]> git.ipfire.org Git - thirdparty/systemd.git/blob - NEWS
Merge pull request #13216 from poettering/busctl-format-table
[thirdparty/systemd.git] / NEWS
1 systemd System and Service Manager
2
3 CHANGES WITH 243 in spe:
4
5 * This release enables unprivileged programs (i.e. requiring neither
6 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
7 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
8 kernel for the whole UNIX group range, i.e. all processes. This
9 change should be reasonably safe, as the kernel support for it was
10 specifically implemented to allow safe access to ICMP Echo for
11 processes lacking any privileges. If this is not desirable, it can be
12 disabled again by setting the parameter to "1 0".
13
14 * Previously, filters defined with SystemCallFilter= would have the
15 effect that any calling of an offending system call would terminate
16 the calling thread. This behaviour never made much sense, since
17 killing individual threads of unsuspecting processes is likely to
18 create more problems than it solves. With this release the default
19 action changed from killing the thread to killing the whole
20 process. For this to work correctly both a kernel version (>= 4.14)
21 and a libseccomp version (>= 2.4.0) supporting this new seccomp
22 action is required. If an older kernel or libseccomp is used the old
23 behaviour continues to be used. This change does not affect any
24 services that have no system call filters defined, or that use
25 SystemCallErrorNumber= (and thus see EPERM or another error instead
26 of being killed when calling an offending system call). Note that
27 systemd documentation always claimed that the whole process is
28 killed. With this change behaviour is thus adjusted to match the
29 documentation.
30
31 * The "kernel.pid_max" sysctl is now bumped to 4194304 by default,
32 i.e. the full 22bit range the kernel allows, up from the old 16bit
33 range. This should improve security and robustness a bit, as PID
34 collisions are made less likely (though certainly still
35 possible). There are rumours this might create compatibility
36 problems, though at this moment no practical ones are known to
37 us. Downstream distributions are hence advised to undo this change in
38 their builds if they are concerned about maximum compatibility, but
39 for everybody else we recommend leaving the value bumped. Besides
40 improving security and robustness this should also simplify things as
41 the maximum number of allowed concurrent tasks was previously bounded
42 by both "kernel.pid_max" and "kernel.threads-max" and now only a
43 single knob is left ("kernel.threads-max"). There have been concerns
44 that usability is affected by this change because larger PID numbers
45 are harder to type, but we believe the change from 5 digit PIDs to 7
46 digit PIDs is not too hampering for usability.
47
48 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
49 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
50 hierarchically set default memory protection values for a particular
51 subtree of the unit hierarchy.
52
53 * Memory protection directives can now take a value of zero, allowing
54 explicit opting out of a default value propagated by an ancestor.
55
56 * A new setting DisableControllers= has been added that may be used to
57 explicitly disable one or more cgroups controllers for a unit and all
58 its children.
59
60 * systemd now defaults to the "unified" cgroup hierarchy setup during
61 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
62 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
63 change reflects the fact that cgroupsv2 support has matured
64 substantially in both systemd and in the kernel, and is clearly the
65 way forward. Downstream production distributions might want to
66 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
67 their builds as unfortunately the popular container managers have not
68 caught up with the kernel API changes.
69
70 * Man pages are not built by default anymore (html pages were already
71 disabled by default), to make development builds quicker. When
72 building systemd for a full installation with documentation, meson
73 should be called with -Dman=true and/or -Dhtml=true as
74 appropriate. The default was changed based on the assumption that
75 quick one-off or repeated development builds are much more common
76 than full optimized builds for installation, and people need to pass
77 various other options to when doing "proper" builds anyway, so the
78 gain from making development builds quicker is bigger than the one
79 time disruption for packagers.
80
81 Two scripts are created in the *build* directory to generate and
82 preview man and html pages on demand, e.g.:
83
84 build/man/man systemctl
85 build/man/html systemd.index
86
87 * libidn2 is used by default if both libidn2 and libidn are installed.
88 Please use -Dlibidn=true when libidn is favorable.
89
90 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
91 big-endian machines. Before, bytes were written and read in native
92 machine order as exposed by the native libc __cpu_mask interface.
93 Now, little-endian order is always used (CPUs 0–7 are described by
94 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
95 This change fixes D-Bus calls that cross endianness boundary.
96
97 The presentation format used for CPUAffinity= by "systemctl show" and
98 "systemd-analyze dump" is changed to present CPU indices instead of
99 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
100 shown as CPUAffinity=03000000000000000000000000000… (on
101 little-endian) or CPUAffinity=00000000000000300000000000000… (on
102 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
103 input format. The maximum integer that will be printed in the new
104 format is 8191 (four digits), while the old format always used a very
105 long number (with the length varying by architecture), so they can be
106 unambiguously distinguished.
107
108 * SuccessExitStatus=, RestartPreventExitStatus=, and
109 RestartForceExitStatus= now accept exit code names (e.g. "DATAERR" is
110 equivalent to "65"). systemd-analyze learnt a new 'exit-codes' verb
111 to display those exit code name mappings.
112
113 * /usr/sbin/halt.local is no longer supported. Implementation in
114 distributions was inconsistent and it seems this functionality was
115 very rarely used.
116
117 To replace this functionality, users should:
118 - either define a new unit and make it a dependency of final.target
119 (systemctl add-wants final.target my-halt-local.service)
120 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
121 and ensure that it accepts "halt", "poweroff", "reboot", and
122 "kexec" as an argument, see the description in systemd-shutdown(8).
123
124 * When a [Match] section in .link or .network file is empty (contains
125 no match patterns), a warning will be emitted. Please add any "match
126 all" pattern instead, e.g. OriginalName=* or Name=* in case all
127 interfaces should really be matched.
128
129 * A new setting NUMAPolicy= may be used to set process memory
130 allocation policy. This setting can be specified in
131 /etc/systemd/system.conf and hence will set the default policy for
132 PID1. The default policy can be overridden on a per-service
133 basis. The related setting NUMAMask= is used to specify NUMA node
134 mask that should be associated with the selected policy.
135
136 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
137 generates when processes it manages are reaching their memory limits,
138 and will place their units in a special state, and optionally kill or
139 stop the whole unit.
140
141 * The service manager will now expose bus properties for the IO
142 resources used by units. This information is also shown in "systemctl
143 status" now (for services that have IOAccounting=yes set). Moreover,
144 the IO accounting data is included in the resource log message
145 generated whenever a unit stops.
146
147 * Units may now configure an explicit time-out to wait for when killed
148 with SIGABRT, for example when a service watchdog is hit. Previously,
149 the regular TimeoutStopSec= time-out was applied in this case too —
150 now a separate time-out may be set using TimeoutAbortSec=.
151
152 * Services may now send a special WATCHDOG=trigger message with
153 sd_notify() to trigger an immediate "watchdog missed" event, and thus
154 request service take down. This is useful both for testing watchdog
155 handling, but also for defining error paths in services, that shall
156 be handled the same way as watchdog events.
157
158 * There are two new per-unit settings IPIngressFilterPath= and
159 IPEgressFilterPath= which allow configuration of a BPF program
160 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
161 to apply to the IP packet ingress/egress path of all processes of a
162 unit. This is useful to allow running systemd services with BPF
163 programs set up externally.
164
165 * systemctl gained a new "clean" verb for removing the state, cache,
166 runtime or logs directories of a service while it is terminated. The
167 new verb may also be used to remove the state maintained on disk for
168 timer units that have Persistent= configured.
169
170 * During the last phase of shutdown systemd will now automatically
171 increase the log level configured in the "kernel.printk" sysctl so
172 that any relevant loggable events happening during late shutdown are
173 made visible. Previously, loggable events happening so late during
174 shutdown were generally lost if the "kernel.printk" sysctl was set to
175 high thresholds, as regular logging daemons are terminated at that
176 time and thus nothing is written to disk.
177
178 * If processes terminated during the last phase of shutdown do not exit
179 quickly systemd will now show their names after a short time, to make
180 debugging easier. After a longer time-out they are forcibly killed,
181 as before.
182
183 * journalctl (and the other tools that display logs) will now highlight
184 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
185 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
186 are now shown in blue color, to separate them visually from regular
187 logs. References to configuration files are now turned into clickable
188 links on terminals that support that.
189
190 * systemd-journald will now stop logging to /var/log/journal during
191 shutdown when /var/ is on a separate mount, so that it can be
192 unmounted safely during shutdown.
193
194 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
195
196 * systemd-resolved "Cache=" configuration option in resolved.conf has
197 been extended to also accept the 'no-negative' value. Previously,
198 only a boolean option was allowed (yes/no), having yes as the
199 default. If this option is set to 'no-negative', negative answers
200 are skipped from being cached while keeping the same cache heuristics
201 for positive answers. The default remains as "yes" (i.e. caching is
202 enabled).
203
204 * The predictable naming scheme for network devices now supports
205 generating predictable names for "netdevsim" devices.
206
207 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
208 interfaces natively.
209
210 * systemd-networkd's bridge FDB support now allows configuration of a
211 destination address for each entry (Destination=), as well as the
212 VXLAN VNI (VNI=), as well as an option to declare what an entry is
213 associated with (AssociatedWith=).
214
215 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
216 option for configuring the maximum number of DHCP lease requests. It
217 also learnt a new BlackList= option for blacklisting DHCP servers (a
218 similar setting has also been added to the IPv6 RA client), as well
219 as a SendRelease= option for configuring whether to send a DHCP
220 RELEASE message when terminating.
221
222 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
223 separately in the [DHCPv4] and [DHCPv6] sections.
224
225 * systemd-networkd's DHCP support will now optionally create an
226 implicit host route to the DNS server specified in the DHCP lease, in
227 addition to the routes listed explicitly in the lease. This should
228 ensure that in multi-homed systems DNS traffic leaves the systems on
229 the interface that acquired the DNS server information even if other
230 routes such as default routes exist. This behaviour may be turned on
231 with the new RoutesToDNS= option.
232
233 * systemd-networkd's VXLAN support gained a new option
234 GenericProtocolExtension= for enabling VXLAN Generic Protocol
235 Extension support, as well as IPDoNotFragment= for setting the IP
236 "Don't fragment" bit on outgoing packets. A similar option has been
237 added to the GENEVE support.
238
239 * In systemd-networkd's [Route] section you may now configure
240 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
241 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
242 propagation. The Type= setting now supports local, broadcast,
243 anycast, multicast, any, xresolve routes, too.
244
245 * systemd-networkd's [Network] section learnt a new option
246 DefaultRouteOnDevice= for automatically configuring a default route
247 onto the network device.
248
249 * systemd-networkd's bridging support gained two new options ProxyARP=
250 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
251 MulticastRouter= for configuring multicast routing behaviour. A new
252 option MulticastIGMPVersion= may be used to change bridge's multicast
253 Internet Group Management Protocol (IGMP) version.
254
255 * systemd-networkd's FooOverUDP support gained the ability to configure
256 local and peer IP addresses via Local= and Peer=. A new option
257 PeerPort= may be used to configure the peer's IP port.
258
259 * systemd-networkd's TUN support gained a new setting VnetHeader= for
260 tweaking Generic Segment Offload support.
261
262 * networkctl gained a new "delete" command for removing virtual network
263 devices, as well as a new "--stats" switch for showing device
264 statistics.
265
266 * networkd.conf gained a new setting SpeedMeter= and
267 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
268 measured speed may be shown by 'networkctl status'.
269
270 * systemd-networkd's .network and .link files gained a new Property=
271 setting in the [Match] section, to match against devices with
272 specific udev properties.
273
274 * systemd-networkd's tunnel support gained a new option
275 AssignToLoopback= for selecting whether to use the loopback device
276 "lo" as underlying device.
277
278 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
279 been renamed to LinkLayerAddress=, and it now allows configuration of
280 IP addresses, too.
281
282 * A new tool systemd-network-generator has been added that may generate
283 .network, .netdev and .link files from IP configuration specified on
284 the kernel command line, compatible with the format Dracut expects.
285
286 * The CriticalConnection= setting in .network files is now deprecated,
287 and replaced by a new KeepConfiguration= setting which allows more
288 detailed configuration of the IP configuration to keep in place.
289
290 * systemd-analyze gained a new "timestamp" verb for parsing and
291 converting timestamps. It's similar to the existing "systemd-analyze
292 calendar" command which does the same for recurring calendar
293 events. It also gained a new "condition" verb for parsing and testing
294 ConditionXYZ= expressions.
295
296 * systemd-logind now exposes a per-session SetBrightness() bus call,
297 which may be used to securely change the brightness of a kernel
298 brightness device, if it belongs to the session's seat. By using this
299 call unprivileged clients can make changes to "backlight" and "leds"
300 devices securely with strict requirements on session
301 membership. Desktop environments may use this to generically make
302 brightness changes to such devices without shipping private SUID
303 binaries or specific udev rules for that purpose.
304
305 * "udevadm info" gained a --wait-for-initialization switch to wait for
306 a device to be initialized.
307
308 * systemd-hibernate-resume-generator will now look for resumeflags= on
309 the kernel command line, which is similar to rootflags= and may be
310 used to configure device timeouts for waiting for the hibernation
311 device to show up.
312
313 * sd-event learnt a new API call sd_event_source_disable_unref() for
314 disabling and unref'ing an event source in a single function. A
315 related call sd_event_source_disable_unrefp() has been added for use
316 with GCC's cleanup extension.
317
318 * The sd-id128.h public API gained a new definition
319 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
320 with printf().
321
322 * "busctl introspect" gained a new switch --xml-interface for dumping
323 XML introspection data unmodified.
324
325 * PID 1 may now show the unit name instead of the unit description
326 string in its status output during boot. This may be configured in
327 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
328 kernel command line option systemd.status_unit_format=.
329
330 * PID 1 now understands a new option KExecWatchdogSec= in
331 /etc/systemd/system.conf. It allows configuration of a watchdog
332 timeout to write to a hardware watchdog device on kexec-based
333 reboots. Previously this functionality was only available for regular
334 reboots. This option defaults to off, since it depends on drivers and
335 software setup whether the watchdog is correctly reset again after
336 the kexec completed, and thus for the general case not clear if safe
337 (since it might cause unwanted watchdog reboots after the kexec
338 completed otherwise). Moreover, the old ShutdownWatchdogSec= setting
339 has been renamed to RebootWatchdogSec= to more clearly communicate
340 what it is about. The old name of the setting is still accepted for
341 compatibility.
342
343 * The systemd.debug_shell kernel command line option now optionally
344 takes a tty name to spawn the debug shell on, which allows selecting
345 a different tty than the built-in default.
346
347 * Service units gained a new ExecCondition= setting which will run
348 before ExecStartPre= and either continue execution of the unit (for
349 clean exit codes), stop execution without marking the unit failed
350 (for exit codes 1 through 254), or stop execution and fail the unit
351 (for exit code 255 or cases of abnormal termination).
352
353 * A new service systemd-pstore.service has been added that pulls data
354 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
355 review.
356
357 * timedatectl gained new verbs for configuring per-interface NTP
358 service configuration for systemd-timesyncd.
359
360 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
361 2019. (You can set non-UTF-8 locales though, if you know their name.)
362
363 * If variable assignments in sysctl.d/ files are prefixed with "-" any
364 failures to apply them are now ignored.
365
366 * systemd-random-seed.service now optionally credits entropy when
367 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
368 true for the service to enable this behaviour, but please consult the
369 documentation first, since this comes with a couple of caveats.
370
371 * systemd-random-seed.service is now a synchronization point for the
372 point in time where the kernel's entropy pool is fully
373 initialized. Order services that require /dev/urandom to be correctly
374 initialized after this service.
375
376 * The systemd-boot boot loader has been updated to optionally maintain
377 a random seed file in the EFI System Partition (ESP). During the boot
378 phase, this random seed is read and updated with a new seed
379 crytographically derived from it. Another derived seed is passed to
380 the OS. The latter seed is then credited to the kernel's entropy pool
381 very early during userspace initialization (from PID 1). This allows
382 systems to boot up with a fully initialized kernel entropy pool from
383 earliest boot on, and thus entirely removes all entropy pool
384 initialization delays from systems using systemd-boot. Special care
385 is taken to ensure different seeds are derived on system images
386 replicated to multiple systems.
387
388 * bootctl gained a new verb "is-installed" that checks whether
389 systemd-boot is currently installed.
390
391 * A new document has been added describing systemd's use and support
392 for the kernel's entropy pool subsystem:
393
394 https://systemd.io/RANDOM_SEEDS
395
396 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Andrej
397 Valek, Anita Zhang, Arian van Putten, Balint Reczey, Bastien Nocera,
398 Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris Chiu, Chris Down,
399 Christian Kellner, Clinton Roy, Connor Reeder, Daniele Medri, Dan
400 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
401 Ray, Dominick Grift, Donald Buczek, Douglas Christman, Eric DeVolder,
402 Evgeny Vereshchagin, Feldwor, Felix Riemann, Florian Dollinger, Franck
403 Bui, Frantisek Sumsal, Franz Pletz, Hans de Goede, Iago López Galeiras,
404 Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob Unterwurzacher,
405 Jan Klötzke, Jan Pokorný, Jan Synacek, Jeka Pats, Jérémy Rosen, Jiri
406 Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
407 Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski, Kai Lüke, Karel
408 Zak, Kashyap Chamarthy, Krayushkin Konstantin, Lennart Poettering,
409 Lubomir Rintel, Luca Boccassi, Luís Ferreira, Marc-André Lureau, Markus
410 Felten, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Prokop,
411 Michael Stapelberg, Michael Zhivich, Michal Koutný, Michal Sekletar,
412 Mike Gilbert, Milan Broz, mpe85, Network Silence, Oliver Harley,
413 pan93412, Paul Menzel, pEJipE, Peter A. Bigot, Philip Withnall, Piotr
414 Drąg, Rafael Fontenelle, Roberto Santalla, root, RussianNeuroMancer,
415 Sebastian Jennen, Shreyas Behera, Simon Schricker, Susant Sahani,
416 Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thomas Haller, Thomas
417 Weißschuh, Tomas Mraz, Topi Miettinen, ven, Wieland Hoffmann, Xi
418 Ruoyao, Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew
419 Jędrzejewski-Szmek, Zhang Xianwei
420
421 – Somewhere, SOME-TI-ME
422
423 CHANGES WITH 242:
424
425 * In .link files, MACAddressPolicy=persistent (the default) is changed
426 to cover more devices. For devices like bridges, tun, tap, bond, and
427 similar interfaces that do not have other identifying information,
428 the interface name is used as the basis for persistent seed for MAC
429 and IPv4LL addresses. The way that devices that were handled
430 previously is not changed, and this change is about covering more
431 devices then previously by the "persistent" policy.
432
433 MACAddressPolicy=random may be used to force randomized MACs and
434 IPv4LL addresses for a device if desired.
435
436 Hint: the log output from udev (at debug level) was enhanced to
437 clarify what policy is followed and which attributes are used.
438 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
439 may be used to view this.
440
441 Hint: if a bridge interface is created without any slaves, and gains
442 a slave later, then now the bridge does not inherit slave's MAC.
443 To inherit slave's MAC, for example, create the following file:
444 ```
445 # /etc/systemd/network/98-bridge-inherit-mac.link
446 [Match]
447 Type=bridge
448
449 [Link]
450 MACAddressPolicy=none
451 ```
452
453 * The .device units generated by systemd-fstab-generator and other
454 generators do not automatically pull in the corresponding .mount unit
455 as a Wants= dependency. This means that simply plugging in the device
456 will not cause the mount unit to be started automatically. But please
457 note that the mount unit may be started for other reasons, in
458 particular if it is part of local-fs.target, and any unit which
459 (transitively) depends on local-fs.target is started.
460
461 * networkctl list/status/lldp now accept globbing wildcards for network
462 interface names to match against all existing interfaces.
463
464 * The $PIDFILE environment variable is set to point the absolute path
465 configured with PIDFile= for processes of that service.
466
467 * The fallback DNS server list was augmented with Cloudflare public DNS
468 servers. Use `-Ddns-servers=` to set a different fallback.
469
470 * A new special target usb-gadget.target will be started automatically
471 when a USB Device Controller is detected (which means that the system
472 is a USB peripheral).
473
474 * A new unit setting CPUQuotaPeriodSec= assigns the time period
475 relatively to which the CPU time quota specified by CPUQuota= is
476 measured.
477
478 * A new unit setting ProtectHostname= may be used to prevent services
479 from modifying hostname information (even if they otherwise would
480 have privileges to do so).
481
482 * A new unit setting NetworkNamespacePath= may be used to specify a
483 namespace for service or socket units through a path referring to a
484 Linux network namespace pseudo-file.
485
486 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
487 have an effect on .socket units: when used the listening socket is
488 created within the configured network namespace instead of the host
489 namespace.
490
491 * ExecStart= command lines in unit files may now be prefixed with ':'
492 in which case environment variable substitution is
493 disabled. (Supported for the other ExecXYZ= settings, too.)
494
495 * .timer units gained two new boolean settings OnClockChange= and
496 OnTimezoneChange= which may be used to also trigger a unit when the
497 system clock is changed or the local timezone is
498 modified. systemd-run has been updated to make these options easily
499 accessible from the command line for transient timers.
500
501 * Two new conditions for units have been added: ConditionMemory= may be
502 used to conditionalize a unit based on installed system
503 RAM. ConditionCPUs= may be used to conditionalize a unit based on
504 installed CPU cores.
505
506 * The @default system call filter group understood by SystemCallFilter=
507 has been updated to include the new rseq() system call introduced in
508 kernel 4.15.
509
510 * A new time-set.target has been added that indicates that the system
511 time has been set from a local source (possibly imprecise). The
512 existing time-sync.target is stronger and indicates that the time has
513 been synchronized with a precise external source. Services where
514 approximate time is sufficient should use the new target.
515
516 * "systemctl start" (and related commands) learnt a new
517 --show-transaction option. If specified brief information about all
518 jobs queued because of the requested operation is shown.
519
520 * systemd-networkd recognizes a new operation state 'enslaved', used
521 (instead of 'degraded' or 'carrier') for interfaces which form a
522 bridge, bond, or similar, and an new 'degraded-carrier' operational
523 state used for the bond or bridge master interface when one of the
524 enslaved devices is not operational.
525
526 * .network files learnt the new IgnoreCarrierLoss= option for leaving
527 networks configured even if the carrier is lost.
528
529 * The RequiredForOnline= setting in .network files may now specify a
530 minimum operational state required for the interface to be considered
531 "online" by systemd-networkd-wait-online. Related to this
532 systemd-networkd-wait-online gained a new option --operational-state=
533 to configure the same, and its --interface= option was updated to
534 optionally also take an operational state specific for an interface.
535
536 * systemd-networkd-wait-online gained a new setting --any for waiting
537 for only one of the requested interfaces instead of all of them.
538
539 * systemd-networkd now implements L2TP tunnels.
540
541 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
542 may be used to cause autonomous and onlink prefixes received in IPv6
543 Router Advertisements to be ignored.
544
545 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
546 file settings may be used to tweak bridge behaviour.
547
548 * The new TripleSampling= option in .network files may be used to
549 configure CAN triple sampling.
550
551 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
552 used to point to private or preshared key for a WireGuard interface.
553
554 * /etc/crypttab now supports the same-cpu-crypt and
555 submit-from-crypt-cpus options to tweak encryption work scheduling
556 details.
557
558 * systemd-tmpfiles will now take a BSD file lock before operating on a
559 contents of directory. This may be used to temporarily exclude
560 directories from aging by taking the same lock (useful for example
561 when extracting a tarball into /tmp or /var/tmp as a privileged user,
562 which might create files with really old timestamps, which
563 nevertheless should not be deleted). For further details, see:
564
565 https://systemd.io/TEMPORARY_DIRECTORIES
566
567 * systemd-tmpfiles' h line type gained support for the
568 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
569 controlling project quota inheritance.
570
571 * sd-boot and bootctl now implement support for an Extended Boot Loader
572 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
573 addition to the ESP partition mounted to /efi or /boot/efi.
574 Configuration file fragments, kernels, initrds and other EFI images
575 to boot will be loaded from both the ESP and XBOOTLDR partitions.
576 The XBOOTLDR partition was previously described by the Boot Loader
577 Specification, but implementation was missing in sd-boot. Support for
578 this concept allows using the sd-boot boot loader in more
579 conservative scenarios where the boot loader itself is placed in the
580 ESP but the kernels to boot (and their metadata) in a separate
581 partition.
582
583 * A system may now be booted with systemd.volatile=overlay on the
584 kernel command line, which causes the root file system to be set up
585 an overlayfs mount combining the root-only root directory with a
586 writable tmpfs. In this setup, the underlying root device is not
587 modified, and any changes are lost at reboot.
588
589 * Similar, systemd-nspawn can now boot containers with a volatile
590 overlayfs root with the new --volatile=overlay switch.
591
592 * systemd-nspawn can now consume OCI runtime bundles using a new
593 --oci-bundle= option. This implementation is fully usable, with most
594 features in the specification implemented, but since this a lot of
595 new code and functionality, this feature should most likely not
596 be used in production yet.
597
598 * systemd-nspawn now supports various options described by the OCI
599 runtime specification on the command-line and in .nspawn files:
600 --inaccessible=/Inaccessible= may be used to mask parts of the file
601 system tree, --console=/--pipe may be used to configure how standard
602 input, output, and error are set up.
603
604 * busctl learned the `emit` verb to generate D-Bus signals.
605
606 * systemd-analyze cat-config may be used to gather and display
607 configuration spread over multiple files, for example system and user
608 presets, tmpfiles.d, sysusers.d, udev rules, etc.
609
610 * systemd-analyze calendar now takes an optional new parameter
611 --iterations= which may be used to show a maximum number of iterations
612 the specified expression will elapse next.
613
614 * The sd-bus C API gained support for naming method parameters in the
615 introspection data.
616
617 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
618 the reboot() system call expects.
619
620 * journalctl learnt a new --cursor-file= option that points to a file
621 from which a cursor should be loaded in the beginning and to which
622 the updated cursor should be stored at the end.
623
624 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
625 detected by systemd-detect-virt (and may also be used in
626 ConditionVirtualization=).
627
628 * The behaviour of systemd-logind may now be modified with environment
629 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
630 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
631 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
632 skip the relevant operation completely (when set to false), or to
633 create a flag file in /run/systemd (when set to true), instead of
634 actually commencing the real operation when requested. The presence
635 of /run/systemd/reboot-to-firmware-setup,
636 /run/systemd/reboot-to-boot-loader-menu, and
637 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
638 boot loader implementations to replace some steps logind performs
639 during reboot with their own operations.
640
641 * systemctl can be used to request a reboot into the boot loader menu
642 or a specific boot loader entry with the new --boot-load-menu= and
643 --boot-loader-entry= options to a reboot command. (This requires a
644 boot loader that supports this, for example sd-boot.)
645
646 * kernel-install will no longer unconditionally create the output
647 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
648 snippets, but will do only if the machine-specific parent directory
649 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
650 to create this parent directory during sd-boot installation.
651
652 This makes it easier to use kernel-install with plugins which support
653 a different layout of the bootloader partitions (for example grub2).
654
655 * During package installation (with `ninja install`), we would create
656 symlinks for getty@tty1.service, systemd-networkd.service,
657 systemd-networkd.socket, systemd-resolved.service,
658 remote-cryptsetup.target, remote-fs.target,
659 systemd-networkd-wait-online.service, and systemd-timesyncd.service
660 in /etc, as if `systemctl enable` was called for those units, to make
661 the system usable immediately after installation. Now this is not
662 done anymore, and instead calling `systemctl preset-all` is
663 recommended after the first installation of systemd.
664
665 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
666 is built on seccomp. When turned on creation of SUID/SGID files is
667 prohibited.
668
669 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
670 implied if DynamicUser= is turned on for a service. This hardens
671 these services, so that they neither can benefit from nor create
672 SUID/SGID executables. This is a minor compatibility breakage, given
673 that when DynamicUser= was first introduced SUID/SGID behaviour was
674 unaffected. However, the security benefit of these two options is
675 substantial, and the setting is still relatively new, hence we opted
676 to make it mandatory for services with dynamic users.
677
678 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
679 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
680 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
681 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
682 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
683 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
684 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
685 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
686 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
687 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
688 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
689 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
690 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
691 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
692 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
693 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
694 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
695 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
696
697 — Warsaw, 2019-04-11
698
699 CHANGES WITH 241:
700
701 * The default locale can now be configured at compile time. Otherwise,
702 a suitable default will be selected automatically (one of C.UTF-8,
703 en_US.UTF-8, and C).
704
705 * The version string shown by systemd and other tools now includes the
706 git commit hash when built from git. An override may be specified
707 during compilation, which is intended to be used by distributions to
708 include the package release information.
709
710 * systemd-cat can now filter standard input and standard error streams
711 for different syslog priorities using the new --stderr-priority=
712 option.
713
714 * systemd-journald and systemd-journal-remote reject entries which
715 contain too many fields (CVE-2018-16865) and set limits on the
716 process' command line length (CVE-2018-16864).
717
718 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
719 again.
720
721 * A new network device NamePolicy "keep" is implemented for link files,
722 and used by default in 99-default.link (the fallback configuration
723 provided by systemd). With this policy, if the network device name
724 was already set by userspace, the device will not be renamed again.
725 This matches the naming scheme that was implemented before
726 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
727 is also enabled by default, even if not specified. Effectively, this
728 means that if naming-scheme >= 240 is specified, network devices will
729 be renamed according to the configuration, even if they have been
730 renamed already, if "keep" is not specified as the naming policy in
731 the .link file. The 99-default.link file provided by systemd includes
732 "keep" for backwards compatibility, but it is recommended for user
733 installed .link files to *not* include it.
734
735 The "kernel" policy, which keeps kernel names declared to be
736 "persistent", now works again as documented.
737
738 * kernel-install script now optionally takes the paths to one or more
739 initrd files, and passes them to all plugins.
740
741 * The mincore() system call has been dropped from the @system-service
742 system call filter group, as it is pretty exotic and may potentially
743 used for side-channel attacks.
744
745 * -fPIE is dropped from compiler and linker options. Please specify
746 -Db_pie=true option to meson to build position-independent
747 executables. Note that the meson option is supported since meson-0.49.
748
749 * The fs.protected_regular and fs.protected_fifos sysctls, which were
750 added in Linux 4.19 to make some data spoofing attacks harder, are
751 now enabled by default. While this will hopefully improve the
752 security of most installations, it is technically a backwards
753 incompatible change; to disable these sysctls again, place the
754 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
755
756 fs.protected_regular = 0
757 fs.protected_fifos = 0
758
759 Note that the similar hardlink and symlink protection has been
760 enabled since v199, and may be disabled likewise.
761
762 * The files read from the EnvironmentFile= setting in unit files now
763 parse backslashes inside quotes literally, matching the behaviour of
764 POSIX shells.
765
766 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
767 now automatically become NOPs when run in a chroot() environment.
768
769 * The tmpfiles.d/ "C" line type will now copy directory trees not only
770 when the destination is so far missing, but also if it already exists
771 as a directory and is empty. This is useful to cater for systems
772 where directory trees are put together from multiple separate mount
773 points but otherwise empty.
774
775 * A new function sd_bus_close_unref() (and the associated
776 sd_bus_close_unrefp()) has been added to libsystemd, that combines
777 sd_bus_close() and sd_bus_unref() in one.
778
779 * udevadm control learnt a new option for --ping for testing whether a
780 systemd-udevd instance is running and reacting.
781
782 * udevadm trigger learnt a new option for --wait-daemon for waiting
783 systemd-udevd daemon to be initialized.
784
785 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
786 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
787 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
788 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
789 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
790 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
791 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
792 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
793 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
794 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
795 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
796 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
797 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
798 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
799 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
800 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
801 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
802
803 — Berlin, 2019-02-14
804
805 CHANGES WITH 240:
806
807 * NoNewPrivileges=yes has been set for all long-running services
808 implemented by systemd. Previously, this was problematic due to
809 SELinux (as this would also prohibit the transition from PID1's label
810 to the service's label). This restriction has since been lifted, but
811 an SELinux policy update is required.
812 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
813
814 * DynamicUser=yes is dropped from systemd-networkd.service,
815 systemd-resolved.service and systemd-timesyncd.service, which was
816 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
817 and since v236 for systemd-timesyncd.service. The users and groups
818 systemd-network, systemd-resolve and systemd-timesync are created
819 by systemd-sysusers again. Distributors or system administrators
820 may need to create these users and groups if they not exist (or need
821 to re-enable DynamicUser= for those units) while upgrading systemd.
822 Also, the clock file for systemd-timesyncd may need to move from
823 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
824
825 * When unit files are loaded from disk, previously systemd would
826 sometimes (depending on the unit loading order) load units from the
827 target path of symlinks in .wants/ or .requires/ directories of other
828 units. This meant that unit could be loaded from different paths
829 depending on whether the unit was requested explicitly or as a
830 dependency of another unit, not honouring the priority of directories
831 in search path. It also meant that it was possible to successfully
832 load and start units which are not found in the unit search path, as
833 long as they were requested as a dependency and linked to from
834 .wants/ or .requires/. The target paths of those symlinks are not
835 used for loading units anymore and the unit file must be found in
836 the search path.
837
838 * A new service type has been added: Type=exec. It's very similar to
839 Type=simple but ensures the service manager will wait for both fork()
840 and execve() of the main service binary to complete before proceeding
841 with follow-up units. This is primarily useful so that the manager
842 propagates any errors in the preparation phase of service execution
843 back to the job that requested the unit to be started. For example,
844 consider a service that has ExecStart= set to a file system binary
845 that doesn't exist. With Type=simple starting the unit would be
846 considered instantly successful, as only fork() has to complete
847 successfully and the manager does not wait for execve(), and hence
848 its failure is seen "too late". With the new Type=exec service type
849 starting the unit will fail, as the manager will wait for the
850 execve() and notice its failure, which is then propagated back to the
851 start job.
852
853 NOTE: with the next release 241 of systemd we intend to change the
854 systemd-run tool to default to Type=exec for transient services
855 started by it. This should be mostly safe, but in specific corner
856 cases might result in problems, as the systemd-run tool will then
857 block on NSS calls (such as user name look-ups due to User=) done
858 between the fork() and execve(), which under specific circumstances
859 might cause problems. It is recommended to specify "-p Type=simple"
860 explicitly in the few cases where this applies. For regular,
861 non-transient services (i.e. those defined with unit files on disk)
862 we will continue to default to Type=simple.
863
864 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
865 userspace processes is set to 1024 (soft) and 4096
866 (hard). Previously, systemd passed this on unmodified to all
867 processes it forked off. With this systemd release the hard limit
868 systemd passes on is increased to 512K, overriding the kernel's
869 defaults and substantially increasing the number of simultaneous file
870 descriptors unprivileged userspace processes can allocate. Note that
871 the soft limit remains at 1024 for compatibility reasons: the
872 traditional UNIX select() call cannot deal with file descriptors >=
873 1024 and increasing the soft limit globally might thus result in
874 programs unexpectedly allocating a high file descriptor and thus
875 failing abnormally when attempting to use it with select() (of
876 course, programs shouldn't use select() anymore, and prefer
877 poll()/epoll, but the call unfortunately remains undeservedly popular
878 at this time). This change reflects the fact that file descriptor
879 handling in the Linux kernel has been optimized in more recent
880 kernels and allocating large numbers of them should be much cheaper
881 both in memory and in performance than it used to be. Programs that
882 want to take benefit of the increased limit have to "opt-in" into
883 high file descriptors explicitly by raising their soft limit. Of
884 course, when they do that they must acknowledge that they cannot use
885 select() anymore (and neither can any shared library they use — or
886 any shared library used by any shared library they use and so on).
887 Which default hard limit is most appropriate is of course hard to
888 decide. However, given reports that ~300K file descriptors are used
889 in real-life applications we believe 512K is sufficiently high as new
890 default for now. Note that there are also reports that using very
891 high hard limits (e.g. 1G) is problematic: some software allocates
892 large arrays with one element for each potential file descriptor
893 (Java, …) — a high hard limit thus triggers excessively large memory
894 allocations in these applications. Hopefully, the new default of 512K
895 is a good middle ground: higher than what real-life applications
896 currently need, and low enough for avoid triggering excessively large
897 allocations in problematic software. (And yes, somebody should fix
898 Java.)
899
900 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
901 to the highest possible values, as separate accounting of file
902 descriptors is no longer necessary, as memcg tracks them correctly as
903 part of the memory accounting anyway. Thus, from the four limits on
904 file descriptors currently enforced (fs.file-max, fs.nr_open,
905 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
906 and keep only the latter two. A set of build-time options
907 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
908 has been added to revert this change in behaviour, which might be
909 an option for systems that turn off memcg in the kernel.
910
911 * When no /etc/locale.conf file exists (and hence no locale settings
912 are in place), systemd will now use the "C.UTF-8" locale by default,
913 and set LANG= to it. This locale is supported by various
914 distributions including Fedora, with clear indications that upstream
915 glibc is going to make it available too. This locale enables UTF-8
916 mode by default, which appears appropriate for 2018.
917
918 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
919 default. This effectively switches the RFC3704 Reverse Path filtering
920 from Strict mode to Loose mode. This is more appropriate for hosts
921 that have multiple links with routes to the same networks (e.g.
922 a client with a Wi-Fi and Ethernet both connected to the internet).
923
924 Consult the kernel documentation for details on this sysctl:
925 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
926
927 * CPUAccounting=yes no longer enables the CPU controller when using
928 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
929 statistics are now provided independently from the CPU controller.
930
931 * Support for disabling a particular cgroup controller within a sub-tree
932 has been added through the DisableControllers= directive.
933
934 * cgroup_no_v1=all on the kernel command line now also implies
935 using the unified cgroup hierarchy, unless one explicitly passes
936 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
937
938 * The new "MemoryMin=" unit file property may now be used to set the
939 memory usage protection limit of processes invoked by the unit. This
940 controls the cgroup v2 memory.min attribute. Similarly, the new
941 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
942 cgroup v2 io.latency cgroup property for configuring per-service I/O
943 latency.
944
945 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
946 to the cgroup v1 "devices" cgroup controller.
947
948 * systemd-escape now is able to combine --unescape with --template. It
949 also learnt a new option --instance for extracting and unescaping the
950 instance part of a unit name.
951
952 * sd-bus now provides the sd_bus_message_readv() which is similar to
953 sd_bus_message_read() but takes a va_list object. The pair
954 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
955 has been added for configuring the default method call timeout to
956 use. sd_bus_error_move() may be used to efficiently move the contents
957 from one sd_bus_error structure to another, invalidating the
958 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
959 be used to control whether a bus connection object is automatically
960 flushed when an sd-event loop is exited.
961
962 * When processing classic BSD syslog log messages, journald will now
963 save the original time-stamp string supplied in the new
964 SYSLOG_TIMESTAMP= journal field. This permits consumers to
965 reconstruct the original BSD syslog message more correctly.
966
967 * StandardOutput=/StandardError= in service files gained support for
968 new "append:…" parameters, for connecting STDOUT/STDERR of a service
969 to a file, and appending to it.
970
971 * The signal to use as last step of killing of unit processes is now
972 configurable. Previously it was hard-coded to SIGKILL, which may now
973 be overridden with the new KillSignal= setting. Note that this is the
974 signal used when regular termination (i.e. SIGTERM) does not suffice.
975 Similarly, the signal used when aborting a program in case of a
976 watchdog timeout may now be configured too (WatchdogSignal=).
977
978 * The XDG_SESSION_DESKTOP environment variable may now be configured in
979 the pam_systemd argument line, using the new desktop= switch. This is
980 useful to initialize it properly from a display manager without
981 having to touch C code.
982
983 * Most configuration options that previously accepted percentage values
984 now also accept permille values with the '‰' suffix (instead of '%').
985
986 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
987 DNS-over-TLS.
988
989 * systemd-resolved's configuration file resolved.conf gained a new
990 option ReadEtcHosts= which may be used to turn off processing and
991 honoring /etc/hosts entries.
992
993 * The "--wait" switch may now be passed to "systemctl
994 is-system-running", in which case the tool will synchronously wait
995 until the system finished start-up.
996
997 * hostnamed gained a new bus call to determine the DMI product UUID.
998
999 * On x86-64 systemd will now prefer using the RDRAND processor
1000 instruction over /dev/urandom whenever it requires randomness that
1001 neither has to be crypto-grade nor should be reproducible. This
1002 should substantially reduce the amount of entropy systemd requests
1003 from the kernel during initialization on such systems, though not
1004 reduce it to zero. (Why not zero? systemd still needs to allocate
1005 UUIDs and such uniquely, which require high-quality randomness.)
1006
1007 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
1008 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
1009 for forcing the "Other Information" bit in IPv6 RA messages. The
1010 bonding logic gained four new options AdActorSystemPriority=,
1011 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
1012 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
1013 shuffling of flows. The tunnel logic gained a new
1014 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
1015 Deployment. The policy rule logic gained four new options IPProtocol=,
1016 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
1017 support for the MulticastToUnicast= option. networkd also gained
1018 support for configuring static IPv4 ARP or IPv6 neighbor entries.
1019
1020 * .preset files (as read by 'systemctl preset') may now be used to
1021 instantiate services.
1022
1023 * /etc/crypttab now understands the sector-size= option to configure
1024 the sector size for an encrypted partition.
1025
1026 * Key material for encrypted disks may now be placed on a formatted
1027 medium, and referenced from /etc/crypttab by the UUID of the file
1028 system, followed by "=" suffixed by the path to the key file.
1029
1030 * The "collect" udev component has been removed without replacement, as
1031 it is neither used nor maintained.
1032
1033 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
1034 LogsDirectory=, ConfigurationDirectory= settings are used in a
1035 service the executed processes will now receive a set of environment
1036 variables containing the full paths of these directories.
1037 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
1038 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
1039 are used. Note that these options may be used multiple times per
1040 service in which case the resulting paths will be concatenated and
1041 separated by colons.
1042
1043 * Predictable interface naming has been extended to cover InfiniBand
1044 NICs. They will be exposed with an "ib" prefix.
1045
1046 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
1047 which case the respective line failing is ignored.
1048
1049 * .link files may now be used to configure the equivalent to the
1050 "ethtool advertise" commands.
1051
1052 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
1053 alternative to libudev.h. Previously, the latter was just an internal
1054 wrapper around the former, but now these two APIs are exposed
1055 directly.
1056
1057 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
1058 which calculates an app-specific boot ID similar to how
1059 sd_id128_get_machine_app_specific() generates an app-specific machine
1060 ID.
1061
1062 * A new tool systemd-id128 has been added that can be used to determine
1063 and generate various 128bit IDs.
1064
1065 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
1066 and LOGO=.
1067
1068 * systemd-hibernate-resume-generator will now honor the "noresume"
1069 kernel command line option, in which case it will bypass resuming
1070 from any hibernated image.
1071
1072 * The systemd-sleep.conf configuration file gained new options
1073 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
1074 AllowHybridSleep= for prohibiting specific sleep modes even if the
1075 kernel exports them.
1076
1077 * portablectl is now officially supported and has thus moved to
1078 /usr/bin/.
1079
1080 * bootctl learnt the two new commands "set-default" and "set-oneshot"
1081 for setting the default boot loader item to boot to (either
1082 persistently or only for the next boot). This is currently only
1083 compatible with sd-boot, but may be implemented on other boot loaders
1084 too, that follow the boot loader interface. The updated interface is
1085 now documented here:
1086
1087 https://systemd.io/BOOT_LOADER_INTERFACE
1088
1089 * A new kernel command line option systemd.early_core_pattern= is now
1090 understood which may be used to influence the core_pattern PID 1
1091 installs during early boot.
1092
1093 * busctl learnt two new options -j and --json= for outputting method
1094 call replies, properties and monitoring output in JSON.
1095
1096 * journalctl's JSON output now supports simple ANSI coloring as well as
1097 a new "json-seq" mode for generating RFC7464 output.
1098
1099 * Unit files now support the %g/%G specifiers that resolve to the UNIX
1100 group/GID of the service manager runs as, similar to the existing
1101 %u/%U specifiers that resolve to the UNIX user/UID.
1102
1103 * systemd-logind learnt a new global configuration option
1104 UserStopDelaySec= that may be set in logind.conf. It specifies how
1105 long the systemd --user instance shall remain started after a user
1106 logs out. This is useful to speed up repetitive re-connections of the
1107 same user, as it means the user's service manager doesn't have to be
1108 stopped/restarted on each iteration, but can be reused between
1109 subsequent options. This setting defaults to 10s. systemd-logind also
1110 exports two new properties on its Manager D-Bus objects indicating
1111 whether the system's lid is currently closed, and whether the system
1112 is on AC power.
1113
1114 * systemd gained support for a generic boot counting logic, which
1115 generically permits automatic reverting to older boot loader entries
1116 if newer updated ones don't work. The boot loader side is implemented
1117 in sd-boot, but is kept open for other boot loaders too. For details
1118 see:
1119
1120 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
1121
1122 * The SuccessAction=/FailureAction= unit file settings now learnt two
1123 new parameters: "exit" and "exit-force", which result in immediate
1124 exiting of the service manager, and are only useful in systemd --user
1125 and container environments.
1126
1127 * Unit files gained support for a pair of options
1128 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
1129 exit status to use as service manager exit status when
1130 SuccessAction=/FailureAction= is set to exit or exit-force.
1131
1132 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
1133 options may now be used to configure the log rate limiting applied by
1134 journald per-service.
1135
1136 * systemd-analyze gained a new verb "timespan" for parsing and
1137 normalizing time span values (i.e. strings like "5min 7s 8us").
1138
1139 * systemd-analyze also gained a new verb "security" for analyzing the
1140 security and sand-boxing settings of services in order to determine an
1141 "exposure level" for them, indicating whether a service would benefit
1142 from more sand-boxing options turned on for them.
1143
1144 * "systemd-analyze syscall-filter" will now also show system calls
1145 supported by the local kernel but not included in any of the defined
1146 groups.
1147
1148 * .nspawn files now understand the Ephemeral= setting, matching the
1149 --ephemeral command line switch.
1150
1151 * sd-event gained the new APIs sd_event_source_get_floating() and
1152 sd_event_source_set_floating() for controlling whether a specific
1153 event source is "floating", i.e. destroyed along with the even loop
1154 object itself.
1155
1156 * Unit objects on D-Bus gained a new "Refs" property that lists all
1157 clients that currently have a reference on the unit (to ensure it is
1158 not unloaded).
1159
1160 * The JoinControllers= option in system.conf is no longer supported, as
1161 it didn't work correctly, is hard to support properly, is legacy (as
1162 the concept only exists on cgroup v1) and apparently wasn't used.
1163
1164 * Journal messages that are generated whenever a unit enters the failed
1165 state are now tagged with a unique MESSAGE_ID. Similarly, messages
1166 generated whenever a service process exits are now made recognizable,
1167 too. A tagged message is also emitted whenever a unit enters the
1168 "dead" state on success.
1169
1170 * systemd-run gained a new switch --working-directory= for configuring
1171 the working directory of the service to start. A shortcut -d is
1172 equivalent, setting the working directory of the service to the
1173 current working directory of the invoking program. The new --shell
1174 (or just -S) option has been added for invoking the $SHELL of the
1175 caller as a service, and implies --pty --same-dir --wait --collect
1176 --service-type=exec. Or in other words, "systemd-run -S" is now the
1177 quickest way to quickly get an interactive in a fully clean and
1178 well-defined system service context.
1179
1180 * machinectl gained a new verb "import-fs" for importing an OS tree
1181 from a directory. Moreover, when a directory or tarball is imported
1182 and single top-level directory found with the OS itself below the OS
1183 tree is automatically mangled and moved one level up.
1184
1185 * systemd-importd will no longer set up an implicit btrfs loop-back
1186 file system on /var/lib/machines. If one is already set up, it will
1187 continue to be used.
1188
1189 * A new generator "systemd-run-generator" has been added. It will
1190 synthesize a unit from one or more program command lines included in
1191 the kernel command line. This is very useful in container managers
1192 for example:
1193
1194 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
1195
1196 This will run "systemd-nspawn" on an image, invoke the specified
1197 command line and immediately shut down the container again, returning
1198 the command line's exit code.
1199
1200 * The block device locking logic is now documented:
1201
1202 https://systemd.io/BLOCK_DEVICE_LOCKING
1203
1204 * loginctl and machinectl now optionally output the various tables in
1205 JSON using the --output= switch. It is our intention to add similar
1206 support to systemctl and all other commands.
1207
1208 * udevadm's query and trigger verb now optionally take a .device unit
1209 name as argument.
1210
1211 * systemd-udevd's network naming logic now understands a new
1212 net.naming-scheme= kernel command line switch, which may be used to
1213 pick a specific version of the naming scheme. This helps stabilizing
1214 interface names even as systemd/udev are updated and the naming logic
1215 is improved.
1216
1217 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
1218 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
1219 initialize one to all 0xFF.
1220
1221 * After loading the SELinux policy systemd will now recursively relabel
1222 all files and directories listed in
1223 /run/systemd/relabel-extra.d/*.relabel (which should be simple
1224 newline separated lists of paths) in addition to the ones it already
1225 implicitly relabels in /run, /dev and /sys. After the relabelling is
1226 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
1227 removed. This is useful to permit initrds (i.e. code running before
1228 the SELinux policy is in effect) to generate files in the host
1229 filesystem safely and ensure that the correct label is applied during
1230 the transition to the host OS.
1231
1232 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
1233 mknod() handling in user namespaces. Previously mknod() would always
1234 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
1235 but device nodes generated that way cannot be opened, and attempts to
1236 open them result in EPERM. This breaks the "graceful fallback" logic
1237 in systemd's PrivateDevices= sand-boxing option. This option is
1238 implemented defensively, so that when systemd detects it runs in a
1239 restricted environment (such as a user namespace, or an environment
1240 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
1241 where device nodes cannot be created the effect of PrivateDevices= is
1242 bypassed (following the logic that 2nd-level sand-boxing is not
1243 essential if the system systemd runs in is itself already sand-boxed
1244 as a whole). This logic breaks with 4.18 in container managers where
1245 user namespacing is used: suddenly PrivateDevices= succeeds setting
1246 up a private /dev/ file system containing devices nodes — but when
1247 these are opened they don't work.
1248
1249 At this point it is recommended that container managers utilizing
1250 user namespaces that intend to run systemd in the payload explicitly
1251 block mknod() with seccomp or similar, so that the graceful fallback
1252 logic works again.
1253
1254 We are very sorry for the breakage and the requirement to change
1255 container configurations for newer kernels. It's purely caused by an
1256 incompatible kernel change. The relevant kernel developers have been
1257 notified about this userspace breakage quickly, but they chose to
1258 ignore it.
1259
1260 * PermissionsStartOnly= setting is deprecated (but is still supported
1261 for backwards compatibility). The same functionality is provided by
1262 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
1263 commands.
1264
1265 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
1266 pam_systemd anymore.
1267
1268 * The naming scheme for network devices was changed to always rename
1269 devices, even if they were already renamed by userspace. The "kernel"
1270 policy was changed to only apply as a fallback, if no other naming
1271 policy took effect.
1272
1273 * The requirements to build systemd is bumped to meson-0.46 and
1274 python-3.5.
1275
1276 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
1277 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
1278 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
1279 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
1280 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
1281 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
1282 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
1283 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
1284 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
1285 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
1286 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
1287 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
1288 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
1289 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
1290 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
1291 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
1292 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
1293 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
1294 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
1295 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
1296 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
1297 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
1298 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
1299 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
1300 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
1301 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
1302 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
1303 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
1304 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
1305 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
1306 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
1307 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
1308 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
1309 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
1310 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
1311 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
1312 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
1313 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
1314 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
1315 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
1316 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
1317 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
1318 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
1319 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
1320 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
1321
1322 — Warsaw, 2018-12-21
1323
1324 CHANGES WITH 239:
1325
1326 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
1327 builtin will name network interfaces differently than in previous
1328 versions for virtual network interfaces created with SR-IOV and NPAR
1329 and for devices where the PCI network controller device does not have
1330 a slot number associated.
1331
1332 SR-IOV virtual devices are now named based on the name of the parent
1333 interface, with a suffix of "v<N>", where <N> is the virtual device
1334 number. Previously those virtual devices were named as if completely
1335 independent.
1336
1337 The ninth and later NPAR virtual devices will be named following the
1338 scheme used for the first eight NPAR partitions. Previously those
1339 devices were not renamed and the kernel default (eth<n>) was used.
1340
1341 "net_id" will also generate names for PCI devices where the PCI
1342 network controller device does not have an associated slot number
1343 itself, but one of its parents does. Previously those devices were
1344 not renamed and the kernel default (eth<n>) was used.
1345
1346 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
1347 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
1348 the unit. So, it is expected that the default behavior of
1349 systemd-logind is not changed. However, if distribution packagers or
1350 administrators disabled or modified IPAddressDeny= setting by a
1351 drop-in config file, then it may be necessary to update the file to
1352 re-enable AF_INET and AF_INET6 to support network user name services,
1353 e.g. NIS.
1354
1355 * When the RestrictNamespaces= unit property is specified multiple
1356 times, then the specified types are merged now. Previously, only the
1357 last assignment was used. So, if distribution packagers or
1358 administrators modified the setting by a drop-in config file, then it
1359 may be necessary to update the file.
1360
1361 * When OnFailure= is used in combination with Restart= on a service
1362 unit, then the specified units will no longer be triggered on
1363 failures that result in restarting. Previously, the specified units
1364 would be activated each time the unit failed, even when the unit was
1365 going to be restarted automatically. This behaviour contradicted the
1366 documentation. With this release the code is adjusted to match the
1367 documentation.
1368
1369 * systemd-tmpfiles will now print a notice whenever it encounters
1370 tmpfiles.d/ lines referencing the /var/run/ directory. It will
1371 recommend reworking them to use the /run/ directory instead (for
1372 which /var/run/ is simply a symlinked compatibility alias). This way
1373 systemd-tmpfiles can properly detect line conflicts and merge lines
1374 referencing the same file by two paths, without having to access
1375 them.
1376
1377 * systemctl disable/unmask/preset/preset-all cannot be used with
1378 --runtime. Previously this was allowed, but resulted in unintuitive
1379 behaviour that wasn't useful. systemctl disable/unmask will now undo
1380 both runtime and persistent enablement/masking, i.e. it will remove
1381 any relevant symlinks both in /run and /etc.
1382
1383 * Note that all long-running system services shipped with systemd will
1384 now default to a system call whitelist (rather than a blacklist, as
1385 before). In particular, systemd-udevd will now enforce one too. For
1386 most cases this should be safe, however downstream distributions
1387 which disabled sandboxing of systemd-udevd (specifically the
1388 MountFlags= setting), might want to disable this security feature
1389 too, as the default whitelisting will prohibit all mount, swap,
1390 reboot and clock changing operations from udev rules.
1391
1392 * sd-boot acquired new loader configuration settings to optionally turn
1393 off Windows and MacOS boot partition discovery as well as
1394 reboot-into-firmware menu items. It is also able to pick a better
1395 screen resolution for HiDPI systems, and now provides loader
1396 configuration settings to change the resolution explicitly.
1397
1398 * systemd-resolved now supports DNS-over-TLS. It's still
1399 turned off by default, use DNSOverTLS=opportunistic to turn it on in
1400 resolved.conf. We intend to make this the default as soon as couple
1401 of additional techniques for optimizing the initial latency caused by
1402 establishing a TLS/TCP connection are implemented.
1403
1404 * systemd-resolved.service and systemd-networkd.service now set
1405 DynamicUser=yes. The users systemd-resolve and systemd-network are
1406 not created by systemd-sysusers anymore.
1407
1408 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
1409 that embed a network facing module into any process using getpwuid()
1410 or related call: the dynamic allocation of the user ID for
1411 systemd-resolved.service means the service manager has to check NSS
1412 if the user name is already taken when forking off the service. Since
1413 the user in the common case won't be defined in /etc/passwd the
1414 lookup is likely to trigger nss-ldap which in turn might use NSS to
1415 ask systemd-resolved for hostname lookups. This will hence result in
1416 a deadlock: a user name lookup in order to start
1417 systemd-resolved.service will result in a host name lookup for which
1418 systemd-resolved.service needs to be started already. There are
1419 multiple ways to work around this problem: pre-allocate the
1420 "systemd-resolve" user on such systems, so that nss-ldap won't be
1421 triggered; or use a different NSS package that doesn't do networking
1422 in-process but provides a local asynchronous name cache; or configure
1423 the NSS package to avoid lookups for UIDs in the range `pkg-config
1424 systemd --variable=dynamicuidmin` … `pkg-config systemd
1425 --variable=dynamicuidmax`, so that it does not consider itself
1426 authoritative for the same UID range systemd allocates dynamic users
1427 from.
1428
1429 * The systemd-resolve tool has been renamed to resolvectl (it also
1430 remains available under the old name, for compatibility), and its
1431 interface is now verb-based, similar in style to the other <xyz>ctl
1432 tools, such as systemctl or loginctl.
1433
1434 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
1435 compatibility. It may be symlinked under the 'resolvconf' name, in
1436 which case it will take arguments and input compatible with the
1437 Debian and FreeBSD resolvconf tool.
1438
1439 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
1440 where the system initially suspends, and after a timeout resumes and
1441 hibernates again.
1442
1443 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
1444 set the client will only send a DUID as client identifier.
1445
1446 * The nss-systemd glibc NSS module will now enumerate dynamic users and
1447 groups in effect. Previously, it could resolve UIDs/GIDs to user
1448 names/groups and vice versa, but did not support enumeration.
1449
1450 * journald's Compress= configuration setting now optionally accepts a
1451 byte threshold value. All journal objects larger than this threshold
1452 will be compressed, smaller ones will not. Previously this threshold
1453 was not configurable and set to 512.
1454
1455 * A new system.conf setting NoNewPrivileges= is now available which may
1456 be used to turn off acquisition of new privileges system-wide
1457 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
1458 for all its children). Note that turning this option on means setuid
1459 binaries and file system capabilities lose their special powers.
1460 While turning on this option is a big step towards a more secure
1461 system, doing so is likely to break numerous pre-existing UNIX tools,
1462 in particular su and sudo.
1463
1464 * A new service systemd-time-sync-wait.service has been added. If
1465 enabled it will delay the time-sync.target unit at boot until time
1466 synchronization has been received from the network. This
1467 functionality is useful on systems lacking a local RTC or where it is
1468 acceptable that the boot process shall be delayed by external network
1469 services.
1470
1471 * When hibernating, systemd will now inform the kernel of the image
1472 write offset, on kernels new enough to support this. This means swap
1473 files should work for hibernation now.
1474
1475 * When loading unit files, systemd will now look for drop-in unit files
1476 extensions in additional places. Previously, for a unit file name
1477 "foo-bar-baz.service" it would look for dropin files in
1478 "foo-bar-baz.service.d/*.conf". Now, it will also look in
1479 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
1480 service name truncated after all inner dashes. This scheme allows
1481 writing drop-ins easily that apply to a whole set of unit files at
1482 once. It's particularly useful for mount and slice units (as their
1483 naming is prefix based), but is also useful for service and other
1484 units, for packages that install multiple unit files at once,
1485 following a strict naming regime of beginning the unit file name with
1486 the package's name. Two new specifiers are now supported in unit
1487 files to match this: %j and %J are replaced by the part of the unit
1488 name following the last dash.
1489
1490 * Unit files and other configuration files that support specifier
1491 expansion now understand another three new specifiers: %T and %V will
1492 resolve to /tmp and /var/tmp respectively, or whatever temporary
1493 directory has been set for the calling user. %E will expand to either
1494 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
1495
1496 * The ExecStart= lines of unit files are no longer required to
1497 reference absolute paths. If non-absolute paths are specified the
1498 specified binary name is searched within the service manager's
1499 built-in $PATH, which may be queried with 'systemd-path
1500 search-binaries-default'. It's generally recommended to continue to
1501 use absolute paths for all binaries specified in unit files.
1502
1503 * Units gained a new load state "bad-setting", which is used when a
1504 unit file was loaded, but contained fatal errors which prevent it
1505 from being started (for example, a service unit has been defined
1506 lacking both ExecStart= and ExecStop= lines).
1507
1508 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
1509 support alternative debuggers, for example lldb. The old name
1510 continues to be available however, for compatibility reasons. Use the
1511 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
1512 to pick an alternative debugger instead of the default gdb.
1513
1514 * systemctl and the other tools will now output escape sequences that
1515 generate proper clickable hyperlinks in various terminal emulators
1516 where useful (for example, in the "systemctl status" output you can
1517 now click on the unit file name to quickly open it in the
1518 editor/viewer of your choice). Note that not all terminal emulators
1519 support this functionality yet, but many do. Unfortunately, the
1520 "less" pager doesn't support this yet, hence this functionality is
1521 currently automatically turned off when a pager is started (which
1522 happens quite often due to auto-paging). We hope to remove this
1523 limitation as soon as "less" learns these escape sequences. This new
1524 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
1525 environment variable. For details on these escape sequences see:
1526 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
1527
1528 * networkd's .network files now support a new IPv6MTUBytes= option for
1529 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
1530 option in the [Route] section to configure the MTU to use for
1531 specific routes. It also gained support for configuration of the DHCP
1532 "UserClass" option through the new UserClass= setting. It gained
1533 three new options in the new [CAN] section for configuring CAN
1534 networks. The MULTICAST and ALLMULTI interface flags may now be
1535 controlled explicitly with the new Multicast= and AllMulticast=
1536 settings.
1537
1538 * networkd will now automatically make use of the kernel's route
1539 expiration feature, if it is available.
1540
1541 * udevd's .link files now support setting the number of receive and
1542 transmit channels, using the RxChannels=, TxChannels=,
1543 OtherChannels=, CombinedChannels= settings.
1544
1545 * Support for UDPSegmentationOffload= has been removed, given its
1546 limited support in hardware, and waning software support.
1547
1548 * networkd's .netdev files now support creating "netdevsim" interfaces.
1549
1550 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
1551 to query the unit belonging to a specific kernel control group.
1552
1553 * systemd-analyze gained a new verb "cat-config", which may be used to
1554 dump the contents of any configuration file, with all its matching
1555 drop-in files added in, and honouring the usual search and masking
1556 logic applied to systemd configuration files. For example use
1557 "systemd-analyze cat-config systemd/system.conf" to get the complete
1558 system configuration file of systemd how it would be loaded by PID 1
1559 itself. Similar to this, various tools such as systemd-tmpfiles or
1560 systemd-sysusers, gained a new option "--cat-config", which does the
1561 corresponding operation for their own configuration settings. For
1562 example, "systemd-tmpfiles --cat-config" will now output the full
1563 list of tmpfiles.d/ lines in place.
1564
1565 * timedatectl gained three new verbs: "show" shows bus properties of
1566 systemd-timedated, "timesync-status" shows the current NTP
1567 synchronization state of systemd-timesyncd, and "show-timesync"
1568 shows bus properties of systemd-timesyncd.
1569
1570 * systemd-timesyncd gained a bus interface on which it exposes details
1571 about its state.
1572
1573 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
1574 understood by systemd-timedated. It takes a colon-separated list of
1575 unit names of NTP client services. The list is used by
1576 "timedatectl set-ntp".
1577
1578 * systemd-nspawn gained a new --rlimit= switch for setting initial
1579 resource limits for the container payload. There's a new switch
1580 --hostname= to explicitly override the container's hostname. A new
1581 --no-new-privileges= switch may be used to control the
1582 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
1583 --oom-score-adjust= switch controls the OOM scoring adjustment value
1584 for the payload. The new --cpu-affinity= switch controls the CPU
1585 affinity of the container payload. The new --resolv-conf= switch
1586 allows more detailed control of /etc/resolv.conf handling of the
1587 container. Similarly, the new --timezone= switch allows more detailed
1588 control of /etc/localtime handling of the container.
1589
1590 * systemd-detect-virt gained a new --list switch, which will print a
1591 list of all currently known VM and container environments.
1592
1593 * Support for "Portable Services" has been added, see
1594 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
1595 experimental, but this is expected to change soon. Reflecting this
1596 experimental state, the "portablectl" binary is not installed into
1597 /usr/bin yet. The binary has to be called with the full path
1598 /usr/lib/systemd/portablectl instead.
1599
1600 * journalctl's and systemctl's -o switch now knows a new log output
1601 mode "with-unit". The output it generates is very similar to the
1602 regular "short" mode, but displays the unit name instead of the
1603 syslog tag for each log line. Also, the date is shown with timezone
1604 information. This mode is probably more useful than the classic
1605 "short" output mode for most purposes, except where pixel-perfect
1606 compatibility with classic /var/log/messages formatting is required.
1607
1608 * A new --dump-bus-properties switch has been added to the systemd
1609 binary, which may be used to dump all supported D-Bus properties.
1610 (Options which are still supported, but are deprecated, are *not*
1611 shown.)
1612
1613 * sd-bus gained a set of new calls:
1614 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
1615 enable/disable the "floating" state of a bus slot object,
1616 i.e. whether the slot object pins the bus it is allocated for into
1617 memory or if the bus slot object gets disconnected when the bus goes
1618 away. sd_bus_open_with_description(),
1619 sd_bus_open_user_with_description(),
1620 sd_bus_open_system_with_description() may be used to allocate bus
1621 objects and set their description string already during allocation.
1622
1623 * sd-event gained support for watching inotify events from the event
1624 loop, in an efficient way, sharing inotify handles between multiple
1625 users. For this a new function sd_event_add_inotify() has been added.
1626
1627 * sd-event and sd-bus gained support for calling special user-supplied
1628 destructor functions for userdata pointers associated with
1629 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
1630 functions sd_bus_slot_set_destroy_callback,
1631 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
1632 sd_bus_track_get_destroy_callback,
1633 sd_event_source_set_destroy_callback,
1634 sd_event_source_get_destroy_callback have been added.
1635
1636 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
1637
1638 * PID 1 will now automatically reschedule .timer units whenever the
1639 local timezone changes. (They previously got rescheduled
1640 automatically when the system clock changed.)
1641
1642 * New documentation has been added to document cgroups delegation,
1643 portable services and the various code quality tools we have set up:
1644
1645 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
1646 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
1647 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
1648
1649 * The Boot Loader Specification has been added to the source tree.
1650
1651 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
1652
1653 While moving it into our source tree we have updated it and further
1654 changes are now accepted through the usual github PR workflow.
1655
1656 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
1657 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
1658 earlier PAM modules. The data in these fields is used to initialize
1659 the session scope's resource properties. Thus external PAM modules
1660 may now configure per-session limits, for example sourced from
1661 external user databases.
1662
1663 * socket units with Accept=yes will now maintain a "refused" counter in
1664 addition to the existing "accepted" counter, counting connections
1665 refused due to the enforced limits.
1666
1667 * The "systemd-path search-binaries-default" command may now be use to
1668 query the default, built-in $PATH PID 1 will pass to the services it
1669 manages.
1670
1671 * A new unit file setting PrivateMounts= has been added. It's a boolean
1672 option. If enabled the unit's processes are invoked in their own file
1673 system namespace. Note that this behaviour is also implied if any
1674 other file system namespacing options (such as PrivateTmp=,
1675 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
1676 primarily useful for services that do not use any of the other file
1677 system namespacing options. One such service is systemd-udevd.service
1678 where this is now used by default.
1679
1680 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
1681 when the system is booted in UEFI "secure mode".
1682
1683 * A new unit "system-update-pre.target" is added, which defines an
1684 optional synchronization point for offline system updates, as
1685 implemented by the pre-existing "system-update.target" unit. It
1686 allows ordering services before the service that executes the actual
1687 update process in a generic way.
1688
1689 * Systemd now emits warnings whenever .include syntax is used.
1690
1691 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
1692 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
1693 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
1694 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
1695 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
1696 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
1697 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
1698 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
1699 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
1700 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
1701 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
1702 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
1703 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
1704 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
1705 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
1706 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
1707 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
1708 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
1709 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
1710 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
1711 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
1712 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
1713 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
1714 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
1715 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
1716 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
1717 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
1718 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
1719 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1720
1721 — Berlin, 2018-06-22
1722
1723 CHANGES WITH 238:
1724
1725 * The MemoryAccounting= unit property now defaults to on. After
1726 discussions with the upstream control group maintainers we learnt
1727 that the negative impact of cgroup memory accounting on current
1728 kernels is finally relatively minimal, so that it should be safe to
1729 enable this by default without affecting system performance. Besides
1730 memory accounting only task accounting is turned on by default, all
1731 other forms of resource accounting (CPU, IO, IP) remain off for now,
1732 because it's not clear yet that their impact is small enough to move
1733 from opt-in to opt-out. We recommend downstreams to leave memory
1734 accounting on by default if kernel 4.14 or higher is primarily
1735 used. On very resource constrained systems or when support for old
1736 kernels is a necessity, -Dmemory-accounting-default=false can be used
1737 to revert this change.
1738
1739 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
1740 %udev_rules_update) and the journal catalog (%journal_catalog_update)
1741 from the upgrade scriptlets of individual packages now do nothing.
1742 Transfiletriggers have been added which will perform those updates
1743 once at the end of the transaction.
1744
1745 Similar transfiletriggers have been added to execute any sysctl.d
1746 and binfmt.d rules. Thus, it should be unnecessary to provide any
1747 scriptlets to execute this configuration from package installation
1748 scripts.
1749
1750 * systemd-sysusers gained a mode where the configuration to execute is
1751 specified on the command line, but this configuration is not executed
1752 directly, but instead it is merged with the configuration on disk,
1753 and the result is executed. This is useful for package installation
1754 scripts which want to create the user before installing any files on
1755 disk (in case some of those files are owned by that user), while
1756 still allowing local admin overrides.
1757
1758 This functionality is exposed to rpm scriptlets through a new
1759 %sysusers_create_package macro. Old %sysusers_create and
1760 %sysusers_create_inline macros are deprecated.
1761
1762 A transfiletrigger for sysusers.d configuration is now installed,
1763 which means that it should be unnecessary to call systemd-sysusers from
1764 package installation scripts, unless the package installs any files
1765 owned by those newly-created users, in which case
1766 %sysusers_create_package should be used.
1767
1768 * Analogous change has been done for systemd-tmpfiles: it gained a mode
1769 where the command-line configuration is merged with the configuration
1770 on disk. This is exposed as the new %tmpfiles_create_package macro,
1771 and %tmpfiles_create is deprecated. A transfiletrigger is installed
1772 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
1773 from package installation scripts.
1774
1775 * sysusers.d configuration for a user may now also specify the group
1776 number, in addition to the user number ("u username 123:456"), or
1777 without the user number ("u username -:456").
1778
1779 * Configution items for systemd-sysusers can now be specified as
1780 positional arguments when the new --inline switch is used.
1781
1782 * The login shell of users created through sysusers.d may now be
1783 specified (previously, it was always /bin/sh for root and
1784 /sbin/nologin for other users).
1785
1786 * systemd-analyze gained a new --global switch to look at global user
1787 configuration. It also gained a unit-paths verb to list the unit load
1788 paths that are compiled into systemd (which can be used with
1789 --systemd, --user, or --global).
1790
1791 * udevadm trigger gained a new --settle/-w option to wait for any
1792 triggered events to finish (but just those, and not any other events
1793 which are triggered meanwhile).
1794
1795 * The action that systemd-logind takes when the lid is closed and the
1796 machine is connected to external power can now be configured using
1797 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
1798 was determined by HandleLidSwitch=, and, for backwards compatibility,
1799 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
1800
1801 * journalctl will periodically call sd_journal_process() to make it
1802 resilient against inotify queue overruns when journal files are
1803 rotated very quickly.
1804
1805 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
1806 sd_bus_get_n_queued_write — may be used to check the number of
1807 pending bus messages.
1808
1809 * systemd gained a new
1810 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
1811 which can be used to migrate foreign processes to scope and service
1812 units. The primary user for this new API is systemd itself: the
1813 systemd --user instance uses this call of the systemd --system
1814 instance to migrate processes if it itself gets the request to
1815 migrate processes and the kernel refuses this due to access
1816 restrictions. Thanks to this "systemd-run --scope --user …" works
1817 again in pure cgroup v2 environments when invoked from the user
1818 session scope.
1819
1820 * A new TemporaryFileSystem= setting can be used to mask out part of
1821 the real file system tree with tmpfs mounts. This may be combined
1822 with BindPaths= and BindReadOnlyPaths= to hide files or directories
1823 not relevant to the unit, while still allowing some paths lower in
1824 the tree to be accessed.
1825
1826 ProtectHome=tmpfs may now be used to hide user home and runtime
1827 directories from units, in a way that is mostly equivalent to
1828 "TemporaryFileSystem=/home /run/user /root".
1829
1830 * Non-service units are now started with KeyringMode=shared by default.
1831 This means that mount and swapon and other mount tools have access
1832 to keys in the main keyring.
1833
1834 * /sys/fs/bpf is now mounted automatically.
1835
1836 * QNX virtualization is now detected by systemd-detect-virt and may
1837 be used in ConditionVirtualization=.
1838
1839 * IPAccounting= may now be enabled also for slice units.
1840
1841 * A new -Dsplit-bin= build configuration switch may be used to specify
1842 whether bin and sbin directories are merged, or if they should be
1843 included separately in $PATH and various listings of executable
1844 directories. The build configuration scripts will try to autodetect
1845 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
1846 system, but distributions are encouraged to configure this
1847 explicitly.
1848
1849 * A new -Dok-color= build configuration switch may be used to change
1850 the colour of "OK" status messages.
1851
1852 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
1853 PrivateNetwork=yes was buggy in previous versions of systemd. This
1854 means that after the upgrade and daemon-reexec, any such units must
1855 be restarted.
1856
1857 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
1858 will not exclude read-only files owned by root from cleanup.
1859
1860 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
1861 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
1862 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
1863 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
1864 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
1865 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
1866 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
1867 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
1868 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
1869 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
1870 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
1871 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
1872 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
1873 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
1874 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
1875 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
1876
1877 — Warsaw, 2018-03-05
1878
1879 CHANGES WITH 237:
1880
1881 * Some keyboards come with a zoom see-saw or rocker which until now got
1882 mapped to the Linux "zoomin/out" keys in hwdb. However, these
1883 keycodes are not recognized by any major desktop. They now produce
1884 Up/Down key events so that they can be used for scrolling.
1885
1886 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
1887 slightly: previously, if an argument was specified for lines of this
1888 type (i.e. the right-most column was set) this string was appended to
1889 existing files each time systemd-tmpfiles was run. This behaviour was
1890 different from what the documentation said, and not particularly
1891 useful, as repeated systemd-tmpfiles invocations would not be
1892 idempotent and grow such files without bounds. With this release
1893 behaviour has been altered to match what the documentation says:
1894 lines of this type only have an effect if the indicated files don't
1895 exist yet, and only then the argument string is written to the file.
1896
1897 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
1898 systemd-tmpfiles behaviour: previously, read-only files owned by root
1899 were always excluded from the file "aging" algorithm (i.e. the
1900 automatic clean-up of directories like /tmp based on
1901 atime/mtime/ctime). We intend to drop this restriction, and age files
1902 by default even when owned by root and read-only. This behaviour was
1903 inherited from older tools, but there have been requests to remove
1904 it, and it's not obvious why this restriction was made in the first
1905 place. Please speak up now, if you are aware of software that reqires
1906 this behaviour, otherwise we'll remove the restriction in v238.
1907
1908 * A new environment variable $SYSTEMD_OFFLINE is now understood by
1909 systemctl. It takes a boolean argument. If on, systemctl assumes it
1910 operates on an "offline" OS tree, and will not attempt to talk to the
1911 service manager. Previously, this mode was implicitly enabled if a
1912 chroot() environment was detected, and this new environment variable
1913 now provides explicit control.
1914
1915 * .path and .socket units may now be created transiently, too.
1916 Previously only service, mount, automount and timer units were
1917 supported as transient units. The systemd-run tool has been updated
1918 to expose this new functionality, you may hence use it now to bind
1919 arbitrary commands to path or socket activation on-the-fly from the
1920 command line. Moreover, almost all properties are now exposed for the
1921 unit types that already supported transient operation.
1922
1923 * The systemd-mount command gained support for a new --owner= parameter
1924 which takes a user name, which is then resolved and included in uid=
1925 and gid= mount options string of the file system to mount.
1926
1927 * A new unit condition ConditionControlGroupController= has been added
1928 that checks whether a specific cgroup controller is available.
1929
1930 * Unit files, udev's .link files, and systemd-networkd's .netdev and
1931 .network files all gained support for a new condition
1932 ConditionKernelVersion= for checking against specific kernel
1933 versions.
1934
1935 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
1936 support for configuring device flags in the Flags= setting. In the
1937 same files, the [Tunnel] section gained support for configuring
1938 AllowLocalRemote=. The [Route] section in .network files gained
1939 support for configuring InitialCongestionWindow=,
1940 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
1941 understands RapidCommit=.
1942
1943 * systemd-networkd's DHCPv6 support gained support for Prefix
1944 Delegation.
1945
1946 * sd-bus gained support for a new "watch-bind" feature. When this
1947 feature is enabled, an sd_bus connection may be set up to connect to
1948 an AF_UNIX socket in the file system as soon as it is created. This
1949 functionality is useful for writing early-boot services that
1950 automatically connect to the system bus as soon as it is started,
1951 without ugly time-based polling. systemd-networkd and
1952 systemd-resolved have been updated to make use of this
1953 functionality. busctl exposes this functionality in a new
1954 --watch-bind= command line switch.
1955
1956 * sd-bus will now optionally synthesize a local "Connected" signal as
1957 soon as a D-Bus connection is set up fully. This message mirrors the
1958 already existing "Disconnected" signal which is synthesized when the
1959 connection is terminated. This signal is generally useful but
1960 particularly handy in combination with the "watch-bind" feature
1961 described above. Synthesizing of this message has to be requested
1962 explicitly through the new API call sd_bus_set_connected_signal(). In
1963 addition a new call sd_bus_is_ready() has been added that checks
1964 whether a connection is fully set up (i.e. between the "Connected" and
1965 "Disconnected" signals).
1966
1967 * sd-bus gained two new calls sd_bus_request_name_async() and
1968 sd_bus_release_name_async() for asynchronously registering bus
1969 names. Similar, there is now sd_bus_add_match_async() for installing
1970 a signal match asynchronously. All of systemd's own services have
1971 been updated to make use of these calls. Doing these operations
1972 asynchronously has two benefits: it reduces the risk of deadlocks in
1973 case of cyclic dependencies between bus services, and it speeds up
1974 service initialization since synchronization points for bus
1975 round-trips are removed.
1976
1977 * sd-bus gained two new calls sd_bus_match_signal() and
1978 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
1979 and sd_bus_add_match_async() but instead of taking a D-Bus match
1980 string take match fields as normal function parameters.
1981
1982 * sd-bus gained two new calls sd_bus_set_sender() and
1983 sd_bus_message_set_sender() for setting the sender name of outgoing
1984 messages (either for all outgoing messages or for just one specific
1985 one). These calls are only useful in direct connections as on
1986 brokered connections the broker fills in the sender anyway,
1987 overwriting whatever the client filled in.
1988
1989 * sd-event gained a new pseudo-handle that may be specified on all API
1990 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
1991 used this refers to the default event loop object of the calling
1992 thread. Note however that this does not implicitly allocate one —
1993 which has to be done prior by using sd_event_default(). Similarly
1994 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
1995 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
1996 to the default bus of the specified type of the calling thread. Here
1997 too this does not implicitly allocate bus connection objects, this
1998 has to be done prior with sd_bus_default() and friends.
1999
2000 * sd-event gained a new call pair
2001 sd_event_source_{get|set}_io_fd_own(). This may be used to request
2002 automatic closure of the file descriptor an IO event source watches
2003 when the event source is destroyed.
2004
2005 * systemd-networkd gained support for natively configuring WireGuard
2006 connections.
2007
2008 * In previous versions systemd synthesized user records both for the
2009 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
2010 internally. In order to simplify distribution-wide renames of the
2011 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
2012 new transitional flag file has been added: if
2013 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
2014 user and group record within the systemd codebase is disabled.
2015
2016 * systemd-notify gained a new --uid= option for selecting the source
2017 user/UID to use for notification messages sent to the service
2018 manager.
2019
2020 * journalctl gained a new --grep= option to list only entries in which
2021 the message matches a certain pattern. By default matching is case
2022 insensitive if the pattern is lowercase, and case sensitive
2023 otherwise. Option --case-sensitive=yes|no can be used to override
2024 this an specify case sensitivity or case insensitivity.
2025
2026 * There's now a "systemd-analyze service-watchdogs" command for printing
2027 the current state of the service runtime watchdog, and optionally
2028 enabling or disabling the per-service watchdogs system-wide if given a
2029 boolean argument (i.e. the concept you configure in WatchdogSec=), for
2030 debugging purposes. There's also a kernel command line option
2031 systemd.service_watchdogs= for controlling the same.
2032
2033 * Two new "log-level" and "log-target" options for systemd-analyze were
2034 added that merge the now deprecated get-log-level, set-log-level and
2035 get-log-target, set-log-target pairs. The deprecated options are still
2036 understood for backwards compatibility. The two new options print the
2037 current value when no arguments are given, and set them when a
2038 level/target is given as an argument.
2039
2040 * sysusers.d's "u" lines now optionally accept both a UID and a GID
2041 specification, separated by a ":" character, in order to create users
2042 where UID and GID do not match.
2043
2044 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
2045 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
2046 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
2047 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
2048 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
2049 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
2050 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
2051 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
2052 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
2053 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
2054 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
2055 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
2056 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
2057 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
2058 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
2059 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
2060 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
2061 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
2062 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
2063 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
2064 Палаузов
2065
2066 — Brno, 2018-01-28
2067
2068 CHANGES WITH 236:
2069
2070 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
2071 in v235 has been extended to also set the dummy.ko module option
2072 numdummies=0, preventing the kernel from automatically creating
2073 dummy0. All dummy interfaces must now be explicitly created.
2074
2075 * Unknown '%' specifiers in configuration files are now rejected. This
2076 applies to units and tmpfiles.d configuration. Any percent characters
2077 that are followed by a letter or digit that are not supposed to be
2078 interpreted as the beginning of a specifier should be escaped by
2079 doubling ("%%"). (So "size=5%" is still accepted, as well as
2080 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
2081 valid specifiers today.)
2082
2083 * systemd-resolved now maintains a new dynamic
2084 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
2085 recommended to make /etc/resolv.conf a symlink to it. This file
2086 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
2087 includes dynamically acquired search domains, achieving more correct
2088 DNS resolution by software that bypasses local DNS APIs such as NSS.
2089
2090 * The "uaccess" udev tag has been dropped from /dev/kvm and
2091 /dev/dri/renderD*. These devices now have the 0666 permissions by
2092 default (but this may be changed at build-time). /dev/dri/renderD*
2093 will now be owned by the "render" group along with /dev/kfd.
2094
2095 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
2096 systemd-journal-gatewayd.service and
2097 systemd-journal-upload.service. This means "nss-systemd" must be
2098 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
2099 services are resolved properly.
2100
2101 * In /etc/fstab two new mount options are now understood:
2102 x-systemd.makefs and x-systemd.growfs. The former has the effect that
2103 the configured file system is formatted before it is mounted, the
2104 latter that the file system is resized to the full block device size
2105 after it is mounted (i.e. if the file system is smaller than the
2106 partition it resides on, it's grown). This is similar to the fsck
2107 logic in /etc/fstab, and pulls in systemd-makefs@.service and
2108 systemd-growfs@.service as necessary, similar to
2109 systemd-fsck@.service. Resizing is currently only supported on ext4
2110 and btrfs.
2111
2112 * In systemd-networkd, the IPv6 RA logic now optionally may announce
2113 DNS server and domain information.
2114
2115 * Support for the LUKS2 on-disk format for encrypted partitions has
2116 been added. This requires libcryptsetup2 during compilation and
2117 runtime.
2118
2119 * The systemd --user instance will now signal "readiness" when its
2120 basic.target unit has been reached, instead of when the run queue ran
2121 empty for the first time.
2122
2123 * Tmpfiles.d with user configuration are now also supported.
2124 systemd-tmpfiles gained a new --user switch, and snippets placed in
2125 ~/.config/user-tmpfiles.d/ and corresponding directories will be
2126 executed by systemd-tmpfiles --user running in the new
2127 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
2128 running in the user session.
2129
2130 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
2131 %S resolves to the top-level state directory (/var/lib for the system
2132 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
2133 top-level cache directory (/var/cache for the system instance,
2134 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
2135 logs directory (/var/log for the system instance,
2136 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
2137 existing %t specifier, that resolves to the top-level runtime
2138 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
2139 user instance).
2140
2141 * journalctl learnt a new parameter --output-fields= for limiting the
2142 set of journal fields to output in verbose and JSON output modes.
2143
2144 * systemd-timesyncd's configuration file gained a new option
2145 RootDistanceMaxSec= for setting the maximum root distance of servers
2146 it'll use, as well as the new options PollIntervalMinSec= and
2147 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
2148
2149 * bootctl gained a new command "list" for listing all available boot
2150 menu items on systems that follow the boot loader specification.
2151
2152 * systemctl gained a new --dry-run switch that shows what would be done
2153 instead of doing it, and is currently supported by the shutdown and
2154 sleep verbs.
2155
2156 * ConditionSecurity= can now detect the TOMOYO security module.
2157
2158 * Unit file [Install] sections are now also respected in unit drop-in
2159 files. This is intended to be used by drop-ins under /usr/lib/.
2160
2161 * systemd-firstboot may now also set the initial keyboard mapping.
2162
2163 * Udev "changed" events for devices which are exposed as systemd
2164 .device units are now propagated to units specified in
2165 ReloadPropagatedFrom= as reload requests.
2166
2167 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
2168 unit template name (i.e. a name in the form of 'foobar@.service',
2169 without the instance component between the '@' and - the '.'), then
2170 the escaped sysfs path of the device is automatically used as the
2171 instance.
2172
2173 * SystemCallFilter= in unit files has been extended so that an "errno"
2174 can be specified individually for each system call. Example:
2175 SystemCallFilter=~uname:EILSEQ.
2176
2177 * The cgroup delegation logic has been substantially updated. Delegate=
2178 now optionally takes a list of controllers (instead of a boolean, as
2179 before), which lists the controllers to delegate at least.
2180
2181 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
2182
2183 * A new LogLevelMax= setting configures the maximum log level any
2184 process of the service may log at (i.e. anything with a lesser
2185 priority than what is specified is automatically dropped). A new
2186 LogExtraFields= setting allows configuration of additional journal
2187 fields to attach to all log records generated by any of the unit's
2188 processes.
2189
2190 * New StandardInputData= and StandardInputText= settings along with the
2191 new option StandardInput=data may be used to configure textual or
2192 binary data that shall be passed to the executed service process via
2193 standard input, encoded in-line in the unit file.
2194
2195 * StandardInput=, StandardOutput= and StandardError= may now be used to
2196 connect stdin/stdout/stderr of executed processes directly with a
2197 file or AF_UNIX socket in the file system, using the new "file:" option.
2198
2199 * A new unit file option CollectMode= has been added, that allows
2200 tweaking the garbage collection logic for units. It may be used to
2201 tell systemd to garbage collect units that have failed automatically
2202 (normally it only GCs units that exited successfully). systemd-run
2203 and systemd-mount expose this new functionality with a new -G option.
2204
2205 * "machinectl bind" may now be used to bind mount non-directories
2206 (i.e. regularfiles, devices, fifos, sockets).
2207
2208 * systemd-analyze gained a new verb "calendar" for validating and
2209 testing calendar time specifications to use for OnCalendar= in timer
2210 units. Besides validating the expression it will calculate the next
2211 time the specified expression would elapse.
2212
2213 * In addition to the pre-existing FailureAction= unit file setting
2214 there's now SuccessAction=, for configuring a shutdown action to
2215 execute when a unit completes successfully. This is useful in
2216 particular inside containers that shall terminate after some workload
2217 has been completed. Also, both options are now supported for all unit
2218 types, not just services.
2219
2220 * networkds's IP rule support gained two new options
2221 IncomingInterface= and OutgoingInterface= for configuring the incoming
2222 and outgoing interfaces of configured rules. systemd-networkd also
2223 gained support for "vxcan" network devices.
2224
2225 * networkd gained a new setting RequiredForOnline=, taking a
2226 boolean. If set, systemd-wait-online will take it into consideration
2227 when determining that the system is up, otherwise it will ignore the
2228 interface for this purpose.
2229
2230 * The sd_notify() protocol gained support for a new operation: with
2231 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
2232 store again, ahead of POLLHUP or POLLERR when they are removed
2233 anyway.
2234
2235 * A new document doc/UIDS-GIDS.md has been added to the source tree,
2236 that documents the UID/GID range and assignment assumptions and
2237 requirements of systemd.
2238
2239 * The watchdog device PID 1 will ping may now be configured through the
2240 WatchdogDevice= configuration file setting, or by setting the
2241 systemd.watchdog_service= kernel commandline option.
2242
2243 * systemd-resolved's gained support for registering DNS-SD services on
2244 the local network using MulticastDNS. Services may either be
2245 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
2246 the same dir below /run, /usr/lib), or through its D-Bus API.
2247
2248 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
2249 extend the effective start, runtime, and stop time. The service must
2250 continue to send EXTEND_TIMEOUT_USEC within the period specified to
2251 prevent the service manager from making the service as timedout.
2252
2253 * systemd-resolved's DNSSEC support gained support for RFC 8080
2254 (Ed25519 keys and signatures).
2255
2256 * The systemd-resolve command line tool gained a new set of options
2257 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
2258 --set-nta= and --revert to configure per-interface DNS configuration
2259 dynamically during runtime. It's useful for pushing DNS information
2260 into systemd-resolved from DNS hook scripts that various interface
2261 managing software supports (such as pppd).
2262
2263 * systemd-nspawn gained a new --network-namespace-path= command line
2264 option, which may be used to make a container join an existing
2265 network namespace, by specifying a path to a "netns" file.
2266
2267 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
2268 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
2269 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
2270 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
2271 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
2272 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
2273 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
2274 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
2275 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
2276 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
2277 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
2278 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
2279 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
2280 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
2281 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
2282 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
2283 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
2284 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2285 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
2286 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
2287 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
2288 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
2289 Jędrzejewski-Szmek, Zeal Jagannatha
2290
2291 — Berlin, 2017-12-14
2292
2293 CHANGES WITH 235:
2294
2295 * INCOMPATIBILITY: systemd-logind.service and other long-running
2296 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
2297 communication with the outside. This generally improves security of
2298 the system, and is in almost all cases a safe and good choice, as
2299 these services do not and should not provide any network-facing
2300 functionality. However, systemd-logind uses the glibc NSS API to
2301 query the user database. This creates problems on systems where NSS
2302 is set up to directly consult network services for user database
2303 lookups. In particular, this creates incompatibilities with the
2304 "nss-nis" module, which attempts to directly contact the NIS/YP
2305 network servers it is configured for, and will now consistently
2306 fail. In such cases, it is possible to turn off IP sandboxing for
2307 systemd-logind.service (set IPAddressDeny= in its [Service] section
2308 to the empty string, via a .d/ unit file drop-in). Downstream
2309 distributions might want to update their nss-nis packaging to include
2310 such a drop-in snippet, accordingly, to hide this incompatibility
2311 from the user. Another option is to make use of glibc's nscd service
2312 to proxy such network requests through a privilege-separated, minimal
2313 local caching daemon, or to switch to more modern technologies such
2314 sssd, whose NSS hook-ups generally do not involve direct network
2315 access. In general, we think it's definitely time to question the
2316 implementation choices of nss-nis, i.e. whether it's a good idea
2317 today to embed a network-facing loadable module into all local
2318 processes that need to query the user database, including the most
2319 trivial and benign ones, such as "ls". For more details about
2320 IPAddressDeny= see below.
2321
2322 * A new modprobe.d drop-in is now shipped by default that sets the
2323 bonding module option max_bonds=0. This overrides the kernel default,
2324 to avoid conflicts and ambiguity as to whether or not bond0 should be
2325 managed by systemd-networkd or not. This resolves multiple issues
2326 with bond0 properties not being applied, when bond0 is configured
2327 with systemd-networkd. Distributors may choose to not package this,
2328 however in that case users will be prevented from correctly managing
2329 bond0 interface using systemd-networkd.
2330
2331 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
2332 which print the logging level and target of the system manager. They
2333 complement the existing "set-log-level" and "set-log-target" verbs
2334 used to change those values.
2335
2336 * journald.conf gained a new boolean setting ReadKMsg= which defaults
2337 to on. If turned off kernel log messages will not be read by
2338 systemd-journald or included in the logs. It also gained a new
2339 setting LineMax= for configuring the maximum line length in
2340 STDOUT/STDERR log streams. The new default for this value is 48K, up
2341 from the previous hardcoded 2048.
2342
2343 * A new unit setting RuntimeDirectoryPreserve= has been added, which
2344 allows more detailed control of what to do with a runtime directory
2345 configured with RuntimeDirectory= (i.e. a directory below /run or
2346 $XDG_RUNTIME_DIR) after a unit is stopped.
2347
2348 * The RuntimeDirectory= setting for units gained support for creating
2349 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
2350 one top-level directory.
2351
2352 * Units gained new options StateDirectory=, CacheDirectory=,
2353 LogsDirectory= and ConfigurationDirectory= which are closely related
2354 to RuntimeDirectory= but manage per-service directories below
2355 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
2356 possible to write unit files which when activated automatically gain
2357 properly owned service specific directories in these locations, thus
2358 making unit files self-contained and increasing compatibility with
2359 stateless systems and factory reset where /etc or /var are
2360 unpopulated at boot. Matching these new settings there's also
2361 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
2362 ConfigurationDirectoryMode= for configuring the access mode of these
2363 directories. These settings are particularly useful in combination
2364 with DynamicUser=yes as they provide secure, properly-owned,
2365 writable, and stateful locations for storage, excluded from the
2366 sandbox that such services live in otherwise.
2367
2368 * Automake support has been removed from this release. systemd is now
2369 Meson-only.
2370
2371 * systemd-journald will now aggressively cache client metadata during
2372 runtime, speeding up log write performance under pressure. This comes
2373 at a small price though: as much of the metadata is read
2374 asynchronously from /proc/ (and isn't implicitly attached to log
2375 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
2376 metadata stored alongside a log entry might be slightly
2377 out-of-date. Previously it could only be slightly newer than the log
2378 message. The time window is small however, and given that the kernel
2379 is unlikely to be improved anytime soon in this regard, this appears
2380 acceptable to us.
2381
2382 * nss-myhostname/systemd-resolved will now by default synthesize an
2383 A/AAAA resource record for the "_gateway" hostname, pointing to the
2384 current default IP gateway. Previously it did that for the "gateway"
2385 name, hampering adoption, as some distributions wanted to leave that
2386 host name open for local use. The old behaviour may still be
2387 requested at build time.
2388
2389 * systemd-networkd's [Address] section in .network files gained a new
2390 Scope= setting for configuring the IP address scope. The [Network]
2391 section gained a new boolean setting ConfigureWithoutCarrier= that
2392 tells systemd-networkd to ignore link sensing when configuring the
2393 device. The [DHCP] section gained a new Anonymize= boolean option for
2394 turning on a number of options suggested in RFC 7844. A new
2395 [RoutingPolicyRule] section has been added for configuring the IP
2396 routing policy. The [Route] section has gained support for a new
2397 Type= setting which permits configuring
2398 blackhole/unreachable/prohibit routes.
2399
2400 * The [VRF] section in .netdev files gained a new Table= setting for
2401 configuring the routing table to use. The [Tunnel] section gained a
2402 new Independent= boolean field for configuring tunnels independent of
2403 an underlying network interface. The [Bridge] section gained a new
2404 GroupForwardMask= option for configuration of propagation of link
2405 local frames between bridge ports.
2406
2407 * The WakeOnLan= setting in .link files gained support for a number of
2408 new modes. A new TCP6SegmentationOffload= setting has been added for
2409 configuring TCP/IPv6 hardware segmentation offload.
2410
2411 * The IPv6 RA sender implementation may now optionally send out RDNSS
2412 and RDNSSL records to supply DNS configuration to peers.
2413
2414 * systemd-nspawn gained support for a new --system-call-filter= command
2415 line option for adding and removing entries in the default system
2416 call filter it applies. Moreover systemd-nspawn has been changed to
2417 implement a system call whitelist instead of a blacklist.
2418
2419 * systemd-run gained support for a new --pipe command line option. If
2420 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
2421 are directly passed on to the activated transient service
2422 executable. This allows invoking arbitrary processes as systemd
2423 services (for example to take benefit of dependency management,
2424 accounting management, resource management or log management that is
2425 done automatically for services) — while still allowing them to be
2426 integrated in a classic UNIX shell pipeline.
2427
2428 * When a service sends RELOAD=1 via sd_notify() and reload propagation
2429 using ReloadPropagationTo= is configured, a reload is now propagated
2430 to configured units. (Previously this was only done on explicitly
2431 requested reloads, using "systemctl reload" or an equivalent
2432 command.)
2433
2434 * For each service unit a restart counter is now kept: it is increased
2435 each time the service is restarted due to Restart=, and may be
2436 queried using "systemctl show -p NRestarts …".
2437
2438 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
2439 @signal and @timer have been added, for usage with SystemCallFilter=
2440 in unit files and the new --system-call-filter= command line option
2441 of systemd-nspawn (see above).
2442
2443 * ExecStart= lines in unit files gained two new modifiers: when a
2444 command line is prefixed with "!" the command will be executed as
2445 configured, except for the credentials applied by
2446 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
2447 "+", but does still apply namespacing options unlike "+". There's
2448 also "!!" now, which is mostly identical, but becomes a NOP on
2449 systems that support ambient capabilities. This is useful to write
2450 unit files that work with ambient capabilities where possible but
2451 automatically fall back to traditional privilege dropping mechanisms
2452 on systems where this is not supported.
2453
2454 * ListenNetlink= settings in socket units now support RDMA netlink
2455 sockets.
2456
2457 * A new unit file setting LockPersonality= has been added which permits
2458 locking down the chosen execution domain ("personality") of a service
2459 during runtime.
2460
2461 * A new special target "getty-pre.target" has been added, which is
2462 ordered before all text logins, and may be used to order services
2463 before textual logins acquire access to the console.
2464
2465 * systemd will now attempt to load the virtio-rng.ko kernel module very
2466 early on if a VM environment supporting this is detected. This should
2467 improve entropy during early boot in virtualized environments.
2468
2469 * A _netdev option is now supported in /etc/crypttab that operates in a
2470 similar way as the same option in /etc/fstab: it permits configuring
2471 encrypted devices that need to be ordered after the network is up.
2472 Following this logic, two new special targets
2473 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
2474 added that are to cryptsetup.target what remote-fs.target and
2475 remote-fs-pre.target are to local-fs.target.
2476
2477 * Service units gained a new UnsetEnvironment= setting which permits
2478 unsetting specific environment variables for services that are
2479 normally passed to it (for example in order to mask out locale
2480 settings for specific services that can't deal with it).
2481
2482 * Units acquired a new boolean option IPAccounting=. When turned on, IP
2483 traffic accounting (packet count as well as byte count) is done for
2484 the service, and shown as part of "systemctl status" or "systemd-run
2485 --wait".
2486
2487 * Service units acquired two new options IPAddressAllow= and
2488 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
2489 for configuring a simple IP access control list for all sockets of
2490 the unit. These options are available also on .slice and .socket
2491 units, permitting flexible access list configuration for individual
2492 services as well as groups of services (as defined by a slice unit),
2493 including system-wide. Note that IP ACLs configured this way are
2494 enforced on every single IPv4 and IPv6 socket created by any process
2495 of the service unit, and apply to ingress as well as egress traffic.
2496
2497 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
2498 structured log message is generated each time the unit is stopped,
2499 containing information about the consumed resources of this
2500 invocation.
2501
2502 * A new setting KeyringMode= has been added to unit files, which may be
2503 used to control how the kernel keyring is set up for executed
2504 processes.
2505
2506 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
2507 "systemctl kexec" and "systemctl exit" are now always asynchronous in
2508 behaviour (that is: these commands return immediately after the
2509 operation was enqueued instead of waiting for the operation to
2510 complete). Previously, "systemctl poweroff" and "systemctl reboot"
2511 were asynchronous on systems using systemd-logind (i.e. almost
2512 always, and like they were on sysvinit), and the other three commands
2513 were unconditionally synchronous. With this release this is cleaned
2514 up, and callers will see the same asynchronous behaviour on all
2515 systems for all five operations.
2516
2517 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
2518 the system.
2519
2520 * .timer units now accept calendar specifications in other timezones
2521 than UTC or the local timezone.
2522
2523 * The tmpfiles snippet var.conf has been changed to create
2524 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
2525 the "utmp" group already, and it appears to be generally understood
2526 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
2527 databases. Previously this was implemented correctly for all these
2528 databases excepts btmp, which has been opened up like this now
2529 too. Note that while the other databases are world-readable
2530 (i.e. 0644), btmp is not and remains more restrictive.
2531
2532 * The systemd-resolve tool gained a new --reset-server-features
2533 switch. When invoked like this systemd-resolved will forget
2534 everything it learnt about the features supported by the configured
2535 upstream DNS servers, and restarts the feature probing logic on the
2536 next resolver look-up for them at the highest feature level
2537 again.
2538
2539 * The status dump systemd-resolved sends to the logs upon receiving
2540 SIGUSR1 now also includes information about all DNS servers it is
2541 configured to use, and the features levels it probed for them.
2542
2543 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
2544 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
2545 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
2546 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
2547 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
2548 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
2549 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
2550 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
2551 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
2552 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
2553 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
2554 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
2555 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
2556 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
2557 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
2558 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
2559 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
2560 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
2561 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
2562 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2563
2564 — Berlin, 2017-10-06
2565
2566 CHANGES WITH 234:
2567
2568 * Meson is now supported as build system in addition to Automake. It is
2569 our plan to remove Automake in one of our next releases, so that
2570 Meson becomes our exclusive build system. Hence, please start using
2571 the Meson build system in your downstream packaging. There's plenty
2572 of documentation around how to use Meson, the extremely brief
2573 summary:
2574
2575 ./autogen.sh && ./configure && make && sudo make install
2576
2577 becomes:
2578
2579 meson build && ninja -C build && sudo ninja -C build install
2580
2581 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
2582 which permits configuring a timeout on the time a job is
2583 running. This is particularly useful for setting timeouts on jobs for
2584 .device units.
2585
2586 * Unit files gained two new options ConditionUser= and ConditionGroup=
2587 for conditionalizing units based on the identity of the user/group
2588 running a systemd user instance.
2589
2590 * systemd-networkd now understands a new FlowLabel= setting in the
2591 [VXLAN] section of .network files, as well as a Priority= in
2592 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
2593 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
2594 gained support for configuration of GENEVE links, and IPv6 address
2595 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
2596
2597 * .link files now understand a new Port= setting.
2598
2599 * systemd-networkd's DHCP support gained support for DHCP option 119
2600 (domain search list).
2601
2602 * systemd-networkd gained support for serving IPv6 address ranges using
2603 the Router Advertisement protocol. The new .network configuration
2604 section [IPv6Prefix] may be used to configure the ranges to
2605 serve. This is implemented based on a new, minimal, native server
2606 implementation of RA.
2607
2608 * journalctl's --output= switch gained support for a new parameter
2609 "short-iso-precise" for a mode where timestamps are shown as precise
2610 ISO date values.
2611
2612 * systemd-udevd's "net_id" builtin may now generate stable network
2613 interface names from IBM PowerVM VIO devices as well as ACPI platform
2614 devices.
2615
2616 * MulticastDNS support in systemd-resolved may now be explicitly
2617 enabled/disabled using the new MulticastDNS= configuration file
2618 option.
2619
2620 * systemd-resolved may now optionally use libidn2 instead of the libidn
2621 for processing internationalized domain names. Support for libidn2
2622 should be considered experimental and should not be enabled by
2623 default yet.
2624
2625 * "machinectl pull-tar" and related call may now do verification of
2626 downloaded images using SUSE-style .sha256 checksum files in addition
2627 to the already existing support for validating using Ubuntu-style
2628 SHA256SUMS files.
2629
2630 * sd-bus gained support for a new sd_bus_message_appendv() call which
2631 is va_list equivalent of sd_bus_message_append().
2632
2633 * sd-boot gained support for validating images using SHIM/MOK.
2634
2635 * The SMACK code learnt support for "onlycap".
2636
2637 * systemd-mount --umount is now much smarter in figuring out how to
2638 properly unmount a device given its mount or device path.
2639
2640 * The code to call libnss_dns as a fallback from libnss_resolve when
2641 the communication with systemd-resolved fails was removed. This
2642 fallback was redundant and interfered with the [!UNAVAIL=return]
2643 suffix. See nss-resolve(8) for the recommended configuration.
2644
2645 * systemd-logind may now be restarted without losing state. It stores
2646 the file descriptors for devices it manages in the system manager
2647 using the FDSTORE= mechanism. Please note that further changes in
2648 other components may be required to make use of this (for example
2649 Xorg has code to listen for stops of systemd-logind and terminate
2650 itself when logind is stopped or restarted, in order to avoid using
2651 stale file descriptors for graphical devices, which is now
2652 counterproductive and must be reverted in order for restarts of
2653 systemd-logind to be safe. See
2654 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
2655
2656 * All kernel-install plugins are called with the environment variable
2657 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
2658 /etc/machine-id. If the machine ID could not be determined,
2659 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
2660 anything in the entry directory (passed as the second argument) if
2661 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
2662 temporary directory is passed as the entry directory and removed
2663 after all the plugins exit.
2664
2665 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
2666 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
2667 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
2668 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
2669 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
2670 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
2671 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
2672 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
2673 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
2674 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
2675 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
2676 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
2677 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
2678 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
2679 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
2680 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
2681 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
2682 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
2683 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
2684 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
2685 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
2686 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
2687 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
2688 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
2689 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
2690 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
2691 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
2692 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
2693 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
2694 Георгиевски
2695
2696 — Berlin, 2017-07-12
2697
2698 CHANGES WITH 233:
2699
2700 * The "hybrid" control group mode has been modified to improve
2701 compatibility with "legacy" cgroups-v1 setups. Specifically, the
2702 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
2703 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
2704 cgroups-v1 hierarchy), the only externally visible change being that
2705 the cgroups-v2 hierarchy is also mounted, to
2706 /sys/fs/cgroup/unified. This should provide a large degree of
2707 compatibility with "legacy" cgroups-v1, while taking benefit of the
2708 better management capabilities of cgroups-v2.
2709
2710 * The default control group setup mode may be selected both a boot-time
2711 via a set of kernel command line parameters (specifically:
2712 systemd.unified_cgroup_hierarchy= and
2713 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
2714 default selected on the configure command line
2715 (--with-default-hierarchy=). The upstream default is "hybrid"
2716 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
2717 this will change in a future systemd version to be "unified" (pure
2718 cgroups-v2 mode). The third option for the compile time option is
2719 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
2720 distributions to default to "hybrid" mode for release distributions,
2721 starting with v233. We recommend "unified" for development
2722 distributions (specifically: distributions such as Fedora's rawhide)
2723 as that's where things are headed in the long run. Use "legacy" for
2724 greatest stability and compatibility only.
2725
2726 * Note one current limitation of "unified" and "hybrid" control group
2727 setup modes: the kernel currently does not permit the systemd --user
2728 instance (i.e. unprivileged code) to migrate processes between two
2729 disconnected cgroup subtrees, even if both are managed and owned by
2730 the user. This effectively means "systemd-run --user --scope" doesn't
2731 work when invoked from outside of any "systemd --user" service or
2732 scope. Specifically, it is not supported from session scopes. We are
2733 working on fixing this in a future systemd version. (See #3388 for
2734 further details about this.)
2735
2736 * DBus policy files are now installed into /usr rather than /etc. Make
2737 sure your system has dbus >= 1.9.18 running before upgrading to this
2738 version, or override the install path with --with-dbuspolicydir= .
2739
2740 * All python scripts shipped with systemd (specifically: the various
2741 tests written in Python) now require Python 3.
2742
2743 * systemd unit tests can now run standalone (without the source or
2744 build directories), and can be installed into /usr/lib/systemd/tests/
2745 with 'make install-tests'.
2746
2747 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
2748 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
2749 kernel.
2750
2751 * Support for the %c, %r, %R specifiers in unit files has been
2752 removed. Specifiers are not supposed to be dependent on configuration
2753 in the unit file itself (so that they resolve the same regardless
2754 where used in the unit files), but these specifiers were influenced
2755 by the Slice= option.
2756
2757 * The shell invoked by debug-shell.service now defaults to /bin/sh in
2758 all cases. If distributions want to use a different shell for this
2759 purpose (for example Fedora's /sbin/sushell) they need to specify
2760 this explicitly at configure time using --with-debug-shell=.
2761
2762 * The confirmation spawn prompt has been reworked to offer the
2763 following choices:
2764
2765 (c)ontinue, proceed without asking anymore
2766 (D)ump, show the state of the unit
2767 (f)ail, don't execute the command and pretend it failed
2768 (h)elp
2769 (i)nfo, show a short summary of the unit
2770 (j)obs, show jobs that are in progress
2771 (s)kip, don't execute the command and pretend it succeeded
2772 (y)es, execute the command
2773
2774 The 'n' choice for the confirmation spawn prompt has been removed,
2775 because its meaning was confusing.
2776
2777 The prompt may now also be redirected to an alternative console by
2778 specifying the console as parameter to systemd.confirm_spawn=.
2779
2780 * Services of Type=notify require a READY=1 notification to be sent
2781 during startup. If no such message is sent, the service now fails,
2782 even if the main process exited with a successful exit code.
2783
2784 * Services that fail to start up correctly now always have their
2785 ExecStopPost= commands executed. Previously, they'd enter "failed"
2786 state directly, without executing these commands.
2787
2788 * The option MulticastDNS= of network configuration files has acquired
2789 an actual implementation. With MulticastDNS=yes a host can resolve
2790 names of remote hosts and reply to mDNS A and AAAA requests.
2791
2792 * When units are about to be started an additional check is now done to
2793 ensure that all dependencies of type BindsTo= (when used in
2794 combination with After=) have been started.
2795
2796 * systemd-analyze gained a new verb "syscall-filter" which shows which
2797 system call groups are defined for the SystemCallFilter= unit file
2798 setting, and which system calls they contain.
2799
2800 * A new system call filter group "@filesystem" has been added,
2801 consisting of various file system related system calls. Group
2802 "@reboot" has been added, covering reboot, kexec and shutdown related
2803 calls. Finally, group "@swap" has been added covering swap
2804 configuration related calls.
2805
2806 * A new unit file option RestrictNamespaces= has been added that may be
2807 used to restrict access to the various process namespace types the
2808 Linux kernel provides. Specifically, it may be used to take away the
2809 right for a service unit to create additional file system, network,
2810 user, and other namespaces. This sandboxing option is particularly
2811 relevant due to the high amount of recently discovered namespacing
2812 related vulnerabilities in the kernel.
2813
2814 * systemd-udev's .link files gained support for a new AutoNegotiation=
2815 setting for configuring Ethernet auto-negotiation.
2816
2817 * systemd-networkd's .network files gained support for a new
2818 ListenPort= setting in the [DHCP] section to explicitly configure the
2819 UDP client port the DHCP client shall listen on.
2820
2821 * .network files gained a new Unmanaged= boolean setting for explicitly
2822 excluding one or more interfaces from management by systemd-networkd.
2823
2824 * The systemd-networkd ProxyARP= option has been renamed to
2825 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
2826 renamed to ReduceARPProxy=. The old names continue to be available
2827 for compatibility.
2828
2829 * systemd-networkd gained support for configuring IPv6 Proxy NDP
2830 addresses via the new IPv6ProxyNDPAddress= .network file setting.
2831
2832 * systemd-networkd's bonding device support gained support for two new
2833 configuration options ActiveSlave= and PrimarySlave=.
2834
2835 * The various options in the [Match] section of .network files gained
2836 support for negative matching.
2837
2838 * New systemd-specific mount options are now understood in /etc/fstab:
2839
2840 x-systemd.mount-timeout= may be used to configure the maximum
2841 permitted runtime of the mount command.
2842
2843 x-systemd.device-bound may be set to bind a mount point to its
2844 backing device unit, in order to automatically remove a mount point
2845 if its backing device is unplugged. This option may also be
2846 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
2847 on the block device, which is now automatically set for all CDROM
2848 drives, so that mounted CDs are automatically unmounted when they are
2849 removed from the drive.
2850
2851 x-systemd.after= and x-systemd.before= may be used to explicitly
2852 order a mount after or before another unit or mount point.
2853
2854 * Enqueued start jobs for device units are now automatically garbage
2855 collected if there are no jobs waiting for them anymore.
2856
2857 * systemctl list-jobs gained two new switches: with --after, for every
2858 queued job the jobs it's waiting for are shown; with --before the
2859 jobs which it's blocking are shown.
2860
2861 * systemd-nspawn gained support for ephemeral boots from disk images
2862 (or in other words: --ephemeral and --image= may now be
2863 combined). Moreover, ephemeral boots are now supported for normal
2864 directories, even if the backing file system is not btrfs. Of course,
2865 if the file system does not support file system snapshots or
2866 reflinks, the initial copy operation will be relatively expensive, but
2867 this should still be suitable for many use cases.
2868
2869 * Calendar time specifications in .timer units now support
2870 specifications relative to the end of a month by using "~" instead of
2871 "-" as separator between month and day. For example, "*-02~03" means
2872 "the third last day in February". In addition a new syntax for
2873 repeated events has been added using the "/" character. For example,
2874 "9..17/2:00" means "every two hours from 9am to 5pm".
2875
2876 * systemd-socket-proxyd gained a new parameter --connections-max= for
2877 configuring the maximum number of concurrent connections.
2878
2879 * sd-id128 gained a new API for generating unique IDs for the host in a
2880 way that does not leak the machine ID. Specifically,
2881 sd_id128_get_machine_app_specific() derives an ID based on the
2882 machine ID a in well-defined, non-reversible, stable way. This is
2883 useful whenever an identifier for the host is needed but where the
2884 identifier shall not be useful to identify the system beyond the
2885 scope of the application itself. (Internally this uses HMAC-SHA256 as
2886 keyed hash function using the machine ID as input.)
2887
2888 * NotifyAccess= gained a new supported value "exec". When set
2889 notifications are accepted from all processes systemd itself invoked,
2890 including all control processes.
2891
2892 * .nspawn files gained support for defining overlay mounts using the
2893 Overlay= and OverlayReadOnly= options. Previously this functionality
2894 was only available on the systemd-nspawn command line.
2895
2896 * systemd-nspawn's --bind= and --overlay= options gained support for
2897 bind/overlay mounts whose source lies within the container tree by
2898 prefixing the source path with "+".
2899
2900 * systemd-nspawn's --bind= and --overlay= options gained support for
2901 automatically allocating a temporary source directory in /var/tmp
2902 that is removed when the container dies. Specifically, if the source
2903 directory is specified as empty string this mechanism is selected. An
2904 example usage is --overlay=+/var::/var, which creates an overlay
2905 mount based on the original /var contained in the image, overlayed
2906 with a temporary directory in the host's /var/tmp. This way changes
2907 to /var are automatically flushed when the container shuts down.
2908
2909 * systemd-nspawn --image= option does now permit raw file system block
2910 devices (in addition to images containing partition tables, as
2911 before).
2912
2913 * The disk image dissection logic in systemd-nspawn gained support for
2914 automatically setting up LUKS encrypted as well as Verity protected
2915 partitions. When a container is booted from an encrypted image the
2916 passphrase is queried at start-up time. When a container with Verity
2917 data is started, the root hash is search in a ".roothash" file
2918 accompanying the disk image (alternatively, pass the root hash via
2919 the new --root-hash= command line option).
2920
2921 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
2922 be used to dissect disk images the same way as systemd-nspawn does
2923 it, following the Bootable Partition Specification. It may even be
2924 used to mount disk images with complex partition setups (including
2925 LUKS and Verity partitions) to a local host directory, in order to
2926 inspect them. This tool is not considered public API (yet), and is
2927 thus not installed into /usr/bin. Please do not rely on its
2928 existence, since it might go away or be changed in later systemd
2929 versions.
2930
2931 * A new generator "systemd-verity-generator" has been added, similar in
2932 style to "systemd-cryptsetup-generator", permitting automatic setup of
2933 Verity root partitions when systemd boots up. In order to make use of
2934 this your partition setup should follow the Discoverable Partitions
2935 Specification, and the GPT partition ID of the root file system
2936 partition should be identical to the upper 128bit of the Verity root
2937 hash. The GPT partition ID of the Verity partition protecting it
2938 should be the lower 128bit of the Verity root hash. If the partition
2939 image follows this model it is sufficient to specify a single
2940 "roothash=" kernel command line argument to both configure which root
2941 image and verity partition to use as well as the root hash for
2942 it. Note that systemd-nspawn's Verity support follows the same
2943 semantics, meaning that disk images with proper Verity data in place
2944 may be booted in containers with systemd-nspawn as well as on
2945 physical systems via the verity generator. Also note that the "mkosi"
2946 tool available at https://github.com/systemd/mkosi has been updated
2947 to generate Verity protected disk images following this scheme. In
2948 fact, it has been updated to generate disk images that optionally
2949 implement a complete UEFI SecureBoot trust chain, involving a signed
2950 kernel and initrd image that incorporates such a root hash as well as
2951 a Verity-enabled root partition.
2952
2953 * The hardware database (hwdb) udev supports has been updated to carry
2954 accelerometer quirks.
2955
2956 * All system services are now run with a fresh kernel keyring set up
2957 for them. The invocation ID is stored by default in it, thus
2958 providing a safe, non-overridable way to determine the invocation
2959 ID of each service.
2960
2961 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
2962 options for bind mounting arbitrary paths in a service-specific
2963 way. When these options are used, arbitrary host or service files and
2964 directories may be mounted to arbitrary locations in the service's
2965 view.
2966
2967 * Documentation has been added that lists all of systemd's low-level
2968 environment variables:
2969
2970 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
2971
2972 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
2973 whether a specific socket file descriptor matches a specified socket
2974 address.
2975
2976 * systemd-firstboot has been updated to check for the
2977 systemd.firstboot= kernel command line option. It accepts a boolean
2978 and when set to false the first boot questions are skipped.
2979
2980 * systemd-fstab-generator has been updated to check for the
2981 systemd.volatile= kernel command line option, which either takes an
2982 optional boolean parameter or the special value "state". If used the
2983 system may be booted in a "volatile" boot mode. Specifically,
2984 "systemd.volatile" is used, the root directory will be mounted as
2985 tmpfs, and only /usr is mounted from the actual root file system. If
2986 "systemd.volatile=state" is used, the root directory will be mounted
2987 as usual, but /var is mounted as tmpfs. This concept provides similar
2988 functionality as systemd-nspawn's --volatile= option, but provides it
2989 on physical boots. Use this option for implementing stateless
2990 systems, or testing systems with all state and/or configuration reset
2991 to the defaults. (Note though that many distributions are not
2992 prepared to boot up without a populated /etc or /var, though.)
2993
2994 * systemd-gpt-auto-generator gained support for LUKS encrypted root
2995 partitions. Previously it only supported LUKS encrypted partitions
2996 for all other uses, except for the root partition itself.
2997
2998 * Socket units gained support for listening on AF_VSOCK sockets for
2999 communication in virtualized QEMU environments.
3000
3001 * The "configure" script gained a new option --with-fallback-hostname=
3002 for specifying the fallback hostname to use if none is configured in
3003 /etc/hostname. For example, by specifying
3004 --with-fallback-hostname=fedora it is possible to default to a
3005 hostname of "fedora" on pristine installations.
3006
3007 * systemd-cgls gained support for a new --unit= switch for listing only
3008 the control groups of a specific unit. Similar --user-unit= has been
3009 added for listing only the control groups of a specific user unit.
3010
3011 * systemd-mount gained a new --umount switch for unmounting a mount or
3012 automount point (and all mount/automount points below it).
3013
3014 * systemd will now refuse full configuration reloads (via systemctl
3015 daemon-reload and related calls) unless at least 16MiB of free space
3016 are available in /run. This is a safety precaution in order to ensure
3017 that generators can safely operate after the reload completed.
3018
3019 * A new unit file option RootImage= has been added, which has a similar
3020 effect as RootDirectory= but mounts the service's root directory from
3021 a disk image instead of plain directory. This logic reuses the same
3022 image dissection and mount logic that systemd-nspawn already uses,
3023 and hence supports any disk images systemd-nspawn supports, including
3024 those following the Discoverable Partition Specification, as well as
3025 Verity enabled images. This option enables systemd to run system
3026 services directly off disk images acting as resource bundles,
3027 possibly even including full integrity data.
3028
3029 * A new MountAPIVFS= unit file option has been added, taking a boolean
3030 argument. If enabled /proc, /sys and /dev (collectively called the
3031 "API VFS") will be mounted for the service. This is only relevant if
3032 RootDirectory= or RootImage= is used for the service, as these mounts
3033 are of course in place in the host mount namespace anyway.
3034
3035 * systemd-nspawn gained support for a new --pivot-root= switch. If
3036 specified the root directory within the container image is pivoted to
3037 the specified mount point, while the original root disk is moved to a
3038 different place. This option enables booting of ostree images
3039 directly with systemd-nspawn.
3040
3041 * The systemd build scripts will no longer complain if the NTP server
3042 addresses are not changed from the defaults. Google now supports
3043 these NTP servers officially. We still recommend downstreams to
3044 properly register an NTP pool with the NTP pool project though.
3045
3046 * coredumpctl gained a new "--reverse" option for printing the list
3047 of coredumps in reverse order.
3048
3049 * coredumpctl will now show additional information about truncated and
3050 inaccessible coredumps, as well as coredumps that are still being
3051 processed. It also gained a new --quiet switch for suppressing
3052 additional informational message in its output.
3053
3054 * coredumpctl gained support for only showing coredumps newer and/or
3055 older than specific timestamps, using the new --since= and --until=
3056 options, reminiscent of journalctl's options by the same name.
3057
3058 * The systemd-coredump logic has been improved so that it may be reused
3059 to collect backtraces in non-compiled languages, for example in
3060 scripting languages such as Python.
3061
3062 * machinectl will now show the UID shift of local containers, if user
3063 namespacing is enabled for them.
3064
3065 * systemd will now optionally run "environment generator" binaries at
3066 configuration load time. They may be used to add environment
3067 variables to the environment block passed to services invoked. One
3068 user environment generator is shipped by default that sets up
3069 environment variables based on files dropped into /etc/environment.d
3070 and ~/.config/environment.d/.
3071
3072 * systemd-resolved now includes the new, recently published 2017 DNSSEC
3073 root key (KSK).
3074
3075 * hostnamed has been updated to report a new chassis type of
3076 "convertible" to cover "foldable" laptops that can both act as a
3077 tablet and as a laptop, such as various Lenovo Yoga devices.
3078
3079 Contributions from: Adrián López, Alexander Galanin, Alexander
3080 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
3081 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
3082 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
3083 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
3084 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
3085 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
3086 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
3087 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
3088 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
3089 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
3090 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
3091 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
3092 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
3093 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
3094 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
3095 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
3096 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
3097 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
3098 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
3099 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
3100 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
3101 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
3102 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
3103 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
3104 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
3105 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
3106 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
3107 Тихонов
3108
3109 — Berlin, 2017-03-01
3110
3111 CHANGES WITH 232:
3112
3113 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
3114 RestrictAddressFamilies= enabled. These sandboxing options should
3115 generally be compatible with the various external udev call-out
3116 binaries we are aware of, however there may be exceptions, in
3117 particular when exotic languages for these call-outs are used. In
3118 this case, consider turning off these settings locally.
3119
3120 * The new RemoveIPC= option can be used to remove IPC objects owned by
3121 the user or group of a service when that service exits.
3122
3123 * The new ProtectKernelModules= option can be used to disable explicit
3124 load and unload operations of kernel modules by a service. In
3125 addition access to /usr/lib/modules is removed if this option is set.
3126
3127 * ProtectSystem= option gained a new value "strict", which causes the
3128 whole file system tree with the exception of /dev, /proc, and /sys,
3129 to be remounted read-only for a service.
3130
3131 * The new ProtectKernelTunables= option can be used to disable
3132 modification of configuration files in /sys and /proc by a service.
3133 Various directories and files are remounted read-only, so access is
3134 restricted even if the file permissions would allow it.
3135
3136 * The new ProtectControlGroups= option can be used to disable write
3137 access by a service to /sys/fs/cgroup.
3138
3139 * Various systemd services have been hardened with
3140 ProtectKernelTunables=yes, ProtectControlGroups=yes,
3141 RestrictAddressFamilies=.
3142
3143 * Support for dynamically creating users for the lifetime of a service
3144 has been added. If DynamicUser=yes is specified, user and group IDs
3145 will be allocated from the range 61184..65519 for the lifetime of the
3146 service. They can be resolved using the new nss-systemd.so NSS
3147 module. The module must be enabled in /etc/nsswitch.conf. Services
3148 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
3149 any resources allocated by the service will be cleaned up when the
3150 service exits. They also have ProtectHome=read-only and
3151 ProtectSystem=strict enabled, so they are not able to make any
3152 permanent modifications to the system.
3153
3154 * The nss-systemd module also always resolves root and nobody, making
3155 it possible to have no /etc/passwd or /etc/group files in minimal
3156 container or chroot environments.
3157
3158 * Services may be started with their own user namespace using the new
3159 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
3160 under which the service is running are mapped. All other users are
3161 mapped to nobody.
3162
3163 * Support for the cgroup namespace has been added to systemd-nspawn. If
3164 supported by kernel, the container system started by systemd-nspawn
3165 will have its own view of the cgroup hierarchy. This new behaviour
3166 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
3167
3168 * The new MemorySwapMax= option can be used to limit the maximum swap
3169 usage under the unified cgroup hierarchy.
3170
3171 * Support for the CPU controller in the unified cgroup hierarchy has
3172 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
3173 options. This controller requires out-of-tree patches for the kernel
3174 and the support is provisional.
3175
3176 * Mount and automount units may now be created transiently
3177 (i.e. dynamically at runtime via the bus API, instead of requiring
3178 unit files in the file system).
3179
3180 * systemd-mount is a new tool which may mount file systems – much like
3181 mount(8), optionally pulling in additional dependencies through
3182 transient .mount and .automount units. For example, this tool
3183 automatically runs fsck on a backing block device before mounting,
3184 and allows the automount logic to be used dynamically from the
3185 command line for establishing mount points. This tool is particularly
3186 useful when dealing with removable media, as it will ensure fsck is
3187 run – if necessary – before the first access and that the file system
3188 is quickly unmounted after each access by utilizing the automount
3189 logic. This maximizes the chance that the file system on the
3190 removable media stays in a clean state, and if it isn't in a clean
3191 state is fixed automatically.
3192
3193 * LazyUnmount=yes option for mount units has been added to expose the
3194 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
3195 option.
3196
3197 * /efi will be used as the mount point of the EFI boot partition, if
3198 the directory is present, and the mount point was not configured
3199 through other means (e.g. fstab). If /efi directory does not exist,
3200 /boot will be used as before. This makes it easier to automatically
3201 mount the EFI partition on systems where /boot is used for something
3202 else.
3203
3204 * When operating on GPT disk images for containers, systemd-nspawn will
3205 now mount the ESP to /boot or /efi according to the same rules as PID
3206 1 running on a host. This allows tools like "bootctl" to operate
3207 correctly within such containers, in order to make container images
3208 bootable on physical systems.
3209
3210 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
3211
3212 * Two new user session targets have been added to support running
3213 graphical sessions under the systemd --user instance:
3214 graphical-session.target and graphical-session-pre.target. See
3215 systemd.special(7) for a description of how those targets should be
3216 used.
3217
3218 * The vconsole initialization code has been significantly reworked to
3219 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
3220 support unicode keymaps. Font and keymap configuration will now be
3221 copied to all allocated virtual consoles.
3222
3223 * FreeBSD's bhyve virtualization is now detected.
3224
3225 * Information recorded in the journal for core dumps now includes the
3226 contents of /proc/mountinfo and the command line of the process at
3227 the top of the process hierarchy (which is usually the init process
3228 of the container).
3229
3230 * systemd-journal-gatewayd learned the --directory= option to serve
3231 files from the specified location.
3232
3233 * journalctl --root=… can be used to peruse the journal in the
3234 /var/log/ directories inside of a container tree. This is similar to
3235 the existing --machine= option, but does not require the container to
3236 be active.
3237
3238 * The hardware database has been extended to support
3239 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
3240 trackball devices.
3241
3242 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
3243 specify the click rate for mice which include a horizontal wheel with
3244 a click rate that is different than the one for the vertical wheel.
3245
3246 * systemd-run gained a new --wait option that makes service execution
3247 synchronous. (Specifically, the command will not return until the
3248 specified service binary exited.)
3249
3250 * systemctl gained a new --wait option that causes the start command to
3251 wait until the units being started have terminated again.
3252
3253 * A new journal output mode "short-full" has been added which displays
3254 timestamps with abbreviated English day names and adds a timezone
3255 suffix. Those timestamps include more information than the default
3256 "short" output mode, and can be passed directly to journalctl's
3257 --since= and --until= options.
3258
3259 * /etc/resolv.conf will be bind-mounted into containers started by
3260 systemd-nspawn, if possible, so any changes to resolv.conf contents
3261 are automatically propagated to the container.
3262
3263 * The number of instances for socket-activated services originating
3264 from a single IP address can be limited with
3265 MaxConnectionsPerSource=, extending the existing setting of
3266 MaxConnections=.
3267
3268 * systemd-networkd gained support for vcan ("Virtual CAN") interface
3269 configuration.
3270
3271 * .netdev and .network configuration can now be extended through
3272 drop-ins.
3273
3274 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
3275 Segmentation Offload, Generic Receive Offload, Large Receive Offload
3276 can be enabled and disabled using the new UDPSegmentationOffload=,
3277 TCPSegmentationOffload=, GenericSegmentationOffload=,
3278 GenericReceiveOffload=, LargeReceiveOffload= options in the
3279 [Link] section of .link files.
3280
3281 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
3282 Port VLAN ID can be configured for bridge devices using the new STP=,
3283 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
3284 section of .netdev files.
3285
3286 * The route table to which routes received over DHCP or RA should be
3287 added can be configured with the new RouteTable= option in the [DHCP]
3288 and [IPv6AcceptRA] sections of .network files.
3289
3290 * The Address Resolution Protocol can be disabled on links managed by
3291 systemd-networkd using the ARP=no setting in the [Link] section of
3292 .network files.
3293
3294 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
3295 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
3296 encode information about the result and exit codes of the current
3297 service runtime cycle.
3298
3299 * systemd-sysctl will now configure kernel parameters in the order
3300 they occur in the configuration files. This matches what sysctl
3301 has been traditionally doing.
3302
3303 * kernel-install "plugins" that are executed to perform various
3304 tasks after a new kernel is added and before an old one is removed
3305 can now return a special value to terminate the procedure and
3306 prevent any later plugins from running.
3307
3308 * Journald's SplitMode=login setting has been deprecated. It has been
3309 removed from documentation, and its use is discouraged. In a future
3310 release it will be completely removed, and made equivalent to current
3311 default of SplitMode=uid.
3312
3313 * Storage=both option setting in /etc/systemd/coredump.conf has been
3314 removed. With fast LZ4 compression storing the core dump twice is not
3315 useful.
3316
3317 * The --share-system systemd-nspawn option has been replaced with an
3318 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
3319 this functionality is discouraged. In addition the variables
3320 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
3321 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
3322 individual namespaces.
3323
3324 * "machinectl list" now shows the IP address of running containers in
3325 the output, as well as OS release information.
3326
3327 * "loginctl list" now shows the TTY of each session in the output.
3328
3329 * sd-bus gained new API calls sd_bus_track_set_recursive(),
3330 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
3331 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
3332 tracking objects in a "recursive" mode, where a single client can be
3333 counted multiple times, if it takes multiple references.
3334
3335 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
3336 sd_bus_get_exit_on_disconnect(). They may be used to make a
3337 process using sd-bus automatically exit if the bus connection is
3338 severed.
3339
3340 * Bus clients of the service manager may now "pin" loaded units into
3341 memory, by taking an explicit reference on them. This is useful to
3342 ensure the client can retrieve runtime data about the service even
3343 after the service completed execution. Taking such a reference is
3344 available only for privileged clients and should be helpful to watch
3345 running services in a race-free manner, and in particular collect
3346 information about exit statuses and results.
3347
3348 * The nss-resolve module has been changed to strictly return UNAVAIL
3349 when communication via D-Bus with resolved failed, and NOTFOUND when
3350 a lookup completed but was negative. This means it is now possible to
3351 neatly configure fallbacks using nsswitch.conf result checking
3352 expressions. Taking benefit of this, the new recommended
3353 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
3354
3355 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
3356
3357 * A new setting CtrlAltDelBurstAction= has been added to
3358 /etc/systemd/system.conf which may be used to configure the precise
3359 behaviour if the user on the console presses Ctrl-Alt-Del more often
3360 than 7 times in 2s. Previously this would unconditionally result in
3361 an expedited, immediate reboot. With this new setting the precise
3362 operation may be configured in more detail, and also turned off
3363 entirely.
3364
3365 * In .netdev files two new settings RemoteChecksumTx= and
3366 RemoteChecksumRx= are now understood that permit configuring the
3367 remote checksumming logic for VXLAN networks.
3368
3369 * The service manager learnt a new "invocation ID" concept for invoked
3370 services. Each runtime cycle of a service will get a new invocation
3371 ID (a 128bit random UUID) assigned that identifies the current
3372 run of the service uniquely and globally. A new invocation ID
3373 is generated each time a service starts up. The journal will store
3374 the invocation ID of a service along with any logged messages, thus
3375 making the invocation ID useful for matching the online runtime of a
3376 service with the offline log data it generated in a safe way without
3377 relying on synchronized timestamps. In many ways this new service
3378 invocation ID concept is similar to the kernel's boot ID concept that
3379 uniquely and globally identifies the runtime of each boot. The
3380 invocation ID of a service is passed to the service itself via an
3381 environment variable ($INVOCATION_ID). A new bus call
3382 GetUnitByInvocationID() has been added that is similar to GetUnit()
3383 but instead of retrieving the bus path for a unit by its name
3384 retrieves it by its invocation ID. The returned path is valid only as
3385 long as the passed invocation ID is current.
3386
3387 * systemd-resolved gained a new "DNSStubListener" setting in
3388 resolved.conf. It either takes a boolean value or the special values
3389 "udp" and "tcp", and configures whether to enable the stub DNS
3390 listener on 127.0.0.53:53.
3391
3392 * IP addresses configured via networkd may now carry additional
3393 configuration settings supported by the kernel. New options include:
3394 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
3395 PrefixRoute=, AutoJoin=.
3396
3397 * The PAM configuration fragment file for "user@.service" shipped with
3398 systemd (i.e. the --user instance of systemd) has been stripped to
3399 the minimum necessary to make the system boot. Previously, it
3400 contained Fedora-specific stanzas that did not apply to other
3401 distributions. It is expected that downstream distributions add
3402 additional configuration lines, matching their needs to this file,
3403 using it only as rough template of what systemd itself needs. Note
3404 that this reduced fragment does not even include an invocation of
3405 pam_limits which most distributions probably want to add, even though
3406 systemd itself does not need it. (There's also the new build time
3407 option --with-pamconfdir=no to disable installation of the PAM
3408 fragment entirely.)
3409
3410 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
3411 capability is now also dropped from its set (in addition to
3412 CAP_SYS_MKNOD as before).
3413
3414 * In service unit files it is now possible to connect a specific named
3415 file descriptor with stdin/stdout/stdout of an executed service. The
3416 name may be specified in matching .socket units using the
3417 FileDescriptorName= setting.
3418
3419 * A number of journal settings may now be configured on the kernel
3420 command line. Specifically, the following options are now understood:
3421 systemd.journald.max_level_console=,
3422 systemd.journald.max_level_store=,
3423 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
3424 systemd.journald.max_level_wall=.
3425
3426 * "systemctl is-enabled --full" will now show by which symlinks a unit
3427 file is enabled in the unit dependency tree.
3428
3429 * Support for VeraCrypt encrypted partitions has been added to the
3430 "cryptsetup" logic and /etc/crypttab.
3431
3432 * systemd-detect-virt gained support for a new --private-users switch
3433 that checks whether the invoking processes are running inside a user
3434 namespace. Similar, a new special value "private-users" for the
3435 existing ConditionVirtualization= setting has been added, permitting
3436 skipping of specific units in user namespace environments.
3437
3438 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
3439 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
3440 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
3441 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
3442 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
3443 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
3444 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
3445 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
3446 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
3447 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
3448 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
3449 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
3450 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
3451 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
3452 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
3453 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
3454 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
3455 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
3456 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
3457 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
3458 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
3459 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
3460 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
3461 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
3462 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
3463 Jędrzejewski-Szmek, Zeal Jagannatha
3464
3465 — Santa Fe, 2016-11-03
3466
3467 CHANGES WITH 231:
3468
3469 * In service units the various ExecXYZ= settings have been extended
3470 with an additional special character as first argument of the
3471 assigned value: if the character '+' is used the specified command
3472 line it will be run with full privileges, regardless of User=,
3473 Group=, CapabilityBoundingSet= and similar options. The effect is
3474 similar to the existing PermissionsStartOnly= option, but allows
3475 configuration of this concept for each executed command line
3476 independently.
3477
3478 * Services may now alter the service watchdog timeout at runtime by
3479 sending a WATCHDOG_USEC= message via sd_notify().
3480
3481 * MemoryLimit= and related unit settings now optionally take percentage
3482 specifications. The percentage is taken relative to the amount of
3483 physical memory in the system (or in case of containers, the assigned
3484 amount of memory). This allows scaling service resources neatly with
3485 the amount of RAM available on the system. Similarly, systemd-logind's
3486 RuntimeDirectorySize= option now also optionally takes percentage
3487 values.
3488
3489 * In similar fashion TasksMax= takes percentage values now, too. The
3490 value is taken relative to the configured maximum number of processes
3491 on the system. The per-service task maximum has been changed to 15%
3492 using this functionality. (Effectively this is an increase of 512 →
3493 4915 for service units, given the kernel's default pid_max setting.)
3494
3495 * Calendar time specifications in .timer units now understand a ".."
3496 syntax for time ranges. Example: "4..7:10" may now be used for
3497 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
3498 7:10am every day.
3499
3500 * The InaccessableDirectories=, ReadOnlyDirectories= and
3501 ReadWriteDirectories= unit file settings have been renamed to
3502 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
3503 applied to all kinds of file nodes, and not just directories, with
3504 the exception of symlinks. Specifically these settings may now be
3505 used on block and character device nodes, UNIX sockets and FIFOS as
3506 well as regular files. The old names of these settings remain
3507 available for compatibility.
3508
3509 * systemd will now log about all service processes it kills forcibly
3510 (using SIGKILL) because they remained after the clean shutdown phase
3511 of the service completed. This should help identifying services that
3512 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
3513 systemd-logind's configuration a similar log message is generated for
3514 processes killed at the end of each session due to this setting.
3515
3516 * systemd will now set the $JOURNAL_STREAM environment variable for all
3517 services whose stdout/stderr are connected to the Journal (which
3518 effectively means by default: all services). The variable contains
3519 the device and inode number of the file descriptor used for
3520 stdout/stderr. This may be used by invoked programs to detect whether
3521 their stdout/stderr is connected to the Journal, in which case they
3522 can switch over to direct Journal communication, thus being able to
3523 pass extended, structured metadata along with their log messages. As
3524 one example, this is now used by glib's logging primitives.
3525
3526 * When using systemd's default tmp.mount unit for /tmp, the mount point
3527 will now be established with the "nosuid" and "nodev" options. This
3528 avoids privilege escalation attacks that put traps and exploits into
3529 /tmp. However, this might cause problems if you e. g. put container
3530 images or overlays into /tmp; if you need this, override tmp.mount's
3531 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
3532 desired options.
3533
3534 * systemd now supports the "memory" cgroup controller also on
3535 cgroup v2.
3536
3537 * The systemd-cgtop tool now optionally takes a control group path as
3538 command line argument. If specified, the control group list shown is
3539 limited to subgroups of that group.
3540
3541 * The SystemCallFilter= unit file setting gained support for
3542 pre-defined, named system call filter sets. For example
3543 SystemCallFilter=@clock is now an effective way to make all clock
3544 changing-related system calls unavailable to a service. A number of
3545 similar pre-defined groups are defined. Writing system call filters
3546 for system services is simplified substantially with this new
3547 concept. Accordingly, all of systemd's own, long-running services now
3548 enable system call filtering based on this, by default.
3549
3550 * A new service setting MemoryDenyWriteExecute= has been added, taking
3551 a boolean value. If turned on, a service may no longer create memory
3552 mappings that are writable and executable at the same time. This
3553 enhances security for services where this is enabled as it becomes
3554 harder to dynamically write and then execute memory in exploited
3555 service processes. This option has been enabled for all of systemd's
3556 own long-running services.
3557
3558 * A new RestrictRealtime= service setting has been added, taking a
3559 boolean argument. If set the service's processes may no longer
3560 acquire realtime scheduling. This improves security as realtime
3561 scheduling may otherwise be used to easily freeze the system.
3562
3563 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
3564 value. This may be used for requesting that the system manager inside
3565 of the container reports start-up completion to nspawn which then
3566 propagates this notification further to the service manager
3567 supervising nspawn itself. A related option NotifyReady= in .nspawn
3568 files has been added too. This functionality allows ordering of the
3569 start-up of multiple containers using the usual systemd ordering
3570 primitives.
3571
3572 * machinectl gained a new command "stop" that is an alias for
3573 "terminate".
3574
3575 * systemd-resolved gained support for contacting DNS servers on
3576 link-local IPv6 addresses.
3577
3578 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
3579 its caches. A method call for requesting the same operation has been
3580 added to the bus API too, and is made available via "systemd-resolve
3581 --flush-caches".
3582
3583 * systemd-resolve gained a new --status switch. If passed a brief
3584 summary of the used DNS configuration with per-interface information
3585 is shown.
3586
3587 * resolved.conf gained a new Cache= boolean option, defaulting to
3588 on. If turned off local DNS caching is disabled. This comes with a
3589 performance penalty in particular when DNSSEC is enabled. Note that
3590 resolved disables its internal caching implicitly anyway, when the
3591 configured DNS server is on a host-local IP address such as ::1 or
3592 127.0.0.1, thus automatically avoiding double local caching.
3593
3594 * systemd-resolved now listens on the local IP address 127.0.0.53:53
3595 for DNS requests. This improves compatibility with local programs
3596 that do not use the libc NSS or systemd-resolved's bus APIs for name
3597 resolution. This minimal DNS service is only available to local
3598 programs and does not implement the full DNS protocol, but enough to
3599 cover local DNS clients. A new, static resolv.conf file, listing just
3600 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
3601 now recommended to make /etc/resolv.conf a symlink to this file in
3602 order to route all DNS lookups to systemd-resolved, regardless if
3603 done via NSS, the bus API or raw DNS packets. Note that this local
3604 DNS service is not as fully featured as the libc NSS or
3605 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
3606 used to deliver link-local address information (as this implies
3607 sending a local interface index along), LLMNR/mDNS support via this
3608 interface is severely restricted. It is thus strongly recommended for
3609 all applications to use the libc NSS API or native systemd-resolved
3610 bus API instead.
3611
3612 * systemd-networkd's bridge support learned a new setting
3613 VLANFiltering= for controlling VLAN filtering. Moreover a new section
3614 in .network files has been added for configuring VLAN bridging in
3615 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
3616
3617 * systemd-networkd's IPv6 Router Advertisement code now makes use of
3618 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
3619 now be acquired without relying on DHCPv6. Two new options
3620 UseDomains= and UseDNS= have been added to configure this behaviour.
3621
3622 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
3623 renamed IPv6AcceptRA=, without altering its behaviour. The old
3624 setting name remains available for compatibility reasons.
3625
3626 * The systemd-networkd VTI/VTI6 tunneling support gained new options
3627 Key=, InputKey= and OutputKey=.
3628
3629 * systemd-networkd gained support for VRF ("Virtual Routing Function")
3630 interface configuration.
3631
3632 * "systemctl edit" may now be used to create new unit files by
3633 specifying the --force switch.
3634
3635 * sd-event gained a new function sd_event_get_iteration() for
3636 requesting the current iteration counter of the event loop. It starts
3637 at zero and is increased by one with each event loop iteration.
3638
3639 * A new rpm macro %systemd_ordering is provided by the macros.systemd
3640 file. It can be used in lieu of %systemd_requires in packages which
3641 don't use any systemd functionality and are intended to be installed
3642 in minimal containers without systemd present. This macro provides
3643 ordering dependencies to ensure that if the package is installed in
3644 the same rpm transaction as systemd, systemd will be installed before
3645 the scriptlets for the package are executed, allowing unit presets
3646 to be handled.
3647
3648 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
3649 been added to simplify packaging of generators.
3650
3651 * The os-release file gained VERSION_CODENAME field for the
3652 distribution nickname (e.g. VERSION_CODENAME=woody).
3653
3654 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
3655 can be set to disable parsing of metadata and the creation
3656 of persistent symlinks for that device.
3657
3658 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
3659 to make them available to logged-in users has been reverted.
3660
3661 * Much of the common code of the various systemd components is now
3662 built into an internal shared library libsystemd-shared-231.so
3663 (incorporating the systemd version number in the name, to be updated
3664 with future releases) that the components link to. This should
3665 decrease systemd footprint both in memory during runtime and on
3666 disk. Note that the shared library is not for public use, and is
3667 neither API nor ABI stable, but is likely to change with every new
3668 released update. Packagers need to make sure that binaries
3669 linking to libsystemd-shared.so are updated in step with the
3670 library.
3671
3672 * Configuration for "mkosi" is now part of the systemd
3673 repository. mkosi is a tool to easily build legacy-free OS images,
3674 and is available on github: https://github.com/systemd/mkosi. If
3675 "mkosi" is invoked in the build tree a new raw OS image is generated
3676 incorporating the systemd sources currently being worked on and a
3677 clean, fresh distribution installation. The generated OS image may be
3678 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
3679 UEFI PC. This functionality is particularly useful to easily test
3680 local changes made to systemd in a pristine, defined environment. See
3681 doc/HACKING for details.
3682
3683 * configure learned the --with-support-url= option to specify the
3684 distribution's bugtracker.
3685
3686 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
3687 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
3688 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
3689 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
3690 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
3691 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
3692 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
3693 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
3694 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
3695 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
3696 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
3697 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
3698 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
3699 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
3700 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
3701 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
3702 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
3703 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
3704 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
3705
3706 — Berlin, 2016-07-25
3707
3708 CHANGES WITH 230:
3709
3710 * DNSSEC is now turned on by default in systemd-resolved (in
3711 "allow-downgrade" mode), but may be turned off during compile time by
3712 passing "--with-default-dnssec=no" to "configure" (and of course,
3713 during runtime with DNSSEC= in resolved.conf). We recommend
3714 downstreams to leave this on at least during development cycles and
3715 report any issues with the DNSSEC logic upstream. We are very
3716 interested in collecting feedback about the DNSSEC validator and its
3717 limitations in the wild. Note however, that DNSSEC support is
3718 probably nothing downstreams should turn on in stable distros just
3719 yet, as it might create incompatibilities with a few DNS servers and
3720 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
3721 automatically whenever we detect such incompatible setups, but there
3722 might be systems we do not cover yet. Hence: please help us testing
3723 the DNSSEC code, leave this on where you can, report back, but then
3724 again don't consider turning this on in your stable, LTS or
3725 production release just yet. (Note that you have to enable
3726 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
3727 and its DNSSEC mode for host name resolution from local
3728 applications.)
3729
3730 * systemd-resolve conveniently resolves DANE records with the --tlsa
3731 option and OPENPGPKEY records with the --openpgp option. It also
3732 supports dumping raw DNS record data via the new --raw= switch.
3733
3734 * systemd-logind will now by default terminate user processes that are
3735 part of the user session scope unit (session-XX.scope) when the user
3736 logs out. This behavior is controlled by the KillUserProcesses=
3737 setting in logind.conf, and the previous default of "no" is now
3738 changed to "yes". This means that user sessions will be properly
3739 cleaned up after, but additional steps are necessary to allow
3740 intentionally long-running processes to survive logout.
3741
3742 While the user is logged in at least once, user@.service is running,
3743 and any service that should survive the end of any individual login
3744 session can be started at a user service or scope using systemd-run.
3745 systemd-run(1) man page has been extended with an example which shows
3746 how to run screen in a scope unit underneath user@.service. The same
3747 command works for tmux.
3748
3749 After the user logs out of all sessions, user@.service will be
3750 terminated too, by default, unless the user has "lingering" enabled.
3751 To effectively allow users to run long-term tasks even if they are
3752 logged out, lingering must be enabled for them. See loginctl(1) for
3753 details. The default polkit policy was modified to allow users to
3754 set lingering for themselves without authentication.
3755
3756 Previous defaults can be restored at compile time by the
3757 --without-kill-user-processes option to "configure".
3758
3759 * systemd-logind gained new configuration settings SessionsMax= and
3760 InhibitorsMax=, both with a default of 8192. It will not register new
3761 user sessions or inhibitors above this limit.
3762
3763 * systemd-logind will now reload configuration on SIGHUP.
3764
3765 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
3766 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
3767 enable. Also, support for the "io" cgroup controller in the unified
3768 hierarchy has been added, so that the "memory", "pids" and "io" are
3769 now the controllers that are supported on the unified hierarchy.
3770
3771 WARNING: it is not possible to use previous systemd versions with
3772 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
3773 is necessary to also update systemd in the initramfs if using the
3774 unified hierarchy. An updated SELinux policy is also required.
3775
3776 * LLDP support has been extended, and both passive (receive-only) and
3777 active (sender) modes are supported. Passive mode ("routers-only") is
3778 enabled by default in systemd-networkd. Active LLDP mode is enabled
3779 by default for containers on the internal network. The "networkctl
3780 lldp" command may be used to list information gathered. "networkctl
3781 status" will also show basic LLDP information on connected peers now.
3782
3783 * The IAID and DUID unique identifier sent in DHCP requests may now be
3784 configured for the system and each .network file managed by
3785 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
3786
3787 * systemd-networkd gained support for configuring proxy ARP support for
3788 each interface, via the ProxyArp= setting in .network files. It also
3789 gained support for configuring the multicast querier feature of
3790 bridge devices, via the new MulticastQuerier= setting in .netdev
3791 files. Similarly, snooping on the IGMP traffic can be controlled
3792 via the new setting MulticastSnooping=.
3793
3794 A new setting PreferredLifetime= has been added for addresses
3795 configured in .network file to configure the lifetime intended for an
3796 address.
3797
3798 The systemd-networkd DHCP server gained the option EmitRouter=, which
3799 defaults to yes, to configure whether the DHCP Option 3 (Router)
3800 should be emitted.
3801
3802 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
3803 systemd-socket-activate and installed into /usr/bin. It is now fully
3804 supported.
3805
3806 * systemd-journald now uses separate threads to flush changes to disk
3807 when closing journal files, thus reducing impact of slow disk I/O on
3808 logging performance.
3809
3810 * The sd-journal API gained two new calls
3811 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
3812 can be used to open journal files using file descriptors instead of
3813 file or directory paths. sd_journal_open_container() has been
3814 deprecated, sd_journal_open_directory_fd() should be used instead
3815 with the flag SD_JOURNAL_OS_ROOT.
3816
3817 * journalctl learned a new output mode "-o short-unix" that outputs log
3818 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
3819 UTC). It also gained support for a new --no-hostname setting to
3820 suppress the hostname column in the family of "short" output modes.
3821
3822 * systemd-ask-password now optionally skips printing of the password to
3823 stdout with --no-output which can be useful in scripts.
3824
3825 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
3826 (devices tagged with ID_MAKER_TOOL) are now tagged with
3827 "uaccess" and are available to logged in users.
3828
3829 * The DeviceAllow= unit setting now supports specifiers (with "%").
3830
3831 * "systemctl show" gained a new --value switch, which allows print a
3832 only the contents of a specific unit property, without also printing
3833 the property's name. Similar support was added to "show*" verbs
3834 of loginctl and machinectl that output "key=value" lists.
3835
3836 * A new unit type "generated" was added for files dynamically generated
3837 by generator tools. Similarly, a new unit type "transient" is used
3838 for unit files created using the runtime API. "systemctl enable" will
3839 refuse to operate on such files.
3840
3841 * A new command "systemctl revert" has been added that may be used to
3842 revert to the vendor version of a unit file, in case local changes
3843 have been made by adding drop-ins or overriding the unit file.
3844
3845 * "machinectl clean" gained a new verb to automatically remove all or
3846 just hidden container images.
3847
3848 * systemd-tmpfiles gained support for a new line type "e" for emptying
3849 directories, if they exist, without creating them if they don't.
3850
3851 * systemd-nspawn gained support for automatically patching the UID/GIDs
3852 of the owners and the ACLs of all files and directories in a
3853 container tree to match the UID/GID user namespacing range selected
3854 for the container invocation. This mode is enabled via the new
3855 --private-users-chown switch. It also gained support for
3856 automatically choosing a free, previously unused UID/GID range when
3857 starting a container, via the new --private-users=pick setting (which
3858 implies --private-users-chown). Together, these options for the first
3859 time make user namespacing for nspawn containers fully automatic and
3860 thus deployable. The systemd-nspawn@.service template unit file has
3861 been changed to use this functionality by default.
3862
3863 * systemd-nspawn gained a new --network-zone= switch, that allows
3864 creating ad-hoc virtual Ethernet links between multiple containers,
3865 that only exist as long as at least one container referencing them is
3866 running. This allows easy connecting of multiple containers with a
3867 common link that implements an Ethernet broadcast domain. Each of
3868 these network "zones" may be named relatively freely by the user, and
3869 may be referenced by any number of containers, but each container may
3870 only reference one of these "zones". On the lower level, this is
3871 implemented by an automatically managed bridge network interface for
3872 each zone, that is created when the first container referencing its
3873 zone is created and removed when the last one referencing its zone
3874 terminates.
3875
3876 * The default start timeout may now be configured on the kernel command
3877 line via systemd.default_timeout_start_sec=. It was already
3878 configurable via the DefaultTimeoutStartSec= option in
3879 /etc/systemd/system.conf.
3880
3881 * Socket units gained a new TriggerLimitIntervalSec= and
3882 TriggerLimitBurst= setting to configure a limit on the activation
3883 rate of the socket unit.
3884
3885 * The LimitNICE= setting now optionally takes normal UNIX nice values
3886 in addition to the raw integer limit value. If the specified
3887 parameter is prefixed with "+" or "-" and is in the range -20..19 the
3888 value is understood as UNIX nice value. If not prefixed like this it
3889 is understood as raw RLIMIT_NICE limit.
3890
3891 * Note that the effect of the PrivateDevices= unit file setting changed
3892 slightly with this release: the per-device /dev file system will be
3893 mounted read-only from this version on, and will have "noexec"
3894 set. This (minor) change of behavior might cause some (exceptional)
3895 legacy software to break, when PrivateDevices=yes is set for its
3896 service. Please leave PrivateDevices= off if you run into problems
3897 with this.
3898
3899 * systemd-bootchart has been split out to a separate repository:
3900 https://github.com/systemd/systemd-bootchart
3901
3902 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
3903 merged into the kernel in its current form.
3904
3905 * The compatibility libraries libsystemd-daemon.so,
3906 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
3907 which have been deprecated since systemd-209 have been removed along
3908 with the corresponding pkg-config files. All symbols provided by
3909 those libraries are provided by libsystemd.so.
3910
3911 * The Capabilities= unit file setting has been removed (it is ignored
3912 for backwards compatibility). AmbientCapabilities= and
3913 CapabilityBoundingSet= should be used instead.
3914
3915 * A new special target has been added, initrd-root-device.target,
3916 which creates a synchronization point for dependencies of the root
3917 device in early userspace. Initramfs builders must ensure that this
3918 target is now included in early userspace.
3919
3920 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
3921 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
3922 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
3923 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
3924 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
3925 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
3926 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
3927 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
3928 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
3929 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
3930 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
3931 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
3932 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
3933 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
3934 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
3935 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
3936 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
3937 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
3938 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
3939 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
3940 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
3941 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
3942 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
3943 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
3944 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
3945 Jędrzejewski-Szmek
3946
3947 — Fairfax, 2016-05-21
3948
3949 CHANGES WITH 229:
3950
3951 * The systemd-resolved DNS resolver service has gained a substantial
3952 set of new features, most prominently it may now act as a DNSSEC
3953 validating stub resolver. DNSSEC mode is currently turned off by
3954 default, but is expected to be turned on by default in one of the
3955 next releases. For now, we invite everybody to test the DNSSEC logic
3956 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
3957 service also gained a full set of D-Bus interfaces, including calls
3958 to configure DNS and DNSSEC settings per link (for use by external
3959 network management software). systemd-resolved and systemd-networkd
3960 now distinguish between "search" and "routing" domains. The former
3961 are used to qualify single-label names, the latter are used purely
3962 for routing lookups within certain domains to specific links.
3963 resolved now also synthesizes RRs for all entries from /etc/hosts.
3964
3965 * The systemd-resolve tool (which is a client utility for
3966 systemd-resolved) has been improved considerably and is now fully
3967 supported and documented. Hence it has moved from /usr/lib/systemd to
3968 /usr/bin.
3969
3970 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
3971 devices.
3972
3973 * The coredump collection logic has been reworked: when a coredump is
3974 collected it is now written to disk, compressed and processed
3975 (including stacktrace extraction) from a new instantiated service
3976 systemd-coredump@.service, instead of directly from the
3977 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
3978 processing large coredumps can take up a substantial amount of
3979 resources and time, and this previously happened entirely outside of
3980 systemd's service supervision. With the new logic the core_pattern
3981 hook only does minimal metadata collection before passing off control
3982 to the new instantiated service, which is configured with a time
3983 limit, a nice level and other settings to minimize negative impact on
3984 the rest of the system. Also note that the new logic will honour the
3985 RLIMIT_CORE setting of the crashed process, which now allows users
3986 and processes to turn off coredumping for their processes by setting
3987 this limit.
3988
3989 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
3990 and all forked processes by default. Previously, PID 1 would leave
3991 the setting at "0" for all processes, as set by the kernel. Note that
3992 the resource limit traditionally has no effect on the generated
3993 coredumps on the system if the /proc/sys/kernel/core_pattern hook
3994 logic is used. Since the limit is now honoured (see above) its
3995 default has been changed so that the coredumping logic is enabled by
3996 default for all processes, while allowing specific opt-out.
3997
3998 * When the stacktrace is extracted from processes of system users, this
3999 is now done as "systemd-coredump" user, in order to sandbox this
4000 potentially security sensitive parsing operation. (Note that when
4001 processing coredumps of normal users this is done under the user ID
4002 of process that crashed, as before.) Packagers should take notice
4003 that it is now necessary to create the "systemd-coredump" system user
4004 and group at package installation time.
4005
4006 * The systemd-activate socket activation testing tool gained support
4007 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
4008 and --seqpacket switches. It also has been extended to support both
4009 new-style and inetd-style file descriptor passing. Use the new
4010 --inetd switch to request inetd-style file descriptor passing.
4011
4012 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
4013 variable, which takes a boolean value. If set to false, ANSI color
4014 output is disabled in the tools even when run on a terminal that
4015 supports it.
4016
4017 * The VXLAN support in networkd now supports two new settings
4018 DestinationPort= and PortRange=.
4019
4020 * A new systemd.machine_id= kernel command line switch has been added,
4021 that may be used to set the machine ID in /etc/machine-id if it is
4022 not initialized yet. This command line option has no effect if the
4023 file is already initialized.
4024
4025 * systemd-nspawn gained a new --as-pid2 switch that invokes any
4026 specified command line as PID 2 rather than PID 1 in the
4027 container. In this mode PID 1 is a minimal stub init process that
4028 implements the special POSIX and Linux semantics of PID 1 regarding
4029 signal and child process management. Note that this stub init process
4030 is implemented in nspawn itself and requires no support from the
4031 container image. This new logic is useful to support running
4032 arbitrary commands in the container, as normal processes are
4033 generally not prepared to run as PID 1.
4034
4035 * systemd-nspawn gained a new --chdir= switch for setting the current
4036 working directory for the process started in the container.
4037
4038 * "journalctl /dev/sda" will now output all kernel log messages for
4039 specified device from the current boot, in addition to all devices
4040 that are parents of it. This should make log output about devices
4041 pretty useful, as long as kernel drivers attach enough metadata to
4042 the log messages. (The usual SATA drivers do.)
4043
4044 * The sd-journal API gained two new calls
4045 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
4046 that report whether log data from /run or /var has been found.
4047
4048 * journalctl gained a new switch "--fields" that prints all journal
4049 record field names currently in use in the journal. This is backed
4050 by two new sd-journal API calls sd_journal_enumerate_fields() and
4051 sd_journal_restart_fields().
4052
4053 * Most configurable timeouts in systemd now expect an argument of
4054 "infinity" to turn them off, instead of "0" as before. The semantics
4055 from now on is that a timeout of "0" means "now", and "infinity"
4056 means "never". To maintain backwards compatibility, "0" continues to
4057 turn off previously existing timeout settings.
4058
4059 * "systemctl reload-or-try-restart" has been renamed to "systemctl
4060 try-reload-or-restart" to clarify what it actually does: the "try"
4061 logic applies to both reloading and restarting, not just restarting.
4062 The old name continues to be accepted for compatibility.
4063
4064 * On boot-up, when PID 1 detects that the system clock is behind the
4065 release date of the systemd version in use, the clock is now set
4066 to the latter. Previously, this was already done in timesyncd, in order
4067 to avoid running with clocks set to the various clock epochs such as
4068 1902, 1938 or 1970. With this change the logic is now done in PID 1
4069 in addition to timesyncd during early boot-up, so that it is enforced
4070 before the first process is spawned by systemd. Note that the logic
4071 in timesyncd remains, as it is more comprehensive and ensures
4072 clock monotonicity by maintaining a persistent timestamp file in
4073 /var. Since /var is generally not available in earliest boot or the
4074 initrd, this part of the logic remains in timesyncd, and is not done
4075 by PID 1.
4076
4077 * Support for tweaking details in net_cls.class_id through the
4078 NetClass= configuration directive has been removed, as the kernel
4079 people have decided to deprecate that controller in cgroup v2.
4080 Userspace tools such as nftables are moving over to setting rules
4081 that are specific to the full cgroup path of a task, which obsoletes
4082 these controllers anyway. The NetClass= directive is kept around for
4083 legacy compatibility reasons. For a more in-depth description of the
4084 kernel change, please refer to the respective upstream commit:
4085
4086 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
4087
4088 * A new service setting RuntimeMaxSec= has been added that may be used
4089 to specify a maximum runtime for a service. If the timeout is hit, the
4090 service is terminated and put into a failure state.
4091
4092 * A new service setting AmbientCapabilities= has been added. It allows
4093 configuration of additional Linux process capabilities that are
4094 passed to the activated processes. This is only available on very
4095 recent kernels.
4096
4097 * The process resource limit settings in service units may now be used
4098 to configure hard and soft limits individually.
4099
4100 * The various libsystemd APIs such as sd-bus or sd-event now publicly
4101 expose support for gcc's __attribute__((cleanup())) C extension.
4102 Specifically, for many object destructor functions alternative
4103 versions have been added that have names suffixed with "p" and take a
4104 pointer to a pointer to the object to destroy, instead of just a
4105 pointer to the object itself. This is useful because these destructor
4106 functions may be used directly as parameters to the cleanup
4107 construct. Internally, systemd has been a heavy user of this GCC
4108 extension for a long time, and with this change similar support is
4109 now available to consumers of the library outside of systemd. Note
4110 that by using this extension in your sources compatibility with old
4111 and strictly ANSI compatible C compilers is lost. However, all gcc or
4112 LLVM versions of recent years support this extension.
4113
4114 * Timer units gained support for a new setting RandomizedDelaySec= that
4115 allows configuring some additional randomized delay to the configured
4116 time. This is useful to spread out timer events to avoid load peaks in
4117 clusters or larger setups.
4118
4119 * Calendar time specifications now support sub-second accuracy.
4120
4121 * Socket units now support listening on SCTP and UDP-lite protocol
4122 sockets.
4123
4124 * The sd-event API now comes with a full set of man pages.
4125
4126 * Older versions of systemd contained experimental support for
4127 compressing journal files and coredumps with the LZ4 compressor that
4128 was not compatible with the lz4 binary (due to API limitations of the
4129 lz4 library). This support has been removed; only support for files
4130 compatible with the lz4 binary remains. This LZ4 logic is now
4131 officially supported and no longer considered experimental.
4132
4133 * The dkr image import logic has been removed again from importd. dkr's
4134 micro-services focus doesn't fit into the machine image focus of
4135 importd, and quickly got out of date with the upstream dkr API.
4136
4137 * Creation of the /run/lock/lockdev/ directory was dropped from
4138 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
4139 been available for many years. If you still need this, you need to
4140 create your own tmpfiles.d config file with:
4141
4142 d /run/lock/lockdev 0775 root lock -
4143
4144 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
4145 and RebootArgument= have been moved from the [Service] section of
4146 unit files to [Unit], and they are now supported on all unit types,
4147 not just service units. Of course, systemd will continue to
4148 understand these settings also at the old location, in order to
4149 maintain compatibility.
4150
4151 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
4152 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
4153 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
4154 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
4155 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
4156 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
4157 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
4158 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
4159 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
4160 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
4161 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
4162 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
4163 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
4164 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
4165 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
4166 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
4167 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
4168 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
4169 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
4170
4171 — Berlin, 2016-02-11
4172
4173 CHANGES WITH 228:
4174
4175 * A number of properties previously only settable in unit
4176 files are now also available as properties to set when
4177 creating transient units programmatically via the bus, as it
4178 is exposed with systemd-run's --property=
4179 setting. Specifically, these are: SyslogIdentifier=,
4180 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
4181 EnvironmentFile=, ReadWriteDirectories=,
4182 ReadOnlyDirectories=, InaccessibleDirectories=,
4183 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
4184
4185 * When creating transient services via the bus API it is now
4186 possible to pass in a set of file descriptors to use as
4187 STDIN/STDOUT/STDERR for the invoked process.
4188
4189 * Slice units may now be created transiently via the bus APIs,
4190 similar to the way service and scope units may already be
4191 created transiently.
4192
4193 * Wherever systemd expects a calendar timestamp specification
4194 (like in journalctl's --since= and --until= switches) UTC
4195 timestamps are now supported. Timestamps suffixed with "UTC"
4196 are now considered to be in Universal Time Coordinated
4197 instead of the local timezone. Also, timestamps may now
4198 optionally be specified with sub-second accuracy. Both of
4199 these additions also apply to recurring calendar event
4200 specification, such as OnCalendar= in timer units.
4201
4202 * journalctl gained a new "--sync" switch that asks the
4203 journal daemon to write all so far unwritten log messages to
4204 disk and sync the files, before returning.
4205
4206 * systemd-tmpfiles learned two new line types "q" and "Q" that
4207 operate like "v", but also set up a basic btrfs quota
4208 hierarchy when used on a btrfs file system with quota
4209 enabled.
4210
4211 * tmpfiles' "v", "q" and "Q" will now create a plain directory
4212 instead of a subvolume (even on a btrfs file system) if the
4213 root directory is a plain directory, and not a
4214 subvolume. This should simplify things with certain chroot()
4215 environments which are not aware of the concept of btrfs
4216 subvolumes.
4217
4218 * systemd-detect-virt gained a new --chroot switch to detect
4219 whether execution takes place in a chroot() environment.
4220
4221 * CPUAffinity= now takes CPU index ranges in addition to
4222 individual indexes.
4223
4224 * The various memory-related resource limit settings (such as
4225 LimitAS=) now understand the usual K, M, G, ... suffixes to
4226 the base of 1024 (IEC). Similar, the time-related resource
4227 limit settings understand the usual min, h, day, ...
4228 suffixes now.
4229
4230 * There's a new system.conf setting DefaultTasksMax= to
4231 control the default TasksMax= setting for services and
4232 scopes running on the system. (TasksMax= is the primary
4233 setting that exposes the "pids" cgroup controller on systemd
4234 and was introduced in the previous systemd release.) The
4235 setting now defaults to 512, which means services that are
4236 not explicitly configured otherwise will only be able to
4237 create 512 processes or threads at maximum, from this
4238 version on. Note that this means that thread- or
4239 process-heavy services might need to be reconfigured to set
4240 TasksMax= to a higher value. It is sufficient to set
4241 TasksMax= in these specific unit files to a higher value, or
4242 even "infinity". Similar, there's now a logind.conf setting
4243 UserTasksMax= that defaults to 4096 and limits the total
4244 number of processes or tasks each user may own
4245 concurrently. nspawn containers also have the TasksMax=
4246 value set by default now, to 8192. Note that all of this
4247 only has an effect if the "pids" cgroup controller is
4248 enabled in the kernel. The general benefit of these changes
4249 should be a more robust and safer system, that provides a
4250 certain amount of per-service fork() bomb protection.
4251
4252 * systemd-nspawn gained the new --network-veth-extra= switch
4253 to define additional and arbitrarily-named virtual Ethernet
4254 links between the host and the container.
4255
4256 * A new service execution setting PassEnvironment= has been
4257 added that allows importing select environment variables
4258 from PID1's environment block into the environment block of
4259 the service.
4260
4261 * Timer units gained support for a new RemainAfterElapse=
4262 setting which takes a boolean argument. It defaults to on,
4263 exposing behaviour unchanged to previous releases. If set to
4264 off, timer units are unloaded after they elapsed if they
4265 cannot elapse again. This is particularly useful for
4266 transient timer units, which shall not stay around longer
4267 than until they first elapse.
4268
4269 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
4270 default now (the kernel default is 16). This is beneficial
4271 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
4272 allows substantially larger numbers of queued
4273 datagrams. This should increase the capability of systemd to
4274 parallelize boot-up, as logging and sd_notify() are unlikely
4275 to stall execution anymore. If you need to change the value
4276 from the new defaults, use the usual sysctl.d/ snippets.
4277
4278 * The compression framing format used by the journal or
4279 coredump processing has changed to be in line with what the
4280 official LZ4 tools generate. LZ4 compression support in
4281 systemd was considered unsupported previously, as the format
4282 was not compatible with the normal tools. With this release
4283 this has changed now, and it is hence safe for downstream
4284 distributions to turn it on. While not compressing as well
4285 as the XZ, LZ4 is substantially faster, which makes
4286 it a good default choice for the compression logic in the
4287 journal and in coredump handling.
4288
4289 * Any reference to /etc/mtab has been dropped from
4290 systemd. The file has been obsolete since a while, but
4291 systemd refused to work on systems where it was incorrectly
4292 set up (it should be a symlink or non-existent). Please make
4293 sure to update to util-linux 2.27.1 or newer in conjunction
4294 with this systemd release, which also drops any reference to
4295 /etc/mtab. If you maintain a distribution make sure that no
4296 software you package still references it, as this is a
4297 likely source of bugs. There's also a glibc bug pending,
4298 asking for removal of any reference to this obsolete file:
4299
4300 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
4301
4302 Note that only util-linux versions built with
4303 --enable-libmount-force-mountinfo are supported.
4304
4305 * Support for the ".snapshot" unit type has been removed. This
4306 feature turned out to be little useful and little used, and
4307 has now been removed from the core and from systemctl.
4308
4309 * The dependency types RequiresOverridable= and
4310 RequisiteOverridable= have been removed from systemd. They
4311 have been used only very sparingly to our knowledge and
4312 other options that provide a similar effect (such as
4313 systemctl --mode=ignore-dependencies) are much more useful
4314 and commonly used. Moreover, they were only half-way
4315 implemented as the option to control behaviour regarding
4316 these dependencies was never added to systemctl. By removing
4317 these dependency types the execution engine becomes a bit
4318 simpler. Unit files that use these dependencies should be
4319 changed to use the non-Overridable dependency types
4320 instead. In fact, when parsing unit files with these
4321 options, that's what systemd will automatically convert them
4322 too, but it will also warn, asking users to fix the unit
4323 files accordingly. Removal of these dependency types should
4324 only affect a negligible number of unit files in the wild.
4325
4326 * Behaviour of networkd's IPForward= option changed
4327 (again). It will no longer maintain a per-interface setting,
4328 but propagate one way from interfaces where this is enabled
4329 to the global kernel setting. The global setting will be
4330 enabled when requested by a network that is set up, but
4331 never be disabled again. This change was made to make sure
4332 IPv4 and IPv6 behaviour regarding packet forwarding is
4333 similar (as the Linux IPv6 stack does not support
4334 per-interface control of this setting) and to minimize
4335 surprises.
4336
4337 * In unit files the behaviour of %u, %U, %h, %s has
4338 changed. These specifiers will now unconditionally resolve
4339 to the various user database fields of the user that the
4340 systemd instance is running as, instead of the user
4341 configured in the specific unit via User=. Note that this
4342 effectively doesn't change much, as resolving of these
4343 specifiers was already turned off in the --system instance
4344 of systemd, as we cannot do NSS lookups from PID 1. In the
4345 --user instance of systemd these specifiers where correctly
4346 resolved, but hardly made any sense, since the user instance
4347 lacks privileges to do user switches anyway, and User= is
4348 hence useless. Moreover, even in the --user instance of
4349 systemd behaviour was awkward as it would only take settings
4350 from User= assignment placed before the specifier into
4351 account. In order to unify and simplify the logic around
4352 this the specifiers will now always resolve to the
4353 credentials of the user invoking the manager (which in case
4354 of PID 1 is the root user).
4355
4356 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
4357 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
4358 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
4359 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
4360 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
4361 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
4362 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
4363 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
4364 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
4365 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
4366 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
4367 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
4368 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
4369 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
4370 Jędrzejewski-Szmek
4371
4372 — Berlin, 2015-11-18
4373
4374 CHANGES WITH 227:
4375
4376 * systemd now depends on util-linux v2.27. More specifically,
4377 the newly added mount monitor feature in libmount now
4378 replaces systemd's former own implementation.
4379
4380 * libmount mandates /etc/mtab not to be regular file, and
4381 systemd now enforces this condition at early boot.
4382 /etc/mtab has been deprecated and warned about for a very
4383 long time, so systems running systemd should already have
4384 stopped having this file around as anything else than a
4385 symlink to /proc/self/mounts.
4386
4387 * Support for the "pids" cgroup controller has been added. It
4388 allows accounting the number of tasks in a cgroup and
4389 enforcing limits on it. This adds two new setting
4390 TasksAccounting= and TasksMax= to each unit, as well as a
4391 global option DefaultTasksAccounting=.
4392
4393 * Support for the "net_cls" cgroup controller has been added.
4394 It allows assigning a net class ID to each task in the
4395 cgroup, which can then be used in firewall rules and traffic
4396 shaping configurations. Note that the kernel netfilter net
4397 class code does not currently work reliably for ingress
4398 packets on unestablished sockets.
4399
4400 This adds a new config directive called NetClass= to CGroup
4401 enabled units. Allowed values are positive numbers for fixed
4402 assignments and "auto" for picking a free value
4403 automatically.
4404
4405 * 'systemctl is-system-running' now returns 'offline' if the
4406 system is not booted with systemd. This command can now be
4407 used as a substitute for 'systemd-notify --booted'.
4408
4409 * Watchdog timeouts have been increased to 3 minutes for all
4410 in-tree service files. Apparently, disk IO issues are more
4411 frequent than we hoped, and user reported >1 minute waiting
4412 for disk IO.
4413
4414 * 'machine-id-commit' functionality has been merged into
4415 'machine-id-setup --commit'. The separate binary has been
4416 removed.
4417
4418 * The WorkingDirectory= directive in unit files may now be set
4419 to the special value '~'. In this case, the working
4420 directory is set to the home directory of the user
4421 configured in User=.
4422
4423 * "machinectl shell" will now open the shell in the home
4424 directory of the selected user by default.
4425
4426 * The CrashChVT= configuration file setting is renamed to
4427 CrashChangeVT=, following our usual logic of not
4428 abbreviating unnecessarily. The old directive is still
4429 supported for compat reasons. Also, this directive now takes
4430 an integer value between 1 and 63, or a boolean value. The
4431 formerly supported '-1' value for disabling stays around for
4432 compat reasons.
4433
4434 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
4435 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
4436 RootDirectory= properties can now be set for transient
4437 units.
4438
4439 * The systemd-analyze tool gained a new "set-log-target" verb
4440 to change the logging target the system manager logs to
4441 dynamically during runtime. This is similar to how
4442 "systemd-analyze set-log-level" already changes the log
4443 level.
4444
4445 * In nspawn /sys is now mounted as tmpfs, with only a selected
4446 set of subdirectories mounted in from the real sysfs. This
4447 enhances security slightly, and is useful for ensuring user
4448 namespaces work correctly.
4449
4450 * Support for USB FunctionFS activation has been added. This
4451 allows implementation of USB gadget services that are
4452 activated as soon as they are requested, so that they don't
4453 have to run continuously, similar to classic socket
4454 activation.
4455
4456 * The "systemctl exit" command now optionally takes an
4457 additional parameter that sets the exit code to return from
4458 the systemd manager when exiting. This is only relevant when
4459 running the systemd user instance, or when running the
4460 system instance in a container.
4461
4462 * sd-bus gained the new API calls sd_bus_path_encode_many()
4463 and sd_bus_path_decode_many() that allow easy encoding and
4464 decoding of multiple identifier strings inside a D-Bus
4465 object path. Another new call sd_bus_default_flush_close()
4466 has been added to flush and close per-thread default
4467 connections.
4468
4469 * systemd-cgtop gained support for a -M/--machine= switch to
4470 show the control groups within a certain container only.
4471
4472 * "systemctl kill" gained support for an optional --fail
4473 switch. If specified the requested operation will fail of no
4474 processes have been killed, because the unit had no
4475 processes attached, or similar.
4476
4477 * A new systemd.crash_reboot=1 kernel command line option has
4478 been added that triggers a reboot after crashing. This can
4479 also be set through CrashReboot= in systemd.conf.
4480
4481 * The RuntimeDirectory= setting now understands unit
4482 specifiers like %i or %f.
4483
4484 * A new (still internal) library API sd-ipv4acd has been added,
4485 that implements address conflict detection for IPv4. It's
4486 based on code from sd-ipv4ll, and will be useful for
4487 detecting DHCP address conflicts.
4488
4489 * File descriptors passed during socket activation may now be
4490 named. A new API sd_listen_fds_with_names() is added to
4491 access the names. The default names may be overridden,
4492 either in the .socket file using the FileDescriptorName=
4493 parameter, or by passing FDNAME= when storing the file
4494 descriptors using sd_notify().
4495
4496 * systemd-networkd gained support for:
4497
4498 - Setting the IPv6 Router Advertisement settings via
4499 IPv6AcceptRouterAdvertisements= in .network files.
4500
4501 - Configuring the HelloTimeSec=, MaxAgeSec= and
4502 ForwardDelaySec= bridge parameters in .netdev files.
4503
4504 - Configuring PreferredSource= for static routes in
4505 .network files.
4506
4507 * The "ask-password" framework used to query for LUKS harddisk
4508 passwords or SSL passwords during boot gained support for
4509 caching passwords in the kernel keyring, if it is
4510 available. This makes sure that the user only has to type in
4511 a passphrase once if there are multiple objects to unlock
4512 with the same one. Previously, such password caching was
4513 available only when Plymouth was used; this moves the
4514 caching logic into the systemd codebase itself. The
4515 "systemd-ask-password" utility gained a new --keyname=
4516 switch to control which kernel keyring key to use for
4517 caching a password in. This functionality is also useful for
4518 enabling display managers such as gdm to automatically
4519 unlock the user's GNOME keyring if its passphrase, the
4520 user's password and the harddisk password are the same, if
4521 gdm-autologin is used.
4522
4523 * When downloading tar or raw images using "machinectl
4524 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
4525 file is now also downloaded, if it is available and stored
4526 next to the image file.
4527
4528 * Units of type ".socket" gained a new boolean setting
4529 Writable= which is only useful in conjunction with
4530 ListenSpecial=. If true, enables opening the specified
4531 special file in O_RDWR mode rather than O_RDONLY mode.
4532
4533 * systemd-rfkill has been reworked to become a singleton
4534 service that is activated through /dev/rfkill on each rfkill
4535 state change and saves the settings to disk. This way,
4536 systemd-rfkill is now compatible with devices that exist
4537 only intermittendly, and even restores state if the previous
4538 system shutdown was abrupt rather than clean.
4539
4540 * The journal daemon gained support for vacuuming old journal
4541 files controlled by the number of files that shall remain,
4542 in addition to the already existing control by size and by
4543 date. This is useful as journal interleaving performance
4544 degrades with too many separate journal files, and allows
4545 putting an effective limit on them. The new setting defaults
4546 to 100, but this may be changed by setting SystemMaxFiles=
4547 and RuntimeMaxFiles= in journald.conf. Also, the
4548 "journalctl" tool gained the new --vacuum-files= switch to
4549 manually vacuum journal files to leave only the specified
4550 number of files in place.
4551
4552 * udev will now create /dev/disk/by-path links for ATA devices
4553 on kernels where that is supported.
4554
4555 * Galician, Serbian, Turkish and Korean translations were added.
4556
4557 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
4558 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
4559 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
4560 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
4561 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
4562 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
4563 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
4564 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
4565 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
4566 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
4567 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
4568 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
4569 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
4570 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
4571 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
4572 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
4573 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
4574 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
4575
4576 — Berlin, 2015-10-07
4577
4578 CHANGES WITH 226:
4579
4580 * The DHCP implementation of systemd-networkd gained a set of
4581 new features:
4582
4583 - The DHCP server now supports emitting DNS and NTP
4584 information. It may be enabled and configured via
4585 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
4586 and NTP information is enabled, but no servers are
4587 configured, the corresponding uplink information (if there
4588 is any) is propagated.
4589
4590 - Server and client now support transmission and reception
4591 of timezone information. It can be configured via the
4592 newly introduced network options UseTimezone=,
4593 EmitTimezone=, and Timezone=. Transmission of timezone
4594 information is enabled between host and containers by
4595 default now: the container will change its local timezone
4596 to what the host has set.
4597
4598 - Lease timeouts can now be configured via
4599 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
4600
4601 - The DHCP server improved on the stability of
4602 leases. Clients are more likely to get the same lease
4603 information back, even if the server loses state.
4604
4605 - The DHCP server supports two new configuration options to
4606 control the lease address pool metrics, PoolOffset= and
4607 PoolSize=.
4608
4609 * The encapsulation limit of tunnels in systemd-networkd may
4610 now be configured via 'EncapsulationLimit='. It allows
4611 modifying the maximum additional levels of encapsulation
4612 that are permitted to be prepended to a packet.
4613
4614 * systemd now supports the concept of user buses replacing
4615 session buses, if used with dbus-1.10 (and enabled via dbus
4616 --enable-user-session). It previously only supported this on
4617 kdbus-enabled systems, and this release expands this to
4618 'dbus-daemon' systems.
4619
4620 * systemd-networkd now supports predictable interface names
4621 for virtio devices.
4622
4623 * systemd now optionally supports the new Linux kernel
4624 "unified" control group hierarchy. If enabled via the kernel
4625 command-line option 'systemd.unified_cgroup_hierarchy=1',
4626 systemd will try to mount the unified cgroup hierarchy
4627 directly on /sys/fs/cgroup. If not enabled, or not
4628 available, systemd will fall back to the legacy cgroup
4629 hierarchy setup, as before. Host system and containers can
4630 mix and match legacy and unified hierarchies as they
4631 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
4632 environment variable to individually select the hierarchy to
4633 use for executed containers. By default, nspawn will use the
4634 unified hierarchy for the containers if the host uses the
4635 unified hierarchy, and the legacy hierarchy otherwise.
4636 Please note that at this point the unified hierarchy is an
4637 experimental kernel feature and is likely to change in one
4638 of the next kernel releases. Therefore, it should not be
4639 enabled by default in downstream distributions yet. The
4640 minimum required kernel version for the unified hierarchy to
4641 work is 4.2. Note that when the unified hierarchy is used
4642 for the first time delegated access to controllers is
4643 safe. Because of this systemd-nspawn containers will get
4644 access to controllers now, as will systemd user
4645 sessions. This means containers and user sessions may now
4646 manage their own resources, partitioning up what the system
4647 grants them.
4648
4649 * A new special scope unit "init.scope" has been introduced
4650 that encapsulates PID 1 of the system. It may be used to
4651 determine resource usage and enforce resource limits on PID
4652 1 itself. PID 1 hence moved out of the root of the control
4653 group tree.
4654
4655 * The cgtop tool gained support for filtering out kernel
4656 threads when counting tasks in a control group. Also, the
4657 count of processes is now recursively summed up by
4658 default. Two options -k and --recursive= have been added to
4659 revert to old behaviour. The tool has also been updated to
4660 work correctly in containers now.
4661
4662 * systemd-nspawn's --bind= and --bind-ro= options have been
4663 extended to allow creation of non-recursive bind mounts.
4664
4665 * libsystemd gained two new calls sd_pid_get_cgroup() and
4666 sd_peer_get_cgroup() which return the control group path of
4667 a process or peer of a connected AF_UNIX socket. This
4668 function call is particularly useful when implementing
4669 delegated subtrees support in the control group hierarchy.
4670
4671 * The "sd-event" event loop API of libsystemd now supports
4672 correct dequeuing of real-time signals, without losing
4673 signal events.
4674
4675 * When systemd requests a polkit decision when managing units it
4676 will now add additional fields to the request, including unit
4677 name and desired operation. This enables more powerful polkit
4678 policies, that make decisions depending on these parameters.
4679
4680 * nspawn learnt support for .nspawn settings files, that may
4681 accompany the image files or directories of containers, and
4682 may contain additional settings for the container. This is
4683 an alternative to configuring container parameters via the
4684 nspawn command line.
4685
4686 Contributions from: Cristian Rodríguez, Daniel Mack, David
4687 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
4688 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
4689 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
4690 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
4691 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
4692 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
4693 Andersen, Tom Gundersen, Torstein Husebø
4694
4695 — Berlin, 2015-09-08
4696
4697 CHANGES WITH 225:
4698
4699 * machinectl gained a new verb 'shell' which opens a fresh
4700 shell on the target container or the host. It is similar to
4701 the existing 'login' command of machinectl, but spawns the
4702 shell directly without prompting for username or
4703 password. The pseudo machine '.host' now refers to the local
4704 host and is used by default. Hence, 'machinectl shell' can
4705 be used as replacement for 'su -' which spawns a session as
4706 a fresh systemd unit in a way that is fully isolated from
4707 the originating session.
4708
4709 * systemd-networkd learned to cope with private-zone DHCP
4710 options and allows other programs to query the values.
4711
4712 * SELinux access control when enabling/disabling units is no
4713 longer enforced with this release. The previous implementation
4714 was incorrect, and a new corrected implementation is not yet
4715 available. As unit file operations are still protected via
4716 polkit and D-Bus policy this is not a security problem. Yet,
4717 distributions which care about optimal SELinux support should
4718 probably not stabilize on this release.
4719
4720 * sd-bus gained support for matches of type "arg0has=", that
4721 test for membership of strings in string arrays sent in bus
4722 messages.
4723
4724 * systemd-resolved now dumps the contents of its DNS and LLMNR
4725 caches to the logs on reception of the SIGUSR1 signal. This
4726 is useful to debug DNS behaviour.
4727
4728 * The coredumpctl tool gained a new --directory= option to
4729 operate on journal files in a specific directory.
4730
4731 * "systemctl reboot" and related commands gained a new
4732 "--message=" option which may be used to set a free-text
4733 wall message when shutting down or rebooting the
4734 system. This message is also logged, which is useful for
4735 figuring out the reason for a reboot or shutdown a
4736 posteriori.
4737
4738 * The "systemd-resolve-host" tool's -i switch now takes
4739 network interface numbers as alternative to interface names.
4740
4741 * A new unit file setting for services has been introduced:
4742 UtmpMode= allows configuration of how precisely systemd
4743 handles utmp and wtmp entries for the service if this is
4744 enabled. This allows writing services that appear similar to
4745 user sessions in the output of the "w", "who", "last" and
4746 "lastlog" tools.
4747
4748 * systemd-resolved will now locally synthesize DNS resource
4749 records for the "localhost" and "gateway" domains as well as
4750 the local hostname. This should ensure that clients querying
4751 RRs via resolved will get similar results as those going via
4752 NSS, if nss-myhostname is enabled.
4753
4754 Contributions from: Alastair Hughes, Alex Crawford, Daniel
4755 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
4756 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
4757 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
4758 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
4759 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
4760 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
4761 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
4762 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
4763 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
4764 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
4765 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4766
4767 — Berlin, 2015-08-27
4768
4769 CHANGES WITH 224:
4770
4771 * The systemd-efi-boot-generator functionality was merged into
4772 systemd-gpt-auto-generator.
4773
4774 * systemd-networkd now supports Group Policy for vxlan
4775 devices. It can be enabled via the new boolean configuration
4776 option called 'GroupPolicyExtension='.
4777
4778 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
4779 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
4780 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
4781
4782 — Berlin, 2015-07-31
4783
4784 CHANGES WITH 223:
4785
4786 * The python-systemd code has been removed from the systemd repository.
4787 A new repository has been created which accommodates the code from
4788 now on, and we kindly ask distributions to create a separate package
4789 for this: https://github.com/systemd/python-systemd
4790
4791 * The systemd daemon will now reload its main configuration
4792 (/etc/systemd/system.conf) on daemon-reload.
4793
4794 * sd-dhcp now exposes vendor specific extensions via
4795 sd_dhcp_lease_get_vendor_specific().
4796
4797 * systemd-networkd gained a number of new configuration options.
4798
4799 - A new boolean configuration option for TAP devices called
4800 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
4801 device, thus allowing to send and receive GSO packets.
4802
4803 - A new tunnel configuration option called 'CopyDSCP='.
4804 If enabled, the DSCP field of ip6 tunnels is copied into the
4805 decapsulated packet.
4806
4807 - A set of boolean bridge configuration options were added.
4808 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
4809 and 'UnicastFlood=' are now parsed by networkd and applied to the
4810 respective bridge link device via the respective IFLA_BRPORT_*
4811 netlink attribute.
4812
4813 - A new string configuration option to override the hostname sent
4814 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
4815 is true, networkd will use the configured hostname instead of the
4816 system hostname when sending DHCP requests.
4817
4818 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
4819 networkd will configure the IPv6 flow-label of the tunnel device
4820 according to RFC2460.
4821
4822 - The 'macvtap' virtual network devices are now supported, similar to
4823 the already supported 'macvlan' devices.
4824
4825 * systemd-resolved now implements RFC5452 to improve resilience against
4826 cache poisoning. Additionally, source port randomization is enabled
4827 by default to further protect against DNS spoofing attacks.
4828
4829 * nss-mymachines now supports translating UIDs and GIDs of running
4830 containers with user-namespaces enabled. If a container 'foo'
4831 translates a host uid 'UID' to the container uid 'TUID', then
4832 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
4833 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
4834 mapped as 'vg-foo-TGID'.
4835
4836 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
4837 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
4838 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
4839 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
4840 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
4841 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
4842 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
4843 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
4844 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
4845 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
4846
4847 — Berlin, 2015-07-29
4848
4849 CHANGES WITH 222:
4850
4851 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
4852 There are no known issues with current sysfs, and udev does not need
4853 or should be used to work around such bugs.
4854
4855 * udev does no longer enable USB HID power management. Several reports
4856 indicate, that some devices cannot handle that setting.
4857
4858 * The udev accelerometer helper was removed. The functionality
4859 is now fully included in iio-sensor-proxy. But this means,
4860 older iio-sensor-proxy versions will no longer provide
4861 accelerometer/orientation data with this systemd version.
4862 Please upgrade iio-sensor-proxy to version 1.0.
4863
4864 * networkd gained a new configuration option IPv6PrivacyExtensions=
4865 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
4866 for Stateless Address") on selected networks.
4867
4868 * For the sake of fewer build-time dependencies and less code in the
4869 main repository, the python bindings are about to be removed in the
4870 next release. A new repository has been created which accommodates
4871 the code from now on, and we kindly ask distributions to create a
4872 separate package for this. The removal will take place in v223.
4873
4874 https://github.com/systemd/python-systemd
4875
4876 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
4877 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
4878 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
4879 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
4880 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
4881 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
4882 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
4883 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
4884 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
4885 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4886
4887 — Berlin, 2015-07-07
4888
4889 CHANGES WITH 221:
4890
4891 * The sd-bus.h and sd-event.h APIs have now been declared
4892 stable and have been added to the official interface of
4893 libsystemd.so. sd-bus implements an alternative D-Bus client
4894 library, that is relatively easy to use, very efficient and
4895 supports both classic D-Bus as well as kdbus as transport
4896 backend. sd-event is a generic event loop abstraction that
4897 is built around Linux epoll, but adds features such as event
4898 prioritization or efficient timer handling. Both APIs are good
4899 choices for C programs looking for a bus and/or event loop
4900 implementation that is minimal and does not have to be
4901 portable to other kernels.
4902
4903 * kdbus support is no longer compile-time optional. It is now
4904 always built-in. However, it can still be disabled at
4905 runtime using the kdbus=0 kernel command line setting, and
4906 that setting may be changed to default to off, by specifying
4907 --disable-kdbus at build-time. Note though that the kernel
4908 command line setting has no effect if the kdbus.ko kernel
4909 module is not installed, in which case kdbus is (obviously)
4910 also disabled. We encourage all downstream distributions to
4911 begin testing kdbus by adding it to the kernel images in the
4912 development distributions, and leaving kdbus support in
4913 systemd enabled.
4914
4915 * The minimal required util-linux version has been bumped to
4916 2.26.
4917
4918 * Support for chkconfig (--enable-chkconfig) was removed in
4919 favor of calling an abstraction tool
4920 /lib/systemd/systemd-sysv-install. This needs to be
4921 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
4922 in README for details.
4923
4924 * If there's a systemd unit and a SysV init script for the
4925 same service name, and the user executes "systemctl enable"
4926 for it (or a related call), then this will now enable both
4927 (or execute the related operation on both), not just the
4928 unit.
4929
4930 * The libudev API documentation has been converted from gtkdoc
4931 into man pages.
4932
4933 * gudev has been removed from the systemd tree, it is now an
4934 external project.
4935
4936 * The systemd-cgtop tool learnt a new --raw switch to generate
4937 "raw" (machine parsable) output.
4938
4939 * networkd's IPForwarding= .network file setting learnt the
4940 new setting "kernel", which ensures that networkd does not
4941 change the IP forwarding sysctl from the default kernel
4942 state.
4943
4944 * The systemd-logind bus API now exposes a new boolean
4945 property "Docked" that reports whether logind considers the
4946 system "docked", i.e. connected to a docking station or not.
4947
4948 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
4949 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
4950 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
4951 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
4952 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
4953 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
4954 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
4955 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
4956 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
4957 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
4958 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
4959 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
4960 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
4961 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
4962 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
4963 Fink, Zbigniew Jędrzejewski-Szmek
4964
4965 — Berlin, 2015-06-19
4966
4967 CHANGES WITH 220:
4968
4969 * The gudev library has been extracted into a separate repository
4970 available at: https://git.gnome.org/browse/libgudev/
4971 It is now managed as part of the Gnome project. Distributions
4972 are recommended to pass --disable-gudev to systemd and use
4973 gudev from the Gnome project instead. gudev is still included
4974 in systemd, for now. It will be removed soon, though. Please
4975 also see the announcement-thread on systemd-devel:
4976 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
4977
4978 * systemd now exposes a CPUUsageNSec= property for each
4979 service unit on the bus, that contains the overall consumed
4980 CPU time of a service (the sum of what each process of the
4981 service consumed). This value is only available if
4982 CPUAccounting= is turned on for a service, and is then shown
4983 in the "systemctl status" output.
4984
4985 * Support for configuring alternative mappings of the old SysV
4986 runlevels to systemd targets has been removed. They are now
4987 hardcoded in a way that runlevels 2, 3, 4 all map to
4988 multi-user.target and 5 to graphical.target (which
4989 previously was already the default behaviour).
4990
4991 * The auto-mounter logic gained support for mount point
4992 expiry, using a new TimeoutIdleSec= setting in .automount
4993 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
4994
4995 * The EFI System Partition (ESP) as mounted to /boot by
4996 systemd-efi-boot-generator will now be unmounted
4997 automatically after 2 minutes of not being used. This should
4998 minimize the risk of ESP corruptions.
4999
5000 * New /etc/fstab options x-systemd.requires= and
5001 x-systemd.requires-mounts-for= are now supported to express
5002 additional dependencies for mounts. This is useful for
5003 journalling file systems that support external journal
5004 devices or overlay file systems that require underlying file
5005 systems to be mounted.
5006
5007 * systemd does not support direct live-upgrades (via systemctl
5008 daemon-reexec) from versions older than v44 anymore. As no
5009 distribution we are aware of shipped such old versions in a
5010 stable release this should not be problematic.
5011
5012 * When systemd forks off a new per-connection service instance
5013 it will now set the $REMOTE_ADDR environment variable to the
5014 remote IP address, and $REMOTE_PORT environment variable to
5015 the remote IP port. This behaviour is similar to the
5016 corresponding environment variables defined by CGI.
5017
5018 * systemd-networkd gained support for uplink failure
5019 detection. The BindCarrier= option allows binding interface
5020 configuration dynamically to the link sense of other
5021 interfaces. This is useful to achieve behaviour like in
5022 network switches.
5023
5024 * systemd-networkd gained support for configuring the DHCP
5025 client identifier to use when requesting leases.
5026
5027 * systemd-networkd now has a per-network UseNTP= option to
5028 configure whether NTP server information acquired via DHCP
5029 is passed on to services like systemd-timesyncd.
5030
5031 * systemd-networkd gained support for vti6 tunnels.
5032
5033 * Note that systemd-networkd manages the sysctl variable
5034 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
5035 it is configured for since v219. The variable controls IP
5036 forwarding, and is a per-interface alternative to the global
5037 /proc/sys/net/ipv[46]/ip_forward. This setting is
5038 configurable in the IPForward= option, which defaults to
5039 "no". This means if networkd is used for an interface it is
5040 no longer sufficient to set the global sysctl option to turn
5041 on IP forwarding! Instead, the .network file option
5042 IPForward= needs to be turned on! Note that the
5043 implementation of this behaviour was broken in v219 and has
5044 been fixed in v220.
5045
5046 * Many bonding and vxlan options are now configurable in
5047 systemd-networkd.
5048
5049 * systemd-nspawn gained a new --property= setting to set unit
5050 properties for the container scope. This is useful for
5051 setting resource parameters (e.g. "CPUShares=500") on
5052 containers started from the command line.
5053
5054 * systemd-nspawn gained a new --private-users= switch to make
5055 use of user namespacing available on recent Linux kernels.
5056
5057 * systemd-nspawn may now be called as part of a shell pipeline
5058 in which case the pipes used for stdin and stdout are passed
5059 directly to the process invoked in the container, without
5060 indirection via a pseudo tty.
5061
5062 * systemd-nspawn gained a new switch to control the UNIX
5063 signal to use when killing the init process of the container
5064 when shutting down.
5065
5066 * systemd-nspawn gained a new --overlay= switch for mounting
5067 overlay file systems into the container using the new kernel
5068 overlayfs support.
5069
5070 * When a container image is imported via systemd-importd and
5071 the host file system is not btrfs, a loopback block device
5072 file is created in /var/lib/machines.raw with a btrfs file
5073 system inside. It is then mounted to /var/lib/machines to
5074 enable btrfs features for container management. The loopback
5075 file and btrfs file system is grown as needed when container
5076 images are imported via systemd-importd.
5077
5078 * systemd-machined/systemd-importd gained support for btrfs
5079 quota, to enforce container disk space limits on disk. This
5080 is exposed in "machinectl set-limit".
5081
5082 * systemd-importd now can import containers from local .tar,
5083 .raw and .qcow2 images, and export them to .tar and .raw. It
5084 can also import dkr v2 images now from the network (on top
5085 of v1 as before).
5086
5087 * systemd-importd gained support for verifying downloaded
5088 images with gpg2 (previously only gpg1 was supported).
5089
5090 * systemd-machined, systemd-logind, systemd: most bus calls are
5091 now accessible to unprivileged processes via polkit. Also,
5092 systemd-logind will now allow users to kill their own sessions
5093 without further privileges or authorization.
5094
5095 * systemd-shutdownd has been removed. This service was
5096 previously responsible for implementing scheduled shutdowns
5097 as exposed in /usr/bin/shutdown's time parameter. This
5098 functionality has now been moved into systemd-logind and is
5099 accessible via a bus interface.
5100
5101 * "systemctl reboot" gained a new switch --firmware-setup that
5102 can be used to reboot into the EFI firmware setup, if that
5103 is available. systemd-logind now exposes an API on the bus
5104 to trigger such reboots, in case graphical desktop UIs want
5105 to cover this functionality.
5106
5107 * "systemctl enable", "systemctl disable" and "systemctl mask"
5108 now support a new "--now" switch. If specified the units
5109 that are enabled will also be started, and the ones
5110 disabled/masked also stopped.
5111
5112 * The Gummiboot EFI boot loader tool has been merged into
5113 systemd, and renamed to "systemd-boot". The bootctl tool has been
5114 updated to support systemd-boot.
5115
5116 * An EFI kernel stub has been added that may be used to create
5117 kernel EFI binaries that contain not only the actual kernel,
5118 but also an initrd, boot splash, command line and OS release
5119 information. This combined binary can then be signed as a
5120 single image, so that the firmware can verify it all in one
5121 step. systemd-boot has special support for EFI binaries created
5122 like this and can extract OS release information from them
5123 and show them in the boot menu. This functionality is useful
5124 to implement cryptographically verified boot schemes.
5125
5126 * Optional support has been added to systemd-fsck to pass
5127 fsck's progress report to an AF_UNIX socket in the file
5128 system.
5129
5130 * udev will no longer create device symlinks for all block
5131 devices by default. A blacklist for excluding special block
5132 devices from this logic has been turned into a whitelist
5133 that requires picking block devices explicitly that require
5134 device symlinks.
5135
5136 * A new (currently still internal) API sd-device.h has been
5137 added to libsystemd. This modernized API is supposed to
5138 replace libudev eventually. In fact, already much of libudev
5139 is now just a wrapper around sd-device.h.
5140
5141 * A new hwdb database for storing metadata about pointing
5142 stick devices has been added.
5143
5144 * systemd-tmpfiles gained support for setting file attributes
5145 similar to the "chattr" tool with new 'h' and 'H' lines.
5146
5147 * systemd-journald will no longer unconditionally set the
5148 btrfs NOCOW flag on new journal files. This is instead done
5149 with tmpfiles snippet using the new 'h' line type. This
5150 allows easy disabling of this logic, by masking the
5151 journal-nocow.conf tmpfiles file.
5152
5153 * systemd-journald will now translate audit message types to
5154 human readable identifiers when writing them to the
5155 journal. This should improve readability of audit messages.
5156
5157 * The LUKS logic gained support for the offset= and skip=
5158 options in /etc/crypttab, as previously implemented by
5159 Debian.
5160
5161 * /usr/lib/os-release gained a new optional field VARIANT= for
5162 distributions that support multiple variants (such as a
5163 desktop edition, a server edition, ...)
5164
5165 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
5166 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
5167 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
5168 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
5169 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
5170 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
5171 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
5172 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
5173 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
5174 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
5175 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
5176 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
5177 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
5178 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
5179 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
5180 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
5181 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
5182 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
5183 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
5184 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
5185 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
5186 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
5187 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
5188 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
5189 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
5190 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
5191 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
5192
5193 — Berlin, 2015-05-22
5194
5195 CHANGES WITH 219:
5196
5197 * Introduce a new API "sd-hwdb.h" for querying the hardware
5198 metadata database. With this minimal interface one can query
5199 and enumerate the udev hwdb, decoupled from the old libudev
5200 library. libudev's interface for this is now only a wrapper
5201 around sd-hwdb. A new tool systemd-hwdb has been added to
5202 interface with and update the database.
5203
5204 * When any of systemd's tools copies files (for example due to
5205 tmpfiles' C lines) a btrfs reflink will attempted first,
5206 before bytewise copying is done.
5207
5208 * systemd-nspawn gained a new --ephemeral switch. When
5209 specified a btrfs snapshot is taken of the container's root
5210 directory, and immediately removed when the container
5211 terminates again. Thus, a container can be started whose
5212 changes never alter the container's root directory, and are
5213 lost on container termination. This switch can also be used
5214 for starting a container off the root file system of the
5215 host without affecting the host OS. This switch is only
5216 available on btrfs file systems.
5217
5218 * systemd-nspawn gained a new --template= switch. It takes the
5219 path to a container tree to use as template for the tree
5220 specified via --directory=, should that directory be
5221 missing. This allows instantiating containers dynamically,
5222 on first run. This switch is only available on btrfs file
5223 systems.
5224
5225 * When a .mount unit refers to a mount point on which multiple
5226 mounts are stacked, and the .mount unit is stopped all of
5227 the stacked mount points will now be unmounted until no
5228 mount point remains.
5229
5230 * systemd now has an explicit notion of supported and
5231 unsupported unit types. Jobs enqueued for unsupported unit
5232 types will now fail with an "unsupported" error code. More
5233 specifically .swap, .automount and .device units are not
5234 supported in containers, .busname units are not supported on
5235 non-kdbus systems. .swap and .automount are also not
5236 supported if their respective kernel compile time options
5237 are disabled.
5238
5239 * machinectl gained support for two new "copy-from" and
5240 "copy-to" commands for copying files from a running
5241 container to the host or vice versa.
5242
5243 * machinectl gained support for a new "bind" command to bind
5244 mount host directories into local containers. This is
5245 currently only supported for nspawn containers.
5246
5247 * networkd gained support for configuring bridge forwarding
5248 database entries (fdb) from .network files.
5249
5250 * A new tiny daemon "systemd-importd" has been added that can
5251 download container images in tar, raw, qcow2 or dkr formats,
5252 and make them available locally in /var/lib/machines, so
5253 that they can run as nspawn containers. The daemon can GPG
5254 verify the downloads (not supported for dkr, since it has no
5255 provisions for verifying downloads). It will transparently
5256 decompress bz2, xz, gzip compressed downloads if necessary,
5257 and restore sparse files on disk. The daemon uses privilege
5258 separation to ensure the actual download logic runs with
5259 fewer privileges than the daemon itself. machinectl has
5260 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
5261 make the functionality of importd available to the
5262 user. With this in place the Fedora and Ubuntu "Cloud"
5263 images can be downloaded and booted as containers unmodified
5264 (the Fedora images lack the appropriate GPG signature files
5265 currently, so they cannot be verified, but this will change
5266 soon, hopefully). Note that downloading images is currently
5267 only fully supported on btrfs.
5268
5269 * machinectl is now able to list container images found in
5270 /var/lib/machines, along with some metadata about sizes of
5271 disk and similar. If the directory is located on btrfs and
5272 quota is enabled, this includes quota display. A new command
5273 "image-status" has been added that shows additional
5274 information about images.
5275
5276 * machinectl is now able to clone container images
5277 efficiently, if the underlying file system (btrfs) supports
5278 it, with the new "machinectl clone" command. It also
5279 gained commands for renaming and removing images, as well as
5280 marking them read-only or read-write (supported also on
5281 legacy file systems).
5282
5283 * networkd gained support for collecting LLDP network
5284 announcements, from hardware that supports this. This is
5285 shown in networkctl output.
5286
5287 * systemd-run gained support for a new -t (--pty) switch for
5288 invoking a binary on a pty whose input and output is
5289 connected to the invoking terminal. This allows executing
5290 processes as system services while interactively
5291 communicating with them via the terminal. Most interestingly
5292 this is supported across container boundaries. Invoking
5293 "systemd-run -t /bin/bash" is an alternative to running a
5294 full login session, the difference being that the former
5295 will not register a session, nor go through the PAM session
5296 setup.
5297
5298 * tmpfiles gained support for a new "v" line type for creating
5299 btrfs subvolumes. If the underlying file system is a legacy
5300 file system, this automatically degrades to creating a
5301 normal directory. Among others /var/lib/machines is now
5302 created like this at boot, should it be missing.
5303
5304 * The directory /var/lib/containers/ has been deprecated and
5305 been replaced by /var/lib/machines. The term "machines" has
5306 been used in the systemd context as generic term for both
5307 VMs and containers, and hence appears more appropriate for
5308 this, as the directory can also contain raw images bootable
5309 via qemu/kvm.
5310
5311 * systemd-nspawn when invoked with -M but without --directory=
5312 or --image= is now capable of searching for the container
5313 root directory, subvolume or disk image automatically, in
5314 /var/lib/machines. systemd-nspawn@.service has been updated
5315 to make use of this, thus allowing it to be used for raw
5316 disk images, too.
5317
5318 * A new machines.target unit has been introduced that is
5319 supposed to group all containers/VMs invoked as services on
5320 the system. systemd-nspawn@.service has been updated to
5321 integrate with that.
5322
5323 * machinectl gained a new "start" command, for invoking a
5324 container as a service. "machinectl start foo" is mostly
5325 equivalent to "systemctl start systemd-nspawn@foo.service",
5326 but handles escaping in a nicer way.
5327
5328 * systemd-nspawn will now mount most of the cgroupfs tree
5329 read-only into each container, with the exception of the
5330 container's own subtree in the name=systemd hierarchy.
5331
5332 * journald now sets the special FS_NOCOW file flag for its
5333 journal files. This should improve performance on btrfs, by
5334 avoiding heavy fragmentation when journald's write-pattern
5335 is used on COW file systems. It degrades btrfs' data
5336 integrity guarantees for the files to the same levels as for
5337 ext3/ext4 however. This should be OK though as journald does
5338 its own data integrity checks and all its objects are
5339 checksummed on disk. Also, journald should handle btrfs disk
5340 full events a lot more gracefully now, by processing SIGBUS
5341 errors, and not relying on fallocate() anymore.
5342
5343 * When journald detects that journal files it is writing to
5344 have been deleted it will immediately start new journal
5345 files.
5346
5347 * systemd now provides a way to store file descriptors
5348 per-service in PID 1. This is useful for daemons to ensure
5349 that fds they require are not lost during a daemon
5350 restart. The fds are passed to the daemon on the next
5351 invocation in the same way socket activation fds are
5352 passed. This is now used by journald to ensure that the
5353 various sockets connected to all the system's stdout/stderr
5354 are not lost when journald is restarted. File descriptors
5355 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
5356 an extension to sd_notify(). Note that a limit is enforced
5357 on the number of fds a service can store in PID 1, and it
5358 defaults to 0, so that no fds may be stored, unless this is
5359 explicitly turned on.
5360
5361 * The default TERM variable to use for units connected to a
5362 terminal, when no other value is explicitly is set is now
5363 vt220 rather than vt102. This should be fairly safe still,
5364 but allows PgUp/PgDn work.
5365
5366 * The /etc/crypttab option header= as known from Debian is now
5367 supported.
5368
5369 * "loginctl user-status" and "loginctl session-status" will
5370 now show the last 10 lines of log messages of the
5371 user/session following the status output. Similar,
5372 "machinectl status" will show the last 10 log lines
5373 associated with a virtual machine or container
5374 service. (Note that this is usually not the log messages
5375 done in the VM/container itself, but simply what the
5376 container manager logs. For nspawn this includes all console
5377 output however.)
5378
5379 * "loginctl session-status" without further argument will now
5380 show the status of the session of the caller. Similar,
5381 "lock-session", "unlock-session", "activate",
5382 "enable-linger", "disable-linger" may now be called without
5383 session/user parameter in which case they apply to the
5384 caller's session/user.
5385
5386 * An X11 session scriptlet is now shipped that uploads
5387 $DISPLAY and $XAUTHORITY into the environment of the systemd
5388 --user daemon if a session begins. This should improve
5389 compatibility with X11 enabled applications run as systemd
5390 user services.
5391
5392 * Generators are now subject to masking via /etc and /run, the
5393 same way as unit files.
5394
5395 * networkd .network files gained support for configuring
5396 per-link IPv4/IPv6 packet forwarding as well as IPv4
5397 masquerading. This is by default turned on for veth links to
5398 containers, as registered by systemd-nspawn. This means that
5399 nspawn containers run with --network-veth will now get
5400 automatic routed access to the host's networks without any
5401 further configuration or setup, as long as networkd runs on
5402 the host.
5403
5404 * systemd-nspawn gained the --port= (-p) switch to expose TCP
5405 or UDP posts of a container on the host. With this in place
5406 it is possible to run containers with private veth links
5407 (--network-veth), and have their functionality exposed on
5408 the host as if their services were running directly on the
5409 host.
5410
5411 * systemd-nspawn's --network-veth switch now gained a short
5412 version "-n", since with the changes above it is now truly
5413 useful out-of-the-box. The systemd-nspawn@.service has been
5414 updated to make use of it too by default.
5415
5416 * systemd-nspawn will now maintain a per-image R/W lock, to
5417 ensure that the same image is not started more than once
5418 writable. (It's OK to run an image multiple times
5419 simultaneously in read-only mode.)
5420
5421 * systemd-nspawn's --image= option is now capable of
5422 dissecting and booting MBR and GPT disk images that contain
5423 only a single active Linux partition. Previously it
5424 supported only GPT disk images with proper GPT type
5425 IDs. This allows running cloud images from major
5426 distributions directly with systemd-nspawn, without
5427 modification.
5428
5429 * In addition to collecting mouse dpi data in the udev
5430 hardware database, there's now support for collecting angle
5431 information for mouse scroll wheels. The database is
5432 supposed to guarantee similar scrolling behavior on mice
5433 that it knows about. There's also support for collecting
5434 information about Touchpad types.
5435
5436 * udev's input_id built-in will now also collect touch screen
5437 dimension data and attach it to probed devices.
5438
5439 * /etc/os-release gained support for a Distribution Privacy
5440 Policy link field.
5441
5442 * networkd gained support for creating "ipvlan", "gretap",
5443 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
5444
5445 * systemd-tmpfiles gained support for "a" lines for setting
5446 ACLs on files.
5447
5448 * systemd-nspawn will now mount /tmp in the container to
5449 tmpfs, automatically.
5450
5451 * systemd now exposes the memory.usage_in_bytes cgroup
5452 attribute and shows it for each service in the "systemctl
5453 status" output, if available.
5454
5455 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
5456 immediate reboot is triggered. This useful if shutdown is
5457 hung and is unable to complete, to expedite the
5458 operation. Note that this kind of reboot will still unmount
5459 all file systems, and hence should not result in fsck being
5460 run on next reboot.
5461
5462 * A .device unit for an optical block device will now be
5463 considered active only when a medium is in the drive. Also,
5464 mount units are now bound to their backing devices thus
5465 triggering automatic unmounting when devices become
5466 unavailable. With this in place systemd will now
5467 automatically unmount left-over mounts when a CD-ROM is
5468 ejected or an USB stick is yanked from the system.
5469
5470 * networkd-wait-online now has support for waiting for
5471 specific interfaces only (with globbing), and for giving up
5472 after a configurable timeout.
5473
5474 * networkd now exits when idle. It will be automatically
5475 restarted as soon as interfaces show up, are removed or
5476 change state. networkd will stay around as long as there is
5477 at least one DHCP state machine or similar around, that keep
5478 it non-idle.
5479
5480 * networkd may now configure IPv6 link-local addressing in
5481 addition to IPv4 link-local addressing.
5482
5483 * The IPv6 "token" for use in SLAAC may now be configured for
5484 each .network interface in networkd.
5485
5486 * Routes configured with networkd may now be assigned a scope
5487 in .network files.
5488
5489 * networkd's [Match] sections now support globbing and lists
5490 of multiple space-separated matches per item.
5491
5492 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
5493 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
5494 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
5495 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
5496 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
5497 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
5498 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
5499 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
5500 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
5501 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
5502 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
5503 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
5504 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
5505 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
5506 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
5507 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
5508 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
5509 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
5510 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
5511 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
5512 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
5513 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
5514 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
5515 Hoffmann, Zbigniew Jędrzejewski-Szmek
5516
5517 — Berlin, 2015-02-16
5518
5519 CHANGES WITH 218:
5520
5521 * When querying unit file enablement status (for example via
5522 "systemctl is-enabled"), a new state "indirect" is now known
5523 which indicates that a unit might not be enabled itself, but
5524 another unit listed in its Also= setting might be.
5525
5526 * Similar to the various existing ConditionXYZ= settings for
5527 units, there are now matching AssertXYZ= settings. While
5528 failing conditions cause a unit to be skipped, but its job
5529 to succeed, failing assertions declared like this will cause
5530 a unit start operation and its job to fail.
5531
5532 * hostnamed now knows a new chassis type "embedded".
5533
5534 * systemctl gained a new "edit" command. When used on a unit
5535 file, this allows extending unit files with .d/ drop-in
5536 configuration snippets or editing the full file (after
5537 copying it from /usr/lib to /etc). This will invoke the
5538 user's editor (as configured with $EDITOR), and reload the
5539 modified configuration after editing.
5540
5541 * "systemctl status" now shows the suggested enablement state
5542 for a unit, as declared in the (usually vendor-supplied)
5543 system preset files.
5544
5545 * nss-myhostname will now resolve the single-label host name
5546 "gateway" to the locally configured default IP routing
5547 gateways, ordered by their metrics. This assigns a stable
5548 name to the used gateways, regardless which ones are
5549 currently configured. Note that the name will only be
5550 resolved after all other name sources (if nss-myhostname is
5551 configured properly) and should hence not negatively impact
5552 systems that use the single-label host name "gateway" in
5553 other contexts.
5554
5555 * systemd-inhibit now allows filtering by mode when listing
5556 inhibitors.
5557
5558 * Scope and service units gained a new "Delegate" boolean
5559 property, which, when set, allows processes running inside the
5560 unit to further partition resources. This is primarily
5561 useful for systemd user instances as well as container
5562 managers.
5563
5564 * journald will now pick up audit messages directly from
5565 the kernel, and log them like any other log message. The
5566 audit fields are split up and fully indexed. This means that
5567 journalctl in many ways is now a (nicer!) alternative to
5568 ausearch, the traditional audit client. Note that this
5569 implements only a minimal audit client. If you want the
5570 special audit modes like reboot-on-log-overflow, please use
5571 the traditional auditd instead, which can be used in
5572 parallel to journald.
5573
5574 * The ConditionSecurity= unit file option now understands the
5575 special string "audit" to check whether auditing is
5576 available.
5577
5578 * journalctl gained two new commands --vacuum-size= and
5579 --vacuum-time= to delete old journal files until the
5580 remaining ones take up no more than the specified size on disk,
5581 or are not older than the specified time.
5582
5583 * A new, native PPPoE library has been added to sd-network,
5584 systemd's library of light-weight networking protocols. This
5585 library will be used in a future version of networkd to
5586 enable PPPoE communication without an external pppd daemon.
5587
5588 * The busctl tool now understands a new "capture" verb that
5589 works similar to "monitor", but writes a packet capture
5590 trace to STDOUT that can be redirected to a file which is
5591 compatible with libcap's capture file format. This can then
5592 be loaded in Wireshark and similar tools to inspect bus
5593 communication.
5594
5595 * The busctl tool now understands a new "tree" verb that shows
5596 the object trees of a specific service on the bus, or of all
5597 services.
5598
5599 * The busctl tool now understands a new "introspect" verb that
5600 shows all interfaces and members of objects on the bus,
5601 including their signature and values. This is particularly
5602 useful to get more information about bus objects shown by
5603 the new "busctl tree" command.
5604
5605 * The busctl tool now understands new verbs "call",
5606 "set-property" and "get-property" for invoking bus method
5607 calls, setting and getting bus object properties in a
5608 friendly way.
5609
5610 * busctl gained a new --augment-creds= argument that controls
5611 whether the tool shall augment credential information it
5612 gets from the bus with data from /proc, in a possibly
5613 race-ful way.
5614
5615 * nspawn's --link-journal= switch gained two new values
5616 "try-guest" and "try-host" that work like "guest" and
5617 "host", but do not fail if the host has no persistent
5618 journalling enabled. -j is now equivalent to
5619 --link-journal=try-guest.
5620
5621 * macvlan network devices created by nspawn will now have
5622 stable MAC addresses.
5623
5624 * A new SmackProcessLabel= unit setting has been added, which
5625 controls the SMACK security label processes forked off by
5626 the respective unit shall use.
5627
5628 * If compiled with --enable-xkbcommon, systemd-localed will
5629 verify x11 keymap settings by compiling the given keymap. It
5630 will spew out warnings if the compilation fails. This
5631 requires libxkbcommon to be installed.
5632
5633 * When a coredump is collected, a larger number of metadata
5634 fields is now collected and included in the journal records
5635 created for it. More specifically, control group membership,
5636 environment variables, memory maps, working directory,
5637 chroot directory, /proc/$PID/status, and a list of open file
5638 descriptors is now stored in the log entry.
5639
5640 * The udev hwdb now contains DPI information for mice. For
5641 details see:
5642
5643 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
5644
5645 * All systemd programs that read standalone configuration
5646 files in /etc now also support a corresponding series of
5647 .conf.d configuration directories in /etc/, /run/,
5648 /usr/local/lib/, /usr/lib/, and (if configured with
5649 --enable-split-usr) /lib/. In particular, the following
5650 configuration files now have corresponding configuration
5651 directories: system.conf user.conf, logind.conf,
5652 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
5653 resolved.conf, timesyncd.conf, journal-remote.conf, and
5654 journal-upload.conf. Note that distributions should use the
5655 configuration directories in /usr/lib/; the directories in
5656 /etc/ are reserved for the system administrator.
5657
5658 * systemd-rfkill will no longer take the rfkill device name
5659 into account when storing rfkill state on disk, as the name
5660 might be dynamically assigned and not stable. Instead, the
5661 ID_PATH udev variable combined with the rfkill type (wlan,
5662 bluetooth, ...) is used.
5663
5664 * A new service systemd-machine-id-commit.service has been
5665 added. When used on systems where /etc is read-only during
5666 boot, and /etc/machine-id is not initialized (but an empty
5667 file), this service will copy the temporary machine ID
5668 created as replacement into /etc after the system is fully
5669 booted up. This is useful for systems that are freshly
5670 installed with a non-initialized machine ID, but should get
5671 a fixed machine ID for subsequent boots.
5672
5673 * networkd's .netdev files now provide a large set of
5674 configuration parameters for VXLAN devices. Similarly, the
5675 bridge port cost parameter is now configurable in .network
5676 files. There's also new support for configuring IP source
5677 routing. networkd .link files gained support for a new
5678 OriginalName= match that is useful to match against the
5679 original interface name the kernel assigned. .network files
5680 may include MTU= and MACAddress= fields for altering the MTU
5681 and MAC address while being connected to a specific network
5682 interface.
5683
5684 * The LUKS logic gained supported for configuring
5685 UUID-specific key files. There's also new support for naming
5686 LUKS device from the kernel command line, using the new
5687 luks.name= argument.
5688
5689 * Timer units may now be transiently created via the bus API
5690 (this was previously already available for scope and service
5691 units). In addition it is now possible to create multiple
5692 transient units at the same time with a single bus call. The
5693 "systemd-run" tool has been updated to make use of this for
5694 running commands on a specified time, in at(1)-style.
5695
5696 * tmpfiles gained support for "t" lines, for assigning
5697 extended attributes to files. Among other uses this may be
5698 used to assign SMACK labels to files.
5699
5700 Contributions from: Alin Rauta, Alison Chaiken, Andrej
5701 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
5702 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
5703 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
5704 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
5705 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
5706 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
5707 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
5708 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
5709 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
5710 Michael Biebl, Michael Chapman, Michael Marineau, Michal
5711 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
5712 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
5713 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
5714 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
5715 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5716 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
5717 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
5718
5719 — Berlin, 2014-12-10
5720
5721 CHANGES WITH 217:
5722
5723 * journalctl gained the new options -t/--identifier= to match
5724 on the syslog identifier (aka "tag"), as well as --utc to
5725 show log timestamps in the UTC timezone. journalctl now also
5726 accepts -n/--lines=all to disable line capping in a pager.
5727
5728 * journalctl gained a new switch, --flush, that synchronously
5729 flushes logs from /run/log/journal to /var/log/journal if
5730 persistent storage is enabled. systemd-journal-flush.service
5731 now waits until the operation is complete.
5732
5733 * Services can notify the manager before they start a reload
5734 (by sending RELOADING=1) or shutdown (by sending
5735 STOPPING=1). This allows the manager to track and show the
5736 internal state of daemons and closes a race condition when
5737 the process is still running but has closed its D-Bus
5738 connection.
5739
5740 * Services with Type=oneshot do not have to have any ExecStart
5741 commands anymore.
5742
5743 * User units are now loaded also from
5744 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
5745 /run/systemd/user directory that was already previously
5746 supported, but is under the control of the user.
5747
5748 * Job timeouts (i.e. timeouts on the time a job that is
5749 queued stays in the run queue) can now optionally result in
5750 immediate reboot or power-off actions (JobTimeoutAction= and
5751 JobTimeoutRebootArgument=). This is useful on ".target"
5752 units, to limit the maximum time a target remains
5753 undispatched in the run queue, and to trigger an emergency
5754 operation in such a case. This is now used by default to
5755 turn off the system if boot-up (as defined by everything in
5756 basic.target) hangs and does not complete for at least
5757 15min. Also, if power-off or reboot hang for at least 30min
5758 an immediate power-off/reboot operation is triggered. This
5759 functionality is particularly useful to increase reliability
5760 on embedded devices, but also on laptops which might
5761 accidentally get powered on when carried in a backpack and
5762 whose boot stays stuck in a hard disk encryption passphrase
5763 question.
5764
5765 * systemd-logind can be configured to also handle lid switch
5766 events even when the machine is docked or multiple displays
5767 are attached (HandleLidSwitchDocked= option).
5768
5769 * A helper binary and a service have been added which can be
5770 used to resume from hibernation in the initramfs. A
5771 generator will parse the resume= option on the kernel
5772 command line to trigger resume.
5773
5774 * A user console daemon systemd-consoled has been
5775 added. Currently, it is a preview, and will so far open a
5776 single terminal on each session of the user marked as
5777 Desktop=systemd-console.
5778
5779 * Route metrics can be specified for DHCP routes added by
5780 systemd-networkd.
5781
5782 * The SELinux context of socket-activated services can be set
5783 from the information provided by the networking stack
5784 (SELinuxContextFromNet= option).
5785
5786 * Userspace firmware loading support has been removed and
5787 the minimum supported kernel version is thus bumped to 3.7.
5788
5789 * Timeout for udev workers has been increased from 1 to 3
5790 minutes, but a warning will be printed after 1 minute to
5791 help diagnose kernel modules that take a long time to load.
5792
5793 * Udev rules can now remove tags on devices with TAG-="foobar".
5794
5795 * systemd's readahead implementation has been removed. In many
5796 circumstances it didn't give expected benefits even for
5797 rotational disk drives and was becoming less relevant in the
5798 age of SSDs. As none of the developers has been using
5799 rotating media anymore, and nobody stepped up to actively
5800 maintain this component of systemd it has now been removed.
5801
5802 * Swap units can use Options= to specify discard options.
5803 Discard options specified for swaps in /etc/fstab are now
5804 respected.
5805
5806 * Docker containers are now detected as a separate type of
5807 virtualization.
5808
5809 * The Password Agent protocol gained support for queries where
5810 the user input is shown, useful e.g. for user names.
5811 systemd-ask-password gained a new --echo option to turn that
5812 on.
5813
5814 * The default sysctl.d/ snippets will now set:
5815
5816 net.core.default_qdisc = fq_codel
5817
5818 This selects Fair Queuing Controlled Delay as the default
5819 queuing discipline for network interfaces. fq_codel helps
5820 fight the network bufferbloat problem. It is believed to be
5821 a good default with no tuning required for most workloads.
5822 Downstream distributions may override this choice. On 10Gbit
5823 servers that do not do forwarding, "fq" may perform better.
5824 Systems without a good clocksource should use "pfifo_fast".
5825
5826 * If kdbus is enabled during build a new option BusPolicy= is
5827 available for service units, that allows locking all service
5828 processes into a stricter bus policy, in order to limit
5829 access to various bus services, or even hide most of them
5830 from the service's view entirely.
5831
5832 * networkctl will now show the .network and .link file
5833 networkd has applied to a specific interface.
5834
5835 * sd-login gained a new API call sd_session_get_desktop() to
5836 query which desktop environment has been selected for a
5837 session.
5838
5839 * UNIX utmp support is now compile-time optional to support
5840 legacy-free systems.
5841
5842 * systemctl gained two new commands "add-wants" and
5843 "add-requires" for pulling in units from specific targets
5844 easily.
5845
5846 * If the word "rescue" is specified on the kernel command line
5847 the system will now boot into rescue mode (aka
5848 rescue.target), which was previously available only by
5849 specifying "1" or "systemd.unit=rescue.target" on the kernel
5850 command line. This new kernel command line option nicely
5851 mirrors the already existing "emergency" kernel command line
5852 option.
5853
5854 * New kernel command line options mount.usr=, mount.usrflags=,
5855 mount.usrfstype= have been added that match root=, rootflags=,
5856 rootfstype= but allow mounting a specific file system to
5857 /usr.
5858
5859 * The $NOTIFY_SOCKET is now also passed to control processes of
5860 services, not only the main process.
5861
5862 * This version reenables support for fsck's -l switch. This
5863 means at least version v2.25 of util-linux is required for
5864 operation, otherwise dead-locks on device nodes may
5865 occur. Again: you need to update util-linux to at least
5866 v2.25 when updating systemd to v217.
5867
5868 * The "multi-seat-x" tool has been removed from systemd, as
5869 its functionality has been integrated into X servers 1.16,
5870 and the tool is hence redundant. It is recommended to update
5871 display managers invoking this tool to simply invoke X
5872 directly from now on, again.
5873
5874 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
5875 message flag has been added for all of systemd's polkit
5876 authenticated method calls has been added. In particular this
5877 now allows optional interactive authorization via polkit for
5878 many of PID1's privileged operations such as unit file
5879 enabling and disabling.
5880
5881 * "udevadm hwdb --update" learnt a new switch "--usr" for
5882 placing the rebuilt hardware database in /usr instead of
5883 /etc. When used only hardware database entries stored in
5884 /usr will be used, and any user database entries in /etc are
5885 ignored. This functionality is useful for vendors to ship a
5886 pre-built database on systems where local configuration is
5887 unnecessary or unlikely.
5888
5889 * Calendar time specifications in .timer units now also
5890 understand the strings "semi-annually", "quarterly" and
5891 "minutely" as shortcuts (in addition to the preexisting
5892 "anually", "hourly", ...).
5893
5894 * systemd-tmpfiles will now correctly create files in /dev
5895 at boot which are marked for creation only at boot. It is
5896 recommended to always create static device nodes with 'c!'
5897 and 'b!', so that they are created only at boot and not
5898 overwritten at runtime.
5899
5900 * When the watchdog logic is used for a service (WatchdogSec=)
5901 and the watchdog timeout is hit the service will now be
5902 terminated with SIGABRT (instead of just SIGTERM), in order
5903 to make sure a proper coredump and backtrace is
5904 generated. This ensures that hanging services will result in
5905 similar coredump/backtrace behaviour as services that hit a
5906 segmentation fault.
5907
5908 Contributions from: Andreas Henriksson, Andrei Borzenkov,
5909 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
5910 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
5911 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
5912 Herrmann, David Sommerseth, David Strauss, Emil Renner
5913 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
5914 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
5915 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
5916 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
5917 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
5918 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
5919 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
5920 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
5921 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
5922 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
5923 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
5924 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
5925 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
5926 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
5927 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
5928 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
5929 Jędrzejewski-Szmek
5930
5931 — Berlin, 2014-10-28
5932
5933 CHANGES WITH 216:
5934
5935 * timedated no longer reads NTP implementation unit names from
5936 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
5937 implementations should add a
5938
5939 Conflicts=systemd-timesyncd.service
5940
5941 to their unit files to take over and replace systemd's NTP
5942 default functionality.
5943
5944 * systemd-sysusers gained a new line type "r" for configuring
5945 which UID/GID ranges to allocate system users/groups
5946 from. Lines of type "u" may now add an additional column
5947 that specifies the home directory for the system user to be
5948 created. Also, systemd-sysusers may now optionally read user
5949 information from STDIN instead of a file. This is useful for
5950 invoking it from RPM preinst scriptlets that need to create
5951 users before the first RPM file is installed since these
5952 files might need to be owned by them. A new
5953 %sysusers_create_inline RPM macro has been introduced to do
5954 just that. systemd-sysusers now updates the shadow files as
5955 well as the user/group databases, which should enhance
5956 compatibility with certain tools like grpck.
5957
5958 * A number of bus APIs of PID 1 now optionally consult polkit to
5959 permit access for otherwise unprivileged clients under certain
5960 conditions. Note that this currently doesn't support
5961 interactive authentication yet, but this is expected to be
5962 added eventually, too.
5963
5964 * /etc/machine-info now has new fields for configuring the
5965 deployment environment of the machine, as well as the
5966 location of the machine. hostnamectl has been updated with
5967 new command to update these fields.
5968
5969 * systemd-timesyncd has been updated to automatically acquire
5970 NTP server information from systemd-networkd, which might
5971 have been discovered via DHCP.
5972
5973 * systemd-resolved now includes a caching DNS stub resolver
5974 and a complete LLMNR name resolution implementation. A new
5975 NSS module "nss-resolve" has been added which can be used
5976 instead of glibc's own "nss-dns" to resolve hostnames via
5977 systemd-resolved. Hostnames, addresses and arbitrary RRs may
5978 be resolved via systemd-resolved D-Bus APIs. In contrast to
5979 the glibc internal resolver systemd-resolved is aware of
5980 multi-homed system, and keeps DNS server and caches separate
5981 and per-interface. Queries are sent simultaneously on all
5982 interfaces that have DNS servers configured, in order to
5983 properly handle VPNs and local LANs which might resolve
5984 separate sets of domain names. systemd-resolved may acquire
5985 DNS server information from systemd-networkd automatically,
5986 which in turn might have discovered them via DHCP. A tool
5987 "systemd-resolve-host" has been added that may be used to
5988 query the DNS logic in resolved. systemd-resolved implements
5989 IDNA and automatically uses IDNA or UTF-8 encoding depending
5990 on whether classic DNS or LLMNR is used as transport. In the
5991 next releases we intend to add a DNSSEC and mDNS/DNS-SD
5992 implementation to systemd-resolved.
5993
5994 * A new NSS module nss-mymachines has been added, that
5995 automatically resolves the names of all local registered
5996 containers to their respective IP addresses.
5997
5998 * A new client tool "networkctl" for systemd-networkd has been
5999 added. It currently is entirely passive and will query
6000 networking configuration from udev, rtnetlink and networkd,
6001 and present it to the user in a very friendly
6002 way. Eventually, we hope to extend it to become a full
6003 control utility for networkd.
6004
6005 * .socket units gained a new DeferAcceptSec= setting that
6006 controls the kernels' TCP_DEFER_ACCEPT sockopt for
6007 TCP. Similarly, support for controlling TCP keep-alive
6008 settings has been added (KeepAliveTimeSec=,
6009 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
6010 turning off Nagle's algorithm on TCP has been added
6011 (NoDelay=).
6012
6013 * logind learned a new session type "web", for use in projects
6014 like Cockpit which register web clients as PAM sessions.
6015
6016 * timer units with at least one OnCalendar= setting will now
6017 be started only after time-sync.target has been
6018 reached. This way they will not elapse before the system
6019 clock has been corrected by a local NTP client or
6020 similar. This is particular useful on RTC-less embedded
6021 machines, that come up with an invalid system clock.
6022
6023 * systemd-nspawn's --network-veth= switch should now result in
6024 stable MAC addresses for both the outer and the inner side
6025 of the link.
6026
6027 * systemd-nspawn gained a new --volatile= switch for running
6028 container instances with /etc or /var unpopulated.
6029
6030 * The kdbus client code has been updated to use the new Linux
6031 3.17 memfd subsystem instead of the old kdbus-specific one.
6032
6033 * systemd-networkd's DHCP client and server now support
6034 FORCERENEW. There are also new configuration options to
6035 configure the vendor client identifier and broadcast mode
6036 for DHCP.
6037
6038 * systemd will no longer inform the kernel about the current
6039 timezone, as this is necessarily incorrect and racy as the
6040 kernel has no understanding of DST and similar
6041 concepts. This hence means FAT timestamps will be always
6042 considered UTC, similar to what Android is already
6043 doing. Also, when the RTC is configured to the local time
6044 (rather than UTC) systemd will never synchronize back to it,
6045 as this might confuse Windows at a later boot.
6046
6047 * systemd-analyze gained a new command "verify" for offline
6048 validation of unit files.
6049
6050 * systemd-networkd gained support for a couple of additional
6051 settings for bonding networking setups. Also, the metric for
6052 statically configured routes may now be configured. For
6053 network interfaces where this is appropriate the peer IP
6054 address may now be configured.
6055
6056 * systemd-networkd's DHCP client will no longer request
6057 broadcasting by default, as this tripped up some networks.
6058 For hardware where broadcast is required the feature should
6059 be switched back on using RequestBroadcast=yes.
6060
6061 * systemd-networkd will now set up IPv4LL addresses (when
6062 enabled) even if DHCP is configured successfully.
6063
6064 * udev will now default to respect network device names given
6065 by the kernel when the kernel indicates that these are
6066 predictable. This behavior can be tweaked by changing
6067 NamePolicy= in the relevant .link file.
6068
6069 * A new library systemd-terminal has been added that
6070 implements full TTY stream parsing and rendering. This
6071 library is supposed to be used later on for implementing a
6072 full userspace VT subsystem, replacing the current kernel
6073 implementation.
6074
6075 * A new tool systemd-journal-upload has been added to push
6076 journal data to a remote system running
6077 systemd-journal-remote.
6078
6079 * journald will no longer forward all local data to another
6080 running syslog daemon. This change has been made because
6081 rsyslog (which appears to be the most commonly used syslog
6082 implementation these days) no longer makes use of this, and
6083 instead pulls the data out of the journal on its own. Since
6084 forwarding the messages to a non-existent syslog server is
6085 more expensive than we assumed we have now turned this
6086 off. If you run a syslog server that is not a recent rsyslog
6087 version, you have to turn this option on again
6088 (ForwardToSyslog= in journald.conf).
6089
6090 * journald now optionally supports the LZ4 compressor for
6091 larger journal fields. This compressor should perform much
6092 better than XZ which was the previous default.
6093
6094 * machinectl now shows the IP addresses of local containers,
6095 if it knows them, plus the interface name of the container.
6096
6097 * A new tool "systemd-escape" has been added that makes it
6098 easy to escape strings to build unit names and similar.
6099
6100 * sd_notify() messages may now include a new ERRNO= field
6101 which is parsed and collected by systemd and shown among the
6102 "systemctl status" output for a service.
6103
6104 * A new component "systemd-firstboot" has been added that
6105 queries the most basic systemd information (timezone,
6106 hostname, root password) interactively on first
6107 boot. Alternatively it may also be used to provision these
6108 things offline on OS images installed into directories.
6109
6110 * The default sysctl.d/ snippets will now set
6111
6112 net.ipv4.conf.default.promote_secondaries=1
6113
6114 This has the benefit of no flushing secondary IP addresses
6115 when primary addresses are removed.
6116
6117 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
6118 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
6119 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
6120 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
6121 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
6122 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
6123 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
6124 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
6125 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
6126 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
6127 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
6128 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
6129 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
6130 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
6131 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
6132
6133 — Berlin, 2014-08-19
6134
6135 CHANGES WITH 215:
6136
6137 * A new tool systemd-sysusers has been added. This tool
6138 creates system users and groups in /etc/passwd and
6139 /etc/group, based on static declarative system user/group
6140 definitions in /usr/lib/sysusers.d/. This is useful to
6141 enable factory resets and volatile systems that boot up with
6142 an empty /etc directory, and thus need system users and
6143 groups created during early boot. systemd now also ships
6144 with two default sysusers.d/ files for the most basic
6145 users and groups systemd and the core operating system
6146 require.
6147
6148 * A new tmpfiles snippet has been added that rebuilds the
6149 essential files in /etc on boot, should they be missing.
6150
6151 * A directive for ensuring automatic clean-up of
6152 /var/cache/man/ has been removed from the default
6153 configuration. This line should now be shipped by the man
6154 implementation. The necessary change has been made to the
6155 man-db implementation. Note that you need to update your man
6156 implementation to one that ships this line, otherwise no
6157 automatic clean-up of /var/cache/man will take place.
6158
6159 * A new condition ConditionNeedsUpdate= has been added that
6160 may conditionalize services to only run when /etc or /var
6161 are "older" than the vendor operating system resources in
6162 /usr. This is useful for reconstructing or updating /etc
6163 after an offline update of /usr or a factory reset, on the
6164 next reboot. Services that want to run once after such an
6165 update or reset should use this condition and order
6166 themselves before the new systemd-update-done.service, which
6167 will mark the two directories as fully updated. A number of
6168 service files have been added making use of this, to rebuild
6169 the udev hardware database, the journald message catalog and
6170 dynamic loader cache (ldconfig). The systemd-sysusers tool
6171 described above also makes use of this now. With this in
6172 place it is now possible to start up a minimal operating
6173 system with /etc empty cleanly. For more information on the
6174 concepts involved see this recent blog story:
6175
6176 http://0pointer.de/blog/projects/stateless.html
6177
6178 * A new system group "input" has been introduced, and all
6179 input device nodes get this group assigned. This is useful
6180 for system-level software to get access to input devices. It
6181 complements what is already done for "audio" and "video".
6182
6183 * systemd-networkd learnt minimal DHCPv4 server support in
6184 addition to the existing DHCPv4 client support. It also
6185 learnt DHCPv6 client and IPv6 Router Solicitation client
6186 support. The DHCPv4 client gained support for static routes
6187 passed in from the server. Note that the [DHCPv4] section
6188 known in older systemd-networkd versions has been renamed to
6189 [DHCP] and is now also used by the DHCPv6 client. Existing
6190 .network files using settings of this section should be
6191 updated, though compatibility is maintained. Optionally, the
6192 client hostname may now be sent to the DHCP server.
6193
6194 * networkd gained support for vxlan virtual networks as well
6195 as tun/tap and dummy devices.
6196
6197 * networkd gained support for automatic allocation of address
6198 ranges for interfaces from a system-wide pool of
6199 addresses. This is useful for dynamically managing a large
6200 number of interfaces with a single network configuration
6201 file. In particular this is useful to easily assign
6202 appropriate IP addresses to the veth links of a large number
6203 of nspawn instances.
6204
6205 * RPM macros for processing sysusers, sysctl and binfmt
6206 drop-in snippets at package installation time have been
6207 added.
6208
6209 * The /etc/os-release file should now be placed in
6210 /usr/lib/os-release. The old location is automatically
6211 created as symlink. /usr/lib is the more appropriate
6212 location of this file, since it shall actually describe the
6213 vendor operating system shipped in /usr, and not the
6214 configuration stored in /etc.
6215
6216 * .mount units gained a new boolean SloppyOptions= setting
6217 that maps to mount(8)'s -s option which enables permissive
6218 parsing of unknown mount options.
6219
6220 * tmpfiles learnt a new "L+" directive which creates a symlink
6221 but (unlike "L") deletes a pre-existing file first, should
6222 it already exist and not already be the correct
6223 symlink. Similarly, "b+", "c+" and "p+" directives have been
6224 added as well, which create block and character devices, as
6225 well as fifos in the filesystem, possibly removing any
6226 pre-existing files of different types.
6227
6228 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
6229 'argument' field (which so far specified the source to
6230 symlink/copy the files from) is now optional. If omitted the
6231 same file os copied from /usr/share/factory/ suffixed by the
6232 full destination path. This is useful for populating /etc
6233 with essential files, by copying them from vendor defaults
6234 shipped in /usr/share/factory/etc.
6235
6236 * A new command "systemctl preset-all" has been added that
6237 applies the service preset settings to all installed unit
6238 files. A new switch --preset-mode= has been added that
6239 controls whether only enable or only disable operations
6240 shall be executed.
6241
6242 * A new command "systemctl is-system-running" has been added
6243 that allows checking the overall state of the system, for
6244 example whether it is fully up and running.
6245
6246 * When the system boots up with an empty /etc, the equivalent
6247 to "systemctl preset-all" is executed during early boot, to
6248 make sure all default services are enabled after a factory
6249 reset.
6250
6251 * systemd now contains a minimal preset file that enables the
6252 most basic services systemd ships by default.
6253
6254 * Unit files' [Install] section gained a new DefaultInstance=
6255 field for defining the default instance to create if a
6256 template unit is enabled with no instance specified.
6257
6258 * A new passive target cryptsetup-pre.target has been added
6259 that may be used by services that need to make they run and
6260 finish before the first LUKS cryptographic device is set up.
6261
6262 * The /dev/loop-control and /dev/btrfs-control device nodes
6263 are now owned by the "disk" group by default, opening up
6264 access to this group.
6265
6266 * systemd-coredump will now automatically generate a
6267 stack trace of all core dumps taking place on the system,
6268 based on elfutils' libdw library. This stack trace is logged
6269 to the journal.
6270
6271 * systemd-coredump may now optionally store coredumps directly
6272 on disk (in /var/lib/systemd/coredump, possibly compressed),
6273 instead of storing them unconditionally in the journal. This
6274 mode is the new default. A new configuration file
6275 /etc/systemd/coredump.conf has been added to configure this
6276 and other parameters of systemd-coredump.
6277
6278 * coredumpctl gained a new "info" verb to show details about a
6279 specific coredump. A new switch "-1" has also been added
6280 that makes sure to only show information about the most
6281 recent entry instead of all entries. Also, as the tool is
6282 generally useful now the "systemd-" prefix of the binary
6283 name has been removed. Distributions that want to maintain
6284 compatibility with the old name should add a symlink from
6285 the old name to the new name.
6286
6287 * journald's SplitMode= now defaults to "uid". This makes sure
6288 that unprivileged users can access their own coredumps with
6289 coredumpctl without restrictions.
6290
6291 * New kernel command line options "systemd.wants=" (for
6292 pulling an additional unit during boot), "systemd.mask="
6293 (for masking a specific unit for the boot), and
6294 "systemd.debug-shell" (for enabling the debug shell on tty9)
6295 have been added. This is implemented in the new generator
6296 "systemd-debug-generator".
6297
6298 * systemd-nspawn will now by default filter a couple of
6299 syscalls for containers, among them those required for
6300 kernel module loading, direct x86 IO port access, swap
6301 management, and kexec. Most importantly though
6302 open_by_handle_at() is now prohibited for containers,
6303 closing a hole similar to a recently discussed vulnerability
6304 in docker regarding access to files on file hierarchies the
6305 container should normally not have access to. Note that, for
6306 nspawn, we generally make no security claims anyway (and
6307 this is explicitly documented in the man page), so this is
6308 just a fix for one of the most obvious problems.
6309
6310 * A new man page file-hierarchy(7) has been added that
6311 contains a minimized, modernized version of the file system
6312 layout systemd expects, similar in style to the FHS
6313 specification or hier(5). A new tool systemd-path(1) has
6314 been added to query many of these paths for the local
6315 machine and user.
6316
6317 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
6318 longer done. Since the directory now has a per-user size
6319 limit, and is cleaned on logout this appears unnecessary,
6320 in particular since this now brings the lifecycle of this
6321 directory closer in line with how IPC objects are handled.
6322
6323 * systemd.pc now exports a number of additional directories,
6324 including $libdir (which is useful to identify the library
6325 path for the primary architecture of the system), and a
6326 couple of drop-in directories.
6327
6328 * udev's predictable network interface names now use the dev_port
6329 sysfs attribute, introduced in linux 3.15 instead of dev_id to
6330 distinguish between ports of the same PCI function. dev_id should
6331 only be used for ports using the same HW address, hence the need
6332 for dev_port.
6333
6334 * machined has been updated to export the OS version of a
6335 container (read from /etc/os-release and
6336 /usr/lib/os-release) on the bus. This is now shown in
6337 "machinectl status" for a machine.
6338
6339 * A new service setting RestartForceExitStatus= has been
6340 added. If configured to a set of exit signals or process
6341 return values, the service will be restarted when the main
6342 daemon process exits with any of them, regardless of the
6343 Restart= setting.
6344
6345 * systemctl's -H switch for connecting to remote systemd
6346 machines has been extended so that it may be used to
6347 directly connect to a specific container on the
6348 host. "systemctl -H root@foobar:waldi" will now connect as
6349 user "root" to host "foobar", and then proceed directly to
6350 the container named "waldi". Note that currently you have to
6351 authenticate as user "root" for this to work, as entering
6352 containers is a privileged operation.
6353
6354 Contributions from: Andreas Henriksson, Benjamin Steinwender,
6355 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
6356 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
6357 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
6358 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
6359 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
6360 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
6361 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
6362 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
6363 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
6364 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
6365 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
6366
6367 — Berlin, 2014-07-03
6368
6369 CHANGES WITH 214:
6370
6371 * As an experimental feature, udev now tries to lock the
6372 disk device node (flock(LOCK_SH|LOCK_NB)) while it
6373 executes events for the disk or any of its partitions.
6374 Applications like partitioning programs can lock the
6375 disk device node (flock(LOCK_EX)) and claim temporary
6376 device ownership that way; udev will entirely skip all event
6377 handling for this disk and its partitions. If the disk
6378 was opened for writing, the close will trigger a partition
6379 table rescan in udev's "watch" facility, and if needed
6380 synthesize "change" events for the disk and all its partitions.
6381 This is now unconditionally enabled, and if it turns out to
6382 cause major problems, we might turn it on only for specific
6383 devices, or might need to disable it entirely. Device Mapper
6384 devices are excluded from this logic.
6385
6386 * We temporarily dropped the "-l" switch for fsck invocations,
6387 since they collide with the flock() logic above. util-linux
6388 upstream has been changed already to avoid this conflict,
6389 and we will re-add "-l" as soon as util-linux with this
6390 change has been released.
6391
6392 * The dependency on libattr has been removed. Since a long
6393 time, the extended attribute calls have moved to glibc, and
6394 libattr is thus unnecessary.
6395
6396 * Virtualization detection works without privileges now. This
6397 means the systemd-detect-virt binary no longer requires
6398 CAP_SYS_PTRACE file capabilities, and our daemons can run
6399 with fewer privileges.
6400
6401 * systemd-networkd now runs under its own "systemd-network"
6402 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
6403 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
6404 loses the ability to write to files owned by root this way.
6405
6406 * Similarly, systemd-resolved now runs under its own
6407 "systemd-resolve" user with no capabilities remaining.
6408
6409 * Similarly, systemd-bus-proxyd now runs under its own
6410 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
6411
6412 * systemd-networkd gained support for setting up "veth"
6413 virtual Ethernet devices for container connectivity, as well
6414 as GRE and VTI tunnels.
6415
6416 * systemd-networkd will no longer automatically attempt to
6417 manually load kernel modules necessary for certain tunnel
6418 transports. Instead, it is assumed the kernel loads them
6419 automatically when required. This only works correctly on
6420 very new kernels. On older kernels, please consider adding
6421 the kernel modules to /etc/modules-load.d/ as a work-around.
6422
6423 * The resolv.conf file systemd-resolved generates has been
6424 moved to /run/systemd/resolve/. If you have a symlink from
6425 /etc/resolv.conf, it might be necessary to correct it.
6426
6427 * Two new service settings, ProtectHome= and ProtectSystem=,
6428 have been added. When enabled, they will make the user data
6429 (such as /home) inaccessible or read-only and the system
6430 (such as /usr) read-only, for specific services. This allows
6431 very light-weight per-service sandboxing to avoid
6432 modifications of user data or system files from
6433 services. These two new switches have been enabled for all
6434 of systemd's long-running services, where appropriate.
6435
6436 * Socket units gained new SocketUser= and SocketGroup=
6437 settings to set the owner user and group of AF_UNIX sockets
6438 and FIFOs in the file system.
6439
6440 * Socket units gained a new RemoveOnStop= setting. If enabled,
6441 all FIFOS and sockets in the file system will be removed
6442 when the specific socket unit is stopped.
6443
6444 * Socket units gained a new Symlinks= setting. It takes a list
6445 of symlinks to create to file system sockets or FIFOs
6446 created by the specific Unix sockets. This is useful to
6447 manage symlinks to socket nodes with the same lifecycle as
6448 the socket itself.
6449
6450 * The /dev/log socket and /dev/initctl FIFO have been moved to
6451 /run, and have been replaced by symlinks. This allows
6452 connecting to these facilities even if PrivateDevices=yes is
6453 used for a service (which makes /dev/log itself unavailable,
6454 but /run is left). This also has the benefit of ensuring
6455 that /dev only contains device nodes, directories and
6456 symlinks, and nothing else.
6457
6458 * sd-daemon gained two new calls sd_pid_notify() and
6459 sd_pid_notifyf(). They are similar to sd_notify() and
6460 sd_notifyf(), but allow overriding of the source PID of
6461 notification messages if permissions permit this. This is
6462 useful to send notify messages on behalf of a different
6463 process (for example, the parent process). The
6464 systemd-notify tool has been updated to make use of this
6465 when sending messages (so that notification messages now
6466 originate from the shell script invoking systemd-notify and
6467 not the systemd-notify process itself. This should minimize
6468 a race where systemd fails to associate notification
6469 messages to services when the originating process already
6470 vanished.
6471
6472 * A new "on-abnormal" setting for Restart= has been added. If
6473 set, it will result in automatic restarts on all "abnormal"
6474 reasons for a process to exit, which includes unclean
6475 signals, core dumps, timeouts and watchdog timeouts, but
6476 does not include clean and unclean exit codes or clean
6477 signals. Restart=on-abnormal is an alternative for
6478 Restart=on-failure for services that shall be able to
6479 terminate and avoid restarts on certain errors, by
6480 indicating so with an unclean exit code. Restart=on-failure
6481 or Restart=on-abnormal is now the recommended setting for
6482 all long-running services.
6483
6484 * If the InaccessibleDirectories= service setting points to a
6485 mount point (or if there are any submounts contained within
6486 it), it is now attempted to completely unmount it, to make
6487 the file systems truly unavailable for the respective
6488 service.
6489
6490 * The ReadOnlyDirectories= service setting and
6491 systemd-nspawn's --read-only parameter are now recursively
6492 applied to all submounts, too.
6493
6494 * Mount units may now be created transiently via the bus APIs.
6495
6496 * The support for SysV and LSB init scripts has been removed
6497 from the systemd daemon itself. Instead, it is now
6498 implemented as a generator that creates native systemd units
6499 from these scripts when needed. This enables us to remove a
6500 substantial amount of legacy code from PID 1, following the
6501 fact that many distributions only ship a very small number
6502 of LSB/SysV init scripts nowadays.
6503
6504 * Privileged Xen (dom0) domains are not considered
6505 virtualization anymore by the virtualization detection
6506 logic. After all, they generally have unrestricted access to
6507 the hardware and usually are used to manage the unprivileged
6508 (domU) domains.
6509
6510 * systemd-tmpfiles gained a new "C" line type, for copying
6511 files or entire directories.
6512
6513 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
6514 lines. So far, they have been non-globbing versions of the
6515 latter, and have thus been redundant. In future, it is
6516 recommended to only use "z". "m" has hence been removed
6517 from the documentation, even though it stays supported.
6518
6519 * A tmpfiles snippet to recreate the most basic structure in
6520 /var has been added. This is enough to create the /var/run →
6521 /run symlink and create a couple of structural
6522 directories. This allows systems to boot up with an empty or
6523 volatile /var. Of course, while with this change, the core OS
6524 now is capable with dealing with a volatile /var, not all
6525 user services are ready for it. However, we hope that sooner
6526 or later, many service daemons will be changed upstream so
6527 that they are able to automatically create their necessary
6528 directories in /var at boot, should they be missing. This is
6529 the first step to allow state-less systems that only require
6530 the vendor image for /usr to boot.
6531
6532 * systemd-nspawn has gained a new --tmpfs= switch to mount an
6533 empty tmpfs instance to a specific directory. This is
6534 particularly useful for making use of the automatic
6535 reconstruction of /var (see above), by passing --tmpfs=/var.
6536
6537 * Access modes specified in tmpfiles snippets may now be
6538 prefixed with "~", which indicates that they shall be masked
6539 by whether the existing file or directory is currently
6540 writable, readable or executable at all. Also, if specified,
6541 the sgid/suid/sticky bits will be masked for all
6542 non-directories.
6543
6544 * A new passive target unit "network-pre.target" has been
6545 added which is useful for services that shall run before any
6546 network is configured, for example firewall scripts.
6547
6548 * The "floppy" group that previously owned the /dev/fd*
6549 devices is no longer used. The "disk" group is now used
6550 instead. Distributions should probably deprecate usage of
6551 this group.
6552
6553 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
6554 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
6555 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
6556 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
6557 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
6558 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
6559 Jędrzejewski-Szmek
6560
6561 — Berlin, 2014-06-11
6562
6563 CHANGES WITH 213:
6564
6565 * A new "systemd-timesyncd" daemon has been added for
6566 synchronizing the system clock across the network. It
6567 implements an SNTP client. In contrast to NTP
6568 implementations such as chrony or the NTP reference server,
6569 this only implements a client side, and does not bother with
6570 the full NTP complexity, focusing only on querying time from
6571 one remote server and synchronizing the local clock to
6572 it. Unless you intend to serve NTP to networked clients or
6573 want to connect to local hardware clocks, this simple NTP
6574 client should be more than appropriate for most
6575 installations. The daemon runs with minimal privileges, and
6576 has been hooked up with networkd to only operate when
6577 network connectivity is available. The daemon saves the
6578 current clock to disk every time a new NTP sync has been
6579 acquired, and uses this to possibly correct the system clock
6580 early at bootup, in order to accommodate for systems that
6581 lack an RTC such as the Raspberry Pi and embedded devices,
6582 and to make sure that time monotonically progresses on these
6583 systems, even if it is not always correct. To make use of
6584 this daemon, a new system user and group "systemd-timesync"
6585 needs to be created on installation of systemd.
6586
6587 * The queue "seqnum" interface of libudev has been disabled, as
6588 it was generally incompatible with device namespacing as
6589 sequence numbers of devices go "missing" if the devices are
6590 part of a different namespace.
6591
6592 * "systemctl list-timers" and "systemctl list-sockets" gained
6593 a --recursive switch for showing units of these types also
6594 for all local containers, similar in style to the already
6595 supported --recursive switch for "systemctl list-units".
6596
6597 * A new RebootArgument= setting has been added for service
6598 units, which may be used to specify a kernel reboot argument
6599 to use when triggering reboots with StartLimitAction=.
6600
6601 * A new FailureAction= setting has been added for service
6602 units which may be used to specify an operation to trigger
6603 when a service fails. This works similarly to
6604 StartLimitAction=, but unlike it, controls what is done
6605 immediately rather than only after several attempts to
6606 restart the service in question.
6607
6608 * hostnamed got updated to also expose the kernel name,
6609 release, and version on the bus. This is useful for
6610 executing commands like hostnamectl with the -H switch.
6611 systemd-analyze makes use of this to properly display
6612 details when running non-locally.
6613
6614 * The bootchart tool can now show cgroup information in the
6615 graphs it generates.
6616
6617 * The CFS CPU quota cgroup attribute is now exposed for
6618 services. The new CPUQuota= switch has been added for this
6619 which takes a percentage value. Setting this will have the
6620 result that a service may never get more CPU time than the
6621 specified percentage, even if the machine is otherwise idle.
6622
6623 * systemd-networkd learned IPIP and SIT tunnel support.
6624
6625 * LSB init scripts exposing a dependency on $network will now
6626 get a dependency on network-online.target rather than simply
6627 network.target. This should bring LSB handling closer to
6628 what it was on SysV systems.
6629
6630 * A new fsck.repair= kernel option has been added to control
6631 how fsck shall deal with unclean file systems at boot.
6632
6633 * The (.ini) configuration file parser will now silently
6634 ignore sections whose name begins with "X-". This may be
6635 used to maintain application-specific extension sections in unit
6636 files.
6637
6638 * machined gained a new API to query the IP addresses of
6639 registered containers. "machinectl status" has been updated
6640 to show these addresses in its output.
6641
6642 * A new call sd_uid_get_display() has been added to the
6643 sd-login APIs for querying the "primary" session of a
6644 user. The "primary" session of the user is elected from the
6645 user's sessions and generally a graphical session is
6646 preferred over a text one.
6647
6648 * A minimal systemd-resolved daemon has been added. It
6649 currently simply acts as a companion to systemd-networkd and
6650 manages resolv.conf based on per-interface DNS
6651 configuration, possibly supplied via DHCP. In the long run
6652 we hope to extend this into a local DNSSEC enabled DNS and
6653 mDNS cache.
6654
6655 * The systemd-networkd-wait-online tool is now enabled by
6656 default. It will delay network-online.target until a network
6657 connection has been configured. The tool primarily integrates
6658 with networkd, but will also make a best effort to make sense
6659 of network configuration performed in some other way.
6660
6661 * Two new service options StartupCPUShares= and
6662 StartupBlockIOWeight= have been added that work similarly to
6663 CPUShares= and BlockIOWeight= however only apply during
6664 system startup. This is useful to prioritize certain services
6665 differently during bootup than during normal runtime.
6666
6667 * hostnamed has been changed to prefer the statically
6668 configured hostname in /etc/hostname (unless set to
6669 'localhost' or empty) over any dynamic one supplied by
6670 dhcp. With this change, the rules for picking the hostname
6671 match more closely the rules of other configuration settings
6672 where the local administrator's configuration in /etc always
6673 overrides any other settings.
6674
6675 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
6676 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
6677 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
6678 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
6679 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
6680 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
6681 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
6682 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
6683 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
6684 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
6685 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
6686 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
6687 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
6688 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
6689 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
6690 Lindskog, WaLyong Cho, Will Woods, Zbigniew
6691 Jędrzejewski-Szmek
6692
6693 — Beijing, 2014-05-28
6694
6695 CHANGES WITH 212:
6696
6697 * When restoring the screen brightness at boot, stay away from
6698 the darkest setting or from the lowest 5% of the available
6699 range, depending on which is the larger value of both. This
6700 should effectively protect the user from rebooting into a
6701 black screen, should the brightness have been set to minimum
6702 by accident.
6703
6704 * sd-login gained a new sd_machine_get_class() call to
6705 determine the class ("vm" or "container") of a machine
6706 registered with machined.
6707
6708 * sd-login gained new calls
6709 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
6710 to query the identity of the peer of a local AF_UNIX
6711 connection. They operate similarly to their sd_pid_get_xyz()
6712 counterparts.
6713
6714 * PID 1 will now maintain a system-wide system state engine
6715 with the states "starting", "running", "degraded",
6716 "maintenance", "stopping". These states are bound to system
6717 startup, normal runtime, runtime with at least one failed
6718 service, rescue/emergency mode and system shutdown. This
6719 state is shown in the "systemctl status" output when no unit
6720 name is passed. It is useful to determine system state, in
6721 particularly when doing so for many systems or containers at
6722 once.
6723
6724 * A new command "list-machines" has been added to "systemctl"
6725 that lists all local OS containers and shows their system
6726 state (see above), if systemd runs inside of them.
6727
6728 * systemctl gained a new "-r" switch to recursively enumerate
6729 units on all local containers, when used with the
6730 "list-unit" command (which is the default one that is
6731 executed when no parameters are specified).
6732
6733 * The GPT automatic partition discovery logic will now honour
6734 two GPT partition flags: one may be set on a partition to
6735 cause it to be mounted read-only, and the other may be set
6736 on a partition to ignore it during automatic discovery.
6737
6738 * Two new GPT type UUIDs have been added for automatic root
6739 partition discovery, for 32-bit and 64-bit ARM. This is not
6740 particularly useful for discovering the root directory on
6741 these architectures during bare-metal boots (since UEFI is
6742 not common there), but still very useful to allow booting of
6743 ARM disk images in nspawn with the -i option.
6744
6745 * MAC addresses of interfaces created with nspawn's
6746 --network-interface= switch will now be generated from the
6747 machine name, and thus be stable between multiple invocations
6748 of the container.
6749
6750 * logind will now automatically remove all IPC objects owned
6751 by a user if she or he fully logs out. This makes sure that
6752 users who are logged out cannot continue to consume IPC
6753 resources. This covers SysV memory, semaphores and message
6754 queues as well as POSIX shared memory and message
6755 queues. Traditionally, SysV and POSIX IPC had no lifecycle
6756 limits. With this functionality, that is corrected. This may
6757 be turned off by using the RemoveIPC= switch of logind.conf.
6758
6759 * The systemd-machine-id-setup and tmpfiles tools gained a
6760 --root= switch to operate on a specific root directory,
6761 instead of /.
6762
6763 * journald can now forward logged messages to the TTYs of all
6764 logged in users ("wall"). This is the default for all
6765 emergency messages now.
6766
6767 * A new tool systemd-journal-remote has been added to stream
6768 journal log messages across the network.
6769
6770 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
6771 controller trees are mounted into it. Note that the
6772 directories mounted beneath it are not read-only. This is a
6773 security measure and is particularly useful because glibc
6774 actually includes a search logic to pick any tmpfs it can
6775 find to implement shm_open() if /dev/shm is not available
6776 (which it might very well be in namespaced setups).
6777
6778 * machinectl gained a new "poweroff" command to cleanly power
6779 down a local OS container.
6780
6781 * The PrivateDevices= unit file setting will now also drop the
6782 CAP_MKNOD capability from the capability bound set, and
6783 imply DevicePolicy=closed.
6784
6785 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
6786 comprehensively on all long-running systemd services where
6787 this is appropriate.
6788
6789 * systemd-udevd will now run in a disassociated mount
6790 namespace. To mount directories from udev rules, make sure to
6791 pull in mount units via SYSTEMD_WANTS properties.
6792
6793 * The kdbus support gained support for uploading policy into
6794 the kernel. sd-bus gained support for creating "monitoring"
6795 connections that can eavesdrop into all bus communication
6796 for debugging purposes.
6797
6798 * Timestamps may now be specified in seconds since the UNIX
6799 epoch Jan 1st, 1970 by specifying "@" followed by the value
6800 in seconds.
6801
6802 * Native tcpwrap support in systemd has been removed. tcpwrap
6803 is old code, not really maintained anymore and has serious
6804 shortcomings, and better options such as firewalls
6805 exist. For setups that require tcpwrap usage, please
6806 consider invoking your socket-activated service via tcpd,
6807 like on traditional inetd.
6808
6809 * A new system.conf configuration option
6810 DefaultTimerAccuracySec= has been added that controls the
6811 default AccuracySec= setting of .timer units.
6812
6813 * Timer units gained a new WakeSystem= switch. If enabled,
6814 timers configured this way will cause the system to resume
6815 from system suspend (if the system supports that, which most
6816 do these days).
6817
6818 * Timer units gained a new Persistent= switch. If enabled,
6819 timers configured this way will save to disk when they have
6820 been last triggered. This information is then used on next
6821 reboot to possible execute overdue timer events, that
6822 could not take place because the system was powered off.
6823 This enables simple anacron-like behaviour for timer units.
6824
6825 * systemctl's "list-timers" will now also list the time a
6826 timer unit was last triggered in addition to the next time
6827 it will be triggered.
6828
6829 * systemd-networkd will now assign predictable IPv4LL
6830 addresses to its local interfaces.
6831
6832 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
6833 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
6834 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
6835 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
6836 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
6837 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
6838 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
6839 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
6840 Jędrzejewski-Szmek
6841
6842 — Berlin, 2014-03-25
6843
6844 CHANGES WITH 211:
6845
6846 * A new unit file setting RestrictAddressFamilies= has been
6847 added to restrict which socket address families unit
6848 processes gain access to. This takes address family names
6849 like "AF_INET" or "AF_UNIX", and is useful to minimize the
6850 attack surface of services via exotic protocol stacks. This
6851 is built on seccomp system call filters.
6852
6853 * Two new unit file settings RuntimeDirectory= and
6854 RuntimeDirectoryMode= have been added that may be used to
6855 manage a per-daemon runtime directories below /run. This is
6856 an alternative for setting up directory permissions with
6857 tmpfiles snippets, and has the advantage that the runtime
6858 directory's lifetime is bound to the daemon runtime and that
6859 the daemon starts up with an empty directory each time. This
6860 is particularly useful when writing services that drop
6861 privileges using the User= or Group= setting.
6862
6863 * The DeviceAllow= unit setting now supports globbing for
6864 matching against device group names.
6865
6866 * The systemd configuration file system.conf gained new
6867 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
6868 DefaultMemoryAccounting= to globally turn on/off accounting
6869 for specific resources (cgroups) for all units. These
6870 settings may still be overridden individually in each unit
6871 though.
6872
6873 * systemd-gpt-auto-generator is now able to discover /srv and
6874 root partitions in addition to /home and swap partitions. It
6875 also supports LUKS-encrypted partitions now. With this in
6876 place, automatic discovery of partitions to mount following
6877 the Discoverable Partitions Specification
6878 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
6879 is now a lot more complete. This allows booting without
6880 /etc/fstab and without root= on the kernel command line on
6881 systems prepared appropriately.
6882
6883 * systemd-nspawn gained a new --image= switch which allows
6884 booting up disk images and Linux installations on any block
6885 device that follow the Discoverable Partitions Specification
6886 (see above). This means that installations made with
6887 appropriately updated installers may now be started and
6888 deployed using container managers, completely
6889 unmodified. (We hope that libvirt-lxc will add support for
6890 this feature soon, too.)
6891
6892 * systemd-nspawn gained a new --network-macvlan= setting to
6893 set up a private macvlan interface for the
6894 container. Similarly, systemd-networkd gained a new
6895 Kind=macvlan setting in .netdev files.
6896
6897 * systemd-networkd now supports configuring local addresses
6898 using IPv4LL.
6899
6900 * A new tool systemd-network-wait-online has been added to
6901 synchronously wait for network connectivity using
6902 systemd-networkd.
6903
6904 * The sd-bus.h bus API gained a new sd_bus_track object for
6905 tracking the lifecycle of bus peers. Note that sd-bus.h is
6906 still not a public API though (unless you specify
6907 --enable-kdbus on the configure command line, which however
6908 voids your warranty and you get no API stability guarantee).
6909
6910 * The $XDG_RUNTIME_DIR runtime directories for each user are
6911 now individual tmpfs instances, which has the benefit of
6912 introducing separate pools for each user, with individual
6913 size limits, and thus making sure that unprivileged clients
6914 can no longer negatively impact the system or other users by
6915 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
6916 RuntimeDirectorySize= has been introduced that allows
6917 controlling the default size limit for all users. It
6918 defaults to 10% of the available physical memory. This is no
6919 replacement for quotas on tmpfs though (which the kernel
6920 still does not support), as /dev/shm and /tmp are still
6921 shared resources used by both the system and unprivileged
6922 users.
6923
6924 * logind will now automatically turn off automatic suspending
6925 on laptop lid close when more than one display is
6926 connected. This was previously expected to be implemented
6927 individually in desktop environments (such as GNOME),
6928 however has been added to logind now, in order to fix a
6929 boot-time race where a desktop environment might not have
6930 been started yet and thus not been able to take an inhibitor
6931 lock at the time where logind already suspends the system
6932 due to a closed lid.
6933
6934 * logind will now wait at least 30s after each system
6935 suspend/resume cycle, and 3min after system boot before
6936 suspending the system due to a closed laptop lid. This
6937 should give USB docking stations and similar enough time to
6938 be probed and configured after system resume and boot in
6939 order to then act as suspend blocker.
6940
6941 * systemd-run gained a new --property= setting which allows
6942 initialization of resource control properties (and others)
6943 for the created scope or service unit. Example: "systemd-run
6944 --property=BlockIOWeight=10 updatedb" may be used to run
6945 updatedb at a low block IO scheduling weight.
6946
6947 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
6948 now also work in --scope mode.
6949
6950 * When systemd is compiled with kdbus support, basic support
6951 for enforced policies is now in place. (Note that enabling
6952 kdbus still voids your warranty and no API compatibility
6953 promises are made.)
6954
6955 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
6956 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
6957 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
6958 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
6959 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
6960 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
6961 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
6962 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
6963 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
6964 Zbigniew Jędrzejewski-Szmek
6965
6966 — Berlin, 2014-03-12
6967
6968 CHANGES WITH 210:
6969
6970 * systemd will now relabel /dev after loading the SMACK policy
6971 according to SMACK rules.
6972
6973 * A new unit file option AppArmorProfile= has been added to
6974 set the AppArmor profile for the processes of a unit.
6975
6976 * A new condition check ConditionArchitecture= has been added
6977 to conditionalize units based on the system architecture, as
6978 reported by uname()'s "machine" field.
6979
6980 * systemd-networkd now supports matching on the system
6981 virtualization, architecture, kernel command line, host name
6982 and machine ID.
6983
6984 * logind is now a lot more aggressive when suspending the
6985 machine due to a closed laptop lid. Instead of acting only
6986 on the lid close action, it will continuously watch the lid
6987 status and act on it. This is useful for laptops where the
6988 power button is on the outside of the chassis so that it can
6989 be reached without opening the lid (such as the Lenovo
6990 Yoga). On those machines, logind will now immediately
6991 re-suspend the machine if the power button has been
6992 accidentally pressed while the laptop was suspended and in a
6993 backpack or similar.
6994
6995 * logind will now watch SW_DOCK switches and inhibit reaction
6996 to the lid switch if it is pressed. This means that logind
6997 will not suspend the machine anymore if the lid is closed
6998 and the system is docked, if the laptop supports SW_DOCK
6999 notifications via the input layer. Note that ACPI docking
7000 stations do not generate this currently. Also note that this
7001 logic is usually not fully sufficient and Desktop
7002 Environments should take a lid switch inhibitor lock when an
7003 external display is connected, as systemd will not watch
7004 this on its own.
7005
7006 * nspawn will now make use of the devices cgroup controller by
7007 default, and only permit creation of and access to the usual
7008 API device nodes like /dev/null or /dev/random, as well as
7009 access to (but not creation of) the pty devices.
7010
7011 * We will now ship a default .network file for
7012 systemd-networkd that automatically configures DHCP for
7013 network interfaces created by nspawn's --network-veth or
7014 --network-bridge= switches.
7015
7016 * systemd will now understand the usual M, K, G, T suffixes
7017 according to SI conventions (i.e. to the base 1000) when
7018 referring to throughput and hardware metrics. It will stay
7019 with IEC conventions (i.e. to the base 1024) for software
7020 metrics, according to what is customary according to
7021 Wikipedia. We explicitly document which base applies for
7022 each configuration option.
7023
7024 * The DeviceAllow= setting in unit files now supports a syntax
7025 to whitelist an entire group of devices node majors at once,
7026 based on the /proc/devices listing. For example, with the
7027 string "char-pts", it is now possible to whitelist all
7028 current and future pseudo-TTYs at once.
7029
7030 * sd-event learned a new "post" event source. Event sources of
7031 this type are triggered by the dispatching of any event
7032 source of a type that is not "post". This is useful for
7033 implementing clean-up and check event sources that are
7034 triggered by other work being done in the program.
7035
7036 * systemd-networkd is no longer statically enabled, but uses
7037 the usual [Install] sections so that it can be
7038 enabled/disabled using systemctl. It still is enabled by
7039 default however.
7040
7041 * When creating a veth interface pair with systemd-nspawn, the
7042 host side will now be prefixed with "vb-" if
7043 --network-bridge= is used, and with "ve-" if --network-veth
7044 is used. This way, it is easy to distinguish these cases on
7045 the host, for example to apply different configuration to
7046 them with systemd-networkd.
7047
7048 * The compatibility libraries for libsystemd-journal.so,
7049 libsystem-id128.so, libsystemd-login.so and
7050 libsystemd-daemon.so do not make use of IFUNC
7051 anymore. Instead, we now build libsystemd.so multiple times
7052 under these alternative names. This means that the footprint
7053 is drastically increased, but given that these are
7054 transitional compatibility libraries, this should not matter
7055 much. This change has been made necessary to support the ARM
7056 platform for these compatibility libraries, as the ARM
7057 toolchain is not really at the same level as the toolchain
7058 for other architectures like x86 and does not support
7059 IFUNC. Please make sure to use --enable-compat-libs only
7060 during a transitional period!
7061
7062 * The .include syntax has been deprecated and is not documented
7063 anymore. Drop-in files in .d directories should be used instead.
7064
7065 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
7066 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
7067 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
7068 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
7069 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
7070 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
7071 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
7072 Zbigniew Jędrzejewski-Szmek
7073
7074 — Berlin, 2014-02-24
7075
7076 CHANGES WITH 209:
7077
7078 * A new component "systemd-networkd" has been added that can
7079 be used to configure local network interfaces statically or
7080 via DHCP. It is capable of bringing up bridges, VLANs, and
7081 bonding. Currently, no hook-ups for interactive network
7082 configuration are provided. Use this for your initrd,
7083 container, embedded, or server setup if you need a simple,
7084 yet powerful, network configuration solution. This
7085 configuration subsystem is quite nifty, as it allows wildcard
7086 hotplug matching in interfaces. For example, with a single
7087 configuration snippet, you can configure that all Ethernet
7088 interfaces showing up are automatically added to a bridge,
7089 or similar. It supports link-sensing and more.
7090
7091 * A new tool "systemd-socket-proxyd" has been added which can
7092 act as a bidirectional proxy for TCP sockets. This is
7093 useful for adding socket activation support to services that
7094 do not actually support socket activation, including virtual
7095 machines and the like.
7096
7097 * Add a new tool to save/restore rfkill state on
7098 shutdown/boot.
7099
7100 * Save/restore state of keyboard backlights in addition to
7101 display backlights on shutdown/boot.
7102
7103 * udev learned a new SECLABEL{} construct to label device
7104 nodes with a specific security label when they appear. For
7105 now, only SECLABEL{selinux} is supported, but the syntax is
7106 prepared for additional security frameworks.
7107
7108 * udev gained a new scheme to configure link-level attributes
7109 from files in /etc/systemd/network/*.link. These files can
7110 match against MAC address, device path, driver name and type,
7111 and will apply attributes like the naming policy, link speed,
7112 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
7113 address assignment policy (randomized, ...).
7114
7115 * The configuration of network interface naming rules for
7116 "permanent interface names" has changed: a new NamePolicy=
7117 setting in the [Link] section of .link files determines the
7118 priority of possible naming schemes (onboard, slot, MAC,
7119 path). The default value of this setting is determined by
7120 /usr/lib/net/links/99-default.link. Old
7121 80-net-name-slot.rules udev configuration file has been
7122 removed, so local configuration overriding this file should
7123 be adapted to override 99-default.link instead.
7124
7125 * When the User= switch is used in a unit file, also
7126 initialize $SHELL= based on the user database entry.
7127
7128 * systemd no longer depends on libdbus. All communication is
7129 now done with sd-bus, systemd's low-level bus library
7130 implementation.
7131
7132 * kdbus support has been added to PID 1 itself. When kdbus is
7133 enabled, this causes PID 1 to set up the system bus and
7134 enable support for a new ".busname" unit type that
7135 encapsulates bus name activation on kdbus. It works a little
7136 bit like ".socket" units, except for bus names. A new
7137 generator has been added that converts classic dbus1 service
7138 activation files automatically into native systemd .busname
7139 and .service units.
7140
7141 * sd-bus: add a light-weight vtable implementation that allows
7142 defining objects on the bus with a simple static const
7143 vtable array of its methods, signals and properties.
7144
7145 * systemd will not generate or install static dbus
7146 introspection data anymore to /usr/share/dbus-1/interfaces,
7147 as the precise format of these files is unclear, and
7148 nothing makes use of it.
7149
7150 * A proxy daemon is now provided to proxy clients connecting
7151 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
7152 compatibility with classic D-Bus.
7153
7154 * A bus driver implementation has been added that supports the
7155 classic D-Bus bus driver calls on kdbus, also for
7156 compatibility purposes.
7157
7158 * A new API "sd-event.h" has been added that implements a
7159 minimal event loop API built around epoll. It provides a
7160 couple of features that direct epoll usage is lacking:
7161 prioritization of events, scales to large numbers of timer
7162 events, per-event timer slack (accuracy), system-wide
7163 coalescing of timer events, exit handlers, watchdog
7164 supervision support using systemd's sd_notify() API, child
7165 process handling.
7166
7167 * A new API "sd-rntl.h" has been added that provides an API
7168 around the route netlink interface of the kernel, similar in
7169 style to "sd-bus.h".
7170
7171 * A new API "sd-dhcp-client.h" has been added that provides a
7172 small DHCPv4 client-side implementation. This is used by
7173 "systemd-networkd".
7174
7175 * There is a new kernel command line option
7176 "systemd.restore_state=0|1". When set to "0", none of the
7177 systemd tools will restore saved runtime state to hardware
7178 devices. More specifically, the rfkill and backlight states
7179 are not restored.
7180
7181 * The FsckPassNo= compatibility option in mount/service units
7182 has been removed. The fstab generator will now add the
7183 necessary dependencies automatically, and does not require
7184 PID1's support for that anymore.
7185
7186 * journalctl gained a new switch, --list-boots, that lists
7187 recent boots with their times and boot IDs.
7188
7189 * The various tools like systemctl, loginctl, timedatectl,
7190 busctl, systemd-run, ... have gained a new switch "-M" to
7191 connect to a specific, local OS container (as direct
7192 connection, without requiring SSH). This works on any
7193 container that is registered with machined, such as those
7194 created by libvirt-lxc or nspawn.
7195
7196 * systemd-run and systemd-analyze also gained support for "-H"
7197 to connect to remote hosts via SSH. This is particularly
7198 useful for systemd-run because it enables queuing of jobs
7199 onto remote systems.
7200
7201 * machinectl gained a new command "login" to open a getty
7202 login in any local container. This works with any container
7203 that is registered with machined (such as those created by
7204 libvirt-lxc or nspawn), and which runs systemd inside.
7205
7206 * machinectl gained a new "reboot" command that may be used to
7207 trigger a reboot on a specific container that is registered
7208 with machined. This works on any container that runs an init
7209 system of some kind.
7210
7211 * systemctl gained a new "list-timers" command to print a nice
7212 listing of installed timer units with the times they elapse
7213 next.
7214
7215 * Alternative reboot() parameters may now be specified on the
7216 "systemctl reboot" command line and are passed to the
7217 reboot() system call.
7218
7219 * systemctl gained a new --job-mode= switch to configure the
7220 mode to queue a job with. This is a more generic version of
7221 --fail, --irreversible, and --ignore-dependencies, which are
7222 still available but not advertised anymore.
7223
7224 * /etc/systemd/system.conf gained new settings to configure
7225 various default timeouts of units, as well as the default
7226 start limit interval and burst. These may still be overridden
7227 within each Unit.
7228
7229 * PID1 will now export on the bus profile data of the security
7230 policy upload process (such as the SELinux policy upload to
7231 the kernel).
7232
7233 * journald: when forwarding logs to the console, include
7234 timestamps (following the setting in
7235 /sys/module/printk/parameters/time).
7236
7237 * OnCalendar= in timer units now understands the special
7238 strings "yearly" and "annually". (Both are equivalent)
7239
7240 * The accuracy of timer units is now configurable with the new
7241 AccuracySec= setting. It defaults to 1min.
7242
7243 * A new dependency type JoinsNamespaceOf= has been added that
7244 allows running two services within the same /tmp and network
7245 namespace, if PrivateNetwork= or PrivateTmp= are used.
7246
7247 * A new command "cat" has been added to systemctl. It outputs
7248 the original unit file of a unit, and concatenates the
7249 contents of additional "drop-in" unit file snippets, so that
7250 the full configuration is shown.
7251
7252 * systemctl now supports globbing on the various "list-xyz"
7253 commands, like "list-units" or "list-sockets", as well as on
7254 those commands which take multiple unit names.
7255
7256 * journalctl's --unit= switch gained support for globbing.
7257
7258 * All systemd daemons now make use of the watchdog logic so
7259 that systemd automatically notices when they hang.
7260
7261 * If the $container_ttys environment variable is set,
7262 getty-generator will automatically spawn a getty for each
7263 listed tty. This is useful for container managers to request
7264 login gettys to be spawned on as many ttys as needed.
7265
7266 * %h, %s, %U specifier support is not available anymore when
7267 used in unit files for PID 1. This is because NSS calls are
7268 not safe from PID 1. They stay available for --user
7269 instances of systemd, and as special case for the root user.
7270
7271 * loginctl gained a new "--no-legend" switch to turn off output
7272 of the legend text.
7273
7274 * The "sd-login.h" API gained three new calls:
7275 sd_session_is_remote(), sd_session_get_remote_user(),
7276 sd_session_get_remote_host() to query information about
7277 remote sessions.
7278
7279 * The udev hardware database now also carries vendor/product
7280 information of SDIO devices.
7281
7282 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
7283 determine whether watchdog notifications are requested by
7284 the system manager.
7285
7286 * Socket-activated per-connection services now include a
7287 short description of the connection parameters in the
7288 description.
7289
7290 * tmpfiles gained a new "--boot" option. When this is not used,
7291 only lines where the command character is not suffixed with
7292 "!" are executed. When this option is specified, those
7293 options are executed too. This partitions tmpfiles
7294 directives into those that can be safely executed at any
7295 time, and those which should be run only at boot (for
7296 example, a line that creates /run/nologin).
7297
7298 * A new API "sd-resolve.h" has been added which provides a simple
7299 asynchronous wrapper around glibc NSS host name resolution
7300 calls, such as getaddrinfo(). In contrast to glibc's
7301 getaddrinfo_a(), it does not use signals. In contrast to most
7302 other asynchronous name resolution libraries, this one does
7303 not reimplement DNS, but reuses NSS, so that alternate
7304 host name resolution systems continue to work, such as mDNS,
7305 LDAP, etc. This API is based on libasyncns, but it has been
7306 cleaned up for inclusion in systemd.
7307
7308 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
7309 "sd-daemon.h" are no longer found in individual libraries
7310 libsystemd-journal.so, libsystemd-login.so,
7311 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
7312 merged them into a single library, libsystemd.so, which
7313 provides all symbols. The reason for this is cyclic
7314 dependencies, as these libraries tend to use each other's
7315 symbols. So far, we have managed to workaround that by linking
7316 a copy of a good part of our code into each of these
7317 libraries again and again, which, however, makes certain
7318 things hard to do, like sharing static variables. Also, it
7319 substantially increases footprint. With this change, there
7320 is only one library for the basic APIs systemd
7321 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
7322 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
7323 library as well, however are subject to the --enable-kdbus
7324 switch (see below). Note that "sd-dhcp-client.h" is not part
7325 of this library (this is because it only consumes, never
7326 provides, services of/to other APIs). To make the transition
7327 easy from the separate libraries to the unified one, we
7328 provide the --enable-compat-libs compile-time switch which
7329 will generate stub libraries that are compatible with the
7330 old ones but redirect all calls to the new one.
7331
7332 * All of the kdbus logic and the new APIs "sd-bus.h",
7333 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
7334 and "sd-utf8.h" are compile-time optional via the
7335 "--enable-kdbus" switch, and they are not compiled in by
7336 default. To make use of kdbus, you have to explicitly enable
7337 the switch. Note however, that neither the kernel nor the
7338 userspace API for all of this is considered stable yet. We
7339 want to maintain the freedom to still change the APIs for
7340 now. By specifying this build-time switch, you acknowledge
7341 that you are aware of the instability of the current
7342 APIs.
7343
7344 * Also, note that while kdbus is pretty much complete,
7345 it lacks one thing: proper policy support. This means you
7346 can build a fully working system with all features; however,
7347 it will be highly insecure. Policy support will be added in
7348 one of the next releases, at the same time that we will
7349 declare the APIs stable.
7350
7351 * When the kernel command line argument "kdbus" is specified,
7352 systemd will automatically load the kdbus.ko kernel module. At
7353 this stage of development, it is only useful for testing kdbus
7354 and should not be used in production. Note: if "--enable-kdbus"
7355 is specified, and the kdbus.ko kernel module is available, and
7356 "kdbus" is added to the kernel command line, the entire system
7357 runs with kdbus instead of dbus-daemon, with the above mentioned
7358 problem of missing the system policy enforcement. Also a future
7359 version of kdbus.ko or a newer systemd will not be compatible with
7360 each other, and will unlikely be able to boot the machine if only
7361 one of them is updated.
7362
7363 * systemctl gained a new "import-environment" command which
7364 uploads the caller's environment (or parts thereof) into the
7365 service manager so that it is inherited by services started
7366 by the manager. This is useful to upload variables like
7367 $DISPLAY into the user service manager.
7368
7369 * A new PrivateDevices= switch has been added to service units
7370 which allows running a service with a namespaced /dev
7371 directory that does not contain any device nodes for
7372 physical devices. More specifically, it only includes devices
7373 such as /dev/null, /dev/urandom, and /dev/zero which are API
7374 entry points.
7375
7376 * logind has been extended to support behaviour like VT
7377 switching on seats that do not support a VT. This makes
7378 multi-session available on seats that are not the first seat
7379 (seat0), and on systems where kernel support for VTs has
7380 been disabled at compile-time.
7381
7382 * If a process holds a delay lock for system sleep or shutdown
7383 and fails to release it in time, we will now log its
7384 identity. This makes it easier to identify processes that
7385 cause slow suspends or power-offs.
7386
7387 * When parsing /etc/crypttab, support for a new key-slot=
7388 option as supported by Debian is added. It allows indicating
7389 which LUKS slot to use on disk, speeding up key loading.
7390
7391 * The sd_journal_sendv() API call has been checked and
7392 officially declared to be async-signal-safe so that it may
7393 be invoked from signal handlers for logging purposes.
7394
7395 * Boot-time status output is now enabled automatically after a
7396 short timeout if boot does not progress, in order to give
7397 the user an indication what she or he is waiting for.
7398
7399 * The boot-time output has been improved to show how much time
7400 remains until jobs expire.
7401
7402 * The KillMode= switch in service units gained a new possible
7403 value "mixed". If set, and the unit is shut down, then the
7404 initial SIGTERM signal is sent only to the main daemon
7405 process, while the following SIGKILL signal is sent to
7406 all remaining processes of the service.
7407
7408 * When a scope unit is registered, a new property "Controller"
7409 may be set. If set to a valid bus name, systemd will send a
7410 RequestStop() signal to this name when it would like to shut
7411 down the scope. This may be used to hook manager logic into
7412 the shutdown logic of scope units. Also, scope units may now
7413 be put in a special "abandoned" state, in which case the
7414 manager process which created them takes no further
7415 responsibilities for it.
7416
7417 * When reading unit files, systemd will now verify
7418 the access mode of these files, and warn about certain
7419 suspicious combinations. This has been added to make it
7420 easier to track down packaging bugs where unit files are
7421 marked executable or world-writable.
7422
7423 * systemd-nspawn gained a new "--setenv=" switch to set
7424 container-wide environment variables. The similar option in
7425 systemd-activate was renamed from "--environment=" to
7426 "--setenv=" for consistency.
7427
7428 * systemd-nspawn has been updated to create a new kdbus domain
7429 for each container that is invoked, thus allowing each
7430 container to have its own set of system and user buses,
7431 independent of the host.
7432
7433 * systemd-nspawn gained a new --drop-capability= switch to run
7434 the container with less capabilities than the default. Both
7435 --drop-capability= and --capability= now take the special
7436 string "all" for dropping or keeping all capabilities.
7437
7438 * systemd-nspawn gained new switches for executing containers
7439 with specific SELinux labels set.
7440
7441 * systemd-nspawn gained a new --quiet switch to not generate
7442 any additional output but the container's own console
7443 output.
7444
7445 * systemd-nspawn gained a new --share-system switch to run a
7446 container without PID namespacing enabled.
7447
7448 * systemd-nspawn gained a new --register= switch to control
7449 whether the container is registered with systemd-machined or
7450 not. This is useful for containers that do not run full
7451 OS images, but only specific apps.
7452
7453 * systemd-nspawn gained a new --keep-unit which may be used
7454 when invoked as the only program from a service unit, and
7455 results in registration of the unit service itself in
7456 systemd-machined, instead of a newly opened scope unit.
7457
7458 * systemd-nspawn gained a new --network-interface= switch for
7459 moving arbitrary interfaces to the container. The new
7460 --network-veth switch creates a virtual Ethernet connection
7461 between host and container. The new --network-bridge=
7462 switch then allows assigning the host side of this virtual
7463 Ethernet connection to a bridge device.
7464
7465 * systemd-nspawn gained a new --personality= switch for
7466 setting the kernel personality for the container. This is
7467 useful when running a 32-bit container on a 64-bit host. A
7468 similar option Personality= is now also available for service
7469 units to use.
7470
7471 * logind will now also track a "Desktop" identifier for each
7472 session which encodes the desktop environment of it. This is
7473 useful for desktop environments that want to identify
7474 multiple running sessions of itself easily.
7475
7476 * A new SELinuxContext= setting for service units has been
7477 added that allows setting a specific SELinux execution
7478 context for a service.
7479
7480 * Most systemd client tools will now honour $SYSTEMD_LESS for
7481 settings of the "less" pager. By default, these tools will
7482 override $LESS to allow certain operations to work, such as
7483 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
7484 influence this logic.
7485
7486 * systemd's "seccomp" hook-up has been changed to make use of
7487 the libseccomp library instead of using its own
7488 implementation. This has benefits for portability among
7489 other things.
7490
7491 * For usage together with SystemCallFilter=, a new
7492 SystemCallErrorNumber= setting has been introduced that
7493 allows configuration of a system error number to be returned
7494 on filtered system calls, instead of immediately killing the
7495 process. Also, SystemCallArchitectures= has been added to
7496 limit access to system calls of a particular architecture
7497 (in order to turn off support for unused secondary
7498 architectures). There is also a global
7499 SystemCallArchitectures= setting in system.conf now to turn
7500 off support for non-native system calls system-wide.
7501
7502 * systemd requires a kernel with a working name_to_handle_at(),
7503 please see the kernel config requirements in the README file.
7504
7505 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
7506 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
7507 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
7508 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
7509 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
7510 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
7511 Elia Pinto, Florian Weimer, George McCollister, Goffredo
7512 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
7513 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
7514 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
7515 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
7516 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
7517 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
7518 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
7519 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
7520 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
7521 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
7522 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
7523 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
7524 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
7525 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
7526 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
7527 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
7528 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
7529
7530 — Berlin, 2014-02-20
7531
7532 CHANGES WITH 208:
7533
7534 * logind has gained support for facilitating privileged input
7535 and drm device access for unprivileged clients. This work is
7536 useful to allow Wayland display servers (and similar
7537 programs, such as kmscon) to run under the user's ID and
7538 access input and drm devices which are normally
7539 protected. When this is used (and the kernel is new enough)
7540 logind will "mute" IO on the file descriptors passed to
7541 Wayland as long as it is in the background and "unmute" it
7542 if it returns into the foreground. This allows secure
7543 session switching without allowing background sessions to
7544 eavesdrop on input and display data. This also introduces
7545 session switching support if VT support is turned off in the
7546 kernel, and on seats that are not seat0.
7547
7548 * A new kernel command line option luks.options= is understood
7549 now which allows specifying LUKS options for usage for LUKS
7550 encrypted partitions specified with luks.uuid=.
7551
7552 * tmpfiles.d(5) snippets may now use specifier expansion in
7553 path names. More specifically %m, %b, %H, %v, are now
7554 replaced by the local machine id, boot id, hostname, and
7555 kernel version number.
7556
7557 * A new tmpfiles.d(5) command "m" has been introduced which
7558 may be used to change the owner/group/access mode of a file
7559 or directory if it exists, but do nothing if it does not.
7560
7561 * This release removes high-level support for the
7562 MemorySoftLimit= cgroup setting. The underlying kernel
7563 cgroup attribute memory.soft_limit= is currently badly
7564 designed and likely to be removed from the kernel API in its
7565 current form, hence we should not expose it for now.
7566
7567 * The memory.use_hierarchy cgroup attribute is now enabled for
7568 all cgroups systemd creates in the memory cgroup
7569 hierarchy. This option is likely to be come the built-in
7570 default in the kernel anyway, and the non-hierarchical mode
7571 never made much sense in the intrinsically hierarchical
7572 cgroup system.
7573
7574 * A new field _SYSTEMD_SLICE= is logged along with all journal
7575 messages containing the slice a message was generated
7576 from. This is useful to allow easy per-customer filtering of
7577 logs among other things.
7578
7579 * systemd-journald will no longer adjust the group of journal
7580 files it creates to the "systemd-journal" group. Instead we
7581 rely on the journal directory to be owned by the
7582 "systemd-journal" group, and its setgid bit set, so that the
7583 kernel file system layer will automatically enforce that
7584 journal files inherit this group assignment. The reason for
7585 this change is that we cannot allow NSS look-ups from
7586 journald which would be necessary to resolve
7587 "systemd-journal" to a numeric GID, because this might
7588 create deadlocks if NSS involves synchronous queries to
7589 other daemons (such as nscd, or sssd) which in turn are
7590 logging clients of journald and might block on it, which
7591 would then dead lock. A tmpfiles.d(5) snippet included in
7592 systemd will make sure the setgid bit and group are
7593 properly set on the journal directory if it exists on every
7594 boot. However, we recommend adjusting it manually after
7595 upgrades too (or from RPM scriptlets), so that the change is
7596 not delayed until next reboot.
7597
7598 * Backlight and random seed files in /var/lib/ have moved into
7599 the /var/lib/systemd/ directory, in order to centralize all
7600 systemd generated files in one directory.
7601
7602 * Boot time performance measurements (as displayed by
7603 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
7604 performance information if that's available to determine how
7605 much time BIOS and boot loader initialization required. With
7606 a sufficiently new BIOS you hence no longer need to boot
7607 with Gummiboot to get access to such information.
7608
7609 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
7610 Cristian Rodríguez, Dave Reisner, David Herrmann, David
7611 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
7612 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
7613 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
7614 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
7615 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7616
7617 — Berlin, 2013-10-02
7618
7619 CHANGES WITH 207:
7620
7621 * The Restart= option for services now understands a new
7622 on-watchdog setting, which will restart the service
7623 automatically if the service stops sending out watchdog keep
7624 alive messages (as configured with WatchdogSec=).
7625
7626 * The getty generator (which is responsible for bringing up a
7627 getty on configured serial consoles) will no longer only
7628 start a getty on the primary kernel console but on all
7629 others, too. This makes the order in which console= is
7630 specified on the kernel command line less important.
7631
7632 * libsystemd-logind gained a new sd_session_get_vt() call to
7633 retrieve the VT number of a session.
7634
7635 * If the option "tries=0" is set for an entry of /etc/crypttab
7636 its passphrase is queried indefinitely instead of any
7637 maximum number of tries.
7638
7639 * If a service with a configure PID file terminates its PID
7640 file will now be removed automatically if it still exists
7641 afterwards. This should put an end to stale PID files.
7642
7643 * systemd-run will now also take relative binary path names
7644 for execution and no longer insists on absolute paths.
7645
7646 * InaccessibleDirectories= and ReadOnlyDirectories= now take
7647 paths that are optionally prefixed with "-" to indicate that
7648 it should not be considered a failure if they do not exist.
7649
7650 * journalctl -o (and similar commands) now understands a new
7651 output mode "short-precise", it is similar to "short" but
7652 shows timestamps with usec accuracy.
7653
7654 * The option "discard" (as known from Debian) is now
7655 synonymous to "allow-discards" in /etc/crypttab. In fact,
7656 "discard" is preferred now (since it is easier to remember
7657 and type).
7658
7659 * Some licensing clean-ups were made, so that more code is now
7660 LGPL-2.1 licensed than before.
7661
7662 * A minimal tool to save/restore the display backlight
7663 brightness across reboots has been added. It will store the
7664 backlight setting as late as possible at shutdown, and
7665 restore it as early as possible during reboot.
7666
7667 * A logic to automatically discover and enable home and swap
7668 partitions on GPT disks has been added. With this in place
7669 /etc/fstab becomes optional for many setups as systemd can
7670 discover certain partitions located on the root disk
7671 automatically. Home partitions are recognized under their
7672 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
7673 partitions are recognized under their GPT type ID
7674 0657fd6da4ab43c484e50933c84b4f4f.
7675
7676 * systemd will no longer pass any environment from the kernel
7677 or initrd to system services. If you want to set an
7678 environment for all services, do so via the kernel command
7679 line systemd.setenv= assignment.
7680
7681 * The systemd-sysctl tool no longer natively reads the file
7682 /etc/sysctl.conf. If desired, the file should be symlinked
7683 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
7684 legacy support by a symlink rather than built-in code, it
7685 also makes the otherwise hidden order of application of the
7686 different files visible. (Note that this partly reverts to a
7687 pre-198 application order of sysctl knobs!)
7688
7689 * The "systemctl set-log-level" and "systemctl dump" commands
7690 have been moved to systemd-analyze.
7691
7692 * systemd-run learned the new --remain-after-exit switch,
7693 which causes the scope unit not to be cleaned up
7694 automatically after the process terminated.
7695
7696 * tmpfiles learned a new --exclude-prefix= switch to exclude
7697 certain paths from operation.
7698
7699 * journald will now automatically flush all messages to disk
7700 as soon as a message at the log level CRIT, ALERT or EMERG
7701 is received.
7702
7703 Contributions from: Andrew Cook, Brandon Philips, Christian
7704 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
7705 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
7706 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
7707 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
7708 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
7709 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
7710 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
7711 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
7712 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
7713 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
7714 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
7715 William Giokas, Zbigniew Jędrzejewski-Szmek
7716
7717 — Berlin, 2013-09-13
7718
7719 CHANGES WITH 206:
7720
7721 * The documentation has been updated to cover the various new
7722 concepts introduced with 205.
7723
7724 * Unit files now understand the new %v specifier which
7725 resolves to the kernel version string as returned by "uname
7726 -r".
7727
7728 * systemctl now supports filtering the unit list output by
7729 load state, active state and sub state, using the new
7730 --state= parameter.
7731
7732 * "systemctl status" will now show the results of the
7733 condition checks (like ConditionPathExists= and similar) of
7734 the last start attempts of the unit. They are also logged to
7735 the journal.
7736
7737 * "journalctl -b" may now be used to look for boot output of a
7738 specific boot. Try "journalctl -b -1" for the previous boot,
7739 but the syntax is substantially more powerful.
7740
7741 * "journalctl --show-cursor" has been added which prints the
7742 cursor string the last shown log line. This may then be used
7743 with the new "journalctl --after-cursor=" switch to continue
7744 browsing logs from that point on.
7745
7746 * "journalctl --force" may now be used to force regeneration
7747 of an FSS key.
7748
7749 * Creation of "dead" device nodes has been moved from udev
7750 into kmod and tmpfiles. Previously, udev would read the kmod
7751 databases to pre-generate dead device nodes based on meta
7752 information contained in kernel modules, so that these would
7753 be auto-loaded on access rather then at boot. As this
7754 does not really have much to do with the exposing actual
7755 kernel devices to userspace this has always been slightly
7756 alien in the udev codebase. Following the new scheme kmod
7757 will now generate a runtime snippet for tmpfiles from the
7758 module meta information and it now is tmpfiles' job to the
7759 create the nodes. This also allows overriding access and
7760 other parameters for the nodes using the usual tmpfiles
7761 facilities. As side effect this allows us to remove the
7762 CAP_SYS_MKNOD capability bit from udevd entirely.
7763
7764 * logind's device ACLs may now be applied to these "dead"
7765 devices nodes too, thus finally allowing managed access to
7766 devices such as /dev/snd/sequencer without loading the
7767 backing module right-away.
7768
7769 * A new RPM macro has been added that may be used to apply
7770 tmpfiles configuration during package installation.
7771
7772 * systemd-detect-virt and ConditionVirtualization= now can
7773 detect User-Mode-Linux machines (UML).
7774
7775 * journald will now implicitly log the effective capabilities
7776 set of processes in the message metadata.
7777
7778 * systemd-cryptsetup has gained support for TrueCrypt volumes.
7779
7780 * The initrd interface has been simplified (more specifically,
7781 support for passing performance data via environment
7782 variables and fsck results via files in /run has been
7783 removed). These features were non-essential, and are
7784 nowadays available in a much nicer way by having systemd in
7785 the initrd serialize its state and have the hosts systemd
7786 deserialize it again.
7787
7788 * The udev "keymap" data files and tools to apply keyboard
7789 specific mappings of scan to key codes, and force-release
7790 scan code lists have been entirely replaced by a udev
7791 "keyboard" builtin and a hwdb data file.
7792
7793 * systemd will now honour the kernel's "quiet" command line
7794 argument also during late shutdown, resulting in a
7795 completely silent shutdown when used.
7796
7797 * There's now an option to control the SO_REUSEPORT socket
7798 option in .socket units.
7799
7800 * Instance units will now automatically get a per-template
7801 subslice of system.slice unless something else is explicitly
7802 configured. For example, instances of sshd@.service will now
7803 implicitly be placed in system-sshd.slice rather than
7804 system.slice as before.
7805
7806 * Test coverage support may now be enabled at build time.
7807
7808 Contributions from: Dave Reisner, Frederic Crozat, Harald
7809 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
7810 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
7811 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
7812 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
7813 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
7814 Giokas, Zbigniew Jędrzejewski-Szmek
7815
7816 — Berlin, 2013-07-23
7817
7818 CHANGES WITH 205:
7819
7820 * Two new unit types have been introduced:
7821
7822 Scope units are very similar to service units, however, are
7823 created out of pre-existing processes — instead of PID 1
7824 forking off the processes. By using scope units it is
7825 possible for system services and applications to group their
7826 own child processes (worker processes) in a powerful way
7827 which then maybe used to organize them, or kill them
7828 together, or apply resource limits on them.
7829
7830 Slice units may be used to partition system resources in an
7831 hierarchical fashion and then assign other units to them. By
7832 default there are now three slices: system.slice (for all
7833 system services), user.slice (for all user sessions),
7834 machine.slice (for VMs and containers).
7835
7836 Slices and scopes have been introduced primarily in
7837 context of the work to move cgroup handling to a
7838 single-writer scheme, where only PID 1
7839 creates/removes/manages cgroups.
7840
7841 * There's a new concept of "transient" units. In contrast to
7842 normal units these units are created via an API at runtime,
7843 not from configuration from disk. More specifically this
7844 means it is now possible to run arbitrary programs as
7845 independent services, with all execution parameters passed
7846 in via bus APIs rather than read from disk. Transient units
7847 make systemd substantially more dynamic then it ever was,
7848 and useful as a general batch manager.
7849
7850 * logind has been updated to make use of scope and slice units
7851 for managing user sessions. As a user logs in he will get
7852 his own private slice unit, to which all sessions are added
7853 as scope units. We also added support for automatically
7854 adding an instance of user@.service for the user into the
7855 slice. Effectively logind will no longer create cgroup
7856 hierarchies on its own now, it will defer entirely to PID 1
7857 for this by means of scope, service and slice units. Since
7858 user sessions this way become entities managed by PID 1
7859 the output of "systemctl" is now a lot more comprehensive.
7860
7861 * A new mini-daemon "systemd-machined" has been added which
7862 may be used by virtualization managers to register local
7863 VMs/containers. nspawn has been updated accordingly, and
7864 libvirt will be updated shortly. machined will collect a bit
7865 of meta information about the VMs/containers, and assign
7866 them their own scope unit (see above). The collected
7867 meta-data is then made available via the "machinectl" tool,
7868 and exposed in "ps" and similar tools. machined/machinectl
7869 is compile-time optional.
7870
7871 * As discussed earlier, the low-level cgroup configuration
7872 options ControlGroup=, ControlGroupModify=,
7873 ControlGroupPersistent=, ControlGroupAttribute= have been
7874 removed. Please use high-level attribute settings instead as
7875 well as slice units.
7876
7877 * A new bus call SetUnitProperties() has been added to alter
7878 various runtime parameters of a unit. This is primarily
7879 useful to alter cgroup parameters dynamically in a nice way,
7880 but will be extended later on to make more properties
7881 modifiable at runtime. systemctl gained a new set-properties
7882 command that wraps this call.
7883
7884 * A new tool "systemd-run" has been added which can be used to
7885 run arbitrary command lines as transient services or scopes,
7886 while configuring a number of settings via the command
7887 line. This tool is currently very basic, however already
7888 very useful. We plan to extend this tool to even allow
7889 queuing of execution jobs with time triggers from the
7890 command line, similar in fashion to "at".
7891
7892 * nspawn will now inform the user explicitly that kernels with
7893 audit enabled break containers, and suggest the user to turn
7894 off audit.
7895
7896 * Support for detecting the IMA and AppArmor security
7897 frameworks with ConditionSecurity= has been added.
7898
7899 * journalctl gained a new "-k" switch for showing only kernel
7900 messages, mimicking dmesg output; in addition to "--user"
7901 and "--system" switches for showing only user's own logs
7902 and system logs.
7903
7904 * systemd-delta can now show information about drop-in
7905 snippets extending unit files.
7906
7907 * libsystemd-bus has been substantially updated but is still
7908 not available as public API.
7909
7910 * systemd will now look for the "debug" argument on the kernel
7911 command line and enable debug logging, similar to what
7912 "systemd.log_level=debug" already did before.
7913
7914 * "systemctl set-default", "systemctl get-default" has been
7915 added to configure the default.target symlink, which
7916 controls what to boot into by default.
7917
7918 * "systemctl set-log-level" has been added as a convenient
7919 way to raise and lower systemd logging threshold.
7920
7921 * "systemd-analyze plot" will now show the time the various
7922 generators needed for execution, as well as information
7923 about the unit file loading.
7924
7925 * libsystemd-journal gained a new sd_journal_open_files() call
7926 for opening specific journal files. journactl also gained a
7927 new switch to expose this new functionality. Previously we
7928 only supported opening all files from a directory, or all
7929 files from the system, as opening individual files only is
7930 racy due to journal file rotation.
7931
7932 * systemd gained the new DefaultEnvironment= setting in
7933 /etc/systemd/system.conf to set environment variables for
7934 all services.
7935
7936 * If a privileged process logs a journal message with the
7937 OBJECT_PID= field set, then journald will automatically
7938 augment this with additional OBJECT_UID=, OBJECT_GID=,
7939 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
7940 system services want to log events about specific client
7941 processes. journactl/systemctl has been updated to make use
7942 of this information if all log messages regarding a specific
7943 unit is requested.
7944
7945 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
7946 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
7947 Reisner, David Coppa, David King, David Strauss, Eelco
7948 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
7949 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
7950 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
7951 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
7952 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
7953 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
7954 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
7955 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
7956 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
7957 Łukasz Stelmach, 장동준
7958
7959 CHANGES WITH 204:
7960
7961 * The Python bindings gained some minimal support for the APIs
7962 exposed by libsystemd-logind.
7963
7964 * ConditionSecurity= gained support for detecting SMACK. Since
7965 this condition already supports SELinux and AppArmor we only
7966 miss IMA for this. Patches welcome!
7967
7968 Contributions from: Karol Lewandowski, Lennart Poettering,
7969 Zbigniew Jędrzejewski-Szmek
7970
7971 CHANGES WITH 203:
7972
7973 * systemd-nspawn will now create /etc/resolv.conf if
7974 necessary, before bind-mounting the host's file onto it.
7975
7976 * systemd-nspawn will now store meta information about a
7977 container on the container's cgroup as extended attribute
7978 fields, including the root directory.
7979
7980 * The cgroup hierarchy has been reworked in many ways. All
7981 objects any of the components systemd creates in the cgroup
7982 tree are now suffixed. More specifically, user sessions are
7983 now placed in cgroups suffixed with ".session", users in
7984 cgroups suffixed with ".user", and nspawn containers in
7985 cgroups suffixed with ".nspawn". Furthermore, all cgroup
7986 names are now escaped in a simple scheme to avoid collision
7987 of userspace object names with kernel filenames. This work
7988 is preparation for making these objects relocatable in the
7989 cgroup tree, in order to allow easy resource partitioning of
7990 these objects without causing naming conflicts.
7991
7992 * systemctl list-dependencies gained the new switches
7993 --plain, --reverse, --after and --before.
7994
7995 * systemd-inhibit now shows the process name of processes that
7996 have taken an inhibitor lock.
7997
7998 * nss-myhostname will now also resolve "localhost"
7999 implicitly. This makes /etc/hosts an optional file and
8000 nicely handles that on IPv6 ::1 maps to both "localhost" and
8001 the local hostname.
8002
8003 * libsystemd-logind.so gained a new call
8004 sd_get_machine_names() to enumerate running containers and
8005 VMs (currently only supported by very new libvirt and
8006 nspawn). sd_login_monitor can now be used to watch
8007 VMs/containers coming and going.
8008
8009 * .include is not allowed recursively anymore, and only in
8010 unit files. Usually it is better to use drop-in snippets in
8011 .d/*.conf anyway, as introduced with systemd 198.
8012
8013 * systemd-analyze gained a new "critical-chain" command that
8014 determines the slowest chain of units run during system
8015 boot-up. It is very useful for tracking down where
8016 optimizing boot time is the most beneficial.
8017
8018 * systemd will no longer allow manipulating service paths in
8019 the name=systemd:/system cgroup tree using ControlGroup= in
8020 units. (But is still fine with it in all other dirs.)
8021
8022 * There's a new systemd-nspawn@.service service file that may
8023 be used to easily run nspawn containers as system
8024 services. With the container's root directory in
8025 /var/lib/container/foobar it is now sufficient to run
8026 "systemctl start systemd-nspawn@foobar.service" to boot it.
8027
8028 * systemd-cgls gained a new parameter "--machine" to list only
8029 the processes within a certain container.
8030
8031 * ConditionSecurity= now can check for "apparmor". We still
8032 are lacking checks for SMACK and IMA for this condition
8033 check though. Patches welcome!
8034
8035 * A new configuration file /etc/systemd/sleep.conf has been
8036 added that may be used to configure which kernel operation
8037 systemd is supposed to execute when "suspend", "hibernate"
8038 or "hybrid-sleep" is requested. This makes the new kernel
8039 "freeze" state accessible to the user.
8040
8041 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
8042 the passed argument if applicable.
8043
8044 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
8045 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
8046 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
8047 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
8048 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
8049 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
8050 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
8051 Jędrzejewski-Szmek
8052
8053 CHANGES WITH 202:
8054
8055 * The output of 'systemctl list-jobs' got some polishing. The
8056 '--type=' argument may now be passed more than once. A new
8057 command 'systemctl list-sockets' has been added which shows
8058 a list of kernel sockets systemd is listening on with the
8059 socket units they belong to, plus the units these socket
8060 units activate.
8061
8062 * The experimental libsystemd-bus library got substantial
8063 updates to work in conjunction with the (also experimental)
8064 kdbus kernel project. It works well enough to exchange
8065 messages with some sophistication. Note that kdbus is not
8066 ready yet, and the library is mostly an elaborate test case
8067 for now, and not installable.
8068
8069 * systemd gained a new unit 'systemd-static-nodes.service'
8070 that generates static device nodes earlier during boot, and
8071 can run in conjunction with udev.
8072
8073 * libsystemd-login gained a new call sd_pid_get_user_unit()
8074 to retrieve the user systemd unit a process is running
8075 in. This is useful for systems where systemd is used as
8076 session manager.
8077
8078 * systemd-nspawn now places all containers in the new /machine
8079 top-level cgroup directory in the name=systemd
8080 hierarchy. libvirt will soon do the same, so that we get a
8081 uniform separation of /system, /user and /machine for system
8082 services, user processes and containers/virtual
8083 machines. This new cgroup hierarchy is also useful to stick
8084 stable names to specific container instances, which can be
8085 recognized later this way (this name may be controlled
8086 via systemd-nspawn's new -M switch). libsystemd-login also
8087 gained a new call sd_pid_get_machine_name() to retrieve the
8088 name of the container/VM a specific process belongs to.
8089
8090 * bootchart can now store its data in the journal.
8091
8092 * libsystemd-journal gained a new call
8093 sd_journal_add_conjunction() for AND expressions to the
8094 matching logic. This can be used to express more complex
8095 logical expressions.
8096
8097 * journactl can now take multiple --unit= and --user-unit=
8098 switches.
8099
8100 * The cryptsetup logic now understands the "luks.key=" kernel
8101 command line switch for specifying a file to read the
8102 decryption key from. Also, if a configured key file is not
8103 found the tool will now automatically fall back to prompting
8104 the user.
8105
8106 * Python systemd.journal module was updated to wrap recently
8107 added functions from libsystemd-journal. The interface was
8108 changed to bring the low level interface in s.j._Reader
8109 closer to the C API, and the high level interface in
8110 s.j.Reader was updated to wrap and convert all data about
8111 an entry.
8112
8113 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
8114 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
8115 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
8116 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
8117 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
8118 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8119
8120 CHANGES WITH 201:
8121
8122 * journalctl --update-catalog now understands a new --root=
8123 option to operate on catalogs found in a different root
8124 directory.
8125
8126 * During shutdown after systemd has terminated all running
8127 services a final killing loop kills all remaining left-over
8128 processes. We will now print the name of these processes
8129 when we send SIGKILL to them, since this usually indicates a
8130 problem.
8131
8132 * If /etc/crypttab refers to password files stored on
8133 configured mount points automatic dependencies will now be
8134 generated to ensure the specific mount is established first
8135 before the key file is attempted to be read.
8136
8137 * 'systemctl status' will now show information about the
8138 network sockets a socket unit is listening on.
8139
8140 * 'systemctl status' will also shown information about any
8141 drop-in configuration file for units. (Drop-In configuration
8142 files in this context are files such as
8143 /etc/systemd/systemd/foobar.service.d/*.conf)
8144
8145 * systemd-cgtop now optionally shows summed up CPU times of
8146 cgroups. Press '%' while running cgtop to switch between
8147 percentage and absolute mode. This is useful to determine
8148 which cgroups use up the most CPU time over the entire
8149 runtime of the system. systemd-cgtop has also been updated
8150 to be 'pipeable' for processing with further shell tools.
8151
8152 * 'hostnamectl set-hostname' will now allow setting of FQDN
8153 hostnames.
8154
8155 * The formatting and parsing of time span values has been
8156 changed. The parser now understands fractional expressions
8157 such as "5.5h". The formatter will now output fractional
8158 expressions for all time spans under 1min, i.e. "5.123456s"
8159 rather than "5s 123ms 456us". For time spans under 1s
8160 millisecond values are shown, for those under 1ms
8161 microsecond values are shown. This should greatly improve
8162 all time-related output of systemd.
8163
8164 * libsystemd-login and libsystemd-journal gained new
8165 functions for querying the poll() events mask and poll()
8166 timeout value for integration into arbitrary event
8167 loops.
8168
8169 * localectl gained the ability to list available X11 keymaps
8170 (models, layouts, variants, options).
8171
8172 * 'systemd-analyze dot' gained the ability to filter for
8173 specific units via shell-style globs, to create smaller,
8174 more useful graphs. I.e. it is now possible to create simple
8175 graphs of all the dependencies between only target units, or
8176 of all units that Avahi has dependencies with.
8177
8178 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
8179 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
8180 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
8181 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
8182 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
8183 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
8184 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
8185
8186 CHANGES WITH 200:
8187
8188 * The boot-time readahead implementation for rotating media
8189 will now read the read-ahead data in multiple passes which
8190 consist of all read requests made in equidistant time
8191 intervals. This means instead of strictly reading read-ahead
8192 data in its physical order on disk we now try to find a
8193 middle ground between physical and access time order.
8194
8195 * /etc/os-release files gained a new BUILD_ID= field for usage
8196 on operating systems that provide continuous builds of OS
8197 images.
8198
8199 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
8200 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
8201 William Douglas, Zbigniew Jędrzejewski-Szmek
8202
8203 CHANGES WITH 199:
8204
8205 * systemd-python gained an API exposing libsystemd-daemon.
8206
8207 * The SMACK setup logic gained support for uploading CIPSO
8208 security policy.
8209
8210 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
8211 ReadOnlyDirectories= and InaccessibleDirectories= has
8212 changed. The private /tmp and /var/tmp directories are now
8213 shared by all processes of a service (which means
8214 ExecStartPre= may now leave data in /tmp that ExecStart= of
8215 the same service can still access). When a service is
8216 stopped its temporary directories are immediately deleted
8217 (normal clean-up with tmpfiles is still done in addition to
8218 this though).
8219
8220 * By default, systemd will now set a couple of sysctl
8221 variables in the kernel: the safe sysrq options are turned
8222 on, IP route verification is turned on, and source routing
8223 disabled. The recently added hardlink and softlink
8224 protection of the kernel is turned on. These settings should
8225 be reasonably safe, and good defaults for all new systems.
8226
8227 * The predictable network naming logic may now be turned off
8228 with a new kernel command line switch: net.ifnames=0.
8229
8230 * A new libsystemd-bus module has been added that implements a
8231 pretty complete D-Bus client library. For details see:
8232
8233 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
8234
8235 * journald will now explicitly flush the journal files to disk
8236 at the latest 5min after each write. The file will then also
8237 be marked offline until the next write. This should increase
8238 reliability in case of a crash. The synchronization delay
8239 can be configured via SyncIntervalSec= in journald.conf.
8240
8241 * There's a new remote-fs-setup.target unit that can be used
8242 to pull in specific services when at least one remote file
8243 system is to be mounted.
8244
8245 * There are new targets timers.target and paths.target as
8246 canonical targets to pull user timer and path units in
8247 from. This complements sockets.target with a similar
8248 purpose for socket units.
8249
8250 * libudev gained a new call udev_device_set_attribute_value()
8251 to set sysfs attributes of a device.
8252
8253 * The udev daemon now sets the default number of worker
8254 processes executed in parallel based on the number of available
8255 CPUs instead of the amount of available RAM. This is supposed
8256 to provide a more reliable default and limit a too aggressive
8257 parallelism for setups with 1000s of devices connected.
8258
8259 Contributions from: Auke Kok, Colin Walters, Cristian
8260 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
8261 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
8262 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
8263 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
8264 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
8265 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
8266 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
8267 Zbigniew Jędrzejewski-Szmek
8268
8269 CHANGES WITH 198:
8270
8271 * Configuration of unit files may now be extended via drop-in
8272 files without having to edit/override the unit files
8273 themselves. More specifically, if the administrator wants to
8274 change one value for a service file foobar.service he can
8275 now do so by dropping in a configuration snippet into
8276 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
8277 will load all these snippets and apply them on top of the
8278 main unit configuration file, possibly extending or
8279 overriding its settings. Using these drop-in snippets is
8280 generally nicer than the two earlier options for changing
8281 unit files locally: copying the files from
8282 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
8283 them there; or creating a new file in /etc/systemd/system/
8284 that incorporates the original one via ".include". Drop-in
8285 snippets into these .d/ directories can be placed in any
8286 directory systemd looks for units in, and the usual
8287 overriding semantics between /usr/lib, /etc and /run apply
8288 for them too.
8289
8290 * Most unit file settings which take lists of items can now be
8291 reset by assigning the empty string to them. For example,
8292 normally, settings such as Environment=FOO=BAR append a new
8293 environment variable assignment to the environment block,
8294 each time they are used. By assigning Environment= the empty
8295 string the environment block can be reset to empty. This is
8296 particularly useful with the .d/*.conf drop-in snippets
8297 mentioned above, since this adds the ability to reset list
8298 settings from vendor unit files via these drop-ins.
8299
8300 * systemctl gained a new "list-dependencies" command for
8301 listing the dependencies of a unit recursively.
8302
8303 * Inhibitors are now honored and listed by "systemctl
8304 suspend", "systemctl poweroff" (and similar) too, not only
8305 GNOME. These commands will also list active sessions by
8306 other users.
8307
8308 * Resource limits (as exposed by the various control group
8309 controllers) can now be controlled dynamically at runtime
8310 for all units. More specifically, you can now use a command
8311 like "systemctl set-cgroup-attr foobar.service cpu.shares
8312 2000" to alter the CPU shares a specific service gets. These
8313 settings are stored persistently on disk, and thus allow the
8314 administrator to easily adjust the resource usage of
8315 services with a few simple commands. This dynamic resource
8316 management logic is also available to other programs via the
8317 bus. Almost any kernel cgroup attribute and controller is
8318 supported.
8319
8320 * systemd-vconsole-setup will now copy all font settings to
8321 all allocated VTs, where it previously applied them only to
8322 the foreground VT.
8323
8324 * libsystemd-login gained the new sd_session_get_tty() API
8325 call.
8326
8327 * This release drops support for a few legacy or
8328 distribution-specific LSB facility names when parsing init
8329 scripts: $x-display-manager, $mail-transfer-agent,
8330 $mail-transport-agent, $mail-transfer-agent, $smtp,
8331 $null. Also, the mail-transfer-agent.target unit backing
8332 this has been removed. Distributions which want to retain
8333 compatibility with this should carry the burden for
8334 supporting this themselves and patch support for these back
8335 in, if they really need to. Also, the facilities $syslog and
8336 $local_fs are now ignored, since systemd does not support
8337 early-boot LSB init scripts anymore, and these facilities
8338 are implied anyway for normal services. syslog.target has
8339 also been removed.
8340
8341 * There are new bus calls on PID1's Manager object for
8342 cancelling jobs, and removing snapshot units. Previously,
8343 both calls were only available on the Job and Snapshot
8344 objects themselves.
8345
8346 * systemd-journal-gatewayd gained SSL support.
8347
8348 * The various "environment" files, such as /etc/locale.conf
8349 now support continuation lines with a backslash ("\") as
8350 last character in the line, similarly in style (but different)
8351 to how this is supported in shells.
8352
8353 * For normal user processes the _SYSTEMD_USER_UNIT= field is
8354 now implicitly appended to every log entry logged. systemctl
8355 has been updated to filter by this field when operating on a
8356 user systemd instance.
8357
8358 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
8359 CAP_AUDIT_CONTROL capabilities to the capabilities set for
8360 the container. This makes it easier to boot unmodified
8361 Fedora systems in a container, which however still requires
8362 audit=0 to be passed on the kernel command line. Auditing in
8363 kernel and userspace is unfortunately still too broken in
8364 context of containers, hence we recommend compiling it out
8365 of the kernel or using audit=0. Hopefully this will be fixed
8366 one day for good in the kernel.
8367
8368 * nspawn gained the new --bind= and --bind-ro= parameters to
8369 bind mount specific directories from the host into the
8370 container.
8371
8372 * nspawn will now mount its own devpts file system instance
8373 into the container, in order not to leak pty devices from
8374 the host into the container.
8375
8376 * systemd will now read the firmware boot time performance
8377 information from the EFI variables, if the used boot loader
8378 supports this, and takes it into account for boot performance
8379 analysis via "systemd-analyze". This is currently supported
8380 only in conjunction with Gummiboot, but could be supported
8381 by other boot loaders too. For details see:
8382
8383 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
8384
8385 * A new generator has been added that automatically mounts the
8386 EFI System Partition (ESP) to /boot, if that directory
8387 exists, is empty, and no other file system has been
8388 configured to be mounted there.
8389
8390 * logind will now send out PrepareForSleep(false) out
8391 unconditionally, after coming back from suspend. This may be
8392 used by applications as asynchronous notification for
8393 system resume events.
8394
8395 * "systemctl unlock-sessions" has been added, that allows
8396 unlocking the screens of all user sessions at once, similar
8397 to how "systemctl lock-sessions" already locked all users
8398 sessions. This is backed by a new D-Bus call UnlockSessions().
8399
8400 * "loginctl seat-status" will now show the master device of a
8401 seat. (i.e. the device of a seat that needs to be around for
8402 the seat to be considered available, usually the graphics
8403 card).
8404
8405 * tmpfiles gained a new "X" line type, that allows
8406 configuration of files and directories (with wildcards) that
8407 shall be excluded from automatic cleanup ("aging").
8408
8409 * udev default rules set the device node permissions now only
8410 at "add" events, and do not change them any longer with a
8411 later "change" event.
8412
8413 * The log messages for lid events and power/sleep keypresses
8414 now carry a message ID.
8415
8416 * We now have a substantially larger unit test suite, but this
8417 continues to be work in progress.
8418
8419 * udevadm hwdb gained a new --root= parameter to change the
8420 root directory to operate relative to.
8421
8422 * logind will now issue a background sync() request to the kernel
8423 early at shutdown, so that dirty buffers are flushed to disk early
8424 instead of at the last moment, in order to optimize shutdown
8425 times a little.
8426
8427 * A new bootctl tool has been added that is an interface for
8428 certain boot loader operations. This is currently a preview
8429 and is likely to be extended into a small mechanism daemon
8430 like timedated, localed, hostnamed, and can be used by
8431 graphical UIs to enumerate available boot options, and
8432 request boot into firmware operations.
8433
8434 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
8435 the rest of the package. It also has been updated to work
8436 correctly in initrds.
8437
8438 * polkit previously has been runtime optional, and is now also
8439 compile time optional via a configure switch.
8440
8441 * systemd-analyze has been reimplemented in C. Also "systemctl
8442 dot" has moved into systemd-analyze.
8443
8444 * "systemctl status" with no further parameters will now print
8445 the status of all active or failed units.
8446
8447 * Operations such as "systemctl start" can now be executed
8448 with a new mode "--irreversible" which may be used to queue
8449 operations that cannot accidentally be reversed by a later
8450 job queuing. This is by default used to make shutdown
8451 requests more robust.
8452
8453 * The Python API of systemd now gained a new module for
8454 reading journal files.
8455
8456 * A new tool kernel-install has been added that can install
8457 kernel images according to the Boot Loader Specification:
8458
8459 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
8460
8461 * Boot time console output has been improved to provide
8462 animated boot time output for hanging jobs.
8463
8464 * A new tool systemd-activate has been added which can be used
8465 to test socket activation with, directly from the command
8466 line. This should make it much easier to test and debug
8467 socket activation in daemons.
8468
8469 * journalctl gained a new "--reverse" (or -r) option to show
8470 journal output in reverse order (i.e. newest line first).
8471
8472 * journalctl gained a new "--pager-end" (or -e) option to jump
8473 to immediately jump to the end of the journal in the
8474 pager. This is only supported in conjunction with "less".
8475
8476 * journalctl gained a new "--user-unit=" option, that works
8477 similarly to "--unit=" but filters for user units rather than
8478 system units.
8479
8480 * A number of unit files to ease adoption of systemd in
8481 initrds has been added. This moves some minimal logic from
8482 the various initrd implementations into systemd proper.
8483
8484 * The journal files are now owned by a new group
8485 "systemd-journal", which exists specifically to allow access
8486 to the journal, and nothing else. Previously, we used the
8487 "adm" group for that, which however possibly covers more
8488 than just journal/log file access. This new group is now
8489 already used by systemd-journal-gatewayd to ensure this
8490 daemon gets access to the journal files and as little else
8491 as possible. Note that "make install" will also set FS ACLs
8492 up for /var/log/journal to give "adm" and "wheel" read
8493 access to it, in addition to "systemd-journal" which owns
8494 the journal files. We recommend that packaging scripts also
8495 add read access to "adm" + "wheel" to /var/log/journal, and
8496 all existing/future journal files. To normal users and
8497 administrators little changes, however packagers need to
8498 ensure to create the "systemd-journal" system group at
8499 package installation time.
8500
8501 * The systemd-journal-gatewayd now runs as unprivileged user
8502 systemd-journal-gateway:systemd-journal-gateway. Packaging
8503 scripts need to create these system user/group at
8504 installation time.
8505
8506 * timedated now exposes a new boolean property CanNTP that
8507 indicates whether a local NTP service is available or not.
8508
8509 * systemd-detect-virt will now also detect xen PVs
8510
8511 * The pstore file system is now mounted by default, if it is
8512 available.
8513
8514 * In addition to the SELinux and IMA policies we will now also
8515 load SMACK policies at early boot.
8516
8517 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
8518 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
8519 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
8520 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
8521 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
8522 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
8523 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
8524 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
8525 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
8526 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
8527 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
8528 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
8529 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
8530 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
8531
8532 CHANGES WITH 197:
8533
8534 * Timer units now support calendar time events in addition to
8535 monotonic time events. That means you can now trigger a unit
8536 based on a calendar time specification such as "Thu,Fri
8537 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
8538 or fifth day of any month of the year 2013, given that it is
8539 a thursday or friday. This brings timer event support
8540 considerably closer to cron's capabilities. For details on
8541 the supported calendar time specification language see
8542 systemd.time(7).
8543
8544 * udev now supports a number of different naming policies for
8545 network interfaces for predictable names, and a combination
8546 of these policies is now the default. Please see this wiki
8547 document for details:
8548
8549 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
8550
8551 * Auke Kok's bootchart implementation has been added to the
8552 systemd tree. It is an optional component that can graph the
8553 boot in quite some detail. It is one of the best bootchart
8554 implementations around and minimal in its code and
8555 dependencies.
8556
8557 * nss-myhostname has been integrated into the systemd source
8558 tree. nss-myhostname guarantees that the local hostname
8559 always stays resolvable via NSS. It has been a weak
8560 requirement of systemd-hostnamed since a long time, and
8561 since its code is actually trivial we decided to just
8562 include it in systemd's source tree. It can be turned off
8563 with a configure switch.
8564
8565 * The read-ahead logic is now capable of properly detecting
8566 whether a btrfs file system is on SSD or rotating media, in
8567 order to optimize the read-ahead scheme. Previously, it was
8568 only capable of detecting this on traditional file systems
8569 such as ext4.
8570
8571 * In udev, additional device properties are now read from the
8572 IAB in addition to the OUI database. Also, Bluetooth company
8573 identities are attached to the devices as well.
8574
8575 * In service files %U may be used as specifier that is
8576 replaced by the configured user name of the service.
8577
8578 * nspawn may now be invoked without a controlling TTY. This
8579 makes it suitable for invocation as its own service. This
8580 may be used to set up a simple containerized server system
8581 using only core OS tools.
8582
8583 * systemd and nspawn can now accept socket file descriptors
8584 when they are started for socket activation. This enables
8585 implementation of socket activated nspawn
8586 containers. i.e. think about autospawning an entire OS image
8587 when the first SSH or HTTP connection is received. We expect
8588 that similar functionality will also be added to libvirt-lxc
8589 eventually.
8590
8591 * journalctl will now suppress ANSI color codes when
8592 presenting log data.
8593
8594 * systemctl will no longer show control group information for
8595 a unit if the control group is empty anyway.
8596
8597 * logind can now automatically suspend/hibernate/shutdown the
8598 system on idle.
8599
8600 * /etc/machine-info and hostnamed now also expose the chassis
8601 type of the system. This can be used to determine whether
8602 the local system is a laptop, desktop, handset or
8603 tablet. This information may either be configured by the
8604 user/vendor or is automatically determined from ACPI and DMI
8605 information if possible.
8606
8607 * A number of polkit actions are now bound together with "imply"
8608 rules. This should simplify creating UIs because many actions
8609 will now authenticate similar ones as well.
8610
8611 * Unit files learnt a new condition ConditionACPower= which
8612 may be used to conditionalize a unit depending on whether an
8613 AC power source is connected or not, of whether the system
8614 is running on battery power.
8615
8616 * systemctl gained a new "is-failed" verb that may be used in
8617 shell scripts and suchlike to check whether a specific unit
8618 is in the "failed" state.
8619
8620 * The EnvironmentFile= setting in unit files now supports file
8621 globbing, and can hence be used to easily read a number of
8622 environment files at once.
8623
8624 * systemd will no longer detect and recognize specific
8625 distributions. All distribution-specific #ifdeffery has been
8626 removed, systemd is now fully generic and
8627 distribution-agnostic. Effectively, not too much is lost as
8628 a lot of the code is still accessible via explicit configure
8629 switches. However, support for some distribution specific
8630 legacy configuration file formats has been dropped. We
8631 recommend distributions to simply adopt the configuration
8632 files everybody else uses now and convert the old
8633 configuration from packaging scripts. Most distributions
8634 already did that. If that's not possible or desirable,
8635 distributions are welcome to forward port the specific
8636 pieces of code locally from the git history.
8637
8638 * When logging a message about a unit systemd will now always
8639 log the unit name in the message meta data.
8640
8641 * localectl will now also discover system locale data that is
8642 not stored in locale archives, but directly unpacked.
8643
8644 * logind will no longer unconditionally use framebuffer
8645 devices as seat masters, i.e. as devices that are required
8646 to be existing before a seat is considered preset. Instead,
8647 it will now look for all devices that are tagged as
8648 "seat-master" in udev. By default, framebuffer devices will
8649 be marked as such, but depending on local systems, other
8650 devices might be marked as well. This may be used to
8651 integrate graphics cards using closed source drivers (such
8652 as NVidia ones) more nicely into logind. Note however, that
8653 we recommend using the open source NVidia drivers instead,
8654 and no udev rules for the closed-source drivers will be
8655 shipped from us upstream.
8656
8657 Contributions from: Adam Williamson, Alessandro Crismani, Auke
8658 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
8659 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
8660 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
8661 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
8662 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
8663 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
8664 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
8665 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
8666 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
8667 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
8668 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
8669 Jędrzejewski-Szmek
8670
8671 CHANGES WITH 196:
8672
8673 * udev gained support for loading additional device properties
8674 from an indexed database that is keyed by vendor/product IDs
8675 and similar device identifiers. For the beginning this
8676 "hwdb" is populated with data from the well-known PCI and
8677 USB database, but also includes PNP, ACPI and OID data. In
8678 the longer run this indexed database shall grow into
8679 becoming the one central database for non-essential
8680 userspace device metadata. Previously, data from the PCI/USB
8681 database was only attached to select devices, since the
8682 lookup was a relatively expensive operation due to O(n) time
8683 complexity (with n being the number of entries in the
8684 database). Since this is now O(1), we decided to add in this
8685 data for all devices where this is available, by
8686 default. Note that the indexed database needs to be rebuilt
8687 when new data files are installed. To achieve this you need
8688 to update your packaging scripts to invoke "udevadm hwdb
8689 --update" after installation of hwdb data files. For
8690 RPM-based distributions we introduced the new
8691 %udev_hwdb_update macro for this purpose.
8692
8693 * The Journal gained support for the "Message Catalog", an
8694 indexed database to link up additional information with
8695 journal entries. For further details please check:
8696
8697 https://www.freedesktop.org/wiki/Software/systemd/catalog
8698
8699 The indexed message catalog database also needs to be
8700 rebuilt after installation of message catalog files. Use
8701 "journalctl --update-catalog" for this. For RPM-based
8702 distributions we introduced the %journal_catalog_update
8703 macro for this purpose.
8704
8705 * The Python Journal bindings gained support for the standard
8706 Python logging framework.
8707
8708 * The Journal API gained new functions for checking whether
8709 the underlying file system of a journal file is capable of
8710 properly reporting file change notifications, or whether
8711 applications that want to reflect journal changes "live"
8712 need to recheck journal files continuously in appropriate
8713 time intervals.
8714
8715 * It is now possible to set the "age" field for tmpfiles
8716 entries to 0, indicating that files matching this entry
8717 shall always be removed when the directories are cleaned up.
8718
8719 * coredumpctl gained a new "gdb" verb which invokes gdb
8720 right-away on the selected coredump.
8721
8722 * There's now support for "hybrid sleep" on kernels that
8723 support this, in addition to "suspend" and "hibernate". Use
8724 "systemctl hybrid-sleep" to make use of this.
8725
8726 * logind's HandleSuspendKey= setting (and related settings)
8727 now gained support for a new "lock" setting to simply
8728 request the screen lock on all local sessions, instead of
8729 actually executing a suspend or hibernation.
8730
8731 * systemd will now mount the EFI variables file system by
8732 default.
8733
8734 * Socket units now gained support for configuration of the
8735 SMACK security label.
8736
8737 * timedatectl will now output the time of the last and next
8738 daylight saving change.
8739
8740 * We dropped support for various legacy and distro-specific
8741 concepts, such as insserv, early-boot SysV services
8742 (i.e. those for non-standard runlevels such as 'b' or 'S')
8743 or ArchLinux /etc/rc.conf support. We recommend the
8744 distributions who still need support this to either continue
8745 to maintain the necessary patches downstream, or find a
8746 different solution. (Talk to us if you have questions!)
8747
8748 * Various systemd components will now bypass polkit checks for
8749 root and otherwise handle properly if polkit is not found to
8750 be around. This should fix most issues for polkit-less
8751 systems. Quite frankly this should have been this way since
8752 day one. It is absolutely our intention to make systemd work
8753 fine on polkit-less systems, and we consider it a bug if
8754 something does not work as it should if polkit is not around.
8755
8756 * For embedded systems it is now possible to build udev and
8757 systemd without blkid and/or kmod support.
8758
8759 * "systemctl switch-root" is now capable of switching root
8760 more than once. I.e. in addition to transitions from the
8761 initrd to the host OS it is now possible to transition to
8762 further OS images from the host. This is useful to implement
8763 offline updating tools.
8764
8765 * Various other additions have been made to the RPM macros
8766 shipped with systemd. Use %udev_rules_update() after
8767 installing new udev rules files. %_udevhwdbdir,
8768 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
8769 %_sysctldir are now available which resolve to the right
8770 directories for packages to place various data files in.
8771
8772 * journalctl gained the new --full switch (in addition to
8773 --all, to disable ellipsation for long messages.
8774
8775 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
8776 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
8777 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
8778 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
8779 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
8780 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
8781 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
8782 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
8783 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
8784
8785 CHANGES WITH 195:
8786
8787 * journalctl gained new --since= and --until= switches to
8788 filter by time. It also now supports nice filtering for
8789 units via --unit=/-u.
8790
8791 * Type=oneshot services may use ExecReload= and do the
8792 right thing.
8793
8794 * The journal daemon now supports time-based rotation and
8795 vacuuming, in addition to the usual disk-space based
8796 rotation.
8797
8798 * The journal will now index the available field values for
8799 each field name. This enables clients to show pretty drop
8800 downs of available match values when filtering. The bash
8801 completion of journalctl has been updated
8802 accordingly. journalctl gained a new switch -F to list all
8803 values a certain field takes in the journal database.
8804
8805 * More service events are now written as structured messages
8806 to the journal, and made recognizable via message IDs.
8807
8808 * The timedated, localed and hostnamed mini-services which
8809 previously only provided support for changing time, locale
8810 and hostname settings from graphical DEs such as GNOME now
8811 also have a minimal (but very useful) text-based client
8812 utility each. This is probably the nicest way to changing
8813 these settings from the command line now, especially since
8814 it lists available options and is fully integrated with bash
8815 completion.
8816
8817 * There's now a new tool "systemd-coredumpctl" to list and
8818 extract coredumps from the journal.
8819
8820 * We now install a README each in /var/log/ and
8821 /etc/rc.d/init.d explaining where the system logs and init
8822 scripts went. This hopefully should help folks who go to
8823 that dirs and look into the otherwise now empty void and
8824 scratch their heads.
8825
8826 * When user-services are invoked (by systemd --user) the
8827 $MANAGERPID env var is set to the PID of systemd.
8828
8829 * SIGRTMIN+24 when sent to a --user instance will now result
8830 in immediate termination of systemd.
8831
8832 * gatewayd received numerous feature additions such as a
8833 "follow" mode, for live syncing and filtering.
8834
8835 * browse.html now allows filtering and showing detailed
8836 information on specific entries. Keyboard navigation and
8837 mouse screen support has been added.
8838
8839 * gatewayd/journalctl now supports HTML5/JSON
8840 Server-Sent-Events as output.
8841
8842 * The SysV init script compatibility logic will now
8843 heuristically determine whether a script supports the
8844 "reload" verb, and only then make this available as
8845 "systemctl reload".
8846
8847 * "systemctl status --follow" has been removed, use "journalctl
8848 -u" instead.
8849
8850 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
8851 have been removed since they are hardly useful to be
8852 configured.
8853
8854 * And I'd like to take the opportunity to specifically mention
8855 Zbigniew for his great contributions. Zbigniew, you rock!
8856
8857 Contributions from: Andrew Eikum, Christian Hesse, Colin
8858 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
8859 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
8860 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
8861 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
8862 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
8863 Jędrzejewski-Szmek, Сковорода Никита Андреевич
8864
8865 CHANGES WITH 194:
8866
8867 * If /etc/vconsole.conf is non-existent or empty we will no
8868 longer load any console font or key map at boot by
8869 default. Instead the kernel defaults will be left
8870 intact. This is definitely the right thing to do, as no
8871 configuration should mean no configuration, and hard-coding
8872 font names that are different on all archs is probably a bad
8873 idea. Also, the kernel default key map and font should be
8874 good enough for most cases anyway, and mostly identical to
8875 the userspace fonts/key maps we previously overloaded them
8876 with. If distributions want to continue to default to a
8877 non-kernel font or key map they should ship a default
8878 /etc/vconsole.conf with the appropriate contents.
8879
8880 Contributions from: Colin Walters, Daniel J Walsh, Dave
8881 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
8882 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8883
8884 CHANGES WITH 193:
8885
8886 * journalctl gained a new --cursor= switch to show entries
8887 starting from the specified location in the journal.
8888
8889 * We now enforce a size limit on journal entry fields exported
8890 with "-o json" in journalctl. Fields larger than 4K will be
8891 assigned null. This can be turned off with --all.
8892
8893 * An (optional) journal gateway daemon is now available as
8894 "systemd-journal-gatewayd.service". This service provides
8895 access to the journal via HTTP and JSON. This functionality
8896 will be used to implement live log synchronization in both
8897 pull and push modes, but has various other users too, such
8898 as easy log access for debugging of embedded devices. Right
8899 now it is already useful to retrieve the journal via HTTP:
8900
8901 # systemctl start systemd-journal-gatewayd.service
8902 # wget http://localhost:19531/entries
8903
8904 This will download the journal contents in a
8905 /var/log/messages compatible format. The same as JSON:
8906
8907 # curl -H"Accept: application/json" http://localhost:19531/entries
8908
8909 This service is also accessible via a web browser where a
8910 single static HTML5 app is served that uses the JSON logic
8911 to enable the user to do some basic browsing of the
8912 journal. This will be extended later on. Here's an example
8913 screenshot of this app in its current state:
8914
8915 http://0pointer.de/public/journal-gatewayd
8916
8917 Contributions from: Kay Sievers, Lennart Poettering, Robert
8918 Milasan, Tom Gundersen
8919
8920 CHANGES WITH 192:
8921
8922 * The bash completion logic is now available for journalctl
8923 too.
8924
8925 * We do not mount the "cpuset" controller anymore together with
8926 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
8927 started if no parameters are assigned to it. "cpuset" hence
8928 broke code that assumed it could create "cpu" groups and
8929 just start them.
8930
8931 * journalctl -f will now subscribe to terminal size changes,
8932 and line break accordingly.
8933
8934 Contributions from: Dave Reisner, Kay Sievers, Lennart
8935 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
8936
8937 CHANGES WITH 191:
8938
8939 * nspawn will now create a symlink /etc/localtime in the
8940 container environment, copying the host's timezone
8941 setting. Previously this has been done via a bind mount, but
8942 since symlinks cannot be bind mounted this has now been
8943 changed to create/update the appropriate symlink.
8944
8945 * journalctl -n's line number argument is now optional, and
8946 will default to 10 if omitted.
8947
8948 * journald will now log the maximum size the journal files may
8949 take up on disk. This is particularly useful if the default
8950 built-in logic of determining this parameter from the file
8951 system size is used. Use "systemctl status
8952 systemd-journald.service" to see this information.
8953
8954 * The multi-seat X wrapper tool has been stripped down. As X
8955 is now capable of enumerating graphics devices via udev in a
8956 seat-aware way the wrapper is not strictly necessary
8957 anymore. A stripped down temporary stop-gap is still shipped
8958 until the upstream display managers have been updated to
8959 fully support the new X logic. Expect this wrapper to be
8960 removed entirely in one of the next releases.
8961
8962 * HandleSleepKey= in logind.conf has been split up into
8963 HandleSuspendKey= and HandleHibernateKey=. The old setting
8964 is not available anymore. X11 and the kernel are
8965 distinguishing between these keys and we should too. This
8966 also means the inhibition lock for these keys has been split
8967 into two.
8968
8969 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
8970 Poettering, Lukas Nykryn, Václav Pavlín
8971
8972 CHANGES WITH 190:
8973
8974 * Whenever a unit changes state we will now log this to the
8975 journal and show along the unit's own log output in
8976 "systemctl status".
8977
8978 * ConditionPathIsMountPoint= can now properly detect bind
8979 mount points too. (Previously, a bind mount of one file
8980 system to another place in the same file system could not be
8981 detected as mount, since they shared struct stat's st_dev
8982 field.)
8983
8984 * We will now mount the cgroup controllers cpu, cpuacct,
8985 cpuset and the controllers net_cls, net_prio together by
8986 default.
8987
8988 * nspawn containers will now have a virtualized boot
8989 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
8990 over with a randomized ID at container initialization). This
8991 has the effect of making "journalctl -b" do the right thing
8992 in a container.
8993
8994 * The JSON output journal serialization has been updated not
8995 to generate "endless" list objects anymore, but rather one
8996 JSON object per line. This is more in line how most JSON
8997 parsers expect JSON objects. The new output mode
8998 "json-pretty" has been added to provide similar output, but
8999 neatly aligned for readability by humans.
9000
9001 * We dropped all explicit sync() invocations in the shutdown
9002 code. The kernel does this implicitly anyway in the kernel
9003 reboot() syscall. halt(8)'s -n option is now a compatibility
9004 no-op.
9005
9006 * We now support virtualized reboot() in containers, as
9007 supported by newer kernels. We will fall back to exit() if
9008 CAP_SYS_REBOOT is not available to the container. Also,
9009 nspawn makes use of this now and will actually reboot the
9010 container if the containerized OS asks for that.
9011
9012 * journalctl will only show local log output by default
9013 now. Use --merge (-m) to show remote log output, too.
9014
9015 * libsystemd-journal gained the new sd_journal_get_usage()
9016 call to determine the current disk usage of all journal
9017 files. This is exposed in the new "journalctl --disk-usage"
9018 command.
9019
9020 * journald gained a new configuration setting SplitMode= in
9021 journald.conf which may be used to control how user journals
9022 are split off. See journald.conf(5) for details.
9023
9024 * A new condition type ConditionFileNotEmpty= has been added.
9025
9026 * tmpfiles' "w" lines now support file globbing, to write
9027 multiple files at once.
9028
9029 * We added Python bindings for the journal submission
9030 APIs. More Python APIs for a number of selected APIs will
9031 likely follow. Note that we intend to add native bindings
9032 only for the Python language, as we consider it common
9033 enough to deserve bindings shipped within systemd. There are
9034 various projects outside of systemd that provide bindings
9035 for languages such as PHP or Lua.
9036
9037 * Many conditions will now resolve specifiers such as %i. In
9038 addition, PathChanged= and related directives of .path units
9039 now support specifiers as well.
9040
9041 * There's now a new RPM macro definition for the system preset
9042 dir: %_presetdir.
9043
9044 * journald will now warn if it ca not forward a message to the
9045 syslog daemon because its socket is full.
9046
9047 * timedated will no longer write or process /etc/timezone,
9048 except on Debian. As we do not support late mounted /usr
9049 anymore /etc/localtime always being a symlink is now safe,
9050 and hence the information in /etc/timezone is not necessary
9051 anymore.
9052
9053 * logind will now always reserve one VT for a text getty (VT6
9054 by default). Previously if more than 6 X sessions where
9055 started they took up all the VTs with auto-spawned gettys,
9056 so that no text gettys were available anymore.
9057
9058 * udev will now automatically inform the btrfs kernel logic
9059 about btrfs RAID components showing up. This should make
9060 simple hotplug based btrfs RAID assembly work.
9061
9062 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
9063 (but not for its children which will stay at the kernel
9064 default). This should allow setups with a lot more listening
9065 sockets.
9066
9067 * systemd will now always pass the configured timezone to the
9068 kernel at boot. timedated will do the same when the timezone
9069 is changed.
9070
9071 * logind's inhibition logic has been updated. By default,
9072 logind will now handle the lid switch, the power and sleep
9073 keys all the time, even in graphical sessions. If DEs want
9074 to handle these events on their own they should take the new
9075 handle-power-key, handle-sleep-key and handle-lid-switch
9076 inhibitors during their runtime. A simple way to achieve
9077 that is to invoke the DE wrapped in an invocation of:
9078
9079 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
9080
9081 * Access to unit operations is now checked via SELinux taking
9082 the unit file label and client process label into account.
9083
9084 * systemd will now notify the administrator in the journal
9085 when he over-mounts a non-empty directory.
9086
9087 * There are new specifiers that are resolved in unit files,
9088 for the host name (%H), the machine ID (%m) and the boot ID
9089 (%b).
9090
9091 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
9092 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
9093 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
9094 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
9095 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
9096 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
9097 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
9098
9099 CHANGES WITH 189:
9100
9101 * Support for reading structured kernel messages from
9102 /dev/kmsg has now been added and is enabled by default.
9103
9104 * Support for reading kernel messages from /proc/kmsg has now
9105 been removed. If you want kernel messages in the journal
9106 make sure to run a recent kernel (>= 3.5) that supports
9107 reading structured messages from /dev/kmsg (see
9108 above). /proc/kmsg is now exclusive property of classic
9109 syslog daemons again.
9110
9111 * The libudev API gained the new
9112 udev_device_new_from_device_id() call.
9113
9114 * The logic for file system namespace (ReadOnlyDirectory=,
9115 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
9116 require pivot_root() anymore. This means fewer temporary
9117 directories are created below /tmp for this feature.
9118
9119 * nspawn containers will now see and receive all submounts
9120 made on the host OS below the root file system of the
9121 container.
9122
9123 * Forward Secure Sealing is now supported for Journal files,
9124 which provide cryptographical sealing of journal files so
9125 that attackers cannot alter log history anymore without this
9126 being detectable. Lennart will soon post a blog story about
9127 this explaining it in more detail.
9128
9129 * There are two new service settings RestartPreventExitStatus=
9130 and SuccessExitStatus= which allow configuration of exit
9131 status (exit code or signal) which will be excepted from the
9132 restart logic, resp. consider successful.
9133
9134 * journalctl gained the new --verify switch that can be used
9135 to check the integrity of the structure of journal files and
9136 (if Forward Secure Sealing is enabled) the contents of
9137 journal files.
9138
9139 * nspawn containers will now be run with /dev/stdin, /dev/fd/
9140 and similar symlinks pre-created. This makes running shells
9141 as container init process a lot more fun.
9142
9143 * The fstab support can now handle PARTUUID= and PARTLABEL=
9144 entries.
9145
9146 * A new ConditionHost= condition has been added to match
9147 against the hostname (with globs) and machine ID. This is
9148 useful for clusters where a single OS image is used to
9149 provision a large number of hosts which shall run slightly
9150 different sets of services.
9151
9152 * Services which hit the restart limit will now be placed in a
9153 failure state.
9154
9155 Contributions from: Bertram Poettering, Dave Reisner, Huang
9156 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
9157 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
9158
9159 CHANGES WITH 188:
9160
9161 * When running in --user mode systemd will now become a
9162 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
9163 tree a lot more organized.
9164
9165 * A new PartOf= unit dependency type has been introduced that
9166 may be used to group services in a natural way.
9167
9168 * "systemctl enable" may now be used to enable instances of
9169 services.
9170
9171 * journalctl now prints error log levels in red, and
9172 warning/notice log levels in bright white. It also supports
9173 filtering by log level now.
9174
9175 * cgtop gained a new -n switch (similar to top), to configure
9176 the maximum number of iterations to run for. It also gained
9177 -b, to run in batch mode (accepting no input).
9178
9179 * The suffix ".service" may now be omitted on most systemctl
9180 command lines involving service unit names.
9181
9182 * There's a new bus call in logind to lock all sessions, as
9183 well as a loginctl verb for it "lock-sessions".
9184
9185 * libsystemd-logind.so gained a new call sd_journal_perror()
9186 that works similar to libc perror() but logs to the journal
9187 and encodes structured information about the error number.
9188
9189 * /etc/crypttab entries now understand the new keyfile-size=
9190 option.
9191
9192 * shutdown(8) now can send a (configurable) wall message when
9193 a shutdown is cancelled.
9194
9195 * The mount propagation mode for the root file system will now
9196 default to "shared", which is useful to make containers work
9197 nicely out-of-the-box so that they receive new mounts from
9198 the host. This can be undone locally by running "mount
9199 --make-rprivate /" if needed.
9200
9201 * The prefdm.service file has been removed. Distributions
9202 should maintain this unit downstream if they intend to keep
9203 it around. However, we recommend writing normal unit files
9204 for display managers instead.
9205
9206 * Since systemd is a crucial part of the OS we will now
9207 default to a number of compiler switches that improve
9208 security (hardening) such as read-only relocations, stack
9209 protection, and suchlike.
9210
9211 * The TimeoutSec= setting for services is now split into
9212 TimeoutStartSec= and TimeoutStopSec= to allow configuration
9213 of individual time outs for the start and the stop phase of
9214 the service.
9215
9216 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
9217 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
9218 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
9219 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
9220 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
9221 Gundersen, Zbigniew Jędrzejewski-Szmek
9222
9223 CHANGES WITH 187:
9224
9225 * The journal and id128 C APIs are now fully documented as man
9226 pages.
9227
9228 * Extra safety checks have been added when transitioning from
9229 the initial RAM disk to the main system to avoid accidental
9230 data loss.
9231
9232 * /etc/crypttab entries now understand the new keyfile-offset=
9233 option.
9234
9235 * systemctl -t can now be used to filter by unit load state.
9236
9237 * The journal C API gained the new sd_journal_wait() call to
9238 make writing synchronous journal clients easier.
9239
9240 * journalctl gained the new -D switch to show journals from a
9241 specific directory.
9242
9243 * journalctl now displays a special marker between log
9244 messages of two different boots.
9245
9246 * The journal is now explicitly flushed to /var via a service
9247 systemd-journal-flush.service, rather than implicitly simply
9248 by seeing /var/log/journal to be writable.
9249
9250 * journalctl (and the journal C APIs) can now match for much
9251 more complex expressions, with alternatives and
9252 disjunctions.
9253
9254 * When transitioning from the initial RAM disk to the main
9255 system we will now kill all processes in a killing spree to
9256 ensure no processes stay around by accident.
9257
9258 * Three new specifiers may be used in unit files: %u, %h, %s
9259 resolve to the user name, user home directory resp. user
9260 shell. This is useful for running systemd user instances.
9261
9262 * We now automatically rotate journal files if their data
9263 object hash table gets a fill level > 75%. We also size the
9264 hash table based on the configured maximum file size. This
9265 together should lower hash collisions drastically and thus
9266 speed things up a bit.
9267
9268 * journalctl gained the new "--header" switch to introspect
9269 header data of journal files.
9270
9271 * A new setting SystemCallFilters= has been added to services
9272 which may be used to apply blacklists or whitelists to
9273 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
9274
9275 * nspawn gained a new --link-journal= switch (and quicker: -j)
9276 to link the container journal with the host. This makes it
9277 very easy to centralize log viewing on the host for all
9278 guests while still keeping the journal files separated.
9279
9280 * Many bugfixes and optimizations
9281
9282 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
9283 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
9284 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
9285 Jędrzejewski-Szmek
9286
9287 CHANGES WITH 186:
9288
9289 * Several tools now understand kernel command line arguments,
9290 which are only read when run in an initial RAM disk. They
9291 usually follow closely their normal counterparts, but are
9292 prefixed with rd.
9293
9294 * There's a new tool to analyze the readahead files that are
9295 automatically generated at boot. Use:
9296
9297 /usr/lib/systemd/systemd-readahead analyze /.readahead
9298
9299 * We now provide an early debug shell on tty9 if this enabled. Use:
9300
9301 systemctl enable debug-shell.service
9302
9303 * All plymouth related units have been moved into the Plymouth
9304 package. Please make sure to upgrade your Plymouth version
9305 as well.
9306
9307 * systemd-tmpfiles now supports getting passed the basename of
9308 a configuration file only, in which case it will look for it
9309 in all appropriate directories automatically.
9310
9311 * udevadm info now takes a /dev or /sys path as argument, and
9312 does the right thing. Example:
9313
9314 udevadm info /dev/sda
9315 udevadm info /sys/class/block/sda
9316
9317 * systemctl now prints a warning if a unit is stopped but a
9318 unit that might trigger it continues to run. Example: a
9319 service is stopped but the socket that activates it is left
9320 running.
9321
9322 * "systemctl status" will now mention if the log output was
9323 shortened due to rotation since a service has been started.
9324
9325 * The journal API now exposes functions to determine the
9326 "cutoff" times due to rotation.
9327
9328 * journald now understands SIGUSR1 and SIGUSR2 for triggering
9329 immediately flushing of runtime logs to /var if possible,
9330 resp. for triggering immediate rotation of the journal
9331 files.
9332
9333 * It is now considered an error if a service is attempted to
9334 be stopped that is not loaded.
9335
9336 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
9337
9338 * systemd-analyze now supports Python 3
9339
9340 * tmpfiles now supports cleaning up directories via aging
9341 where the first level dirs are always kept around but
9342 directories beneath it automatically aged. This is enabled
9343 by prefixing the age field with '~'.
9344
9345 * Seat objects now expose CanGraphical, CanTTY properties
9346 which is required to deal with very fast bootups where the
9347 display manager might be running before the graphics drivers
9348 completed initialization.
9349
9350 * Seat objects now expose a State property.
9351
9352 * We now include RPM macros for service enabling/disabling
9353 based on the preset logic. We recommend RPM based
9354 distributions to make use of these macros if possible. This
9355 makes it simpler to reuse RPM spec files across
9356 distributions.
9357
9358 * We now make sure that the collected systemd unit name is
9359 always valid when services log to the journal via
9360 STDOUT/STDERR.
9361
9362 * There's a new man page kernel-command-line(7) detailing all
9363 command line options we understand.
9364
9365 * The fstab generator may now be disabled at boot by passing
9366 fstab=0 on the kernel command line.
9367
9368 * A new kernel command line option modules-load= is now understood
9369 to load a specific kernel module statically, early at boot.
9370
9371 * Unit names specified on the systemctl command line are now
9372 automatically escaped as needed. Also, if file system or
9373 device paths are specified they are automatically turned
9374 into the appropriate mount or device unit names. Example:
9375
9376 systemctl status /home
9377 systemctl status /dev/sda
9378
9379 * The SysVConsole= configuration option has been removed from
9380 system.conf parsing.
9381
9382 * The SysV search path is no longer exported on the D-Bus
9383 Manager object.
9384
9385 * The Names= option has been removed from unit file parsing.
9386
9387 * There's a new man page bootup(7) detailing the boot process.
9388
9389 * Every unit and every generator we ship with systemd now
9390 comes with full documentation. The self-explanatory boot is
9391 complete.
9392
9393 * A couple of services gained "systemd-" prefixes in their
9394 name if they wrap systemd code, rather than only external
9395 code. Among them fsck@.service which is now
9396 systemd-fsck@.service.
9397
9398 * The HaveWatchdog property has been removed from the D-Bus
9399 Manager object.
9400
9401 * systemd.confirm_spawn= on the kernel command line should now
9402 work sensibly.
9403
9404 * There's a new man page crypttab(5) which details all options
9405 we actually understand.
9406
9407 * systemd-nspawn gained a new --capability= switch to pass
9408 additional capabilities to the container.
9409
9410 * timedated will now read known NTP implementation unit names
9411 from /usr/lib/systemd/ntp-units.d/*.list,
9412 systemd-timedated-ntp.target has been removed.
9413
9414 * journalctl gained a new switch "-b" that lists log data of
9415 the current boot only.
9416
9417 * The notify socket is in the abstract namespace again, in
9418 order to support daemons which chroot() at start-up.
9419
9420 * There is a new Storage= configuration option for journald
9421 which allows configuration of where log data should go. This
9422 also provides a way to disable journal logging entirely, so
9423 that data collected is only forwarded to the console, the
9424 kernel log buffer or another syslog implementation.
9425
9426 * Many bugfixes and optimizations
9427
9428 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
9429 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
9430 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
9431 Shawn Landden, Tom Gundersen
9432
9433 CHANGES WITH 185:
9434
9435 * "systemctl help <unit>" now shows the man page if one is
9436 available.
9437
9438 * Several new man pages have been added.
9439
9440 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
9441 MaxLevelConsole= can now be specified in
9442 journald.conf. These options allow reducing the amount of
9443 data stored on disk or forwarded by the log level.
9444
9445 * TimerSlackNSec= can now be specified in system.conf for
9446 PID1. This allows system-wide power savings.
9447
9448 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
9449 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
9450 Matthias Clasen
9451
9452 CHANGES WITH 184:
9453
9454 * logind is now capable of (optionally) handling power and
9455 sleep keys as well as the lid switch.
9456
9457 * journalctl now understands the syntax "journalctl
9458 /usr/bin/avahi-daemon" to get all log output of a specific
9459 daemon.
9460
9461 * CapabilityBoundingSet= in system.conf now also influences
9462 the capability bound set of usermode helpers of the kernel.
9463
9464 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
9465 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
9466 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
9467 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
9468
9469 CHANGES WITH 183:
9470
9471 * Note that we skipped 139 releases here in order to set the
9472 new version to something that is greater than both udev's
9473 and systemd's most recent version number.
9474
9475 * udev: all udev sources are merged into the systemd source tree now.
9476 All future udev development will happen in the systemd tree. It
9477 is still fully supported to use the udev daemon and tools without
9478 systemd running, like in initramfs or other init systems. Building
9479 udev though, will require the *build* of the systemd tree, but
9480 udev can be properly *run* without systemd.
9481
9482 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
9483 should be used to create dead device nodes as workarounds for broken
9484 subsystems.
9485
9486 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
9487 no longer supported. udev_monitor_new_from_netlink() needs to be
9488 used to subscribe to events.
9489
9490 * udev: when udevd is started by systemd, processes which are left
9491 behind by forking them off of udev rules, are unconditionally cleaned
9492 up and killed now after the event handling has finished. Services or
9493 daemons must be started as systemd services. Services can be
9494 pulled-in by udev to get started, but they can no longer be directly
9495 forked by udev rules.
9496
9497 * udev: the daemon binary is called systemd-udevd now and installed
9498 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
9499 to adapt to that, create symlink, or rename the binary after building
9500 it.
9501
9502 * libudev no longer provides these symbols:
9503 udev_monitor_from_socket()
9504 udev_queue_get_failed_list_entry()
9505 udev_get_{dev,sys,run}_path()
9506 The versions number was bumped and symbol versioning introduced.
9507
9508 * systemd-loginctl and systemd-journalctl have been renamed
9509 to loginctl and journalctl to match systemctl.
9510
9511 * The config files: /etc/systemd/systemd-logind.conf and
9512 /etc/systemd/systemd-journald.conf have been renamed to
9513 logind.conf and journald.conf. Package updates should rename
9514 the files to the new names on upgrade.
9515
9516 * For almost all files the license is now LGPL2.1+, changed
9517 from the previous GPL2.0+. Exceptions are some minor stuff
9518 of udev (which will be changed to LGPL2.1 eventually, too),
9519 and the MIT licensed sd-daemon.[ch] library that is suitable
9520 to be used as drop-in files.
9521
9522 * systemd and logind now handle system sleep states, in
9523 particular suspending and hibernating.
9524
9525 * logind now implements a sleep/shutdown/idle inhibiting logic
9526 suitable for a variety of uses. Soonishly Lennart will blog
9527 about this in more detail.
9528
9529 * var-run.mount and var-lock.mount are no longer provided
9530 (which previously bind mounted these directories to their new
9531 places). Distributions which have not converted these
9532 directories to symlinks should consider stealing these files
9533 from git history and add them downstream.
9534
9535 * We introduced the Documentation= field for units and added
9536 this to all our shipped units. This is useful to make it
9537 easier to explore the boot and the purpose of the various
9538 units.
9539
9540 * All smaller setup units (such as
9541 systemd-vconsole-setup.service) now detect properly if they
9542 are run in a container and are skipped when
9543 appropriate. This guarantees an entirely noise-free boot in
9544 Linux container environments such as systemd-nspawn.
9545
9546 * A framework for implementing offline system updates is now
9547 integrated, for details see:
9548 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
9549
9550 * A new service type Type=idle is available now which helps us
9551 avoiding ugly interleaving of getty output and boot status
9552 messages.
9553
9554 * There's now a system-wide CapabilityBoundingSet= option to
9555 globally reduce the set of capabilities for the
9556 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
9557 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
9558 even CAP_NET_ADMIN system-wide for secure systems.
9559
9560 * There are now system-wide DefaultLimitXXX= options to
9561 globally change the defaults of the various resource limits
9562 for all units started by PID 1.
9563
9564 * Harald Hoyer's systemd test suite has been integrated into
9565 systemd which allows easy testing of systemd builds in qemu
9566 and nspawn. (This is really awesome! Ask us for details!)
9567
9568 * The fstab parser is now implemented as generator, not inside
9569 of PID 1 anymore.
9570
9571 * systemctl will now warn you if .mount units generated from
9572 /etc/fstab are out of date due to changes in fstab that
9573 have not been read by systemd yet.
9574
9575 * systemd is now suitable for usage in initrds. Dracut has
9576 already been updated to make use of this. With this in place
9577 initrds get a slight bit faster but primarily are much
9578 easier to introspect and debug since "systemctl status" in
9579 the host system can be used to introspect initrd services,
9580 and the journal from the initrd is kept around too.
9581
9582 * systemd-delta has been added, a tool to explore differences
9583 between user/admin configuration and vendor defaults.
9584
9585 * PrivateTmp= now affects both /tmp and /var/tmp.
9586
9587 * Boot time status messages are now much prettier and feature
9588 proper english language. Booting up systemd has never been
9589 so sexy.
9590
9591 * Read-ahead pack files now include the inode number of all
9592 files to pre-cache. When the inode changes the pre-caching
9593 is not attempted. This should be nicer to deal with updated
9594 packages which might result in changes of read-ahead
9595 patterns.
9596
9597 * We now temporaritly lower the kernel's read_ahead_kb variable
9598 when collecting read-ahead data to ensure the kernel's
9599 built-in read-ahead does not add noise to our measurements
9600 of necessary blocks to pre-cache.
9601
9602 * There's now RequiresMountsFor= to add automatic dependencies
9603 for all mounts necessary for a specific file system path.
9604
9605 * MountAuto= and SwapAuto= have been removed from
9606 system.conf. Mounting file systems at boot has to take place
9607 in systemd now.
9608
9609 * nspawn now learned a new switch --uuid= to set the machine
9610 ID on the command line.
9611
9612 * nspawn now learned the -b switch to automatically search
9613 for an init system.
9614
9615 * vt102 is now the default TERM for serial TTYs, upgraded from
9616 vt100.
9617
9618 * systemd-logind now works on VT-less systems.
9619
9620 * The build tree has been reorganized. The individual
9621 components now have directories of their own.
9622
9623 * A new condition type ConditionPathIsReadWrite= is now available.
9624
9625 * nspawn learned the new -C switch to create cgroups for the
9626 container in other hierarchies.
9627
9628 * We now have support for hardware watchdogs, configurable in
9629 system.conf.
9630
9631 * The scheduled shutdown logic now has a public API.
9632
9633 * We now mount /tmp as tmpfs by default, but this can be
9634 masked and /etc/fstab can override it.
9635
9636 * Since udisks does not make use of /media anymore we are not
9637 mounting a tmpfs on it anymore.
9638
9639 * journalctl gained a new --local switch to only interleave
9640 locally generated journal files.
9641
9642 * We can now load the IMA policy at boot automatically.
9643
9644 * The GTK tools have been split off into a systemd-ui.
9645
9646 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
9647 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
9648 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
9649 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
9650 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
9651 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
9652 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
9653 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
9654 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
9655 Gundersen
9656
9657 CHANGES WITH 44:
9658
9659 * This is mostly a bugfix release
9660
9661 * Support optional initialization of the machine ID from the
9662 KVM or container configured UUID.
9663
9664 * Support immediate reboots with "systemctl reboot -ff"
9665
9666 * Show /etc/os-release data in systemd-analyze output
9667
9668 * Many bugfixes for the journal, including endianness fixes and
9669 ensuring that disk space enforcement works
9670
9671 * sd-login.h is C++ compatible again
9672
9673 * Extend the /etc/os-release format on request of the Debian
9674 folks
9675
9676 * We now refuse non-UTF8 strings used in various configuration
9677 and unit files. This is done to ensure we do not pass invalid
9678 data over D-Bus or expose it elsewhere.
9679
9680 * Register Mimo USB Screens as suitable for automatic seat
9681 configuration
9682
9683 * Read SELinux client context from journal clients in a race
9684 free fashion
9685
9686 * Reorder configuration file lookup order. /etc now always
9687 overrides /run in order to allow the administrator to always
9688 and unconditionally override vendor-supplied or
9689 automatically generated data.
9690
9691 * The various user visible bits of the journal now have man
9692 pages. We still lack man pages for the journal API calls
9693 however.
9694
9695 * We now ship all man pages in HTML format again in the
9696 tarball.
9697
9698 Contributions from: Dave Reisner, Dirk Eibach, Frederic
9699 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
9700 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
9701 Reding
9702
9703 CHANGES WITH 43:
9704
9705 * This is mostly a bugfix release
9706
9707 * systems lacking /etc/os-release are no longer supported.
9708
9709 * Various functionality updates to libsystemd-login.so
9710
9711 * Track class of PAM logins to distinguish greeters from
9712 normal user logins.
9713
9714 Contributions from: Kay Sievers, Lennart Poettering, Michael
9715 Biebl
9716
9717 CHANGES WITH 42:
9718
9719 * This is an important bugfix release for v41.
9720
9721 * Building man pages is now optional which should be useful
9722 for those building systemd from git but unwilling to install
9723 xsltproc.
9724
9725 * Watchdog support for supervising services is now usable. In
9726 a future release support for hardware watchdogs
9727 (i.e. /dev/watchdog) will be added building on this.
9728
9729 * Service start rate limiting is now configurable and can be
9730 turned off per service. When a start rate limit is hit a
9731 reboot can automatically be triggered.
9732
9733 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
9734
9735 Contributions from: Benjamin Franzke, Bill Nottingham,
9736 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
9737 Schmidt, Michał Górny, Piotr Drąg
9738
9739 CHANGES WITH 41:
9740
9741 * The systemd binary is installed /usr/lib/systemd/systemd now;
9742 An existing /sbin/init symlink needs to be adapted with the
9743 package update.
9744
9745 * The code that loads kernel modules has been ported to invoke
9746 libkmod directly, instead of modprobe. This means we do not
9747 support systems with module-init-tools anymore.
9748
9749 * Watchdog support is now already useful, but still not
9750 complete.
9751
9752 * A new kernel command line option systemd.setenv= is
9753 understood to set system wide environment variables
9754 dynamically at boot.
9755
9756 * We now limit the set of capabilities of systemd-journald.
9757
9758 * We now set SIGPIPE to ignore by default, since it only is
9759 useful in shell pipelines, and has little use in general
9760 code. This can be disabled with IgnoreSIPIPE=no in unit
9761 files.
9762
9763 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
9764 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
9765 William Douglas
9766
9767 CHANGES WITH 40:
9768
9769 * This is mostly a bugfix release
9770
9771 * We now expose the reason why a service failed in the
9772 "Result" D-Bus property.
9773
9774 * Rudimentary service watchdog support (will be completed over
9775 the next few releases.)
9776
9777 * When systemd forks off in order execute some service we will
9778 now immediately changes its argv[0] to reflect which process
9779 it will execute. This is useful to minimize the time window
9780 with a generic argv[0], which makes bootcharts more useful
9781
9782 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
9783 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
9784 Mike Kazantsev, Ray Strode
9785
9786 CHANGES WITH 39:
9787
9788 * This is mostly a test release, but incorporates many
9789 bugfixes.
9790
9791 * New systemd-cgtop tool to show control groups by their
9792 resource usage.
9793
9794 * Linking against libacl for ACLs is optional again. If
9795 disabled, support tracking device access for active logins
9796 goes becomes unavailable, and so does access to the user
9797 journals by the respective users.
9798
9799 * If a group "adm" exists, journal files are automatically
9800 owned by them, thus allow members of this group full access
9801 to the system journal as well as all user journals.
9802
9803 * The journal now stores the SELinux context of the logging
9804 client for all entries.
9805
9806 * Add C++ inclusion guards to all public headers
9807
9808 * New output mode "cat" in the journal to print only text
9809 messages, without any meta data like date or time.
9810
9811 * Include tiny X server wrapper as a temporary stop-gap to
9812 teach XOrg udev display enumeration. This is used by display
9813 managers such as gdm, and will go away as soon as XOrg
9814 learned native udev hotplugging for display devices.
9815
9816 * Add new systemd-cat tool for executing arbitrary programs
9817 with STDERR/STDOUT connected to the journal. Can also act as
9818 BSD logger replacement, and does so by default.
9819
9820 * Optionally store all locally generated coredumps in the
9821 journal along with meta data.
9822
9823 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
9824 writing short strings to files (for usage for /sys), and for
9825 creating symlinks, character and block device nodes.
9826
9827 * New unit file option ControlGroupPersistent= to make cgroups
9828 persistent, following the mechanisms outlined in
9829 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
9830
9831 * Support multiple local RTCs in a sane way
9832
9833 * No longer monopolize IO when replaying readahead data on
9834 rotating disks, since we might starve non-file-system IO to
9835 death, since fanotify() will not see accesses done by blkid,
9836 or fsck.
9837
9838 * Do not show kernel threads in systemd-cgls anymore, unless
9839 requested with new -k switch.
9840
9841 Contributions from: Dan Horák, Kay Sievers, Lennart
9842 Poettering, Michal Schmidt
9843
9844 CHANGES WITH 38:
9845
9846 * This is mostly a test release, but incorporates many
9847 bugfixes.
9848
9849 * The git repository moved to:
9850 git://anongit.freedesktop.org/systemd/systemd
9851 ssh://git.freedesktop.org/git/systemd/systemd
9852
9853 * First release with the journal
9854 http://0pointer.de/blog/projects/the-journal.html
9855
9856 * The journal replaces both systemd-kmsg-syslogd and
9857 systemd-stdout-bridge.
9858
9859 * New sd_pid_get_unit() API call in libsystemd-logind
9860
9861 * Many systemadm clean-ups
9862
9863 * Introduce remote-fs-pre.target which is ordered before all
9864 remote mounts and may be used to start services before all
9865 remote mounts.
9866
9867 * Added Mageia support
9868
9869 * Add bash completion for systemd-loginctl
9870
9871 * Actively monitor PID file creation for daemons which exit in
9872 the parent process before having finished writing the PID
9873 file in the daemon process. Daemons which do this need to be
9874 fixed (i.e. PID file creation must have finished before the
9875 parent exits), but we now react a bit more gracefully to them.
9876
9877 * Add colourful boot output, mimicking the well-known output
9878 of existing distributions.
9879
9880 * New option PassCredentials= for socket units, for
9881 compatibility with a recent kernel ABI breakage.
9882
9883 * /etc/rc.local is now hooked in via a generator binary, and
9884 thus will no longer act as synchronization point during
9885 boot.
9886
9887 * systemctl list-unit-files now supports --root=.
9888
9889 * systemd-tmpfiles now understands two new commands: z, Z for
9890 relabelling files according to the SELinux database. This is
9891 useful to apply SELinux labels to specific files in /sys,
9892 among other things.
9893
9894 * Output of SysV services is now forwarded to both the console
9895 and the journal by default, not only just the console.
9896
9897 * New man pages for all APIs from libsystemd-login.
9898
9899 * The build tree got reorganized and the build system is a
9900 lot more modular allowing embedded setups to specifically
9901 select the components of systemd they are interested in.
9902
9903 * Support for Linux systems lacking the kernel VT subsystem is
9904 restored.
9905
9906 * configure's --with-rootdir= got renamed to
9907 --with-rootprefix= to follow the naming used by udev and
9908 kmod
9909
9910 * Unless specified otherwise we will now install to /usr instead
9911 of /usr/local by default.
9912
9913 * Processes with '@' in argv[0][0] are now excluded from the
9914 final shut-down killing spree, following the logic explained
9915 in:
9916 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
9917
9918 * All processes remaining in a service cgroup when we enter
9919 the START or START_PRE states are now killed with
9920 SIGKILL. That means it is no longer possible to spawn
9921 background processes from ExecStart= lines (which was never
9922 supported anyway, and bad style).
9923
9924 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
9925 reloading of units together.
9926
9927 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
9928 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
9929 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
9930 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
9931 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek