming method parameters in the
1593 introspection data.
1594
1595 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
1596 the reboot() system call expects.
1597
1598 * journalctl learnt a new --cursor-file= option that points to a file
9b89e602
ZJS
1599 from which a cursor should be loaded in the beginning and to which
1600 the updated cursor should be stored at the end.
1601
1602 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
1603 detected by systemd-detect-virt (and may also be used in
1604 ConditionVirtualization=).
1605
1606 * The behaviour of systemd-logind may now be modified with environment
1607 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
1608 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
1609 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
1610 skip the relevant operation completely (when set to false), or to
1611 create a flag file in /run/systemd (when set to true), instead of
1612 actually commencing the real operation when requested. The presence
1613 of /run/systemd/reboot-to-firmware-setup,
1614 /run/systemd/reboot-to-boot-loader-menu, and
1615 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
1616 boot loader implementations to replace some steps logind performs
1617 during reboot with their own operations.
1618
1619 * systemctl can be used to request a reboot into the boot loader menu
5787c509
LP
1620 or a specific boot loader entry with the new --boot-load-menu= and
1621 --boot-loader-entry= options to a reboot command. (This requires a
1622 boot loader that supports this, for example sd-boot.)
9b89e602
ZJS
1623
1624 * kernel-install will no longer unconditionally create the output
1625 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
1626 snippets, but will do only if the machine-specific parent directory
1627 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
1628 to create this parent directory during sd-boot installation.
1629
1630 This makes it easier to use kernel-install with plugins which support
1631 a different layout of the bootloader partitions (for example grub2).
1632
a3134241 1633 * During package installation (with `ninja install`), we would create
1fa3ba90
PM
1634 symlinks for getty@tty1.service, systemd-networkd.service,
1635 systemd-networkd.socket, systemd-resolved.service,
1636 remote-cryptsetup.target, remote-fs.target,
a3134241
ZJS
1637 systemd-networkd-wait-online.service, and systemd-timesyncd.service
1638 in /etc, as if `systemctl enable` was called for those units, to make
1639 the system usable immediately after installation. Now this is not
1640 done anymore, and instead calling `systemctl preset-all` is
1641 recommended after the first installation of systemd.
1642
bf65b7e0
LP
1643 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
1644 is built on seccomp. When turned on creation of SUID/SGID files is
1645 prohibited.
1646
1647 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
1648 implied if DynamicUser= is turned on for a service. This hardens
1649 these services, so that they neither can benefit from nor create
1650 SUID/SGID executables. This is a minor compatibility breakage, given
1651 that when DynamicUser= was first introduced SUID/SGID behaviour was
1652 unaffected. However, the security benefit of these two options is
1653 substantial, and the setting is still relatively new, hence we opted
1654 to make it mandatory for services with dynamic users.
1655
5b2fc74f
LP
1656 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
1657 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
1658 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
1659 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
1660 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
1661 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
1662 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
c3287a42
LP
1663 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
1664 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
1665 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
1666 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
ab80eca1
ZJS
1667 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
1668 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
1e5d2d65
ZJS
1669 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
1670 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
1671 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
1672 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
1673 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
c3287a42 1674
1e5d2d65 1675 — Warsaw, 2019-04-11
bf65b7e0 1676
d0f71749 1677CHANGES WITH 241:
b4ff3dbb
ZJS
1678
1679 * The default locale can now be configured at compile time. Otherwise,
1680 a suitable default will be selected automatically (one of C.UTF-8,
1681 en_US.UTF-8, and C).
1682
1683 * The version string shown by systemd and other tools now includes the
1684 git commit hash when built from git. An override may be specified
1685 during compilation, which is intended to be used by distributions to
1686 include the package release information.
1687
1688 * systemd-cat can now filter standard input and standard error streams
1689 for different syslog priorities using the new --stderr-priority=
1690 option.
1691
1692 * systemd-journald and systemd-journal-remote reject entries which
1693 contain too many fields (CVE-2018-16865) and set limits on the
1694 process' command line length (CVE-2018-16864).
1695
1696 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
1697 again.
1698
08e1fe42
ZJS
1699 * A new network device NamePolicy "keep" is implemented for link files,
1700 and used by default in 99-default.link (the fallback configuration
1701 provided by systemd). With this policy, if the network device name
1702 was already set by userspace, the device will not be renamed again.
1703 This matches the naming scheme that was implemented before
1704 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
1705 is also enabled by default, even if not specified. Effectively, this
1706 means that if naming-scheme >= 240 is specified, network devices will
1707 be renamed according to the configuration, even if they have been
1708 renamed already, if "keep" is not specified as the naming policy in
1709 the .link file. The 99-default.link file provided by systemd includes
1710 "keep" for backwards compatibility, but it is recommended for user
1711 installed .link files to *not* include it.
1712
1713 The "kernel" policy, which keeps kernel names declared to be
1714 "persistent", now works again as documented.
1715
ba7a6b8c
LP
1716 * kernel-install script now optionally takes the paths to one or more
1717 initrd files, and passes them to all plugins.
bd36ef0a 1718
57c03b1e
LP
1719 * The mincore() system call has been dropped from the @system-service
1720 system call filter group, as it is pretty exotic and may potentially
1721 used for side-channel attacks.
1722
774d6375
ZJS
1723 * -fPIE is dropped from compiler and linker options. Please specify
1724 -Db_pie=true option to meson to build position-independent
bd36ef0a
YW
1725 executables. Note that the meson option is supported since meson-0.49.
1726
27325875
LW
1727 * The fs.protected_regular and fs.protected_fifos sysctls, which were
1728 added in Linux 4.19 to make some data spoofing attacks harder, are
1729 now enabled by default. While this will hopefully improve the
1730 security of most installations, it is technically a backwards
1731 incompatible change; to disable these sysctls again, place the
1732 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
1733
1734 fs.protected_regular = 0
1735 fs.protected_fifos = 0
1736
1737 Note that the similar hardlink and symlink protection has been
1738 enabled since v199, and may be disabled likewise.
1739
a77f438b
LT
1740 * The files read from the EnvironmentFile= setting in unit files now
1741 parse backslashes inside quotes literally, matching the behaviour of
1742 POSIX shells.
1743
ba7a6b8c
LP
1744 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
1745 now automatically become NOPs when run in a chroot() environment.
1746
1747 * The tmpfiles.d/ "C" line type will now copy directory trees not only
1748 when the destination is so far missing, but also if it already exists
1749 as a directory and is empty. This is useful to cater for systems
1750 where directory trees are put together from multiple separate mount
1751 points but otherwise empty.
1752
1753 * A new function sd_bus_close_unref() (and the associated
1754 sd_bus_close_unrefp()) has been added to libsystemd, that combines
1755 sd_bus_close() and sd_bus_unref() in one.
1756
1757 * udevadm control learnt a new option for --ping for testing whether a
1758 systemd-udevd instance is running and reacting.
1759
ecebd1ec
YW
1760 * udevadm trigger learnt a new option for --wait-daemon for waiting
1761 systemd-udevd daemon to be initialized.
1762
d0f71749
LP
1763 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
1764 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
1765 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
1766 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
1767 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
1768 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
1769 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
1770 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
1771 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
1772 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
1773 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
1774 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
1775 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
1776 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
1777 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
1778 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
1779 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
1780
36d28ebc 1781 — Berlin, 2019-02-14
ba7a6b8c 1782
32673162 1783CHANGES WITH 240:
fcb97512 1784
e68a35a7
ZJS
1785 * NoNewPrivileges=yes has been set for all long-running services
1786 implemented by systemd. Previously, this was problematic due to
1787 SELinux (as this would also prohibit the transition from PID1's label
1788 to the service's label). This restriction has since been lifted, but
1789 an SELinux policy update is required.
1790 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
1791
aa2437e2
YW
1792 * DynamicUser=yes is dropped from systemd-networkd.service,
1793 systemd-resolved.service and systemd-timesyncd.service, which was
1794 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
1795 and since v236 for systemd-timesyncd.service. The users and groups
1796 systemd-network, systemd-resolve and systemd-timesync are created
1797 by systemd-sysusers again. Distributors or system administrators
1798 may need to create these users and groups if they not exist (or need
1799 to re-enable DynamicUser= for those units) while upgrading systemd.
787a133f
YW
1800 Also, the clock file for systemd-timesyncd may need to move from
1801 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
aa2437e2 1802
b1a082cd
ZJS
1803 * When unit files are loaded from disk, previously systemd would
1804 sometimes (depending on the unit loading order) load units from the
1805 target path of symlinks in .wants/ or .requires/ directories of other
1806 units. This meant that unit could be loaded from different paths
1807 depending on whether the unit was requested explicitly or as a
1808 dependency of another unit, not honouring the priority of directories
1809 in search path. It also meant that it was possible to successfully
1810 load and start units which are not found in the unit search path, as
1811 long as they were requested as a dependency and linked to from
1812 .wants/ or .requires/. The target paths of those symlinks are not
1813 used for loading units anymore and the unit file must be found in
1814 the search path.
1815
fcb97512 1816 * A new service type has been added: Type=exec. It's very similar to
421e3b45 1817 Type=simple but ensures the service manager will wait for both fork()
fcb97512
LP
1818 and execve() of the main service binary to complete before proceeding
1819 with follow-up units. This is primarily useful so that the manager
1820 propagates any errors in the preparation phase of service execution
1821 back to the job that requested the unit to be started. For example,
1822 consider a service that has ExecStart= set to a file system binary
421e3b45
ZJS
1823 that doesn't exist. With Type=simple starting the unit would be
1824 considered instantly successful, as only fork() has to complete
1825 successfully and the manager does not wait for execve(), and hence
1826 its failure is seen "too late". With the new Type=exec service type
1827 starting the unit will fail, as the manager will wait for the
1828 execve() and notice its failure, which is then propagated back to the
1829 start job.
fcb97512
LP
1830
1831 NOTE: with the next release 241 of systemd we intend to change the
1832 systemd-run tool to default to Type=exec for transient services
1833 started by it. This should be mostly safe, but in specific corner
1834 cases might result in problems, as the systemd-run tool will then
6b1ab752 1835 block on NSS calls (such as user name look-ups due to User=) done
fcb97512
LP
1836 between the fork() and execve(), which under specific circumstances
1837 might cause problems. It is recommended to specify "-p Type=simple"
1838 explicitly in the few cases where this applies. For regular,
1839 non-transient services (i.e. those defined with unit files on disk)
1840 we will continue to default to Type=simple.
1841
0972c1ae
LP
1842 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
1843 userspace processes is set to 1024 (soft) and 4096
1844 (hard). Previously, systemd passed this on unmodified to all
1845 processes it forked off. With this systemd release the hard limit
0abf9492 1846 systemd passes on is increased to 512K, overriding the kernel's
0972c1ae
LP
1847 defaults and substantially increasing the number of simultaneous file
1848 descriptors unprivileged userspace processes can allocate. Note that
1849 the soft limit remains at 1024 for compatibility reasons: the
1850 traditional UNIX select() call cannot deal with file descriptors >=
1851 1024 and increasing the soft limit globally might thus result in
1852 programs unexpectedly allocating a high file descriptor and thus
1853 failing abnormally when attempting to use it with select() (of
1854 course, programs shouldn't use select() anymore, and prefer
1855 poll()/epoll, but the call unfortunately remains undeservedly popular
1856 at this time). This change reflects the fact that file descriptor
1857 handling in the Linux kernel has been optimized in more recent
1858 kernels and allocating large numbers of them should be much cheaper
1859 both in memory and in performance than it used to be. Programs that
1860 want to take benefit of the increased limit have to "opt-in" into
421e3b45
ZJS
1861 high file descriptors explicitly by raising their soft limit. Of
1862 course, when they do that they must acknowledge that they cannot use
1863 select() anymore (and neither can any shared library they use — or
1864 any shared library used by any shared library they use and so on).
1865 Which default hard limit is most appropriate is of course hard to
1866 decide. However, given reports that ~300K file descriptors are used
1867 in real-life applications we believe 512K is sufficiently high as new
1868 default for now. Note that there are also reports that using very
1869 high hard limits (e.g. 1G) is problematic: some software allocates
1870 large arrays with one element for each potential file descriptor
1871 (Java, …) — a high hard limit thus triggers excessively large memory
1872 allocations in these applications. Hopefully, the new default of 512K
1873 is a good middle ground: higher than what real-life applications
1874 currently need, and low enough for avoid triggering excessively large
1875 allocations in problematic software. (And yes, somebody should fix
1876 Java.)
0972c1ae 1877
a8b627aa
LP
1878 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
1879 to the highest possible values, as separate accounting of file
1880 descriptors is no longer necessary, as memcg tracks them correctly as
1881 part of the memory accounting anyway. Thus, from the four limits on
1882 file descriptors currently enforced (fs.file-max, fs.nr_open,
1883 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
1884 and keep only the latter two. A set of build-time options
a579d42a 1885 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
a8b627aa
LP
1886 has been added to revert this change in behaviour, which might be
1887 an option for systems that turn off memcg in the kernel.
1888
4f7dc24f
LP
1889 * When no /etc/locale.conf file exists (and hence no locale settings
1890 are in place), systemd will now use the "C.UTF-8" locale by default,
1891 and set LANG= to it. This locale is supported by various
1892 distributions including Fedora, with clear indications that upstream
1893 glibc is going to make it available too. This locale enables UTF-8
1894 mode by default, which appears appropriate for 2018.
1895
230450d4
LR
1896 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
1897 default. This effectively switches the RFC3704 Reverse Path filtering
1898 from Strict mode to Loose mode. This is more appropriate for hosts
1899 that have multiple links with routes to the same networks (e.g.
1900 a client with a Wi-Fi and Ethernet both connected to the internet).
1901
6b1ab752 1902 Consult the kernel documentation for details on this sysctl:
230450d4
LR
1903 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
1904
23305a29
CD
1905 * CPUAccounting=yes no longer enables the CPU controller when using
1906 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
1907 statistics are now provided independently from the CPU controller.
1908
6b1ab752 1909 * Support for disabling a particular cgroup controller within a sub-tree
a8467688
CD
1910 has been added through the DisableControllers= directive.
1911
8f044cf9
CD
1912 * cgroup_no_v1=all on the kernel command line now also implies
1913 using the unified cgroup hierarchy, unless one explicitly passes
1914 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
1915
6b1ab752
LP
1916 * The new "MemoryMin=" unit file property may now be used to set the
1917 memory usage protection limit of processes invoked by the unit. This
4e1dfa45 1918 controls the cgroup v2 memory.min attribute. Similarly, the new
6b1ab752 1919 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
4e1dfa45 1920 cgroup v2 io.latency cgroup property for configuring per-service I/O
6b1ab752
LP
1921 latency.
1922
4e1dfa45
CD
1923 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
1924 to the cgroup v1 "devices" cgroup controller.
6b1ab752
LP
1925
1926 * systemd-escape now is able to combine --unescape with --template. It
1927 also learnt a new option --instance for extracting and unescaping the
1928 instance part of a unit name.
1929
1930 * sd-bus now provides the sd_bus_message_readv() which is similar to
1931 sd_bus_message_read() but takes a va_list object. The pair
1932 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
421e3b45 1933 has been added for configuring the default method call timeout to
6b1ab752
LP
1934 use. sd_bus_error_move() may be used to efficiently move the contents
1935 from one sd_bus_error structure to another, invalidating the
1936 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
1937 be used to control whether a bus connection object is automatically
1938 flushed when an sd-event loop is exited.
1939
1940 * When processing classic BSD syslog log messages, journald will now
1941 save the original time-stamp string supplied in the new
1942 SYSLOG_TIMESTAMP= journal field. This permits consumers to
1943 reconstruct the original BSD syslog message more correctly.
1944
1945 * StandardOutput=/StandardError= in service files gained support for
1946 new "append:…" parameters, for connecting STDOUT/STDERR of a service
1947 to a file, and appending to it.
1948
1949 * The signal to use as last step of killing of unit processes is now
1950 configurable. Previously it was hard-coded to SIGKILL, which may now
1951 be overridden with the new KillSignal= setting. Note that this is the
46b028f2 1952 signal used when regular termination (i.e. SIGTERM) does not suffice.
421e3b45
ZJS
1953 Similarly, the signal used when aborting a program in case of a
1954 watchdog timeout may now be configured too (WatchdogSignal=).
6b1ab752
LP
1955
1956 * The XDG_SESSION_DESKTOP environment variable may now be configured in
1957 the pam_systemd argument line, using the new desktop= switch. This is
1958 useful to initialize it properly from a display manager without
1959 having to touch C code.
1960
421e3b45
ZJS
1961 * Most configuration options that previously accepted percentage values
1962 now also accept permille values with the '‰' suffix (instead of '%').
6b1ab752 1963
6b1ab752
LP
1964 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
1965 DNS-over-TLS.
1966
1967 * systemd-resolved's configuration file resolved.conf gained a new
1968 option ReadEtcHosts= which may be used to turn off processing and
1969 honoring /etc/hosts entries.
1970
1971 * The "--wait" switch may now be passed to "systemctl
1972 is-system-running", in which case the tool will synchronously wait
1973 until the system finished start-up.
1974
1975 * hostnamed gained a new bus call to determine the DMI product UUID.
1976
1977 * On x86-64 systemd will now prefer using the RDRAND processor
1978 instruction over /dev/urandom whenever it requires randomness that
1979 neither has to be crypto-grade nor should be reproducible. This
1980 should substantially reduce the amount of entropy systemd requests
1981 from the kernel during initialization on such systems, though not
1982 reduce it to zero. (Why not zero? systemd still needs to allocate
1983 UUIDs and such uniquely, which require high-quality randomness.)
1984
1985 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
1986 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
1987 for forcing the "Other Information" bit in IPv6 RA messages. The
d6131be9 1988 bonding logic gained four new options AdActorSystemPriority=,
6b1ab752 1989 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
d6131be9
YW
1990 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
1991 shuffling of flows. The tunnel logic gained a new
1992 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
1993 Deployment. The policy rule logic gained four new options IPProtocol=,
1994 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
1995 support for the MulticastToUnicast= option. networkd also gained
1996 support for configuring static IPv4 ARP or IPv6 neighbor entries.
6b1ab752
LP
1997
1998 * .preset files (as read by 'systemctl preset') may now be used to
1999 instantiate services.
2000
2001 * /etc/crypttab now understands the sector-size= option to configure
2002 the sector size for an encrypted partition.
2003
2004 * Key material for encrypted disks may now be placed on a formatted
421e3b45
ZJS
2005 medium, and referenced from /etc/crypttab by the UUID of the file
2006 system, followed by "=" suffixed by the path to the key file.
6b1ab752
LP
2007
2008 * The "collect" udev component has been removed without replacement, as
421e3b45 2009 it is neither used nor maintained.
6b1ab752
LP
2010
2011 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
2012 LogsDirectory=, ConfigurationDirectory= settings are used in a
2013 service the executed processes will now receive a set of environment
421e3b45
ZJS
2014 variables containing the full paths of these directories.
2015 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
2016 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
2017 are used. Note that these options may be used multiple times per
2018 service in which case the resulting paths will be concatenated and
2019 separated by colons.
6b1ab752
LP
2020
2021 * Predictable interface naming has been extended to cover InfiniBand
2022 NICs. They will be exposed with an "ib" prefix.
2023
2024 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
2025 which case the respective line failing is ignored.
2026
2027 * .link files may now be used to configure the equivalent to the
2028 "ethtool advertise" commands.
2029
2030 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
2031 alternative to libudev.h. Previously, the latter was just an internal
2032 wrapper around the former, but now these two APIs are exposed
2033 directly.
2034
2035 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
2036 which calculates an app-specific boot ID similar to how
2037 sd_id128_get_machine_app_specific() generates an app-specific machine
2038 ID.
2039
2040 * A new tool systemd-id128 has been added that can be used to determine
2041 and generate various 128bit IDs.
2042
2043 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
2044 and LOGO=.
2045
2046 * systemd-hibernate-resume-generator will now honor the "noresume"
2047 kernel command line option, in which case it will bypass resuming
2048 from any hibernated image.
2049
2050 * The systemd-sleep.conf configuration file gained new options
2051 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
2052 AllowHybridSleep= for prohibiting specific sleep modes even if the
421e3b45 2053 kernel exports them.
6b1ab752
LP
2054
2055 * portablectl is now officially supported and has thus moved to
2056 /usr/bin/.
2057
2058 * bootctl learnt the two new commands "set-default" and "set-oneshot"
2059 for setting the default boot loader item to boot to (either
2060 persistently or only for the next boot). This is currently only
2061 compatible with sd-boot, but may be implemented on other boot loaders
2062 too, that follow the boot loader interface. The updated interface is
2063 now documented here:
2064
2065 https://systemd.io/BOOT_LOADER_INTERFACE
2066
2067 * A new kernel command line option systemd.early_core_pattern= is now
2068 understood which may be used to influence the core_pattern PID 1
2069 installs during early boot.
2070
2071 * busctl learnt two new options -j and --json= for outputting method
2072 call replies, properties and monitoring output in JSON.
2073
2074 * journalctl's JSON output now supports simple ANSI coloring as well as
2075 a new "json-seq" mode for generating RFC7464 output.
2076
2077 * Unit files now support the %g/%G specifiers that resolve to the UNIX
2078 group/GID of the service manager runs as, similar to the existing
2079 %u/%U specifiers that resolve to the UNIX user/UID.
2080
2081 * systemd-logind learnt a new global configuration option
2082 UserStopDelaySec= that may be set in logind.conf. It specifies how
2083 long the systemd --user instance shall remain started after a user
2084 logs out. This is useful to speed up repetitive re-connections of the
2085 same user, as it means the user's service manager doesn't have to be
2086 stopped/restarted on each iteration, but can be reused between
2087 subsequent options. This setting defaults to 10s. systemd-logind also
2088 exports two new properties on its Manager D-Bus objects indicating
421e3b45
ZJS
2089 whether the system's lid is currently closed, and whether the system
2090 is on AC power.
6b1ab752
LP
2091
2092 * systemd gained support for a generic boot counting logic, which
2093 generically permits automatic reverting to older boot loader entries
2094 if newer updated ones don't work. The boot loader side is implemented
2095 in sd-boot, but is kept open for other boot loaders too. For details
2096 see:
2097
2098 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
2099
2100 * The SuccessAction=/FailureAction= unit file settings now learnt two
2101 new parameters: "exit" and "exit-force", which result in immediate
2102 exiting of the service manager, and are only useful in systemd --user
2103 and container environments.
2104
2105 * Unit files gained support for a pair of options
2106 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
2107 exit status to use as service manager exit status when
2108 SuccessAction=/FailureAction= is set to exit or exit-force.
2109
2110 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
2111 options may now be used to configure the log rate limiting applied by
2112 journald per-service.
2113
2114 * systemd-analyze gained a new verb "timespan" for parsing and
2115 normalizing time span values (i.e. strings like "5min 7s 8us").
2116
2117 * systemd-analyze also gained a new verb "security" for analyzing the
2118 security and sand-boxing settings of services in order to determine an
2119 "exposure level" for them, indicating whether a service would benefit
2120 from more sand-boxing options turned on for them.
2121
2122 * "systemd-analyze syscall-filter" will now also show system calls
2123 supported by the local kernel but not included in any of the defined
2124 groups.
2125
2126 * .nspawn files now understand the Ephemeral= setting, matching the
2127 --ephemeral command line switch.
2128
2129 * sd-event gained the new APIs sd_event_source_get_floating() and
2130 sd_event_source_set_floating() for controlling whether a specific
2131 event source is "floating", i.e. destroyed along with the even loop
2132 object itself.
2133
2134 * Unit objects on D-Bus gained a new "Refs" property that lists all
421e3b45
ZJS
2135 clients that currently have a reference on the unit (to ensure it is
2136 not unloaded).
6b1ab752
LP
2137
2138 * The JoinControllers= option in system.conf is no longer supported, as
2139 it didn't work correctly, is hard to support properly, is legacy (as
4e1dfa45 2140 the concept only exists on cgroup v1) and apparently wasn't used.
6b1ab752
LP
2141
2142 * Journal messages that are generated whenever a unit enters the failed
421e3b45
ZJS
2143 state are now tagged with a unique MESSAGE_ID. Similarly, messages
2144 generated whenever a service process exits are now made recognizable,
5238e957 2145 too. A tagged message is also emitted whenever a unit enters the
421e3b45 2146 "dead" state on success.
6b1ab752
LP
2147
2148 * systemd-run gained a new switch --working-directory= for configuring
2149 the working directory of the service to start. A shortcut -d is
2150 equivalent, setting the working directory of the service to the
2151 current working directory of the invoking program. The new --shell
2152 (or just -S) option has been added for invoking the $SHELL of the
2153 caller as a service, and implies --pty --same-dir --wait --collect
421e3b45 2154 --service-type=exec. Or in other words, "systemd-run -S" is now the
6b1ab752
LP
2155 quickest way to quickly get an interactive in a fully clean and
2156 well-defined system service context.
2157
2158 * machinectl gained a new verb "import-fs" for importing an OS tree
2159 from a directory. Moreover, when a directory or tarball is imported
2160 and single top-level directory found with the OS itself below the OS
2161 tree is automatically mangled and moved one level up.
2162
421e3b45
ZJS
2163 * systemd-importd will no longer set up an implicit btrfs loop-back
2164 file system on /var/lib/machines. If one is already set up, it will
2165 continue to be used.
6b1ab752
LP
2166
2167 * A new generator "systemd-run-generator" has been added. It will
2168 synthesize a unit from one or more program command lines included in
2169 the kernel command line. This is very useful in container managers
2170 for example:
2171
2172 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
2173
2174 This will run "systemd-nspawn" on an image, invoke the specified
421e3b45
ZJS
2175 command line and immediately shut down the container again, returning
2176 the command line's exit code.
6b1ab752 2177
421e3b45 2178 * The block device locking logic is now documented:
6b1ab752
LP
2179
2180 https://systemd.io/BLOCK_DEVICE_LOCKING
2181
2182 * loginctl and machinectl now optionally output the various tables in
2183 JSON using the --output= switch. It is our intention to add similar
2184 support to systemctl and all other commands.
2185
2186 * udevadm's query and trigger verb now optionally take a .device unit
2187 name as argument.
2188
2189 * systemd-udevd's network naming logic now understands a new
421e3b45 2190 net.naming-scheme= kernel command line switch, which may be used to
6b1ab752
LP
2191 pick a specific version of the naming scheme. This helps stabilizing
2192 interface names even as systemd/udev are updated and the naming logic
2193 is improved.
2194
67081438
LP
2195 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
2196 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
2197 initialize one to all 0xFF.
2198
144d7f1d
LP
2199 * After loading the SELinux policy systemd will now recursively relabel
2200 all files and directories listed in
2201 /run/systemd/relabel-extra.d/*.relabel (which should be simple
2202 newline separated lists of paths) in addition to the ones it already
2203 implicitly relabels in /run, /dev and /sys. After the relabelling is
2204 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
2205 removed. This is useful to permit initrds (i.e. code running before
2206 the SELinux policy is in effect) to generate files in the host
2207 filesystem safely and ensure that the correct label is applied during
2208 the transition to the host OS.
2209
98a7b55a
LP
2210 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
2211 mknod() handling in user namespaces. Previously mknod() would always
2212 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
2213 but device nodes generated that way cannot be opened, and attempts to
2214 open them result in EPERM. This breaks the "graceful fallback" logic
2215 in systemd's PrivateDevices= sand-boxing option. This option is
2216 implemented defensively, so that when systemd detects it runs in a
2217 restricted environment (such as a user namespace, or an environment
2218 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
2219 where device nodes cannot be created the effect of PrivateDevices= is
2220 bypassed (following the logic that 2nd-level sand-boxing is not
2221 essential if the system systemd runs in is itself already sand-boxed
2222 as a whole). This logic breaks with 4.18 in container managers where
2223 user namespacing is used: suddenly PrivateDevices= succeeds setting
2224 up a private /dev/ file system containing devices nodes — but when
2225 these are opened they don't work.
2226
d238709c 2227 At this point it is recommended that container managers utilizing
98a7b55a
LP
2228 user namespaces that intend to run systemd in the payload explicitly
2229 block mknod() with seccomp or similar, so that the graceful fallback
2230 logic works again.
2231
2232 We are very sorry for the breakage and the requirement to change
2233 container configurations for newer kernels. It's purely caused by an
2234 incompatible kernel change. The relevant kernel developers have been
2235 notified about this userspace breakage quickly, but they chose to
2236 ignore it.
2237
455027c9
ZJS
2238 * PermissionsStartOnly= setting is deprecated (but is still supported
2239 for backwards compatibility). The same functionality is provided by
2240 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
2241 commands.
2242
b4ff3dbb
ZJS
2243 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
2244 pam_systemd anymore.
2245
08e1fe42
ZJS
2246 * The naming scheme for network devices was changed to always rename
2247 devices, even if they were already renamed by userspace. The "kernel"
2248 policy was changed to only apply as a fallback, if no other naming
2249 policy took effect.
2250
bd36ef0a
YW
2251 * The requirements to build systemd is bumped to meson-0.46 and
2252 python-3.5.
2253
6b1ab752
LP
2254 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
2255 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
2256 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
2257 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
c37e2358
LP
2258 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
2259 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
2260 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
2261 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
2262 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
1742aae2
ZJS
2263 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
2264 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
2265 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
2266 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
2267 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
2268 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
2269 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
2270 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
2271 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
2272 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
2273 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
2274 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
2275 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
2276 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
2277 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
2278 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
2279 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
2280 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
2281 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
2282 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
2283 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
2284 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
2285 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
2286 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
2287 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
2288 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
2289 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
2290 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
2291 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
2292 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
2293 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
2294 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
2295 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
2296 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
2297 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
2298 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
2299
2300 — Warsaw, 2018-12-21
6b1ab752 2301
e8498f82 2302CHANGES WITH 239:
019cb3ab
SH
2303
2304 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
d69f5282
ZJS
2305 builtin will name network interfaces differently than in previous
2306 versions for virtual network interfaces created with SR-IOV and NPAR
2307 and for devices where the PCI network controller device does not have
2308 a slot number associated.
2309
2310 SR-IOV virtual devices are now named based on the name of the parent
2311 interface, with a suffix of "v<N>", where <N> is the virtual device
2312 number. Previously those virtual devices were named as if completely
2313 independent.
2314
2315 The ninth and later NPAR virtual devices will be named following the
2316 scheme used for the first eight NPAR partitions. Previously those
2317 devices were not renamed and the kernel default (eth<n>) was used.
2318
2319 "net_id" will also generate names for PCI devices where the PCI
2320 network controller device does not have an associated slot number
2321 itself, but one of its parents does. Previously those devices were
2322 not renamed and the kernel default (eth<n>) was used.
019cb3ab 2323
6e2d744b
YW
2324 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
2325 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
fe903cf4
LP
2326 the unit. So, it is expected that the default behavior of
2327 systemd-logind is not changed. However, if distribution packagers or
2328 administrators disabled or modified IPAddressDeny= setting by a
2329 drop-in config file, then it may be necessary to update the file to
2330 re-enable AF_INET and AF_INET6 to support network user name services,
2331 e.g. NIS.
2332
2333 * When the RestrictNamespaces= unit property is specified multiple
2334 times, then the specified types are merged now. Previously, only the
2335 last assignment was used. So, if distribution packagers or
2336 administrators modified the setting by a drop-in config file, then it
2337 may be necessary to update the file.
e0eee477 2338
1fc83d09
LP
2339 * When OnFailure= is used in combination with Restart= on a service
2340 unit, then the specified units will no longer be triggered on
2341 failures that result in restarting. Previously, the specified units
2342 would be activated each time the unit failed, even when the unit was
2343 going to be restarted automatically. This behaviour contradicted the
2344 documentation. With this release the code is adjusted to match the
2345 documentation.
2346
41a4c3ec
LP
2347 * systemd-tmpfiles will now print a notice whenever it encounters
2348 tmpfiles.d/ lines referencing the /var/run/ directory. It will
2349 recommend reworking them to use the /run/ directory instead (for
5cadf58e
ZJS
2350 which /var/run/ is simply a symlinked compatibility alias). This way
2351 systemd-tmpfiles can properly detect line conflicts and merge lines
2352 referencing the same file by two paths, without having to access
2353 them.
41a4c3ec 2354
ce55bd5e
ZJS
2355 * systemctl disable/unmask/preset/preset-all cannot be used with
2356 --runtime. Previously this was allowed, but resulted in unintuitive
fe903cf4
LP
2357 behaviour that wasn't useful. systemctl disable/unmask will now undo
2358 both runtime and persistent enablement/masking, i.e. it will remove
2359 any relevant symlinks both in /run and /etc.
ce55bd5e 2360
e01d9e21
LP
2361 * Note that all long-running system services shipped with systemd will
2362 now default to a system call whitelist (rather than a blacklist, as
2363 before). In particular, systemd-udevd will now enforce one too. For
2364 most cases this should be safe, however downstream distributions
2365 which disabled sandboxing of systemd-udevd (specifically the
2366 MountFlags= setting), might want to disable this security feature
2367 too, as the default whitelisting will prohibit all mount, swap,
2368 reboot and clock changing operations from udev rules.
2369
5cadf58e
ZJS
2370 * sd-boot acquired new loader configuration settings to optionally turn
2371 off Windows and MacOS boot partition discovery as well as
2372 reboot-into-firmware menu items. It is also able to pick a better
2373 screen resolution for HiDPI systems, and now provides loader
41a4c3ec
LP
2374 configuration settings to change the resolution explicitly.
2375
c9299be2
IT
2376 * systemd-resolved now supports DNS-over-TLS. It's still
2377 turned off by default, use DNSOverTLS=opportunistic to turn it on in
c086ce8c
LP
2378 resolved.conf. We intend to make this the default as soon as couple
2379 of additional techniques for optimizing the initial latency caused by
2380 establishing a TLS/TCP connection are implemented.
2381
73c718a9
YW
2382 * systemd-resolved.service and systemd-networkd.service now set
2383 DynamicUser=yes. The users systemd-resolve and systemd-network are
abc291aa
LP
2384 not created by systemd-sysusers anymore.
2385
2386 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
5238e957 2387 that embed a network facing module into any process using getpwuid()
abc291aa
LP
2388 or related call: the dynamic allocation of the user ID for
2389 systemd-resolved.service means the service manager has to check NSS
2390 if the user name is already taken when forking off the service. Since
2391 the user in the common case won't be defined in /etc/passwd the
2392 lookup is likely to trigger nss-ldap which in turn might use NSS to
2393 ask systemd-resolved for hostname lookups. This will hence result in
2394 a deadlock: a user name lookup in order to start
38b38500 2395 systemd-resolved.service will result in a hostname lookup for which
abc291aa
LP
2396 systemd-resolved.service needs to be started already. There are
2397 multiple ways to work around this problem: pre-allocate the
2398 "systemd-resolve" user on such systems, so that nss-ldap won't be
2399 triggered; or use a different NSS package that doesn't do networking
2400 in-process but provides a local asynchronous name cache; or configure
2401 the NSS package to avoid lookups for UIDs in the range `pkg-config
2402 systemd --variable=dynamicuidmin` … `pkg-config systemd
2403 --variable=dynamicuidmax`, so that it does not consider itself
2404 authoritative for the same UID range systemd allocates dynamic users
2405 from.
73c718a9 2406
41a4c3ec
LP
2407 * The systemd-resolve tool has been renamed to resolvectl (it also
2408 remains available under the old name, for compatibility), and its
2409 interface is now verb-based, similar in style to the other <xyz>ctl
5cadf58e
ZJS
2410 tools, such as systemctl or loginctl.
2411
75da262a
LP
2412 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
2413 compatibility. It may be symlinked under the 'resolvconf' name, in
5cadf58e
ZJS
2414 which case it will take arguments and input compatible with the
2415 Debian and FreeBSD resolvconf tool.
41a4c3ec
LP
2416
2417 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
3f9a0a52 2418 where the system initially suspends, and after a timeout resumes and
41a4c3ec
LP
2419 hibernates again.
2420
2421 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
2422 set the client will only send a DUID as client identifier.
2423
2424 * The nss-systemd glibc NSS module will now enumerate dynamic users and
2425 groups in effect. Previously, it could resolve UIDs/GIDs to user
2426 names/groups and vice versa, but did not support enumeration.
2427
2428 * journald's Compress= configuration setting now optionally accepts a
2429 byte threshold value. All journal objects larger than this threshold
2430 will be compressed, smaller ones will not. Previously this threshold
2431 was not configurable and set to 512.
2432
5cadf58e
ZJS
2433 * A new system.conf setting NoNewPrivileges= is now available which may
2434 be used to turn off acquisition of new privileges system-wide
2435 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
2436 for all its children). Note that turning this option on means setuid
2437 binaries and file system capabilities lose their special powers.
2438 While turning on this option is a big step towards a more secure
2439 system, doing so is likely to break numerous pre-existing UNIX tools,
2440 in particular su and sudo.
41a4c3ec
LP
2441
2442 * A new service systemd-time-sync-wait.service has been added. If
2443 enabled it will delay the time-sync.target unit at boot until time
bc99dac5 2444 synchronization has been received from the network. This
41a4c3ec
LP
2445 functionality is useful on systems lacking a local RTC or where it is
2446 acceptable that the boot process shall be delayed by external network
2447 services.
2448
2449 * When hibernating, systemd will now inform the kernel of the image
2450 write offset, on kernels new enough to support this. This means swap
2451 files should work for hibernation now.
2452
5cadf58e
ZJS
2453 * When loading unit files, systemd will now look for drop-in unit files
2454 extensions in additional places. Previously, for a unit file name
41a4c3ec
LP
2455 "foo-bar-baz.service" it would look for dropin files in
2456 "foo-bar-baz.service.d/*.conf". Now, it will also look in
2457 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
2458 service name truncated after all inner dashes. This scheme allows
2459 writing drop-ins easily that apply to a whole set of unit files at
2460 once. It's particularly useful for mount and slice units (as their
5cadf58e
ZJS
2461 naming is prefix based), but is also useful for service and other
2462 units, for packages that install multiple unit files at once,
41a4c3ec 2463 following a strict naming regime of beginning the unit file name with
5cadf58e
ZJS
2464 the package's name. Two new specifiers are now supported in unit
2465 files to match this: %j and %J are replaced by the part of the unit
2466 name following the last dash.
2467
2468 * Unit files and other configuration files that support specifier
88099359 2469 expansion now understand another three new specifiers: %T and %V will
5cadf58e 2470 resolve to /tmp and /var/tmp respectively, or whatever temporary
88099359
ZJS
2471 directory has been set for the calling user. %E will expand to either
2472 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
41a4c3ec
LP
2473
2474 * The ExecStart= lines of unit files are no longer required to
2475 reference absolute paths. If non-absolute paths are specified the
2476 specified binary name is searched within the service manager's
5cadf58e
ZJS
2477 built-in $PATH, which may be queried with 'systemd-path
2478 search-binaries-default'. It's generally recommended to continue to
2479 use absolute paths for all binaries specified in unit files.
41a4c3ec 2480
c7f93e28
ZJS
2481 * Units gained a new load state "bad-setting", which is used when a
2482 unit file was loaded, but contained fatal errors which prevent it
ba1dc1a1
LP
2483 from being started (for example, a service unit has been defined
2484 lacking both ExecStart= and ExecStop= lines).
c7f93e28 2485
41a4c3ec
LP
2486 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
2487 support alternative debuggers, for example lldb. The old name
2488 continues to be available however, for compatibility reasons. Use the
5cadf58e
ZJS
2489 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
2490 to pick an alternative debugger instead of the default gdb.
41a4c3ec
LP
2491
2492 * systemctl and the other tools will now output escape sequences that
2493 generate proper clickable hyperlinks in various terminal emulators
2494 where useful (for example, in the "systemctl status" output you can
2495 now click on the unit file name to quickly open it in the
2496 editor/viewer of your choice). Note that not all terminal emulators
2497 support this functionality yet, but many do. Unfortunately, the
2498 "less" pager doesn't support this yet, hence this functionality is
2499 currently automatically turned off when a pager is started (which
2500 happens quite often due to auto-paging). We hope to remove this
5cadf58e
ZJS
2501 limitation as soon as "less" learns these escape sequences. This new
2502 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
2503 environment variable. For details on these escape sequences see:
41a4c3ec
LP
2504 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
2505
2506 * networkd's .network files now support a new IPv6MTUBytes= option for
2507 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
2508 option in the [Route] section to configure the MTU to use for
2509 specific routes. It also gained support for configuration of the DHCP
2510 "UserClass" option through the new UserClass= setting. It gained
2511 three new options in the new [CAN] section for configuring CAN
2512 networks. The MULTICAST and ALLMULTI interface flags may now be
2513 controlled explicitly with the new Multicast= and AllMulticast=
2514 settings.
2515
2516 * networkd will now automatically make use of the kernel's route
2517 expiration feature, if it is available.
2518
5cadf58e
ZJS
2519 * udevd's .link files now support setting the number of receive and
2520 transmit channels, using the RxChannels=, TxChannels=,
2521 OtherChannels=, CombinedChannels= settings.
2522
2523 * Support for UDPSegmentationOffload= has been removed, given its
2524 limited support in hardware, and waning software support.
41a4c3ec
LP
2525
2526 * networkd's .netdev files now support creating "netdevsim" interfaces.
2527
2528 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
2529 to query the unit belonging to a specific kernel control group.
2530
5cadf58e 2531 * systemd-analyze gained a new verb "cat-config", which may be used to
41a4c3ec
LP
2532 dump the contents of any configuration file, with all its matching
2533 drop-in files added in, and honouring the usual search and masking
2534 logic applied to systemd configuration files. For example use
2535 "systemd-analyze cat-config systemd/system.conf" to get the complete
2536 system configuration file of systemd how it would be loaded by PID 1
5cadf58e
ZJS
2537 itself. Similar to this, various tools such as systemd-tmpfiles or
2538 systemd-sysusers, gained a new option "--cat-config", which does the
41a4c3ec
LP
2539 corresponding operation for their own configuration settings. For
2540 example, "systemd-tmpfiles --cat-config" will now output the full
2541 list of tmpfiles.d/ lines in place.
2542
704ae536
YW
2543 * timedatectl gained three new verbs: "show" shows bus properties of
2544 systemd-timedated, "timesync-status" shows the current NTP
2545 synchronization state of systemd-timesyncd, and "show-timesync"
2546 shows bus properties of systemd-timesyncd.
41a4c3ec
LP
2547
2548 * systemd-timesyncd gained a bus interface on which it exposes details
2549 about its state.
2550
73c718a9
YW
2551 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
2552 understood by systemd-timedated. It takes a colon-separated list of
2553 unit names of NTP client services. The list is used by
2554 "timedatectl set-ntp".
2555
41a4c3ec
LP
2556 * systemd-nspawn gained a new --rlimit= switch for setting initial
2557 resource limits for the container payload. There's a new switch
5cadf58e 2558 --hostname= to explicitly override the container's hostname. A new
41a4c3ec
LP
2559 --no-new-privileges= switch may be used to control the
2560 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
2561 --oom-score-adjust= switch controls the OOM scoring adjustment value
2562 for the payload. The new --cpu-affinity= switch controls the CPU
2563 affinity of the container payload. The new --resolv-conf= switch
2564 allows more detailed control of /etc/resolv.conf handling of the
5cadf58e 2565 container. Similarly, the new --timezone= switch allows more detailed
41a4c3ec
LP
2566 control of /etc/localtime handling of the container.
2567
5cadf58e 2568 * systemd-detect-virt gained a new --list switch, which will print a
41a4c3ec
LP
2569 list of all currently known VM and container environments.
2570
5cadf58e 2571 * Support for "Portable Services" has been added, see
41a4c3ec 2572 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
5cadf58e
ZJS
2573 experimental, but this is expected to change soon. Reflecting this
2574 experimental state, the "portablectl" binary is not installed into
41a4c3ec
LP
2575 /usr/bin yet. The binary has to be called with the full path
2576 /usr/lib/systemd/portablectl instead.
2577
2578 * journalctl's and systemctl's -o switch now knows a new log output
2579 mode "with-unit". The output it generates is very similar to the
2580 regular "short" mode, but displays the unit name instead of the
2581 syslog tag for each log line. Also, the date is shown with timezone
2582 information. This mode is probably more useful than the classic
2583 "short" output mode for most purposes, except where pixel-perfect
2584 compatibility with classic /var/log/messages formatting is required.
2585
2586 * A new --dump-bus-properties switch has been added to the systemd
2587 binary, which may be used to dump all supported D-Bus properties.
c7f93e28
ZJS
2588 (Options which are still supported, but are deprecated, are *not*
2589 shown.)
41a4c3ec 2590
41a4c3ec
LP
2591 * sd-bus gained a set of new calls:
2592 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
2593 enable/disable the "floating" state of a bus slot object,
2594 i.e. whether the slot object pins the bus it is allocated for into
2595 memory or if the bus slot object gets disconnected when the bus goes
2596 away. sd_bus_open_with_description(),
2597 sd_bus_open_user_with_description(),
2598 sd_bus_open_system_with_description() may be used to allocate bus
2599 objects and set their description string already during allocation.
2600
2601 * sd-event gained support for watching inotify events from the event
2602 loop, in an efficient way, sharing inotify handles between multiple
2603 users. For this a new function sd_event_add_inotify() has been added.
2604
2605 * sd-event and sd-bus gained support for calling special user-supplied
2606 destructor functions for userdata pointers associated with
c7f93e28
ZJS
2607 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
2608 functions sd_bus_slot_set_destroy_callback,
2609 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
2610 sd_bus_track_get_destroy_callback,
2611 sd_event_source_set_destroy_callback,
2612 sd_event_source_get_destroy_callback have been added.
41a4c3ec
LP
2613
2614 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
2615
2616 * PID 1 will now automatically reschedule .timer units whenever the
5cadf58e 2617 local timezone changes. (They previously got rescheduled
41a4c3ec
LP
2618 automatically when the system clock changed.)
2619
2620 * New documentation has been added to document cgroups delegation,
2621 portable services and the various code quality tools we have set up:
2622
a8a27374
SK
2623 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
2624 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
2625 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
41a4c3ec 2626
d6906108
LP
2627 * The Boot Loader Specification has been added to the source tree.
2628
a8a27374 2629 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
d6906108
LP
2630
2631 While moving it into our source tree we have updated it and further
2632 changes are now accepted through the usual github PR workflow.
2633
41a4c3ec
LP
2634 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
2635 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
2636 earlier PAM modules. The data in these fields is used to initialize
2637 the session scope's resource properties. Thus external PAM modules
2638 may now configure per-session limits, for example sourced from
2639 external user databases.
2640
2641 * socket units with Accept=yes will now maintain a "refused" counter in
2642 addition to the existing "accepted" counter, counting connections
2643 refused due to the enforced limits.
2644
2645 * The "systemd-path search-binaries-default" command may now be use to
2646 query the default, built-in $PATH PID 1 will pass to the services it
2647 manages.
2648
c49a7cbd
LP
2649 * A new unit file setting PrivateMounts= has been added. It's a boolean
2650 option. If enabled the unit's processes are invoked in their own file
2651 system namespace. Note that this behaviour is also implied if any
2652 other file system namespacing options (such as PrivateTmp=,
2653 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
2654 primarily useful for services that do not use any of the other file
2655 system namespacing options. One such service is systemd-udevd.service
5238e957 2656 where this is now used by default.
c49a7cbd 2657
57ab451e
ZJS
2658 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
2659 when the system is booted in UEFI "secure mode".
2660
c7668c1c
LP
2661 * A new unit "system-update-pre.target" is added, which defines an
2662 optional synchronization point for offline system updates, as
2663 implemented by the pre-existing "system-update.target" unit. It
2664 allows ordering services before the service that executes the actual
2665 update process in a generic way.
2666
f26ad321
ZJS
2667 * Systemd now emits warnings whenever .include syntax is used.
2668
41a4c3ec 2669 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
ec53d48c 2670 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
bb6f071f
LP
2671 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
2672 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
2673 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
2674 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
2675 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
2676 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
2677 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
2678 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
2679 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
2680 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
2681 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
2682 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
2683 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
2684 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
2685 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
2686 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
2687 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
2688 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
2689 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
2690 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
61d0025d 2691 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
bb6f071f
LP
2692 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
2693 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
2694 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
2695 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
2696 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
2697 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
41a4c3ec 2698
e8498f82 2699 — Berlin, 2018-06-22
41a4c3ec 2700
c657bff1 2701CHANGES WITH 238:
e0c46a73
LP
2702
2703 * The MemoryAccounting= unit property now defaults to on. After
2704 discussions with the upstream control group maintainers we learnt
2705 that the negative impact of cgroup memory accounting on current
2706 kernels is finally relatively minimal, so that it should be safe to
444d5863
ZJS
2707 enable this by default without affecting system performance. Besides
2708 memory accounting only task accounting is turned on by default, all
2709 other forms of resource accounting (CPU, IO, IP) remain off for now,
2710 because it's not clear yet that their impact is small enough to move
2711 from opt-in to opt-out. We recommend downstreams to leave memory
07a35e84 2712 accounting on by default if kernel 4.14 or higher is primarily
444d5863
ZJS
2713 used. On very resource constrained systems or when support for old
2714 kernels is a necessity, -Dmemory-accounting-default=false can be used
2715 to revert this change.
e0c46a73 2716
313c32c3
ZJS
2717 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
2718 %udev_rules_update) and the journal catalog (%journal_catalog_update)
2719 from the upgrade scriptlets of individual packages now do nothing.
2720 Transfiletriggers have been added which will perform those updates
2721 once at the end of the transaction.
2722
2723 Similar transfiletriggers have been added to execute any sysctl.d
2724 and binfmt.d rules. Thus, it should be unnecessary to provide any
2725 scriptlets to execute this configuration from package installation
2726 scripts.
2727
2728 * systemd-sysusers gained a mode where the configuration to execute is
2729 specified on the command line, but this configuration is not executed
2730 directly, but instead it is merged with the configuration on disk,
2731 and the result is executed. This is useful for package installation
2732 scripts which want to create the user before installing any files on
2733 disk (in case some of those files are owned by that user), while
2734 still allowing local admin overrides.
2735
07a35e84 2736 This functionality is exposed to rpm scriptlets through a new
313c32c3
ZJS
2737 %sysusers_create_package macro. Old %sysusers_create and
2738 %sysusers_create_inline macros are deprecated.
2739
2740 A transfiletrigger for sysusers.d configuration is now installed,
07a35e84 2741 which means that it should be unnecessary to call systemd-sysusers from
313c32c3
ZJS
2742 package installation scripts, unless the package installs any files
2743 owned by those newly-created users, in which case
2744 %sysusers_create_package should be used.
2745
2746 * Analogous change has been done for systemd-tmpfiles: it gained a mode
2747 where the command-line configuration is merged with the configuration
2748 on disk. This is exposed as the new %tmpfiles_create_package macro,
2749 and %tmpfiles_create is deprecated. A transfiletrigger is installed
2750 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
2751 from package installation scripts.
2752
2753 * sysusers.d configuration for a user may now also specify the group
2754 number, in addition to the user number ("u username 123:456"), or
2755 without the user number ("u username -:456").
2756
2757 * Configution items for systemd-sysusers can now be specified as
2758 positional arguments when the new --inline switch is used.
2759
2760 * The login shell of users created through sysusers.d may now be
2761 specified (previously, it was always /bin/sh for root and
2762 /sbin/nologin for other users).
2763
2764 * systemd-analyze gained a new --global switch to look at global user
2765 configuration. It also gained a unit-paths verb to list the unit load
2766 paths that are compiled into systemd (which can be used with
2767 --systemd, --user, or --global).
2768
2769 * udevadm trigger gained a new --settle/-w option to wait for any
2770 triggered events to finish (but just those, and not any other events
2771 which are triggered meanwhile).
2772
2773 * The action that systemd-logind takes when the lid is closed and the
2774 machine is connected to external power can now be configured using
2775 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
2776 was determined by HandleLidSwitch=, and, for backwards compatibility,
2777 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
2778
2779 * journalctl will periodically call sd_journal_process() to make it
2780 resilient against inotify queue overruns when journal files are
2781 rotated very quickly.
2782
2783 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
2784 sd_bus_get_n_queued_write — may be used to check the number of
2785 pending bus messages.
2786
2787 * systemd gained a new
2788 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
2789 which can be used to migrate foreign processes to scope and service
2790 units. The primary user for this new API is systemd itself: the
2791 systemd --user instance uses this call of the systemd --system
2792 instance to migrate processes if it itself gets the request to
2793 migrate processes and the kernel refuses this due to access
2794 restrictions. Thanks to this "systemd-run --scope --user …" works
4e1dfa45 2795 again in pure cgroup v2 environments when invoked from the user
313c32c3
ZJS
2796 session scope.
2797
2798 * A new TemporaryFileSystem= setting can be used to mask out part of
2799 the real file system tree with tmpfs mounts. This may be combined
2800 with BindPaths= and BindReadOnlyPaths= to hide files or directories
2801 not relevant to the unit, while still allowing some paths lower in
2802 the tree to be accessed.
2803
2804 ProtectHome=tmpfs may now be used to hide user home and runtime
2805 directories from units, in a way that is mostly equivalent to
2806 "TemporaryFileSystem=/home /run/user /root".
2807
2808 * Non-service units are now started with KeyringMode=shared by default.
2809 This means that mount and swapon and other mount tools have access
2810 to keys in the main keyring.
2811
2812 * /sys/fs/bpf is now mounted automatically.
2813
2814 * QNX virtualization is now detected by systemd-detect-virt and may
2815 be used in ConditionVirtualization=.
2816
2817 * IPAccounting= may now be enabled also for slice units.
2818
2819 * A new -Dsplit-bin= build configuration switch may be used to specify
2820 whether bin and sbin directories are merged, or if they should be
2821 included separately in $PATH and various listings of executable
2822 directories. The build configuration scripts will try to autodetect
2823 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
2824 system, but distributions are encouraged to configure this
2825 explicitly.
2826
2827 * A new -Dok-color= build configuration switch may be used to change
2828 the colour of "OK" status messages.
2829
2830 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
2831 PrivateNetwork=yes was buggy in previous versions of systemd. This
2832 means that after the upgrade and daemon-reexec, any such units must
2833 be restarted.
2834
2835 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
2836 will not exclude read-only files owned by root from cleanup.
2837
c657bff1
ZJS
2838 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
2839 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
2840 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
2841 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
2842 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
2843 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
2844 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
2845 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
2846 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
2847 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
2848 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
2849 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
2850 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2851 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
2852 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
2853 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
2854
2855 — Warsaw, 2018-03-05
2856
82c8e3e6 2857CHANGES WITH 237:
2b0c59ba
MP
2858
2859 * Some keyboards come with a zoom see-saw or rocker which until now got
2860 mapped to the Linux "zoomin/out" keys in hwdb. However, these
2861 keycodes are not recognized by any major desktop. They now produce
2862 Up/Down key events so that they can be used for scrolling.
2863
49e87292
LP
2864 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
2865 slightly: previously, if an argument was specified for lines of this
2866 type (i.e. the right-most column was set) this string was appended to
2867 existing files each time systemd-tmpfiles was run. This behaviour was
2868 different from what the documentation said, and not particularly
2869 useful, as repeated systemd-tmpfiles invocations would not be
2870 idempotent and grow such files without bounds. With this release
15c5594b
ZJS
2871 behaviour has been altered to match what the documentation says:
2872 lines of this type only have an effect if the indicated files don't
2873 exist yet, and only then the argument string is written to the file.
49e87292 2874
82c8e3e6
LP
2875 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
2876 systemd-tmpfiles behaviour: previously, read-only files owned by root
2877 were always excluded from the file "aging" algorithm (i.e. the
2878 automatic clean-up of directories like /tmp based on
2879 atime/mtime/ctime). We intend to drop this restriction, and age files
2880 by default even when owned by root and read-only. This behaviour was
2881 inherited from older tools, but there have been requests to remove
2882 it, and it's not obvious why this restriction was made in the first
2883 place. Please speak up now, if you are aware of software that reqires
2884 this behaviour, otherwise we'll remove the restriction in v238.
2885
95894b91
LP
2886 * A new environment variable $SYSTEMD_OFFLINE is now understood by
2887 systemctl. It takes a boolean argument. If on, systemctl assumes it
2888 operates on an "offline" OS tree, and will not attempt to talk to the
2889 service manager. Previously, this mode was implicitly enabled if a
2890 chroot() environment was detected, and this new environment variable
2891 now provides explicit control.
2892
1a0cd2c7
ZJS
2893 * .path and .socket units may now be created transiently, too.
2894 Previously only service, mount, automount and timer units were
95894b91
LP
2895 supported as transient units. The systemd-run tool has been updated
2896 to expose this new functionality, you may hence use it now to bind
2897 arbitrary commands to path or socket activation on-the-fly from the
1a0cd2c7
ZJS
2898 command line. Moreover, almost all properties are now exposed for the
2899 unit types that already supported transient operation.
95894b91
LP
2900
2901 * The systemd-mount command gained support for a new --owner= parameter
2902 which takes a user name, which is then resolved and included in uid=
2903 and gid= mount options string of the file system to mount.
2904
2905 * A new unit condition ConditionControlGroupController= has been added
2906 that checks whether a specific cgroup controller is available.
2907
2908 * Unit files, udev's .link files, and systemd-networkd's .netdev and
2909 .network files all gained support for a new condition
2910 ConditionKernelVersion= for checking against specific kernel
2911 versions.
2912
2913 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
6cddc792 2914 support for configuring device flags in the Flags= setting. In the
95894b91
LP
2915 same files, the [Tunnel] section gained support for configuring
2916 AllowLocalRemote=. The [Route] section in .network files gained
2917 support for configuring InitialCongestionWindow=,
2918 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
2919 understands RapidCommit=.
2920
2921 * systemd-networkd's DHCPv6 support gained support for Prefix
2922 Delegation.
2923
2924 * sd-bus gained support for a new "watch-bind" feature. When this
2925 feature is enabled, an sd_bus connection may be set up to connect to
2926 an AF_UNIX socket in the file system as soon as it is created. This
2927 functionality is useful for writing early-boot services that
2928 automatically connect to the system bus as soon as it is started,
2929 without ugly time-based polling. systemd-networkd and
2930 systemd-resolved have been updated to make use of this
2931 functionality. busctl exposes this functionality in a new
2932 --watch-bind= command line switch.
2933
2934 * sd-bus will now optionally synthesize a local "Connected" signal as
2935 soon as a D-Bus connection is set up fully. This message mirrors the
2936 already existing "Disconnected" signal which is synthesized when the
2937 connection is terminated. This signal is generally useful but
2938 particularly handy in combination with the "watch-bind" feature
2939 described above. Synthesizing of this message has to be requested
2940 explicitly through the new API call sd_bus_set_connected_signal(). In
2941 addition a new call sd_bus_is_ready() has been added that checks
caf2a2d8 2942 whether a connection is fully set up (i.e. between the "Connected" and
95894b91
LP
2943 "Disconnected" signals).
2944
2945 * sd-bus gained two new calls sd_bus_request_name_async() and
2946 sd_bus_release_name_async() for asynchronously registering bus
2947 names. Similar, there is now sd_bus_add_match_async() for installing
2948 a signal match asynchronously. All of systemd's own services have
2949 been updated to make use of these calls. Doing these operations
2950 asynchronously has two benefits: it reduces the risk of deadlocks in
2951 case of cyclic dependencies between bus services, and it speeds up
2952 service initialization since synchronization points for bus
2953 round-trips are removed.
2954
2955 * sd-bus gained two new calls sd_bus_match_signal() and
2956 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
2957 and sd_bus_add_match_async() but instead of taking a D-Bus match
2958 string take match fields as normal function parameters.
2959
2960 * sd-bus gained two new calls sd_bus_set_sender() and
2961 sd_bus_message_set_sender() for setting the sender name of outgoing
2962 messages (either for all outgoing messages or for just one specific
2963 one). These calls are only useful in direct connections as on
2964 brokered connections the broker fills in the sender anyway,
2965 overwriting whatever the client filled in.
2966
2967 * sd-event gained a new pseudo-handle that may be specified on all API
2968 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
2969 used this refers to the default event loop object of the calling
2970 thread. Note however that this does not implicitly allocate one —
6cddc792
CR
2971 which has to be done prior by using sd_event_default(). Similarly
2972 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
95894b91
LP
2973 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
2974 to the default bus of the specified type of the calling thread. Here
2975 too this does not implicitly allocate bus connection objects, this
2976 has to be done prior with sd_bus_default() and friends.
2977
2978 * sd-event gained a new call pair
6cddc792
CR
2979 sd_event_source_{get|set}_io_fd_own(). This may be used to request
2980 automatic closure of the file descriptor an IO event source watches
95894b91
LP
2981 when the event source is destroyed.
2982
2983 * systemd-networkd gained support for natively configuring WireGuard
2984 connections.
2985
6cddc792
CR
2986 * In previous versions systemd synthesized user records both for the
2987 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
2988 internally. In order to simplify distribution-wide renames of the
95894b91
LP
2989 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
2990 new transitional flag file has been added: if
2991 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
2992 user and group record within the systemd codebase is disabled.
2993
2994 * systemd-notify gained a new --uid= option for selecting the source
2995 user/UID to use for notification messages sent to the service
2996 manager.
2997
31751f7e 2998 * journalctl gained a new --grep= option to list only entries in which
e6501af8
ZJS
2999 the message matches a certain pattern. By default matching is case
3000 insensitive if the pattern is lowercase, and case sensitive
3001 otherwise. Option --case-sensitive=yes|no can be used to override
3002 this an specify case sensitivity or case insensitivity.
3003
56a29112 3004 * There's now a "systemd-analyze service-watchdogs" command for printing
508058c9 3005 the current state of the service runtime watchdog, and optionally
56a29112 3006 enabling or disabling the per-service watchdogs system-wide if given a
508058c9
LP
3007 boolean argument (i.e. the concept you configure in WatchdogSec=), for
3008 debugging purposes. There's also a kernel command line option
56a29112 3009 systemd.service_watchdogs= for controlling the same.
508058c9
LP
3010
3011 * Two new "log-level" and "log-target" options for systemd-analyze were
bc99dac5 3012 added that merge the now deprecated get-log-level, set-log-level and
508058c9
LP
3013 get-log-target, set-log-target pairs. The deprecated options are still
3014 understood for backwards compatibility. The two new options print the
3015 current value when no arguments are given, and set them when a
56a29112 3016 level/target is given as an argument.
95894b91 3017
508058c9
LP
3018 * sysusers.d's "u" lines now optionally accept both a UID and a GID
3019 specification, separated by a ":" character, in order to create users
3020 where UID and GID do not match.
3021
95894b91 3022 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
508058c9
LP
3023 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
3024 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
3025 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
3026 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
3027 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
3028 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
3029 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
3030 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
3031 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
3032 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
3033 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
3034 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
3035 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
3036 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
3037 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
3038 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
3039 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
3040 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
3041 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
3042 Палаузов
3043
3044 — Brno, 2018-01-28
2b0c59ba 3045
a1b2c92d 3046CHANGES WITH 236:
195b943d 3047
89780840
ZJS
3048 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
3049 in v235 has been extended to also set the dummy.ko module option
3050 numdummies=0, preventing the kernel from automatically creating
3051 dummy0. All dummy interfaces must now be explicitly created.
195b943d 3052
3925496a
LP
3053 * Unknown '%' specifiers in configuration files are now rejected. This
3054 applies to units and tmpfiles.d configuration. Any percent characters
3055 that are followed by a letter or digit that are not supposed to be
3056 interpreted as the beginning of a specifier should be escaped by
3057 doubling ("%%"). (So "size=5%" is still accepted, as well as
3058 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
3059 valid specifiers today.)
751223fe 3060
e6b2d948 3061 * systemd-resolved now maintains a new dynamic
89780840
ZJS
3062 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
3063 recommended to make /etc/resolv.conf a symlink to it. This file
3064 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
3065 includes dynamically acquired search domains, achieving more correct
3066 DNS resolution by software that bypasses local DNS APIs such as NSS.
e6b2d948 3067
67eb5b38
LP
3068 * The "uaccess" udev tag has been dropped from /dev/kvm and
3069 /dev/dri/renderD*. These devices now have the 0666 permissions by
3070 default (but this may be changed at build-time). /dev/dri/renderD*
3071 will now be owned by the "render" group along with /dev/kfd.
3072
89780840
ZJS
3073 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
3074 systemd-journal-gatewayd.service and
3075 systemd-journal-upload.service. This means "nss-systemd" must be
3076 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
3077 services are resolved properly.
67eb5b38 3078
3925496a
LP
3079 * In /etc/fstab two new mount options are now understood:
3080 x-systemd.makefs and x-systemd.growfs. The former has the effect that
3081 the configured file system is formatted before it is mounted, the
3082 latter that the file system is resized to the full block device size
3083 after it is mounted (i.e. if the file system is smaller than the
3084 partition it resides on, it's grown). This is similar to the fsck
3085 logic in /etc/fstab, and pulls in systemd-makefs@.service and
3086 systemd-growfs@.service as necessary, similar to
3087 systemd-fsck@.service. Resizing is currently only supported on ext4
3088 and btrfs.
3089
67eb5b38
LP
3090 * In systemd-networkd, the IPv6 RA logic now optionally may announce
3091 DNS server and domain information.
3092
3093 * Support for the LUKS2 on-disk format for encrypted partitions has
3094 been added. This requires libcryptsetup2 during compilation and
3095 runtime.
3096
89780840 3097 * The systemd --user instance will now signal "readiness" when its
67eb5b38
LP
3098 basic.target unit has been reached, instead of when the run queue ran
3099 empty for the first time.
3100
8ea2dcb0
ZJS
3101 * Tmpfiles.d with user configuration are now also supported.
3102 systemd-tmpfiles gained a new --user switch, and snippets placed in
3103 ~/.config/user-tmpfiles.d/ and corresponding directories will be
3104 executed by systemd-tmpfiles --user running in the new
3105 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
3106 running in the user session.
3107
3108 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
3109 %S resolves to the top-level state directory (/var/lib for the system
3110 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
3111 top-level cache directory (/var/cache for the system instance,
3112 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
3113 logs directory (/var/log for the system instance,
67eb5b38 3114 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
8ea2dcb0 3115 existing %t specifier, that resolves to the top-level runtime
67eb5b38
LP
3116 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
3117 user instance).
3118
3119 * journalctl learnt a new parameter --output-fields= for limiting the
3120 set of journal fields to output in verbose and JSON output modes.
3121
3122 * systemd-timesyncd's configuration file gained a new option
89780840
ZJS
3123 RootDistanceMaxSec= for setting the maximum root distance of servers
3124 it'll use, as well as the new options PollIntervalMinSec= and
3125 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
67eb5b38
LP
3126
3127 * bootctl gained a new command "list" for listing all available boot
89780840 3128 menu items on systems that follow the boot loader specification.
67eb5b38
LP
3129
3130 * systemctl gained a new --dry-run switch that shows what would be done
3131 instead of doing it, and is currently supported by the shutdown and
3132 sleep verbs.
3133
e9ad86d5 3134 * ConditionSecurity= can now detect the TOMOYO security module.
67eb5b38
LP
3135
3136 * Unit file [Install] sections are now also respected in unit drop-in
89780840 3137 files. This is intended to be used by drop-ins under /usr/lib/.
67eb5b38 3138
89780840 3139 * systemd-firstboot may now also set the initial keyboard mapping.
67eb5b38 3140
89780840
ZJS
3141 * Udev "changed" events for devices which are exposed as systemd
3142 .device units are now propagated to units specified in
3143 ReloadPropagatedFrom= as reload requests.
67eb5b38 3144
89780840
ZJS
3145 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
3146 unit template name (i.e. a name in the form of 'foobar@.service',
3147 without the instance component between the '@' and - the '.'), then
3148 the escaped sysfs path of the device is automatically used as the
3149 instance.
67eb5b38
LP
3150
3151 * SystemCallFilter= in unit files has been extended so that an "errno"
3152 can be specified individually for each system call. Example:
3153 SystemCallFilter=~uname:EILSEQ.
3154
3155 * The cgroup delegation logic has been substantially updated. Delegate=
3156 now optionally takes a list of controllers (instead of a boolean, as
3157 before), which lists the controllers to delegate at least.
3158
89780840 3159 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
67eb5b38 3160
89780840
ZJS
3161 * A new LogLevelMax= setting configures the maximum log level any
3162 process of the service may log at (i.e. anything with a lesser
3163 priority than what is specified is automatically dropped). A new
3164 LogExtraFields= setting allows configuration of additional journal
3165 fields to attach to all log records generated by any of the unit's
3166 processes.
67eb5b38 3167
89780840
ZJS
3168 * New StandardInputData= and StandardInputText= settings along with the
3169 new option StandardInput=data may be used to configure textual or
3170 binary data that shall be passed to the executed service process via
3171 standard input, encoded in-line in the unit file.
67eb5b38
LP
3172
3173 * StandardInput=, StandardOutput= and StandardError= may now be used to
3174 connect stdin/stdout/stderr of executed processes directly with a
3175 file or AF_UNIX socket in the file system, using the new "file:" option.
3176
89780840
ZJS
3177 * A new unit file option CollectMode= has been added, that allows
3178 tweaking the garbage collection logic for units. It may be used to
3179 tell systemd to garbage collect units that have failed automatically
3180 (normally it only GCs units that exited successfully). systemd-run
3181 and systemd-mount expose this new functionality with a new -G option.
3182
67eb5b38
LP
3183 * "machinectl bind" may now be used to bind mount non-directories
3184 (i.e. regularfiles, devices, fifos, sockets).
3185
3186 * systemd-analyze gained a new verb "calendar" for validating and
3187 testing calendar time specifications to use for OnCalendar= in timer
3188 units. Besides validating the expression it will calculate the next
3189 time the specified expression would elapse.
3190
3191 * In addition to the pre-existing FailureAction= unit file setting
89780840
ZJS
3192 there's now SuccessAction=, for configuring a shutdown action to
3193 execute when a unit completes successfully. This is useful in
3194 particular inside containers that shall terminate after some workload
3195 has been completed. Also, both options are now supported for all unit
3196 types, not just services.
67eb5b38
LP
3197
3198 * networkds's IP rule support gained two new options
dd014eeb 3199 IncomingInterface= and OutgoingInterface= for configuring the incoming
67eb5b38
LP
3200 and outgoing interfaces of configured rules. systemd-networkd also
3201 gained support for "vxcan" network devices.
3202
3203 * networkd gained a new setting RequiredForOnline=, taking a
3204 boolean. If set, systemd-wait-online will take it into consideration
3205 when determining that the system is up, otherwise it will ignore the
3206 interface for this purpose.
3207
3208 * The sd_notify() protocol gained support for a new operation: with
3209 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
3210 store again, ahead of POLLHUP or POLLERR when they are removed
3211 anyway.
3212
f09eb768
LP
3213 * A new document doc/UIDS-GIDS.md has been added to the source tree,
3214 that documents the UID/GID range and assignment assumptions and
3925496a
LP
3215 requirements of systemd.
3216
3217 * The watchdog device PID 1 will ping may now be configured through the
3218 WatchdogDevice= configuration file setting, or by setting the
3219 systemd.watchdog_service= kernel commandline option.
3220
3221 * systemd-resolved's gained support for registering DNS-SD services on
3222 the local network using MulticastDNS. Services may either be
3223 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
3224 the same dir below /run, /usr/lib), or through its D-Bus API.
3225
a327431b
DB
3226 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
3227 extend the effective start, runtime, and stop time. The service must
3228 continue to send EXTEND_TIMEOUT_USEC within the period specified to
3229 prevent the service manager from making the service as timedout.
3230
ea2a3c9e
LP
3231 * systemd-resolved's DNSSEC support gained support for RFC 8080
3232 (Ed25519 keys and signatures).
3233
a1b2c92d
LP
3234 * The systemd-resolve command line tool gained a new set of options
3235 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
3236 --set-nta= and --revert to configure per-interface DNS configuration
3237 dynamically during runtime. It's useful for pushing DNS information
3238 into systemd-resolved from DNS hook scripts that various interface
3239 managing software supports (such as pppd).
3240
3241 * systemd-nspawn gained a new --network-namespace-path= command line
3242 option, which may be used to make a container join an existing
3243 network namespace, by specifying a path to a "netns" file.
3244
3925496a
LP
3245 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
3246 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
3247 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
ea2a3c9e 3248 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
a1b2c92d
LP
3249 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
3250 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
3251 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
3252 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
3253 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
3254 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
3255 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
3256 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
3257 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
3258 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
3259 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
3260 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
3261 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
3262 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
3263 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
3264 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
3265 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
3266 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
3267 Jędrzejewski-Szmek, Zeal Jagannatha
67eb5b38 3268
ea2a3c9e 3269 — Berlin, 2017-12-14
3754abc5 3270
582faeb4
DJL
3271CHANGES WITH 235:
3272
2bcbffd6
LP
3273 * INCOMPATIBILITY: systemd-logind.service and other long-running
3274 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
3275 communication with the outside. This generally improves security of
3276 the system, and is in almost all cases a safe and good choice, as
23d37367 3277 these services do not and should not provide any network-facing
2bcbffd6
LP
3278 functionality. However, systemd-logind uses the glibc NSS API to
3279 query the user database. This creates problems on systems where NSS
3280 is set up to directly consult network services for user database
3281 lookups. In particular, this creates incompatibilities with the
3282 "nss-nis" module, which attempts to directly contact the NIS/YP
3283 network servers it is configured for, and will now consistently
3284 fail. In such cases, it is possible to turn off IP sandboxing for
3285 systemd-logind.service (set IPAddressDeny= in its [Service] section
3286 to the empty string, via a .d/ unit file drop-in). Downstream
3287 distributions might want to update their nss-nis packaging to include
3288 such a drop-in snippet, accordingly, to hide this incompatibility
3289 from the user. Another option is to make use of glibc's nscd service
3290 to proxy such network requests through a privilege-separated, minimal
3291 local caching daemon, or to switch to more modern technologies such
3292 sssd, whose NSS hook-ups generally do not involve direct network
3293 access. In general, we think it's definitely time to question the
3294 implementation choices of nss-nis, i.e. whether it's a good idea
3295 today to embed a network-facing loadable module into all local
3296 processes that need to query the user database, including the most
3297 trivial and benign ones, such as "ls". For more details about
3298 IPAddressDeny= see below.
3299
fccf5419
LP
3300 * A new modprobe.d drop-in is now shipped by default that sets the
3301 bonding module option max_bonds=0. This overrides the kernel default,
3302 to avoid conflicts and ambiguity as to whether or not bond0 should be
3303 managed by systemd-networkd or not. This resolves multiple issues
3304 with bond0 properties not being applied, when bond0 is configured
3305 with systemd-networkd. Distributors may choose to not package this,
3306 however in that case users will be prevented from correctly managing
3307 bond0 interface using systemd-networkd.
582faeb4 3308
ef5a8cb1 3309 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
21723f53
ZJS
3310 which print the logging level and target of the system manager. They
3311 complement the existing "set-log-level" and "set-log-target" verbs
3312 used to change those values.
ef5a8cb1 3313
fccf5419
LP
3314 * journald.conf gained a new boolean setting ReadKMsg= which defaults
3315 to on. If turned off kernel log messages will not be read by
21723f53
ZJS
3316 systemd-journald or included in the logs. It also gained a new
3317 setting LineMax= for configuring the maximum line length in
3318 STDOUT/STDERR log streams. The new default for this value is 48K, up
3319 from the previous hardcoded 2048.
fccf5419 3320
21723f53
ZJS
3321 * A new unit setting RuntimeDirectoryPreserve= has been added, which
3322 allows more detailed control of what to do with a runtime directory
3323 configured with RuntimeDirectory= (i.e. a directory below /run or
3324 $XDG_RUNTIME_DIR) after a unit is stopped.
fccf5419
LP
3325
3326 * The RuntimeDirectory= setting for units gained support for creating
3327 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
3328 one top-level directory.
3329
3330 * Units gained new options StateDirectory=, CacheDirectory=,
3331 LogsDirectory= and ConfigurationDirectory= which are closely related
3332 to RuntimeDirectory= but manage per-service directories below
21723f53 3333 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
fccf5419
LP
3334 possible to write unit files which when activated automatically gain
3335 properly owned service specific directories in these locations, thus
3336 making unit files self-contained and increasing compatibility with
3337 stateless systems and factory reset where /etc or /var are
3338 unpopulated at boot. Matching these new settings there's also
3339 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
3340 ConfigurationDirectoryMode= for configuring the access mode of these
75dfbbac
LP
3341 directories. These settings are particularly useful in combination
3342 with DynamicUser=yes as they provide secure, properly-owned,
3343 writable, and stateful locations for storage, excluded from the
3344 sandbox that such services live in otherwise.
fccf5419
LP
3345
3346 * Automake support has been removed from this release. systemd is now
3347 Meson-only.
3348
3349 * systemd-journald will now aggressively cache client metadata during
3350 runtime, speeding up log write performance under pressure. This comes
3351 at a small price though: as much of the metadata is read
3352 asynchronously from /proc/ (and isn't implicitly attached to log
3353 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
3354 metadata stored alongside a log entry might be slightly
3355 out-of-date. Previously it could only be slightly newer than the log
3356 message. The time window is small however, and given that the kernel
3357 is unlikely to be improved anytime soon in this regard, this appears
3358 acceptable to us.
3359
3360 * nss-myhostname/systemd-resolved will now by default synthesize an
3361 A/AAAA resource record for the "_gateway" hostname, pointing to the
3362 current default IP gateway. Previously it did that for the "gateway"
3363 name, hampering adoption, as some distributions wanted to leave that
38b38500 3364 hostname open for local use. The old behaviour may still be
fccf5419
LP
3365 requested at build time.
3366
3367 * systemd-networkd's [Address] section in .network files gained a new
3368 Scope= setting for configuring the IP address scope. The [Network]
3369 section gained a new boolean setting ConfigureWithoutCarrier= that
3370 tells systemd-networkd to ignore link sensing when configuring the
3371 device. The [DHCP] section gained a new Anonymize= boolean option for
3372 turning on a number of options suggested in RFC 7844. A new
3373 [RoutingPolicyRule] section has been added for configuring the IP
3374 routing policy. The [Route] section has gained support for a new
3375 Type= setting which permits configuring
3376 blackhole/unreachable/prohibit routes.
3377
3378 * The [VRF] section in .netdev files gained a new Table= setting for
3379 configuring the routing table to use. The [Tunnel] section gained a
3380 new Independent= boolean field for configuring tunnels independent of
3381 an underlying network interface. The [Bridge] section gained a new
3382 GroupForwardMask= option for configuration of propagation of link
3383 local frames between bridge ports.
3384
3385 * The WakeOnLan= setting in .link files gained support for a number of
3386 new modes. A new TCP6SegmentationOffload= setting has been added for
3387 configuring TCP/IPv6 hardware segmentation offload.
3388
3389 * The IPv6 RA sender implementation may now optionally send out RDNSS
21723f53 3390 and RDNSSL records to supply DNS configuration to peers.
fccf5419
LP
3391
3392 * systemd-nspawn gained support for a new --system-call-filter= command
21723f53
ZJS
3393 line option for adding and removing entries in the default system
3394 call filter it applies. Moreover systemd-nspawn has been changed to
fccf5419
LP
3395 implement a system call whitelist instead of a blacklist.
3396
3397 * systemd-run gained support for a new --pipe command line option. If
3398 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
3399 are directly passed on to the activated transient service
21723f53
ZJS
3400 executable. This allows invoking arbitrary processes as systemd
3401 services (for example to take benefit of dependency management,
3402 accounting management, resource management or log management that is
3403 done automatically for services) — while still allowing them to be
fccf5419
LP
3404 integrated in a classic UNIX shell pipeline.
3405
3406 * When a service sends RELOAD=1 via sd_notify() and reload propagation
3407 using ReloadPropagationTo= is configured, a reload is now propagated
3408 to configured units. (Previously this was only done on explicitly
3409 requested reloads, using "systemctl reload" or an equivalent
3410 command.)
3411
3412 * For each service unit a restart counter is now kept: it is increased
3413 each time the service is restarted due to Restart=, and may be
3414 queried using "systemctl show -p NRestarts …".
3415
44898c53
LP
3416 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
3417 @signal and @timer have been added, for usage with SystemCallFilter=
fccf5419
LP
3418 in unit files and the new --system-call-filter= command line option
3419 of systemd-nspawn (see above).
3420
3421 * ExecStart= lines in unit files gained two new modifiers: when a
3422 command line is prefixed with "!" the command will be executed as
3423 configured, except for the credentials applied by
3424 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
3425 "+", but does still apply namespacing options unlike "+". There's
3426 also "!!" now, which is mostly identical, but becomes a NOP on
3427 systems that support ambient capabilities. This is useful to write
3428 unit files that work with ambient capabilities where possible but
3429 automatically fall back to traditional privilege dropping mechanisms
3430 on systems where this is not supported.
3431
3432 * ListenNetlink= settings in socket units now support RDMA netlink
3433 sockets.
3434
3435 * A new unit file setting LockPersonality= has been added which permits
3436 locking down the chosen execution domain ("personality") of a service
3437 during runtime.
3438
3439 * A new special target "getty-pre.target" has been added, which is
3440 ordered before all text logins, and may be used to order services
21723f53 3441 before textual logins acquire access to the console.
fccf5419
LP
3442
3443 * systemd will now attempt to load the virtio-rng.ko kernel module very
3444 early on if a VM environment supporting this is detected. This should
3445 improve entropy during early boot in virtualized environments.
3446
3447 * A _netdev option is now supported in /etc/crypttab that operates in a
3448 similar way as the same option in /etc/fstab: it permits configuring
21723f53
ZJS
3449 encrypted devices that need to be ordered after the network is up.
3450 Following this logic, two new special targets
fccf5419 3451 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
21723f53
ZJS
3452 added that are to cryptsetup.target what remote-fs.target and
3453 remote-fs-pre.target are to local-fs.target.
fccf5419
LP
3454
3455 * Service units gained a new UnsetEnvironment= setting which permits
21723f53
ZJS
3456 unsetting specific environment variables for services that are
3457 normally passed to it (for example in order to mask out locale
fccf5419
LP
3458 settings for specific services that can't deal with it).
3459
3460 * Units acquired a new boolean option IPAccounting=. When turned on, IP
3461 traffic accounting (packet count as well as byte count) is done for
3462 the service, and shown as part of "systemctl status" or "systemd-run
3463 --wait".
3464
3465 * Service units acquired two new options IPAddressAllow= and
3466 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
3467 for configuring a simple IP access control list for all sockets of
3468 the unit. These options are available also on .slice and .socket
3469 units, permitting flexible access list configuration for individual
3470 services as well as groups of services (as defined by a slice unit),
3471 including system-wide. Note that IP ACLs configured this way are
3472 enforced on every single IPv4 and IPv6 socket created by any process
3473 of the service unit, and apply to ingress as well as egress traffic.
3474
21723f53 3475 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
608f70e6 3476 structured log message is generated each time the unit is stopped,
fccf5419
LP
3477 containing information about the consumed resources of this
3478 invocation.
3479
3480 * A new setting KeyringMode= has been added to unit files, which may be
3481 used to control how the kernel keyring is set up for executed
3482 processes.
3483
e06fafb2
LP
3484 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
3485 "systemctl kexec" and "systemctl exit" are now always asynchronous in
3486 behaviour (that is: these commands return immediately after the
21723f53
ZJS
3487 operation was enqueued instead of waiting for the operation to
3488 complete). Previously, "systemctl poweroff" and "systemctl reboot"
e06fafb2
LP
3489 were asynchronous on systems using systemd-logind (i.e. almost
3490 always, and like they were on sysvinit), and the other three commands
3491 were unconditionally synchronous. With this release this is cleaned
3492 up, and callers will see the same asynchronous behaviour on all
3493 systems for all five operations.
3494
3495 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
3496 the system.
3497
fccf5419
LP
3498 * .timer units now accept calendar specifications in other timezones
3499 than UTC or the local timezone.
3500
f6e64b78 3501 * The tmpfiles snippet var.conf has been changed to create
21723f53
ZJS
3502 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
3503 the "utmp" group already, and it appears to be generally understood
3504 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
3505 databases. Previously this was implemented correctly for all these
3506 databases excepts btmp, which has been opened up like this now
3507 too. Note that while the other databases are world-readable
3508 (i.e. 0644), btmp is not and remains more restrictive.
f6e64b78 3509
d55b0463
LP
3510 * The systemd-resolve tool gained a new --reset-server-features
3511 switch. When invoked like this systemd-resolved will forget
3512 everything it learnt about the features supported by the configured
3513 upstream DNS servers, and restarts the feature probing logic on the
cf84484a
LP
3514 next resolver look-up for them at the highest feature level
3515 again.
3516
3517 * The status dump systemd-resolved sends to the logs upon receiving
3518 SIGUSR1 now also includes information about all DNS servers it is
3519 configured to use, and the features levels it probed for them.
d55b0463 3520
fccf5419
LP
3521 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
3522 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
76451c1d
LP
3523 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
3524 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
3525 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
3526 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
3527 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
3528 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
3529 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
3530 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
3531 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
3532 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
3533 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
3534 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
3535 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
3536 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
3537 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
3538 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
3539 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
3540 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
fccf5419 3541
c1719d8b 3542 — Berlin, 2017-10-06
fccf5419 3543
4b4da299
LP
3544CHANGES WITH 234:
3545
3546 * Meson is now supported as build system in addition to Automake. It is
3547 our plan to remove Automake in one of our next releases, so that
3548 Meson becomes our exclusive build system. Hence, please start using
3549 the Meson build system in your downstream packaging. There's plenty
3550 of documentation around how to use Meson, the extremely brief
3551 summary:
3552
3553 ./autogen.sh && ./configure && make && sudo make install
3554
3555 becomes:
3556
3557 meson build && ninja -C build && sudo ninja -C build install
3558
3559 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
3560 which permits configuring a timeout on the time a job is
3561 running. This is particularly useful for setting timeouts on jobs for
3562 .device units.
3563
3564 * Unit files gained two new options ConditionUser= and ConditionGroup=
3565 for conditionalizing units based on the identity of the user/group
3566 running a systemd user instance.
3567
3568 * systemd-networkd now understands a new FlowLabel= setting in the
3569 [VXLAN] section of .network files, as well as a Priority= in
3570 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
3571 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
3572 gained support for configuration of GENEVE links, and IPv6 address
3573 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
3574
9f09a95a 3575 * .link files now understand a new Port= setting.
4b4da299
LP
3576
3577 * systemd-networkd's DHCP support gained support for DHCP option 119
3578 (domain search list).
3579
3580 * systemd-networkd gained support for serving IPv6 address ranges using
bc99dac5 3581 the Router Advertisement protocol. The new .network configuration
4b4da299
LP
3582 section [IPv6Prefix] may be used to configure the ranges to
3583 serve. This is implemented based on a new, minimal, native server
3584 implementation of RA.
3585
3586 * journalctl's --output= switch gained support for a new parameter
3587 "short-iso-precise" for a mode where timestamps are shown as precise
3588 ISO date values.
3589
3590 * systemd-udevd's "net_id" builtin may now generate stable network
3591 interface names from IBM PowerVM VIO devices as well as ACPI platform
3592 devices.
3593
3594 * MulticastDNS support in systemd-resolved may now be explicitly
3595 enabled/disabled using the new MulticastDNS= configuration file
3596 option.
3597
3598 * systemd-resolved may now optionally use libidn2 instead of the libidn
7f7ab228
ZJS
3599 for processing internationalized domain names. Support for libidn2
3600 should be considered experimental and should not be enabled by
3601 default yet.
4b4da299
LP
3602
3603 * "machinectl pull-tar" and related call may now do verification of
3604 downloaded images using SUSE-style .sha256 checksum files in addition
3605 to the already existing support for validating using Ubuntu-style
3606 SHA256SUMS files.
3607
3608 * sd-bus gained support for a new sd_bus_message_appendv() call which
3609 is va_list equivalent of sd_bus_message_append().
3610
3611 * sd-boot gained support for validating images using SHIM/MOK.
3612
3613 * The SMACK code learnt support for "onlycap".
3614
3615 * systemd-mount --umount is now much smarter in figuring out how to
3616 properly unmount a device given its mount or device path.
5486a31d
ZJS
3617
3618 * The code to call libnss_dns as a fallback from libnss_resolve when
3619 the communication with systemd-resolved fails was removed. This
3620 fallback was redundant and interfered with the [!UNAVAIL=return]
3621 suffix. See nss-resolve(8) for the recommended configuration.
3622
9f09a95a
ZJS
3623 * systemd-logind may now be restarted without losing state. It stores
3624 the file descriptors for devices it manages in the system manager
38d93385 3625 using the FDSTORE= mechanism. Please note that further changes in
9f09a95a
ZJS
3626 other components may be required to make use of this (for example
3627 Xorg has code to listen for stops of systemd-logind and terminate
3628 itself when logind is stopped or restarted, in order to avoid using
3629 stale file descriptors for graphical devices, which is now
3630 counterproductive and must be reverted in order for restarts of
3631 systemd-logind to be safe. See
3632 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
3633
d271c5d3 3634 * All kernel-install plugins are called with the environment variable
9d8813b3 3635 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
d271c5d3
ZJS
3636 /etc/machine-id. If the machine ID could not be determined,
3637 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
3638 anything in the entry directory (passed as the second argument) if
5238e957 3639 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
d271c5d3
ZJS
3640 temporary directory is passed as the entry directory and removed
3641 after all the plugins exit.
9d8813b3 3642
184d2c15 3643 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
ac172e52
LP
3644 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
3645 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
3646 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
3647 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
184d2c15
LP
3648 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
3649 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
3650 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
3651 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
ac172e52
LP
3652 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
3653 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
3654 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
3655 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
3656 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
3657 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
3658 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
3659 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
3660 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
3661 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
3662 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
3663 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
3664 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
3665 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
3666 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
3667 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
184d2c15
LP
3668 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
3669 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
ac172e52
LP
3670 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
3671 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
3672 Георгиевски
4b4da299 3673
ac172e52 3674 — Berlin, 2017-07-12
4b4da299 3675
a2b53448 3676CHANGES WITH 233:
d08ee7cb 3677
23eb30b3
ZJS
3678 * The "hybrid" control group mode has been modified to improve
3679 compatibility with "legacy" cgroups-v1 setups. Specifically, the
3680 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
3681 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
3682 cgroups-v1 hierarchy), the only externally visible change being that
3683 the cgroups-v2 hierarchy is also mounted, to
3684 /sys/fs/cgroup/unified. This should provide a large degree of
3685 compatibility with "legacy" cgroups-v1, while taking benefit of the
3686 better management capabilities of cgroups-v2.
3687
3688 * The default control group setup mode may be selected both a boot-time
3689 via a set of kernel command line parameters (specifically:
3690 systemd.unified_cgroup_hierarchy= and
3691 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
3692 default selected on the configure command line
3693 (--with-default-hierarchy=). The upstream default is "hybrid"
3694 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
3695 this will change in a future systemd version to be "unified" (pure
3696 cgroups-v2 mode). The third option for the compile time option is
3697 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
3698 distributions to default to "hybrid" mode for release distributions,
3699 starting with v233. We recommend "unified" for development
3700 distributions (specifically: distributions such as Fedora's rawhide)
3701 as that's where things are headed in the long run. Use "legacy" for
3702 greatest stability and compatibility only.
3703
3704 * Note one current limitation of "unified" and "hybrid" control group
3705 setup modes: the kernel currently does not permit the systemd --user
3706 instance (i.e. unprivileged code) to migrate processes between two
3707 disconnected cgroup subtrees, even if both are managed and owned by
3708 the user. This effectively means "systemd-run --user --scope" doesn't
3709 work when invoked from outside of any "systemd --user" service or
3710 scope. Specifically, it is not supported from session scopes. We are
3711 working on fixing this in a future systemd version. (See #3388 for
3712 further details about this.)
3713
fb7c4eff
MG
3714 * DBus policy files are now installed into /usr rather than /etc. Make
3715 sure your system has dbus >= 1.9.18 running before upgrading to this
3716 version, or override the install path with --with-dbuspolicydir= .
3717
23eb30b3
ZJS
3718 * All python scripts shipped with systemd (specifically: the various
3719 tests written in Python) now require Python 3.
3720
d60c5270 3721 * systemd unit tests can now run standalone (without the source or
4dfe64f8
ZJS
3722 build directories), and can be installed into /usr/lib/systemd/tests/
3723 with 'make install-tests'.
3724
23eb30b3
ZJS
3725 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
3726 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
3727 kernel.
3728
3729 * Support for the %c, %r, %R specifiers in unit files has been
3730 removed. Specifiers are not supposed to be dependent on configuration
3731 in the unit file itself (so that they resolve the same regardless
3732 where used in the unit files), but these specifiers were influenced
3733 by the Slice= option.
3734
5cfc0a84
LP
3735 * The shell invoked by debug-shell.service now defaults to /bin/sh in
3736 all cases. If distributions want to use a different shell for this
3737 purpose (for example Fedora's /sbin/sushell) they need to specify
3738 this explicitly at configure time using --with-debug-shell=.
3739
2bcc3309
FB
3740 * The confirmation spawn prompt has been reworked to offer the
3741 following choices:
3742
b0eb2944 3743 (c)ontinue, proceed without asking anymore
dd6f9ac0 3744 (D)ump, show the state of the unit
2bcc3309 3745 (f)ail, don't execute the command and pretend it failed
d172b175 3746 (h)elp
eedf223a 3747 (i)nfo, show a short summary of the unit
56fde33a 3748 (j)obs, show jobs that are in progress
2bcc3309
FB
3749 (s)kip, don't execute the command and pretend it succeeded
3750 (y)es, execute the command
3751
3752 The 'n' choice for the confirmation spawn prompt has been removed,
3753 because its meaning was confusing.
3754
d08ee7cb
LP
3755 The prompt may now also be redirected to an alternative console by
3756 specifying the console as parameter to systemd.confirm_spawn=.
3757
8e458bfe
JW
3758 * Services of Type=notify require a READY=1 notification to be sent
3759 during startup. If no such message is sent, the service now fails,
3760 even if the main process exited with a successful exit code.
3761
85266f9b
LP
3762 * Services that fail to start up correctly now always have their
3763 ExecStopPost= commands executed. Previously, they'd enter "failed"
3764 state directly, without executing these commands.
3765
baf32786
MP
3766 * The option MulticastDNS= of network configuration files has acquired
3767 an actual implementation. With MulticastDNS=yes a host can resolve
23eb30b3 3768 names of remote hosts and reply to mDNS A and AAAA requests.
fa8b4499 3769
631b676b
LP
3770 * When units are about to be started an additional check is now done to
3771 ensure that all dependencies of type BindsTo= (when used in
3772 combination with After=) have been started.
3773
d08ee7cb
LP
3774 * systemd-analyze gained a new verb "syscall-filter" which shows which
3775 system call groups are defined for the SystemCallFilter= unit file
23eb30b3 3776 setting, and which system calls they contain.
d08ee7cb
LP
3777
3778 * A new system call filter group "@filesystem" has been added,
23eb30b3 3779 consisting of various file system related system calls. Group
d08ee7cb 3780 "@reboot" has been added, covering reboot, kexec and shutdown related
23eb30b3 3781 calls. Finally, group "@swap" has been added covering swap
d08ee7cb
LP
3782 configuration related calls.
3783
3784 * A new unit file option RestrictNamespaces= has been added that may be
3785 used to restrict access to the various process namespace types the
3786 Linux kernel provides. Specifically, it may be used to take away the
23eb30b3
ZJS
3787 right for a service unit to create additional file system, network,
3788 user, and other namespaces. This sandboxing option is particularly
3789 relevant due to the high amount of recently discovered namespacing
3790 related vulnerabilities in the kernel.
d08ee7cb 3791
23eb30b3
ZJS
3792 * systemd-udev's .link files gained support for a new AutoNegotiation=
3793 setting for configuring Ethernet auto-negotiation.
d08ee7cb
LP
3794
3795 * systemd-networkd's .network files gained support for a new
3796 ListenPort= setting in the [DHCP] section to explicitly configure the
3797 UDP client port the DHCP client shall listen on.
3798
23eb30b3
ZJS
3799 * .network files gained a new Unmanaged= boolean setting for explicitly
3800 excluding one or more interfaces from management by systemd-networkd.
3801
3802 * The systemd-networkd ProxyARP= option has been renamed to
3803 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
3804 renamed to ReduceARPProxy=. The old names continue to be available
3805 for compatibility.
3806
3807 * systemd-networkd gained support for configuring IPv6 Proxy NDP
3808 addresses via the new IPv6ProxyNDPAddress= .network file setting.
3809
3810 * systemd-networkd's bonding device support gained support for two new
3811 configuration options ActiveSlave= and PrimarySlave=.
3812
3813 * The various options in the [Match] section of .network files gained
3814 support for negative matching.
3815
d08ee7cb
LP
3816 * New systemd-specific mount options are now understood in /etc/fstab:
3817
3818 x-systemd.mount-timeout= may be used to configure the maximum
3819 permitted runtime of the mount command.
3820
3821 x-systemd.device-bound may be set to bind a mount point to its
3822 backing device unit, in order to automatically remove a mount point
3823 if its backing device is unplugged. This option may also be
3824 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
3825 on the block device, which is now automatically set for all CDROM
3826 drives, so that mounted CDs are automatically unmounted when they are
3827 removed from the drive.
3828
23eb30b3
ZJS
3829 x-systemd.after= and x-systemd.before= may be used to explicitly
3830 order a mount after or before another unit or mount point.
d08ee7cb
LP
3831
3832 * Enqueued start jobs for device units are now automatically garbage
3833 collected if there are no jobs waiting for them anymore.
3834
23eb30b3
ZJS
3835 * systemctl list-jobs gained two new switches: with --after, for every
3836 queued job the jobs it's waiting for are shown; with --before the
3837 jobs which it's blocking are shown.
d08ee7cb
LP
3838
3839 * systemd-nspawn gained support for ephemeral boots from disk images
3840 (or in other words: --ephemeral and --image= may now be
3841 combined). Moreover, ephemeral boots are now supported for normal
3842 directories, even if the backing file system is not btrfs. Of course,
baf32786
MP
3843 if the file system does not support file system snapshots or
3844 reflinks, the initial copy operation will be relatively expensive, but
3845 this should still be suitable for many use cases.
d08ee7cb
LP
3846
3847 * Calendar time specifications in .timer units now support
3848 specifications relative to the end of a month by using "~" instead of
3849 "-" as separator between month and day. For example, "*-02~03" means
23eb30b3 3850 "the third last day in February". In addition a new syntax for
d08ee7cb
LP
3851 repeated events has been added using the "/" character. For example,
3852 "9..17/2:00" means "every two hours from 9am to 5pm".
3853
3854 * systemd-socket-proxyd gained a new parameter --connections-max= for
3855 configuring the maximum number of concurrent connections.
3856
23eb30b3
ZJS
3857 * sd-id128 gained a new API for generating unique IDs for the host in a
3858 way that does not leak the machine ID. Specifically,
d08ee7cb 3859 sd_id128_get_machine_app_specific() derives an ID based on the
baf32786 3860 machine ID a in well-defined, non-reversible, stable way. This is
d08ee7cb
LP
3861 useful whenever an identifier for the host is needed but where the
3862 identifier shall not be useful to identify the system beyond the
3863 scope of the application itself. (Internally this uses HMAC-SHA256 as
3864 keyed hash function using the machine ID as input.)
3865
3866 * NotifyAccess= gained a new supported value "exec". When set
3867 notifications are accepted from all processes systemd itself invoked,
3868 including all control processes.
3869
3870 * .nspawn files gained support for defining overlay mounts using the
3871 Overlay= and OverlayReadOnly= options. Previously this functionality
3872 was only available on the systemd-nspawn command line.
3873
3874 * systemd-nspawn's --bind= and --overlay= options gained support for
3875 bind/overlay mounts whose source lies within the container tree by
3876 prefixing the source path with "+".
3877
3878 * systemd-nspawn's --bind= and --overlay= options gained support for
3879 automatically allocating a temporary source directory in /var/tmp
3880 that is removed when the container dies. Specifically, if the source
3881 directory is specified as empty string this mechanism is selected. An
3882 example usage is --overlay=+/var::/var, which creates an overlay
86b52a39 3883 mount based on the original /var contained in the image, overlaid
d08ee7cb
LP
3884 with a temporary directory in the host's /var/tmp. This way changes
3885 to /var are automatically flushed when the container shuts down.
3886
baf32786
MP
3887 * systemd-nspawn --image= option does now permit raw file system block
3888 devices (in addition to images containing partition tables, as
3889 before).
d08ee7cb
LP
3890
3891 * The disk image dissection logic in systemd-nspawn gained support for
3892 automatically setting up LUKS encrypted as well as Verity protected
3893 partitions. When a container is booted from an encrypted image the
3894 passphrase is queried at start-up time. When a container with Verity
3895 data is started, the root hash is search in a ".roothash" file
3896 accompanying the disk image (alternatively, pass the root hash via
3897 the new --root-hash= command line option).
3898
3899 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
3900 be used to dissect disk images the same way as systemd-nspawn does
3901 it, following the Bootable Partition Specification. It may even be
3902 used to mount disk images with complex partition setups (including
3903 LUKS and Verity partitions) to a local host directory, in order to
3904 inspect them. This tool is not considered public API (yet), and is
3905 thus not installed into /usr/bin. Please do not rely on its
3b31c466 3906 existence, since it might go away or be changed in later systemd
d08ee7cb
LP
3907 versions.
3908
3909 * A new generator "systemd-verity-generator" has been added, similar in
baf32786 3910 style to "systemd-cryptsetup-generator", permitting automatic setup of
d08ee7cb
LP
3911 Verity root partitions when systemd boots up. In order to make use of
3912 this your partition setup should follow the Discoverable Partitions
3913 Specification, and the GPT partition ID of the root file system
3914 partition should be identical to the upper 128bit of the Verity root
3915 hash. The GPT partition ID of the Verity partition protecting it
3916 should be the lower 128bit of the Verity root hash. If the partition
3917 image follows this model it is sufficient to specify a single
3918 "roothash=" kernel command line argument to both configure which root
3919 image and verity partition to use as well as the root hash for
3920 it. Note that systemd-nspawn's Verity support follows the same
3921 semantics, meaning that disk images with proper Verity data in place
3922 may be booted in containers with systemd-nspawn as well as on
3923 physical systems via the verity generator. Also note that the "mkosi"
3924 tool available at https://github.com/systemd/mkosi has been updated
3925 to generate Verity protected disk images following this scheme. In
3926 fact, it has been updated to generate disk images that optionally
3927 implement a complete UEFI SecureBoot trust chain, involving a signed
3928 kernel and initrd image that incorporates such a root hash as well as
3929 a Verity-enabled root partition.
3930
d08ee7cb
LP
3931 * The hardware database (hwdb) udev supports has been updated to carry
3932 accelerometer quirks.
3933
3934 * All system services are now run with a fresh kernel keyring set up
3935 for them. The invocation ID is stored by default in it, thus
3936 providing a safe, non-overridable way to determine the invocation
3937 ID of each service.
3938
3939 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
3940 options for bind mounting arbitrary paths in a service-specific
3941 way. When these options are used, arbitrary host or service files and
3942 directories may be mounted to arbitrary locations in the service's
3943 view.
3944
3945 * Documentation has been added that lists all of systemd's low-level
3946 environment variables:
3947
a8a27374 3948 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
d08ee7cb
LP
3949
3950 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
3951 whether a specific socket file descriptor matches a specified socket
3952 address.
3953
3954 * systemd-firstboot has been updated to check for the
3955 systemd.firstboot= kernel command line option. It accepts a boolean
3956 and when set to false the first boot questions are skipped.
3957
d08ee7cb 3958 * systemd-fstab-generator has been updated to check for the
23eb30b3
ZJS
3959 systemd.volatile= kernel command line option, which either takes an
3960 optional boolean parameter or the special value "state". If used the
3961 system may be booted in a "volatile" boot mode. Specifically,
3962 "systemd.volatile" is used, the root directory will be mounted as
d08ee7cb 3963 tmpfs, and only /usr is mounted from the actual root file system. If
23eb30b3
ZJS
3964 "systemd.volatile=state" is used, the root directory will be mounted
3965 as usual, but /var is mounted as tmpfs. This concept provides similar
d08ee7cb
LP
3966 functionality as systemd-nspawn's --volatile= option, but provides it
3967 on physical boots. Use this option for implementing stateless
3968 systems, or testing systems with all state and/or configuration reset
3969 to the defaults. (Note though that many distributions are not
23eb30b3 3970 prepared to boot up without a populated /etc or /var, though.)
d08ee7cb
LP
3971
3972 * systemd-gpt-auto-generator gained support for LUKS encrypted root
3973 partitions. Previously it only supported LUKS encrypted partitions
3974 for all other uses, except for the root partition itself.
3975
3976 * Socket units gained support for listening on AF_VSOCK sockets for
3977 communication in virtualized QEMU environments.
3978
3979 * The "configure" script gained a new option --with-fallback-hostname=
3980 for specifying the fallback hostname to use if none is configured in
3981 /etc/hostname. For example, by specifying
3982 --with-fallback-hostname=fedora it is possible to default to a
23eb30b3 3983 hostname of "fedora" on pristine installations.
d08ee7cb
LP
3984
3985 * systemd-cgls gained support for a new --unit= switch for listing only
3986 the control groups of a specific unit. Similar --user-unit= has been
3987 added for listing only the control groups of a specific user unit.
3988
3989 * systemd-mount gained a new --umount switch for unmounting a mount or
3990 automount point (and all mount/automount points below it).
3991
3992 * systemd will now refuse full configuration reloads (via systemctl
3993 daemon-reload and related calls) unless at least 16MiB of free space
3994 are available in /run. This is a safety precaution in order to ensure
3995 that generators can safely operate after the reload completed.
3996
3997 * A new unit file option RootImage= has been added, which has a similar
3998 effect as RootDirectory= but mounts the service's root directory from
3999 a disk image instead of plain directory. This logic reuses the same
4000 image dissection and mount logic that systemd-nspawn already uses,
4001 and hence supports any disk images systemd-nspawn supports, including
4002 those following the Discoverable Partition Specification, as well as
4003 Verity enabled images. This option enables systemd to run system
4004 services directly off disk images acting as resource bundles,
4005 possibly even including full integrity data.
4006
4007 * A new MountAPIVFS= unit file option has been added, taking a boolean
baf32786 4008 argument. If enabled /proc, /sys and /dev (collectively called the
d08ee7cb
LP
4009 "API VFS") will be mounted for the service. This is only relevant if
4010 RootDirectory= or RootImage= is used for the service, as these mounts
4011 are of course in place in the host mount namespace anyway.
4012
4013 * systemd-nspawn gained support for a new --pivot-root= switch. If
4014 specified the root directory within the container image is pivoted to
4015 the specified mount point, while the original root disk is moved to a
4016 different place. This option enables booting of ostree images
4017 directly with systemd-nspawn.
4018
d08ee7cb 4019 * The systemd build scripts will no longer complain if the NTP server
23eb30b3 4020 addresses are not changed from the defaults. Google now supports
d08ee7cb
LP
4021 these NTP servers officially. We still recommend downstreams to
4022 properly register an NTP pool with the NTP pool project though.
4023
c1ec34d1 4024 * coredumpctl gained a new "--reverse" option for printing the list
d08ee7cb
LP
4025 of coredumps in reverse order.
4026
23eb30b3
ZJS
4027 * coredumpctl will now show additional information about truncated and
4028 inaccessible coredumps, as well as coredumps that are still being
4029 processed. It also gained a new --quiet switch for suppressing
4030 additional informational message in its output.
4031
4032 * coredumpctl gained support for only showing coredumps newer and/or
4033 older than specific timestamps, using the new --since= and --until=
4034 options, reminiscent of journalctl's options by the same name.
4035
d08ee7cb 4036 * The systemd-coredump logic has been improved so that it may be reused
23eb30b3 4037 to collect backtraces in non-compiled languages, for example in
d08ee7cb
LP
4038 scripting languages such as Python.
4039
4040 * machinectl will now show the UID shift of local containers, if user
4041 namespacing is enabled for them.
4042
baf32786 4043 * systemd will now optionally run "environment generator" binaries at
d08ee7cb
LP
4044 configuration load time. They may be used to add environment
4045 variables to the environment block passed to services invoked. One
baf32786 4046 user environment generator is shipped by default that sets up
23eb30b3
ZJS
4047 environment variables based on files dropped into /etc/environment.d
4048 and ~/.config/environment.d/.
d08ee7cb 4049
a2b53448
LP
4050 * systemd-resolved now includes the new, recently published 2017 DNSSEC
4051 root key (KSK).
4052
a2b53448
LP
4053 * hostnamed has been updated to report a new chassis type of
4054 "convertible" to cover "foldable" laptops that can both act as a
4055 tablet and as a laptop, such as various Lenovo Yoga devices.
4056
d08ee7cb
LP
4057 Contributions from: Adrián López, Alexander Galanin, Alexander
4058 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
4059 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
4060 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
4061 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
4062 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
4063 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
4064 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
4065 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
a2b53448
LP
4066 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
4067 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
4068 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
4069 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
4070 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
4071 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
4072 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
4073 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
4074 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
4075 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
4076 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
4077 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
4078 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
4079 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
4080 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
4081 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
4082 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
4083 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
4084 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
4085 Тихонов
4086
4087 — Berlin, 2017-03-01
d08ee7cb 4088
54b24597 4089CHANGES WITH 232:
76153ad4 4090
05f426d2
LP
4091 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
4092 RestrictAddressFamilies= enabled. These sandboxing options should
4093 generally be compatible with the various external udev call-out
4094 binaries we are aware of, however there may be exceptions, in
4095 particular when exotic languages for these call-outs are used. In
4096 this case, consider turning off these settings locally.
4097
4ffe2479
ZJS
4098 * The new RemoveIPC= option can be used to remove IPC objects owned by
4099 the user or group of a service when that service exits.
4100
6fa44114 4101 * The new ProtectKernelModules= option can be used to disable explicit
4c37970d
LP
4102 load and unload operations of kernel modules by a service. In
4103 addition access to /usr/lib/modules is removed if this option is set.
6fa44114 4104
4a77c53d
ZJS
4105 * ProtectSystem= option gained a new value "strict", which causes the
4106 whole file system tree with the exception of /dev, /proc, and /sys,
4107 to be remounted read-only for a service.
4108
e49e2c25 4109 * The new ProtectKernelTunables= option can be used to disable
4a77c53d
ZJS
4110 modification of configuration files in /sys and /proc by a service.
4111 Various directories and files are remounted read-only, so access is
4112 restricted even if the file permissions would allow it.
4113
6fa44114 4114 * The new ProtectControlGroups= option can be used to disable write
4a77c53d
ZJS
4115 access by a service to /sys/fs/cgroup.
4116
4117 * Various systemd services have been hardened with
4118 ProtectKernelTunables=yes, ProtectControlGroups=yes,
4119 RestrictAddressFamilies=.
4120
4ffe2479
ZJS
4121 * Support for dynamically creating users for the lifetime of a service
4122 has been added. If DynamicUser=yes is specified, user and group IDs
4123 will be allocated from the range 61184..65519 for the lifetime of the
4124 service. They can be resolved using the new nss-systemd.so NSS
4125 module. The module must be enabled in /etc/nsswitch.conf. Services
4126 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
4127 any resources allocated by the service will be cleaned up when the
4a77c53d
ZJS
4128 service exits. They also have ProtectHome=read-only and
4129 ProtectSystem=strict enabled, so they are not able to make any
4130 permanent modifications to the system.
4ffe2479 4131
171ae2cd 4132 * The nss-systemd module also always resolves root and nobody, making
4ffe2479 4133 it possible to have no /etc/passwd or /etc/group files in minimal
171ae2cd 4134 container or chroot environments.
4ffe2479
ZJS
4135
4136 * Services may be started with their own user namespace using the new
171ae2cd
LP
4137 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
4138 under which the service is running are mapped. All other users are
4139 mapped to nobody.
4ffe2479
ZJS
4140
4141 * Support for the cgroup namespace has been added to systemd-nspawn. If
4142 supported by kernel, the container system started by systemd-nspawn
4143 will have its own view of the cgroup hierarchy. This new behaviour
4144 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
4145
4146 * The new MemorySwapMax= option can be used to limit the maximum swap
4147 usage under the unified cgroup hierarchy.
4148
4149 * Support for the CPU controller in the unified cgroup hierarchy has
4150 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
4151 options. This controller requires out-of-tree patches for the kernel
4152 and the support is provisional.
4153
171ae2cd
LP
4154 * Mount and automount units may now be created transiently
4155 (i.e. dynamically at runtime via the bus API, instead of requiring
4156 unit files in the file system).
4157
4158 * systemd-mount is a new tool which may mount file systems – much like
4159 mount(8), optionally pulling in additional dependencies through
4160 transient .mount and .automount units. For example, this tool
4161 automatically runs fsck on a backing block device before mounting,
4162 and allows the automount logic to be used dynamically from the
4163 command line for establishing mount points. This tool is particularly
4164 useful when dealing with removable media, as it will ensure fsck is
4165 run – if necessary – before the first access and that the file system
4166 is quickly unmounted after each access by utilizing the automount
4167 logic. This maximizes the chance that the file system on the
4168 removable media stays in a clean state, and if it isn't in a clean
4169 state is fixed automatically.
4ffe2479
ZJS
4170
4171 * LazyUnmount=yes option for mount units has been added to expose the
4172 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
4173 option.
4174
4175 * /efi will be used as the mount point of the EFI boot partition, if
4176 the directory is present, and the mount point was not configured
4177 through other means (e.g. fstab). If /efi directory does not exist,
4178 /boot will be used as before. This makes it easier to automatically
4179 mount the EFI partition on systems where /boot is used for something
4180 else.
4181
171ae2cd
LP
4182 * When operating on GPT disk images for containers, systemd-nspawn will
4183 now mount the ESP to /boot or /efi according to the same rules as PID
4184 1 running on a host. This allows tools like "bootctl" to operate
4185 correctly within such containers, in order to make container images
4186 bootable on physical systems.
4187
4a77c53d 4188 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
4ffe2479
ZJS
4189
4190 * Two new user session targets have been added to support running
4191 graphical sessions under the systemd --user instance:
4192 graphical-session.target and graphical-session-pre.target. See
4193 systemd.special(7) for a description of how those targets should be
4194 used.
4195
4196 * The vconsole initialization code has been significantly reworked to
d4c08299 4197 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
4ffe2479
ZJS
4198 support unicode keymaps. Font and keymap configuration will now be
4199 copied to all allocated virtual consoles.
4200
05ecf467 4201 * FreeBSD's bhyve virtualization is now detected.
4ffe2479 4202
d4c08299 4203 * Information recorded in the journal for core dumps now includes the
4ffe2479
ZJS
4204 contents of /proc/mountinfo and the command line of the process at
4205 the top of the process hierarchy (which is usually the init process
4206 of the container).
4207
171ae2cd 4208 * systemd-journal-gatewayd learned the --directory= option to serve
4ffe2479
ZJS
4209 files from the specified location.
4210
4211 * journalctl --root=… can be used to peruse the journal in the
4212 /var/log/ directories inside of a container tree. This is similar to
4213 the existing --machine= option, but does not require the container to
4214 be active.
4215
4216 * The hardware database has been extended to support
4217 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
4218 trackball devices.
4219
4220 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
4221 specify the click rate for mice which include a horizontal wheel with
4222 a click rate that is different than the one for the vertical wheel.
4223
4224 * systemd-run gained a new --wait option that makes service execution
171ae2cd
LP
4225 synchronous. (Specifically, the command will not return until the
4226 specified service binary exited.)
4ffe2479 4227
171ae2cd 4228 * systemctl gained a new --wait option that causes the start command to
4a77c53d
ZJS
4229 wait until the units being started have terminated again.
4230
171ae2cd 4231 * A new journal output mode "short-full" has been added which displays
4ffe2479 4232 timestamps with abbreviated English day names and adds a timezone
171ae2cd
LP
4233 suffix. Those timestamps include more information than the default
4234 "short" output mode, and can be passed directly to journalctl's
4235 --since= and --until= options.
4ffe2479
ZJS
4236
4237 * /etc/resolv.conf will be bind-mounted into containers started by
4238 systemd-nspawn, if possible, so any changes to resolv.conf contents
4239 are automatically propagated to the container.
4240
4241 * The number of instances for socket-activated services originating
171ae2cd
LP
4242 from a single IP address can be limited with
4243 MaxConnectionsPerSource=, extending the existing setting of
4244 MaxConnections=.
4ffe2479 4245
4a77c53d
ZJS
4246 * systemd-networkd gained support for vcan ("Virtual CAN") interface
4247 configuration.
4248
4249 * .netdev and .network configuration can now be extended through
4250 drop-ins.
4251
4ffe2479
ZJS
4252 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
4253 Segmentation Offload, Generic Receive Offload, Large Receive Offload
4254 can be enabled and disabled using the new UDPSegmentationOffload=,
4255 TCPSegmentationOffload=, GenericSegmentationOffload=,
4256 GenericReceiveOffload=, LargeReceiveOffload= options in the
4257 [Link] section of .link files.
4258
171ae2cd
LP
4259 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
4260 Port VLAN ID can be configured for bridge devices using the new STP=,
4261 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
4262 section of .netdev files.
4ffe2479 4263
171ae2cd 4264 * The route table to which routes received over DHCP or RA should be
4a77c53d
ZJS
4265 added can be configured with the new RouteTable= option in the [DHCP]
4266 and [IPv6AcceptRA] sections of .network files.
4267
171ae2cd 4268 * The Address Resolution Protocol can be disabled on links managed by
4ffe2479
ZJS
4269 systemd-networkd using the ARP=no setting in the [Link] section of
4270 .network files.
4271
171ae2cd
LP
4272 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
4273 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
4274 encode information about the result and exit codes of the current
4275 service runtime cycle.
4ffe2479 4276
4a77c53d 4277 * systemd-sysctl will now configure kernel parameters in the order
1f4f4cf7 4278 they occur in the configuration files. This matches what sysctl
4a77c53d
ZJS
4279 has been traditionally doing.
4280
4281 * kernel-install "plugins" that are executed to perform various
4282 tasks after a new kernel is added and before an old one is removed
4283 can now return a special value to terminate the procedure and
4284 prevent any later plugins from running.
4285
76153ad4 4286 * Journald's SplitMode=login setting has been deprecated. It has been
d4c08299 4287 removed from documentation, and its use is discouraged. In a future
76153ad4
ZJS
4288 release it will be completely removed, and made equivalent to current
4289 default of SplitMode=uid.
4290
4a77c53d
ZJS
4291 * Storage=both option setting in /etc/systemd/coredump.conf has been
4292 removed. With fast LZ4 compression storing the core dump twice is not
4293 useful.
4294
4ffe2479
ZJS
4295 * The --share-system systemd-nspawn option has been replaced with an
4296 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
4297 this functionality is discouraged. In addition the variables
4298 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
4299 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
4300 individual namespaces.
4301
171ae2cd
LP
4302 * "machinectl list" now shows the IP address of running containers in
4303 the output, as well as OS release information.
4304
4305 * "loginctl list" now shows the TTY of each session in the output.
4306
4307 * sd-bus gained new API calls sd_bus_track_set_recursive(),
4308 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
4309 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
4310 tracking objects in a "recursive" mode, where a single client can be
4311 counted multiple times, if it takes multiple references.
4312
4313 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
bc99dac5 4314 sd_bus_get_exit_on_disconnect(). They may be used to make a
171ae2cd
LP
4315 process using sd-bus automatically exit if the bus connection is
4316 severed.
4317
4318 * Bus clients of the service manager may now "pin" loaded units into
4319 memory, by taking an explicit reference on them. This is useful to
4320 ensure the client can retrieve runtime data about the service even
4321 after the service completed execution. Taking such a reference is
4322 available only for privileged clients and should be helpful to watch
4323 running services in a race-free manner, and in particular collect
4324 information about exit statuses and results.
4325
4c37970d
LP
4326 * The nss-resolve module has been changed to strictly return UNAVAIL
4327 when communication via D-Bus with resolved failed, and NOTFOUND when
4328 a lookup completed but was negative. This means it is now possible to
4329 neatly configure fallbacks using nsswitch.conf result checking
4330 expressions. Taking benefit of this, the new recommended
4331 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
4332
4333 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
4334
4335 * A new setting CtrlAltDelBurstAction= has been added to
4336 /etc/systemd/system.conf which may be used to configure the precise
4337 behaviour if the user on the console presses Ctrl-Alt-Del more often
4338 than 7 times in 2s. Previously this would unconditionally result in
4339 an expedited, immediate reboot. With this new setting the precise
4340 operation may be configured in more detail, and also turned off
4341 entirely.
4342
4343 * In .netdev files two new settings RemoteChecksumTx= and
4344 RemoteChecksumRx= are now understood that permit configuring the
4345 remote checksumming logic for VXLAN networks.
4346
4347 * The service manager learnt a new "invocation ID" concept for invoked
4348 services. Each runtime cycle of a service will get a new invocation
4349 ID (a 128bit random UUID) assigned that identifies the current
4350 run of the service uniquely and globally. A new invocation ID
4351 is generated each time a service starts up. The journal will store
4352 the invocation ID of a service along with any logged messages, thus
4353 making the invocation ID useful for matching the online runtime of a
4354 service with the offline log data it generated in a safe way without
4355 relying on synchronized timestamps. In many ways this new service
4356 invocation ID concept is similar to the kernel's boot ID concept that
4357 uniquely and globally identifies the runtime of each boot. The
4358 invocation ID of a service is passed to the service itself via an
4359 environment variable ($INVOCATION_ID). A new bus call
4360 GetUnitByInvocationID() has been added that is similar to GetUnit()
4361 but instead of retrieving the bus path for a unit by its name
4362 retrieves it by its invocation ID. The returned path is valid only as
4363 long as the passed invocation ID is current.
4364
4365 * systemd-resolved gained a new "DNSStubListener" setting in
4366 resolved.conf. It either takes a boolean value or the special values
4367 "udp" and "tcp", and configures whether to enable the stub DNS
4368 listener on 127.0.0.53:53.
4369
4370 * IP addresses configured via networkd may now carry additional
4371 configuration settings supported by the kernel. New options include:
4372 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
4373 PrefixRoute=, AutoJoin=.
4374
4375 * The PAM configuration fragment file for "user@.service" shipped with
4376 systemd (i.e. the --user instance of systemd) has been stripped to
4377 the minimum necessary to make the system boot. Previously, it
4378 contained Fedora-specific stanzas that did not apply to other
4379 distributions. It is expected that downstream distributions add
4380 additional configuration lines, matching their needs to this file,
4381 using it only as rough template of what systemd itself needs. Note
4382 that this reduced fragment does not even include an invocation of
4383 pam_limits which most distributions probably want to add, even though
4384 systemd itself does not need it. (There's also the new build time
4385 option --with-pamconfdir=no to disable installation of the PAM
4386 fragment entirely.)
4387
4388 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
4389 capability is now also dropped from its set (in addition to
4390 CAP_SYS_MKNOD as before).
4391
4392 * In service unit files it is now possible to connect a specific named
4393 file descriptor with stdin/stdout/stdout of an executed service. The
4394 name may be specified in matching .socket units using the
4395 FileDescriptorName= setting.
4396
4397 * A number of journal settings may now be configured on the kernel
4398 command line. Specifically, the following options are now understood:
4399 systemd.journald.max_level_console=,
4400 systemd.journald.max_level_store=,
4401 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
4402 systemd.journald.max_level_wall=.
4403
4404 * "systemctl is-enabled --full" will now show by which symlinks a unit
4405 file is enabled in the unit dependency tree.
4406
b4eed568
LP
4407 * Support for VeraCrypt encrypted partitions has been added to the
4408 "cryptsetup" logic and /etc/crypttab.
4409
4410 * systemd-detect-virt gained support for a new --private-users switch
4411 that checks whether the invoking processes are running inside a user
4412 namespace. Similar, a new special value "private-users" for the
4413 existing ConditionVirtualization= setting has been added, permitting
4414 skipping of specific units in user namespace environments.
4415
07393b6e
LP
4416 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
4417 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
4418 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
4419 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
4420 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
4421 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
4422 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
4423 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
4424 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
4425 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
4426 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
4427 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
4428 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
4429 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
4430 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
4431 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
4432 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
4433 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
4434 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
4435 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
4436 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
4437 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
4438 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
4439 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
4440 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
4441 Jędrzejewski-Szmek, Zeal Jagannatha
4442
54b24597 4443 — Santa Fe, 2016-11-03
07393b6e 4444
5cd118ba
MP
4445CHANGES WITH 231:
4446
fcd30826
LP
4447 * In service units the various ExecXYZ= settings have been extended
4448 with an additional special character as first argument of the
43eb109a 4449 assigned value: if the character '+' is used the specified command
fcd30826
LP
4450 line it will be run with full privileges, regardless of User=,
4451 Group=, CapabilityBoundingSet= and similar options. The effect is
4452 similar to the existing PermissionsStartOnly= option, but allows
4453 configuration of this concept for each executed command line
4454 independently.
4455
4456 * Services may now alter the service watchdog timeout at runtime by
4457 sending a WATCHDOG_USEC= message via sd_notify().
4458
4459 * MemoryLimit= and related unit settings now optionally take percentage
4460 specifications. The percentage is taken relative to the amount of
4461 physical memory in the system (or in case of containers, the assigned
4462 amount of memory). This allows scaling service resources neatly with
771de3f5 4463 the amount of RAM available on the system. Similarly, systemd-logind's
fcd30826
LP
4464 RuntimeDirectorySize= option now also optionally takes percentage
4465 values.
4466
4467 * In similar fashion TasksMax= takes percentage values now, too. The
4468 value is taken relative to the configured maximum number of processes
4469 on the system. The per-service task maximum has been changed to 15%
4470 using this functionality. (Effectively this is an increase of 512 →
4471 4915 for service units, given the kernel's default pid_max setting.)
4472
4473 * Calendar time specifications in .timer units now understand a ".."
4474 syntax for time ranges. Example: "4..7:10" may now be used for
4475 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
4476 7:10am every day.
4477
4478 * The InaccessableDirectories=, ReadOnlyDirectories= and
4479 ReadWriteDirectories= unit file settings have been renamed to
4480 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
4481 applied to all kinds of file nodes, and not just directories, with
4482 the exception of symlinks. Specifically these settings may now be
4483 used on block and character device nodes, UNIX sockets and FIFOS as
4484 well as regular files. The old names of these settings remain
4485 available for compatibility.
4486
4487 * systemd will now log about all service processes it kills forcibly
4488 (using SIGKILL) because they remained after the clean shutdown phase
4489 of the service completed. This should help identifying services that
4490 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
4491 systemd-logind's configuration a similar log message is generated for
4492 processes killed at the end of each session due to this setting.
4493
4494 * systemd will now set the $JOURNAL_STREAM environment variable for all
4495 services whose stdout/stderr are connected to the Journal (which
4496 effectively means by default: all services). The variable contains
4497 the device and inode number of the file descriptor used for
4498 stdout/stderr. This may be used by invoked programs to detect whether
4499 their stdout/stderr is connected to the Journal, in which case they
4500 can switch over to direct Journal communication, thus being able to
4501 pass extended, structured metadata along with their log messages. As
4502 one example, this is now used by glib's logging primitives.
4503
4504 * When using systemd's default tmp.mount unit for /tmp, the mount point
4505 will now be established with the "nosuid" and "nodev" options. This
4506 avoids privilege escalation attacks that put traps and exploits into
4507 /tmp. However, this might cause problems if you e. g. put container
5cd118ba
MP
4508 images or overlays into /tmp; if you need this, override tmp.mount's
4509 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
4510 desired options.
4511
fcd30826 4512 * systemd now supports the "memory" cgroup controller also on
4e1dfa45 4513 cgroup v2.
fcd30826
LP
4514
4515 * The systemd-cgtop tool now optionally takes a control group path as
4516 command line argument. If specified, the control group list shown is
4517 limited to subgroups of that group.
4518
4519 * The SystemCallFilter= unit file setting gained support for
4520 pre-defined, named system call filter sets. For example
4521 SystemCallFilter=@clock is now an effective way to make all clock
771de3f5 4522 changing-related system calls unavailable to a service. A number of
fcd30826
LP
4523 similar pre-defined groups are defined. Writing system call filters
4524 for system services is simplified substantially with this new
4525 concept. Accordingly, all of systemd's own, long-running services now
4526 enable system call filtering based on this, by default.
4527
4528 * A new service setting MemoryDenyWriteExecute= has been added, taking
4529 a boolean value. If turned on, a service may no longer create memory
4530 mappings that are writable and executable at the same time. This
4531 enhances security for services where this is enabled as it becomes
4532 harder to dynamically write and then execute memory in exploited
4533 service processes. This option has been enabled for all of systemd's
4534 own long-running services.
4535
4536 * A new RestrictRealtime= service setting has been added, taking a
4537 boolean argument. If set the service's processes may no longer
4538 acquire realtime scheduling. This improves security as realtime
4539 scheduling may otherwise be used to easily freeze the system.
4540
4541 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
4542 value. This may be used for requesting that the system manager inside
4543 of the container reports start-up completion to nspawn which then
4544 propagates this notification further to the service manager
4545 supervising nspawn itself. A related option NotifyReady= in .nspawn
4546 files has been added too. This functionality allows ordering of the
4547 start-up of multiple containers using the usual systemd ordering
4548 primitives.
4549
4550 * machinectl gained a new command "stop" that is an alias for
4551 "terminate".
4552
4553 * systemd-resolved gained support for contacting DNS servers on
4554 link-local IPv6 addresses.
4555
4556 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
4557 its caches. A method call for requesting the same operation has been
4558 added to the bus API too, and is made available via "systemd-resolve
4559 --flush-caches".
4560
771de3f5 4561 * systemd-resolve gained a new --status switch. If passed a brief
fcd30826
LP
4562 summary of the used DNS configuration with per-interface information
4563 is shown.
4564
4565 * resolved.conf gained a new Cache= boolean option, defaulting to
4566 on. If turned off local DNS caching is disabled. This comes with a
4567 performance penalty in particular when DNSSEC is enabled. Note that
771de3f5 4568 resolved disables its internal caching implicitly anyway, when the
fcd30826
LP
4569 configured DNS server is on a host-local IP address such as ::1 or
4570 127.0.0.1, thus automatically avoiding double local caching.
4571
4572 * systemd-resolved now listens on the local IP address 127.0.0.53:53
4573 for DNS requests. This improves compatibility with local programs
4574 that do not use the libc NSS or systemd-resolved's bus APIs for name
4575 resolution. This minimal DNS service is only available to local
4576 programs and does not implement the full DNS protocol, but enough to
4577 cover local DNS clients. A new, static resolv.conf file, listing just
4578 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
4579 now recommended to make /etc/resolv.conf a symlink to this file in
4580 order to route all DNS lookups to systemd-resolved, regardless if
4581 done via NSS, the bus API or raw DNS packets. Note that this local
4582 DNS service is not as fully featured as the libc NSS or
4583 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
4584 used to deliver link-local address information (as this implies
4585 sending a local interface index along), LLMNR/mDNS support via this
4586 interface is severely restricted. It is thus strongly recommended for
4587 all applications to use the libc NSS API or native systemd-resolved
4588 bus API instead.
4589
4590 * systemd-networkd's bridge support learned a new setting
4591 VLANFiltering= for controlling VLAN filtering. Moreover a new section
4592 in .network files has been added for configuring VLAN bridging in
4593 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
4594
4595 * systemd-networkd's IPv6 Router Advertisement code now makes use of
4596 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
4597 now be acquired without relying on DHCPv6. Two new options
4598 UseDomains= and UseDNS= have been added to configure this behaviour.
4599
4600 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
4601 renamed IPv6AcceptRA=, without altering its behaviour. The old
4602 setting name remains available for compatibility reasons.
4603
4604 * The systemd-networkd VTI/VTI6 tunneling support gained new options
4605 Key=, InputKey= and OutputKey=.
4606
4607 * systemd-networkd gained support for VRF ("Virtual Routing Function")
4608 interface configuration.
4609
4610 * "systemctl edit" may now be used to create new unit files by
4611 specifying the --force switch.
4612
4613 * sd-event gained a new function sd_event_get_iteration() for
4614 requesting the current iteration counter of the event loop. It starts
4615 at zero and is increased by one with each event loop iteration.
4616
43a569a1
ZJS
4617 * A new rpm macro %systemd_ordering is provided by the macros.systemd
4618 file. It can be used in lieu of %systemd_requires in packages which
4619 don't use any systemd functionality and are intended to be installed
4620 in minimal containers without systemd present. This macro provides
ce830873 4621 ordering dependencies to ensure that if the package is installed in
43a569a1
ZJS
4622 the same rpm transaction as systemd, systemd will be installed before
4623 the scriptlets for the package are executed, allowing unit presets
4624 to be handled.
4625
4626 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
4627 been added to simplify packaging of generators.
4628
4629 * The os-release file gained VERSION_CODENAME field for the
4630 distribution nickname (e.g. VERSION_CODENAME=woody).
4631
4632 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
4633 can be set to disable parsing of metadata and the creation
4634 of persistent symlinks for that device.
4635
0f1da52b
LP
4636 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
4637 to make them available to logged-in users has been reverted.
4638
4639 * Much of the common code of the various systemd components is now
4640 built into an internal shared library libsystemd-shared-231.so
4641 (incorporating the systemd version number in the name, to be updated
4642 with future releases) that the components link to. This should
4643 decrease systemd footprint both in memory during runtime and on
4644 disk. Note that the shared library is not for public use, and is
ead6bd25 4645 neither API nor ABI stable, but is likely to change with every new
1ecbf32f
ZJS
4646 released update. Packagers need to make sure that binaries
4647 linking to libsystemd-shared.so are updated in step with the
4648 library.
43a569a1 4649
fcd30826
LP
4650 * Configuration for "mkosi" is now part of the systemd
4651 repository. mkosi is a tool to easily build legacy-free OS images,
4652 and is available on github: https://github.com/systemd/mkosi. If
4653 "mkosi" is invoked in the build tree a new raw OS image is generated
4654 incorporating the systemd sources currently being worked on and a
4655 clean, fresh distribution installation. The generated OS image may be
ce830873 4656 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
fcd30826
LP
4657 UEFI PC. This functionality is particularly useful to easily test
4658 local changes made to systemd in a pristine, defined environment. See
f09eb768 4659 doc/HACKING for details.
ceeddf79 4660
4ffe2479
ZJS
4661 * configure learned the --with-support-url= option to specify the
4662 distribution's bugtracker.
4663
38b383d9
LP
4664 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
4665 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
4666 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
4667 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
4668 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
4669 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
4670 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
4671 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
4672 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
4673 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
4674 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
4675 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
4676 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
4677 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
4678 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
4679 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
771de3f5
ZJS
4680 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
4681 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
38b383d9 4682 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
5cd118ba 4683
38b383d9 4684 — Berlin, 2016-07-25
5cd118ba 4685
46e40fab 4686CHANGES WITH 230:
7f6e8043 4687
61ecb465
LP
4688 * DNSSEC is now turned on by default in systemd-resolved (in
4689 "allow-downgrade" mode), but may be turned off during compile time by
4690 passing "--with-default-dnssec=no" to "configure" (and of course,
4691 during runtime with DNSSEC= in resolved.conf). We recommend
4692 downstreams to leave this on at least during development cycles and
4693 report any issues with the DNSSEC logic upstream. We are very
4694 interested in collecting feedback about the DNSSEC validator and its
4695 limitations in the wild. Note however, that DNSSEC support is
4696 probably nothing downstreams should turn on in stable distros just
96d49011 4697 yet, as it might create incompatibilities with a few DNS servers and
61ecb465
LP
4698 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
4699 automatically whenever we detect such incompatible setups, but there
4700 might be systems we do not cover yet. Hence: please help us testing
4701 the DNSSEC code, leave this on where you can, report back, but then
4702 again don't consider turning this on in your stable, LTS or
e40a326c
LP
4703 production release just yet. (Note that you have to enable
4704 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
38b38500 4705 and its DNSSEC mode for hostname resolution from local
e40a326c 4706 applications.)
61ecb465 4707
96515dbf 4708 * systemd-resolve conveniently resolves DANE records with the --tlsa
e40a326c 4709 option and OPENPGPKEY records with the --openpgp option. It also
e75690c3 4710 supports dumping raw DNS record data via the new --raw= switch.
96515dbf 4711
97e5530c
ZJS
4712 * systemd-logind will now by default terminate user processes that are
4713 part of the user session scope unit (session-XX.scope) when the user
977f2bea 4714 logs out. This behavior is controlled by the KillUserProcesses=
e40a326c
LP
4715 setting in logind.conf, and the previous default of "no" is now
4716 changed to "yes". This means that user sessions will be properly
4717 cleaned up after, but additional steps are necessary to allow
4718 intentionally long-running processes to survive logout.
97e5530c
ZJS
4719
4720 While the user is logged in at least once, user@.service is running,
4721 and any service that should survive the end of any individual login
4722 session can be started at a user service or scope using systemd-run.
e40a326c 4723 systemd-run(1) man page has been extended with an example which shows
8951eaec 4724 how to run screen in a scope unit underneath user@.service. The same
e40a326c 4725 command works for tmux.
97e5530c
ZJS
4726
4727 After the user logs out of all sessions, user@.service will be
4728 terminated too, by default, unless the user has "lingering" enabled.
4729 To effectively allow users to run long-term tasks even if they are
152199f2
ZJS
4730 logged out, lingering must be enabled for them. See loginctl(1) for
4731 details. The default polkit policy was modified to allow users to
4732 set lingering for themselves without authentication.
7f6e8043 4733
95365a57 4734 Previous defaults can be restored at compile time by the
e40a326c 4735 --without-kill-user-processes option to "configure".
7f6e8043 4736
e75690c3
ZJS
4737 * systemd-logind gained new configuration settings SessionsMax= and
4738 InhibitorsMax=, both with a default of 8192. It will not register new
188d3082 4739 user sessions or inhibitors above this limit.
e75690c3
ZJS
4740
4741 * systemd-logind will now reload configuration on SIGHUP.
4742
96515dbf 4743 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
e40a326c 4744 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
25b0e6cb
LP
4745 enable. Also, support for the "io" cgroup controller in the unified
4746 hierarchy has been added, so that the "memory", "pids" and "io" are
4747 now the controllers that are supported on the unified hierarchy.
e40a326c 4748
96515dbf
ZJS
4749 WARNING: it is not possible to use previous systemd versions with
4750 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
4751 is necessary to also update systemd in the initramfs if using the
e40a326c 4752 unified hierarchy. An updated SELinux policy is also required.
96515dbf 4753
e40a326c
LP
4754 * LLDP support has been extended, and both passive (receive-only) and
4755 active (sender) modes are supported. Passive mode ("routers-only") is
8951eaec
ZJS
4756 enabled by default in systemd-networkd. Active LLDP mode is enabled
4757 by default for containers on the internal network. The "networkctl
e40a326c
LP
4758 lldp" command may be used to list information gathered. "networkctl
4759 status" will also show basic LLDP information on connected peers now.
96515dbf 4760
e40a326c
LP
4761 * The IAID and DUID unique identifier sent in DHCP requests may now be
4762 configured for the system and each .network file managed by
e75690c3
ZJS
4763 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
4764
4765 * systemd-networkd gained support for configuring proxy ARP support for
4766 each interface, via the ProxyArp= setting in .network files. It also
4767 gained support for configuring the multicast querier feature of
4768 bridge devices, via the new MulticastQuerier= setting in .netdev
4769 files. Similarly, snooping on the IGMP traffic can be controlled
4770 via the new setting MulticastSnooping=.
4771
4772 A new setting PreferredLifetime= has been added for addresses
4773 configured in .network file to configure the lifetime intended for an
4774 address.
4775
4776 The systemd-networkd DHCP server gained the option EmitRouter=, which
4777 defaults to yes, to configure whether the DHCP Option 3 (Router)
4778 should be emitted.
96515dbf 4779
e40a326c 4780 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
97e5530c
ZJS
4781 systemd-socket-activate and installed into /usr/bin. It is now fully
4782 supported.
4783
e40a326c
LP
4784 * systemd-journald now uses separate threads to flush changes to disk
4785 when closing journal files, thus reducing impact of slow disk I/O on
4786 logging performance.
96515dbf 4787
e75690c3
ZJS
4788 * The sd-journal API gained two new calls
4789 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
4790 can be used to open journal files using file descriptors instead of
4791 file or directory paths. sd_journal_open_container() has been
4792 deprecated, sd_journal_open_directory_fd() should be used instead
4793 with the flag SD_JOURNAL_OS_ROOT.
4794
4795 * journalctl learned a new output mode "-o short-unix" that outputs log
4796 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
4797 UTC). It also gained support for a new --no-hostname setting to
4798 suppress the hostname column in the family of "short" output modes.
4799
e40a326c
LP
4800 * systemd-ask-password now optionally skips printing of the password to
4801 stdout with --no-output which can be useful in scripts.
96515dbf
ZJS
4802
4803 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
4804 (devices tagged with ID_MAKER_TOOL) are now tagged with
4805 "uaccess" and are available to logged in users.
4806
e75690c3 4807 * The DeviceAllow= unit setting now supports specifiers (with "%").
e40a326c
LP
4808
4809 * "systemctl show" gained a new --value switch, which allows print a
4810 only the contents of a specific unit property, without also printing
8951eaec
ZJS
4811 the property's name. Similar support was added to "show*" verbs
4812 of loginctl and machinectl that output "key=value" lists.
e40a326c 4813
e75690c3
ZJS
4814 * A new unit type "generated" was added for files dynamically generated
4815 by generator tools. Similarly, a new unit type "transient" is used
4816 for unit files created using the runtime API. "systemctl enable" will
4817 refuse to operate on such files.
4818
e40a326c
LP
4819 * A new command "systemctl revert" has been added that may be used to
4820 revert to the vendor version of a unit file, in case local changes
4821 have been made by adding drop-ins or overriding the unit file.
4822
4823 * "machinectl clean" gained a new verb to automatically remove all or
4824 just hidden container images.
4825
e40a326c
LP
4826 * systemd-tmpfiles gained support for a new line type "e" for emptying
4827 directories, if they exist, without creating them if they don't.
4828
e40a326c
LP
4829 * systemd-nspawn gained support for automatically patching the UID/GIDs
4830 of the owners and the ACLs of all files and directories in a
4831 container tree to match the UID/GID user namespacing range selected
4832 for the container invocation. This mode is enabled via the new
24597ee0
ZJS
4833 --private-users-chown switch. It also gained support for
4834 automatically choosing a free, previously unused UID/GID range when
4835 starting a container, via the new --private-users=pick setting (which
4836 implies --private-users-chown). Together, these options for the first
4837 time make user namespacing for nspawn containers fully automatic and
4838 thus deployable. The systemd-nspawn@.service template unit file has
4839 been changed to use this functionality by default.
e40a326c 4840
25b0e6cb
LP
4841 * systemd-nspawn gained a new --network-zone= switch, that allows
4842 creating ad-hoc virtual Ethernet links between multiple containers,
4843 that only exist as long as at least one container referencing them is
4844 running. This allows easy connecting of multiple containers with a
4845 common link that implements an Ethernet broadcast domain. Each of
4846 these network "zones" may be named relatively freely by the user, and
4847 may be referenced by any number of containers, but each container may
4848 only reference one of these "zones". On the lower level, this is
4849 implemented by an automatically managed bridge network interface for
4850 each zone, that is created when the first container referencing its
4851 zone is created and removed when the last one referencing its zone
4852 terminates.
4853
e40a326c 4854 * The default start timeout may now be configured on the kernel command
8951eaec
ZJS
4855 line via systemd.default_timeout_start_sec=. It was already
4856 configurable via the DefaultTimeoutStartSec= option in
4857 /etc/systemd/system.conf.
e40a326c 4858
030bd839 4859 * Socket units gained a new TriggerLimitIntervalSec= and
e40a326c
LP
4860 TriggerLimitBurst= setting to configure a limit on the activation
4861 rate of the socket unit.
4862
4863 * The LimitNICE= setting now optionally takes normal UNIX nice values
4864 in addition to the raw integer limit value. If the specified
4865 parameter is prefixed with "+" or "-" and is in the range -20..19 the
4866 value is understood as UNIX nice value. If not prefixed like this it
4867 is understood as raw RLIMIT_NICE limit.
4868
999a43f8
LP
4869 * Note that the effect of the PrivateDevices= unit file setting changed
4870 slightly with this release: the per-device /dev file system will be
4871 mounted read-only from this version on, and will have "noexec"
188d3082 4872 set. This (minor) change of behavior might cause some (exceptional)
999a43f8
LP
4873 legacy software to break, when PrivateDevices=yes is set for its
4874 service. Please leave PrivateDevices= off if you run into problems
4875 with this.
4876
e75690c3
ZJS
4877 * systemd-bootchart has been split out to a separate repository:
4878 https://github.com/systemd/systemd-bootchart
4879
4880 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
4881 merged into the kernel in its current form.
4882
4883 * The compatibility libraries libsystemd-daemon.so,
4884 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
4885 which have been deprecated since systemd-209 have been removed along
4886 with the corresponding pkg-config files. All symbols provided by
4887 those libraries are provided by libsystemd.so.
4888
4889 * The Capabilities= unit file setting has been removed (it is ignored
4890 for backwards compatibility). AmbientCapabilities= and
4891 CapabilityBoundingSet= should be used instead.
4892
4f9020fa
DR
4893 * A new special target has been added, initrd-root-device.target,
4894 which creates a synchronization point for dependencies of the root
4895 device in early userspace. Initramfs builders must ensure that this
4896 target is now included in early userspace.
4897
e75690c3
ZJS
4898 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
4899 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
4900 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
4901 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
4902 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
4903 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
4904 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
4905 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
77ff6022
CG
4906 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
4907 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
e75690c3
ZJS
4908 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
4909 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
4910 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
77ff6022
CG
4911 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
4912 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
4913 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
e75690c3
ZJS
4914 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
4915 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
4916 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
4917 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
4918 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
4919 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
46e40fab
ZJS
4920 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
4921 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
4922 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
4923 Jędrzejewski-Szmek
e40a326c 4924
46e40fab 4925 — Fairfax, 2016-05-21
96515dbf 4926
61f32bff
MP
4927CHANGES WITH 229:
4928
d5f8b295
LP
4929 * The systemd-resolved DNS resolver service has gained a substantial
4930 set of new features, most prominently it may now act as a DNSSEC
4931 validating stub resolver. DNSSEC mode is currently turned off by
ed5f8840
ZJS
4932 default, but is expected to be turned on by default in one of the
4933 next releases. For now, we invite everybody to test the DNSSEC logic
4934 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
4935 service also gained a full set of D-Bus interfaces, including calls
4936 to configure DNS and DNSSEC settings per link (for use by external
4937 network management software). systemd-resolved and systemd-networkd
4938 now distinguish between "search" and "routing" domains. The former
4939 are used to qualify single-label names, the latter are used purely
4940 for routing lookups within certain domains to specific links.
4941 resolved now also synthesizes RRs for all entries from /etc/hosts.
d5f8b295
LP
4942
4943 * The systemd-resolve tool (which is a client utility for
ed5f8840
ZJS
4944 systemd-resolved) has been improved considerably and is now fully
4945 supported and documented. Hence it has moved from /usr/lib/systemd to
4946 /usr/bin.
d5f8b295
LP
4947
4948 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
4949 devices.
4950
a7c723c0
LP
4951 * The coredump collection logic has been reworked: when a coredump is
4952 collected it is now written to disk, compressed and processed
4953 (including stacktrace extraction) from a new instantiated service
4954 systemd-coredump@.service, instead of directly from the
4955 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
4956 processing large coredumps can take up a substantial amount of
4957 resources and time, and this previously happened entirely outside of
4958 systemd's service supervision. With the new logic the core_pattern
4959 hook only does minimal metadata collection before passing off control
4960 to the new instantiated service, which is configured with a time
4961 limit, a nice level and other settings to minimize negative impact on
4962 the rest of the system. Also note that the new logic will honour the
4963 RLIMIT_CORE setting of the crashed process, which now allows users
4964 and processes to turn off coredumping for their processes by setting
4965 this limit.
4966
4967 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
4968 and all forked processes by default. Previously, PID 1 would leave
4969 the setting at "0" for all processes, as set by the kernel. Note that
4970 the resource limit traditionally has no effect on the generated
4971 coredumps on the system if the /proc/sys/kernel/core_pattern hook
4972 logic is used. Since the limit is now honoured (see above) its
4973 default has been changed so that the coredumping logic is enabled by
4974 default for all processes, while allowing specific opt-out.
4975
4976 * When the stacktrace is extracted from processes of system users, this
4977 is now done as "systemd-coredump" user, in order to sandbox this
4978 potentially security sensitive parsing operation. (Note that when
4979 processing coredumps of normal users this is done under the user ID
4980 of process that crashed, as before.) Packagers should take notice
4981 that it is now necessary to create the "systemd-coredump" system user
4982 and group at package installation time.
4983
d5f8b295
LP
4984 * The systemd-activate socket activation testing tool gained support
4985 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
4986 and --seqpacket switches. It also has been extended to support both
4987 new-style and inetd-style file descriptor passing. Use the new
4988 --inetd switch to request inetd-style file descriptor passing.
4989
8968aea0
MP
4990 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
4991 variable, which takes a boolean value. If set to false, ANSI color
d5f8b295
LP
4992 output is disabled in the tools even when run on a terminal that
4993 supports it.
4994
4995 * The VXLAN support in networkd now supports two new settings
4996 DestinationPort= and PortRange=.
4997
4998 * A new systemd.machine_id= kernel command line switch has been added,
4999 that may be used to set the machine ID in /etc/machine-id if it is
5000 not initialized yet. This command line option has no effect if the
5001 file is already initialized.
5002
5003 * systemd-nspawn gained a new --as-pid2 switch that invokes any
5004 specified command line as PID 2 rather than PID 1 in the
ed5f8840
ZJS
5005 container. In this mode PID 1 is a minimal stub init process that
5006 implements the special POSIX and Linux semantics of PID 1 regarding
5007 signal and child process management. Note that this stub init process
5008 is implemented in nspawn itself and requires no support from the
5009 container image. This new logic is useful to support running
5010 arbitrary commands in the container, as normal processes are
d5f8b295
LP
5011 generally not prepared to run as PID 1.
5012
5013 * systemd-nspawn gained a new --chdir= switch for setting the current
5014 working directory for the process started in the container.
5015
ed5f8840
ZJS
5016 * "journalctl /dev/sda" will now output all kernel log messages for
5017 specified device from the current boot, in addition to all devices
5018 that are parents of it. This should make log output about devices
5019 pretty useful, as long as kernel drivers attach enough metadata to
5020 the log messages. (The usual SATA drivers do.)
d5f8b295
LP
5021
5022 * The sd-journal API gained two new calls
5023 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
5024 that report whether log data from /run or /var has been found.
5025
5026 * journalctl gained a new switch "--fields" that prints all journal
5027 record field names currently in use in the journal. This is backed
5028 by two new sd-journal API calls sd_journal_enumerate_fields() and
5029 sd_journal_restart_fields().
5030
5031 * Most configurable timeouts in systemd now expect an argument of
8968aea0
MP
5032 "infinity" to turn them off, instead of "0" as before. The semantics
5033 from now on is that a timeout of "0" means "now", and "infinity"
5034 means "never". To maintain backwards compatibility, "0" continues to
5035 turn off previously existing timeout settings.
d5f8b295
LP
5036
5037 * "systemctl reload-or-try-restart" has been renamed to "systemctl
8968aea0
MP
5038 try-reload-or-restart" to clarify what it actually does: the "try"
5039 logic applies to both reloading and restarting, not just restarting.
5040 The old name continues to be accepted for compatibility.
5041
5042 * On boot-up, when PID 1 detects that the system clock is behind the
5043 release date of the systemd version in use, the clock is now set
5044 to the latter. Previously, this was already done in timesyncd, in order
d5f8b295
LP
5045 to avoid running with clocks set to the various clock epochs such as
5046 1902, 1938 or 1970. With this change the logic is now done in PID 1
5047 in addition to timesyncd during early boot-up, so that it is enforced
5048 before the first process is spawned by systemd. Note that the logic
5049 in timesyncd remains, as it is more comprehensive and ensures
ed5f8840 5050 clock monotonicity by maintaining a persistent timestamp file in
d5f8b295
LP
5051 /var. Since /var is generally not available in earliest boot or the
5052 initrd, this part of the logic remains in timesyncd, and is not done
5053 by PID 1.
5054
50f48ad3
DM
5055 * Support for tweaking details in net_cls.class_id through the
5056 NetClass= configuration directive has been removed, as the kernel
5057 people have decided to deprecate that controller in cgroup v2.
5058 Userspace tools such as nftables are moving over to setting rules
5059 that are specific to the full cgroup path of a task, which obsoletes
5060 these controllers anyway. The NetClass= directive is kept around for
5061 legacy compatibility reasons. For a more in-depth description of the
5062 kernel change, please refer to the respective upstream commit:
5063
5064 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
5065
d5f8b295 5066 * A new service setting RuntimeMaxSec= has been added that may be used
8968aea0 5067 to specify a maximum runtime for a service. If the timeout is hit, the
d5f8b295
LP
5068 service is terminated and put into a failure state.
5069
8968aea0
MP
5070 * A new service setting AmbientCapabilities= has been added. It allows
5071 configuration of additional Linux process capabilities that are
5072 passed to the activated processes. This is only available on very
d5f8b295
LP
5073 recent kernels.
5074
5075 * The process resource limit settings in service units may now be used
5076 to configure hard and soft limits individually.
5077
8968aea0 5078 * The various libsystemd APIs such as sd-bus or sd-event now publicly
ed5f8840
ZJS
5079 expose support for gcc's __attribute__((cleanup())) C extension.
5080 Specifically, for many object destructor functions alternative
5081 versions have been added that have names suffixed with "p" and take a
5082 pointer to a pointer to the object to destroy, instead of just a
5083 pointer to the object itself. This is useful because these destructor
5084 functions may be used directly as parameters to the cleanup
5085 construct. Internally, systemd has been a heavy user of this GCC
5086 extension for a long time, and with this change similar support is
5087 now available to consumers of the library outside of systemd. Note
8968aea0 5088 that by using this extension in your sources compatibility with old
ed5f8840
ZJS
5089 and strictly ANSI compatible C compilers is lost. However, all gcc or
5090 LLVM versions of recent years support this extension.
d5f8b295
LP
5091
5092 * Timer units gained support for a new setting RandomizedDelaySec= that
8968aea0
MP
5093 allows configuring some additional randomized delay to the configured
5094 time. This is useful to spread out timer events to avoid load peaks in
5095 clusters or larger setups.
d5f8b295
LP
5096
5097 * Calendar time specifications now support sub-second accuracy.
5098
5099 * Socket units now support listening on SCTP and UDP-lite protocol
5100 sockets.
5101
5102 * The sd-event API now comes with a full set of man pages.
5103
5104 * Older versions of systemd contained experimental support for
5105 compressing journal files and coredumps with the LZ4 compressor that
5106 was not compatible with the lz4 binary (due to API limitations of the
5107 lz4 library). This support has been removed; only support for files
5108 compatible with the lz4 binary remains. This LZ4 logic is now
5109 officially supported and no longer considered experimental.
5110
5111 * The dkr image import logic has been removed again from importd. dkr's
5112 micro-services focus doesn't fit into the machine image focus of
5113 importd, and quickly got out of date with the upstream dkr API.
5114
5115 * Creation of the /run/lock/lockdev/ directory was dropped from
5116 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
61f32bff
MP
5117 been available for many years. If you still need this, you need to
5118 create your own tmpfiles.d config file with:
d5f8b295
LP
5119
5120 d /run/lock/lockdev 0775 root lock -
61f32bff 5121
dd95b381
LP
5122 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
5123 and RebootArgument= have been moved from the [Service] section of
5124 unit files to [Unit], and they are now supported on all unit types,
5125 not just service units. Of course, systemd will continue to
5126 understand these settings also at the old location, in order to
5127 maintain compatibility.
5128
3545ab35
LP
5129 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
5130 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
5131 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
5132 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
5133 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
5134 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
5135 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
5136 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
5137 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
5138 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
5139 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
5140 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
5141 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
5142 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
5143 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
5144 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
5145 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
5146 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
5147 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
5148
ccddd104 5149 — Berlin, 2016-02-11
61f32bff 5150
a11c7ea5
LP
5151CHANGES WITH 228:
5152
a11c7ea5
LP
5153 * A number of properties previously only settable in unit
5154 files are now also available as properties to set when
5155 creating transient units programmatically via the bus, as it
5156 is exposed with systemd-run's --property=
5157 setting. Specifically, these are: SyslogIdentifier=,
5158 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
5159 EnvironmentFile=, ReadWriteDirectories=,
5160 ReadOnlyDirectories=, InaccessibleDirectories=,
5161 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
5162
28c85daf
LP
5163 * When creating transient services via the bus API it is now
5164 possible to pass in a set of file descriptors to use as
5165 STDIN/STDOUT/STDERR for the invoked process.
a11c7ea5 5166
f1f8a5a5
LP
5167 * Slice units may now be created transiently via the bus APIs,
5168 similar to the way service and scope units may already be
5169 created transiently.
5170
a11c7ea5
LP
5171 * Wherever systemd expects a calendar timestamp specification
5172 (like in journalctl's --since= and --until= switches) UTC
5173 timestamps are now supported. Timestamps suffixed with "UTC"
5174 are now considered to be in Universal Time Coordinated
5175 instead of the local timezone. Also, timestamps may now
815bb5bd 5176 optionally be specified with sub-second accuracy. Both of
a11c7ea5
LP
5177 these additions also apply to recurring calendar event
5178 specification, such as OnCalendar= in timer units.
5179
28c85daf
LP
5180 * journalctl gained a new "--sync" switch that asks the
5181 journal daemon to write all so far unwritten log messages to
5182 disk and sync the files, before returning.
5183
a11c7ea5
LP
5184 * systemd-tmpfiles learned two new line types "q" and "Q" that
5185 operate like "v", but also set up a basic btrfs quota
5186 hierarchy when used on a btrfs file system with quota
5187 enabled.
5188
f1f8a5a5
LP
5189 * tmpfiles' "v", "q" and "Q" will now create a plain directory
5190 instead of a subvolume (even on a btrfs file system) if the
5191 root directory is a plain directory, and not a
5192 subvolume. This should simplify things with certain chroot()
5193 environments which are not aware of the concept of btrfs
5194 subvolumes.
5195
a11c7ea5
LP
5196 * systemd-detect-virt gained a new --chroot switch to detect
5197 whether execution takes place in a chroot() environment.
5198
28c85daf 5199 * CPUAffinity= now takes CPU index ranges in addition to
a11c7ea5
LP
5200 individual indexes.
5201
28c85daf
LP
5202 * The various memory-related resource limit settings (such as
5203 LimitAS=) now understand the usual K, M, G, ... suffixes to
5204 the base of 1024 (IEC). Similar, the time-related resource
5205 limit settings understand the usual min, h, day, ...
5206 suffixes now.
5207
f1f8a5a5
LP
5208 * There's a new system.conf setting DefaultTasksMax= to
5209 control the default TasksMax= setting for services and
5210 scopes running on the system. (TasksMax= is the primary
5211 setting that exposes the "pids" cgroup controller on systemd
5212 and was introduced in the previous systemd release.) The
5213 setting now defaults to 512, which means services that are
5214 not explicitly configured otherwise will only be able to
5215 create 512 processes or threads at maximum, from this
5216 version on. Note that this means that thread- or
5217 process-heavy services might need to be reconfigured to set
5218 TasksMax= to a higher value. It is sufficient to set
5219 TasksMax= in these specific unit files to a higher value, or
5220 even "infinity". Similar, there's now a logind.conf setting
5221 UserTasksMax= that defaults to 4096 and limits the total
5222 number of processes or tasks each user may own
5223 concurrently. nspawn containers also have the TasksMax=
5224 value set by default now, to 8192. Note that all of this
5225 only has an effect if the "pids" cgroup controller is
5226 enabled in the kernel. The general benefit of these changes
5227 should be a more robust and safer system, that provides a
5228 certain amount of per-service fork() bomb protection.
5229
28c85daf
LP
5230 * systemd-nspawn gained the new --network-veth-extra= switch
5231 to define additional and arbitrarily-named virtual Ethernet
5232 links between the host and the container.
5233
5234 * A new service execution setting PassEnvironment= has been
5235 added that allows importing select environment variables
5236 from PID1's environment block into the environment block of
5237 the service.
5238
ddb4b0d3 5239 * Timer units gained support for a new RemainAfterElapse=
595bfe7d 5240 setting which takes a boolean argument. It defaults to on,
ddb4b0d3
LP
5241 exposing behaviour unchanged to previous releases. If set to
5242 off, timer units are unloaded after they elapsed if they
5243 cannot elapse again. This is particularly useful for
5244 transient timer units, which shall not stay around longer
5245 than until they first elapse.
5246
a11c7ea5 5247 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
28c85daf
LP
5248 default now (the kernel default is 16). This is beneficial
5249 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
a11c7ea5
LP
5250 allows substantially larger numbers of queued
5251 datagrams. This should increase the capability of systemd to
5252 parallelize boot-up, as logging and sd_notify() are unlikely
5253 to stall execution anymore. If you need to change the value
5254 from the new defaults, use the usual sysctl.d/ snippets.
5255
28c85daf
LP
5256 * The compression framing format used by the journal or
5257 coredump processing has changed to be in line with what the
5258 official LZ4 tools generate. LZ4 compression support in
5259 systemd was considered unsupported previously, as the format
5260 was not compatible with the normal tools. With this release
5261 this has changed now, and it is hence safe for downstream
5262 distributions to turn it on. While not compressing as well
815bb5bd 5263 as the XZ, LZ4 is substantially faster, which makes
28c85daf
LP
5264 it a good default choice for the compression logic in the
5265 journal and in coredump handling.
a11c7ea5 5266
28c85daf
LP
5267 * Any reference to /etc/mtab has been dropped from
5268 systemd. The file has been obsolete since a while, but
5269 systemd refused to work on systems where it was incorrectly
815bb5bd 5270 set up (it should be a symlink or non-existent). Please make
28c85daf
LP
5271 sure to update to util-linux 2.27.1 or newer in conjunction
5272 with this systemd release, which also drops any reference to
5273 /etc/mtab. If you maintain a distribution make sure that no
5274 software you package still references it, as this is a
5275 likely source of bugs. There's also a glibc bug pending,
5276 asking for removal of any reference to this obsolete file:
5277
5278 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
a11c7ea5 5279
d5bd92bb
LP
5280 Note that only util-linux versions built with
5281 --enable-libmount-force-mountinfo are supported.
5282
a11c7ea5
LP
5283 * Support for the ".snapshot" unit type has been removed. This
5284 feature turned out to be little useful and little used, and
5285 has now been removed from the core and from systemctl.
5286
b9e2f7eb
LP
5287 * The dependency types RequiresOverridable= and
5288 RequisiteOverridable= have been removed from systemd. They
5289 have been used only very sparingly to our knowledge and
5290 other options that provide a similar effect (such as
5291 systemctl --mode=ignore-dependencies) are much more useful
5292 and commonly used. Moreover, they were only half-way
5293 implemented as the option to control behaviour regarding
5294 these dependencies was never added to systemctl. By removing
5295 these dependency types the execution engine becomes a bit
5296 simpler. Unit files that use these dependencies should be
5297 changed to use the non-Overridable dependency types
5298 instead. In fact, when parsing unit files with these
5299 options, that's what systemd will automatically convert them
5300 too, but it will also warn, asking users to fix the unit
5301 files accordingly. Removal of these dependency types should
5302 only affect a negligible number of unit files in the wild.
5303
5304 * Behaviour of networkd's IPForward= option changed
5305 (again). It will no longer maintain a per-interface setting,
5306 but propagate one way from interfaces where this is enabled
5307 to the global kernel setting. The global setting will be
5308 enabled when requested by a network that is set up, but
5309 never be disabled again. This change was made to make sure
5310 IPv4 and IPv6 behaviour regarding packet forwarding is
5311 similar (as the Linux IPv6 stack does not support
5312 per-interface control of this setting) and to minimize
5313 surprises.
5314
28c85daf
LP
5315 * In unit files the behaviour of %u, %U, %h, %s has
5316 changed. These specifiers will now unconditionally resolve
5317 to the various user database fields of the user that the
5318 systemd instance is running as, instead of the user
5319 configured in the specific unit via User=. Note that this
5320 effectively doesn't change much, as resolving of these
5321 specifiers was already turned off in the --system instance
5322 of systemd, as we cannot do NSS lookups from PID 1. In the
5323 --user instance of systemd these specifiers where correctly
5324 resolved, but hardly made any sense, since the user instance
5325 lacks privileges to do user switches anyway, and User= is
ce830873 5326 hence useless. Moreover, even in the --user instance of
28c85daf
LP
5327 systemd behaviour was awkward as it would only take settings
5328 from User= assignment placed before the specifier into
5329 account. In order to unify and simplify the logic around
5330 this the specifiers will now always resolve to the
5331 credentials of the user invoking the manager (which in case
5332 of PID 1 is the root user).
5333
5334 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
5335 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
5336 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
f1f8a5a5
LP
5337 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
5338 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
5339 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
5340 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
5341 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
5342 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
5343 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
5344 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
5345 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
5346 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
5347 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
5348 Jędrzejewski-Szmek
28c85daf 5349
ccddd104 5350 — Berlin, 2015-11-18
a11c7ea5 5351
c97e586d
DM
5352CHANGES WITH 227:
5353
5354 * systemd now depends on util-linux v2.27. More specifically,
5355 the newly added mount monitor feature in libmount now
5356 replaces systemd's former own implementation.
5357
5358 * libmount mandates /etc/mtab not to be regular file, and
5359 systemd now enforces this condition at early boot.
5360 /etc/mtab has been deprecated and warned about for a very
5361 long time, so systems running systemd should already have
5362 stopped having this file around as anything else than a
5363 symlink to /proc/self/mounts.
5364
d046fb93
LP
5365 * Support for the "pids" cgroup controller has been added. It
5366 allows accounting the number of tasks in a cgroup and
c97e586d
DM
5367 enforcing limits on it. This adds two new setting
5368 TasksAccounting= and TasksMax= to each unit, as well as a
6fd5517b 5369 global option DefaultTasksAccounting=.
c97e586d
DM
5370
5371 * Support for the "net_cls" cgroup controller has been added.
fe08a30b
LP
5372 It allows assigning a net class ID to each task in the
5373 cgroup, which can then be used in firewall rules and traffic
5374 shaping configurations. Note that the kernel netfilter net
5375 class code does not currently work reliably for ingress
5376 packets on unestablished sockets.
c97e586d
DM
5377
5378 This adds a new config directive called NetClass= to CGroup
6fd5517b 5379 enabled units. Allowed values are positive numbers for fixed
c97e586d
DM
5380 assignments and "auto" for picking a free value
5381 automatically.
5382
21d86c61
DM
5383 * 'systemctl is-system-running' now returns 'offline' if the
5384 system is not booted with systemd. This command can now be
5385 used as a substitute for 'systemd-notify --booted'.
5386
5387 * Watchdog timeouts have been increased to 3 minutes for all
5388 in-tree service files. Apparently, disk IO issues are more
5389 frequent than we hoped, and user reported >1 minute waiting
5390 for disk IO.
5391
5392 * 'machine-id-commit' functionality has been merged into
5393 'machine-id-setup --commit'. The separate binary has been
5394 removed.
5395
d046fb93
LP
5396 * The WorkingDirectory= directive in unit files may now be set
5397 to the special value '~'. In this case, the working
5398 directory is set to the home directory of the user
5399 configured in User=.
21d86c61 5400
fe08a30b
LP
5401 * "machinectl shell" will now open the shell in the home
5402 directory of the selected user by default.
5403
21d86c61 5404 * The CrashChVT= configuration file setting is renamed to
d046fb93
LP
5405 CrashChangeVT=, following our usual logic of not
5406 abbreviating unnecessarily. The old directive is still
5407 supported for compat reasons. Also, this directive now takes
5408 an integer value between 1 and 63, or a boolean value. The
5409 formerly supported '-1' value for disabling stays around for
5410 compat reasons.
21d86c61 5411
fe08a30b 5412 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
8b5f9d15 5413 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
fe08a30b
LP
5414 RootDirectory= properties can now be set for transient
5415 units.
5416
5417 * The systemd-analyze tool gained a new "set-log-target" verb
5418 to change the logging target the system manager logs to
5419 dynamically during runtime. This is similar to how
5420 "systemd-analyze set-log-level" already changes the log
5421 level.
5422
5423 * In nspawn /sys is now mounted as tmpfs, with only a selected
5424 set of subdirectories mounted in from the real sysfs. This
5425 enhances security slightly, and is useful for ensuring user
5426 namespaces work correctly.
5427
5428 * Support for USB FunctionFS activation has been added. This
5429 allows implementation of USB gadget services that are
5430 activated as soon as they are requested, so that they don't
595bfe7d 5431 have to run continuously, similar to classic socket
fe08a30b
LP
5432 activation.
5433
5434 * The "systemctl exit" command now optionally takes an
5435 additional parameter that sets the exit code to return from
5436 the systemd manager when exiting. This is only relevant when
5437 running the systemd user instance, or when running the
5438 system instance in a container.
5439
5440 * sd-bus gained the new API calls sd_bus_path_encode_many()
5441 and sd_bus_path_decode_many() that allow easy encoding and
5442 decoding of multiple identifier strings inside a D-Bus
5443 object path. Another new call sd_bus_default_flush_close()
5444 has been added to flush and close per-thread default
5445 connections.
5446
5447 * systemd-cgtop gained support for a -M/--machine= switch to
5448 show the control groups within a certain container only.
5449
5450 * "systemctl kill" gained support for an optional --fail
5451 switch. If specified the requested operation will fail of no
5452 processes have been killed, because the unit had no
5453 processes attached, or similar.
5454
bdba9227
DM
5455 * A new systemd.crash_reboot=1 kernel command line option has
5456 been added that triggers a reboot after crashing. This can
5457 also be set through CrashReboot= in systemd.conf.
5458
5459 * The RuntimeDirectory= setting now understands unit
5460 specifiers like %i or %f.
5461
ce830873 5462 * A new (still internal) library API sd-ipv4acd has been added,
fe08a30b
LP
5463 that implements address conflict detection for IPv4. It's
5464 based on code from sd-ipv4ll, and will be useful for
5465 detecting DHCP address conflicts.
5466
bdba9227
DM
5467 * File descriptors passed during socket activation may now be
5468 named. A new API sd_listen_fds_with_names() is added to
a8eaaee7 5469 access the names. The default names may be overridden,
bdba9227
DM
5470 either in the .socket file using the FileDescriptorName=
5471 parameter, or by passing FDNAME= when storing the file
5472 descriptors using sd_notify().
fe08a30b 5473
d046fb93
LP
5474 * systemd-networkd gained support for:
5475
0053598f 5476 - Setting the IPv6 Router Advertisement settings via
edf4126f 5477 IPv6AcceptRouterAdvertisements= in .network files.
d046fb93
LP
5478
5479 - Configuring the HelloTimeSec=, MaxAgeSec= and
5480 ForwardDelaySec= bridge parameters in .netdev files.
5481
5482 - Configuring PreferredSource= for static routes in
edf4126f 5483 .network files.
fe08a30b 5484
bdba9227
DM
5485 * The "ask-password" framework used to query for LUKS harddisk
5486 passwords or SSL passwords during boot gained support for
5487 caching passwords in the kernel keyring, if it is
5488 available. This makes sure that the user only has to type in
5489 a passphrase once if there are multiple objects to unlock
5490 with the same one. Previously, such password caching was
5491 available only when Plymouth was used; this moves the
5492 caching logic into the systemd codebase itself. The
5493 "systemd-ask-password" utility gained a new --keyname=
5494 switch to control which kernel keyring key to use for
5495 caching a password in. This functionality is also useful for
5496 enabling display managers such as gdm to automatically
5497 unlock the user's GNOME keyring if its passphrase, the
5498 user's password and the harddisk password are the same, if
5499 gdm-autologin is used.
fe08a30b
LP
5500
5501 * When downloading tar or raw images using "machinectl
5502 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
5503 file is now also downloaded, if it is available and stored
5504 next to the image file.
c97e586d 5505
91d0d699
LP
5506 * Units of type ".socket" gained a new boolean setting
5507 Writable= which is only useful in conjunction with
5508 ListenSpecial=. If true, enables opening the specified
5509 special file in O_RDWR mode rather than O_RDONLY mode.
5510
5511 * systemd-rfkill has been reworked to become a singleton
5512 service that is activated through /dev/rfkill on each rfkill
5513 state change and saves the settings to disk. This way,
5514 systemd-rfkill is now compatible with devices that exist
5515 only intermittendly, and even restores state if the previous
5516 system shutdown was abrupt rather than clean.
5517
d046fb93
LP
5518 * The journal daemon gained support for vacuuming old journal
5519 files controlled by the number of files that shall remain,
5520 in addition to the already existing control by size and by
5521 date. This is useful as journal interleaving performance
6dd6a9c4 5522 degrades with too many separate journal files, and allows
d046fb93
LP
5523 putting an effective limit on them. The new setting defaults
5524 to 100, but this may be changed by setting SystemMaxFiles=
5525 and RuntimeMaxFiles= in journald.conf. Also, the
5526 "journalctl" tool gained the new --vacuum-files= switch to
5527 manually vacuum journal files to leave only the specified
5528 number of files in place.
c48eb61f 5529
bdba9227
DM
5530 * udev will now create /dev/disk/by-path links for ATA devices
5531 on kernels where that is supported.
c30f086f 5532
efce0ffe 5533 * Galician, Serbian, Turkish and Korean translations were added.
c97e586d 5534
61e6771c
LP
5535 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
5536 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
5537 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
5538 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
5539 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
5540 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
5541 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
5542 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
5543 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
5544 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
5545 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
5546 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
5547 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
5548 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
5549 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
5550 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
5551 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
5552 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
5553
ccddd104 5554 — Berlin, 2015-10-07
c97e586d 5555
c9912c5e
DH
5556CHANGES WITH 226:
5557
5e8d4254
LP
5558 * The DHCP implementation of systemd-networkd gained a set of
5559 new features:
5560
5561 - The DHCP server now supports emitting DNS and NTP
5562 information. It may be enabled and configured via
5563 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
5564 and NTP information is enabled, but no servers are
5565 configured, the corresponding uplink information (if there
5566 is any) is propagated.
5567
5568 - Server and client now support transmission and reception
5569 of timezone information. It can be configured via the
5570 newly introduced network options UseTimezone=,
5571 EmitTimezone=, and Timezone=. Transmission of timezone
5572 information is enabled between host and containers by
5573 default now: the container will change its local timezone
5574 to what the host has set.
5575
5576 - Lease timeouts can now be configured via
5577 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
5578
5579 - The DHCP server improved on the stability of
5580 leases. Clients are more likely to get the same lease
5581 information back, even if the server loses state.
5582
5583 - The DHCP server supports two new configuration options to
5584 control the lease address pool metrics, PoolOffset= and
5585 PoolSize=.
5586
5587 * The encapsulation limit of tunnels in systemd-networkd may
5588 now be configured via 'EncapsulationLimit='. It allows
5589 modifying the maximum additional levels of encapsulation
5590 that are permitted to be prepended to a packet.
5591
5592 * systemd now supports the concept of user buses replacing
5593 session buses, if used with dbus-1.10 (and enabled via dbus
5594 --enable-user-session). It previously only supported this on
5595 kdbus-enabled systems, and this release expands this to
5596 'dbus-daemon' systems.
5597
5598 * systemd-networkd now supports predictable interface names
5599 for virtio devices.
5600
5601 * systemd now optionally supports the new Linux kernel
5602 "unified" control group hierarchy. If enabled via the kernel
5603 command-line option 'systemd.unified_cgroup_hierarchy=1',
5604 systemd will try to mount the unified cgroup hierarchy
5605 directly on /sys/fs/cgroup. If not enabled, or not
5606 available, systemd will fall back to the legacy cgroup
5607 hierarchy setup, as before. Host system and containers can
5608 mix and match legacy and unified hierarchies as they
856ca72b 5609 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
5e8d4254
LP
5610 environment variable to individually select the hierarchy to
5611 use for executed containers. By default, nspawn will use the
5612 unified hierarchy for the containers if the host uses the
5613 unified hierarchy, and the legacy hierarchy otherwise.
5614 Please note that at this point the unified hierarchy is an
5615 experimental kernel feature and is likely to change in one
5616 of the next kernel releases. Therefore, it should not be
5617 enabled by default in downstream distributions yet. The
5618 minimum required kernel version for the unified hierarchy to
5619 work is 4.2. Note that when the unified hierarchy is used
5620 for the first time delegated access to controllers is
5621 safe. Because of this systemd-nspawn containers will get
5622 access to controllers now, as will systemd user
5623 sessions. This means containers and user sessions may now
5624 manage their own resources, partitioning up what the system
5625 grants them.
5626
5627 * A new special scope unit "init.scope" has been introduced
5628 that encapsulates PID 1 of the system. It may be used to
5629 determine resource usage and enforce resource limits on PID
5630 1 itself. PID 1 hence moved out of the root of the control
5631 group tree.
5632
5633 * The cgtop tool gained support for filtering out kernel
5634 threads when counting tasks in a control group. Also, the
5635 count of processes is now recursively summed up by
5636 default. Two options -k and --recursive= have been added to
5637 revert to old behaviour. The tool has also been updated to
5638 work correctly in containers now.
5639
5640 * systemd-nspawn's --bind= and --bind-ro= options have been
5641 extended to allow creation of non-recursive bind mounts.
5642
c626bf1d
DM
5643 * libsystemd gained two new calls sd_pid_get_cgroup() and
5644 sd_peer_get_cgroup() which return the control group path of
5e8d4254
LP
5645 a process or peer of a connected AF_UNIX socket. This
5646 function call is particularly useful when implementing
5647 delegated subtrees support in the control group hierarchy.
5648
5649 * The "sd-event" event loop API of libsystemd now supports
5650 correct dequeuing of real-time signals, without losing
5651 signal events.
5652
d35f51ea
ZJS
5653 * When systemd requests a polkit decision when managing units it
5654 will now add additional fields to the request, including unit
5655 name and desired operation. This enables more powerful polkit
5656 policies, that make decisions depending on these parameters.
c9912c5e 5657
47f5a38c
LP
5658 * nspawn learnt support for .nspawn settings files, that may
5659 accompany the image files or directories of containers, and
5660 may contain additional settings for the container. This is
5661 an alternative to configuring container parameters via the
5662 nspawn command line.
5663
2f77decc
LP
5664 Contributions from: Cristian Rodríguez, Daniel Mack, David
5665 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
5666 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
5667 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
5668 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
5669 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
5670 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
23d08d1b 5671 Andersen, Tom Gundersen, Torstein Husebø
c9912c5e 5672
ccddd104 5673 — Berlin, 2015-09-08
c9912c5e 5674
ec5249a2
DM
5675CHANGES WITH 225:
5676
5e8d4254
LP
5677 * machinectl gained a new verb 'shell' which opens a fresh
5678 shell on the target container or the host. It is similar to
5679 the existing 'login' command of machinectl, but spawns the
5680 shell directly without prompting for username or
5681 password. The pseudo machine '.host' now refers to the local
5682 host and is used by default. Hence, 'machinectl shell' can
5683 be used as replacement for 'su -' which spawns a session as
5684 a fresh systemd unit in a way that is fully isolated from
5685 the originating session.
5686
5687 * systemd-networkd learned to cope with private-zone DHCP
5688 options and allows other programs to query the values.
5689
5690 * SELinux access control when enabling/disabling units is no
d35f51ea
ZJS
5691 longer enforced with this release. The previous implementation
5692 was incorrect, and a new corrected implementation is not yet
5693 available. As unit file operations are still protected via
5694 polkit and D-Bus policy this is not a security problem. Yet,
5695 distributions which care about optimal SELinux support should
5696 probably not stabilize on this release.
5e8d4254
LP
5697
5698 * sd-bus gained support for matches of type "arg0has=", that
5699 test for membership of strings in string arrays sent in bus
5700 messages.
5701
5702 * systemd-resolved now dumps the contents of its DNS and LLMNR
5703 caches to the logs on reception of the SIGUSR1 signal. This
5704 is useful to debug DNS behaviour.
5705
5706 * The coredumpctl tool gained a new --directory= option to
5707 operate on journal files in a specific directory.
5708
5709 * "systemctl reboot" and related commands gained a new
5710 "--message=" option which may be used to set a free-text
5711 wall message when shutting down or rebooting the
5712 system. This message is also logged, which is useful for
5713 figuring out the reason for a reboot or shutdown a
5714 posteriori.
5715
5716 * The "systemd-resolve-host" tool's -i switch now takes
5717 network interface numbers as alternative to interface names.
5718
5719 * A new unit file setting for services has been introduced:
5720 UtmpMode= allows configuration of how precisely systemd
5721 handles utmp and wtmp entries for the service if this is
5722 enabled. This allows writing services that appear similar to
5723 user sessions in the output of the "w", "who", "last" and
5724 "lastlog" tools.
5725
5726 * systemd-resolved will now locally synthesize DNS resource
5727 records for the "localhost" and "gateway" domains as well as
5728 the local hostname. This should ensure that clients querying
5729 RRs via resolved will get similar results as those going via
5730 NSS, if nss-myhostname is enabled.
5731
5732 Contributions from: Alastair Hughes, Alex Crawford, Daniel
5733 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
5734 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
5735 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
5736 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
5737 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
5738 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
5739 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
5740 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
5741 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
5742 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
5743 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
e1439a14 5744
ccddd104 5745 — Berlin, 2015-08-27
ec5249a2 5746
11811e85
DH
5747CHANGES WITH 224:
5748
10fa421c
DH
5749 * The systemd-efi-boot-generator functionality was merged into
5750 systemd-gpt-auto-generator.
5751
5e8d4254
LP
5752 * systemd-networkd now supports Group Policy for vxlan
5753 devices. It can be enabled via the new boolean configuration
5754 option called 'GroupPolicyExtension='.
10fa421c 5755
11811e85
DH
5756 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
5757 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
5758 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
5759
ccddd104 5760 — Berlin, 2015-07-31
11811e85 5761
e57eaef8
DH
5762CHANGES WITH 223:
5763
5764 * The python-systemd code has been removed from the systemd repository.
5765 A new repository has been created which accommodates the code from
5766 now on, and we kindly ask distributions to create a separate package
5767 for this: https://github.com/systemd/python-systemd
5768
01608bc8 5769 * The systemd daemon will now reload its main configuration
e57eaef8
DH
5770 (/etc/systemd/system.conf) on daemon-reload.
5771
5772 * sd-dhcp now exposes vendor specific extensions via
5773 sd_dhcp_lease_get_vendor_specific().
5774
931618d0
DM
5775 * systemd-networkd gained a number of new configuration options.
5776
5777 - A new boolean configuration option for TAP devices called
37d54b93 5778 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
931618d0
DM
5779 device, thus allowing to send and receive GSO packets.
5780
5781 - A new tunnel configuration option called 'CopyDSCP='.
5782 If enabled, the DSCP field of ip6 tunnels is copied into the
5783 decapsulated packet.
5784
5785 - A set of boolean bridge configuration options were added.
5786 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
5787 and 'UnicastFlood=' are now parsed by networkd and applied to the
5788 respective bridge link device via the respective IFLA_BRPORT_*
5789 netlink attribute.
5790
5791 - A new string configuration option to override the hostname sent
5792 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
5793 is true, networkd will use the configured hostname instead of the
5794 system hostname when sending DHCP requests.
5795
5796 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
5797 networkd will configure the IPv6 flow-label of the tunnel device
5798 according to RFC2460.
e57eaef8 5799
f5f113f6
DH
5800 - The 'macvtap' virtual network devices are now supported, similar to
5801 the already supported 'macvlan' devices.
5802
e57eaef8 5803 * systemd-resolved now implements RFC5452 to improve resilience against
01608bc8 5804 cache poisoning. Additionally, source port randomization is enabled
e57eaef8
DH
5805 by default to further protect against DNS spoofing attacks.
5806
5807 * nss-mymachines now supports translating UIDs and GIDs of running
5808 containers with user-namespaces enabled. If a container 'foo'
5809 translates a host uid 'UID' to the container uid 'TUID', then
5810 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
5811 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
5812 mapped as 'vg-foo-TGID'.
5813
5814 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
e4e66993
DH
5815 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
5816 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
5817 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
5818 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
5819 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
5820 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
5821 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
5822 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
5823 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
5824
ccddd104 5825 — Berlin, 2015-07-29
e57eaef8 5826
0db83ad7 5827CHANGES WITH 222:
5541c889 5828
861b02eb
KS
5829 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
5830 There are no known issues with current sysfs, and udev does not need
5831 or should be used to work around such bugs.
5832
5833 * udev does no longer enable USB HID power management. Several reports
5834 indicate, that some devices cannot handle that setting.
0db83ad7
DH
5835
5836 * The udev accelerometer helper was removed. The functionality
5837 is now fully included in iio-sensor-proxy. But this means,
5838 older iio-sensor-proxy versions will no longer provide
5839 accelerometer/orientation data with this systemd version.
5840 Please upgrade iio-sensor-proxy to version 1.0.
5841
5541c889
DH
5842 * networkd gained a new configuration option IPv6PrivacyExtensions=
5843 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
5844 for Stateless Address") on selected networks.
5845
9b361114
DM
5846 * For the sake of fewer build-time dependencies and less code in the
5847 main repository, the python bindings are about to be removed in the
5848 next release. A new repository has been created which accommodates
5849 the code from now on, and we kindly ask distributions to create a
5850 separate package for this. The removal will take place in v223.
5851
5852 https://github.com/systemd/python-systemd
5853
0db83ad7
DH
5854 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
5855 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
5856 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
5857 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
5541c889
DH
5858 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
5859 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
0db83ad7
DH
5860 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
5861 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2d1ca112
DH
5862 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5863 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
0db83ad7 5864
ccddd104 5865 — Berlin, 2015-07-07
0db83ad7 5866
0f0467e6
MP
5867CHANGES WITH 221:
5868
470e72d4 5869 * The sd-bus.h and sd-event.h APIs have now been declared
5f92d24f 5870 stable and have been added to the official interface of
470e72d4
LP
5871 libsystemd.so. sd-bus implements an alternative D-Bus client
5872 library, that is relatively easy to use, very efficient and
5873 supports both classic D-Bus as well as kdbus as transport
5874 backend. sd-event is a generic event loop abstraction that
5875 is built around Linux epoll, but adds features such as event
0aee49d5 5876 prioritization or efficient timer handling. Both APIs are good
470e72d4
LP
5877 choices for C programs looking for a bus and/or event loop
5878 implementation that is minimal and does not have to be
5f92d24f 5879 portable to other kernels.
0f0467e6 5880
470e72d4
LP
5881 * kdbus support is no longer compile-time optional. It is now
5882 always built-in. However, it can still be disabled at
5883 runtime using the kdbus=0 kernel command line setting, and
c6551464 5884 that setting may be changed to default to off, by specifying
470e72d4
LP
5885 --disable-kdbus at build-time. Note though that the kernel
5886 command line setting has no effect if the kdbus.ko kernel
5887 module is not installed, in which case kdbus is (obviously)
5888 also disabled. We encourage all downstream distributions to
0aee49d5 5889 begin testing kdbus by adding it to the kernel images in the
470e72d4
LP
5890 development distributions, and leaving kdbus support in
5891 systemd enabled.
0f0467e6 5892
470e72d4
LP
5893 * The minimal required util-linux version has been bumped to
5894 2.26.
5895
5896 * Support for chkconfig (--enable-chkconfig) was removed in
0aee49d5 5897 favor of calling an abstraction tool
470e72d4
LP
5898 /lib/systemd/systemd-sysv-install. This needs to be
5899 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
5900 in README for details.
5901
5902 * If there's a systemd unit and a SysV init script for the
5903 same service name, and the user executes "systemctl enable"
5904 for it (or a related call), then this will now enable both
5905 (or execute the related operation on both), not just the
5906 unit.
5907
5908 * The libudev API documentation has been converted from gtkdoc
5909 into man pages.
5910
5911 * gudev has been removed from the systemd tree, it is now an
5912 external project.
5913
5914 * The systemd-cgtop tool learnt a new --raw switch to generate
0aee49d5 5915 "raw" (machine parsable) output.
470e72d4
LP
5916
5917 * networkd's IPForwarding= .network file setting learnt the
5918 new setting "kernel", which ensures that networkd does not
5919 change the IP forwarding sysctl from the default kernel
5920 state.
5921
5922 * The systemd-logind bus API now exposes a new boolean
5923 property "Docked" that reports whether logind considers the
5924 system "docked", i.e. connected to a docking station or not.
5925
5926 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
5927 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
5928 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
5929 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
5930 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
5931 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
5932 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
5933 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
5934 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
5935 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
5936 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
b912e251
LP
5937 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
5938 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
5939 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5940 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
5941 Fink, Zbigniew Jędrzejewski-Szmek
470e72d4 5942
ccddd104 5943 — Berlin, 2015-06-19
0f0467e6 5944
481a0aa2
LP
5945CHANGES WITH 220:
5946
f7a73a25
DH
5947 * The gudev library has been extracted into a separate repository
5948 available at: https://git.gnome.org/browse/libgudev/
5949 It is now managed as part of the Gnome project. Distributions
5950 are recommended to pass --disable-gudev to systemd and use
5951 gudev from the Gnome project instead. gudev is still included
5952 in systemd, for now. It will be removed soon, though. Please
5953 also see the announcement-thread on systemd-devel:
56cadcb6 5954 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
f7a73a25 5955
481a0aa2
LP
5956 * systemd now exposes a CPUUsageNSec= property for each
5957 service unit on the bus, that contains the overall consumed
5958 CPU time of a service (the sum of what each process of the
5959 service consumed). This value is only available if
5960 CPUAccounting= is turned on for a service, and is then shown
5961 in the "systemctl status" output.
5962
5963 * Support for configuring alternative mappings of the old SysV
5964 runlevels to systemd targets has been removed. They are now
29d1fcb4 5965 hardcoded in a way that runlevels 2, 3, 4 all map to
481a0aa2
LP
5966 multi-user.target and 5 to graphical.target (which
5967 previously was already the default behaviour).
5968
5969 * The auto-mounter logic gained support for mount point
5970 expiry, using a new TimeoutIdleSec= setting in .automount
5971 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
5972
5973 * The EFI System Partition (ESP) as mounted to /boot by
5974 systemd-efi-boot-generator will now be unmounted
29d1fcb4 5975 automatically after 2 minutes of not being used. This should
481a0aa2
LP
5976 minimize the risk of ESP corruptions.
5977
5978 * New /etc/fstab options x-systemd.requires= and
5979 x-systemd.requires-mounts-for= are now supported to express
5980 additional dependencies for mounts. This is useful for
5981 journalling file systems that support external journal
5982 devices or overlay file systems that require underlying file
5983 systems to be mounted.
5984
5985 * systemd does not support direct live-upgrades (via systemctl
5986 daemon-reexec) from versions older than v44 anymore. As no
5987 distribution we are aware of shipped such old versions in a
5988 stable release this should not be problematic.
5989
5990 * When systemd forks off a new per-connection service instance
5991 it will now set the $REMOTE_ADDR environment variable to the
5992 remote IP address, and $REMOTE_PORT environment variable to
5993 the remote IP port. This behaviour is similar to the
5994 corresponding environment variables defined by CGI.
5995
5996 * systemd-networkd gained support for uplink failure
5997 detection. The BindCarrier= option allows binding interface
5998 configuration dynamically to the link sense of other
5999 interfaces. This is useful to achieve behaviour like in
6000 network switches.
6001
6002 * systemd-networkd gained support for configuring the DHCP
6003 client identifier to use when requesting leases.
6004
6005 * systemd-networkd now has a per-network UseNTP= option to
6006 configure whether NTP server information acquired via DHCP
6007 is passed on to services like systemd-timesyncd.
6008
6009 * systemd-networkd gained support for vti6 tunnels.
6010
1579dd2c
LP
6011 * Note that systemd-networkd manages the sysctl variable
6012 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
6013 it is configured for since v219. The variable controls IP
6014 forwarding, and is a per-interface alternative to the global
6015 /proc/sys/net/ipv[46]/ip_forward. This setting is
6016 configurable in the IPForward= option, which defaults to
6017 "no". This means if networkd is used for an interface it is
6018 no longer sufficient to set the global sysctl option to turn
6019 on IP forwarding! Instead, the .network file option
6020 IPForward= needs to be turned on! Note that the
6021 implementation of this behaviour was broken in v219 and has
6022 been fixed in v220.
6023
481a0aa2
LP
6024 * Many bonding and vxlan options are now configurable in
6025 systemd-networkd.
6026
6027 * systemd-nspawn gained a new --property= setting to set unit
6028 properties for the container scope. This is useful for
ce830873 6029 setting resource parameters (e.g. "CPUShares=500") on
481a0aa2
LP
6030 containers started from the command line.
6031
6032 * systemd-nspawn gained a new --private-users= switch to make
6033 use of user namespacing available on recent Linux kernels.
6034
6035 * systemd-nspawn may now be called as part of a shell pipeline
6036 in which case the pipes used for stdin and stdout are passed
6037 directly to the process invoked in the container, without
6038 indirection via a pseudo tty.
6039
6040 * systemd-nspawn gained a new switch to control the UNIX
6041 signal to use when killing the init process of the container
6042 when shutting down.
6043
6044 * systemd-nspawn gained a new --overlay= switch for mounting
6045 overlay file systems into the container using the new kernel
6046 overlayfs support.
6047
6048 * When a container image is imported via systemd-importd and
6049 the host file system is not btrfs, a loopback block device
6050 file is created in /var/lib/machines.raw with a btrfs file
6051 system inside. It is then mounted to /var/lib/machines to
6052 enable btrfs features for container management. The loopback
6053 file and btrfs file system is grown as needed when container
6054 images are imported via systemd-importd.
6055
6056 * systemd-machined/systemd-importd gained support for btrfs
6057 quota, to enforce container disk space limits on disk. This
6058 is exposed in "machinectl set-limit".
6059
6060 * systemd-importd now can import containers from local .tar,
6061 .raw and .qcow2 images, and export them to .tar and .raw. It
6062 can also import dkr v2 images now from the network (on top
6063 of v1 as before).
6064
6065 * systemd-importd gained support for verifying downloaded
6066 images with gpg2 (previously only gpg1 was supported).
6067
d35f51ea
ZJS
6068 * systemd-machined, systemd-logind, systemd: most bus calls are
6069 now accessible to unprivileged processes via polkit. Also,
6070 systemd-logind will now allow users to kill their own sessions
6071 without further privileges or authorization.
481a0aa2
LP
6072
6073 * systemd-shutdownd has been removed. This service was
6074 previously responsible for implementing scheduled shutdowns
6075 as exposed in /usr/bin/shutdown's time parameter. This
6076 functionality has now been moved into systemd-logind and is
6077 accessible via a bus interface.
6078
6079 * "systemctl reboot" gained a new switch --firmware-setup that
6080 can be used to reboot into the EFI firmware setup, if that
6081 is available. systemd-logind now exposes an API on the bus
6082 to trigger such reboots, in case graphical desktop UIs want
6083 to cover this functionality.
6084
6085 * "systemctl enable", "systemctl disable" and "systemctl mask"
1579dd2c 6086 now support a new "--now" switch. If specified the units
481a0aa2
LP
6087 that are enabled will also be started, and the ones
6088 disabled/masked also stopped.
6089
6090 * The Gummiboot EFI boot loader tool has been merged into
1a2d5fbe
DH
6091 systemd, and renamed to "systemd-boot". The bootctl tool has been
6092 updated to support systemd-boot.
481a0aa2
LP
6093
6094 * An EFI kernel stub has been added that may be used to create
6095 kernel EFI binaries that contain not only the actual kernel,
6096 but also an initrd, boot splash, command line and OS release
6097 information. This combined binary can then be signed as a
6098 single image, so that the firmware can verify it all in one
1a2d5fbe 6099 step. systemd-boot has special support for EFI binaries created
481a0aa2
LP
6100 like this and can extract OS release information from them
6101 and show them in the boot menu. This functionality is useful
6102 to implement cryptographically verified boot schemes.
6103
6104 * Optional support has been added to systemd-fsck to pass
6105 fsck's progress report to an AF_UNIX socket in the file
6106 system.
6107
6108 * udev will no longer create device symlinks for all block
6109 devices by default. A blacklist for excluding special block
6110 devices from this logic has been turned into a whitelist
6111 that requires picking block devices explicitly that require
6112 device symlinks.
6113
6114 * A new (currently still internal) API sd-device.h has been
6115 added to libsystemd. This modernized API is supposed to
6116 replace libudev eventually. In fact, already much of libudev
6117 is now just a wrapper around sd-device.h.
6118
6119 * A new hwdb database for storing metadata about pointing
6120 stick devices has been added.
6121
6122 * systemd-tmpfiles gained support for setting file attributes
6123 similar to the "chattr" tool with new 'h' and 'H' lines.
6124
6125 * systemd-journald will no longer unconditionally set the
6126 btrfs NOCOW flag on new journal files. This is instead done
6127 with tmpfiles snippet using the new 'h' line type. This
6128 allows easy disabling of this logic, by masking the
6129 journal-nocow.conf tmpfiles file.
6130
6131 * systemd-journald will now translate audit message types to
6132 human readable identifiers when writing them to the
6133 journal. This should improve readability of audit messages.
6134
6135 * The LUKS logic gained support for the offset= and skip=
6136 options in /etc/crypttab, as previously implemented by
6137 Debian.
6138
6139 * /usr/lib/os-release gained a new optional field VARIANT= for
6140 distributions that support multiple variants (such as a
6141 desktop edition, a server edition, ...)
6142
6143 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
6144 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
6145 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
6146 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
6147 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
6148 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
6149 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
6150 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
6151 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
6152 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
6153 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
6154 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
6155 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
6156 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
6157 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
6158 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
6159 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
6160 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
6161 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
6162 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
6163 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
6164 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
6165 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
6166 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
6167 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
6168 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
6169 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
6170
ccddd104 6171 — Berlin, 2015-05-22
481a0aa2 6172
615aaf41
LP
6173CHANGES WITH 219:
6174
615aaf41
LP
6175 * Introduce a new API "sd-hwdb.h" for querying the hardware
6176 metadata database. With this minimal interface one can query
6177 and enumerate the udev hwdb, decoupled from the old libudev
6178 library. libudev's interface for this is now only a wrapper
6179 around sd-hwdb. A new tool systemd-hwdb has been added to
6180 interface with and update the database.
6181
6182 * When any of systemd's tools copies files (for example due to
6183 tmpfiles' C lines) a btrfs reflink will attempted first,
6184 before bytewise copying is done.
6185
6186 * systemd-nspawn gained a new --ephemeral switch. When
6187 specified a btrfs snapshot is taken of the container's root
6188 directory, and immediately removed when the container
6189 terminates again. Thus, a container can be started whose
6190 changes never alter the container's root directory, and are
6191 lost on container termination. This switch can also be used
6192 for starting a container off the root file system of the
6193 host without affecting the host OS. This switch is only
6194 available on btrfs file systems.
6195
6196 * systemd-nspawn gained a new --template= switch. It takes the
6197 path to a container tree to use as template for the tree
7edecf21 6198 specified via --directory=, should that directory be
615aaf41
LP
6199 missing. This allows instantiating containers dynamically,
6200 on first run. This switch is only available on btrfs file
6201 systems.
6202
6203 * When a .mount unit refers to a mount point on which multiple
6204 mounts are stacked, and the .mount unit is stopped all of
6205 the stacked mount points will now be unmounted until no
6206 mount point remains.
6207
6208 * systemd now has an explicit notion of supported and
6209 unsupported unit types. Jobs enqueued for unsupported unit
6210 types will now fail with an "unsupported" error code. More
6211 specifically .swap, .automount and .device units are not
6212 supported in containers, .busname units are not supported on
6213 non-kdbus systems. .swap and .automount are also not
6214 supported if their respective kernel compile time options
6215 are disabled.
6216
6217 * machinectl gained support for two new "copy-from" and
6218 "copy-to" commands for copying files from a running
6219 container to the host or vice versa.
6220
6221 * machinectl gained support for a new "bind" command to bind
6222 mount host directories into local containers. This is
6223 currently only supported for nspawn containers.
6224
6225 * networkd gained support for configuring bridge forwarding
6226 database entries (fdb) from .network files.
6227
6228 * A new tiny daemon "systemd-importd" has been added that can
6229 download container images in tar, raw, qcow2 or dkr formats,
6230 and make them available locally in /var/lib/machines, so
6231 that they can run as nspawn containers. The daemon can GPG
6232 verify the downloads (not supported for dkr, since it has no
6233 provisions for verifying downloads). It will transparently
6234 decompress bz2, xz, gzip compressed downloads if necessary,
6235 and restore sparse files on disk. The daemon uses privilege
6236 separation to ensure the actual download logic runs with
94e5ba37 6237 fewer privileges than the daemon itself. machinectl has
615aaf41
LP
6238 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
6239 make the functionality of importd available to the
6240 user. With this in place the Fedora and Ubuntu "Cloud"
6241 images can be downloaded and booted as containers unmodified
6242 (the Fedora images lack the appropriate GPG signature files
6243 currently, so they cannot be verified, but this will change
6244 soon, hopefully). Note that downloading images is currently
6245 only fully supported on btrfs.
6246
6247 * machinectl is now able to list container images found in
6248 /var/lib/machines, along with some metadata about sizes of
6249 disk and similar. If the directory is located on btrfs and
6250 quota is enabled, this includes quota display. A new command
6251 "image-status" has been added that shows additional
6252 information about images.
6253
6254 * machinectl is now able to clone container images
6255 efficiently, if the underlying file system (btrfs) supports
f59dba26 6256 it, with the new "machinectl clone" command. It also
615aaf41
LP
6257 gained commands for renaming and removing images, as well as
6258 marking them read-only or read-write (supported also on
6259 legacy file systems).
6260
6261 * networkd gained support for collecting LLDP network
6262 announcements, from hardware that supports this. This is
6263 shown in networkctl output.
6264
6265 * systemd-run gained support for a new -t (--pty) switch for
6266 invoking a binary on a pty whose input and output is
6267 connected to the invoking terminal. This allows executing
6268 processes as system services while interactively
6269 communicating with them via the terminal. Most interestingly
6270 this is supported across container boundaries. Invoking
6271 "systemd-run -t /bin/bash" is an alternative to running a
6272 full login session, the difference being that the former
6273 will not register a session, nor go through the PAM session
6274 setup.
6275
6276 * tmpfiles gained support for a new "v" line type for creating
6277 btrfs subvolumes. If the underlying file system is a legacy
6278 file system, this automatically degrades to creating a
6279 normal directory. Among others /var/lib/machines is now
6280 created like this at boot, should it be missing.
6281
6282 * The directory /var/lib/containers/ has been deprecated and
6283 been replaced by /var/lib/machines. The term "machines" has
6284 been used in the systemd context as generic term for both
6285 VMs and containers, and hence appears more appropriate for
6286 this, as the directory can also contain raw images bootable
6287 via qemu/kvm.
6288
6289 * systemd-nspawn when invoked with -M but without --directory=
6290 or --image= is now capable of searching for the container
6291 root directory, subvolume or disk image automatically, in
6292 /var/lib/machines. systemd-nspawn@.service has been updated
6293 to make use of this, thus allowing it to be used for raw
6294 disk images, too.
6295
6296 * A new machines.target unit has been introduced that is
6297 supposed to group all containers/VMs invoked as services on
6298 the system. systemd-nspawn@.service has been updated to
6299 integrate with that.
6300
6301 * machinectl gained a new "start" command, for invoking a
6302 container as a service. "machinectl start foo" is mostly
6303 equivalent to "systemctl start systemd-nspawn@foo.service",
6304 but handles escaping in a nicer way.
6305
6306 * systemd-nspawn will now mount most of the cgroupfs tree
6307 read-only into each container, with the exception of the
6308 container's own subtree in the name=systemd hierarchy.
6309
6310 * journald now sets the special FS_NOCOW file flag for its
6311 journal files. This should improve performance on btrfs, by
6312 avoiding heavy fragmentation when journald's write-pattern
6313 is used on COW file systems. It degrades btrfs' data
6314 integrity guarantees for the files to the same levels as for
6315 ext3/ext4 however. This should be OK though as journald does
6316 its own data integrity checks and all its objects are
6317 checksummed on disk. Also, journald should handle btrfs disk
6318 full events a lot more gracefully now, by processing SIGBUS
6319 errors, and not relying on fallocate() anymore.
6320
6321 * When journald detects that journal files it is writing to
6322 have been deleted it will immediately start new journal
6323 files.
6324
6325 * systemd now provides a way to store file descriptors
4c37970d 6326 per-service in PID 1. This is useful for daemons to ensure
615aaf41 6327 that fds they require are not lost during a daemon
94e5ba37 6328 restart. The fds are passed to the daemon on the next
615aaf41
LP
6329 invocation in the same way socket activation fds are
6330 passed. This is now used by journald to ensure that the
6331 various sockets connected to all the system's stdout/stderr
6332 are not lost when journald is restarted. File descriptors
6333 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
6334 an extension to sd_notify(). Note that a limit is enforced
6335 on the number of fds a service can store in PID 1, and it
6336 defaults to 0, so that no fds may be stored, unless this is
6337 explicitly turned on.
6338
6339 * The default TERM variable to use for units connected to a
6340 terminal, when no other value is explicitly is set is now
6341 vt220 rather than vt102. This should be fairly safe still,
6342 but allows PgUp/PgDn work.
6343
6344 * The /etc/crypttab option header= as known from Debian is now
6345 supported.
6346
6347 * "loginctl user-status" and "loginctl session-status" will
6348 now show the last 10 lines of log messages of the
6349 user/session following the status output. Similar,
6350 "machinectl status" will show the last 10 log lines
6351 associated with a virtual machine or container
6352 service. (Note that this is usually not the log messages
6353 done in the VM/container itself, but simply what the
6354 container manager logs. For nspawn this includes all console
6355 output however.)
6356
6357 * "loginctl session-status" without further argument will now
6358 show the status of the session of the caller. Similar,
6359 "lock-session", "unlock-session", "activate",
6360 "enable-linger", "disable-linger" may now be called without
6361 session/user parameter in which case they apply to the
6362 caller's session/user.
6363
6364 * An X11 session scriptlet is now shipped that uploads
6365 $DISPLAY and $XAUTHORITY into the environment of the systemd
6366 --user daemon if a session begins. This should improve
6367 compatibility with X11 enabled applications run as systemd
6368 user services.
6369
6370 * Generators are now subject to masking via /etc and /run, the
6371 same way as unit files.
6372
6373 * networkd .network files gained support for configuring
6374 per-link IPv4/IPv6 packet forwarding as well as IPv4
6375 masquerading. This is by default turned on for veth links to
6376 containers, as registered by systemd-nspawn. This means that
6377 nspawn containers run with --network-veth will now get
6378 automatic routed access to the host's networks without any
6379 further configuration or setup, as long as networkd runs on
6380 the host.
6381
6382 * systemd-nspawn gained the --port= (-p) switch to expose TCP
6383 or UDP posts of a container on the host. With this in place
6384 it is possible to run containers with private veth links
6385 (--network-veth), and have their functionality exposed on
6386 the host as if their services were running directly on the
6387 host.
6388
dd2fd155 6389 * systemd-nspawn's --network-veth switch now gained a short
615aaf41
LP
6390 version "-n", since with the changes above it is now truly
6391 useful out-of-the-box. The systemd-nspawn@.service has been
6392 updated to make use of it too by default.
6393
6394 * systemd-nspawn will now maintain a per-image R/W lock, to
6395 ensure that the same image is not started more than once
6396 writable. (It's OK to run an image multiple times
6397 simultaneously in read-only mode.)
6398
6399 * systemd-nspawn's --image= option is now capable of
6400 dissecting and booting MBR and GPT disk images that contain
6401 only a single active Linux partition. Previously it
6402 supported only GPT disk images with proper GPT type
6403 IDs. This allows running cloud images from major
6404 distributions directly with systemd-nspawn, without
6405 modification.
6406
6407 * In addition to collecting mouse dpi data in the udev
6408 hardware database, there's now support for collecting angle
6409 information for mouse scroll wheels. The database is
7edecf21 6410 supposed to guarantee similar scrolling behavior on mice
615aaf41
LP
6411 that it knows about. There's also support for collecting
6412 information about Touchpad types.
6413
6414 * udev's input_id built-in will now also collect touch screen
6415 dimension data and attach it to probed devices.
6416
6417 * /etc/os-release gained support for a Distribution Privacy
6418 Policy link field.
6419
6420 * networkd gained support for creating "ipvlan", "gretap",
6421 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
6422
6423 * systemd-tmpfiles gained support for "a" lines for setting
6424 ACLs on files.
6425
6426 * systemd-nspawn will now mount /tmp in the container to
6427 tmpfs, automatically.
6428
6429 * systemd now exposes the memory.usage_in_bytes cgroup
6430 attribute and shows it for each service in the "systemctl
6431 status" output, if available.
6432
6433 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
6434 immediate reboot is triggered. This useful if shutdown is
6435 hung and is unable to complete, to expedite the
6436 operation. Note that this kind of reboot will still unmount
6437 all file systems, and hence should not result in fsck being
6438 run on next reboot.
6439
6440 * A .device unit for an optical block device will now be
6441 considered active only when a medium is in the drive. Also,
6442 mount units are now bound to their backing devices thus
6443 triggering automatic unmounting when devices become
6444 unavailable. With this in place systemd will now
6445 automatically unmount left-over mounts when a CD-ROM is
6446 ejected or an USB stick is yanked from the system.
6447
6448 * networkd-wait-online now has support for waiting for
6449 specific interfaces only (with globbing), and for giving up
6450 after a configurable timeout.
6451
6452 * networkd now exits when idle. It will be automatically
6453 restarted as soon as interfaces show up, are removed or
6454 change state. networkd will stay around as long as there is
6455 at least one DHCP state machine or similar around, that keep
6456 it non-idle.
6457
6458 * networkd may now configure IPv6 link-local addressing in
6459 addition to IPv4 link-local addressing.
6460
6461 * The IPv6 "token" for use in SLAAC may now be configured for
6462 each .network interface in networkd.
6463
6464 * Routes configured with networkd may now be assigned a scope
6465 in .network files.
6466
6467 * networkd's [Match] sections now support globbing and lists
6468 of multiple space-separated matches per item.
6469
11ea2781 6470 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
d2c643c6
LP
6471 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
6472 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
6473 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
6474 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
6475 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
6476 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
6477 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
6478 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
6479 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
6480 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
6481 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
6482 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
6483 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
6484 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
11ea2781
LP
6485 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
6486 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
6487 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
6488 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
6489 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
6490 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
6491 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
d2c643c6
LP
6492 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
6493 Hoffmann, Zbigniew Jędrzejewski-Szmek
11ea2781 6494
ccddd104 6495 — Berlin, 2015-02-16
11ea2781 6496
d4f5a1f4
DH
6497CHANGES WITH 218:
6498
f9e00a9f
LP
6499 * When querying unit file enablement status (for example via
6500 "systemctl is-enabled"), a new state "indirect" is now known
6501 which indicates that a unit might not be enabled itself, but
c7683ffb 6502 another unit listed in its Also= setting might be.
f9e00a9f
LP
6503
6504 * Similar to the various existing ConditionXYZ= settings for
b938cb90 6505 units, there are now matching AssertXYZ= settings. While
f9e00a9f
LP
6506 failing conditions cause a unit to be skipped, but its job
6507 to succeed, failing assertions declared like this will cause
6508 a unit start operation and its job to fail.
6509
6510 * hostnamed now knows a new chassis type "embedded".
6511
6512 * systemctl gained a new "edit" command. When used on a unit
b938cb90 6513 file, this allows extending unit files with .d/ drop-in
f9e00a9f
LP
6514 configuration snippets or editing the full file (after
6515 copying it from /usr/lib to /etc). This will invoke the
6516 user's editor (as configured with $EDITOR), and reload the
6517 modified configuration after editing.
6518
6519 * "systemctl status" now shows the suggested enablement state
6520 for a unit, as declared in the (usually vendor-supplied)
6521 system preset files.
6522
38b38500 6523 * nss-myhostname will now resolve the single-label hostname
f9e00a9f
LP
6524 "gateway" to the locally configured default IP routing
6525 gateways, ordered by their metrics. This assigns a stable
6526 name to the used gateways, regardless which ones are
6527 currently configured. Note that the name will only be
6528 resolved after all other name sources (if nss-myhostname is
6529 configured properly) and should hence not negatively impact
38b38500 6530 systems that use the single-label hostname "gateway" in
f9e00a9f
LP
6531 other contexts.
6532
6533 * systemd-inhibit now allows filtering by mode when listing
6534 inhibitors.
6535
122676c9 6536 * Scope and service units gained a new "Delegate" boolean
b938cb90 6537 property, which, when set, allows processes running inside the
122676c9
LP
6538 unit to further partition resources. This is primarily
6539 useful for systemd user instances as well as container
6540 managers.
f9e00a9f
LP
6541
6542 * journald will now pick up audit messages directly from
6543 the kernel, and log them like any other log message. The
6544 audit fields are split up and fully indexed. This means that
6545 journalctl in many ways is now a (nicer!) alternative to
6546 ausearch, the traditional audit client. Note that this
b938cb90 6547 implements only a minimal audit client. If you want the
f9e00a9f
LP
6548 special audit modes like reboot-on-log-overflow, please use
6549 the traditional auditd instead, which can be used in
6550 parallel to journald.
6551
6552 * The ConditionSecurity= unit file option now understands the
6553 special string "audit" to check whether auditing is
6554 available.
6555
6556 * journalctl gained two new commands --vacuum-size= and
6557 --vacuum-time= to delete old journal files until the
a8eaaee7 6558 remaining ones take up no more than the specified size on disk,
f9e00a9f
LP
6559 or are not older than the specified time.
6560
6561 * A new, native PPPoE library has been added to sd-network,
6562 systemd's library of light-weight networking protocols. This
6563 library will be used in a future version of networkd to
6564 enable PPPoE communication without an external pppd daemon.
6565
6566 * The busctl tool now understands a new "capture" verb that
6567 works similar to "monitor", but writes a packet capture
6568 trace to STDOUT that can be redirected to a file which is
6569 compatible with libcap's capture file format. This can then
6570 be loaded in Wireshark and similar tools to inspect bus
6571 communication.
6572
6573 * The busctl tool now understands a new "tree" verb that shows
6574 the object trees of a specific service on the bus, or of all
6575 services.
6576
6577 * The busctl tool now understands a new "introspect" verb that
6578 shows all interfaces and members of objects on the bus,
6579 including their signature and values. This is particularly
6580 useful to get more information about bus objects shown by
6581 the new "busctl tree" command.
6582
6583 * The busctl tool now understands new verbs "call",
6584 "set-property" and "get-property" for invoking bus method
6585 calls, setting and getting bus object properties in a
6586 friendly way.
6587
6588 * busctl gained a new --augment-creds= argument that controls
6589 whether the tool shall augment credential information it
6590 gets from the bus with data from /proc, in a possibly
6591 race-ful way.
6592
6593 * nspawn's --link-journal= switch gained two new values
6594 "try-guest" and "try-host" that work like "guest" and
17c29493 6595 "host", but do not fail if the host has no persistent
f9e00a9f
LP
6596 journalling enabled. -j is now equivalent to
6597 --link-journal=try-guest.
6598
6599 * macvlan network devices created by nspawn will now have
6600 stable MAC addresses.
6601
6602 * A new SmackProcessLabel= unit setting has been added, which
6603 controls the SMACK security label processes forked off by
6604 the respective unit shall use.
6605
d4f5a1f4
DH
6606 * If compiled with --enable-xkbcommon, systemd-localed will
6607 verify x11 keymap settings by compiling the given keymap. It
6608 will spew out warnings if the compilation fails. This
6609 requires libxkbcommon to be installed.
6610
b938cb90 6611 * When a coredump is collected, a larger number of metadata
f9e00a9f 6612 fields is now collected and included in the journal records
b938cb90 6613 created for it. More specifically, control group membership,
f9e00a9f
LP
6614 environment variables, memory maps, working directory,
6615 chroot directory, /proc/$PID/status, and a list of open file
6616 descriptors is now stored in the log entry.
6617
17c29493 6618 * The udev hwdb now contains DPI information for mice. For
f9e00a9f
LP
6619 details see:
6620
6621 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
6622
6623 * All systemd programs that read standalone configuration
6624 files in /etc now also support a corresponding series of
997b2b43
JT
6625 .conf.d configuration directories in /etc/, /run/,
6626 /usr/local/lib/, /usr/lib/, and (if configured with
6627 --enable-split-usr) /lib/. In particular, the following
6628 configuration files now have corresponding configuration
6629 directories: system.conf user.conf, logind.conf,
6630 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
6631 resolved.conf, timesyncd.conf, journal-remote.conf, and
6632 journal-upload.conf. Note that distributions should use the
6633 configuration directories in /usr/lib/; the directories in
6634 /etc/ are reserved for the system administrator.
6635
f9e00a9f
LP
6636 * systemd-rfkill will no longer take the rfkill device name
6637 into account when storing rfkill state on disk, as the name
6638 might be dynamically assigned and not stable. Instead, the
6639 ID_PATH udev variable combined with the rfkill type (wlan,
6640 bluetooth, ...) is used.
6641
6642 * A new service systemd-machine-id-commit.service has been
6643 added. When used on systems where /etc is read-only during
6644 boot, and /etc/machine-id is not initialized (but an empty
6645 file), this service will copy the temporary machine ID
6646 created as replacement into /etc after the system is fully
6647 booted up. This is useful for systems that are freshly
6648 installed with a non-initialized machine ID, but should get
6649 a fixed machine ID for subsequent boots.
6650
6651 * networkd's .netdev files now provide a large set of
a8eaaee7 6652 configuration parameters for VXLAN devices. Similarly, the
f9e00a9f
LP
6653 bridge port cost parameter is now configurable in .network
6654 files. There's also new support for configuring IP source
6655 routing. networkd .link files gained support for a new
6656 OriginalName= match that is useful to match against the
6657 original interface name the kernel assigned. .network files
6658 may include MTU= and MACAddress= fields for altering the MTU
6659 and MAC address while being connected to a specific network
6660 interface.
6661
6662 * The LUKS logic gained supported for configuring
6663 UUID-specific key files. There's also new support for naming
6664 LUKS device from the kernel command line, using the new
6665 luks.name= argument.
6666
6667 * Timer units may now be transiently created via the bus API
6668 (this was previously already available for scope and service
6669 units). In addition it is now possible to create multiple
6670 transient units at the same time with a single bus call. The
6671 "systemd-run" tool has been updated to make use of this for
6672 running commands on a specified time, in at(1)-style.
6673
6674 * tmpfiles gained support for "t" lines, for assigning
6675 extended attributes to files. Among other uses this may be
6676 used to assign SMACK labels to files.
6677
13e92f39
LP
6678 Contributions from: Alin Rauta, Alison Chaiken, Andrej
6679 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
6680 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
6681 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
6682 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
6683 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
6684 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
6685 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
6686 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
6687 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
6688 Michael Biebl, Michael Chapman, Michael Marineau, Michal
7da81d33
LP
6689 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
6690 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
6691 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
6692 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
6693 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6694 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
6695 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
13e92f39 6696
ccddd104 6697 — Berlin, 2014-12-10
f9e00a9f 6698
b62a309a
ZJS
6699CHANGES WITH 217:
6700
78b6b7ce
LP
6701 * journalctl gained the new options -t/--identifier= to match
6702 on the syslog identifier (aka "tag"), as well as --utc to
6703 show log timestamps in the UTC timezone. journalctl now also
6704 accepts -n/--lines=all to disable line capping in a pager.
b62a309a 6705
a65b8245
ZJS
6706 * journalctl gained a new switch, --flush, that synchronously
6707 flushes logs from /run/log/journal to /var/log/journal if
6708 persistent storage is enabled. systemd-journal-flush.service
6709 now waits until the operation is complete.
2a97b03b 6710
b62a309a
ZJS
6711 * Services can notify the manager before they start a reload
6712 (by sending RELOADING=1) or shutdown (by sending
4bdc60cb
LP
6713 STOPPING=1). This allows the manager to track and show the
6714 internal state of daemons and closes a race condition when
78b6b7ce 6715 the process is still running but has closed its D-Bus
4bdc60cb 6716 connection.
b62a309a 6717
78b6b7ce
LP
6718 * Services with Type=oneshot do not have to have any ExecStart
6719 commands anymore.
b62a309a
ZJS
6720
6721 * User units are now loaded also from
6722 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
6723 /run/systemd/user directory that was already previously
6724 supported, but is under the control of the user.
6725
3f9a0a52 6726 * Job timeouts (i.e. timeouts on the time a job that is
4ffd29fd
LP
6727 queued stays in the run queue) can now optionally result in
6728 immediate reboot or power-off actions (JobTimeoutAction= and
6729 JobTimeoutRebootArgument=). This is useful on ".target"
6730 units, to limit the maximum time a target remains
6731 undispatched in the run queue, and to trigger an emergency
6732 operation in such a case. This is now used by default to
6733 turn off the system if boot-up (as defined by everything in
6734 basic.target) hangs and does not complete for at least
6735 15min. Also, if power-off or reboot hang for at least 30min
6736 an immediate power-off/reboot operation is triggered. This
6737 functionality is particularly useful to increase reliability
6738 on embedded devices, but also on laptops which might
6739 accidentally get powered on when carried in a backpack and
6740 whose boot stays stuck in a hard disk encryption passphrase
6741 question.
6742
b62a309a
ZJS
6743 * systemd-logind can be configured to also handle lid switch
6744 events even when the machine is docked or multiple displays
6745 are attached (HandleLidSwitchDocked= option).
6746
6747 * A helper binary and a service have been added which can be
6748 used to resume from hibernation in the initramfs. A
6749 generator will parse the resume= option on the kernel
81c7dd89 6750 command line to trigger resume.
b62a309a 6751
78b6b7ce
LP
6752 * A user console daemon systemd-consoled has been
6753 added. Currently, it is a preview, and will so far open a
6754 single terminal on each session of the user marked as
09077149 6755 Desktop=systemd-console.
b62a309a
ZJS
6756
6757 * Route metrics can be specified for DHCP routes added by
6758 systemd-networkd.
6759
ba8df74b 6760 * The SELinux context of socket-activated services can be set
78b6b7ce 6761 from the information provided by the networking stack
b62a309a
ZJS
6762 (SELinuxContextFromNet= option).
6763
6764 * Userspace firmware loading support has been removed and
6765 the minimum supported kernel version is thus bumped to 3.7.
6766
6767 * Timeout for udev workers has been increased from 1 to 3
6768 minutes, but a warning will be printed after 1 minute to
6769 help diagnose kernel modules that take a long time to load.
6770
78b6b7ce 6771 * Udev rules can now remove tags on devices with TAG-="foobar".
b62a309a 6772
4bdc60cb 6773 * systemd's readahead implementation has been removed. In many
f6d1de85 6774 circumstances it didn't give expected benefits even for
b62a309a 6775 rotational disk drives and was becoming less relevant in the
78b6b7ce
LP
6776 age of SSDs. As none of the developers has been using
6777 rotating media anymore, and nobody stepped up to actively
6778 maintain this component of systemd it has now been removed.
b62a309a 6779
c4ac9900 6780 * Swap units can use Options= to specify discard options.
b62a309a
ZJS
6781 Discard options specified for swaps in /etc/fstab are now
6782 respected.
6783
6784 * Docker containers are now detected as a separate type of
6785 virtualization.
6786
6787 * The Password Agent protocol gained support for queries where
ba8df74b 6788 the user input is shown, useful e.g. for user names.
78b6b7ce
LP
6789 systemd-ask-password gained a new --echo option to turn that
6790 on.
b62a309a 6791
e6c253e3
MS
6792 * The default sysctl.d/ snippets will now set:
6793
6794 net.core.default_qdisc = fq_codel
6795
ba8df74b
KS
6796 This selects Fair Queuing Controlled Delay as the default
6797 queuing discipline for network interfaces. fq_codel helps
e6c253e3
MS
6798 fight the network bufferbloat problem. It is believed to be
6799 a good default with no tuning required for most workloads.
6800 Downstream distributions may override this choice. On 10Gbit
6801 servers that do not do forwarding, "fq" may perform better.
6802 Systems without a good clocksource should use "pfifo_fast".
6803
4bdc60cb
LP
6804 * If kdbus is enabled during build a new option BusPolicy= is
6805 available for service units, that allows locking all service
6806 processes into a stricter bus policy, in order to limit
6807 access to various bus services, or even hide most of them
6808 from the service's view entirely.
6809
6810 * networkctl will now show the .network and .link file
6811 networkd has applied to a specific interface.
6812
6813 * sd-login gained a new API call sd_session_get_desktop() to
6814 query which desktop environment has been selected for a
6815 session.
6816
6817 * UNIX utmp support is now compile-time optional to support
6818 legacy-free systems.
6819
78b6b7ce
LP
6820 * systemctl gained two new commands "add-wants" and
6821 "add-requires" for pulling in units from specific targets
6822 easily.
6823
6824 * If the word "rescue" is specified on the kernel command line
6825 the system will now boot into rescue mode (aka
6826 rescue.target), which was previously available only by
6827 specifying "1" or "systemd.unit=rescue.target" on the kernel
6828 command line. This new kernel command line option nicely
6829 mirrors the already existing "emergency" kernel command line
6830 option.
6831
6832 * New kernel command line options mount.usr=, mount.usrflags=,
d4474c41 6833 mount.usrfstype= have been added that match root=, rootflags=,
78b6b7ce
LP
6834 rootfstype= but allow mounting a specific file system to
6835 /usr.
6836
f6d1de85 6837 * The $NOTIFY_SOCKET is now also passed to control processes of
78b6b7ce
LP
6838 services, not only the main process.
6839
6840 * This version reenables support for fsck's -l switch. This
6841 means at least version v2.25 of util-linux is required for
6842 operation, otherwise dead-locks on device nodes may
6843 occur. Again: you need to update util-linux to at least
6844 v2.25 when updating systemd to v217.
6845
3769415e
TT
6846 * The "multi-seat-x" tool has been removed from systemd, as
6847 its functionality has been integrated into X servers 1.16,
6848 and the tool is hence redundant. It is recommended to update
6849 display managers invoking this tool to simply invoke X
6850 directly from now on, again.
6851
fae9332b 6852 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
d35f51ea
ZJS
6853 message flag has been added for all of systemd's polkit
6854 authenticated method calls has been added. In particular this
6855 now allows optional interactive authorization via polkit for
6856 many of PID1's privileged operations such as unit file
6857 enabling and disabling.
fae9332b 6858
cfa1571b
LP
6859 * "udevadm hwdb --update" learnt a new switch "--usr" for
6860 placing the rebuilt hardware database in /usr instead of
6861 /etc. When used only hardware database entries stored in
6862 /usr will be used, and any user database entries in /etc are
6863 ignored. This functionality is useful for vendors to ship a
6864 pre-built database on systems where local configuration is
6865 unnecessary or unlikely.
6866
7e63dd10
LP
6867 * Calendar time specifications in .timer units now also
6868 understand the strings "semi-annually", "quarterly" and
ba8df74b 6869 "minutely" as shortcuts (in addition to the preexisting
86b52a39 6870 "annually", "hourly", ...).
7e63dd10 6871
d4474c41
TG
6872 * systemd-tmpfiles will now correctly create files in /dev
6873 at boot which are marked for creation only at boot. It is
6874 recommended to always create static device nodes with 'c!'
6875 and 'b!', so that they are created only at boot and not
6876 overwritten at runtime.
6877
3b187c5c
LP
6878 * When the watchdog logic is used for a service (WatchdogSec=)
6879 and the watchdog timeout is hit the service will now be
6880 terminated with SIGABRT (instead of just SIGTERM), in order
6881 to make sure a proper coredump and backtrace is
6882 generated. This ensures that hanging services will result in
6883 similar coredump/backtrace behaviour as services that hit a
6884 segmentation fault.
6885
4b08dd87
LP
6886 Contributions from: Andreas Henriksson, Andrei Borzenkov,
6887 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
6888 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
6889 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
6890 Herrmann, David Sommerseth, David Strauss, Emil Renner
6891 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
6892 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
6893 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
6894 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
6895 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
6896 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
6897 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
6898 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
6899 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
6900 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
6901 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
6902 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
6903 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
6904 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
6905 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
6906 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
13e92f39 6907 Jędrzejewski-Szmek
4b08dd87 6908
ccddd104 6909 — Berlin, 2014-10-28
4b08dd87 6910
b72ddf0f 6911CHANGES WITH 216:
b2ca0d63
LP
6912
6913 * timedated no longer reads NTP implementation unit names from
b72ddf0f 6914 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
b2ca0d63
LP
6915 implementations should add a
6916
b72ddf0f 6917 Conflicts=systemd-timesyncd.service
b2ca0d63
LP
6918
6919 to their unit files to take over and replace systemd's NTP
6920 default functionality.
6921
6922 * systemd-sysusers gained a new line type "r" for configuring
6923 which UID/GID ranges to allocate system users/groups
6924 from. Lines of type "u" may now add an additional column
6925 that specifies the home directory for the system user to be
6926 created. Also, systemd-sysusers may now optionally read user
6927 information from STDIN instead of a file. This is useful for
6928 invoking it from RPM preinst scriptlets that need to create
6929 users before the first RPM file is installed since these
6930 files might need to be owned by them. A new
6931 %sysusers_create_inline RPM macro has been introduced to do
6932 just that. systemd-sysusers now updates the shadow files as
6933 well as the user/group databases, which should enhance
6934 compatibility with certain tools like grpck.
6935
d35f51ea
ZJS
6936 * A number of bus APIs of PID 1 now optionally consult polkit to
6937 permit access for otherwise unprivileged clients under certain
6938 conditions. Note that this currently doesn't support
6939 interactive authentication yet, but this is expected to be
6940 added eventually, too.
b2ca0d63
LP
6941
6942 * /etc/machine-info now has new fields for configuring the
6943 deployment environment of the machine, as well as the
6944 location of the machine. hostnamectl has been updated with
6945 new command to update these fields.
6946
6947 * systemd-timesyncd has been updated to automatically acquire
6948 NTP server information from systemd-networkd, which might
6949 have been discovered via DHCP.
6950
6951 * systemd-resolved now includes a caching DNS stub resolver
6952 and a complete LLMNR name resolution implementation. A new
daa05349
AB
6953 NSS module "nss-resolve" has been added which can be used
6954 instead of glibc's own "nss-dns" to resolve hostnames via
b2ca0d63
LP
6955 systemd-resolved. Hostnames, addresses and arbitrary RRs may
6956 be resolved via systemd-resolved D-Bus APIs. In contrast to
6957 the glibc internal resolver systemd-resolved is aware of
6958 multi-homed system, and keeps DNS server and caches separate
5f02e26c 6959 and per-interface. Queries are sent simultaneously on all
b2ca0d63
LP
6960 interfaces that have DNS servers configured, in order to
6961 properly handle VPNs and local LANs which might resolve
6962 separate sets of domain names. systemd-resolved may acquire
a1a4a25e 6963 DNS server information from systemd-networkd automatically,
b2ca0d63
LP
6964 which in turn might have discovered them via DHCP. A tool
6965 "systemd-resolve-host" has been added that may be used to
6966 query the DNS logic in resolved. systemd-resolved implements
6967 IDNA and automatically uses IDNA or UTF-8 encoding depending
6968 on whether classic DNS or LLMNR is used as transport. In the
6969 next releases we intend to add a DNSSEC and mDNS/DNS-SD
6970 implementation to systemd-resolved.
6971
6972 * A new NSS module nss-mymachines has been added, that
6973 automatically resolves the names of all local registered
6974 containers to their respective IP addresses.
6975
6976 * A new client tool "networkctl" for systemd-networkd has been
6977 added. It currently is entirely passive and will query
6978 networking configuration from udev, rtnetlink and networkd,
5f02e26c 6979 and present it to the user in a very friendly
b2ca0d63
LP
6980 way. Eventually, we hope to extend it to become a full
6981 control utility for networkd.
6982
6983 * .socket units gained a new DeferAcceptSec= setting that
6984 controls the kernels' TCP_DEFER_ACCEPT sockopt for
a8eaaee7 6985 TCP. Similarly, support for controlling TCP keep-alive
b2ca0d63
LP
6986 settings has been added (KeepAliveTimeSec=,
6987 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
6988 turning off Nagle's algorithm on TCP has been added
6989 (NoDelay=).
6990
a1a4a25e 6991 * logind learned a new session type "web", for use in projects
b2ca0d63
LP
6992 like Cockpit which register web clients as PAM sessions.
6993
6994 * timer units with at least one OnCalendar= setting will now
46ae28d8 6995 be started only after time-sync.target has been
b2ca0d63
LP
6996 reached. This way they will not elapse before the system
6997 clock has been corrected by a local NTP client or
6998 similar. This is particular useful on RTC-less embedded
6999 machines, that come up with an invalid system clock.
7000
7001 * systemd-nspawn's --network-veth= switch should now result in
7002 stable MAC addresses for both the outer and the inner side
7003 of the link.
7004
7005 * systemd-nspawn gained a new --volatile= switch for running
7006 container instances with /etc or /var unpopulated.
7007
7008 * The kdbus client code has been updated to use the new Linux
7009 3.17 memfd subsystem instead of the old kdbus-specific one.
7010
7011 * systemd-networkd's DHCP client and server now support
01da80b1
LP
7012 FORCERENEW. There are also new configuration options to
7013 configure the vendor client identifier and broadcast mode
7014 for DHCP.
b2ca0d63
LP
7015
7016 * systemd will no longer inform the kernel about the current
7017 timezone, as this is necessarily incorrect and racy as the
7018 kernel has no understanding of DST and similar
7019 concepts. This hence means FAT timestamps will be always
7020 considered UTC, similar to what Android is already
7021 doing. Also, when the RTC is configured to the local time
7022 (rather than UTC) systemd will never synchronize back to it,
7023 as this might confuse Windows at a later boot.
7024
7025 * systemd-analyze gained a new command "verify" for offline
7026 validation of unit files.
7027
7028 * systemd-networkd gained support for a couple of additional
7029 settings for bonding networking setups. Also, the metric for
7030 statically configured routes may now be configured. For
7031 network interfaces where this is appropriate the peer IP
7032 address may now be configured.
7033
26568403
TG
7034 * systemd-networkd's DHCP client will no longer request
7035 broadcasting by default, as this tripped up some networks.
7036 For hardware where broadcast is required the feature should
7037 be switched back on using RequestBroadcast=yes.
7038
7039 * systemd-networkd will now set up IPv4LL addresses (when
7040 enabled) even if DHCP is configured successfully.
7041
7042 * udev will now default to respect network device names given
7043 by the kernel when the kernel indicates that these are
7044 predictable. This behavior can be tweaked by changing
7045 NamePolicy= in the relevant .link file.
7046
b2ca0d63
LP
7047 * A new library systemd-terminal has been added that
7048 implements full TTY stream parsing and rendering. This
7049 library is supposed to be used later on for implementing a
7050 full userspace VT subsystem, replacing the current kernel
7051 implementation.
7052
7053 * A new tool systemd-journal-upload has been added to push
7054 journal data to a remote system running
7055 systemd-journal-remote.
7056
7057 * journald will no longer forward all local data to another
7058 running syslog daemon. This change has been made because
7059 rsyslog (which appears to be the most commonly used syslog
7060 implementation these days) no longer makes use of this, and
7061 instead pulls the data out of the journal on its own. Since
5f02e26c 7062 forwarding the messages to a non-existent syslog server is
b2ca0d63
LP
7063 more expensive than we assumed we have now turned this
7064 off. If you run a syslog server that is not a recent rsyslog
7065 version, you have to turn this option on again
7066 (ForwardToSyslog= in journald.conf).
7067
7068 * journald now optionally supports the LZ4 compressor for
7069 larger journal fields. This compressor should perform much
7070 better than XZ which was the previous default.
7071
7072 * machinectl now shows the IP addresses of local containers,
7073 if it knows them, plus the interface name of the container.
7074
7075 * A new tool "systemd-escape" has been added that makes it
7076 easy to escape strings to build unit names and similar.
7077
7078 * sd_notify() messages may now include a new ERRNO= field
7079 which is parsed and collected by systemd and shown among the
7080 "systemctl status" output for a service.
7081
7082 * A new component "systemd-firstboot" has been added that
7083 queries the most basic systemd information (timezone,
a1a4a25e 7084 hostname, root password) interactively on first
b2ca0d63
LP
7085 boot. Alternatively it may also be used to provision these
7086 things offline on OS images installed into directories.
7087
01da80b1
LP
7088 * The default sysctl.d/ snippets will now set
7089
7090 net.ipv4.conf.default.promote_secondaries=1
7091
7092 This has the benefit of no flushing secondary IP addresses
7093 when primary addresses are removed.
7094
b2ca0d63
LP
7095 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
7096 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
7097 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
7098 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
7099 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
7100 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
7101 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
7102 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
7103 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
7104 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
7105 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
7106 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
7107 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
7108 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
7109 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
7110
ccddd104 7111 — Berlin, 2014-08-19
b72ddf0f 7112
3dff3e00 7113CHANGES WITH 215:
24a2bf4c
LP
7114
7115 * A new tool systemd-sysusers has been added. This tool
7116 creates system users and groups in /etc/passwd and
7117 /etc/group, based on static declarative system user/group
7118 definitions in /usr/lib/sysusers.d/. This is useful to
7119 enable factory resets and volatile systems that boot up with
7120 an empty /etc directory, and thus need system users and
7121 groups created during early boot. systemd now also ships
7122 with two default sysusers.d/ files for the most basic
7123 users and groups systemd and the core operating system
7124 require.
7125
7126 * A new tmpfiles snippet has been added that rebuilds the
7127 essential files in /etc on boot, should they be missing.
7128
7129 * A directive for ensuring automatic clean-up of
7130 /var/cache/man/ has been removed from the default
7131 configuration. This line should now be shipped by the man
7132 implementation. The necessary change has been made to the
7133 man-db implementation. Note that you need to update your man
7134 implementation to one that ships this line, otherwise no
7135 automatic clean-up of /var/cache/man will take place.
7136
7137 * A new condition ConditionNeedsUpdate= has been added that
7138 may conditionalize services to only run when /etc or /var
7139 are "older" than the vendor operating system resources in
7140 /usr. This is useful for reconstructing or updating /etc
7141 after an offline update of /usr or a factory reset, on the
7142 next reboot. Services that want to run once after such an
7143 update or reset should use this condition and order
7144 themselves before the new systemd-update-done.service, which
7145 will mark the two directories as fully updated. A number of
7146 service files have been added making use of this, to rebuild
7147 the udev hardware database, the journald message catalog and
7148 dynamic loader cache (ldconfig). The systemd-sysusers tool
7149 described above also makes use of this now. With this in
7150 place it is now possible to start up a minimal operating
ce1dde29 7151 system with /etc empty cleanly. For more information on the
24a2bf4c
LP
7152 concepts involved see this recent blog story:
7153
7154 http://0pointer.de/blog/projects/stateless.html
7155
7156 * A new system group "input" has been introduced, and all
7157 input device nodes get this group assigned. This is useful
7158 for system-level software to get access to input devices. It
3dff3e00
KS
7159 complements what is already done for "audio" and "video".
7160
24a2bf4c
LP
7161 * systemd-networkd learnt minimal DHCPv4 server support in
7162 addition to the existing DHCPv4 client support. It also
7163 learnt DHCPv6 client and IPv6 Router Solicitation client
7164 support. The DHCPv4 client gained support for static routes
7165 passed in from the server. Note that the [DHCPv4] section
7166 known in older systemd-networkd versions has been renamed to
7167 [DHCP] and is now also used by the DHCPv6 client. Existing
c7435cc9
LP
7168 .network files using settings of this section should be
7169 updated, though compatibility is maintained. Optionally, the
7170 client hostname may now be sent to the DHCP server.
24a2bf4c 7171
c7435cc9
LP
7172 * networkd gained support for vxlan virtual networks as well
7173 as tun/tap and dummy devices.
24a2bf4c
LP
7174
7175 * networkd gained support for automatic allocation of address
7176 ranges for interfaces from a system-wide pool of
7177 addresses. This is useful for dynamically managing a large
7178 number of interfaces with a single network configuration
7179 file. In particular this is useful to easily assign
7180 appropriate IP addresses to the veth links of a large number
7181 of nspawn instances.
7182
7183 * RPM macros for processing sysusers, sysctl and binfmt
7184 drop-in snippets at package installation time have been
7185 added.
7186
7187 * The /etc/os-release file should now be placed in
7188 /usr/lib/os-release. The old location is automatically
7189 created as symlink. /usr/lib is the more appropriate
7190 location of this file, since it shall actually describe the
7191 vendor operating system shipped in /usr, and not the
7192 configuration stored in /etc.
7193
7194 * .mount units gained a new boolean SloppyOptions= setting
7195 that maps to mount(8)'s -s option which enables permissive
7196 parsing of unknown mount options.
7197
7198 * tmpfiles learnt a new "L+" directive which creates a symlink
7199 but (unlike "L") deletes a pre-existing file first, should
7200 it already exist and not already be the correct
a8eaaee7 7201 symlink. Similarly, "b+", "c+" and "p+" directives have been
24a2bf4c
LP
7202 added as well, which create block and character devices, as
7203 well as fifos in the filesystem, possibly removing any
7204 pre-existing files of different types.
7205
7206 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
7207 'argument' field (which so far specified the source to
ce1dde29 7208 symlink/copy the files from) is now optional. If omitted the
24a2bf4c
LP
7209 same file os copied from /usr/share/factory/ suffixed by the
7210 full destination path. This is useful for populating /etc
7211 with essential files, by copying them from vendor defaults
7212 shipped in /usr/share/factory/etc.
7213
7214 * A new command "systemctl preset-all" has been added that
7215 applies the service preset settings to all installed unit
7216 files. A new switch --preset-mode= has been added that
7217 controls whether only enable or only disable operations
7218 shall be executed.
7219
7220 * A new command "systemctl is-system-running" has been added
7221 that allows checking the overall state of the system, for
ce1dde29 7222 example whether it is fully up and running.
24a2bf4c
LP
7223
7224 * When the system boots up with an empty /etc, the equivalent
7225 to "systemctl preset-all" is executed during early boot, to
7226 make sure all default services are enabled after a factory
7227 reset.
7228
7229 * systemd now contains a minimal preset file that enables the
7230 most basic services systemd ships by default.
7231
7232 * Unit files' [Install] section gained a new DefaultInstance=
7233 field for defining the default instance to create if a
7234 template unit is enabled with no instance specified.
7235
7236 * A new passive target cryptsetup-pre.target has been added
7237 that may be used by services that need to make they run and
7238 finish before the first LUKS cryptographic device is set up.
7239
7240 * The /dev/loop-control and /dev/btrfs-control device nodes
7241 are now owned by the "disk" group by default, opening up
7242 access to this group.
7243
7244 * systemd-coredump will now automatically generate a
7245 stack trace of all core dumps taking place on the system,
7246 based on elfutils' libdw library. This stack trace is logged
7247 to the journal.
7248
7249 * systemd-coredump may now optionally store coredumps directly
7250 on disk (in /var/lib/systemd/coredump, possibly compressed),
7251 instead of storing them unconditionally in the journal. This
7252 mode is the new default. A new configuration file
7253 /etc/systemd/coredump.conf has been added to configure this
7254 and other parameters of systemd-coredump.
7255
7256 * coredumpctl gained a new "info" verb to show details about a
7257 specific coredump. A new switch "-1" has also been added
7258 that makes sure to only show information about the most
7259 recent entry instead of all entries. Also, as the tool is
7260 generally useful now the "systemd-" prefix of the binary
7261 name has been removed. Distributions that want to maintain
7262 compatibility with the old name should add a symlink from
7263 the old name to the new name.
7264
7265 * journald's SplitMode= now defaults to "uid". This makes sure
ce1dde29 7266 that unprivileged users can access their own coredumps with
24a2bf4c
LP
7267 coredumpctl without restrictions.
7268
7269 * New kernel command line options "systemd.wants=" (for
7270 pulling an additional unit during boot), "systemd.mask="
7271 (for masking a specific unit for the boot), and
7272 "systemd.debug-shell" (for enabling the debug shell on tty9)
7273 have been added. This is implemented in the new generator
7274 "systemd-debug-generator".
7275
7276 * systemd-nspawn will now by default filter a couple of
7277 syscalls for containers, among them those required for
7278 kernel module loading, direct x86 IO port access, swap
7279 management, and kexec. Most importantly though
7280 open_by_handle_at() is now prohibited for containers,
7281 closing a hole similar to a recently discussed vulnerability
7282 in docker regarding access to files on file hierarchies the
b938cb90
JE
7283 container should normally not have access to. Note that, for
7284 nspawn, we generally make no security claims anyway (and
24a2bf4c
LP
7285 this is explicitly documented in the man page), so this is
7286 just a fix for one of the most obvious problems.
7287
7288 * A new man page file-hierarchy(7) has been added that
7289 contains a minimized, modernized version of the file system
7290 layout systemd expects, similar in style to the FHS
c7435cc9
LP
7291 specification or hier(5). A new tool systemd-path(1) has
7292 been added to query many of these paths for the local
7293 machine and user.
24a2bf4c
LP
7294
7295 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
7296 longer done. Since the directory now has a per-user size
7297 limit, and is cleaned on logout this appears unnecessary,
7298 in particular since this now brings the lifecycle of this
7299 directory closer in line with how IPC objects are handled.
7300
7301 * systemd.pc now exports a number of additional directories,
7302 including $libdir (which is useful to identify the library
7303 path for the primary architecture of the system), and a
7304 couple of drop-in directories.
7305
3058e017
TLSC
7306 * udev's predictable network interface names now use the dev_port
7307 sysfs attribute, introduced in linux 3.15 instead of dev_id to
7308 distinguish between ports of the same PCI function. dev_id should
7309 only be used for ports using the same HW address, hence the need
7310 for dev_port.
7311
c7435cc9
LP
7312 * machined has been updated to export the OS version of a
7313 container (read from /etc/os-release and
7314 /usr/lib/os-release) on the bus. This is now shown in
7315 "machinectl status" for a machine.
7316
7317 * A new service setting RestartForceExitStatus= has been
7318 added. If configured to a set of exit signals or process
7319 return values, the service will be restarted when the main
7320 daemon process exits with any of them, regardless of the
7321 Restart= setting.
7322
7323 * systemctl's -H switch for connecting to remote systemd
7324 machines has been extended so that it may be used to
7325 directly connect to a specific container on the
7326 host. "systemctl -H root@foobar:waldi" will now connect as
7327 user "root" to host "foobar", and then proceed directly to
7328 the container named "waldi". Note that currently you have to
7329 authenticate as user "root" for this to work, as entering
7330 containers is a privileged operation.
7331
7332 Contributions from: Andreas Henriksson, Benjamin Steinwender,
7333 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
7334 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
7335 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
7336 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
7337 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
7338 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
7339 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
7340 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
7341 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
7342 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
7343 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
7344
ccddd104 7345 — Berlin, 2014-07-03
c7435cc9 7346
4196a3ea
KS
7347CHANGES WITH 214:
7348
7349 * As an experimental feature, udev now tries to lock the
7350 disk device node (flock(LOCK_SH|LOCK_NB)) while it
7351 executes events for the disk or any of its partitions.
7352 Applications like partitioning programs can lock the
7353 disk device node (flock(LOCK_EX)) and claim temporary
7354 device ownership that way; udev will entirely skip all event
7355 handling for this disk and its partitions. If the disk
7356 was opened for writing, the close will trigger a partition
7357 table rescan in udev's "watch" facility, and if needed
71449caf 7358 synthesize "change" events for the disk and all its partitions.
8d0e0ddd 7359 This is now unconditionally enabled, and if it turns out to
4196a3ea 7360 cause major problems, we might turn it on only for specific
45df8656 7361 devices, or might need to disable it entirely. Device Mapper
4196a3ea
KS
7362 devices are excluded from this logic.
7363
04e91da2
LP
7364 * We temporarily dropped the "-l" switch for fsck invocations,
7365 since they collide with the flock() logic above. util-linux
7366 upstream has been changed already to avoid this conflict,
5238e957 7367 and we will re-add "-l" as soon as util-linux with this
04e91da2
LP
7368 change has been released.
7369
7370 * The dependency on libattr has been removed. Since a long
8d0e0ddd 7371 time, the extended attribute calls have moved to glibc, and
04e91da2
LP
7372 libattr is thus unnecessary.
7373
ce830873 7374 * Virtualization detection works without privileges now. This
04e91da2
LP
7375 means the systemd-detect-virt binary no longer requires
7376 CAP_SYS_PTRACE file capabilities, and our daemons can run
71449caf 7377 with fewer privileges.
04e91da2
LP
7378
7379 * systemd-networkd now runs under its own "systemd-network"
7380 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
7381 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
7382 loses the ability to write to files owned by root this way.
7383
a8eaaee7 7384 * Similarly, systemd-resolved now runs under its own
04e91da2
LP
7385 "systemd-resolve" user with no capabilities remaining.
7386
a8eaaee7 7387 * Similarly, systemd-bus-proxyd now runs under its own
04e91da2
LP
7388 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
7389
7390 * systemd-networkd gained support for setting up "veth"
a8eaaee7 7391 virtual Ethernet devices for container connectivity, as well
04e91da2
LP
7392 as GRE and VTI tunnels.
7393
7394 * systemd-networkd will no longer automatically attempt to
7395 manually load kernel modules necessary for certain tunnel
8d0e0ddd 7396 transports. Instead, it is assumed the kernel loads them
04e91da2
LP
7397 automatically when required. This only works correctly on
7398 very new kernels. On older kernels, please consider adding
c54bed5d 7399 the kernel modules to /etc/modules-load.d/ as a work-around.
04e91da2 7400
cd14eda3 7401 * The resolv.conf file systemd-resolved generates has been
8d0e0ddd
JE
7402 moved to /run/systemd/resolve/. If you have a symlink from
7403 /etc/resolv.conf, it might be necessary to correct it.
cd14eda3 7404
ef392da6 7405 * Two new service settings, ProtectHome= and ProtectSystem=,
8d0e0ddd 7406 have been added. When enabled, they will make the user data
04e91da2
LP
7407 (such as /home) inaccessible or read-only and the system
7408 (such as /usr) read-only, for specific services. This allows
7409 very light-weight per-service sandboxing to avoid
7410 modifications of user data or system files from
7411 services. These two new switches have been enabled for all
7412 of systemd's long-running services, where appropriate.
7413
7414 * Socket units gained new SocketUser= and SocketGroup=
7415 settings to set the owner user and group of AF_UNIX sockets
7416 and FIFOs in the file system.
7417
8d0e0ddd 7418 * Socket units gained a new RemoveOnStop= setting. If enabled,
04e91da2
LP
7419 all FIFOS and sockets in the file system will be removed
7420 when the specific socket unit is stopped.
7421
7422 * Socket units gained a new Symlinks= setting. It takes a list
7423 of symlinks to create to file system sockets or FIFOs
45df8656 7424 created by the specific Unix sockets. This is useful to
de04bbdc 7425 manage symlinks to socket nodes with the same lifecycle as
04e91da2
LP
7426 the socket itself.
7427
7428 * The /dev/log socket and /dev/initctl FIFO have been moved to
7429 /run, and have been replaced by symlinks. This allows
7430 connecting to these facilities even if PrivateDevices=yes is
7431 used for a service (which makes /dev/log itself unavailable,
7432 but /run is left). This also has the benefit of ensuring
7433 that /dev only contains device nodes, directories and
7434 symlinks, and nothing else.
7435
7436 * sd-daemon gained two new calls sd_pid_notify() and
7437 sd_pid_notifyf(). They are similar to sd_notify() and
7438 sd_notifyf(), but allow overriding of the source PID of
7439 notification messages if permissions permit this. This is
7440 useful to send notify messages on behalf of a different
7441 process (for example, the parent process). The
7442 systemd-notify tool has been updated to make use of this
7443 when sending messages (so that notification messages now
7444 originate from the shell script invoking systemd-notify and
7445 not the systemd-notify process itself. This should minimize
7446 a race where systemd fails to associate notification
7447 messages to services when the originating process already
7448 vanished.
7449
7450 * A new "on-abnormal" setting for Restart= has been added. If
8d0e0ddd 7451 set, it will result in automatic restarts on all "abnormal"
04e91da2
LP
7452 reasons for a process to exit, which includes unclean
7453 signals, core dumps, timeouts and watchdog timeouts, but
7454 does not include clean and unclean exit codes or clean
7455 signals. Restart=on-abnormal is an alternative for
7456 Restart=on-failure for services that shall be able to
7457 terminate and avoid restarts on certain errors, by
7458 indicating so with an unclean exit code. Restart=on-failure
7459 or Restart=on-abnormal is now the recommended setting for
7460 all long-running services.
7461
7462 * If the InaccessibleDirectories= service setting points to a
7463 mount point (or if there are any submounts contained within
7464 it), it is now attempted to completely unmount it, to make
7465 the file systems truly unavailable for the respective
7466 service.
7467
7468 * The ReadOnlyDirectories= service setting and
7469 systemd-nspawn's --read-only parameter are now recursively
7470 applied to all submounts, too.
7471
7472 * Mount units may now be created transiently via the bus APIs.
7473
7474 * The support for SysV and LSB init scripts has been removed
7475 from the systemd daemon itself. Instead, it is now
7476 implemented as a generator that creates native systemd units
7477 from these scripts when needed. This enables us to remove a
7478 substantial amount of legacy code from PID 1, following the
7479 fact that many distributions only ship a very small number
7480 of LSB/SysV init scripts nowadays.
7481
cc98b302 7482 * Privileged Xen (dom0) domains are not considered
04e91da2
LP
7483 virtualization anymore by the virtualization detection
7484 logic. After all, they generally have unrestricted access to
71449caf 7485 the hardware and usually are used to manage the unprivileged
04e91da2
LP
7486 (domU) domains.
7487
7488 * systemd-tmpfiles gained a new "C" line type, for copying
7489 files or entire directories.
7490
7491 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
8d0e0ddd
JE
7492 lines. So far, they have been non-globbing versions of the
7493 latter, and have thus been redundant. In future, it is
7494 recommended to only use "z". "m" has hence been removed
04e91da2
LP
7495 from the documentation, even though it stays supported.
7496
7497 * A tmpfiles snippet to recreate the most basic structure in
7498 /var has been added. This is enough to create the /var/run →
7499 /run symlink and create a couple of structural
7500 directories. This allows systems to boot up with an empty or
8d0e0ddd
JE
7501 volatile /var. Of course, while with this change, the core OS
7502 now is capable with dealing with a volatile /var, not all
04e91da2 7503 user services are ready for it. However, we hope that sooner
8d0e0ddd 7504 or later, many service daemons will be changed upstream so
04e91da2
LP
7505 that they are able to automatically create their necessary
7506 directories in /var at boot, should they be missing. This is
7507 the first step to allow state-less systems that only require
7508 the vendor image for /usr to boot.
7509
7510 * systemd-nspawn has gained a new --tmpfs= switch to mount an
7511 empty tmpfs instance to a specific directory. This is
7512 particularly useful for making use of the automatic
7513 reconstruction of /var (see above), by passing --tmpfs=/var.
7514
7515 * Access modes specified in tmpfiles snippets may now be
7516 prefixed with "~", which indicates that they shall be masked
daa05349 7517 by whether the existing file or directory is currently
8d0e0ddd 7518 writable, readable or executable at all. Also, if specified,
04e91da2
LP
7519 the sgid/suid/sticky bits will be masked for all
7520 non-directories.
7521
7522 * A new passive target unit "network-pre.target" has been
7523 added which is useful for services that shall run before any
7524 network is configured, for example firewall scripts.
7525
4c0d13bd
LP
7526 * The "floppy" group that previously owned the /dev/fd*
7527 devices is no longer used. The "disk" group is now used
7528 instead. Distributions should probably deprecate usage of
7529 this group.
7530
dc1d6c02
LP
7531 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
7532 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
7533 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
7534 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
7535 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
7536 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
7537 Jędrzejewski-Szmek
7538
ccddd104 7539 — Berlin, 2014-06-11
dc1d6c02 7540
6936cd89
LP
7541CHANGES WITH 213:
7542
7543 * A new "systemd-timesyncd" daemon has been added for
69beda1f 7544 synchronizing the system clock across the network. It
6936cd89 7545 implements an SNTP client. In contrast to NTP
8d0e0ddd 7546 implementations such as chrony or the NTP reference server,
6936cd89 7547 this only implements a client side, and does not bother with
c9679c65
LP
7548 the full NTP complexity, focusing only on querying time from
7549 one remote server and synchronizing the local clock to
6936cd89 7550 it. Unless you intend to serve NTP to networked clients or
8d0e0ddd 7551 want to connect to local hardware clocks, this simple NTP
6936cd89
LP
7552 client should be more than appropriate for most
7553 installations. The daemon runs with minimal privileges, and
7554 has been hooked up with networkd to only operate when
7555 network connectivity is available. The daemon saves the
7556 current clock to disk every time a new NTP sync has been
7557 acquired, and uses this to possibly correct the system clock
69beda1f 7558 early at bootup, in order to accommodate for systems that
6936cd89 7559 lack an RTC such as the Raspberry Pi and embedded devices,
8d0e0ddd 7560 and to make sure that time monotonically progresses on these
c9679c65 7561 systems, even if it is not always correct. To make use of
8d0e0ddd 7562 this daemon, a new system user and group "systemd-timesync"
c9679c65 7563 needs to be created on installation of systemd.
6936cd89 7564
69beda1f
KS
7565 * The queue "seqnum" interface of libudev has been disabled, as
7566 it was generally incompatible with device namespacing as
6936cd89
LP
7567 sequence numbers of devices go "missing" if the devices are
7568 part of a different namespace.
7569
7570 * "systemctl list-timers" and "systemctl list-sockets" gained
7571 a --recursive switch for showing units of these types also
499b604b
ZJS
7572 for all local containers, similar in style to the already
7573 supported --recursive switch for "systemctl list-units".
6936cd89
LP
7574
7575 * A new RebootArgument= setting has been added for service
7576 units, which may be used to specify a kernel reboot argument
499b604b 7577 to use when triggering reboots with StartLimitAction=.
6936cd89
LP
7578
7579 * A new FailureAction= setting has been added for service
7580 units which may be used to specify an operation to trigger
499b604b 7581 when a service fails. This works similarly to
8d0e0ddd 7582 StartLimitAction=, but unlike it, controls what is done
6936cd89
LP
7583 immediately rather than only after several attempts to
7584 restart the service in question.
7585
7586 * hostnamed got updated to also expose the kernel name,
499b604b
ZJS
7587 release, and version on the bus. This is useful for
7588 executing commands like hostnamectl with the -H switch.
7589 systemd-analyze makes use of this to properly display
7590 details when running non-locally.
6936cd89
LP
7591
7592 * The bootchart tool can now show cgroup information in the
7593 graphs it generates.
7594
7595 * The CFS CPU quota cgroup attribute is now exposed for
7596 services. The new CPUQuota= switch has been added for this
7597 which takes a percentage value. Setting this will have the
7598 result that a service may never get more CPU time than the
7599 specified percentage, even if the machine is otherwise idle.
7600
7601 * systemd-networkd learned IPIP and SIT tunnel support.
7602
7603 * LSB init scripts exposing a dependency on $network will now
7604 get a dependency on network-online.target rather than simply
7605 network.target. This should bring LSB handling closer to
7606 what it was on SysV systems.
7607
7608 * A new fsck.repair= kernel option has been added to control
7609 how fsck shall deal with unclean file systems at boot.
7610
2ad98889
ZJS
7611 * The (.ini) configuration file parser will now silently ignore
7612 sections whose names begin with "X-". This may be used to maintain
7613 application-specific extension sections in unit files.
6936cd89
LP
7614
7615 * machined gained a new API to query the IP addresses of
7616 registered containers. "machinectl status" has been updated
7617 to show these addresses in its output.
7618
7619 * A new call sd_uid_get_display() has been added to the
7620 sd-login APIs for querying the "primary" session of a
7621 user. The "primary" session of the user is elected from the
7622 user's sessions and generally a graphical session is
7623 preferred over a text one.
7624
7625 * A minimal systemd-resolved daemon has been added. It
7626 currently simply acts as a companion to systemd-networkd and
7627 manages resolv.conf based on per-interface DNS
7628 configuration, possibly supplied via DHCP. In the long run
7629 we hope to extend this into a local DNSSEC enabled DNS and
7630 mDNS cache.
7631
68dd0956
TG
7632 * The systemd-networkd-wait-online tool is now enabled by
7633 default. It will delay network-online.target until a network
7634 connection has been configured. The tool primarily integrates
7635 with networkd, but will also make a best effort to make sense
7636 of network configuration performed in some other way.
7637
6936cd89 7638 * Two new service options StartupCPUShares= and
499b604b 7639 StartupBlockIOWeight= have been added that work similarly to
6936cd89 7640 CPUShares= and BlockIOWeight= however only apply during
69beda1f 7641 system startup. This is useful to prioritize certain services
6936cd89
LP
7642 differently during bootup than during normal runtime.
7643
8e7acf67
LP
7644 * hostnamed has been changed to prefer the statically
7645 configured hostname in /etc/hostname (unless set to
7646 'localhost' or empty) over any dynamic one supplied by
8d0e0ddd 7647 dhcp. With this change, the rules for picking the hostname
8e7acf67
LP
7648 match more closely the rules of other configuration settings
7649 where the local administrator's configuration in /etc always
7650 overrides any other settings.
7651
5238e957 7652 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
6936cd89
LP
7653 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
7654 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
7655 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
7656 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
7657 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
7658 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
7659 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
7660 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
8e7acf67
LP
7661 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
7662 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
7663 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
7664 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
7665 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
7666 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
7667 Lindskog, WaLyong Cho, Will Woods, Zbigniew
6936cd89
LP
7668 Jędrzejewski-Szmek
7669
ccddd104 7670 — Beijing, 2014-05-28
6936cd89 7671
51c61cda
LP
7672CHANGES WITH 212:
7673
7674 * When restoring the screen brightness at boot, stay away from
7675 the darkest setting or from the lowest 5% of the available
7676 range, depending on which is the larger value of both. This
7677 should effectively protect the user from rebooting into a
7678 black screen, should the brightness have been set to minimum
7679 by accident.
7680
7681 * sd-login gained a new sd_machine_get_class() call to
7682 determine the class ("vm" or "container") of a machine
7683 registered with machined.
7684
7685 * sd-login gained new calls
7686 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
7687 to query the identity of the peer of a local AF_UNIX
499b604b 7688 connection. They operate similarly to their sd_pid_get_xyz()
51c61cda
LP
7689 counterparts.
7690
7691 * PID 1 will now maintain a system-wide system state engine
7692 with the states "starting", "running", "degraded",
7693 "maintenance", "stopping". These states are bound to system
7694 startup, normal runtime, runtime with at least one failed
7695 service, rescue/emergency mode and system shutdown. This
7696 state is shown in the "systemctl status" output when no unit
7697 name is passed. It is useful to determine system state, in
7698 particularly when doing so for many systems or containers at
7699 once.
7700
7701 * A new command "list-machines" has been added to "systemctl"
7702 that lists all local OS containers and shows their system
7703 state (see above), if systemd runs inside of them.
7704
7705 * systemctl gained a new "-r" switch to recursively enumerate
7706 units on all local containers, when used with the
7707 "list-unit" command (which is the default one that is
7708 executed when no parameters are specified).
7709
7710 * The GPT automatic partition discovery logic will now honour
7711 two GPT partition flags: one may be set on a partition to
7712 cause it to be mounted read-only, and the other may be set
7713 on a partition to ignore it during automatic discovery.
7714
7715 * Two new GPT type UUIDs have been added for automatic root
70a44afe 7716 partition discovery, for 32-bit and 64-bit ARM. This is not
51c61cda
LP
7717 particularly useful for discovering the root directory on
7718 these architectures during bare-metal boots (since UEFI is
7719 not common there), but still very useful to allow booting of
7720 ARM disk images in nspawn with the -i option.
7721
7722 * MAC addresses of interfaces created with nspawn's
7723 --network-interface= switch will now be generated from the
7724 machine name, and thus be stable between multiple invocations
7725 of the container.
7726
7727 * logind will now automatically remove all IPC objects owned
7728 by a user if she or he fully logs out. This makes sure that
7729 users who are logged out cannot continue to consume IPC
7730 resources. This covers SysV memory, semaphores and message
7731 queues as well as POSIX shared memory and message
de04bbdc 7732 queues. Traditionally, SysV and POSIX IPC had no lifecycle
b8bde116
JE
7733 limits. With this functionality, that is corrected. This may
7734 be turned off by using the RemoveIPC= switch of logind.conf.
51c61cda
LP
7735
7736 * The systemd-machine-id-setup and tmpfiles tools gained a
7737 --root= switch to operate on a specific root directory,
7738 instead of /.
7739
7740 * journald can now forward logged messages to the TTYs of all
7741 logged in users ("wall"). This is the default for all
7742 emergency messages now.
7743
7744 * A new tool systemd-journal-remote has been added to stream
7745 journal log messages across the network.
7746
7747 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
7748 controller trees are mounted into it. Note that the
7749 directories mounted beneath it are not read-only. This is a
7750 security measure and is particularly useful because glibc
7751 actually includes a search logic to pick any tmpfs it can
7752 find to implement shm_open() if /dev/shm is not available
7753 (which it might very well be in namespaced setups).
7754
7755 * machinectl gained a new "poweroff" command to cleanly power
7756 down a local OS container.
7757
7758 * The PrivateDevices= unit file setting will now also drop the
7759 CAP_MKNOD capability from the capability bound set, and
7760 imply DevicePolicy=closed.
7761
7762 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
7763 comprehensively on all long-running systemd services where
7764 this is appropriate.
7765
7766 * systemd-udevd will now run in a disassociated mount
b8bde116 7767 namespace. To mount directories from udev rules, make sure to
51c61cda
LP
7768 pull in mount units via SYSTEMD_WANTS properties.
7769
7770 * The kdbus support gained support for uploading policy into
7771 the kernel. sd-bus gained support for creating "monitoring"
7772 connections that can eavesdrop into all bus communication
7773 for debugging purposes.
7774
7775 * Timestamps may now be specified in seconds since the UNIX
7776 epoch Jan 1st, 1970 by specifying "@" followed by the value
7777 in seconds.
7778
7779 * Native tcpwrap support in systemd has been removed. tcpwrap
7780 is old code, not really maintained anymore and has serious
7781 shortcomings, and better options such as firewalls
7782 exist. For setups that require tcpwrap usage, please
7783 consider invoking your socket-activated service via tcpd,
7784 like on traditional inetd.
7785
7786 * A new system.conf configuration option
7787 DefaultTimerAccuracySec= has been added that controls the
7788 default AccuracySec= setting of .timer units.
7789
b8bde116 7790 * Timer units gained a new WakeSystem= switch. If enabled,
51c61cda
LP
7791 timers configured this way will cause the system to resume
7792 from system suspend (if the system supports that, which most
7793 do these days).
7794
b8bde116 7795 * Timer units gained a new Persistent= switch. If enabled,
51c61cda
LP
7796 timers configured this way will save to disk when they have
7797 been last triggered. This information is then used on next
7798 reboot to possible execute overdue timer events, that
d28315e4
JE
7799 could not take place because the system was powered off.
7800 This enables simple anacron-like behaviour for timer units.
51c61cda
LP
7801
7802 * systemctl's "list-timers" will now also list the time a
7803 timer unit was last triggered in addition to the next time
7804 it will be triggered.
7805
7806 * systemd-networkd will now assign predictable IPv4LL
7807 addresses to its local interfaces.
7808
7809 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
7810 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
7811 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
7812 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
7813 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
7814 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
7815 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
7816 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
7817 Jędrzejewski-Szmek
7818
ccddd104 7819 — Berlin, 2014-03-25
51c61cda 7820
699b6b34
LP
7821CHANGES WITH 211:
7822
7823 * A new unit file setting RestrictAddressFamilies= has been
7824 added to restrict which socket address families unit
7825 processes gain access to. This takes address family names
7826 like "AF_INET" or "AF_UNIX", and is useful to minimize the
7827 attack surface of services via exotic protocol stacks. This
7828 is built on seccomp system call filters.
7829
7830 * Two new unit file settings RuntimeDirectory= and
7831 RuntimeDirectoryMode= have been added that may be used to
7832 manage a per-daemon runtime directories below /run. This is
7833 an alternative for setting up directory permissions with
7834 tmpfiles snippets, and has the advantage that the runtime
7835 directory's lifetime is bound to the daemon runtime and that
7836 the daemon starts up with an empty directory each time. This
7837 is particularly useful when writing services that drop
f1721625 7838 privileges using the User= or Group= setting.
699b6b34
LP
7839
7840 * The DeviceAllow= unit setting now supports globbing for
7841 matching against device group names.
7842
7843 * The systemd configuration file system.conf gained new
7844 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
7845 DefaultMemoryAccounting= to globally turn on/off accounting
7846 for specific resources (cgroups) for all units. These
22e7062d 7847 settings may still be overridden individually in each unit
699b6b34
LP
7848 though.
7849
7850 * systemd-gpt-auto-generator is now able to discover /srv and
7851 root partitions in addition to /home and swap partitions. It
7852 also supports LUKS-encrypted partitions now. With this in
b8bde116 7853 place, automatic discovery of partitions to mount following
699b6b34 7854 the Discoverable Partitions Specification
56cadcb6 7855 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
699b6b34
LP
7856 is now a lot more complete. This allows booting without
7857 /etc/fstab and without root= on the kernel command line on
b8bde116 7858 systems prepared appropriately.
699b6b34
LP
7859
7860 * systemd-nspawn gained a new --image= switch which allows
7861 booting up disk images and Linux installations on any block
7862 device that follow the Discoverable Partitions Specification
7863 (see above). This means that installations made with
7864 appropriately updated installers may now be started and
7865 deployed using container managers, completely
7866 unmodified. (We hope that libvirt-lxc will add support for
7867 this feature soon, too.)
7868
7869 * systemd-nspawn gained a new --network-macvlan= setting to
7870 set up a private macvlan interface for the
499b604b 7871 container. Similarly, systemd-networkd gained a new
699b6b34
LP
7872 Kind=macvlan setting in .netdev files.
7873
7874 * systemd-networkd now supports configuring local addresses
7875 using IPv4LL.
7876
7877 * A new tool systemd-network-wait-online has been added to
7878 synchronously wait for network connectivity using
7879 systemd-networkd.
7880
7881 * The sd-bus.h bus API gained a new sd_bus_track object for
de04bbdc 7882 tracking the lifecycle of bus peers. Note that sd-bus.h is
699b6b34
LP
7883 still not a public API though (unless you specify
7884 --enable-kdbus on the configure command line, which however
7885 voids your warranty and you get no API stability guarantee).
7886
7887 * The $XDG_RUNTIME_DIR runtime directories for each user are
7888 now individual tmpfs instances, which has the benefit of
7889 introducing separate pools for each user, with individual
4ef6e535 7890 size limits, and thus making sure that unprivileged clients
699b6b34
LP
7891 can no longer negatively impact the system or other users by
7892 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
7893 RuntimeDirectorySize= has been introduced that allows
7894 controlling the default size limit for all users. It
7895 defaults to 10% of the available physical memory. This is no
7896 replacement for quotas on tmpfs though (which the kernel
7897 still does not support), as /dev/shm and /tmp are still
4ef6e535 7898 shared resources used by both the system and unprivileged
699b6b34
LP
7899 users.
7900
7901 * logind will now automatically turn off automatic suspending
7902 on laptop lid close when more than one display is
7903 connected. This was previously expected to be implemented
7904 individually in desktop environments (such as GNOME),
7905 however has been added to logind now, in order to fix a
7906 boot-time race where a desktop environment might not have
7907 been started yet and thus not been able to take an inhibitor
7908 lock at the time where logind already suspends the system
7909 due to a closed lid.
7910
7911 * logind will now wait at least 30s after each system
7912 suspend/resume cycle, and 3min after system boot before
7913 suspending the system due to a closed laptop lid. This
7914 should give USB docking stations and similar enough time to
4ef6e535 7915 be probed and configured after system resume and boot in
699b6b34
LP
7916 order to then act as suspend blocker.
7917
7918 * systemd-run gained a new --property= setting which allows
7919 initialization of resource control properties (and others)
7920 for the created scope or service unit. Example: "systemd-run
7921 --property=BlockIOWeight=10 updatedb" may be used to run
7922 updatedb at a low block IO scheduling weight.
7923
7924 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
7925 now also work in --scope mode.
7926
7927 * When systemd is compiled with kdbus support, basic support
7928 for enforced policies is now in place. (Note that enabling
7929 kdbus still voids your warranty and no API compatibility
7930 promises are made.)
7931
7932 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
7933 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
7934 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
7935 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
7936 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
7937 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
7938 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
7939 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
7940 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
7941 Zbigniew Jędrzejewski-Szmek
7942
ccddd104 7943 — Berlin, 2014-03-12
699b6b34 7944
43c71255
LP
7945CHANGES WITH 210:
7946
7947 * systemd will now relabel /dev after loading the SMACK policy
7948 according to SMACK rules.
7949
67dd87c5 7950 * A new unit file option AppArmorProfile= has been added to
43c71255
LP
7951 set the AppArmor profile for the processes of a unit.
7952
7953 * A new condition check ConditionArchitecture= has been added
7954 to conditionalize units based on the system architecture, as
7955 reported by uname()'s "machine" field.
7956
7957 * systemd-networkd now supports matching on the system
38b38500 7958 virtualization, architecture, kernel command line, hostname
43c71255
LP
7959 and machine ID.
7960
ed28905e 7961 * logind is now a lot more aggressive when suspending the
43c71255 7962 machine due to a closed laptop lid. Instead of acting only
b8bde116 7963 on the lid close action, it will continuously watch the lid
43c71255
LP
7964 status and act on it. This is useful for laptops where the
7965 power button is on the outside of the chassis so that it can
ed28905e 7966 be reached without opening the lid (such as the Lenovo
b8bde116 7967 Yoga). On those machines, logind will now immediately
ed28905e 7968 re-suspend the machine if the power button has been
43c71255
LP
7969 accidentally pressed while the laptop was suspended and in a
7970 backpack or similar.
7971
7972 * logind will now watch SW_DOCK switches and inhibit reaction
7973 to the lid switch if it is pressed. This means that logind
d27893ef 7974 will not suspend the machine anymore if the lid is closed
949138cc 7975 and the system is docked, if the laptop supports SW_DOCK
43c71255
LP
7976 notifications via the input layer. Note that ACPI docking
7977 stations do not generate this currently. Also note that this
7978 logic is usually not fully sufficient and Desktop
7979 Environments should take a lid switch inhibitor lock when an
7980 external display is connected, as systemd will not watch
7981 this on its own.
7982
7983 * nspawn will now make use of the devices cgroup controller by
7984 default, and only permit creation of and access to the usual
7985 API device nodes like /dev/null or /dev/random, as well as
7986 access to (but not creation of) the pty devices.
7987
7988 * We will now ship a default .network file for
7989 systemd-networkd that automatically configures DHCP for
7990 network interfaces created by nspawn's --network-veth or
7991 --network-bridge= switches.
7992
7993 * systemd will now understand the usual M, K, G, T suffixes
7994 according to SI conventions (i.e. to the base 1000) when
7995 referring to throughput and hardware metrics. It will stay
7996 with IEC conventions (i.e. to the base 1024) for software
7997 metrics, according to what is customary according to
7998 Wikipedia. We explicitly document which base applies for
7999 each configuration option.
8000
8001 * The DeviceAllow= setting in unit files now supports a syntax
ed28905e 8002 to whitelist an entire group of devices node majors at once,
43c71255 8003 based on the /proc/devices listing. For example, with the
b8bde116 8004 string "char-pts", it is now possible to whitelist all
43c71255
LP
8005 current and future pseudo-TTYs at once.
8006
8007 * sd-event learned a new "post" event source. Event sources of
8008 this type are triggered by the dispatching of any event
8009 source of a type that is not "post". This is useful for
8010 implementing clean-up and check event sources that are
8011 triggered by other work being done in the program.
8012
8013 * systemd-networkd is no longer statically enabled, but uses
8014 the usual [Install] sections so that it can be
8015 enabled/disabled using systemctl. It still is enabled by
8016 default however.
8017
b8bde116 8018 * When creating a veth interface pair with systemd-nspawn, the
43c71255
LP
8019 host side will now be prefixed with "vb-" if
8020 --network-bridge= is used, and with "ve-" if --network-veth
b8bde116 8021 is used. This way, it is easy to distinguish these cases on
43c71255
LP
8022 the host, for example to apply different configuration to
8023 them with systemd-networkd.
8024
d27893ef
LP
8025 * The compatibility libraries for libsystemd-journal.so,
8026 libsystem-id128.so, libsystemd-login.so and
8027 libsystemd-daemon.so do not make use of IFUNC
b8bde116 8028 anymore. Instead, we now build libsystemd.so multiple times
d27893ef
LP
8029 under these alternative names. This means that the footprint
8030 is drastically increased, but given that these are
b8bde116 8031 transitional compatibility libraries, this should not matter
d27893ef
LP
8032 much. This change has been made necessary to support the ARM
8033 platform for these compatibility libraries, as the ARM
d28315e4 8034 toolchain is not really at the same level as the toolchain
ed28905e 8035 for other architectures like x86 and does not support
d27893ef
LP
8036 IFUNC. Please make sure to use --enable-compat-libs only
8037 during a transitional period!
8038
f26ad321
ZJS
8039 * The .include syntax has been deprecated and is not documented
8040 anymore. Drop-in files in .d directories should be used instead.
8041
13b28d82 8042 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
43c71255
LP
8043 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
8044 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
8045 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
8046 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
8047 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
8048 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
8049 Zbigniew Jędrzejewski-Szmek
8050
ccddd104 8051 — Berlin, 2014-02-24
43c71255 8052
e49b5aad
LP
8053CHANGES WITH 209:
8054
8055 * A new component "systemd-networkd" has been added that can
8056 be used to configure local network interfaces statically or
8b7d0494
JSJ
8057 via DHCP. It is capable of bringing up bridges, VLANs, and
8058 bonding. Currently, no hook-ups for interactive network
4670e9d5 8059 configuration are provided. Use this for your initrd,
8b7d0494
JSJ
8060 container, embedded, or server setup if you need a simple,
8061 yet powerful, network configuration solution. This
4670e9d5 8062 configuration subsystem is quite nifty, as it allows wildcard
1e190502 8063 hotplug matching in interfaces. For example, with a single
4670e9d5 8064 configuration snippet, you can configure that all Ethernet
1e190502
ZJS
8065 interfaces showing up are automatically added to a bridge,
8066 or similar. It supports link-sensing and more.
e49b5aad
LP
8067
8068 * A new tool "systemd-socket-proxyd" has been added which can
4c2413bf 8069 act as a bidirectional proxy for TCP sockets. This is
e49b5aad
LP
8070 useful for adding socket activation support to services that
8071 do not actually support socket activation, including virtual
4c2413bf 8072 machines and the like.
e49b5aad
LP
8073
8074 * Add a new tool to save/restore rfkill state on
8075 shutdown/boot.
8076
8b7d0494
JSJ
8077 * Save/restore state of keyboard backlights in addition to
8078 display backlights on shutdown/boot.
e49b5aad
LP
8079
8080 * udev learned a new SECLABEL{} construct to label device
8081 nodes with a specific security label when they appear. For
4c2413bf 8082 now, only SECLABEL{selinux} is supported, but the syntax is
e49b5aad
LP
8083 prepared for additional security frameworks.
8084
8085 * udev gained a new scheme to configure link-level attributes
8086 from files in /etc/systemd/network/*.link. These files can
8b7d0494 8087 match against MAC address, device path, driver name and type,
4c2413bf 8088 and will apply attributes like the naming policy, link speed,
8b7d0494 8089 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
e49b5aad
LP
8090 address assignment policy (randomized, ...).
8091
dfb08b05
ZJS
8092 * The configuration of network interface naming rules for
8093 "permanent interface names" has changed: a new NamePolicy=
8094 setting in the [Link] section of .link files determines the
a8eaaee7 8095 priority of possible naming schemes (onboard, slot, MAC,
dfb08b05
ZJS
8096 path). The default value of this setting is determined by
8097 /usr/lib/net/links/99-default.link. Old
8098 80-net-name-slot.rules udev configuration file has been
8099 removed, so local configuration overriding this file should
ce830873 8100 be adapted to override 99-default.link instead.
dfb08b05 8101
e49b5aad 8102 * When the User= switch is used in a unit file, also
4c2413bf 8103 initialize $SHELL= based on the user database entry.
e49b5aad
LP
8104
8105 * systemd no longer depends on libdbus. All communication is
8106 now done with sd-bus, systemd's low-level bus library
8107 implementation.
8108
8109 * kdbus support has been added to PID 1 itself. When kdbus is
4c2413bf 8110 enabled, this causes PID 1 to set up the system bus and
e49b5aad
LP
8111 enable support for a new ".busname" unit type that
8112 encapsulates bus name activation on kdbus. It works a little
8113 bit like ".socket" units, except for bus names. A new
8114 generator has been added that converts classic dbus1 service
8115 activation files automatically into native systemd .busname
8116 and .service units.
8117
8118 * sd-bus: add a light-weight vtable implementation that allows
8119 defining objects on the bus with a simple static const
8120 vtable array of its methods, signals and properties.
8121
8b7d0494 8122 * systemd will not generate or install static dbus
e49b5aad 8123 introspection data anymore to /usr/share/dbus-1/interfaces,
1e190502 8124 as the precise format of these files is unclear, and
e49b5aad
LP
8125 nothing makes use of it.
8126
8127 * A proxy daemon is now provided to proxy clients connecting
8128 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
8129 compatibility with classic D-Bus.
8130
8131 * A bus driver implementation has been added that supports the
8132 classic D-Bus bus driver calls on kdbus, also for
8133 compatibility purposes.
8134
8135 * A new API "sd-event.h" has been added that implements a
8136 minimal event loop API built around epoll. It provides a
8137 couple of features that direct epoll usage is lacking:
b9761003 8138 prioritization of events, scales to large numbers of timer
e49b5aad
LP
8139 events, per-event timer slack (accuracy), system-wide
8140 coalescing of timer events, exit handlers, watchdog
8141 supervision support using systemd's sd_notify() API, child
8142 process handling.
8143
8144 * A new API "sd-rntl.h" has been added that provides an API
8145 around the route netlink interface of the kernel, similar in
8146 style to "sd-bus.h".
8147
7e95eda5
PF
8148 * A new API "sd-dhcp-client.h" has been added that provides a
8149 small DHCPv4 client-side implementation. This is used by
e49b5aad
LP
8150 "systemd-networkd".
8151
4c2413bf 8152 * There is a new kernel command line option
8b7d0494
JSJ
8153 "systemd.restore_state=0|1". When set to "0", none of the
8154 systemd tools will restore saved runtime state to hardware
8155 devices. More specifically, the rfkill and backlight states
8156 are not restored.
e49b5aad
LP
8157
8158 * The FsckPassNo= compatibility option in mount/service units
8159 has been removed. The fstab generator will now add the
8160 necessary dependencies automatically, and does not require
8161 PID1's support for that anymore.
8162
8b7d0494 8163 * journalctl gained a new switch, --list-boots, that lists
e49b5aad
LP
8164 recent boots with their times and boot IDs.
8165
8166 * The various tools like systemctl, loginctl, timedatectl,
8167 busctl, systemd-run, ... have gained a new switch "-M" to
8168 connect to a specific, local OS container (as direct
8169 connection, without requiring SSH). This works on any
8170 container that is registered with machined, such as those
8171 created by libvirt-lxc or nspawn.
8172
8173 * systemd-run and systemd-analyze also gained support for "-H"
4c2413bf 8174 to connect to remote hosts via SSH. This is particularly
8b7d0494
JSJ
8175 useful for systemd-run because it enables queuing of jobs
8176 onto remote systems.
e49b5aad
LP
8177
8178 * machinectl gained a new command "login" to open a getty
8179 login in any local container. This works with any container
8180 that is registered with machined (such as those created by
8e420494 8181 libvirt-lxc or nspawn), and which runs systemd inside.
e49b5aad
LP
8182
8183 * machinectl gained a new "reboot" command that may be used to
8184 trigger a reboot on a specific container that is registered
8185 with machined. This works on any container that runs an init
8186 system of some kind.
8187
8188 * systemctl gained a new "list-timers" command to print a nice
8189 listing of installed timer units with the times they elapse
8190 next.
8191
8192 * Alternative reboot() parameters may now be specified on the
8193 "systemctl reboot" command line and are passed to the
8194 reboot() system call.
8195
8196 * systemctl gained a new --job-mode= switch to configure the
8197 mode to queue a job with. This is a more generic version of
8b7d0494 8198 --fail, --irreversible, and --ignore-dependencies, which are
e49b5aad
LP
8199 still available but not advertised anymore.
8200
e49b5aad
LP
8201 * /etc/systemd/system.conf gained new settings to configure
8202 various default timeouts of units, as well as the default
b9761003 8203 start limit interval and burst. These may still be overridden
e49b5aad
LP
8204 within each Unit.
8205
270f1624
LP
8206 * PID1 will now export on the bus profile data of the security
8207 policy upload process (such as the SELinux policy upload to
8e420494 8208 the kernel).
e49b5aad 8209
4670e9d5 8210 * journald: when forwarding logs to the console, include
1e190502
ZJS
8211 timestamps (following the setting in
8212 /sys/module/printk/parameters/time).
e49b5aad
LP
8213
8214 * OnCalendar= in timer units now understands the special
8215 strings "yearly" and "annually". (Both are equivalent)
8216
8217 * The accuracy of timer units is now configurable with the new
8218 AccuracySec= setting. It defaults to 1min.
8219
8220 * A new dependency type JoinsNamespaceOf= has been added that
8221 allows running two services within the same /tmp and network
8222 namespace, if PrivateNetwork= or PrivateTmp= are used.
8223
8224 * A new command "cat" has been added to systemctl. It outputs
8225 the original unit file of a unit, and concatenates the
1e190502
ZJS
8226 contents of additional "drop-in" unit file snippets, so that
8227 the full configuration is shown.
e49b5aad
LP
8228
8229 * systemctl now supports globbing on the various "list-xyz"
8230 commands, like "list-units" or "list-sockets", as well as on
1e190502
ZJS
8231 those commands which take multiple unit names.
8232
8233 * journalctl's --unit= switch gained support for globbing.
e49b5aad
LP
8234
8235 * All systemd daemons now make use of the watchdog logic so
8236 that systemd automatically notices when they hang.
8237
4c2413bf 8238 * If the $container_ttys environment variable is set,
e49b5aad
LP
8239 getty-generator will automatically spawn a getty for each
8240 listed tty. This is useful for container managers to request
8241 login gettys to be spawned on as many ttys as needed.
8242
8243 * %h, %s, %U specifier support is not available anymore when
8244 used in unit files for PID 1. This is because NSS calls are
8245 not safe from PID 1. They stay available for --user
8246 instances of systemd, and as special case for the root user.
8247
e49b5aad
LP
8248 * loginctl gained a new "--no-legend" switch to turn off output
8249 of the legend text.
8250
8251 * The "sd-login.h" API gained three new calls:
8252 sd_session_is_remote(), sd_session_get_remote_user(),
8253 sd_session_get_remote_host() to query information about
8254 remote sessions.
8255
8e420494
LP
8256 * The udev hardware database now also carries vendor/product
8257 information of SDIO devices.
e49b5aad
LP
8258
8259 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
8260 determine whether watchdog notifications are requested by
8261 the system manager.
8262
1e190502 8263 * Socket-activated per-connection services now include a
e49b5aad
LP
8264 short description of the connection parameters in the
8265 description.
8266
4c2413bf 8267 * tmpfiles gained a new "--boot" option. When this is not used,
e49b5aad 8268 only lines where the command character is not suffixed with
4670e9d5 8269 "!" are executed. When this option is specified, those
1e190502
ZJS
8270 options are executed too. This partitions tmpfiles
8271 directives into those that can be safely executed at any
8272 time, and those which should be run only at boot (for
8273 example, a line that creates /run/nologin).
e49b5aad 8274
c0c5af00 8275 * A new API "sd-resolve.h" has been added which provides a simple
38b38500 8276 asynchronous wrapper around glibc NSS hostname resolution
e49b5aad 8277 calls, such as getaddrinfo(). In contrast to glibc's
4c2413bf
JE
8278 getaddrinfo_a(), it does not use signals. In contrast to most
8279 other asynchronous name resolution libraries, this one does
8280 not reimplement DNS, but reuses NSS, so that alternate
38b38500 8281 hostname resolution systems continue to work, such as mDNS,
8b7d0494 8282 LDAP, etc. This API is based on libasyncns, but it has been
e49b5aad
LP
8283 cleaned up for inclusion in systemd.
8284
6300b3ec
LP
8285 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
8286 "sd-daemon.h" are no longer found in individual libraries
8287 libsystemd-journal.so, libsystemd-login.so,
8288 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
8b7d0494
JSJ
8289 merged them into a single library, libsystemd.so, which
8290 provides all symbols. The reason for this is cyclic
e49b5aad 8291 dependencies, as these libraries tend to use each other's
d28315e4 8292 symbols. So far, we have managed to workaround that by linking
6300b3ec
LP
8293 a copy of a good part of our code into each of these
8294 libraries again and again, which, however, makes certain
8295 things hard to do, like sharing static variables. Also, it
8296 substantially increases footprint. With this change, there
8297 is only one library for the basic APIs systemd
8298 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
8299 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
8300 library as well, however are subject to the --enable-kdbus
8301 switch (see below). Note that "sd-dhcp-client.h" is not part
8302 of this library (this is because it only consumes, never
8303 provides, services of/to other APIs). To make the transition
8b7d0494 8304 easy from the separate libraries to the unified one, we
4c2413bf 8305 provide the --enable-compat-libs compile-time switch which
e49b5aad
LP
8306 will generate stub libraries that are compatible with the
8307 old ones but redirect all calls to the new one.
8308
8b7d0494 8309 * All of the kdbus logic and the new APIs "sd-bus.h",
e49b5aad 8310 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
8b7d0494
JSJ
8311 and "sd-utf8.h" are compile-time optional via the
8312 "--enable-kdbus" switch, and they are not compiled in by
8313 default. To make use of kdbus, you have to explicitly enable
4c2413bf 8314 the switch. Note however, that neither the kernel nor the
e49b5aad
LP
8315 userspace API for all of this is considered stable yet. We
8316 want to maintain the freedom to still change the APIs for
4c2413bf 8317 now. By specifying this build-time switch, you acknowledge
e49b5aad 8318 that you are aware of the instability of the current
ad42cf73
KS
8319 APIs.
8320
8321 * Also, note that while kdbus is pretty much complete,
e49b5aad 8322 it lacks one thing: proper policy support. This means you
8b7d0494 8323 can build a fully working system with all features; however,
4c2413bf
JE
8324 it will be highly insecure. Policy support will be added in
8325 one of the next releases, at the same time that we will
8326 declare the APIs stable.
e49b5aad 8327
81c7dd89 8328 * When the kernel command line argument "kdbus" is specified,
ad42cf73 8329 systemd will automatically load the kdbus.ko kernel module. At
8b7d0494 8330 this stage of development, it is only useful for testing kdbus
ad42cf73 8331 and should not be used in production. Note: if "--enable-kdbus"
8b7d0494 8332 is specified, and the kdbus.ko kernel module is available, and
ad42cf73
KS
8333 "kdbus" is added to the kernel command line, the entire system
8334 runs with kdbus instead of dbus-daemon, with the above mentioned
8335 problem of missing the system policy enforcement. Also a future
8336 version of kdbus.ko or a newer systemd will not be compatible with
8337 each other, and will unlikely be able to boot the machine if only
8338 one of them is updated.
8339
e49b5aad 8340 * systemctl gained a new "import-environment" command which
4c2413bf 8341 uploads the caller's environment (or parts thereof) into the
e49b5aad
LP
8342 service manager so that it is inherited by services started
8343 by the manager. This is useful to upload variables like
8344 $DISPLAY into the user service manager.
8345
8346 * A new PrivateDevices= switch has been added to service units
8347 which allows running a service with a namespaced /dev
8348 directory that does not contain any device nodes for
4c2413bf 8349 physical devices. More specifically, it only includes devices
8b7d0494 8350 such as /dev/null, /dev/urandom, and /dev/zero which are API
e49b5aad
LP
8351 entry points.
8352
8353 * logind has been extended to support behaviour like VT
8354 switching on seats that do not support a VT. This makes
8355 multi-session available on seats that are not the first seat
8356 (seat0), and on systems where kernel support for VTs has
8b7d0494 8357 been disabled at compile-time.
e49b5aad
LP
8358
8359 * If a process holds a delay lock for system sleep or shutdown
1e190502 8360 and fails to release it in time, we will now log its
e49b5aad
LP
8361 identity. This makes it easier to identify processes that
8362 cause slow suspends or power-offs.
8363
1e190502
ZJS
8364 * When parsing /etc/crypttab, support for a new key-slot=
8365 option as supported by Debian is added. It allows indicating
8366 which LUKS slot to use on disk, speeding up key loading.
e49b5aad 8367
000b1ba5 8368 * The sd_journal_sendv() API call has been checked and
1e190502
ZJS
8369 officially declared to be async-signal-safe so that it may
8370 be invoked from signal handlers for logging purposes.
e49b5aad
LP
8371
8372 * Boot-time status output is now enabled automatically after a
8373 short timeout if boot does not progress, in order to give
8e420494 8374 the user an indication what she or he is waiting for.
1e190502
ZJS
8375
8376 * The boot-time output has been improved to show how much time
8377 remains until jobs expire.
e49b5aad
LP
8378
8379 * The KillMode= switch in service units gained a new possible
8b7d0494 8380 value "mixed". If set, and the unit is shut down, then the
e49b5aad 8381 initial SIGTERM signal is sent only to the main daemon
8e420494 8382 process, while the following SIGKILL signal is sent to
e49b5aad
LP
8383 all remaining processes of the service.
8384
4c2413bf
JE
8385 * When a scope unit is registered, a new property "Controller"
8386 may be set. If set to a valid bus name, systemd will send a
e49b5aad
LP
8387 RequestStop() signal to this name when it would like to shut
8388 down the scope. This may be used to hook manager logic into
8389 the shutdown logic of scope units. Also, scope units may now
8b7d0494 8390 be put in a special "abandoned" state, in which case the
e49b5aad
LP
8391 manager process which created them takes no further
8392 responsibilities for it.
8393
1e190502 8394 * When reading unit files, systemd will now verify
e49b5aad
LP
8395 the access mode of these files, and warn about certain
8396 suspicious combinations. This has been added to make it
8397 easier to track down packaging bugs where unit files are
8398 marked executable or world-writable.
8399
8400 * systemd-nspawn gained a new "--setenv=" switch to set
8b7d0494 8401 container-wide environment variables. The similar option in
1e190502
ZJS
8402 systemd-activate was renamed from "--environment=" to
8403 "--setenv=" for consistency.
e49b5aad
LP
8404
8405 * systemd-nspawn has been updated to create a new kdbus domain
8406 for each container that is invoked, thus allowing each
b9761003 8407 container to have its own set of system and user buses,
8b7d0494 8408 independent of the host.
e49b5aad
LP
8409
8410 * systemd-nspawn gained a new --drop-capability= switch to run
8411 the container with less capabilities than the default. Both
b9761003 8412 --drop-capability= and --capability= now take the special
e49b5aad
LP
8413 string "all" for dropping or keeping all capabilities.
8414
8415 * systemd-nspawn gained new switches for executing containers
8416 with specific SELinux labels set.
8417
8418 * systemd-nspawn gained a new --quiet switch to not generate
8419 any additional output but the container's own console
8420 output.
8421
8422 * systemd-nspawn gained a new --share-system switch to run a
8423 container without PID namespacing enabled.
8424
8425 * systemd-nspawn gained a new --register= switch to control
1e190502 8426 whether the container is registered with systemd-machined or
8e420494 8427 not. This is useful for containers that do not run full
e49b5aad
LP
8428 OS images, but only specific apps.
8429
8430 * systemd-nspawn gained a new --keep-unit which may be used
8b7d0494 8431 when invoked as the only program from a service unit, and
e49b5aad 8432 results in registration of the unit service itself in
1e190502 8433 systemd-machined, instead of a newly opened scope unit.
e49b5aad
LP
8434
8435 * systemd-nspawn gained a new --network-interface= switch for
8436 moving arbitrary interfaces to the container. The new
4c2413bf 8437 --network-veth switch creates a virtual Ethernet connection
8b7d0494
JSJ
8438 between host and container. The new --network-bridge=
8439 switch then allows assigning the host side of this virtual
8440 Ethernet connection to a bridge device.
e49b5aad 8441
6afc95b7
LP
8442 * systemd-nspawn gained a new --personality= switch for
8443 setting the kernel personality for the container. This is
70a44afe 8444 useful when running a 32-bit container on a 64-bit host. A
b8bde116
JE
8445 similar option Personality= is now also available for service
8446 units to use.
6afc95b7 8447
e49b5aad
LP
8448 * logind will now also track a "Desktop" identifier for each
8449 session which encodes the desktop environment of it. This is
8450 useful for desktop environments that want to identify
8451 multiple running sessions of itself easily.
8452
8453 * A new SELinuxContext= setting for service units has been
8454 added that allows setting a specific SELinux execution
8455 context for a service.
8456
8457 * Most systemd client tools will now honour $SYSTEMD_LESS for
8458 settings of the "less" pager. By default, these tools will
8b7d0494
JSJ
8459 override $LESS to allow certain operations to work, such as
8460 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
e49b5aad
LP
8461 influence this logic.
8462
8463 * systemd's "seccomp" hook-up has been changed to make use of
8464 the libseccomp library instead of using its own
8465 implementation. This has benefits for portability among
8466 other things.
8467
4c2413bf 8468 * For usage together with SystemCallFilter=, a new
8b7d0494 8469 SystemCallErrorNumber= setting has been introduced that
b8bde116
JE
8470 allows configuration of a system error number to be returned
8471 on filtered system calls, instead of immediately killing the
e49b5aad
LP
8472 process. Also, SystemCallArchitectures= has been added to
8473 limit access to system calls of a particular architecture
8474 (in order to turn off support for unused secondary
4c2413bf 8475 architectures). There is also a global
8b7d0494 8476 SystemCallArchitectures= setting in system.conf now to turn
e49b5aad
LP
8477 off support for non-native system calls system-wide.
8478
210054d7
KS
8479 * systemd requires a kernel with a working name_to_handle_at(),
8480 please see the kernel config requirements in the README file.
8481
e49b5aad
LP
8482 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
8483 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
8484 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
8485 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
8486 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
8487 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
8488 Elia Pinto, Florian Weimer, George McCollister, Goffredo
8489 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
8490 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
8491 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
8492 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
8493 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
8494 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
8495 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
8496 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
8497 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
8498 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
8499 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
8500 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
8501 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
8502 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
8503 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
8504 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
8505 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
8506
ccddd104 8507 — Berlin, 2014-02-20
e49b5aad 8508
cd4010b3
LP
8509CHANGES WITH 208:
8510
8511 * logind has gained support for facilitating privileged input
8512 and drm device access for unprivileged clients. This work is
8513 useful to allow Wayland display servers (and similar
8514 programs, such as kmscon) to run under the user's ID and
8515 access input and drm devices which are normally
8516 protected. When this is used (and the kernel is new enough)
8517 logind will "mute" IO on the file descriptors passed to
8518 Wayland as long as it is in the background and "unmute" it
8519 if it returns into the foreground. This allows secure
8520 session switching without allowing background sessions to
8521 eavesdrop on input and display data. This also introduces
8522 session switching support if VT support is turned off in the
8523 kernel, and on seats that are not seat0.
8524
8525 * A new kernel command line option luks.options= is understood
06b643e7 8526 now which allows specifying LUKS options for usage for LUKS
cd4010b3
LP
8527 encrypted partitions specified with luks.uuid=.
8528
8529 * tmpfiles.d(5) snippets may now use specifier expansion in
8530 path names. More specifically %m, %b, %H, %v, are now
8531 replaced by the local machine id, boot id, hostname, and
8532 kernel version number.
8533
8534 * A new tmpfiles.d(5) command "m" has been introduced which
8535 may be used to change the owner/group/access mode of a file
d28315e4 8536 or directory if it exists, but do nothing if it does not.
cd4010b3
LP
8537
8538 * This release removes high-level support for the
8539 MemorySoftLimit= cgroup setting. The underlying kernel
8540 cgroup attribute memory.soft_limit= is currently badly
8541 designed and likely to be removed from the kernel API in its
d28315e4 8542 current form, hence we should not expose it for now.
cd4010b3
LP
8543
8544 * The memory.use_hierarchy cgroup attribute is now enabled for
8545 all cgroups systemd creates in the memory cgroup
8546 hierarchy. This option is likely to be come the built-in
cc98b302
TH
8547 default in the kernel anyway, and the non-hierarchical mode
8548 never made much sense in the intrinsically hierarchical
cd4010b3
LP
8549 cgroup system.
8550
8551 * A new field _SYSTEMD_SLICE= is logged along with all journal
8552 messages containing the slice a message was generated
8553 from. This is useful to allow easy per-customer filtering of
8554 logs among other things.
8555
8556 * systemd-journald will no longer adjust the group of journal
8557 files it creates to the "systemd-journal" group. Instead we
8558 rely on the journal directory to be owned by the
8559 "systemd-journal" group, and its setgid bit set, so that the
8560 kernel file system layer will automatically enforce that
8561 journal files inherit this group assignment. The reason for
8562 this change is that we cannot allow NSS look-ups from
8563 journald which would be necessary to resolve
8564 "systemd-journal" to a numeric GID, because this might
8565 create deadlocks if NSS involves synchronous queries to
8566 other daemons (such as nscd, or sssd) which in turn are
8567 logging clients of journald and might block on it, which
8568 would then dead lock. A tmpfiles.d(5) snippet included in
8569 systemd will make sure the setgid bit and group are
8570 properly set on the journal directory if it exists on every
8571 boot. However, we recommend adjusting it manually after
8572 upgrades too (or from RPM scriptlets), so that the change is
8573 not delayed until next reboot.
8574
8575 * Backlight and random seed files in /var/lib/ have moved into
8576 the /var/lib/systemd/ directory, in order to centralize all
8577 systemd generated files in one directory.
8578
8579 * Boot time performance measurements (as displayed by
8580 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
8581 performance information if that's available to determine how
8582 much time BIOS and boot loader initialization required. With
8583 a sufficiently new BIOS you hence no longer need to boot
8584 with Gummiboot to get access to such information.
8585
8586 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
8587 Cristian Rodríguez, Dave Reisner, David Herrmann, David
8588 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
8589 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
8590 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
8591 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
8592 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8593
ccddd104 8594 — Berlin, 2013-10-02
cd4010b3 8595
4f0be680
LP
8596CHANGES WITH 207:
8597
8598 * The Restart= option for services now understands a new
f3a165b0 8599 on-watchdog setting, which will restart the service
4f0be680
LP
8600 automatically if the service stops sending out watchdog keep
8601 alive messages (as configured with WatchdogSec=).
8602
8603 * The getty generator (which is responsible for bringing up a
8604 getty on configured serial consoles) will no longer only
8605 start a getty on the primary kernel console but on all
8606 others, too. This makes the order in which console= is
8607 specified on the kernel command line less important.
8608
8609 * libsystemd-logind gained a new sd_session_get_vt() call to
8610 retrieve the VT number of a session.
8611
8612 * If the option "tries=0" is set for an entry of /etc/crypttab
8613 its passphrase is queried indefinitely instead of any
8614 maximum number of tries.
8615
8616 * If a service with a configure PID file terminates its PID
8617 file will now be removed automatically if it still exists
8618 afterwards. This should put an end to stale PID files.
8619
8620 * systemd-run will now also take relative binary path names
8621 for execution and no longer insists on absolute paths.
8622
8623 * InaccessibleDirectories= and ReadOnlyDirectories= now take
8624 paths that are optionally prefixed with "-" to indicate that
d28315e4 8625 it should not be considered a failure if they do not exist.
4f0be680 8626
f3a165b0
KS
8627 * journalctl -o (and similar commands) now understands a new
8628 output mode "short-precise", it is similar to "short" but
4f0be680
LP
8629 shows timestamps with usec accuracy.
8630
8631 * The option "discard" (as known from Debian) is now
8632 synonymous to "allow-discards" in /etc/crypttab. In fact,
387abf80 8633 "discard" is preferred now (since it is easier to remember
4f0be680
LP
8634 and type).
8635
f3a165b0 8636 * Some licensing clean-ups were made, so that more code is now
4f0be680
LP
8637 LGPL-2.1 licensed than before.
8638
8639 * A minimal tool to save/restore the display backlight
8640 brightness across reboots has been added. It will store the
f3a165b0 8641 backlight setting as late as possible at shutdown, and
4f0be680
LP
8642 restore it as early as possible during reboot.
8643
8644 * A logic to automatically discover and enable home and swap
8645 partitions on GPT disks has been added. With this in place
8646 /etc/fstab becomes optional for many setups as systemd can
8647 discover certain partitions located on the root disk
8648 automatically. Home partitions are recognized under their
8649 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
8650 partitions are recognized under their GPT type ID
8651 0657fd6da4ab43c484e50933c84b4f4f.
8652
8653 * systemd will no longer pass any environment from the kernel
8654 or initrd to system services. If you want to set an
8655 environment for all services, do so via the kernel command
8656 line systemd.setenv= assignment.
8657
387abf80
LP
8658 * The systemd-sysctl tool no longer natively reads the file
8659 /etc/sysctl.conf. If desired, the file should be symlinked
8660 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
8661 legacy support by a symlink rather than built-in code, it
8662 also makes the otherwise hidden order of application of the
8663 different files visible. (Note that this partly reverts to a
8664 pre-198 application order of sysctl knobs!)
04bf3c1a 8665
4f0be680
LP
8666 * The "systemctl set-log-level" and "systemctl dump" commands
8667 have been moved to systemd-analyze.
8668
8669 * systemd-run learned the new --remain-after-exit switch,
8670 which causes the scope unit not to be cleaned up
8671 automatically after the process terminated.
8672
8673 * tmpfiles learned a new --exclude-prefix= switch to exclude
8674 certain paths from operation.
8675
8676 * journald will now automatically flush all messages to disk
f47ad593
ZJS
8677 as soon as a message at the log level CRIT, ALERT or EMERG
8678 is received.
4f0be680
LP
8679
8680 Contributions from: Andrew Cook, Brandon Philips, Christian
8681 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
8682 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
8683 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
8684 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
8685 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
8686 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
8687 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
8688 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
8689 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
8690 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
8691 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
8692 William Giokas, Zbigniew Jędrzejewski-Szmek
8693
ccddd104 8694 — Berlin, 2013-09-13
4f0be680 8695
408f281b
LP
8696CHANGES WITH 206:
8697
8698 * The documentation has been updated to cover the various new
8699 concepts introduced with 205.
8700
8701 * Unit files now understand the new %v specifier which
8702 resolves to the kernel version string as returned by "uname
8703 -r".
8704
8705 * systemctl now supports filtering the unit list output by
8706 load state, active state and sub state, using the new
33b521be 8707 --state= parameter.
408f281b
LP
8708
8709 * "systemctl status" will now show the results of the
8710 condition checks (like ConditionPathExists= and similar) of
8711 the last start attempts of the unit. They are also logged to
8712 the journal.
8713
8714 * "journalctl -b" may now be used to look for boot output of a
8715 specific boot. Try "journalctl -b -1" for the previous boot,
8716 but the syntax is substantially more powerful.
8717
8718 * "journalctl --show-cursor" has been added which prints the
8719 cursor string the last shown log line. This may then be used
8720 with the new "journalctl --after-cursor=" switch to continue
8721 browsing logs from that point on.
8722
8723 * "journalctl --force" may now be used to force regeneration
8724 of an FSS key.
8725
251cc819
LP
8726 * Creation of "dead" device nodes has been moved from udev
8727 into kmod and tmpfiles. Previously, udev would read the kmod
8728 databases to pre-generate dead device nodes based on meta
8729 information contained in kernel modules, so that these would
8730 be auto-loaded on access rather then at boot. As this
d28315e4 8731 does not really have much to do with the exposing actual
251cc819
LP
8732 kernel devices to userspace this has always been slightly
8733 alien in the udev codebase. Following the new scheme kmod
8734 will now generate a runtime snippet for tmpfiles from the
8735 module meta information and it now is tmpfiles' job to the
8736 create the nodes. This also allows overriding access and
8737 other parameters for the nodes using the usual tmpfiles
8738 facilities. As side effect this allows us to remove the
8739 CAP_SYS_MKNOD capability bit from udevd entirely.
8740
8741 * logind's device ACLs may now be applied to these "dead"
8742 devices nodes too, thus finally allowing managed access to
ce830873 8743 devices such as /dev/snd/sequencer without loading the
251cc819 8744 backing module right-away.
408f281b
LP
8745
8746 * A new RPM macro has been added that may be used to apply
8747 tmpfiles configuration during package installation.
8748
8749 * systemd-detect-virt and ConditionVirtualization= now can
8750 detect User-Mode-Linux machines (UML).
8751
251cc819
LP
8752 * journald will now implicitly log the effective capabilities
8753 set of processes in the message metadata.
408f281b
LP
8754
8755 * systemd-cryptsetup has gained support for TrueCrypt volumes.
8756
8757 * The initrd interface has been simplified (more specifically,
8758 support for passing performance data via environment
8759 variables and fsck results via files in /run has been
8760 removed). These features were non-essential, and are
8761 nowadays available in a much nicer way by having systemd in
8762 the initrd serialize its state and have the hosts systemd
8763 deserialize it again.
8764
28f5c779
KS
8765 * The udev "keymap" data files and tools to apply keyboard
8766 specific mappings of scan to key codes, and force-release
8767 scan code lists have been entirely replaced by a udev
8768 "keyboard" builtin and a hwdb data file.
408f281b 8769
251cc819
LP
8770 * systemd will now honour the kernel's "quiet" command line
8771 argument also during late shutdown, resulting in a
8772 completely silent shutdown when used.
8773
8774 * There's now an option to control the SO_REUSEPORT socket
8775 option in .socket units.
8776
8777 * Instance units will now automatically get a per-template
8778 subslice of system.slice unless something else is explicitly
8779 configured. For example, instances of sshd@.service will now
8780 implicitly be placed in system-sshd.slice rather than
8781 system.slice as before.
8782
8783 * Test coverage support may now be enabled at build time.
8784
8785 Contributions from: Dave Reisner, Frederic Crozat, Harald
8786 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
8787 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
8788 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
8789 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
8790 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
8791 Giokas, Zbigniew Jędrzejewski-Szmek
8792
ccddd104 8793 — Berlin, 2013-07-23
4f0be680 8794
00aa832b
LP
8795CHANGES WITH 205:
8796
8797 * Two new unit types have been introduced:
8798
8799 Scope units are very similar to service units, however, are
ccddd104 8800 created out of pre-existing processes — instead of PID 1
00aa832b
LP
8801 forking off the processes. By using scope units it is
8802 possible for system services and applications to group their
8803 own child processes (worker processes) in a powerful way
8804 which then maybe used to organize them, or kill them
8805 together, or apply resource limits on them.
8806
8807 Slice units may be used to partition system resources in an
cc98b302 8808 hierarchical fashion and then assign other units to them. By
00aa832b
LP
8809 default there are now three slices: system.slice (for all
8810 system services), user.slice (for all user sessions),
8811 machine.slice (for VMs and containers).
8812
8813 Slices and scopes have been introduced primarily in
8814 context of the work to move cgroup handling to a
8815 single-writer scheme, where only PID 1
8816 creates/removes/manages cgroups.
8817
8818 * There's a new concept of "transient" units. In contrast to
8819 normal units these units are created via an API at runtime,
8820 not from configuration from disk. More specifically this
8821 means it is now possible to run arbitrary programs as
8822 independent services, with all execution parameters passed
8823 in via bus APIs rather than read from disk. Transient units
8824 make systemd substantially more dynamic then it ever was,
8825 and useful as a general batch manager.
8826
8827 * logind has been updated to make use of scope and slice units
8828 for managing user sessions. As a user logs in he will get
8829 his own private slice unit, to which all sessions are added
8830 as scope units. We also added support for automatically
8831 adding an instance of user@.service for the user into the
8832 slice. Effectively logind will no longer create cgroup
8833 hierarchies on its own now, it will defer entirely to PID 1
8834 for this by means of scope, service and slice units. Since
8835 user sessions this way become entities managed by PID 1
8836 the output of "systemctl" is now a lot more comprehensive.
8837
8838 * A new mini-daemon "systemd-machined" has been added which
8839 may be used by virtualization managers to register local
8840 VMs/containers. nspawn has been updated accordingly, and
8841 libvirt will be updated shortly. machined will collect a bit
8842 of meta information about the VMs/containers, and assign
8843 them their own scope unit (see above). The collected
8844 meta-data is then made available via the "machinectl" tool,
8845 and exposed in "ps" and similar tools. machined/machinectl
8846 is compile-time optional.
8847
8848 * As discussed earlier, the low-level cgroup configuration
8849 options ControlGroup=, ControlGroupModify=,
8850 ControlGroupPersistent=, ControlGroupAttribute= have been
8851 removed. Please use high-level attribute settings instead as
8852 well as slice units.
8853
8854 * A new bus call SetUnitProperties() has been added to alter
8855 various runtime parameters of a unit. This is primarily
8856 useful to alter cgroup parameters dynamically in a nice way,
8857 but will be extended later on to make more properties
8858 modifiable at runtime. systemctl gained a new set-properties
8859 command that wraps this call.
8860
8861 * A new tool "systemd-run" has been added which can be used to
8862 run arbitrary command lines as transient services or scopes,
8863 while configuring a number of settings via the command
8864 line. This tool is currently very basic, however already
8865 very useful. We plan to extend this tool to even allow
8866 queuing of execution jobs with time triggers from the
8867 command line, similar in fashion to "at".
8868
8869 * nspawn will now inform the user explicitly that kernels with
8870 audit enabled break containers, and suggest the user to turn
8871 off audit.
8872
8873 * Support for detecting the IMA and AppArmor security
8874 frameworks with ConditionSecurity= has been added.
8875
8876 * journalctl gained a new "-k" switch for showing only kernel
1fda0ab5
ZJS
8877 messages, mimicking dmesg output; in addition to "--user"
8878 and "--system" switches for showing only user's own logs
8879 and system logs.
00aa832b
LP
8880
8881 * systemd-delta can now show information about drop-in
8882 snippets extending unit files.
8883
8884 * libsystemd-bus has been substantially updated but is still
8885 not available as public API.
8886
8887 * systemd will now look for the "debug" argument on the kernel
499b604b 8888 command line and enable debug logging, similar to what
00aa832b
LP
8889 "systemd.log_level=debug" already did before.
8890
8891 * "systemctl set-default", "systemctl get-default" has been
8892 added to configure the default.target symlink, which
8893 controls what to boot into by default.
8894
1fda0ab5
ZJS
8895 * "systemctl set-log-level" has been added as a convenient
8896 way to raise and lower systemd logging threshold.
8897
00aa832b
LP
8898 * "systemd-analyze plot" will now show the time the various
8899 generators needed for execution, as well as information
8900 about the unit file loading.
8901
00aa832b
LP
8902 * libsystemd-journal gained a new sd_journal_open_files() call
8903 for opening specific journal files. journactl also gained a
8904 new switch to expose this new functionality. Previously we
8905 only supported opening all files from a directory, or all
8906 files from the system, as opening individual files only is
8907 racy due to journal file rotation.
8908
8909 * systemd gained the new DefaultEnvironment= setting in
8910 /etc/systemd/system.conf to set environment variables for
8911 all services.
8912
8913 * If a privileged process logs a journal message with the
8914 OBJECT_PID= field set, then journald will automatically
8915 augment this with additional OBJECT_UID=, OBJECT_GID=,
8916 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
8917 system services want to log events about specific client
8918 processes. journactl/systemctl has been updated to make use
8919 of this information if all log messages regarding a specific
8920 unit is requested.
8921
8922 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
8923 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
8924 Reisner, David Coppa, David King, David Strauss, Eelco
8925 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
8926 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
8927 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
8928 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
8929 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
8930 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
8931 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
8932 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
8933 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
8934 Łukasz Stelmach, 장동준
8935
606c24e3
LP
8936CHANGES WITH 204:
8937
8938 * The Python bindings gained some minimal support for the APIs
8939 exposed by libsystemd-logind.
8940
8941 * ConditionSecurity= gained support for detecting SMACK. Since
8942 this condition already supports SELinux and AppArmor we only
8943 miss IMA for this. Patches welcome!
8944
8945 Contributions from: Karol Lewandowski, Lennart Poettering,
8946 Zbigniew Jędrzejewski-Szmek
8947
2f3fcf85
LP
8948CHANGES WITH 203:
8949
8950 * systemd-nspawn will now create /etc/resolv.conf if
8951 necessary, before bind-mounting the host's file onto it.
8952
8953 * systemd-nspawn will now store meta information about a
8954 container on the container's cgroup as extended attribute
8955 fields, including the root directory.
8956
8957 * The cgroup hierarchy has been reworked in many ways. All
8958 objects any of the components systemd creates in the cgroup
b82eed9a 8959 tree are now suffixed. More specifically, user sessions are
2f3fcf85
LP
8960 now placed in cgroups suffixed with ".session", users in
8961 cgroups suffixed with ".user", and nspawn containers in
8962 cgroups suffixed with ".nspawn". Furthermore, all cgroup
8963 names are now escaped in a simple scheme to avoid collision
8964 of userspace object names with kernel filenames. This work
8965 is preparation for making these objects relocatable in the
8966 cgroup tree, in order to allow easy resource partitioning of
8967 these objects without causing naming conflicts.
8968
8969 * systemctl list-dependencies gained the new switches
8970 --plain, --reverse, --after and --before.
8971
8972 * systemd-inhibit now shows the process name of processes that
8973 have taken an inhibitor lock.
8974
8975 * nss-myhostname will now also resolve "localhost"
8976 implicitly. This makes /etc/hosts an optional file and
8977 nicely handles that on IPv6 ::1 maps to both "localhost" and
8978 the local hostname.
8979
8980 * libsystemd-logind.so gained a new call
8981 sd_get_machine_names() to enumerate running containers and
8982 VMs (currently only supported by very new libvirt and
8983 nspawn). sd_login_monitor can now be used to watch
8984 VMs/containers coming and going.
8985
8986 * .include is not allowed recursively anymore, and only in
8987 unit files. Usually it is better to use drop-in snippets in
8988 .d/*.conf anyway, as introduced with systemd 198.
8989
8990 * systemd-analyze gained a new "critical-chain" command that
8991 determines the slowest chain of units run during system
8992 boot-up. It is very useful for tracking down where
8993 optimizing boot time is the most beneficial.
8994
8995 * systemd will no longer allow manipulating service paths in
8996 the name=systemd:/system cgroup tree using ControlGroup= in
8997 units. (But is still fine with it in all other dirs.)
8998
8999 * There's a new systemd-nspawn@.service service file that may
9000 be used to easily run nspawn containers as system
9001 services. With the container's root directory in
9002 /var/lib/container/foobar it is now sufficient to run
9003 "systemctl start systemd-nspawn@foobar.service" to boot it.
9004
9005 * systemd-cgls gained a new parameter "--machine" to list only
9006 the processes within a certain container.
9007
9008 * ConditionSecurity= now can check for "apparmor". We still
9009 are lacking checks for SMACK and IMA for this condition
9010 check though. Patches welcome!
9011
9012 * A new configuration file /etc/systemd/sleep.conf has been
9013 added that may be used to configure which kernel operation
9014 systemd is supposed to execute when "suspend", "hibernate"
9015 or "hybrid-sleep" is requested. This makes the new kernel
9016 "freeze" state accessible to the user.
9017
9018 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
9019 the passed argument if applicable.
9020
9021 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
9022 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
9023 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
9024 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
9025 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
9026 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
9027 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
9028 Jędrzejewski-Szmek
9029
ef3b5246
LP
9030CHANGES WITH 202:
9031
9032 * The output of 'systemctl list-jobs' got some polishing. The
9033 '--type=' argument may now be passed more than once. A new
9034 command 'systemctl list-sockets' has been added which shows
9035 a list of kernel sockets systemd is listening on with the
9036 socket units they belong to, plus the units these socket
9037 units activate.
9038
9039 * The experimental libsystemd-bus library got substantial
9040 updates to work in conjunction with the (also experimental)
9041 kdbus kernel project. It works well enough to exchange
9042 messages with some sophistication. Note that kdbus is not
9043 ready yet, and the library is mostly an elaborate test case
9044 for now, and not installable.
9045
9046 * systemd gained a new unit 'systemd-static-nodes.service'
9047 that generates static device nodes earlier during boot, and
9048 can run in conjunction with udev.
9049
9050 * libsystemd-login gained a new call sd_pid_get_user_unit()
9051 to retrieve the user systemd unit a process is running
9052 in. This is useful for systems where systemd is used as
9053 session manager.
9054
9055 * systemd-nspawn now places all containers in the new /machine
9056 top-level cgroup directory in the name=systemd
9057 hierarchy. libvirt will soon do the same, so that we get a
9058 uniform separation of /system, /user and /machine for system
9059 services, user processes and containers/virtual
9060 machines. This new cgroup hierarchy is also useful to stick
9061 stable names to specific container instances, which can be
7c04ad2d 9062 recognized later this way (this name may be controlled
ef3b5246
LP
9063 via systemd-nspawn's new -M switch). libsystemd-login also
9064 gained a new call sd_pid_get_machine_name() to retrieve the
9065 name of the container/VM a specific process belongs to.
9066
9067 * bootchart can now store its data in the journal.
9068
9069 * libsystemd-journal gained a new call
9070 sd_journal_add_conjunction() for AND expressions to the
9071 matching logic. This can be used to express more complex
9072 logical expressions.
9073
9074 * journactl can now take multiple --unit= and --user-unit=
9075 switches.
9076
9077 * The cryptsetup logic now understands the "luks.key=" kernel
9078 command line switch for specifying a file to read the
7c04ad2d 9079 decryption key from. Also, if a configured key file is not
ef3b5246
LP
9080 found the tool will now automatically fall back to prompting
9081 the user.
9082
cbeabcfb
ZJS
9083 * Python systemd.journal module was updated to wrap recently
9084 added functions from libsystemd-journal. The interface was
9085 changed to bring the low level interface in s.j._Reader
9086 closer to the C API, and the high level interface in
9087 s.j.Reader was updated to wrap and convert all data about
9088 an entry.
9089
ef3b5246
LP
9090 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
9091 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
9092 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
9093 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
9094 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
9095 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9096
d3a86981
LP
9097CHANGES WITH 201:
9098
9099 * journalctl --update-catalog now understands a new --root=
9100 option to operate on catalogs found in a different root
9101 directory.
9102
9103 * During shutdown after systemd has terminated all running
9104 services a final killing loop kills all remaining left-over
9105 processes. We will now print the name of these processes
9106 when we send SIGKILL to them, since this usually indicates a
9107 problem.
9108
9109 * If /etc/crypttab refers to password files stored on
9110 configured mount points automatic dependencies will now be
9111 generated to ensure the specific mount is established first
9112 before the key file is attempted to be read.
9113
9114 * 'systemctl status' will now show information about the
9115 network sockets a socket unit is listening on.
9116
9117 * 'systemctl status' will also shown information about any
9118 drop-in configuration file for units. (Drop-In configuration
9119 files in this context are files such as
9120 /etc/systemd/systemd/foobar.service.d/*.conf)
9121
9122 * systemd-cgtop now optionally shows summed up CPU times of
9123 cgroups. Press '%' while running cgtop to switch between
9124 percentage and absolute mode. This is useful to determine
9125 which cgroups use up the most CPU time over the entire
9126 runtime of the system. systemd-cgtop has also been updated
9127 to be 'pipeable' for processing with further shell tools.
9128
9129 * 'hostnamectl set-hostname' will now allow setting of FQDN
9130 hostnames.
9131
9132 * The formatting and parsing of time span values has been
9133 changed. The parser now understands fractional expressions
9134 such as "5.5h". The formatter will now output fractional
9135 expressions for all time spans under 1min, i.e. "5.123456s"
9136 rather than "5s 123ms 456us". For time spans under 1s
9137 millisecond values are shown, for those under 1ms
9138 microsecond values are shown. This should greatly improve
9139 all time-related output of systemd.
9140
9141 * libsystemd-login and libsystemd-journal gained new
9142 functions for querying the poll() events mask and poll()
9143 timeout value for integration into arbitrary event
9144 loops.
9145
9146 * localectl gained the ability to list available X11 keymaps
9147 (models, layouts, variants, options).
9148
9149 * 'systemd-analyze dot' gained the ability to filter for
9150 specific units via shell-style globs, to create smaller,
d28315e4 9151 more useful graphs. I.e. it is now possible to create simple
d3a86981
LP
9152 graphs of all the dependencies between only target units, or
9153 of all units that Avahi has dependencies with.
9154
9155 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
9156 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
9157 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
9158 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
9159 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
9160 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
9161 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
9162
9ca3c17f
LP
9163CHANGES WITH 200:
9164
9165 * The boot-time readahead implementation for rotating media
9166 will now read the read-ahead data in multiple passes which
9167 consist of all read requests made in equidistant time
9168 intervals. This means instead of strictly reading read-ahead
9169 data in its physical order on disk we now try to find a
9170 middle ground between physical and access time order.
9171
9172 * /etc/os-release files gained a new BUILD_ID= field for usage
9173 on operating systems that provide continuous builds of OS
9174 images.
9175
9176 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
9177 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
9178 William Douglas, Zbigniew Jędrzejewski-Szmek
9179
35911459
LP
9180CHANGES WITH 199:
9181
9182 * systemd-python gained an API exposing libsystemd-daemon.
9183
9184 * The SMACK setup logic gained support for uploading CIPSO
9185 security policy.
9186
9187 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
9188 ReadOnlyDirectories= and InaccessibleDirectories= has
9189 changed. The private /tmp and /var/tmp directories are now
9190 shared by all processes of a service (which means
9191 ExecStartPre= may now leave data in /tmp that ExecStart= of
9192 the same service can still access). When a service is
9193 stopped its temporary directories are immediately deleted
a87197f5 9194 (normal clean-up with tmpfiles is still done in addition to
35911459
LP
9195 this though).
9196
9197 * By default, systemd will now set a couple of sysctl
9198 variables in the kernel: the safe sysrq options are turned
9199 on, IP route verification is turned on, and source routing
9200 disabled. The recently added hardlink and softlink
9201 protection of the kernel is turned on. These settings should
9202 be reasonably safe, and good defaults for all new systems.
9203
9204 * The predictable network naming logic may now be turned off
a87197f5 9205 with a new kernel command line switch: net.ifnames=0.
35911459
LP
9206
9207 * A new libsystemd-bus module has been added that implements a
9208 pretty complete D-Bus client library. For details see:
9209
56cadcb6 9210 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
35911459 9211
c20d8298 9212 * journald will now explicitly flush the journal files to disk
a87197f5
ZJS
9213 at the latest 5min after each write. The file will then also
9214 be marked offline until the next write. This should increase
9215 reliability in case of a crash. The synchronization delay
9216 can be configured via SyncIntervalSec= in journald.conf.
35911459
LP
9217
9218 * There's a new remote-fs-setup.target unit that can be used
9219 to pull in specific services when at least one remote file
9220 system is to be mounted.
9221
9222 * There are new targets timers.target and paths.target as
9223 canonical targets to pull user timer and path units in
9224 from. This complements sockets.target with a similar
9225 purpose for socket units.
9226
6a7d3d68
LP
9227 * libudev gained a new call udev_device_set_attribute_value()
9228 to set sysfs attributes of a device.
9229
a87197f5
ZJS
9230 * The udev daemon now sets the default number of worker
9231 processes executed in parallel based on the number of available
c20d8298 9232 CPUs instead of the amount of available RAM. This is supposed
ab06eef8 9233 to provide a more reliable default and limit a too aggressive
ce830873 9234 parallelism for setups with 1000s of devices connected.
c20d8298 9235
35911459
LP
9236 Contributions from: Auke Kok, Colin Walters, Cristian
9237 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
9238 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
9239 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
9240 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
9241 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
9242 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
9243 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
9244 Zbigniew Jędrzejewski-Szmek
9245
85d68397
LP
9246CHANGES WITH 198:
9247
9248 * Configuration of unit files may now be extended via drop-in
9249 files without having to edit/override the unit files
9250 themselves. More specifically, if the administrator wants to
9251 change one value for a service file foobar.service he can
9252 now do so by dropping in a configuration snippet into
ad88e758 9253 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
85d68397
LP
9254 will load all these snippets and apply them on top of the
9255 main unit configuration file, possibly extending or
9256 overriding its settings. Using these drop-in snippets is
40e21da8
KS
9257 generally nicer than the two earlier options for changing
9258 unit files locally: copying the files from
85d68397
LP
9259 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
9260 them there; or creating a new file in /etc/systemd/system/
9261 that incorporates the original one via ".include". Drop-in
9262 snippets into these .d/ directories can be placed in any
fd868975 9263 directory systemd looks for units in, and the usual
85d68397
LP
9264 overriding semantics between /usr/lib, /etc and /run apply
9265 for them too.
9266
9267 * Most unit file settings which take lists of items can now be
6aa8d43a 9268 reset by assigning the empty string to them. For example,
85d68397
LP
9269 normally, settings such as Environment=FOO=BAR append a new
9270 environment variable assignment to the environment block,
9271 each time they are used. By assigning Environment= the empty
9272 string the environment block can be reset to empty. This is
9273 particularly useful with the .d/*.conf drop-in snippets
156f7d09
KS
9274 mentioned above, since this adds the ability to reset list
9275 settings from vendor unit files via these drop-ins.
85d68397
LP
9276
9277 * systemctl gained a new "list-dependencies" command for
9278 listing the dependencies of a unit recursively.
9279
40e21da8 9280 * Inhibitors are now honored and listed by "systemctl
85d68397
LP
9281 suspend", "systemctl poweroff" (and similar) too, not only
9282 GNOME. These commands will also list active sessions by
9283 other users.
9284
9285 * Resource limits (as exposed by the various control group
9286 controllers) can now be controlled dynamically at runtime
9287 for all units. More specifically, you can now use a command
9288 like "systemctl set-cgroup-attr foobar.service cpu.shares
9289 2000" to alter the CPU shares a specific service gets. These
6aa8d43a 9290 settings are stored persistently on disk, and thus allow the
85d68397
LP
9291 administrator to easily adjust the resource usage of
9292 services with a few simple commands. This dynamic resource
6aa8d43a 9293 management logic is also available to other programs via the
85d68397
LP
9294 bus. Almost any kernel cgroup attribute and controller is
9295 supported.
9296
9297 * systemd-vconsole-setup will now copy all font settings to
6aa8d43a
LP
9298 all allocated VTs, where it previously applied them only to
9299 the foreground VT.
85d68397
LP
9300
9301 * libsystemd-login gained the new sd_session_get_tty() API
9302 call.
9303
6aa8d43a
LP
9304 * This release drops support for a few legacy or
9305 distribution-specific LSB facility names when parsing init
9306 scripts: $x-display-manager, $mail-transfer-agent,
85d68397
LP
9307 $mail-transport-agent, $mail-transfer-agent, $smtp,
9308 $null. Also, the mail-transfer-agent.target unit backing
9309 this has been removed. Distributions which want to retain
6aa8d43a
LP
9310 compatibility with this should carry the burden for
9311 supporting this themselves and patch support for these back
9312 in, if they really need to. Also, the facilities $syslog and
9313 $local_fs are now ignored, since systemd does not support
9314 early-boot LSB init scripts anymore, and these facilities
9315 are implied anyway for normal services. syslog.target has
9316 also been removed.
85d68397 9317
40e21da8 9318 * There are new bus calls on PID1's Manager object for
6aa8d43a 9319 cancelling jobs, and removing snapshot units. Previously,
85d68397
LP
9320 both calls were only available on the Job and Snapshot
9321 objects themselves.
9322
9323 * systemd-journal-gatewayd gained SSL support.
9324
9325 * The various "environment" files, such as /etc/locale.conf
9326 now support continuation lines with a backslash ("\") as
499b604b 9327 last character in the line, similarly in style (but different)
85d68397
LP
9328 to how this is supported in shells.
9329
9330 * For normal user processes the _SYSTEMD_USER_UNIT= field is
9331 now implicitly appended to every log entry logged. systemctl
9332 has been updated to filter by this field when operating on a
9333 user systemd instance.
9334
9335 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
9336 CAP_AUDIT_CONTROL capabilities to the capabilities set for
9337 the container. This makes it easier to boot unmodified
9338 Fedora systems in a container, which however still requires
9339 audit=0 to be passed on the kernel command line. Auditing in
9340 kernel and userspace is unfortunately still too broken in
9341 context of containers, hence we recommend compiling it out
9342 of the kernel or using audit=0. Hopefully this will be fixed
9343 one day for good in the kernel.
9344
9345 * nspawn gained the new --bind= and --bind-ro= parameters to
9346 bind mount specific directories from the host into the
9347 container.
9348
40e21da8 9349 * nspawn will now mount its own devpts file system instance
6aa8d43a 9350 into the container, in order not to leak pty devices from
85d68397
LP
9351 the host into the container.
9352
9353 * systemd will now read the firmware boot time performance
6aa8d43a
LP
9354 information from the EFI variables, if the used boot loader
9355 supports this, and takes it into account for boot performance
9356 analysis via "systemd-analyze". This is currently supported
9357 only in conjunction with Gummiboot, but could be supported
9358 by other boot loaders too. For details see:
85d68397 9359
56cadcb6 9360 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
85d68397
LP
9361
9362 * A new generator has been added that automatically mounts the
9363 EFI System Partition (ESP) to /boot, if that directory
6aa8d43a
LP
9364 exists, is empty, and no other file system has been
9365 configured to be mounted there.
85d68397
LP
9366
9367 * logind will now send out PrepareForSleep(false) out
9368 unconditionally, after coming back from suspend. This may be
9369 used by applications as asynchronous notification for
9370 system resume events.
9371
9372 * "systemctl unlock-sessions" has been added, that allows
9373 unlocking the screens of all user sessions at once, similar
499b604b 9374 to how "systemctl lock-sessions" already locked all users
40e21da8 9375 sessions. This is backed by a new D-Bus call UnlockSessions().
85d68397
LP
9376
9377 * "loginctl seat-status" will now show the master device of a
9378 seat. (i.e. the device of a seat that needs to be around for
9379 the seat to be considered available, usually the graphics
9380 card).
9381
9382 * tmpfiles gained a new "X" line type, that allows
9383 configuration of files and directories (with wildcards) that
9384 shall be excluded from automatic cleanup ("aging").
9385
bf933560
KS
9386 * udev default rules set the device node permissions now only
9387 at "add" events, and do not change them any longer with a
9388 later "change" event.
85d68397
LP
9389
9390 * The log messages for lid events and power/sleep keypresses
9391 now carry a message ID.
9392
9393 * We now have a substantially larger unit test suite, but this
9394 continues to be work in progress.
9395
9396 * udevadm hwdb gained a new --root= parameter to change the
9397 root directory to operate relative to.
9398
40e21da8
KS
9399 * logind will now issue a background sync() request to the kernel
9400 early at shutdown, so that dirty buffers are flushed to disk early
85d68397
LP
9401 instead of at the last moment, in order to optimize shutdown
9402 times a little.
9403
9404 * A new bootctl tool has been added that is an interface for
9405 certain boot loader operations. This is currently a preview
9406 and is likely to be extended into a small mechanism daemon
9407 like timedated, localed, hostnamed, and can be used by
9408 graphical UIs to enumerate available boot options, and
9409 request boot into firmware operations.
9410
9411 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
9412 the rest of the package. It also has been updated to work
9413 correctly in initrds.
9414
d35f51ea
ZJS
9415 * polkit previously has been runtime optional, and is now also
9416 compile time optional via a configure switch.
85d68397
LP
9417
9418 * systemd-analyze has been reimplemented in C. Also "systemctl
9419 dot" has moved into systemd-analyze.
9420
9421 * "systemctl status" with no further parameters will now print
9422 the status of all active or failed units.
9423
9424 * Operations such as "systemctl start" can now be executed
9425 with a new mode "--irreversible" which may be used to queue
9426 operations that cannot accidentally be reversed by a later
6aa8d43a 9427 job queuing. This is by default used to make shutdown
85d68397
LP
9428 requests more robust.
9429
9430 * The Python API of systemd now gained a new module for
9431 reading journal files.
9432
9433 * A new tool kernel-install has been added that can install
9434 kernel images according to the Boot Loader Specification:
9435
56cadcb6 9436 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
85d68397
LP
9437
9438 * Boot time console output has been improved to provide
6aa8d43a 9439 animated boot time output for hanging jobs.
85d68397
LP
9440
9441 * A new tool systemd-activate has been added which can be used
9442 to test socket activation with, directly from the command
9443 line. This should make it much easier to test and debug
9444 socket activation in daemons.
9445
9446 * journalctl gained a new "--reverse" (or -r) option to show
9447 journal output in reverse order (i.e. newest line first).
9448
43447fb7
LP
9449 * journalctl gained a new "--pager-end" (or -e) option to jump
9450 to immediately jump to the end of the journal in the
9451 pager. This is only supported in conjunction with "less".
9452
85d68397 9453 * journalctl gained a new "--user-unit=" option, that works
499b604b 9454 similarly to "--unit=" but filters for user units rather than
85d68397
LP
9455 system units.
9456
9457 * A number of unit files to ease adoption of systemd in
9458 initrds has been added. This moves some minimal logic from
9459 the various initrd implementations into systemd proper.
9460
9461 * The journal files are now owned by a new group
9462 "systemd-journal", which exists specifically to allow access
9463 to the journal, and nothing else. Previously, we used the
6aa8d43a 9464 "adm" group for that, which however possibly covers more
85d68397
LP
9465 than just journal/log file access. This new group is now
9466 already used by systemd-journal-gatewayd to ensure this
9467 daemon gets access to the journal files and as little else
9468 as possible. Note that "make install" will also set FS ACLs
9469 up for /var/log/journal to give "adm" and "wheel" read
9470 access to it, in addition to "systemd-journal" which owns
9471 the journal files. We recommend that packaging scripts also
6aa8d43a 9472 add read access to "adm" + "wheel" to /var/log/journal, and
85d68397
LP
9473 all existing/future journal files. To normal users and
9474 administrators little changes, however packagers need to
9475 ensure to create the "systemd-journal" system group at
9476 package installation time.
9477
9478 * The systemd-journal-gatewayd now runs as unprivileged user
9479 systemd-journal-gateway:systemd-journal-gateway. Packaging
9480 scripts need to create these system user/group at
9481 installation time.
9482
9483 * timedated now exposes a new boolean property CanNTP that
9484 indicates whether a local NTP service is available or not.
9485
9486 * systemd-detect-virt will now also detect xen PVs
9487
40e21da8
KS
9488 * The pstore file system is now mounted by default, if it is
9489 available.
85d68397 9490
1aed4590
LP
9491 * In addition to the SELinux and IMA policies we will now also
9492 load SMACK policies at early boot.
9493
85d68397
LP
9494 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
9495 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
9496 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
9497 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
9498 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
9499 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
9500 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
9501 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
9502 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
9503 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
9504 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
9505 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
9506 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
9507 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
9508
8ad26859
LP
9509CHANGES WITH 197:
9510
9511 * Timer units now support calendar time events in addition to
9512 monotonic time events. That means you can now trigger a unit
9513 based on a calendar time specification such as "Thu,Fri
9514 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
9515 or fifth day of any month of the year 2013, given that it is
9516 a thursday or friday. This brings timer event support
9517 considerably closer to cron's capabilities. For details on
9518 the supported calendar time specification language see
9519 systemd.time(7).
9520
9521 * udev now supports a number of different naming policies for
9522 network interfaces for predictable names, and a combination
9523 of these policies is now the default. Please see this wiki
9524 document for details:
9525
56cadcb6 9526 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
8ad26859
LP
9527
9528 * Auke Kok's bootchart implementation has been added to the
d28315e4
JE
9529 systemd tree. It is an optional component that can graph the
9530 boot in quite some detail. It is one of the best bootchart
8ad26859
LP
9531 implementations around and minimal in its code and
9532 dependencies.
9533
9534 * nss-myhostname has been integrated into the systemd source
9535 tree. nss-myhostname guarantees that the local hostname
9536 always stays resolvable via NSS. It has been a weak
9537 requirement of systemd-hostnamed since a long time, and
9538 since its code is actually trivial we decided to just
9539 include it in systemd's source tree. It can be turned off
9540 with a configure switch.
9541
9542 * The read-ahead logic is now capable of properly detecting
9543 whether a btrfs file system is on SSD or rotating media, in
9544 order to optimize the read-ahead scheme. Previously, it was
9545 only capable of detecting this on traditional file systems
9546 such as ext4.
9547
9548 * In udev, additional device properties are now read from the
9549 IAB in addition to the OUI database. Also, Bluetooth company
9550 identities are attached to the devices as well.
9551
9552 * In service files %U may be used as specifier that is
9553 replaced by the configured user name of the service.
9554
9555 * nspawn may now be invoked without a controlling TTY. This
9556 makes it suitable for invocation as its own service. This
9557 may be used to set up a simple containerized server system
9558 using only core OS tools.
9559
9560 * systemd and nspawn can now accept socket file descriptors
9561 when they are started for socket activation. This enables
9562 implementation of socket activated nspawn
9563 containers. i.e. think about autospawning an entire OS image
9564 when the first SSH or HTTP connection is received. We expect
9565 that similar functionality will also be added to libvirt-lxc
9566 eventually.
9567
9568 * journalctl will now suppress ANSI color codes when
9569 presenting log data.
9570
9571 * systemctl will no longer show control group information for
ce830873 9572 a unit if the control group is empty anyway.
8ad26859
LP
9573
9574 * logind can now automatically suspend/hibernate/shutdown the
9575 system on idle.
9576
9577 * /etc/machine-info and hostnamed now also expose the chassis
9578 type of the system. This can be used to determine whether
9579 the local system is a laptop, desktop, handset or
9580 tablet. This information may either be configured by the
9581 user/vendor or is automatically determined from ACPI and DMI
9582 information if possible.
9583
d35f51ea
ZJS
9584 * A number of polkit actions are now bound together with "imply"
9585 rules. This should simplify creating UIs because many actions
9586 will now authenticate similar ones as well.
8ad26859
LP
9587
9588 * Unit files learnt a new condition ConditionACPower= which
9589 may be used to conditionalize a unit depending on whether an
9590 AC power source is connected or not, of whether the system
9591 is running on battery power.
9592
9593 * systemctl gained a new "is-failed" verb that may be used in
9594 shell scripts and suchlike to check whether a specific unit
9595 is in the "failed" state.
9596
9597 * The EnvironmentFile= setting in unit files now supports file
9598 globbing, and can hence be used to easily read a number of
9599 environment files at once.
9600
9601 * systemd will no longer detect and recognize specific
9602 distributions. All distribution-specific #ifdeffery has been
9603 removed, systemd is now fully generic and
9604 distribution-agnostic. Effectively, not too much is lost as
9605 a lot of the code is still accessible via explicit configure
9606 switches. However, support for some distribution specific
9607 legacy configuration file formats has been dropped. We
9608 recommend distributions to simply adopt the configuration
9609 files everybody else uses now and convert the old
9610 configuration from packaging scripts. Most distributions
9611 already did that. If that's not possible or desirable,
9612 distributions are welcome to forward port the specific
9613 pieces of code locally from the git history.
9614
9615 * When logging a message about a unit systemd will now always
9616 log the unit name in the message meta data.
9617
9618 * localectl will now also discover system locale data that is
9619 not stored in locale archives, but directly unpacked.
9620
9621 * logind will no longer unconditionally use framebuffer
9622 devices as seat masters, i.e. as devices that are required
9623 to be existing before a seat is considered preset. Instead,
9624 it will now look for all devices that are tagged as
b938cb90
JE
9625 "seat-master" in udev. By default, framebuffer devices will
9626 be marked as such, but depending on local systems, other
8ad26859
LP
9627 devices might be marked as well. This may be used to
9628 integrate graphics cards using closed source drivers (such
9629 as NVidia ones) more nicely into logind. Note however, that
9630 we recommend using the open source NVidia drivers instead,
9631 and no udev rules for the closed-source drivers will be
9632 shipped from us upstream.
9633
9634 Contributions from: Adam Williamson, Alessandro Crismani, Auke
9635 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
9636 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
9637 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
9638 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
9639 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
9640 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
9641 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
9642 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
9643 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
9644 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
9645 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
9646 Jędrzejewski-Szmek
9647
0428ddb7
LP
9648CHANGES WITH 196:
9649
9650 * udev gained support for loading additional device properties
9651 from an indexed database that is keyed by vendor/product IDs
9652 and similar device identifiers. For the beginning this
9653 "hwdb" is populated with data from the well-known PCI and
9654 USB database, but also includes PNP, ACPI and OID data. In
9655 the longer run this indexed database shall grow into
9656 becoming the one central database for non-essential
9657 userspace device metadata. Previously, data from the PCI/USB
96ec33c0 9658 database was only attached to select devices, since the
0428ddb7 9659 lookup was a relatively expensive operation due to O(n) time
96ec33c0
LP
9660 complexity (with n being the number of entries in the
9661 database). Since this is now O(1), we decided to add in this
9662 data for all devices where this is available, by
0428ddb7
LP
9663 default. Note that the indexed database needs to be rebuilt
9664 when new data files are installed. To achieve this you need
9665 to update your packaging scripts to invoke "udevadm hwdb
9666 --update" after installation of hwdb data files. For
9667 RPM-based distributions we introduced the new
9668 %udev_hwdb_update macro for this purpose.
9669
9670 * The Journal gained support for the "Message Catalog", an
9671 indexed database to link up additional information with
9672 journal entries. For further details please check:
9673
56cadcb6 9674 https://www.freedesktop.org/wiki/Software/systemd/catalog
0428ddb7
LP
9675
9676 The indexed message catalog database also needs to be
9677 rebuilt after installation of message catalog files. Use
9678 "journalctl --update-catalog" for this. For RPM-based
9679 distributions we introduced the %journal_catalog_update
9680 macro for this purpose.
9681
9682 * The Python Journal bindings gained support for the standard
9683 Python logging framework.
9684
9685 * The Journal API gained new functions for checking whether
9686 the underlying file system of a journal file is capable of
9687 properly reporting file change notifications, or whether
9688 applications that want to reflect journal changes "live"
ab06eef8 9689 need to recheck journal files continuously in appropriate
0428ddb7
LP
9690 time intervals.
9691
9692 * It is now possible to set the "age" field for tmpfiles
9693 entries to 0, indicating that files matching this entry
9694 shall always be removed when the directories are cleaned up.
9695
9696 * coredumpctl gained a new "gdb" verb which invokes gdb
9697 right-away on the selected coredump.
9698
9699 * There's now support for "hybrid sleep" on kernels that
9700 support this, in addition to "suspend" and "hibernate". Use
9701 "systemctl hybrid-sleep" to make use of this.
9702
9703 * logind's HandleSuspendKey= setting (and related settings)
9704 now gained support for a new "lock" setting to simply
9705 request the screen lock on all local sessions, instead of
9706 actually executing a suspend or hibernation.
9707
9708 * systemd will now mount the EFI variables file system by
9709 default.
9710
9711 * Socket units now gained support for configuration of the
9712 SMACK security label.
9713
9714 * timedatectl will now output the time of the last and next
9715 daylight saving change.
9716
9717 * We dropped support for various legacy and distro-specific
9718 concepts, such as insserv, early-boot SysV services
9719 (i.e. those for non-standard runlevels such as 'b' or 'S')
9720 or ArchLinux /etc/rc.conf support. We recommend the
9721 distributions who still need support this to either continue
9722 to maintain the necessary patches downstream, or find a
9723 different solution. (Talk to us if you have questions!)
9724
d35f51ea
ZJS
9725 * Various systemd components will now bypass polkit checks for
9726 root and otherwise handle properly if polkit is not found to
9727 be around. This should fix most issues for polkit-less
9728 systems. Quite frankly this should have been this way since
9729 day one. It is absolutely our intention to make systemd work
9730 fine on polkit-less systems, and we consider it a bug if
9731 something does not work as it should if polkit is not around.
0428ddb7
LP
9732
9733 * For embedded systems it is now possible to build udev and
9734 systemd without blkid and/or kmod support.
9735
9736 * "systemctl switch-root" is now capable of switching root
9737 more than once. I.e. in addition to transitions from the
9738 initrd to the host OS it is now possible to transition to
9739 further OS images from the host. This is useful to implement
9740 offline updating tools.
9741
9742 * Various other additions have been made to the RPM macros
9743 shipped with systemd. Use %udev_rules_update() after
9744 installing new udev rules files. %_udevhwdbdir,
9745 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
9746 %_sysctldir are now available which resolve to the right
9747 directories for packages to place various data files in.
9748
9749 * journalctl gained the new --full switch (in addition to
9750 --all, to disable ellipsation for long messages.
9751
9752 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
9753 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
9754 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
9755 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
9756 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
9757 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
9758 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
9759 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
9760 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
9761
139ee8cc
LP
9762CHANGES WITH 195:
9763
6827101a 9764 * journalctl gained new --since= and --until= switches to
139ee8cc
LP
9765 filter by time. It also now supports nice filtering for
9766 units via --unit=/-u.
9767
6827101a 9768 * Type=oneshot services may use ExecReload= and do the
139ee8cc
LP
9769 right thing.
9770
9771 * The journal daemon now supports time-based rotation and
9772 vacuuming, in addition to the usual disk-space based
9773 rotation.
9774
9775 * The journal will now index the available field values for
9776 each field name. This enables clients to show pretty drop
9777 downs of available match values when filtering. The bash
9778 completion of journalctl has been updated
9779 accordingly. journalctl gained a new switch -F to list all
9780 values a certain field takes in the journal database.
9781
9782 * More service events are now written as structured messages
9783 to the journal, and made recognizable via message IDs.
9784
9785 * The timedated, localed and hostnamed mini-services which
9786 previously only provided support for changing time, locale
9787 and hostname settings from graphical DEs such as GNOME now
9788 also have a minimal (but very useful) text-based client
9789 utility each. This is probably the nicest way to changing
9790 these settings from the command line now, especially since
9791 it lists available options and is fully integrated with bash
9792 completion.
9793
9794 * There's now a new tool "systemd-coredumpctl" to list and
9795 extract coredumps from the journal.
9796
9797 * We now install a README each in /var/log/ and
9798 /etc/rc.d/init.d explaining where the system logs and init
9799 scripts went. This hopefully should help folks who go to
9800 that dirs and look into the otherwise now empty void and
9801 scratch their heads.
9802
9803 * When user-services are invoked (by systemd --user) the
9804 $MANAGERPID env var is set to the PID of systemd.
9805
9806 * SIGRTMIN+24 when sent to a --user instance will now result
9807 in immediate termination of systemd.
9808
9809 * gatewayd received numerous feature additions such as a
9810 "follow" mode, for live syncing and filtering.
9811
9812 * browse.html now allows filtering and showing detailed
9813 information on specific entries. Keyboard navigation and
9814 mouse screen support has been added.
9815
9816 * gatewayd/journalctl now supports HTML5/JSON
9817 Server-Sent-Events as output.
9818
1cb88f2c 9819 * The SysV init script compatibility logic will now
139ee8cc
LP
9820 heuristically determine whether a script supports the
9821 "reload" verb, and only then make this available as
9822 "systemctl reload".
9823
15f47220 9824 * "systemctl status --follow" has been removed, use "journalctl
139ee8cc
LP
9825 -u" instead.
9826
9827 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
9828 have been removed since they are hardly useful to be
9829 configured.
9830
9831 * And I'd like to take the opportunity to specifically mention
9832 Zbigniew for his great contributions. Zbigniew, you rock!
9833
9834 Contributions from: Andrew Eikum, Christian Hesse, Colin
9835 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4d92e078
LP
9836 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
9837 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
9838 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
9839 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
9840 Jędrzejewski-Szmek, Сковорода Никита Андреевич
139ee8cc 9841
f9b55720
LP
9842CHANGES WITH 194:
9843
9844 * If /etc/vconsole.conf is non-existent or empty we will no
9845 longer load any console font or key map at boot by
9846 default. Instead the kernel defaults will be left
9847 intact. This is definitely the right thing to do, as no
9848 configuration should mean no configuration, and hard-coding
9849 font names that are different on all archs is probably a bad
9850 idea. Also, the kernel default key map and font should be
9851 good enough for most cases anyway, and mostly identical to
9852 the userspace fonts/key maps we previously overloaded them
9853 with. If distributions want to continue to default to a
9854 non-kernel font or key map they should ship a default
9855 /etc/vconsole.conf with the appropriate contents.
9856
9857 Contributions from: Colin Walters, Daniel J Walsh, Dave
9858 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
9859 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9860
597c52cf
LP
9861CHANGES WITH 193:
9862
9863 * journalctl gained a new --cursor= switch to show entries
9864 starting from the specified location in the journal.
9865
9866 * We now enforce a size limit on journal entry fields exported
9867 with "-o json" in journalctl. Fields larger than 4K will be
9868 assigned null. This can be turned off with --all.
9869
9870 * An (optional) journal gateway daemon is now available as
9871 "systemd-journal-gatewayd.service". This service provides
9872 access to the journal via HTTP and JSON. This functionality
9873 will be used to implement live log synchronization in both
9874 pull and push modes, but has various other users too, such
9875 as easy log access for debugging of embedded devices. Right
9876 now it is already useful to retrieve the journal via HTTP:
9877
9878 # systemctl start systemd-journal-gatewayd.service
9879 # wget http://localhost:19531/entries
9880
9881 This will download the journal contents in a
9882 /var/log/messages compatible format. The same as JSON:
9883
9884 # curl -H"Accept: application/json" http://localhost:19531/entries
9885
9886 This service is also accessible via a web browser where a
9887 single static HTML5 app is served that uses the JSON logic
9888 to enable the user to do some basic browsing of the
9889 journal. This will be extended later on. Here's an example
9890 screenshot of this app in its current state:
9891
9892 http://0pointer.de/public/journal-gatewayd
9893
9894 Contributions from: Kay Sievers, Lennart Poettering, Robert
9895 Milasan, Tom Gundersen
9896
075d4ecb
LP
9897CHANGES WITH 192:
9898
9899 * The bash completion logic is now available for journalctl
9900 too.
9901
d28315e4 9902 * We do not mount the "cpuset" controller anymore together with
075d4ecb
LP
9903 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
9904 started if no parameters are assigned to it. "cpuset" hence
61233823 9905 broke code that assumed it could create "cpu" groups and
075d4ecb
LP
9906 just start them.
9907
9908 * journalctl -f will now subscribe to terminal size changes,
9909 and line break accordingly.
9910
597c52cf
LP
9911 Contributions from: Dave Reisner, Kay Sievers, Lennart
9912 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
075d4ecb 9913
b6a86739
LP
9914CHANGES WITH 191:
9915
9916 * nspawn will now create a symlink /etc/localtime in the
9917 container environment, copying the host's timezone
9918 setting. Previously this has been done via a bind mount, but
9919 since symlinks cannot be bind mounted this has now been
9920 changed to create/update the appropriate symlink.
9921
9922 * journalctl -n's line number argument is now optional, and
9923 will default to 10 if omitted.
9924
9925 * journald will now log the maximum size the journal files may
9926 take up on disk. This is particularly useful if the default
9927 built-in logic of determining this parameter from the file
9928 system size is used. Use "systemctl status
6563b535 9929 systemd-journald.service" to see this information.
b6a86739
LP
9930
9931 * The multi-seat X wrapper tool has been stripped down. As X
9932 is now capable of enumerating graphics devices via udev in a
9933 seat-aware way the wrapper is not strictly necessary
9934 anymore. A stripped down temporary stop-gap is still shipped
9935 until the upstream display managers have been updated to
9936 fully support the new X logic. Expect this wrapper to be
6563b535 9937 removed entirely in one of the next releases.
b6a86739
LP
9938
9939 * HandleSleepKey= in logind.conf has been split up into
9940 HandleSuspendKey= and HandleHibernateKey=. The old setting
6563b535 9941 is not available anymore. X11 and the kernel are
45afd519 9942 distinguishing between these keys and we should too. This
b6a86739
LP
9943 also means the inhibition lock for these keys has been split
9944 into two.
9945
597c52cf
LP
9946 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
9947 Poettering, Lukas Nykryn, Václav Pavlín
b6a86739 9948
0c11f949
LP
9949CHANGES WITH 190:
9950
d28315e4 9951 * Whenever a unit changes state we will now log this to the
0c11f949
LP
9952 journal and show along the unit's own log output in
9953 "systemctl status".
9954
9955 * ConditionPathIsMountPoint= can now properly detect bind
9956 mount points too. (Previously, a bind mount of one file
8d0256b7 9957 system to another place in the same file system could not be
0c11f949
LP
9958 detected as mount, since they shared struct stat's st_dev
9959 field.)
9960
9961 * We will now mount the cgroup controllers cpu, cpuacct,
9962 cpuset and the controllers net_cls, net_prio together by
9963 default.
9964
9965 * nspawn containers will now have a virtualized boot
9966 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
9967 over with a randomized ID at container initialization). This
9968 has the effect of making "journalctl -b" do the right thing
9969 in a container.
9970
9971 * The JSON output journal serialization has been updated not
9972 to generate "endless" list objects anymore, but rather one
9973 JSON object per line. This is more in line how most JSON
9974 parsers expect JSON objects. The new output mode
9975 "json-pretty" has been added to provide similar output, but
9976 neatly aligned for readability by humans.
9977
9978 * We dropped all explicit sync() invocations in the shutdown
9979 code. The kernel does this implicitly anyway in the kernel
9980 reboot() syscall. halt(8)'s -n option is now a compatibility
9981 no-op.
9982
9983 * We now support virtualized reboot() in containers, as
9984 supported by newer kernels. We will fall back to exit() if
9985 CAP_SYS_REBOOT is not available to the container. Also,
9986 nspawn makes use of this now and will actually reboot the
9987 container if the containerized OS asks for that.
9988
9989 * journalctl will only show local log output by default
9990 now. Use --merge (-m) to show remote log output, too.
9991
9992 * libsystemd-journal gained the new sd_journal_get_usage()
9993 call to determine the current disk usage of all journal
9994 files. This is exposed in the new "journalctl --disk-usage"
9995 command.
9996
9997 * journald gained a new configuration setting SplitMode= in
9998 journald.conf which may be used to control how user journals
9999 are split off. See journald.conf(5) for details.
10000
10001 * A new condition type ConditionFileNotEmpty= has been added.
10002
10003 * tmpfiles' "w" lines now support file globbing, to write
10004 multiple files at once.
10005
10006 * We added Python bindings for the journal submission
10007 APIs. More Python APIs for a number of selected APIs will
10008 likely follow. Note that we intend to add native bindings
10009 only for the Python language, as we consider it common
10010 enough to deserve bindings shipped within systemd. There are
10011 various projects outside of systemd that provide bindings
10012 for languages such as PHP or Lua.
10013
a98d5d64
LP
10014 * Many conditions will now resolve specifiers such as %i. In
10015 addition, PathChanged= and related directives of .path units
10016 now support specifiers as well.
0c11f949
LP
10017
10018 * There's now a new RPM macro definition for the system preset
10019 dir: %_presetdir.
10020
d28315e4 10021 * journald will now warn if it ca not forward a message to the
dca348bc 10022 syslog daemon because its socket is full.
0c11f949
LP
10023
10024 * timedated will no longer write or process /etc/timezone,
10025 except on Debian. As we do not support late mounted /usr
10026 anymore /etc/localtime always being a symlink is now safe,
10027 and hence the information in /etc/timezone is not necessary
10028 anymore.
10029
aaccc32c 10030 * logind will now always reserve one VT for a text getty (VT6
0c11f949
LP
10031 by default). Previously if more than 6 X sessions where
10032 started they took up all the VTs with auto-spawned gettys,
10033 so that no text gettys were available anymore.
10034
10035 * udev will now automatically inform the btrfs kernel logic
10036 about btrfs RAID components showing up. This should make
10037 simple hotplug based btrfs RAID assembly work.
10038
10039 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
10040 (but not for its children which will stay at the kernel
10041 default). This should allow setups with a lot more listening
10042 sockets.
10043
10044 * systemd will now always pass the configured timezone to the
10045 kernel at boot. timedated will do the same when the timezone
10046 is changed.
10047
10048 * logind's inhibition logic has been updated. By default,
10049 logind will now handle the lid switch, the power and sleep
10050 keys all the time, even in graphical sessions. If DEs want
10051 to handle these events on their own they should take the new
10052 handle-power-key, handle-sleep-key and handle-lid-switch
f131770b 10053 inhibitors during their runtime. A simple way to achieve
0c11f949
LP
10054 that is to invoke the DE wrapped in an invocation of:
10055
10056 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
10057
10058 * Access to unit operations is now checked via SELinux taking
10059 the unit file label and client process label into account.
10060
aad803af
LP
10061 * systemd will now notify the administrator in the journal
10062 when he over-mounts a non-empty directory.
10063
10064 * There are new specifiers that are resolved in unit files,
38b38500 10065 for the hostname (%H), the machine ID (%m) and the boot ID
aad803af
LP
10066 (%b).
10067
b6a86739 10068 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
0c11f949
LP
10069 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
10070 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
10071 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
10072 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
10073 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
10074 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
10075
38a60d71
LP
10076CHANGES WITH 189:
10077
10078 * Support for reading structured kernel messages from
10079 /dev/kmsg has now been added and is enabled by default.
10080
10081 * Support for reading kernel messages from /proc/kmsg has now
10082 been removed. If you want kernel messages in the journal
10083 make sure to run a recent kernel (>= 3.5) that supports
10084 reading structured messages from /dev/kmsg (see
10085 above). /proc/kmsg is now exclusive property of classic
10086 syslog daemons again.
10087
10088 * The libudev API gained the new
10089 udev_device_new_from_device_id() call.
10090
10091 * The logic for file system namespace (ReadOnlyDirectory=,
10092 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
10093 require pivot_root() anymore. This means fewer temporary
10094 directories are created below /tmp for this feature.
10095
10096 * nspawn containers will now see and receive all submounts
10097 made on the host OS below the root file system of the
10098 container.
10099
10100 * Forward Secure Sealing is now supported for Journal files,
10101 which provide cryptographical sealing of journal files so
10102 that attackers cannot alter log history anymore without this
10103 being detectable. Lennart will soon post a blog story about
10104 this explaining it in more detail.
10105
10106 * There are two new service settings RestartPreventExitStatus=
10107 and SuccessExitStatus= which allow configuration of exit
10108 status (exit code or signal) which will be excepted from the
10109 restart logic, resp. consider successful.
10110
10111 * journalctl gained the new --verify switch that can be used
10112 to check the integrity of the structure of journal files and
10113 (if Forward Secure Sealing is enabled) the contents of
10114 journal files.
10115
10116 * nspawn containers will now be run with /dev/stdin, /dev/fd/
10117 and similar symlinks pre-created. This makes running shells
10118 as container init process a lot more fun.
10119
10120 * The fstab support can now handle PARTUUID= and PARTLABEL=
10121 entries.
10122
10123 * A new ConditionHost= condition has been added to match
10124 against the hostname (with globs) and machine ID. This is
10125 useful for clusters where a single OS image is used to
10126 provision a large number of hosts which shall run slightly
10127 different sets of services.
10128
10129 * Services which hit the restart limit will now be placed in a
10130 failure state.
10131
b6a86739 10132 Contributions from: Bertram Poettering, Dave Reisner, Huang
38a60d71
LP
10133 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
10134 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
10135
c269cec3
LP
10136CHANGES WITH 188:
10137
10138 * When running in --user mode systemd will now become a
10139 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
10140 tree a lot more organized.
10141
10142 * A new PartOf= unit dependency type has been introduced that
10143 may be used to group services in a natural way.
10144
10145 * "systemctl enable" may now be used to enable instances of
10146 services.
10147
10148 * journalctl now prints error log levels in red, and
10149 warning/notice log levels in bright white. It also supports
10150 filtering by log level now.
10151
10152 * cgtop gained a new -n switch (similar to top), to configure
10153 the maximum number of iterations to run for. It also gained
10154 -b, to run in batch mode (accepting no input).
10155
ab06eef8 10156 * The suffix ".service" may now be omitted on most systemctl
c269cec3
LP
10157 command lines involving service unit names.
10158
10159 * There's a new bus call in logind to lock all sessions, as
10160 well as a loginctl verb for it "lock-sessions".
10161
10162 * libsystemd-logind.so gained a new call sd_journal_perror()
10163 that works similar to libc perror() but logs to the journal
10164 and encodes structured information about the error number.
10165
10166 * /etc/crypttab entries now understand the new keyfile-size=
10167 option.
10168
10169 * shutdown(8) now can send a (configurable) wall message when
10170 a shutdown is cancelled.
10171
10172 * The mount propagation mode for the root file system will now
10173 default to "shared", which is useful to make containers work
10174 nicely out-of-the-box so that they receive new mounts from
10175 the host. This can be undone locally by running "mount
10176 --make-rprivate /" if needed.
10177
10178 * The prefdm.service file has been removed. Distributions
10179 should maintain this unit downstream if they intend to keep
10180 it around. However, we recommend writing normal unit files
10181 for display managers instead.
10182
10183 * Since systemd is a crucial part of the OS we will now
10184 default to a number of compiler switches that improve
10185 security (hardening) such as read-only relocations, stack
10186 protection, and suchlike.
10187
10188 * The TimeoutSec= setting for services is now split into
10189 TimeoutStartSec= and TimeoutStopSec= to allow configuration
10190 of individual time outs for the start and the stop phase of
10191 the service.
10192
10193 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
10194 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
10195 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
10196 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
10197 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
10198 Gundersen, Zbigniew Jędrzejewski-Szmek
10199
c4f1b862
LP
10200CHANGES WITH 187:
10201
10202 * The journal and id128 C APIs are now fully documented as man
10203 pages.
10204
10205 * Extra safety checks have been added when transitioning from
10206 the initial RAM disk to the main system to avoid accidental
10207 data loss.
10208
c269cec3 10209 * /etc/crypttab entries now understand the new keyfile-offset=
c4f1b862
LP
10210 option.
10211
10212 * systemctl -t can now be used to filter by unit load state.
10213
10214 * The journal C API gained the new sd_journal_wait() call to
10215 make writing synchronous journal clients easier.
10216
10217 * journalctl gained the new -D switch to show journals from a
10218 specific directory.
10219
10220 * journalctl now displays a special marker between log
10221 messages of two different boots.
10222
10223 * The journal is now explicitly flushed to /var via a service
10224 systemd-journal-flush.service, rather than implicitly simply
10225 by seeing /var/log/journal to be writable.
10226
10227 * journalctl (and the journal C APIs) can now match for much
10228 more complex expressions, with alternatives and
10229 disjunctions.
10230
10231 * When transitioning from the initial RAM disk to the main
10232 system we will now kill all processes in a killing spree to
10233 ensure no processes stay around by accident.
10234
10235 * Three new specifiers may be used in unit files: %u, %h, %s
10236 resolve to the user name, user home directory resp. user
10237 shell. This is useful for running systemd user instances.
10238
10239 * We now automatically rotate journal files if their data
10240 object hash table gets a fill level > 75%. We also size the
10241 hash table based on the configured maximum file size. This
10242 together should lower hash collisions drastically and thus
10243 speed things up a bit.
10244
10245 * journalctl gained the new "--header" switch to introspect
10246 header data of journal files.
10247
10248 * A new setting SystemCallFilters= has been added to services
10249 which may be used to apply blacklists or whitelists to
10250 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
10251
10252 * nspawn gained a new --link-journal= switch (and quicker: -j)
10253 to link the container journal with the host. This makes it
10254 very easy to centralize log viewing on the host for all
10255 guests while still keeping the journal files separated.
10256
10257 * Many bugfixes and optimizations
10258
10259 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
10260 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
10261 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
10262 Jędrzejewski-Szmek
10263
b5b4c94a
LP
10264CHANGES WITH 186:
10265
10266 * Several tools now understand kernel command line arguments,
10267 which are only read when run in an initial RAM disk. They
10268 usually follow closely their normal counterparts, but are
10269 prefixed with rd.
10270
10271 * There's a new tool to analyze the readahead files that are
10272 automatically generated at boot. Use:
10273
10274 /usr/lib/systemd/systemd-readahead analyze /.readahead
10275
10276 * We now provide an early debug shell on tty9 if this enabled. Use:
10277
d1f9edaf 10278 systemctl enable debug-shell.service
b5b4c94a
LP
10279
10280 * All plymouth related units have been moved into the Plymouth
10281 package. Please make sure to upgrade your Plymouth version
10282 as well.
10283
10284 * systemd-tmpfiles now supports getting passed the basename of
10285 a configuration file only, in which case it will look for it
10286 in all appropriate directories automatically.
10287
10288 * udevadm info now takes a /dev or /sys path as argument, and
10289 does the right thing. Example:
10290
10291 udevadm info /dev/sda
10292 udevadm info /sys/class/block/sda
10293
10294 * systemctl now prints a warning if a unit is stopped but a
10295 unit that might trigger it continues to run. Example: a
10296 service is stopped but the socket that activates it is left
10297 running.
10298
10299 * "systemctl status" will now mention if the log output was
10300 shortened due to rotation since a service has been started.
10301
10302 * The journal API now exposes functions to determine the
10303 "cutoff" times due to rotation.
10304
10305 * journald now understands SIGUSR1 and SIGUSR2 for triggering
10306 immediately flushing of runtime logs to /var if possible,
10307 resp. for triggering immediate rotation of the journal
10308 files.
10309
10310 * It is now considered an error if a service is attempted to
10311 be stopped that is not loaded.
10312
10313 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
10314
10315 * systemd-analyze now supports Python 3
10316
10317 * tmpfiles now supports cleaning up directories via aging
10318 where the first level dirs are always kept around but
10319 directories beneath it automatically aged. This is enabled
10320 by prefixing the age field with '~'.
10321
10322 * Seat objects now expose CanGraphical, CanTTY properties
10323 which is required to deal with very fast bootups where the
10324 display manager might be running before the graphics drivers
10325 completed initialization.
10326
10327 * Seat objects now expose a State property.
10328
10329 * We now include RPM macros for service enabling/disabling
10330 based on the preset logic. We recommend RPM based
10331 distributions to make use of these macros if possible. This
10332 makes it simpler to reuse RPM spec files across
10333 distributions.
10334
10335 * We now make sure that the collected systemd unit name is
10336 always valid when services log to the journal via
10337 STDOUT/STDERR.
10338
10339 * There's a new man page kernel-command-line(7) detailing all
10340 command line options we understand.
10341
10342 * The fstab generator may now be disabled at boot by passing
10343 fstab=0 on the kernel command line.
10344
91ac7425 10345 * A new kernel command line option modules-load= is now understood
b5b4c94a
LP
10346 to load a specific kernel module statically, early at boot.
10347
10348 * Unit names specified on the systemctl command line are now
10349 automatically escaped as needed. Also, if file system or
10350 device paths are specified they are automatically turned
10351 into the appropriate mount or device unit names. Example:
10352
10353 systemctl status /home
10354 systemctl status /dev/sda
10355
10356 * The SysVConsole= configuration option has been removed from
10357 system.conf parsing.
10358
10359 * The SysV search path is no longer exported on the D-Bus
10360 Manager object.
10361
ce830873 10362 * The Names= option has been removed from unit file parsing.
b5b4c94a
LP
10363
10364 * There's a new man page bootup(7) detailing the boot process.
10365
10366 * Every unit and every generator we ship with systemd now
10367 comes with full documentation. The self-explanatory boot is
10368 complete.
10369
10370 * A couple of services gained "systemd-" prefixes in their
10371 name if they wrap systemd code, rather than only external
10372 code. Among them fsck@.service which is now
10373 systemd-fsck@.service.
10374
10375 * The HaveWatchdog property has been removed from the D-Bus
10376 Manager object.
10377
10378 * systemd.confirm_spawn= on the kernel command line should now
10379 work sensibly.
10380
10381 * There's a new man page crypttab(5) which details all options
10382 we actually understand.
10383
10384 * systemd-nspawn gained a new --capability= switch to pass
10385 additional capabilities to the container.
10386
10387 * timedated will now read known NTP implementation unit names
5b00c016 10388 from /usr/lib/systemd/ntp-units.d/*.list,
b5b4c94a
LP
10389 systemd-timedated-ntp.target has been removed.
10390
10391 * journalctl gained a new switch "-b" that lists log data of
10392 the current boot only.
10393
10394 * The notify socket is in the abstract namespace again, in
10395 order to support daemons which chroot() at start-up.
10396
10397 * There is a new Storage= configuration option for journald
10398 which allows configuration of where log data should go. This
10399 also provides a way to disable journal logging entirely, so
10400 that data collected is only forwarded to the console, the
10401 kernel log buffer or another syslog implementation.
10402
c4f1b862 10403 * Many bugfixes and optimizations
b5b4c94a 10404
2d938ac7
LP
10405 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
10406 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
10407 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
10408 Shawn Landden, Tom Gundersen
b5b4c94a 10409
2d197285 10410CHANGES WITH 185:
b6a86739 10411
2d197285
KS
10412 * "systemctl help <unit>" now shows the man page if one is
10413 available.
10414
10415 * Several new man pages have been added.
10416
b5b4c94a
LP
10417 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
10418 MaxLevelConsole= can now be specified in
10419 journald.conf. These options allow reducing the amount of
10420 data stored on disk or forwarded by the log level.
2d197285 10421
b5b4c94a
LP
10422 * TimerSlackNSec= can now be specified in system.conf for
10423 PID1. This allows system-wide power savings.
2d197285
KS
10424
10425 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
10426 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
10427 Matthias Clasen
10428
4c8cd173 10429CHANGES WITH 184:
b6a86739 10430
4c8cd173
LP
10431 * logind is now capable of (optionally) handling power and
10432 sleep keys as well as the lid switch.
10433
10434 * journalctl now understands the syntax "journalctl
10435 /usr/bin/avahi-daemon" to get all log output of a specific
10436 daemon.
10437
10438 * CapabilityBoundingSet= in system.conf now also influences
10439 the capability bound set of usermode helpers of the kernel.
10440
10441 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
10442 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
10443 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
10444 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
10445
ea5943d3 10446CHANGES WITH 183:
b6a86739 10447
187076d4
LP
10448 * Note that we skipped 139 releases here in order to set the
10449 new version to something that is greater than both udev's
10450 and systemd's most recent version number.
10451
194bbe33
KS
10452 * udev: all udev sources are merged into the systemd source tree now.
10453 All future udev development will happen in the systemd tree. It
10454 is still fully supported to use the udev daemon and tools without
10455 systemd running, like in initramfs or other init systems. Building
10456 udev though, will require the *build* of the systemd tree, but
ea5943d3 10457 udev can be properly *run* without systemd.
07cd4fc1 10458
91cf7e5c 10459 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
f13b388f
KS
10460 should be used to create dead device nodes as workarounds for broken
10461 subsystems.
64661ee7 10462
2d13da88
KS
10463 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
10464 no longer supported. udev_monitor_new_from_netlink() needs to be
10465 used to subscribe to events.
10466
194bbe33
KS
10467 * udev: when udevd is started by systemd, processes which are left
10468 behind by forking them off of udev rules, are unconditionally cleaned
10469 up and killed now after the event handling has finished. Services or
10470 daemons must be started as systemd services. Services can be
ea5943d3 10471 pulled-in by udev to get started, but they can no longer be directly
194bbe33
KS
10472 forked by udev rules.
10473
f13b388f
KS
10474 * udev: the daemon binary is called systemd-udevd now and installed
10475 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
10476 to adapt to that, create symlink, or rename the binary after building
10477 it.
10478
ea5943d3 10479 * libudev no longer provides these symbols:
c1959569
KS
10480 udev_monitor_from_socket()
10481 udev_queue_get_failed_list_entry()
10482 udev_get_{dev,sys,run}_path()
ea5943d3 10483 The versions number was bumped and symbol versioning introduced.
c1959569 10484
ea5943d3 10485 * systemd-loginctl and systemd-journalctl have been renamed
9ae9afce 10486 to loginctl and journalctl to match systemctl.
18b754d3
KS
10487
10488 * The config files: /etc/systemd/systemd-logind.conf and
10489 /etc/systemd/systemd-journald.conf have been renamed to
10490 logind.conf and journald.conf. Package updates should rename
10491 the files to the new names on upgrade.
10492
ea5943d3
LP
10493 * For almost all files the license is now LGPL2.1+, changed
10494 from the previous GPL2.0+. Exceptions are some minor stuff
10495 of udev (which will be changed to LGPL2.1 eventually, too),
10496 and the MIT licensed sd-daemon.[ch] library that is suitable
10497 to be used as drop-in files.
10498
10499 * systemd and logind now handle system sleep states, in
49f43d5f 10500 particular suspending and hibernating.
ea5943d3
LP
10501
10502 * logind now implements a sleep/shutdown/idle inhibiting logic
10503 suitable for a variety of uses. Soonishly Lennart will blog
10504 about this in more detail.
10505
10506 * var-run.mount and var-lock.mount are no longer provided
ce830873 10507 (which previously bind mounted these directories to their new
ea5943d3
LP
10508 places). Distributions which have not converted these
10509 directories to symlinks should consider stealing these files
10510 from git history and add them downstream.
10511
10512 * We introduced the Documentation= field for units and added
10513 this to all our shipped units. This is useful to make it
3943231c 10514 easier to explore the boot and the purpose of the various
ea5943d3
LP
10515 units.
10516
10517 * All smaller setup units (such as
10518 systemd-vconsole-setup.service) now detect properly if they
10519 are run in a container and are skipped when
10520 appropriate. This guarantees an entirely noise-free boot in
10521 Linux container environments such as systemd-nspawn.
10522
10523 * A framework for implementing offline system updates is now
10524 integrated, for details see:
c6749ba5 10525 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
ea5943d3
LP
10526
10527 * A new service type Type=idle is available now which helps us
10528 avoiding ugly interleaving of getty output and boot status
10529 messages.
10530
439d6dfd
LP
10531 * There's now a system-wide CapabilityBoundingSet= option to
10532 globally reduce the set of capabilities for the
ea5943d3
LP
10533 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
10534 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
10535 even CAP_NET_ADMIN system-wide for secure systems.
10536
10537 * There are now system-wide DefaultLimitXXX= options to
10538 globally change the defaults of the various resource limits
10539 for all units started by PID 1.
10540
10541 * Harald Hoyer's systemd test suite has been integrated into
10542 systemd which allows easy testing of systemd builds in qemu
10543 and nspawn. (This is really awesome! Ask us for details!)
10544
3943231c
LP
10545 * The fstab parser is now implemented as generator, not inside
10546 of PID 1 anymore.
ea5943d3
LP
10547
10548 * systemctl will now warn you if .mount units generated from
10549 /etc/fstab are out of date due to changes in fstab that
d28315e4 10550 have not been read by systemd yet.
ea5943d3
LP
10551
10552 * systemd is now suitable for usage in initrds. Dracut has
10553 already been updated to make use of this. With this in place
10554 initrds get a slight bit faster but primarily are much
10555 easier to introspect and debug since "systemctl status" in
10556 the host system can be used to introspect initrd services,
10557 and the journal from the initrd is kept around too.
10558
10559 * systemd-delta has been added, a tool to explore differences
10560 between user/admin configuration and vendor defaults.
10561
10562 * PrivateTmp= now affects both /tmp and /var/tmp.
10563
10564 * Boot time status messages are now much prettier and feature
10565 proper english language. Booting up systemd has never been
10566 so sexy.
10567
10568 * Read-ahead pack files now include the inode number of all
10569 files to pre-cache. When the inode changes the pre-caching
10570 is not attempted. This should be nicer to deal with updated
10571 packages which might result in changes of read-ahead
10572 patterns.
10573
10574 * We now temporaritly lower the kernel's read_ahead_kb variable
10575 when collecting read-ahead data to ensure the kernel's
10576 built-in read-ahead does not add noise to our measurements
10577 of necessary blocks to pre-cache.
10578
10579 * There's now RequiresMountsFor= to add automatic dependencies
10580 for all mounts necessary for a specific file system path.
10581
10582 * MountAuto= and SwapAuto= have been removed from
10583 system.conf. Mounting file systems at boot has to take place
10584 in systemd now.
10585
10586 * nspawn now learned a new switch --uuid= to set the machine
10587 ID on the command line.
10588
f8c0a2cb 10589 * nspawn now learned the -b switch to automatically search
ea5943d3
LP
10590 for an init system.
10591
10592 * vt102 is now the default TERM for serial TTYs, upgraded from
10593 vt100.
10594
10595 * systemd-logind now works on VT-less systems.
10596
10597 * The build tree has been reorganized. The individual
3943231c 10598 components now have directories of their own.
ea5943d3
LP
10599
10600 * A new condition type ConditionPathIsReadWrite= is now available.
10601
10602 * nspawn learned the new -C switch to create cgroups for the
10603 container in other hierarchies.
10604
10605 * We now have support for hardware watchdogs, configurable in
10606 system.conf.
10607
10608 * The scheduled shutdown logic now has a public API.
10609
10610 * We now mount /tmp as tmpfs by default, but this can be
10611 masked and /etc/fstab can override it.
10612
d28315e4 10613 * Since udisks does not make use of /media anymore we are not
ea5943d3
LP
10614 mounting a tmpfs on it anymore.
10615
10616 * journalctl gained a new --local switch to only interleave
10617 locally generated journal files.
10618
10619 * We can now load the IMA policy at boot automatically.
10620
10621 * The GTK tools have been split off into a systemd-ui.
10622
79849bf9
LP
10623 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
10624 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
10625 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
10626 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
10627 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
10628 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
10629 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
10630 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
10631 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
10632 Gundersen
10633
16f1239e 10634CHANGES WITH 44:
b6a86739 10635
16f1239e
LP
10636 * This is mostly a bugfix release
10637
10638 * Support optional initialization of the machine ID from the
10639 KVM or container configured UUID.
10640
10641 * Support immediate reboots with "systemctl reboot -ff"
10642
10643 * Show /etc/os-release data in systemd-analyze output
10644
ab06eef8 10645 * Many bugfixes for the journal, including endianness fixes and
16f1239e
LP
10646 ensuring that disk space enforcement works
10647
ce830873 10648 * sd-login.h is C++ compatible again
16f1239e
LP
10649
10650 * Extend the /etc/os-release format on request of the Debian
10651 folks
10652
10653 * We now refuse non-UTF8 strings used in various configuration
d28315e4 10654 and unit files. This is done to ensure we do not pass invalid
16f1239e
LP
10655 data over D-Bus or expose it elsewhere.
10656
10657 * Register Mimo USB Screens as suitable for automatic seat
10658 configuration
10659
10660 * Read SELinux client context from journal clients in a race
10661 free fashion
10662
10663 * Reorder configuration file lookup order. /etc now always
10664 overrides /run in order to allow the administrator to always
b938cb90 10665 and unconditionally override vendor-supplied or
16f1239e
LP
10666 automatically generated data.
10667
10668 * The various user visible bits of the journal now have man
10669 pages. We still lack man pages for the journal API calls
10670 however.
10671
10672 * We now ship all man pages in HTML format again in the
10673 tarball.
10674
10675 Contributions from: Dave Reisner, Dirk Eibach, Frederic
10676 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
10677 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
10678 Reding
10679
437b7dee 10680CHANGES WITH 43:
b6a86739 10681
437b7dee
LP
10682 * This is mostly a bugfix release
10683
10684 * systems lacking /etc/os-release are no longer supported.
10685
10686 * Various functionality updates to libsystemd-login.so
10687
45afd519 10688 * Track class of PAM logins to distinguish greeters from
437b7dee
LP
10689 normal user logins.
10690
10691 Contributions from: Kay Sievers, Lennart Poettering, Michael
10692 Biebl
10693
204fa33c 10694CHANGES WITH 42:
b6a86739 10695
204fa33c
LP
10696 * This is an important bugfix release for v41.
10697
10698 * Building man pages is now optional which should be useful
10699 for those building systemd from git but unwilling to install
10700 xsltproc.
10701
10702 * Watchdog support for supervising services is now usable. In
10703 a future release support for hardware watchdogs
10704 (i.e. /dev/watchdog) will be added building on this.
10705
10706 * Service start rate limiting is now configurable and can be
10707 turned off per service. When a start rate limit is hit a
10708 reboot can automatically be triggered.
10709
10710 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
10711
10712 Contributions from: Benjamin Franzke, Bill Nottingham,
10713 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
10714 Schmidt, Michał Górny, Piotr Drąg
10715
e0d25329 10716CHANGES WITH 41:
b6a86739 10717
e0d25329
KS
10718 * The systemd binary is installed /usr/lib/systemd/systemd now;
10719 An existing /sbin/init symlink needs to be adapted with the
10720 package update.
10721
b13df964
LP
10722 * The code that loads kernel modules has been ported to invoke
10723 libkmod directly, instead of modprobe. This means we do not
10724 support systems with module-init-tools anymore.
10725
10726 * Watchdog support is now already useful, but still not
10727 complete.
10728
10729 * A new kernel command line option systemd.setenv= is
10730 understood to set system wide environment variables
10731 dynamically at boot.
10732
e9c1ea9d 10733 * We now limit the set of capabilities of systemd-journald.
ccd07a08 10734
353e12c2
LP
10735 * We now set SIGPIPE to ignore by default, since it only is
10736 useful in shell pipelines, and has little use in general
10737 code. This can be disabled with IgnoreSIPIPE=no in unit
10738 files.
10739
b13df964
LP
10740 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
10741 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
10742 William Douglas
10743
d26e4270 10744CHANGES WITH 40:
b6a86739 10745
d26e4270
LP
10746 * This is mostly a bugfix release
10747
10748 * We now expose the reason why a service failed in the
10749 "Result" D-Bus property.
10750
10751 * Rudimentary service watchdog support (will be completed over
10752 the next few releases.)
10753
10754 * When systemd forks off in order execute some service we will
10755 now immediately changes its argv[0] to reflect which process
10756 it will execute. This is useful to minimize the time window
10757 with a generic argv[0], which makes bootcharts more useful
10758
b13df964
LP
10759 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
10760 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
10761 Mike Kazantsev, Ray Strode
10762
220a21d3 10763CHANGES WITH 39:
b6a86739 10764
220a21d3
LP
10765 * This is mostly a test release, but incorporates many
10766 bugfixes.
10767
10768 * New systemd-cgtop tool to show control groups by their
10769 resource usage.
10770
10771 * Linking against libacl for ACLs is optional again. If
10772 disabled, support tracking device access for active logins
10773 goes becomes unavailable, and so does access to the user
10774 journals by the respective users.
10775
10776 * If a group "adm" exists, journal files are automatically
10777 owned by them, thus allow members of this group full access
10778 to the system journal as well as all user journals.
10779
10780 * The journal now stores the SELinux context of the logging
10781 client for all entries.
10782
10783 * Add C++ inclusion guards to all public headers
10784
10785 * New output mode "cat" in the journal to print only text
10786 messages, without any meta data like date or time.
10787
10788 * Include tiny X server wrapper as a temporary stop-gap to
10789 teach XOrg udev display enumeration. This is used by display
10790 managers such as gdm, and will go away as soon as XOrg
10791 learned native udev hotplugging for display devices.
10792
10793 * Add new systemd-cat tool for executing arbitrary programs
10794 with STDERR/STDOUT connected to the journal. Can also act as
10795 BSD logger replacement, and does so by default.
10796
10797 * Optionally store all locally generated coredumps in the
10798 journal along with meta data.
10799
10800 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
10801 writing short strings to files (for usage for /sys), and for
10802 creating symlinks, character and block device nodes.
10803
10804 * New unit file option ControlGroupPersistent= to make cgroups
10805 persistent, following the mechanisms outlined in
56cadcb6 10806 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
220a21d3
LP
10807
10808 * Support multiple local RTCs in a sane way
10809
10810 * No longer monopolize IO when replaying readahead data on
10811 rotating disks, since we might starve non-file-system IO to
10812 death, since fanotify() will not see accesses done by blkid,
10813 or fsck.
10814
d28315e4 10815 * Do not show kernel threads in systemd-cgls anymore, unless
220a21d3
LP
10816 requested with new -k switch.
10817
10818 Contributions from: Dan Horák, Kay Sievers, Lennart
10819 Poettering, Michal Schmidt
10820
10821CHANGES WITH 38:
b6a86739 10822
220a21d3
LP
10823 * This is mostly a test release, but incorporates many
10824 bugfixes.
10825
10826 * The git repository moved to:
10827 git://anongit.freedesktop.org/systemd/systemd
10828 ssh://git.freedesktop.org/git/systemd/systemd
10829
10830 * First release with the journal
10831 http://0pointer.de/blog/projects/the-journal.html
10832
10833 * The journal replaces both systemd-kmsg-syslogd and
10834 systemd-stdout-bridge.
10835
10836 * New sd_pid_get_unit() API call in libsystemd-logind
10837
10838 * Many systemadm clean-ups
10839
10840 * Introduce remote-fs-pre.target which is ordered before all
10841 remote mounts and may be used to start services before all
10842 remote mounts.
10843
10844 * Added Mageia support
10845
10846 * Add bash completion for systemd-loginctl
10847
10848 * Actively monitor PID file creation for daemons which exit in
10849 the parent process before having finished writing the PID
10850 file in the daemon process. Daemons which do this need to be
10851 fixed (i.e. PID file creation must have finished before the
10852 parent exits), but we now react a bit more gracefully to them.
10853
10854 * Add colourful boot output, mimicking the well-known output
10855 of existing distributions.
10856
10857 * New option PassCredentials= for socket units, for
10858 compatibility with a recent kernel ABI breakage.
10859
10860 * /etc/rc.local is now hooked in via a generator binary, and
10861 thus will no longer act as synchronization point during
10862 boot.
10863
10864 * systemctl list-unit-files now supports --root=.
10865
10866 * systemd-tmpfiles now understands two new commands: z, Z for
10867 relabelling files according to the SELinux database. This is
10868 useful to apply SELinux labels to specific files in /sys,
10869 among other things.
10870
10871 * Output of SysV services is now forwarded to both the console
10872 and the journal by default, not only just the console.
10873
10874 * New man pages for all APIs from libsystemd-login.
10875
ce830873 10876 * The build tree got reorganized and the build system is a
220a21d3
LP
10877 lot more modular allowing embedded setups to specifically
10878 select the components of systemd they are interested in.
10879
10880 * Support for Linux systems lacking the kernel VT subsystem is
10881 restored.
10882
10883 * configure's --with-rootdir= got renamed to
10884 --with-rootprefix= to follow the naming used by udev and
10885 kmod
10886
d28315e4 10887 * Unless specified otherwise we will now install to /usr instead
220a21d3
LP
10888 of /usr/local by default.
10889
10890 * Processes with '@' in argv[0][0] are now excluded from the
10891 final shut-down killing spree, following the logic explained
10892 in:
56cadcb6 10893 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
220a21d3
LP
10894
10895 * All processes remaining in a service cgroup when we enter
10896 the START or START_PRE states are now killed with
10897 SIGKILL. That means it is no longer possible to spawn
10898 background processes from ExecStart= lines (which was never
10899 supported anyway, and bad style).
10900
10901 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
10902 reloading of units together.
10903
4c8cd173 10904 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
220a21d3
LP
10905 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
10906 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
10907 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
10908 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek