]>
Commit | Line | Data |
---|---|---|
1 | systemd System and Service Manager | |
2 | ||
3 | CHANGES WITH 238 in spe: | |
4 | ||
5 | * The MemoryAccounting= unit property now defaults to on. After | |
6 | discussions with the upstream control group maintainers we learnt | |
7 | that the negative impact of cgroup memory accounting on current | |
8 | kernels is finally relatively minimal, so that it should be safe to | |
9 | enable this by default without affecting system performance. Besides | |
10 | memory accounting only task accounting is turned on by default, all | |
11 | other forms of resource accounting (CPU, IO, IP) remain off for now, | |
12 | because it's not clear yet that their impact is small enough to move | |
13 | from opt-in to opt-out. We recommend downstreams to leave memory | |
14 | accounting on by default if kernel 4.14 or higher is are primarily | |
15 | used. On very resource constrained systems or when support for old | |
16 | kernels is a necessity, -Dmemory-accounting-default=false can be used | |
17 | to revert this change. | |
18 | ||
19 | * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update, | |
20 | %udev_rules_update) and the journal catalog (%journal_catalog_update) | |
21 | from the upgrade scriptlets of individual packages now do nothing. | |
22 | Transfiletriggers have been added which will perform those updates | |
23 | once at the end of the transaction. | |
24 | ||
25 | Similar transfiletriggers have been added to execute any sysctl.d | |
26 | and binfmt.d rules. Thus, it should be unnecessary to provide any | |
27 | scriptlets to execute this configuration from package installation | |
28 | scripts. | |
29 | ||
30 | * systemd-sysusers gained a mode where the configuration to execute is | |
31 | specified on the command line, but this configuration is not executed | |
32 | directly, but instead it is merged with the configuration on disk, | |
33 | and the result is executed. This is useful for package installation | |
34 | scripts which want to create the user before installing any files on | |
35 | disk (in case some of those files are owned by that user), while | |
36 | still allowing local admin overrides. | |
37 | ||
38 | This functionality is exposed to rpm scriplets through a new | |
39 | %sysusers_create_package macro. Old %sysusers_create and | |
40 | %sysusers_create_inline macros are deprecated. | |
41 | ||
42 | A transfiletrigger for sysusers.d configuration is now installed, | |
43 | which means that it should be uncessary to call systemd-sysusers from | |
44 | package installation scripts, unless the package installs any files | |
45 | owned by those newly-created users, in which case | |
46 | %sysusers_create_package should be used. | |
47 | ||
48 | * Analogous change has been done for systemd-tmpfiles: it gained a mode | |
49 | where the command-line configuration is merged with the configuration | |
50 | on disk. This is exposed as the new %tmpfiles_create_package macro, | |
51 | and %tmpfiles_create is deprecated. A transfiletrigger is installed | |
52 | for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles | |
53 | from package installation scripts. | |
54 | ||
55 | * sysusers.d configuration for a user may now also specify the group | |
56 | number, in addition to the user number ("u username 123:456"), or | |
57 | without the user number ("u username -:456"). | |
58 | ||
59 | * Configution items for systemd-sysusers can now be specified as | |
60 | positional arguments when the new --inline switch is used. | |
61 | ||
62 | * The login shell of users created through sysusers.d may now be | |
63 | specified (previously, it was always /bin/sh for root and | |
64 | /sbin/nologin for other users). | |
65 | ||
66 | * systemd-analyze gained a new --global switch to look at global user | |
67 | configuration. It also gained a unit-paths verb to list the unit load | |
68 | paths that are compiled into systemd (which can be used with | |
69 | --systemd, --user, or --global). | |
70 | ||
71 | * udevadm trigger gained a new --settle/-w option to wait for any | |
72 | triggered events to finish (but just those, and not any other events | |
73 | which are triggered meanwhile). | |
74 | ||
75 | * The action that systemd-logind takes when the lid is closed and the | |
76 | machine is connected to external power can now be configured using | |
77 | HandleLidSwitchExternalPower= in logind.conf. Previously, this action | |
78 | was determined by HandleLidSwitch=, and, for backwards compatibility, | |
79 | is still is, if HandleLidSwitchExternalPower= is not explicitly set. | |
80 | ||
81 | * journalctl will periodically call sd_journal_process() to make it | |
82 | resilient against inotify queue overruns when journal files are | |
83 | rotated very quickly. | |
84 | ||
85 | * Two new functions in libsystemd — sd_bus_get_n_queued_read and | |
86 | sd_bus_get_n_queued_write — may be used to check the number of | |
87 | pending bus messages. | |
88 | ||
89 | * systemd gained a new | |
90 | org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call | |
91 | which can be used to migrate foreign processes to scope and service | |
92 | units. The primary user for this new API is systemd itself: the | |
93 | systemd --user instance uses this call of the systemd --system | |
94 | instance to migrate processes if it itself gets the request to | |
95 | migrate processes and the kernel refuses this due to access | |
96 | restrictions. Thanks to this "systemd-run --scope --user …" works | |
97 | again in pure cgroups v2 environments when invoked from the user | |
98 | session scope. | |
99 | ||
100 | * A new TemporaryFileSystem= setting can be used to mask out part of | |
101 | the real file system tree with tmpfs mounts. This may be combined | |
102 | with BindPaths= and BindReadOnlyPaths= to hide files or directories | |
103 | not relevant to the unit, while still allowing some paths lower in | |
104 | the tree to be accessed. | |
105 | ||
106 | ProtectHome=tmpfs may now be used to hide user home and runtime | |
107 | directories from units, in a way that is mostly equivalent to | |
108 | "TemporaryFileSystem=/home /run/user /root". | |
109 | ||
110 | * Non-service units are now started with KeyringMode=shared by default. | |
111 | This means that mount and swapon and other mount tools have access | |
112 | to keys in the main keyring. | |
113 | ||
114 | * /sys/fs/bpf is now mounted automatically. | |
115 | ||
116 | * QNX virtualization is now detected by systemd-detect-virt and may | |
117 | be used in ConditionVirtualization=. | |
118 | ||
119 | * IPAccounting= may now be enabled also for slice units. | |
120 | ||
121 | * A new -Dsplit-bin= build configuration switch may be used to specify | |
122 | whether bin and sbin directories are merged, or if they should be | |
123 | included separately in $PATH and various listings of executable | |
124 | directories. The build configuration scripts will try to autodetect | |
125 | the proper values of -Dsplit-usr= and -Dsplit-bin= based on build | |
126 | system, but distributions are encouraged to configure this | |
127 | explicitly. | |
128 | ||
129 | * A new -Dok-color= build configuration switch may be used to change | |
130 | the colour of "OK" status messages. | |
131 | ||
132 | * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with | |
133 | PrivateNetwork=yes was buggy in previous versions of systemd. This | |
134 | means that after the upgrade and daemon-reexec, any such units must | |
135 | be restarted. | |
136 | ||
137 | * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles | |
138 | will not exclude read-only files owned by root from cleanup. | |
139 | ||
140 | CHANGES WITH 237: | |
141 | ||
142 | * Some keyboards come with a zoom see-saw or rocker which until now got | |
143 | mapped to the Linux "zoomin/out" keys in hwdb. However, these | |
144 | keycodes are not recognized by any major desktop. They now produce | |
145 | Up/Down key events so that they can be used for scrolling. | |
146 | ||
147 | * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour | |
148 | slightly: previously, if an argument was specified for lines of this | |
149 | type (i.e. the right-most column was set) this string was appended to | |
150 | existing files each time systemd-tmpfiles was run. This behaviour was | |
151 | different from what the documentation said, and not particularly | |
152 | useful, as repeated systemd-tmpfiles invocations would not be | |
153 | idempotent and grow such files without bounds. With this release | |
154 | behaviour has been altered slightly, to match what the documentation | |
155 | says: lines of this type only have an effect if the indicated files | |
156 | don't exist yet, and only then the argument string is written to the | |
157 | file. | |
158 | ||
159 | * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change | |
160 | systemd-tmpfiles behaviour: previously, read-only files owned by root | |
161 | were always excluded from the file "aging" algorithm (i.e. the | |
162 | automatic clean-up of directories like /tmp based on | |
163 | atime/mtime/ctime). We intend to drop this restriction, and age files | |
164 | by default even when owned by root and read-only. This behaviour was | |
165 | inherited from older tools, but there have been requests to remove | |
166 | it, and it's not obvious why this restriction was made in the first | |
167 | place. Please speak up now, if you are aware of software that reqires | |
168 | this behaviour, otherwise we'll remove the restriction in v238. | |
169 | ||
170 | * A new environment variable $SYSTEMD_OFFLINE is now understood by | |
171 | systemctl. It takes a boolean argument. If on, systemctl assumes it | |
172 | operates on an "offline" OS tree, and will not attempt to talk to the | |
173 | service manager. Previously, this mode was implicitly enabled if a | |
174 | chroot() environment was detected, and this new environment variable | |
175 | now provides explicit control. | |
176 | ||
177 | * .path and .socket units may now be created transiently, too. | |
178 | Previously only service, mount, automount and timer units were | |
179 | supported as transient units. The systemd-run tool has been updated | |
180 | to expose this new functionality, you may hence use it now to bind | |
181 | arbitrary commands to path or socket activation on-the-fly from the | |
182 | command line. Moreover, almost all properties are now exposed for the | |
183 | unit types that already supported transient operation. | |
184 | ||
185 | * The systemd-mount command gained support for a new --owner= parameter | |
186 | which takes a user name, which is then resolved and included in uid= | |
187 | and gid= mount options string of the file system to mount. | |
188 | ||
189 | * A new unit condition ConditionControlGroupController= has been added | |
190 | that checks whether a specific cgroup controller is available. | |
191 | ||
192 | * Unit files, udev's .link files, and systemd-networkd's .netdev and | |
193 | .network files all gained support for a new condition | |
194 | ConditionKernelVersion= for checking against specific kernel | |
195 | versions. | |
196 | ||
197 | * In systemd-networkd, the [IPVLAN] section in .netdev files gained | |
198 | support for configuring device flags in the Flags= setting. In the | |
199 | same files, the [Tunnel] section gained support for configuring | |
200 | AllowLocalRemote=. The [Route] section in .network files gained | |
201 | support for configuring InitialCongestionWindow=, | |
202 | InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now | |
203 | understands RapidCommit=. | |
204 | ||
205 | * systemd-networkd's DHCPv6 support gained support for Prefix | |
206 | Delegation. | |
207 | ||
208 | * sd-bus gained support for a new "watch-bind" feature. When this | |
209 | feature is enabled, an sd_bus connection may be set up to connect to | |
210 | an AF_UNIX socket in the file system as soon as it is created. This | |
211 | functionality is useful for writing early-boot services that | |
212 | automatically connect to the system bus as soon as it is started, | |
213 | without ugly time-based polling. systemd-networkd and | |
214 | systemd-resolved have been updated to make use of this | |
215 | functionality. busctl exposes this functionality in a new | |
216 | --watch-bind= command line switch. | |
217 | ||
218 | * sd-bus will now optionally synthesize a local "Connected" signal as | |
219 | soon as a D-Bus connection is set up fully. This message mirrors the | |
220 | already existing "Disconnected" signal which is synthesized when the | |
221 | connection is terminated. This signal is generally useful but | |
222 | particularly handy in combination with the "watch-bind" feature | |
223 | described above. Synthesizing of this message has to be requested | |
224 | explicitly through the new API call sd_bus_set_connected_signal(). In | |
225 | addition a new call sd_bus_is_ready() has been added that checks | |
226 | whether a connection is fully set up (i.e. between the "Connected" and | |
227 | "Disconnected" signals). | |
228 | ||
229 | * sd-bus gained two new calls sd_bus_request_name_async() and | |
230 | sd_bus_release_name_async() for asynchronously registering bus | |
231 | names. Similar, there is now sd_bus_add_match_async() for installing | |
232 | a signal match asynchronously. All of systemd's own services have | |
233 | been updated to make use of these calls. Doing these operations | |
234 | asynchronously has two benefits: it reduces the risk of deadlocks in | |
235 | case of cyclic dependencies between bus services, and it speeds up | |
236 | service initialization since synchronization points for bus | |
237 | round-trips are removed. | |
238 | ||
239 | * sd-bus gained two new calls sd_bus_match_signal() and | |
240 | sd_bus_match_signal_async(), which are similar to sd_bus_add_match() | |
241 | and sd_bus_add_match_async() but instead of taking a D-Bus match | |
242 | string take match fields as normal function parameters. | |
243 | ||
244 | * sd-bus gained two new calls sd_bus_set_sender() and | |
245 | sd_bus_message_set_sender() for setting the sender name of outgoing | |
246 | messages (either for all outgoing messages or for just one specific | |
247 | one). These calls are only useful in direct connections as on | |
248 | brokered connections the broker fills in the sender anyway, | |
249 | overwriting whatever the client filled in. | |
250 | ||
251 | * sd-event gained a new pseudo-handle that may be specified on all API | |
252 | calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When | |
253 | used this refers to the default event loop object of the calling | |
254 | thread. Note however that this does not implicitly allocate one — | |
255 | which has to be done prior by using sd_event_default(). Similarly | |
256 | sd-bus gained three new pseudo-handles SD_BUS_DEFAULT, | |
257 | SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer | |
258 | to the default bus of the specified type of the calling thread. Here | |
259 | too this does not implicitly allocate bus connection objects, this | |
260 | has to be done prior with sd_bus_default() and friends. | |
261 | ||
262 | * sd-event gained a new call pair | |
263 | sd_event_source_{get|set}_io_fd_own(). This may be used to request | |
264 | automatic closure of the file descriptor an IO event source watches | |
265 | when the event source is destroyed. | |
266 | ||
267 | * systemd-networkd gained support for natively configuring WireGuard | |
268 | connections. | |
269 | ||
270 | * In previous versions systemd synthesized user records both for the | |
271 | "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and | |
272 | internally. In order to simplify distribution-wide renames of the | |
273 | "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a | |
274 | new transitional flag file has been added: if | |
275 | /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534 | |
276 | user and group record within the systemd codebase is disabled. | |
277 | ||
278 | * systemd-notify gained a new --uid= option for selecting the source | |
279 | user/UID to use for notification messages sent to the service | |
280 | manager. | |
281 | ||
282 | * journalctl gained a new --grep= option to list only entries in which | |
283 | the message matches a certain pattern. By default matching is case | |
284 | insensitive if the pattern is lowercase, and case sensitive | |
285 | otherwise. Option --case-sensitive=yes|no can be used to override | |
286 | this an specify case sensitivity or case insensitivity. | |
287 | ||
288 | * There's now a "systemd-analyze service-watchdogs" command for printing | |
289 | the current state of the service runtime watchdog, and optionally | |
290 | enabling or disabling the per-service watchdogs system-wide if given a | |
291 | boolean argument (i.e. the concept you configure in WatchdogSec=), for | |
292 | debugging purposes. There's also a kernel command line option | |
293 | systemd.service_watchdogs= for controlling the same. | |
294 | ||
295 | * Two new "log-level" and "log-target" options for systemd-analyze were | |
296 | addded that merge the now deprecated get-log-level, set-log-level and | |
297 | get-log-target, set-log-target pairs. The deprecated options are still | |
298 | understood for backwards compatibility. The two new options print the | |
299 | current value when no arguments are given, and set them when a | |
300 | level/target is given as an argument. | |
301 | ||
302 | * sysusers.d's "u" lines now optionally accept both a UID and a GID | |
303 | specification, separated by a ":" character, in order to create users | |
304 | where UID and GID do not match. | |
305 | ||
306 | Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov, | |
307 | Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman | |
308 | Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton | |
309 | Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov, | |
310 | Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui, | |
311 | Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian | |
312 | Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander | |
313 | Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen, | |
314 | Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg | |
315 | Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering, | |
316 | Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt, | |
317 | Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy, | |
318 | Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał | |
319 | Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf | |
320 | Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer, | |
321 | Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer, | |
322 | Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani, | |
323 | Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz | |
324 | Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary | |
325 | Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян | |
326 | Палаузов | |
327 | ||
328 | — Brno, 2018-01-28 | |
329 | ||
330 | CHANGES WITH 236: | |
331 | ||
332 | * The modprobe.d/ drop-in for the bonding.ko kernel module introduced | |
333 | in v235 has been extended to also set the dummy.ko module option | |
334 | numdummies=0, preventing the kernel from automatically creating | |
335 | dummy0. All dummy interfaces must now be explicitly created. | |
336 | ||
337 | * Unknown '%' specifiers in configuration files are now rejected. This | |
338 | applies to units and tmpfiles.d configuration. Any percent characters | |
339 | that are followed by a letter or digit that are not supposed to be | |
340 | interpreted as the beginning of a specifier should be escaped by | |
341 | doubling ("%%"). (So "size=5%" is still accepted, as well as | |
342 | "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not | |
343 | valid specifiers today.) | |
344 | ||
345 | * systemd-resolved now maintains a new dynamic | |
346 | /run/systemd/resolve/stub-resolv.conf compatibility file. It is | |
347 | recommended to make /etc/resolv.conf a symlink to it. This file | |
348 | points at the systemd-resolved stub DNS 127.0.0.53 resolver and | |
349 | includes dynamically acquired search domains, achieving more correct | |
350 | DNS resolution by software that bypasses local DNS APIs such as NSS. | |
351 | ||
352 | * The "uaccess" udev tag has been dropped from /dev/kvm and | |
353 | /dev/dri/renderD*. These devices now have the 0666 permissions by | |
354 | default (but this may be changed at build-time). /dev/dri/renderD* | |
355 | will now be owned by the "render" group along with /dev/kfd. | |
356 | ||
357 | * "DynamicUser=yes" has been enabled for systemd-timesyncd.service, | |
358 | systemd-journal-gatewayd.service and | |
359 | systemd-journal-upload.service. This means "nss-systemd" must be | |
360 | enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these | |
361 | services are resolved properly. | |
362 | ||
363 | * In /etc/fstab two new mount options are now understood: | |
364 | x-systemd.makefs and x-systemd.growfs. The former has the effect that | |
365 | the configured file system is formatted before it is mounted, the | |
366 | latter that the file system is resized to the full block device size | |
367 | after it is mounted (i.e. if the file system is smaller than the | |
368 | partition it resides on, it's grown). This is similar to the fsck | |
369 | logic in /etc/fstab, and pulls in systemd-makefs@.service and | |
370 | systemd-growfs@.service as necessary, similar to | |
371 | systemd-fsck@.service. Resizing is currently only supported on ext4 | |
372 | and btrfs. | |
373 | ||
374 | * In systemd-networkd, the IPv6 RA logic now optionally may announce | |
375 | DNS server and domain information. | |
376 | ||
377 | * Support for the LUKS2 on-disk format for encrypted partitions has | |
378 | been added. This requires libcryptsetup2 during compilation and | |
379 | runtime. | |
380 | ||
381 | * The systemd --user instance will now signal "readiness" when its | |
382 | basic.target unit has been reached, instead of when the run queue ran | |
383 | empty for the first time. | |
384 | ||
385 | * Tmpfiles.d with user configuration are now also supported. | |
386 | systemd-tmpfiles gained a new --user switch, and snippets placed in | |
387 | ~/.config/user-tmpfiles.d/ and corresponding directories will be | |
388 | executed by systemd-tmpfiles --user running in the new | |
389 | systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service | |
390 | running in the user session. | |
391 | ||
392 | * Unit files and tmpfiles.d snippets learnt three new % specifiers: | |
393 | %S resolves to the top-level state directory (/var/lib for the system | |
394 | instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the | |
395 | top-level cache directory (/var/cache for the system instance, | |
396 | $XDG_CACHE_HOME for the user instance), %L resolves to the top-level | |
397 | logs directory (/var/log for the system instance, | |
398 | $XDG_CONFIG_HOME/log/ for the user instance). This matches the | |
399 | existing %t specifier, that resolves to the top-level runtime | |
400 | directory (/run for the system instance, and $XDG_RUNTIME_DIR for the | |
401 | user instance). | |
402 | ||
403 | * journalctl learnt a new parameter --output-fields= for limiting the | |
404 | set of journal fields to output in verbose and JSON output modes. | |
405 | ||
406 | * systemd-timesyncd's configuration file gained a new option | |
407 | RootDistanceMaxSec= for setting the maximum root distance of servers | |
408 | it'll use, as well as the new options PollIntervalMinSec= and | |
409 | PollIntervalMaxSec= to tweak the minimum and maximum poll interval. | |
410 | ||
411 | * bootctl gained a new command "list" for listing all available boot | |
412 | menu items on systems that follow the boot loader specification. | |
413 | ||
414 | * systemctl gained a new --dry-run switch that shows what would be done | |
415 | instead of doing it, and is currently supported by the shutdown and | |
416 | sleep verbs. | |
417 | ||
418 | * ConditionSecurity= can now detect the TOMOYO security module. | |
419 | ||
420 | * Unit file [Install] sections are now also respected in unit drop-in | |
421 | files. This is intended to be used by drop-ins under /usr/lib/. | |
422 | ||
423 | * systemd-firstboot may now also set the initial keyboard mapping. | |
424 | ||
425 | * Udev "changed" events for devices which are exposed as systemd | |
426 | .device units are now propagated to units specified in | |
427 | ReloadPropagatedFrom= as reload requests. | |
428 | ||
429 | * If a udev device has a SYSTEMD_WANTS= property containing a systemd | |
430 | unit template name (i.e. a name in the form of 'foobar@.service', | |
431 | without the instance component between the '@' and - the '.'), then | |
432 | the escaped sysfs path of the device is automatically used as the | |
433 | instance. | |
434 | ||
435 | * SystemCallFilter= in unit files has been extended so that an "errno" | |
436 | can be specified individually for each system call. Example: | |
437 | SystemCallFilter=~uname:EILSEQ. | |
438 | ||
439 | * The cgroup delegation logic has been substantially updated. Delegate= | |
440 | now optionally takes a list of controllers (instead of a boolean, as | |
441 | before), which lists the controllers to delegate at least. | |
442 | ||
443 | * The networkd DHCPv6 client now implements the FQDN option (RFC 4704). | |
444 | ||
445 | * A new LogLevelMax= setting configures the maximum log level any | |
446 | process of the service may log at (i.e. anything with a lesser | |
447 | priority than what is specified is automatically dropped). A new | |
448 | LogExtraFields= setting allows configuration of additional journal | |
449 | fields to attach to all log records generated by any of the unit's | |
450 | processes. | |
451 | ||
452 | * New StandardInputData= and StandardInputText= settings along with the | |
453 | new option StandardInput=data may be used to configure textual or | |
454 | binary data that shall be passed to the executed service process via | |
455 | standard input, encoded in-line in the unit file. | |
456 | ||
457 | * StandardInput=, StandardOutput= and StandardError= may now be used to | |
458 | connect stdin/stdout/stderr of executed processes directly with a | |
459 | file or AF_UNIX socket in the file system, using the new "file:" option. | |
460 | ||
461 | * A new unit file option CollectMode= has been added, that allows | |
462 | tweaking the garbage collection logic for units. It may be used to | |
463 | tell systemd to garbage collect units that have failed automatically | |
464 | (normally it only GCs units that exited successfully). systemd-run | |
465 | and systemd-mount expose this new functionality with a new -G option. | |
466 | ||
467 | * "machinectl bind" may now be used to bind mount non-directories | |
468 | (i.e. regularfiles, devices, fifos, sockets). | |
469 | ||
470 | * systemd-analyze gained a new verb "calendar" for validating and | |
471 | testing calendar time specifications to use for OnCalendar= in timer | |
472 | units. Besides validating the expression it will calculate the next | |
473 | time the specified expression would elapse. | |
474 | ||
475 | * In addition to the pre-existing FailureAction= unit file setting | |
476 | there's now SuccessAction=, for configuring a shutdown action to | |
477 | execute when a unit completes successfully. This is useful in | |
478 | particular inside containers that shall terminate after some workload | |
479 | has been completed. Also, both options are now supported for all unit | |
480 | types, not just services. | |
481 | ||
482 | * networkds's IP rule support gained two new options | |
483 | IncomingInterface= and OutgoingInterface= for configuring the incoming | |
484 | and outgoing interfaces of configured rules. systemd-networkd also | |
485 | gained support for "vxcan" network devices. | |
486 | ||
487 | * networkd gained a new setting RequiredForOnline=, taking a | |
488 | boolean. If set, systemd-wait-online will take it into consideration | |
489 | when determining that the system is up, otherwise it will ignore the | |
490 | interface for this purpose. | |
491 | ||
492 | * The sd_notify() protocol gained support for a new operation: with | |
493 | FDSTOREREMOVE=1 file descriptors may be removed from the per-service | |
494 | store again, ahead of POLLHUP or POLLERR when they are removed | |
495 | anyway. | |
496 | ||
497 | * A new document doc/UIDS-GIDS.md has been added to the source tree, | |
498 | that documents the UID/GID range and assignment assumptions and | |
499 | requirements of systemd. | |
500 | ||
501 | * The watchdog device PID 1 will ping may now be configured through the | |
502 | WatchdogDevice= configuration file setting, or by setting the | |
503 | systemd.watchdog_service= kernel commandline option. | |
504 | ||
505 | * systemd-resolved's gained support for registering DNS-SD services on | |
506 | the local network using MulticastDNS. Services may either be | |
507 | registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or | |
508 | the same dir below /run, /usr/lib), or through its D-Bus API. | |
509 | ||
510 | * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond | |
511 | extend the effective start, runtime, and stop time. The service must | |
512 | continue to send EXTEND_TIMEOUT_USEC within the period specified to | |
513 | prevent the service manager from making the service as timedout. | |
514 | ||
515 | * systemd-resolved's DNSSEC support gained support for RFC 8080 | |
516 | (Ed25519 keys and signatures). | |
517 | ||
518 | * The systemd-resolve command line tool gained a new set of options | |
519 | --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=, | |
520 | --set-nta= and --revert to configure per-interface DNS configuration | |
521 | dynamically during runtime. It's useful for pushing DNS information | |
522 | into systemd-resolved from DNS hook scripts that various interface | |
523 | managing software supports (such as pppd). | |
524 | ||
525 | * systemd-nspawn gained a new --network-namespace-path= command line | |
526 | option, which may be used to make a container join an existing | |
527 | network namespace, by specifying a path to a "netns" file. | |
528 | ||
529 | Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini, | |
530 | Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten | |
531 | Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin | |
532 | Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri | |
533 | John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny | |
534 | Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de | |
535 | Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty, | |
536 | Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef | |
537 | Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars | |
538 | Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel, | |
539 | Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz | |
540 | Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson, | |
541 | Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala, | |
542 | Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal | |
543 | Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf | |
544 | Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer, | |
545 | Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran | |
546 | Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon | |
547 | Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant | |
548 | Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem | |
549 | Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich, | |
550 | Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew | |
551 | Jędrzejewski-Szmek, Zeal Jagannatha | |
552 | ||
553 | — Berlin, 2017-12-14 | |
554 | ||
555 | CHANGES WITH 235: | |
556 | ||
557 | * INCOMPATIBILITY: systemd-logind.service and other long-running | |
558 | services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP | |
559 | communication with the outside. This generally improves security of | |
560 | the system, and is in almost all cases a safe and good choice, as | |
561 | these services do not and should not provide any network-facing | |
562 | functionality. However, systemd-logind uses the glibc NSS API to | |
563 | query the user database. This creates problems on systems where NSS | |
564 | is set up to directly consult network services for user database | |
565 | lookups. In particular, this creates incompatibilities with the | |
566 | "nss-nis" module, which attempts to directly contact the NIS/YP | |
567 | network servers it is configured for, and will now consistently | |
568 | fail. In such cases, it is possible to turn off IP sandboxing for | |
569 | systemd-logind.service (set IPAddressDeny= in its [Service] section | |
570 | to the empty string, via a .d/ unit file drop-in). Downstream | |
571 | distributions might want to update their nss-nis packaging to include | |
572 | such a drop-in snippet, accordingly, to hide this incompatibility | |
573 | from the user. Another option is to make use of glibc's nscd service | |
574 | to proxy such network requests through a privilege-separated, minimal | |
575 | local caching daemon, or to switch to more modern technologies such | |
576 | sssd, whose NSS hook-ups generally do not involve direct network | |
577 | access. In general, we think it's definitely time to question the | |
578 | implementation choices of nss-nis, i.e. whether it's a good idea | |
579 | today to embed a network-facing loadable module into all local | |
580 | processes that need to query the user database, including the most | |
581 | trivial and benign ones, such as "ls". For more details about | |
582 | IPAddressDeny= see below. | |
583 | ||
584 | * A new modprobe.d drop-in is now shipped by default that sets the | |
585 | bonding module option max_bonds=0. This overrides the kernel default, | |
586 | to avoid conflicts and ambiguity as to whether or not bond0 should be | |
587 | managed by systemd-networkd or not. This resolves multiple issues | |
588 | with bond0 properties not being applied, when bond0 is configured | |
589 | with systemd-networkd. Distributors may choose to not package this, | |
590 | however in that case users will be prevented from correctly managing | |
591 | bond0 interface using systemd-networkd. | |
592 | ||
593 | * systemd-analyze gained new verbs "get-log-level" and "get-log-target" | |
594 | which print the logging level and target of the system manager. They | |
595 | complement the existing "set-log-level" and "set-log-target" verbs | |
596 | used to change those values. | |
597 | ||
598 | * journald.conf gained a new boolean setting ReadKMsg= which defaults | |
599 | to on. If turned off kernel log messages will not be read by | |
600 | systemd-journald or included in the logs. It also gained a new | |
601 | setting LineMax= for configuring the maximum line length in | |
602 | STDOUT/STDERR log streams. The new default for this value is 48K, up | |
603 | from the previous hardcoded 2048. | |
604 | ||
605 | * A new unit setting RuntimeDirectoryPreserve= has been added, which | |
606 | allows more detailed control of what to do with a runtime directory | |
607 | configured with RuntimeDirectory= (i.e. a directory below /run or | |
608 | $XDG_RUNTIME_DIR) after a unit is stopped. | |
609 | ||
610 | * The RuntimeDirectory= setting for units gained support for creating | |
611 | deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just | |
612 | one top-level directory. | |
613 | ||
614 | * Units gained new options StateDirectory=, CacheDirectory=, | |
615 | LogsDirectory= and ConfigurationDirectory= which are closely related | |
616 | to RuntimeDirectory= but manage per-service directories below | |
617 | /var/lib, /var/cache, /var/log and /etc. By making use of them it is | |
618 | possible to write unit files which when activated automatically gain | |
619 | properly owned service specific directories in these locations, thus | |
620 | making unit files self-contained and increasing compatibility with | |
621 | stateless systems and factory reset where /etc or /var are | |
622 | unpopulated at boot. Matching these new settings there's also | |
623 | StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=, | |
624 | ConfigurationDirectoryMode= for configuring the access mode of these | |
625 | directories. These settings are particularly useful in combination | |
626 | with DynamicUser=yes as they provide secure, properly-owned, | |
627 | writable, and stateful locations for storage, excluded from the | |
628 | sandbox that such services live in otherwise. | |
629 | ||
630 | * Automake support has been removed from this release. systemd is now | |
631 | Meson-only. | |
632 | ||
633 | * systemd-journald will now aggressively cache client metadata during | |
634 | runtime, speeding up log write performance under pressure. This comes | |
635 | at a small price though: as much of the metadata is read | |
636 | asynchronously from /proc/ (and isn't implicitly attached to log | |
637 | datagrams by the kernel, like UID/GID/PID/SELinux are) this means the | |
638 | metadata stored alongside a log entry might be slightly | |
639 | out-of-date. Previously it could only be slightly newer than the log | |
640 | message. The time window is small however, and given that the kernel | |
641 | is unlikely to be improved anytime soon in this regard, this appears | |
642 | acceptable to us. | |
643 | ||
644 | * nss-myhostname/systemd-resolved will now by default synthesize an | |
645 | A/AAAA resource record for the "_gateway" hostname, pointing to the | |
646 | current default IP gateway. Previously it did that for the "gateway" | |
647 | name, hampering adoption, as some distributions wanted to leave that | |
648 | host name open for local use. The old behaviour may still be | |
649 | requested at build time. | |
650 | ||
651 | * systemd-networkd's [Address] section in .network files gained a new | |
652 | Scope= setting for configuring the IP address scope. The [Network] | |
653 | section gained a new boolean setting ConfigureWithoutCarrier= that | |
654 | tells systemd-networkd to ignore link sensing when configuring the | |
655 | device. The [DHCP] section gained a new Anonymize= boolean option for | |
656 | turning on a number of options suggested in RFC 7844. A new | |
657 | [RoutingPolicyRule] section has been added for configuring the IP | |
658 | routing policy. The [Route] section has gained support for a new | |
659 | Type= setting which permits configuring | |
660 | blackhole/unreachable/prohibit routes. | |
661 | ||
662 | * The [VRF] section in .netdev files gained a new Table= setting for | |
663 | configuring the routing table to use. The [Tunnel] section gained a | |
664 | new Independent= boolean field for configuring tunnels independent of | |
665 | an underlying network interface. The [Bridge] section gained a new | |
666 | GroupForwardMask= option for configuration of propagation of link | |
667 | local frames between bridge ports. | |
668 | ||
669 | * The WakeOnLan= setting in .link files gained support for a number of | |
670 | new modes. A new TCP6SegmentationOffload= setting has been added for | |
671 | configuring TCP/IPv6 hardware segmentation offload. | |
672 | ||
673 | * The IPv6 RA sender implementation may now optionally send out RDNSS | |
674 | and RDNSSL records to supply DNS configuration to peers. | |
675 | ||
676 | * systemd-nspawn gained support for a new --system-call-filter= command | |
677 | line option for adding and removing entries in the default system | |
678 | call filter it applies. Moreover systemd-nspawn has been changed to | |
679 | implement a system call whitelist instead of a blacklist. | |
680 | ||
681 | * systemd-run gained support for a new --pipe command line option. If | |
682 | used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run | |
683 | are directly passed on to the activated transient service | |
684 | executable. This allows invoking arbitrary processes as systemd | |
685 | services (for example to take benefit of dependency management, | |
686 | accounting management, resource management or log management that is | |
687 | done automatically for services) — while still allowing them to be | |
688 | integrated in a classic UNIX shell pipeline. | |
689 | ||
690 | * When a service sends RELOAD=1 via sd_notify() and reload propagation | |
691 | using ReloadPropagationTo= is configured, a reload is now propagated | |
692 | to configured units. (Previously this was only done on explicitly | |
693 | requested reloads, using "systemctl reload" or an equivalent | |
694 | command.) | |
695 | ||
696 | * For each service unit a restart counter is now kept: it is increased | |
697 | each time the service is restarted due to Restart=, and may be | |
698 | queried using "systemctl show -p NRestarts …". | |
699 | ||
700 | * New system call filter groups @aio, @sync, @chown, @setuid, @memlock, | |
701 | @signal and @timer have been added, for usage with SystemCallFilter= | |
702 | in unit files and the new --system-call-filter= command line option | |
703 | of systemd-nspawn (see above). | |
704 | ||
705 | * ExecStart= lines in unit files gained two new modifiers: when a | |
706 | command line is prefixed with "!" the command will be executed as | |
707 | configured, except for the credentials applied by | |
708 | setuid()/setgid()/setgroups(). It is very similar to the pre-existing | |
709 | "+", but does still apply namespacing options unlike "+". There's | |
710 | also "!!" now, which is mostly identical, but becomes a NOP on | |
711 | systems that support ambient capabilities. This is useful to write | |
712 | unit files that work with ambient capabilities where possible but | |
713 | automatically fall back to traditional privilege dropping mechanisms | |
714 | on systems where this is not supported. | |
715 | ||
716 | * ListenNetlink= settings in socket units now support RDMA netlink | |
717 | sockets. | |
718 | ||
719 | * A new unit file setting LockPersonality= has been added which permits | |
720 | locking down the chosen execution domain ("personality") of a service | |
721 | during runtime. | |
722 | ||
723 | * A new special target "getty-pre.target" has been added, which is | |
724 | ordered before all text logins, and may be used to order services | |
725 | before textual logins acquire access to the console. | |
726 | ||
727 | * systemd will now attempt to load the virtio-rng.ko kernel module very | |
728 | early on if a VM environment supporting this is detected. This should | |
729 | improve entropy during early boot in virtualized environments. | |
730 | ||
731 | * A _netdev option is now supported in /etc/crypttab that operates in a | |
732 | similar way as the same option in /etc/fstab: it permits configuring | |
733 | encrypted devices that need to be ordered after the network is up. | |
734 | Following this logic, two new special targets | |
735 | remote-cryptsetup-pre.target and remote-cryptsetup.target have been | |
736 | added that are to cryptsetup.target what remote-fs.target and | |
737 | remote-fs-pre.target are to local-fs.target. | |
738 | ||
739 | * Service units gained a new UnsetEnvironment= setting which permits | |
740 | unsetting specific environment variables for services that are | |
741 | normally passed to it (for example in order to mask out locale | |
742 | settings for specific services that can't deal with it). | |
743 | ||
744 | * Units acquired a new boolean option IPAccounting=. When turned on, IP | |
745 | traffic accounting (packet count as well as byte count) is done for | |
746 | the service, and shown as part of "systemctl status" or "systemd-run | |
747 | --wait". | |
748 | ||
749 | * Service units acquired two new options IPAddressAllow= and | |
750 | IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks, | |
751 | for configuring a simple IP access control list for all sockets of | |
752 | the unit. These options are available also on .slice and .socket | |
753 | units, permitting flexible access list configuration for individual | |
754 | services as well as groups of services (as defined by a slice unit), | |
755 | including system-wide. Note that IP ACLs configured this way are | |
756 | enforced on every single IPv4 and IPv6 socket created by any process | |
757 | of the service unit, and apply to ingress as well as egress traffic. | |
758 | ||
759 | * If CPUAccounting= or IPAccounting= is turned on for a unit a new | |
760 | structured log message is generated each time the unit is stopped, | |
761 | containing information about the consumed resources of this | |
762 | invocation. | |
763 | ||
764 | * A new setting KeyringMode= has been added to unit files, which may be | |
765 | used to control how the kernel keyring is set up for executed | |
766 | processes. | |
767 | ||
768 | * "systemctl poweroff", "systemctl reboot", "systemctl halt", | |
769 | "systemctl kexec" and "systemctl exit" are now always asynchronous in | |
770 | behaviour (that is: these commands return immediately after the | |
771 | operation was enqueued instead of waiting for the operation to | |
772 | complete). Previously, "systemctl poweroff" and "systemctl reboot" | |
773 | were asynchronous on systems using systemd-logind (i.e. almost | |
774 | always, and like they were on sysvinit), and the other three commands | |
775 | were unconditionally synchronous. With this release this is cleaned | |
776 | up, and callers will see the same asynchronous behaviour on all | |
777 | systems for all five operations. | |
778 | ||
779 | * systemd-logind gained new Halt() and CanHalt() bus calls for halting | |
780 | the system. | |
781 | ||
782 | * .timer units now accept calendar specifications in other timezones | |
783 | than UTC or the local timezone. | |
784 | ||
785 | * The tmpfiles snippet var.conf has been changed to create | |
786 | /var/log/btmp with access mode 0660 instead of 0600. It was owned by | |
787 | the "utmp" group already, and it appears to be generally understood | |
788 | that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp | |
789 | databases. Previously this was implemented correctly for all these | |
790 | databases excepts btmp, which has been opened up like this now | |
791 | too. Note that while the other databases are world-readable | |
792 | (i.e. 0644), btmp is not and remains more restrictive. | |
793 | ||
794 | * The systemd-resolve tool gained a new --reset-server-features | |
795 | switch. When invoked like this systemd-resolved will forget | |
796 | everything it learnt about the features supported by the configured | |
797 | upstream DNS servers, and restarts the feature probing logic on the | |
798 | next resolver look-up for them at the highest feature level | |
799 | again. | |
800 | ||
801 | * The status dump systemd-resolved sends to the logs upon receiving | |
802 | SIGUSR1 now also includes information about all DNS servers it is | |
803 | configured to use, and the features levels it probed for them. | |
804 | ||
805 | Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander | |
806 | Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar | |
807 | Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles | |
808 | Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel | |
809 | Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John | |
810 | Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov, | |
811 | ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui, | |
812 | Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov, | |
813 | Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen, | |
814 | John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg | |
815 | Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud, | |
816 | Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas | |
817 | Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin | |
818 | Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert, | |
819 | Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell | |
820 | Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo, | |
821 | Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom | |
822 | Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid, | |
823 | Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang | |
824 | Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek | |
825 | ||
826 | — Berlin, 2017-10-06 | |
827 | ||
828 | CHANGES WITH 234: | |
829 | ||
830 | * Meson is now supported as build system in addition to Automake. It is | |
831 | our plan to remove Automake in one of our next releases, so that | |
832 | Meson becomes our exclusive build system. Hence, please start using | |
833 | the Meson build system in your downstream packaging. There's plenty | |
834 | of documentation around how to use Meson, the extremely brief | |
835 | summary: | |
836 | ||
837 | ./autogen.sh && ./configure && make && sudo make install | |
838 | ||
839 | becomes: | |
840 | ||
841 | meson build && ninja -C build && sudo ninja -C build install | |
842 | ||
843 | * Unit files gained support for a new JobRunningTimeoutUSec= setting, | |
844 | which permits configuring a timeout on the time a job is | |
845 | running. This is particularly useful for setting timeouts on jobs for | |
846 | .device units. | |
847 | ||
848 | * Unit files gained two new options ConditionUser= and ConditionGroup= | |
849 | for conditionalizing units based on the identity of the user/group | |
850 | running a systemd user instance. | |
851 | ||
852 | * systemd-networkd now understands a new FlowLabel= setting in the | |
853 | [VXLAN] section of .network files, as well as a Priority= in | |
854 | [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN] | |
855 | and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also | |
856 | gained support for configuration of GENEVE links, and IPv6 address | |
857 | labels. The [Network] section gained the new IPv6ProxyNDP= setting. | |
858 | ||
859 | * .link files now understand a new Port= setting. | |
860 | ||
861 | * systemd-networkd's DHCP support gained support for DHCP option 119 | |
862 | (domain search list). | |
863 | ||
864 | * systemd-networkd gained support for serving IPv6 address ranges using | |
865 | the Router Advertisment protocol. The new .network configuration | |
866 | section [IPv6Prefix] may be used to configure the ranges to | |
867 | serve. This is implemented based on a new, minimal, native server | |
868 | implementation of RA. | |
869 | ||
870 | * journalctl's --output= switch gained support for a new parameter | |
871 | "short-iso-precise" for a mode where timestamps are shown as precise | |
872 | ISO date values. | |
873 | ||
874 | * systemd-udevd's "net_id" builtin may now generate stable network | |
875 | interface names from IBM PowerVM VIO devices as well as ACPI platform | |
876 | devices. | |
877 | ||
878 | * MulticastDNS support in systemd-resolved may now be explicitly | |
879 | enabled/disabled using the new MulticastDNS= configuration file | |
880 | option. | |
881 | ||
882 | * systemd-resolved may now optionally use libidn2 instead of the libidn | |
883 | for processing internationalized domain names. Support for libidn2 | |
884 | should be considered experimental and should not be enabled by | |
885 | default yet. | |
886 | ||
887 | * "machinectl pull-tar" and related call may now do verification of | |
888 | downloaded images using SUSE-style .sha256 checksum files in addition | |
889 | to the already existing support for validating using Ubuntu-style | |
890 | SHA256SUMS files. | |
891 | ||
892 | * sd-bus gained support for a new sd_bus_message_appendv() call which | |
893 | is va_list equivalent of sd_bus_message_append(). | |
894 | ||
895 | * sd-boot gained support for validating images using SHIM/MOK. | |
896 | ||
897 | * The SMACK code learnt support for "onlycap". | |
898 | ||
899 | * systemd-mount --umount is now much smarter in figuring out how to | |
900 | properly unmount a device given its mount or device path. | |
901 | ||
902 | * The code to call libnss_dns as a fallback from libnss_resolve when | |
903 | the communication with systemd-resolved fails was removed. This | |
904 | fallback was redundant and interfered with the [!UNAVAIL=return] | |
905 | suffix. See nss-resolve(8) for the recommended configuration. | |
906 | ||
907 | * systemd-logind may now be restarted without losing state. It stores | |
908 | the file descriptors for devices it manages in the system manager | |
909 | using the FDSTORE= mechanism. Please note that further changes in | |
910 | other components may be required to make use of this (for example | |
911 | Xorg has code to listen for stops of systemd-logind and terminate | |
912 | itself when logind is stopped or restarted, in order to avoid using | |
913 | stale file descriptors for graphical devices, which is now | |
914 | counterproductive and must be reverted in order for restarts of | |
915 | systemd-logind to be safe. See | |
916 | https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.) | |
917 | ||
918 | * All kernel install plugins are called with the environment variable | |
919 | KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by | |
920 | /etc/machine-id. If the file is missing or empty, the variable is | |
921 | empty and BOOT_DIR_ABS is the path of a temporary directory which is | |
922 | removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID | |
923 | is empty, all plugins should not put anything in BOOT_DIR_ABS. | |
924 | ||
925 | Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander | |
926 | Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir | |
927 | Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert, | |
928 | Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb, | |
929 | Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake, | |
930 | Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide | |
931 | Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John | |
932 | Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin, | |
933 | Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary | |
934 | Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede, | |
935 | hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan | |
936 | Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason | |
937 | Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg | |
938 | Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow, | |
939 | Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili, | |
940 | Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas, | |
941 | Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala, | |
942 | Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin, | |
943 | Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal | |
944 | Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis, | |
945 | Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik | |
946 | Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr | |
947 | Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes, | |
948 | Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan | |
949 | Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas | |
950 | H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom | |
951 | Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog, | |
952 | userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu, | |
953 | Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан | |
954 | Георгиевски | |
955 | ||
956 | — Berlin, 2017-07-12 | |
957 | ||
958 | CHANGES WITH 233: | |
959 | ||
960 | * The "hybrid" control group mode has been modified to improve | |
961 | compatibility with "legacy" cgroups-v1 setups. Specifically, the | |
962 | "hybrid" setup of /sys/fs/cgroup is now pretty much identical to | |
963 | "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named | |
964 | cgroups-v1 hierarchy), the only externally visible change being that | |
965 | the cgroups-v2 hierarchy is also mounted, to | |
966 | /sys/fs/cgroup/unified. This should provide a large degree of | |
967 | compatibility with "legacy" cgroups-v1, while taking benefit of the | |
968 | better management capabilities of cgroups-v2. | |
969 | ||
970 | * The default control group setup mode may be selected both a boot-time | |
971 | via a set of kernel command line parameters (specifically: | |
972 | systemd.unified_cgroup_hierarchy= and | |
973 | systemd.legacy_systemd_cgroup_controller=), as well as a compile-time | |
974 | default selected on the configure command line | |
975 | (--with-default-hierarchy=). The upstream default is "hybrid" | |
976 | (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but | |
977 | this will change in a future systemd version to be "unified" (pure | |
978 | cgroups-v2 mode). The third option for the compile time option is | |
979 | "legacy", to enter pure cgroups-v1 mode. We recommend downstream | |
980 | distributions to default to "hybrid" mode for release distributions, | |
981 | starting with v233. We recommend "unified" for development | |
982 | distributions (specifically: distributions such as Fedora's rawhide) | |
983 | as that's where things are headed in the long run. Use "legacy" for | |
984 | greatest stability and compatibility only. | |
985 | ||
986 | * Note one current limitation of "unified" and "hybrid" control group | |
987 | setup modes: the kernel currently does not permit the systemd --user | |
988 | instance (i.e. unprivileged code) to migrate processes between two | |
989 | disconnected cgroup subtrees, even if both are managed and owned by | |
990 | the user. This effectively means "systemd-run --user --scope" doesn't | |
991 | work when invoked from outside of any "systemd --user" service or | |
992 | scope. Specifically, it is not supported from session scopes. We are | |
993 | working on fixing this in a future systemd version. (See #3388 for | |
994 | further details about this.) | |
995 | ||
996 | * DBus policy files are now installed into /usr rather than /etc. Make | |
997 | sure your system has dbus >= 1.9.18 running before upgrading to this | |
998 | version, or override the install path with --with-dbuspolicydir= . | |
999 | ||
1000 | * All python scripts shipped with systemd (specifically: the various | |
1001 | tests written in Python) now require Python 3. | |
1002 | ||
1003 | * systemd unit tests can now run standalone (without the source or | |
1004 | build directories), and can be installed into /usr/lib/systemd/tests/ | |
1005 | with 'make install-tests'. | |
1006 | ||
1007 | * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH, | |
1008 | CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the | |
1009 | kernel. | |
1010 | ||
1011 | * Support for the %c, %r, %R specifiers in unit files has been | |
1012 | removed. Specifiers are not supposed to be dependent on configuration | |
1013 | in the unit file itself (so that they resolve the same regardless | |
1014 | where used in the unit files), but these specifiers were influenced | |
1015 | by the Slice= option. | |
1016 | ||
1017 | * The shell invoked by debug-shell.service now defaults to /bin/sh in | |
1018 | all cases. If distributions want to use a different shell for this | |
1019 | purpose (for example Fedora's /sbin/sushell) they need to specify | |
1020 | this explicitly at configure time using --with-debug-shell=. | |
1021 | ||
1022 | * The confirmation spawn prompt has been reworked to offer the | |
1023 | following choices: | |
1024 | ||
1025 | (c)ontinue, proceed without asking anymore | |
1026 | (D)ump, show the state of the unit | |
1027 | (f)ail, don't execute the command and pretend it failed | |
1028 | (h)elp | |
1029 | (i)nfo, show a short summary of the unit | |
1030 | (j)obs, show jobs that are in progress | |
1031 | (s)kip, don't execute the command and pretend it succeeded | |
1032 | (y)es, execute the command | |
1033 | ||
1034 | The 'n' choice for the confirmation spawn prompt has been removed, | |
1035 | because its meaning was confusing. | |
1036 | ||
1037 | The prompt may now also be redirected to an alternative console by | |
1038 | specifying the console as parameter to systemd.confirm_spawn=. | |
1039 | ||
1040 | * Services of Type=notify require a READY=1 notification to be sent | |
1041 | during startup. If no such message is sent, the service now fails, | |
1042 | even if the main process exited with a successful exit code. | |
1043 | ||
1044 | * Services that fail to start up correctly now always have their | |
1045 | ExecStopPost= commands executed. Previously, they'd enter "failed" | |
1046 | state directly, without executing these commands. | |
1047 | ||
1048 | * The option MulticastDNS= of network configuration files has acquired | |
1049 | an actual implementation. With MulticastDNS=yes a host can resolve | |
1050 | names of remote hosts and reply to mDNS A and AAAA requests. | |
1051 | ||
1052 | * When units are about to be started an additional check is now done to | |
1053 | ensure that all dependencies of type BindsTo= (when used in | |
1054 | combination with After=) have been started. | |
1055 | ||
1056 | * systemd-analyze gained a new verb "syscall-filter" which shows which | |
1057 | system call groups are defined for the SystemCallFilter= unit file | |
1058 | setting, and which system calls they contain. | |
1059 | ||
1060 | * A new system call filter group "@filesystem" has been added, | |
1061 | consisting of various file system related system calls. Group | |
1062 | "@reboot" has been added, covering reboot, kexec and shutdown related | |
1063 | calls. Finally, group "@swap" has been added covering swap | |
1064 | configuration related calls. | |
1065 | ||
1066 | * A new unit file option RestrictNamespaces= has been added that may be | |
1067 | used to restrict access to the various process namespace types the | |
1068 | Linux kernel provides. Specifically, it may be used to take away the | |
1069 | right for a service unit to create additional file system, network, | |
1070 | user, and other namespaces. This sandboxing option is particularly | |
1071 | relevant due to the high amount of recently discovered namespacing | |
1072 | related vulnerabilities in the kernel. | |
1073 | ||
1074 | * systemd-udev's .link files gained support for a new AutoNegotiation= | |
1075 | setting for configuring Ethernet auto-negotiation. | |
1076 | ||
1077 | * systemd-networkd's .network files gained support for a new | |
1078 | ListenPort= setting in the [DHCP] section to explicitly configure the | |
1079 | UDP client port the DHCP client shall listen on. | |
1080 | ||
1081 | * .network files gained a new Unmanaged= boolean setting for explicitly | |
1082 | excluding one or more interfaces from management by systemd-networkd. | |
1083 | ||
1084 | * The systemd-networkd ProxyARP= option has been renamed to | |
1085 | IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been | |
1086 | renamed to ReduceARPProxy=. The old names continue to be available | |
1087 | for compatibility. | |
1088 | ||
1089 | * systemd-networkd gained support for configuring IPv6 Proxy NDP | |
1090 | addresses via the new IPv6ProxyNDPAddress= .network file setting. | |
1091 | ||
1092 | * systemd-networkd's bonding device support gained support for two new | |
1093 | configuration options ActiveSlave= and PrimarySlave=. | |
1094 | ||
1095 | * The various options in the [Match] section of .network files gained | |
1096 | support for negative matching. | |
1097 | ||
1098 | * New systemd-specific mount options are now understood in /etc/fstab: | |
1099 | ||
1100 | x-systemd.mount-timeout= may be used to configure the maximum | |
1101 | permitted runtime of the mount command. | |
1102 | ||
1103 | x-systemd.device-bound may be set to bind a mount point to its | |
1104 | backing device unit, in order to automatically remove a mount point | |
1105 | if its backing device is unplugged. This option may also be | |
1106 | configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property | |
1107 | on the block device, which is now automatically set for all CDROM | |
1108 | drives, so that mounted CDs are automatically unmounted when they are | |
1109 | removed from the drive. | |
1110 | ||
1111 | x-systemd.after= and x-systemd.before= may be used to explicitly | |
1112 | order a mount after or before another unit or mount point. | |
1113 | ||
1114 | * Enqueued start jobs for device units are now automatically garbage | |
1115 | collected if there are no jobs waiting for them anymore. | |
1116 | ||
1117 | * systemctl list-jobs gained two new switches: with --after, for every | |
1118 | queued job the jobs it's waiting for are shown; with --before the | |
1119 | jobs which it's blocking are shown. | |
1120 | ||
1121 | * systemd-nspawn gained support for ephemeral boots from disk images | |
1122 | (or in other words: --ephemeral and --image= may now be | |
1123 | combined). Moreover, ephemeral boots are now supported for normal | |
1124 | directories, even if the backing file system is not btrfs. Of course, | |
1125 | if the file system does not support file system snapshots or | |
1126 | reflinks, the initial copy operation will be relatively expensive, but | |
1127 | this should still be suitable for many use cases. | |
1128 | ||
1129 | * Calendar time specifications in .timer units now support | |
1130 | specifications relative to the end of a month by using "~" instead of | |
1131 | "-" as separator between month and day. For example, "*-02~03" means | |
1132 | "the third last day in February". In addition a new syntax for | |
1133 | repeated events has been added using the "/" character. For example, | |
1134 | "9..17/2:00" means "every two hours from 9am to 5pm". | |
1135 | ||
1136 | * systemd-socket-proxyd gained a new parameter --connections-max= for | |
1137 | configuring the maximum number of concurrent connections. | |
1138 | ||
1139 | * sd-id128 gained a new API for generating unique IDs for the host in a | |
1140 | way that does not leak the machine ID. Specifically, | |
1141 | sd_id128_get_machine_app_specific() derives an ID based on the | |
1142 | machine ID a in well-defined, non-reversible, stable way. This is | |
1143 | useful whenever an identifier for the host is needed but where the | |
1144 | identifier shall not be useful to identify the system beyond the | |
1145 | scope of the application itself. (Internally this uses HMAC-SHA256 as | |
1146 | keyed hash function using the machine ID as input.) | |
1147 | ||
1148 | * NotifyAccess= gained a new supported value "exec". When set | |
1149 | notifications are accepted from all processes systemd itself invoked, | |
1150 | including all control processes. | |
1151 | ||
1152 | * .nspawn files gained support for defining overlay mounts using the | |
1153 | Overlay= and OverlayReadOnly= options. Previously this functionality | |
1154 | was only available on the systemd-nspawn command line. | |
1155 | ||
1156 | * systemd-nspawn's --bind= and --overlay= options gained support for | |
1157 | bind/overlay mounts whose source lies within the container tree by | |
1158 | prefixing the source path with "+". | |
1159 | ||
1160 | * systemd-nspawn's --bind= and --overlay= options gained support for | |
1161 | automatically allocating a temporary source directory in /var/tmp | |
1162 | that is removed when the container dies. Specifically, if the source | |
1163 | directory is specified as empty string this mechanism is selected. An | |
1164 | example usage is --overlay=+/var::/var, which creates an overlay | |
1165 | mount based on the original /var contained in the image, overlayed | |
1166 | with a temporary directory in the host's /var/tmp. This way changes | |
1167 | to /var are automatically flushed when the container shuts down. | |
1168 | ||
1169 | * systemd-nspawn --image= option does now permit raw file system block | |
1170 | devices (in addition to images containing partition tables, as | |
1171 | before). | |
1172 | ||
1173 | * The disk image dissection logic in systemd-nspawn gained support for | |
1174 | automatically setting up LUKS encrypted as well as Verity protected | |
1175 | partitions. When a container is booted from an encrypted image the | |
1176 | passphrase is queried at start-up time. When a container with Verity | |
1177 | data is started, the root hash is search in a ".roothash" file | |
1178 | accompanying the disk image (alternatively, pass the root hash via | |
1179 | the new --root-hash= command line option). | |
1180 | ||
1181 | * A new tool /usr/lib/systemd/systemd-dissect has been added that may | |
1182 | be used to dissect disk images the same way as systemd-nspawn does | |
1183 | it, following the Bootable Partition Specification. It may even be | |
1184 | used to mount disk images with complex partition setups (including | |
1185 | LUKS and Verity partitions) to a local host directory, in order to | |
1186 | inspect them. This tool is not considered public API (yet), and is | |
1187 | thus not installed into /usr/bin. Please do not rely on its | |
1188 | existence, since it might go away or be changed in later systemd | |
1189 | versions. | |
1190 | ||
1191 | * A new generator "systemd-verity-generator" has been added, similar in | |
1192 | style to "systemd-cryptsetup-generator", permitting automatic setup of | |
1193 | Verity root partitions when systemd boots up. In order to make use of | |
1194 | this your partition setup should follow the Discoverable Partitions | |
1195 | Specification, and the GPT partition ID of the root file system | |
1196 | partition should be identical to the upper 128bit of the Verity root | |
1197 | hash. The GPT partition ID of the Verity partition protecting it | |
1198 | should be the lower 128bit of the Verity root hash. If the partition | |
1199 | image follows this model it is sufficient to specify a single | |
1200 | "roothash=" kernel command line argument to both configure which root | |
1201 | image and verity partition to use as well as the root hash for | |
1202 | it. Note that systemd-nspawn's Verity support follows the same | |
1203 | semantics, meaning that disk images with proper Verity data in place | |
1204 | may be booted in containers with systemd-nspawn as well as on | |
1205 | physical systems via the verity generator. Also note that the "mkosi" | |
1206 | tool available at https://github.com/systemd/mkosi has been updated | |
1207 | to generate Verity protected disk images following this scheme. In | |
1208 | fact, it has been updated to generate disk images that optionally | |
1209 | implement a complete UEFI SecureBoot trust chain, involving a signed | |
1210 | kernel and initrd image that incorporates such a root hash as well as | |
1211 | a Verity-enabled root partition. | |
1212 | ||
1213 | * The hardware database (hwdb) udev supports has been updated to carry | |
1214 | accelerometer quirks. | |
1215 | ||
1216 | * All system services are now run with a fresh kernel keyring set up | |
1217 | for them. The invocation ID is stored by default in it, thus | |
1218 | providing a safe, non-overridable way to determine the invocation | |
1219 | ID of each service. | |
1220 | ||
1221 | * Service unit files gained new BindPaths= and BindReadOnlyPaths= | |
1222 | options for bind mounting arbitrary paths in a service-specific | |
1223 | way. When these options are used, arbitrary host or service files and | |
1224 | directories may be mounted to arbitrary locations in the service's | |
1225 | view. | |
1226 | ||
1227 | * Documentation has been added that lists all of systemd's low-level | |
1228 | environment variables: | |
1229 | ||
1230 | https://github.com/systemd/systemd/blob/master/doc/ENVIRONMENT.md | |
1231 | ||
1232 | * sd-daemon gained a new API sd_is_socket_sockaddr() for determining | |
1233 | whether a specific socket file descriptor matches a specified socket | |
1234 | address. | |
1235 | ||
1236 | * systemd-firstboot has been updated to check for the | |
1237 | systemd.firstboot= kernel command line option. It accepts a boolean | |
1238 | and when set to false the first boot questions are skipped. | |
1239 | ||
1240 | * systemd-fstab-generator has been updated to check for the | |
1241 | systemd.volatile= kernel command line option, which either takes an | |
1242 | optional boolean parameter or the special value "state". If used the | |
1243 | system may be booted in a "volatile" boot mode. Specifically, | |
1244 | "systemd.volatile" is used, the root directory will be mounted as | |
1245 | tmpfs, and only /usr is mounted from the actual root file system. If | |
1246 | "systemd.volatile=state" is used, the root directory will be mounted | |
1247 | as usual, but /var is mounted as tmpfs. This concept provides similar | |
1248 | functionality as systemd-nspawn's --volatile= option, but provides it | |
1249 | on physical boots. Use this option for implementing stateless | |
1250 | systems, or testing systems with all state and/or configuration reset | |
1251 | to the defaults. (Note though that many distributions are not | |
1252 | prepared to boot up without a populated /etc or /var, though.) | |
1253 | ||
1254 | * systemd-gpt-auto-generator gained support for LUKS encrypted root | |
1255 | partitions. Previously it only supported LUKS encrypted partitions | |
1256 | for all other uses, except for the root partition itself. | |
1257 | ||
1258 | * Socket units gained support for listening on AF_VSOCK sockets for | |
1259 | communication in virtualized QEMU environments. | |
1260 | ||
1261 | * The "configure" script gained a new option --with-fallback-hostname= | |
1262 | for specifying the fallback hostname to use if none is configured in | |
1263 | /etc/hostname. For example, by specifying | |
1264 | --with-fallback-hostname=fedora it is possible to default to a | |
1265 | hostname of "fedora" on pristine installations. | |
1266 | ||
1267 | * systemd-cgls gained support for a new --unit= switch for listing only | |
1268 | the control groups of a specific unit. Similar --user-unit= has been | |
1269 | added for listing only the control groups of a specific user unit. | |
1270 | ||
1271 | * systemd-mount gained a new --umount switch for unmounting a mount or | |
1272 | automount point (and all mount/automount points below it). | |
1273 | ||
1274 | * systemd will now refuse full configuration reloads (via systemctl | |
1275 | daemon-reload and related calls) unless at least 16MiB of free space | |
1276 | are available in /run. This is a safety precaution in order to ensure | |
1277 | that generators can safely operate after the reload completed. | |
1278 | ||
1279 | * A new unit file option RootImage= has been added, which has a similar | |
1280 | effect as RootDirectory= but mounts the service's root directory from | |
1281 | a disk image instead of plain directory. This logic reuses the same | |
1282 | image dissection and mount logic that systemd-nspawn already uses, | |
1283 | and hence supports any disk images systemd-nspawn supports, including | |
1284 | those following the Discoverable Partition Specification, as well as | |
1285 | Verity enabled images. This option enables systemd to run system | |
1286 | services directly off disk images acting as resource bundles, | |
1287 | possibly even including full integrity data. | |
1288 | ||
1289 | * A new MountAPIVFS= unit file option has been added, taking a boolean | |
1290 | argument. If enabled /proc, /sys and /dev (collectively called the | |
1291 | "API VFS") will be mounted for the service. This is only relevant if | |
1292 | RootDirectory= or RootImage= is used for the service, as these mounts | |
1293 | are of course in place in the host mount namespace anyway. | |
1294 | ||
1295 | * systemd-nspawn gained support for a new --pivot-root= switch. If | |
1296 | specified the root directory within the container image is pivoted to | |
1297 | the specified mount point, while the original root disk is moved to a | |
1298 | different place. This option enables booting of ostree images | |
1299 | directly with systemd-nspawn. | |
1300 | ||
1301 | * The systemd build scripts will no longer complain if the NTP server | |
1302 | addresses are not changed from the defaults. Google now supports | |
1303 | these NTP servers officially. We still recommend downstreams to | |
1304 | properly register an NTP pool with the NTP pool project though. | |
1305 | ||
1306 | * coredumpctl gained a new "--reverse" option for printing the list | |
1307 | of coredumps in reverse order. | |
1308 | ||
1309 | * coredumpctl will now show additional information about truncated and | |
1310 | inaccessible coredumps, as well as coredumps that are still being | |
1311 | processed. It also gained a new --quiet switch for suppressing | |
1312 | additional informational message in its output. | |
1313 | ||
1314 | * coredumpctl gained support for only showing coredumps newer and/or | |
1315 | older than specific timestamps, using the new --since= and --until= | |
1316 | options, reminiscent of journalctl's options by the same name. | |
1317 | ||
1318 | * The systemd-coredump logic has been improved so that it may be reused | |
1319 | to collect backtraces in non-compiled languages, for example in | |
1320 | scripting languages such as Python. | |
1321 | ||
1322 | * machinectl will now show the UID shift of local containers, if user | |
1323 | namespacing is enabled for them. | |
1324 | ||
1325 | * systemd will now optionally run "environment generator" binaries at | |
1326 | configuration load time. They may be used to add environment | |
1327 | variables to the environment block passed to services invoked. One | |
1328 | user environment generator is shipped by default that sets up | |
1329 | environment variables based on files dropped into /etc/environment.d | |
1330 | and ~/.config/environment.d/. | |
1331 | ||
1332 | * systemd-resolved now includes the new, recently published 2017 DNSSEC | |
1333 | root key (KSK). | |
1334 | ||
1335 | * hostnamed has been updated to report a new chassis type of | |
1336 | "convertible" to cover "foldable" laptops that can both act as a | |
1337 | tablet and as a laptop, such as various Lenovo Yoga devices. | |
1338 | ||
1339 | Contributions from: Adrián López, Alexander Galanin, Alexander | |
1340 | Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch | |
1341 | Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric | |
1342 | Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri, | |
1343 | Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner, | |
1344 | David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry | |
1345 | Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly, | |
1346 | Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn | |
1347 | Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter, | |
1348 | Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede, | |
1349 | Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan | |
1350 | Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen, | |
1351 | Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart | |
1352 | Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de | |
1353 | Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry, | |
1354 | Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de | |
1355 | Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal | |
1356 | Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak, | |
1357 | Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip | |
1358 | Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin | |
1359 | Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx, | |
1360 | Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi, | |
1361 | Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève, | |
1362 | Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor | |
1363 | Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar | |
1364 | Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb, | |
1365 | Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun, | |
1366 | YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр | |
1367 | Тихонов | |
1368 | ||
1369 | — Berlin, 2017-03-01 | |
1370 | ||
1371 | CHANGES WITH 232: | |
1372 | ||
1373 | * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and | |
1374 | RestrictAddressFamilies= enabled. These sandboxing options should | |
1375 | generally be compatible with the various external udev call-out | |
1376 | binaries we are aware of, however there may be exceptions, in | |
1377 | particular when exotic languages for these call-outs are used. In | |
1378 | this case, consider turning off these settings locally. | |
1379 | ||
1380 | * The new RemoveIPC= option can be used to remove IPC objects owned by | |
1381 | the user or group of a service when that service exits. | |
1382 | ||
1383 | * The new ProtectKernelModules= option can be used to disable explicit | |
1384 | load and unload operations of kernel modules by a service. In | |
1385 | addition access to /usr/lib/modules is removed if this option is set. | |
1386 | ||
1387 | * ProtectSystem= option gained a new value "strict", which causes the | |
1388 | whole file system tree with the exception of /dev, /proc, and /sys, | |
1389 | to be remounted read-only for a service. | |
1390 | ||
1391 | * The new ProtectKernelTunables= option can be used to disable | |
1392 | modification of configuration files in /sys and /proc by a service. | |
1393 | Various directories and files are remounted read-only, so access is | |
1394 | restricted even if the file permissions would allow it. | |
1395 | ||
1396 | * The new ProtectControlGroups= option can be used to disable write | |
1397 | access by a service to /sys/fs/cgroup. | |
1398 | ||
1399 | * Various systemd services have been hardened with | |
1400 | ProtectKernelTunables=yes, ProtectControlGroups=yes, | |
1401 | RestrictAddressFamilies=. | |
1402 | ||
1403 | * Support for dynamically creating users for the lifetime of a service | |
1404 | has been added. If DynamicUser=yes is specified, user and group IDs | |
1405 | will be allocated from the range 61184..65519 for the lifetime of the | |
1406 | service. They can be resolved using the new nss-systemd.so NSS | |
1407 | module. The module must be enabled in /etc/nsswitch.conf. Services | |
1408 | started in this way have PrivateTmp= and RemoveIPC= enabled, so that | |
1409 | any resources allocated by the service will be cleaned up when the | |
1410 | service exits. They also have ProtectHome=read-only and | |
1411 | ProtectSystem=strict enabled, so they are not able to make any | |
1412 | permanent modifications to the system. | |
1413 | ||
1414 | * The nss-systemd module also always resolves root and nobody, making | |
1415 | it possible to have no /etc/passwd or /etc/group files in minimal | |
1416 | container or chroot environments. | |
1417 | ||
1418 | * Services may be started with their own user namespace using the new | |
1419 | boolean PrivateUsers= option. Only root, nobody, and the uid/gid | |
1420 | under which the service is running are mapped. All other users are | |
1421 | mapped to nobody. | |
1422 | ||
1423 | * Support for the cgroup namespace has been added to systemd-nspawn. If | |
1424 | supported by kernel, the container system started by systemd-nspawn | |
1425 | will have its own view of the cgroup hierarchy. This new behaviour | |
1426 | can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable. | |
1427 | ||
1428 | * The new MemorySwapMax= option can be used to limit the maximum swap | |
1429 | usage under the unified cgroup hierarchy. | |
1430 | ||
1431 | * Support for the CPU controller in the unified cgroup hierarchy has | |
1432 | been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting= | |
1433 | options. This controller requires out-of-tree patches for the kernel | |
1434 | and the support is provisional. | |
1435 | ||
1436 | * Mount and automount units may now be created transiently | |
1437 | (i.e. dynamically at runtime via the bus API, instead of requiring | |
1438 | unit files in the file system). | |
1439 | ||
1440 | * systemd-mount is a new tool which may mount file systems – much like | |
1441 | mount(8), optionally pulling in additional dependencies through | |
1442 | transient .mount and .automount units. For example, this tool | |
1443 | automatically runs fsck on a backing block device before mounting, | |
1444 | and allows the automount logic to be used dynamically from the | |
1445 | command line for establishing mount points. This tool is particularly | |
1446 | useful when dealing with removable media, as it will ensure fsck is | |
1447 | run – if necessary – before the first access and that the file system | |
1448 | is quickly unmounted after each access by utilizing the automount | |
1449 | logic. This maximizes the chance that the file system on the | |
1450 | removable media stays in a clean state, and if it isn't in a clean | |
1451 | state is fixed automatically. | |
1452 | ||
1453 | * LazyUnmount=yes option for mount units has been added to expose the | |
1454 | umount --lazy option. Similarly, ForceUnmount=yes exposes the --force | |
1455 | option. | |
1456 | ||
1457 | * /efi will be used as the mount point of the EFI boot partition, if | |
1458 | the directory is present, and the mount point was not configured | |
1459 | through other means (e.g. fstab). If /efi directory does not exist, | |
1460 | /boot will be used as before. This makes it easier to automatically | |
1461 | mount the EFI partition on systems where /boot is used for something | |
1462 | else. | |
1463 | ||
1464 | * When operating on GPT disk images for containers, systemd-nspawn will | |
1465 | now mount the ESP to /boot or /efi according to the same rules as PID | |
1466 | 1 running on a host. This allows tools like "bootctl" to operate | |
1467 | correctly within such containers, in order to make container images | |
1468 | bootable on physical systems. | |
1469 | ||
1470 | * disk/by-id and disk/by-path symlinks are now created for NVMe drives. | |
1471 | ||
1472 | * Two new user session targets have been added to support running | |
1473 | graphical sessions under the systemd --user instance: | |
1474 | graphical-session.target and graphical-session-pre.target. See | |
1475 | systemd.special(7) for a description of how those targets should be | |
1476 | used. | |
1477 | ||
1478 | * The vconsole initialization code has been significantly reworked to | |
1479 | use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better | |
1480 | support unicode keymaps. Font and keymap configuration will now be | |
1481 | copied to all allocated virtual consoles. | |
1482 | ||
1483 | * FreeBSD's bhyve virtualization is now detected. | |
1484 | ||
1485 | * Information recorded in the journal for core dumps now includes the | |
1486 | contents of /proc/mountinfo and the command line of the process at | |
1487 | the top of the process hierarchy (which is usually the init process | |
1488 | of the container). | |
1489 | ||
1490 | * systemd-journal-gatewayd learned the --directory= option to serve | |
1491 | files from the specified location. | |
1492 | ||
1493 | * journalctl --root=… can be used to peruse the journal in the | |
1494 | /var/log/ directories inside of a container tree. This is similar to | |
1495 | the existing --machine= option, but does not require the container to | |
1496 | be active. | |
1497 | ||
1498 | * The hardware database has been extended to support | |
1499 | ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify | |
1500 | trackball devices. | |
1501 | ||
1502 | MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to | |
1503 | specify the click rate for mice which include a horizontal wheel with | |
1504 | a click rate that is different than the one for the vertical wheel. | |
1505 | ||
1506 | * systemd-run gained a new --wait option that makes service execution | |
1507 | synchronous. (Specifically, the command will not return until the | |
1508 | specified service binary exited.) | |
1509 | ||
1510 | * systemctl gained a new --wait option that causes the start command to | |
1511 | wait until the units being started have terminated again. | |
1512 | ||
1513 | * A new journal output mode "short-full" has been added which displays | |
1514 | timestamps with abbreviated English day names and adds a timezone | |
1515 | suffix. Those timestamps include more information than the default | |
1516 | "short" output mode, and can be passed directly to journalctl's | |
1517 | --since= and --until= options. | |
1518 | ||
1519 | * /etc/resolv.conf will be bind-mounted into containers started by | |
1520 | systemd-nspawn, if possible, so any changes to resolv.conf contents | |
1521 | are automatically propagated to the container. | |
1522 | ||
1523 | * The number of instances for socket-activated services originating | |
1524 | from a single IP address can be limited with | |
1525 | MaxConnectionsPerSource=, extending the existing setting of | |
1526 | MaxConnections=. | |
1527 | ||
1528 | * systemd-networkd gained support for vcan ("Virtual CAN") interface | |
1529 | configuration. | |
1530 | ||
1531 | * .netdev and .network configuration can now be extended through | |
1532 | drop-ins. | |
1533 | ||
1534 | * UDP Segmentation Offload, TCP Segmentation Offload, Generic | |
1535 | Segmentation Offload, Generic Receive Offload, Large Receive Offload | |
1536 | can be enabled and disabled using the new UDPSegmentationOffload=, | |
1537 | TCPSegmentationOffload=, GenericSegmentationOffload=, | |
1538 | GenericReceiveOffload=, LargeReceiveOffload= options in the | |
1539 | [Link] section of .link files. | |
1540 | ||
1541 | * The Spanning Tree Protocol, Priority, Aging Time, and the Default | |
1542 | Port VLAN ID can be configured for bridge devices using the new STP=, | |
1543 | Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge] | |
1544 | section of .netdev files. | |
1545 | ||
1546 | * The route table to which routes received over DHCP or RA should be | |
1547 | added can be configured with the new RouteTable= option in the [DHCP] | |
1548 | and [IPv6AcceptRA] sections of .network files. | |
1549 | ||
1550 | * The Address Resolution Protocol can be disabled on links managed by | |
1551 | systemd-networkd using the ARP=no setting in the [Link] section of | |
1552 | .network files. | |
1553 | ||
1554 | * New environment variables $SERVICE_RESULT, $EXIT_CODE and | |
1555 | $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and | |
1556 | encode information about the result and exit codes of the current | |
1557 | service runtime cycle. | |
1558 | ||
1559 | * systemd-sysctl will now configure kernel parameters in the order | |
1560 | they occur in the configuration files. This matches what sysctl | |
1561 | has been traditionally doing. | |
1562 | ||
1563 | * kernel-install "plugins" that are executed to perform various | |
1564 | tasks after a new kernel is added and before an old one is removed | |
1565 | can now return a special value to terminate the procedure and | |
1566 | prevent any later plugins from running. | |
1567 | ||
1568 | * Journald's SplitMode=login setting has been deprecated. It has been | |
1569 | removed from documentation, and its use is discouraged. In a future | |
1570 | release it will be completely removed, and made equivalent to current | |
1571 | default of SplitMode=uid. | |
1572 | ||
1573 | * Storage=both option setting in /etc/systemd/coredump.conf has been | |
1574 | removed. With fast LZ4 compression storing the core dump twice is not | |
1575 | useful. | |
1576 | ||
1577 | * The --share-system systemd-nspawn option has been replaced with an | |
1578 | (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of | |
1579 | this functionality is discouraged. In addition the variables | |
1580 | $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID, | |
1581 | $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of | |
1582 | individual namespaces. | |
1583 | ||
1584 | * "machinectl list" now shows the IP address of running containers in | |
1585 | the output, as well as OS release information. | |
1586 | ||
1587 | * "loginctl list" now shows the TTY of each session in the output. | |
1588 | ||
1589 | * sd-bus gained new API calls sd_bus_track_set_recursive(), | |
1590 | sd_bus_track_get_recursive(), sd_bus_track_count_name(), | |
1591 | sd_bus_track_count_sender(). They permit usage of sd_bus_track peer | |
1592 | tracking objects in a "recursive" mode, where a single client can be | |
1593 | counted multiple times, if it takes multiple references. | |
1594 | ||
1595 | * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and | |
1596 | sd_bus_get_exit_on_disconnect(). They may be used to to make a | |
1597 | process using sd-bus automatically exit if the bus connection is | |
1598 | severed. | |
1599 | ||
1600 | * Bus clients of the service manager may now "pin" loaded units into | |
1601 | memory, by taking an explicit reference on them. This is useful to | |
1602 | ensure the client can retrieve runtime data about the service even | |
1603 | after the service completed execution. Taking such a reference is | |
1604 | available only for privileged clients and should be helpful to watch | |
1605 | running services in a race-free manner, and in particular collect | |
1606 | information about exit statuses and results. | |
1607 | ||
1608 | * The nss-resolve module has been changed to strictly return UNAVAIL | |
1609 | when communication via D-Bus with resolved failed, and NOTFOUND when | |
1610 | a lookup completed but was negative. This means it is now possible to | |
1611 | neatly configure fallbacks using nsswitch.conf result checking | |
1612 | expressions. Taking benefit of this, the new recommended | |
1613 | configuration line for the "hosts" entry in /etc/nsswitch.conf is: | |
1614 | ||
1615 | hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname | |
1616 | ||
1617 | * A new setting CtrlAltDelBurstAction= has been added to | |
1618 | /etc/systemd/system.conf which may be used to configure the precise | |
1619 | behaviour if the user on the console presses Ctrl-Alt-Del more often | |
1620 | than 7 times in 2s. Previously this would unconditionally result in | |
1621 | an expedited, immediate reboot. With this new setting the precise | |
1622 | operation may be configured in more detail, and also turned off | |
1623 | entirely. | |
1624 | ||
1625 | * In .netdev files two new settings RemoteChecksumTx= and | |
1626 | RemoteChecksumRx= are now understood that permit configuring the | |
1627 | remote checksumming logic for VXLAN networks. | |
1628 | ||
1629 | * The service manager learnt a new "invocation ID" concept for invoked | |
1630 | services. Each runtime cycle of a service will get a new invocation | |
1631 | ID (a 128bit random UUID) assigned that identifies the current | |
1632 | run of the service uniquely and globally. A new invocation ID | |
1633 | is generated each time a service starts up. The journal will store | |
1634 | the invocation ID of a service along with any logged messages, thus | |
1635 | making the invocation ID useful for matching the online runtime of a | |
1636 | service with the offline log data it generated in a safe way without | |
1637 | relying on synchronized timestamps. In many ways this new service | |
1638 | invocation ID concept is similar to the kernel's boot ID concept that | |
1639 | uniquely and globally identifies the runtime of each boot. The | |
1640 | invocation ID of a service is passed to the service itself via an | |
1641 | environment variable ($INVOCATION_ID). A new bus call | |
1642 | GetUnitByInvocationID() has been added that is similar to GetUnit() | |
1643 | but instead of retrieving the bus path for a unit by its name | |
1644 | retrieves it by its invocation ID. The returned path is valid only as | |
1645 | long as the passed invocation ID is current. | |
1646 | ||
1647 | * systemd-resolved gained a new "DNSStubListener" setting in | |
1648 | resolved.conf. It either takes a boolean value or the special values | |
1649 | "udp" and "tcp", and configures whether to enable the stub DNS | |
1650 | listener on 127.0.0.53:53. | |
1651 | ||
1652 | * IP addresses configured via networkd may now carry additional | |
1653 | configuration settings supported by the kernel. New options include: | |
1654 | HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=, | |
1655 | PrefixRoute=, AutoJoin=. | |
1656 | ||
1657 | * The PAM configuration fragment file for "user@.service" shipped with | |
1658 | systemd (i.e. the --user instance of systemd) has been stripped to | |
1659 | the minimum necessary to make the system boot. Previously, it | |
1660 | contained Fedora-specific stanzas that did not apply to other | |
1661 | distributions. It is expected that downstream distributions add | |
1662 | additional configuration lines, matching their needs to this file, | |
1663 | using it only as rough template of what systemd itself needs. Note | |
1664 | that this reduced fragment does not even include an invocation of | |
1665 | pam_limits which most distributions probably want to add, even though | |
1666 | systemd itself does not need it. (There's also the new build time | |
1667 | option --with-pamconfdir=no to disable installation of the PAM | |
1668 | fragment entirely.) | |
1669 | ||
1670 | * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO | |
1671 | capability is now also dropped from its set (in addition to | |
1672 | CAP_SYS_MKNOD as before). | |
1673 | ||
1674 | * In service unit files it is now possible to connect a specific named | |
1675 | file descriptor with stdin/stdout/stdout of an executed service. The | |
1676 | name may be specified in matching .socket units using the | |
1677 | FileDescriptorName= setting. | |
1678 | ||
1679 | * A number of journal settings may now be configured on the kernel | |
1680 | command line. Specifically, the following options are now understood: | |
1681 | systemd.journald.max_level_console=, | |
1682 | systemd.journald.max_level_store=, | |
1683 | systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=, | |
1684 | systemd.journald.max_level_wall=. | |
1685 | ||
1686 | * "systemctl is-enabled --full" will now show by which symlinks a unit | |
1687 | file is enabled in the unit dependency tree. | |
1688 | ||
1689 | * Support for VeraCrypt encrypted partitions has been added to the | |
1690 | "cryptsetup" logic and /etc/crypttab. | |
1691 | ||
1692 | * systemd-detect-virt gained support for a new --private-users switch | |
1693 | that checks whether the invoking processes are running inside a user | |
1694 | namespace. Similar, a new special value "private-users" for the | |
1695 | existing ConditionVirtualization= setting has been added, permitting | |
1696 | skipping of specific units in user namespace environments. | |
1697 | ||
1698 | Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John, | |
1699 | Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin | |
1700 | Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner, | |
1701 | Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez, | |
1702 | Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick, | |
1703 | Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg, | |
1704 | Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric | |
1705 | Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang, | |
1706 | Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke, | |
1707 | Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan | |
1708 | Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker, | |
1709 | Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski, | |
1710 | Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering, | |
1711 | Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš | |
1712 | Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou, | |
1713 | Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej | |
1714 | Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy, | |
1715 | Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike | |
1716 | Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer, | |
1717 | Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny | |
1718 | Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant | |
1719 | Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit, | |
1720 | Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut | |
1721 | Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann | |
1722 | E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew | |
1723 | Jędrzejewski-Szmek, Zeal Jagannatha | |
1724 | ||
1725 | — Santa Fe, 2016-11-03 | |
1726 | ||
1727 | CHANGES WITH 231: | |
1728 | ||
1729 | * In service units the various ExecXYZ= settings have been extended | |
1730 | with an additional special character as first argument of the | |
1731 | assigned value: if the character '+' is used the specified command | |
1732 | line it will be run with full privileges, regardless of User=, | |
1733 | Group=, CapabilityBoundingSet= and similar options. The effect is | |
1734 | similar to the existing PermissionsStartOnly= option, but allows | |
1735 | configuration of this concept for each executed command line | |
1736 | independently. | |
1737 | ||
1738 | * Services may now alter the service watchdog timeout at runtime by | |
1739 | sending a WATCHDOG_USEC= message via sd_notify(). | |
1740 | ||
1741 | * MemoryLimit= and related unit settings now optionally take percentage | |
1742 | specifications. The percentage is taken relative to the amount of | |
1743 | physical memory in the system (or in case of containers, the assigned | |
1744 | amount of memory). This allows scaling service resources neatly with | |
1745 | the amount of RAM available on the system. Similarly, systemd-logind's | |
1746 | RuntimeDirectorySize= option now also optionally takes percentage | |
1747 | values. | |
1748 | ||
1749 | * In similar fashion TasksMax= takes percentage values now, too. The | |
1750 | value is taken relative to the configured maximum number of processes | |
1751 | on the system. The per-service task maximum has been changed to 15% | |
1752 | using this functionality. (Effectively this is an increase of 512 → | |
1753 | 4915 for service units, given the kernel's default pid_max setting.) | |
1754 | ||
1755 | * Calendar time specifications in .timer units now understand a ".." | |
1756 | syntax for time ranges. Example: "4..7:10" may now be used for | |
1757 | defining a timer that is triggered at 4:10am, 5:10am, 6:10am and | |
1758 | 7:10am every day. | |
1759 | ||
1760 | * The InaccessableDirectories=, ReadOnlyDirectories= and | |
1761 | ReadWriteDirectories= unit file settings have been renamed to | |
1762 | InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be | |
1763 | applied to all kinds of file nodes, and not just directories, with | |
1764 | the exception of symlinks. Specifically these settings may now be | |
1765 | used on block and character device nodes, UNIX sockets and FIFOS as | |
1766 | well as regular files. The old names of these settings remain | |
1767 | available for compatibility. | |
1768 | ||
1769 | * systemd will now log about all service processes it kills forcibly | |
1770 | (using SIGKILL) because they remained after the clean shutdown phase | |
1771 | of the service completed. This should help identifying services that | |
1772 | shut down uncleanly. Moreover if KillUserProcesses= is enabled in | |
1773 | systemd-logind's configuration a similar log message is generated for | |
1774 | processes killed at the end of each session due to this setting. | |
1775 | ||
1776 | * systemd will now set the $JOURNAL_STREAM environment variable for all | |
1777 | services whose stdout/stderr are connected to the Journal (which | |
1778 | effectively means by default: all services). The variable contains | |
1779 | the device and inode number of the file descriptor used for | |
1780 | stdout/stderr. This may be used by invoked programs to detect whether | |
1781 | their stdout/stderr is connected to the Journal, in which case they | |
1782 | can switch over to direct Journal communication, thus being able to | |
1783 | pass extended, structured metadata along with their log messages. As | |
1784 | one example, this is now used by glib's logging primitives. | |
1785 | ||
1786 | * When using systemd's default tmp.mount unit for /tmp, the mount point | |
1787 | will now be established with the "nosuid" and "nodev" options. This | |
1788 | avoids privilege escalation attacks that put traps and exploits into | |
1789 | /tmp. However, this might cause problems if you e. g. put container | |
1790 | images or overlays into /tmp; if you need this, override tmp.mount's | |
1791 | "Options=" with a drop-in, or mount /tmp from /etc/fstab with your | |
1792 | desired options. | |
1793 | ||
1794 | * systemd now supports the "memory" cgroup controller also on | |
1795 | cgroupsv2. | |
1796 | ||
1797 | * The systemd-cgtop tool now optionally takes a control group path as | |
1798 | command line argument. If specified, the control group list shown is | |
1799 | limited to subgroups of that group. | |
1800 | ||
1801 | * The SystemCallFilter= unit file setting gained support for | |
1802 | pre-defined, named system call filter sets. For example | |
1803 | SystemCallFilter=@clock is now an effective way to make all clock | |
1804 | changing-related system calls unavailable to a service. A number of | |
1805 | similar pre-defined groups are defined. Writing system call filters | |
1806 | for system services is simplified substantially with this new | |
1807 | concept. Accordingly, all of systemd's own, long-running services now | |
1808 | enable system call filtering based on this, by default. | |
1809 | ||
1810 | * A new service setting MemoryDenyWriteExecute= has been added, taking | |
1811 | a boolean value. If turned on, a service may no longer create memory | |
1812 | mappings that are writable and executable at the same time. This | |
1813 | enhances security for services where this is enabled as it becomes | |
1814 | harder to dynamically write and then execute memory in exploited | |
1815 | service processes. This option has been enabled for all of systemd's | |
1816 | own long-running services. | |
1817 | ||
1818 | * A new RestrictRealtime= service setting has been added, taking a | |
1819 | boolean argument. If set the service's processes may no longer | |
1820 | acquire realtime scheduling. This improves security as realtime | |
1821 | scheduling may otherwise be used to easily freeze the system. | |
1822 | ||
1823 | * systemd-nspawn gained a new switch --notify-ready= taking a boolean | |
1824 | value. This may be used for requesting that the system manager inside | |
1825 | of the container reports start-up completion to nspawn which then | |
1826 | propagates this notification further to the service manager | |
1827 | supervising nspawn itself. A related option NotifyReady= in .nspawn | |
1828 | files has been added too. This functionality allows ordering of the | |
1829 | start-up of multiple containers using the usual systemd ordering | |
1830 | primitives. | |
1831 | ||
1832 | * machinectl gained a new command "stop" that is an alias for | |
1833 | "terminate". | |
1834 | ||
1835 | * systemd-resolved gained support for contacting DNS servers on | |
1836 | link-local IPv6 addresses. | |
1837 | ||
1838 | * If systemd-resolved receives the SIGUSR2 signal it will now flush all | |
1839 | its caches. A method call for requesting the same operation has been | |
1840 | added to the bus API too, and is made available via "systemd-resolve | |
1841 | --flush-caches". | |
1842 | ||
1843 | * systemd-resolve gained a new --status switch. If passed a brief | |
1844 | summary of the used DNS configuration with per-interface information | |
1845 | is shown. | |
1846 | ||
1847 | * resolved.conf gained a new Cache= boolean option, defaulting to | |
1848 | on. If turned off local DNS caching is disabled. This comes with a | |
1849 | performance penalty in particular when DNSSEC is enabled. Note that | |
1850 | resolved disables its internal caching implicitly anyway, when the | |
1851 | configured DNS server is on a host-local IP address such as ::1 or | |
1852 | 127.0.0.1, thus automatically avoiding double local caching. | |
1853 | ||
1854 | * systemd-resolved now listens on the local IP address 127.0.0.53:53 | |
1855 | for DNS requests. This improves compatibility with local programs | |
1856 | that do not use the libc NSS or systemd-resolved's bus APIs for name | |
1857 | resolution. This minimal DNS service is only available to local | |
1858 | programs and does not implement the full DNS protocol, but enough to | |
1859 | cover local DNS clients. A new, static resolv.conf file, listing just | |
1860 | this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is | |
1861 | now recommended to make /etc/resolv.conf a symlink to this file in | |
1862 | order to route all DNS lookups to systemd-resolved, regardless if | |
1863 | done via NSS, the bus API or raw DNS packets. Note that this local | |
1864 | DNS service is not as fully featured as the libc NSS or | |
1865 | systemd-resolved's bus APIs. For example, as unicast DNS cannot be | |
1866 | used to deliver link-local address information (as this implies | |
1867 | sending a local interface index along), LLMNR/mDNS support via this | |
1868 | interface is severely restricted. It is thus strongly recommended for | |
1869 | all applications to use the libc NSS API or native systemd-resolved | |
1870 | bus API instead. | |
1871 | ||
1872 | * systemd-networkd's bridge support learned a new setting | |
1873 | VLANFiltering= for controlling VLAN filtering. Moreover a new section | |
1874 | in .network files has been added for configuring VLAN bridging in | |
1875 | more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN]. | |
1876 | ||
1877 | * systemd-networkd's IPv6 Router Advertisement code now makes use of | |
1878 | the DNSSL and RDNSS options. This means IPv6 DNS configuration may | |
1879 | now be acquired without relying on DHCPv6. Two new options | |
1880 | UseDomains= and UseDNS= have been added to configure this behaviour. | |
1881 | ||
1882 | * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been | |
1883 | renamed IPv6AcceptRA=, without altering its behaviour. The old | |
1884 | setting name remains available for compatibility reasons. | |
1885 | ||
1886 | * The systemd-networkd VTI/VTI6 tunneling support gained new options | |
1887 | Key=, InputKey= and OutputKey=. | |
1888 | ||
1889 | * systemd-networkd gained support for VRF ("Virtual Routing Function") | |
1890 | interface configuration. | |
1891 | ||
1892 | * "systemctl edit" may now be used to create new unit files by | |
1893 | specifying the --force switch. | |
1894 | ||
1895 | * sd-event gained a new function sd_event_get_iteration() for | |
1896 | requesting the current iteration counter of the event loop. It starts | |
1897 | at zero and is increased by one with each event loop iteration. | |
1898 | ||
1899 | * A new rpm macro %systemd_ordering is provided by the macros.systemd | |
1900 | file. It can be used in lieu of %systemd_requires in packages which | |
1901 | don't use any systemd functionality and are intended to be installed | |
1902 | in minimal containers without systemd present. This macro provides | |
1903 | ordering dependencies to ensure that if the package is installed in | |
1904 | the same rpm transaction as systemd, systemd will be installed before | |
1905 | the scriptlets for the package are executed, allowing unit presets | |
1906 | to be handled. | |
1907 | ||
1908 | New macros %_systemdgeneratordir and %_systemdusergeneratordir have | |
1909 | been added to simplify packaging of generators. | |
1910 | ||
1911 | * The os-release file gained VERSION_CODENAME field for the | |
1912 | distribution nickname (e.g. VERSION_CODENAME=woody). | |
1913 | ||
1914 | * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1 | |
1915 | can be set to disable parsing of metadata and the creation | |
1916 | of persistent symlinks for that device. | |
1917 | ||
1918 | * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess" | |
1919 | to make them available to logged-in users has been reverted. | |
1920 | ||
1921 | * Much of the common code of the various systemd components is now | |
1922 | built into an internal shared library libsystemd-shared-231.so | |
1923 | (incorporating the systemd version number in the name, to be updated | |
1924 | with future releases) that the components link to. This should | |
1925 | decrease systemd footprint both in memory during runtime and on | |
1926 | disk. Note that the shared library is not for public use, and is | |
1927 | neither API not ABI stable, but is likely to change with every new | |
1928 | released update. Packagers need to make sure that binaries | |
1929 | linking to libsystemd-shared.so are updated in step with the | |
1930 | library. | |
1931 | ||
1932 | * Configuration for "mkosi" is now part of the systemd | |
1933 | repository. mkosi is a tool to easily build legacy-free OS images, | |
1934 | and is available on github: https://github.com/systemd/mkosi. If | |
1935 | "mkosi" is invoked in the build tree a new raw OS image is generated | |
1936 | incorporating the systemd sources currently being worked on and a | |
1937 | clean, fresh distribution installation. The generated OS image may be | |
1938 | booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical | |
1939 | UEFI PC. This functionality is particularly useful to easily test | |
1940 | local changes made to systemd in a pristine, defined environment. See | |
1941 | doc/HACKING for details. | |
1942 | ||
1943 | * configure learned the --with-support-url= option to specify the | |
1944 | distribution's bugtracker. | |
1945 | ||
1946 | Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor | |
1947 | Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika | |
1948 | Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar | |
1949 | Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse, | |
1950 | Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David | |
1951 | Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias | |
1952 | Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler, | |
1953 | Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan | |
1954 | Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke | |
1955 | Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart | |
1956 | Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel | |
1957 | Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov, | |
1958 | Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz, | |
1959 | Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran, | |
1960 | Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier, | |
1961 | Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas | |
1962 | Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan, | |
1963 | Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič, | |
1964 | WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek | |
1965 | ||
1966 | — Berlin, 2016-07-25 | |
1967 | ||
1968 | CHANGES WITH 230: | |
1969 | ||
1970 | * DNSSEC is now turned on by default in systemd-resolved (in | |
1971 | "allow-downgrade" mode), but may be turned off during compile time by | |
1972 | passing "--with-default-dnssec=no" to "configure" (and of course, | |
1973 | during runtime with DNSSEC= in resolved.conf). We recommend | |
1974 | downstreams to leave this on at least during development cycles and | |
1975 | report any issues with the DNSSEC logic upstream. We are very | |
1976 | interested in collecting feedback about the DNSSEC validator and its | |
1977 | limitations in the wild. Note however, that DNSSEC support is | |
1978 | probably nothing downstreams should turn on in stable distros just | |
1979 | yet, as it might create incompatibilities with a few DNS servers and | |
1980 | networks. We tried hard to make sure we downgrade to non-DNSSEC mode | |
1981 | automatically whenever we detect such incompatible setups, but there | |
1982 | might be systems we do not cover yet. Hence: please help us testing | |
1983 | the DNSSEC code, leave this on where you can, report back, but then | |
1984 | again don't consider turning this on in your stable, LTS or | |
1985 | production release just yet. (Note that you have to enable | |
1986 | nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved | |
1987 | and its DNSSEC mode for host name resolution from local | |
1988 | applications.) | |
1989 | ||
1990 | * systemd-resolve conveniently resolves DANE records with the --tlsa | |
1991 | option and OPENPGPKEY records with the --openpgp option. It also | |
1992 | supports dumping raw DNS record data via the new --raw= switch. | |
1993 | ||
1994 | * systemd-logind will now by default terminate user processes that are | |
1995 | part of the user session scope unit (session-XX.scope) when the user | |
1996 | logs out. This behavior is controlled by the KillUserProcesses= | |
1997 | setting in logind.conf, and the previous default of "no" is now | |
1998 | changed to "yes". This means that user sessions will be properly | |
1999 | cleaned up after, but additional steps are necessary to allow | |
2000 | intentionally long-running processes to survive logout. | |
2001 | ||
2002 | While the user is logged in at least once, user@.service is running, | |
2003 | and any service that should survive the end of any individual login | |
2004 | session can be started at a user service or scope using systemd-run. | |
2005 | systemd-run(1) man page has been extended with an example which shows | |
2006 | how to run screen in a scope unit underneath user@.service. The same | |
2007 | command works for tmux. | |
2008 | ||
2009 | After the user logs out of all sessions, user@.service will be | |
2010 | terminated too, by default, unless the user has "lingering" enabled. | |
2011 | To effectively allow users to run long-term tasks even if they are | |
2012 | logged out, lingering must be enabled for them. See loginctl(1) for | |
2013 | details. The default polkit policy was modified to allow users to | |
2014 | set lingering for themselves without authentication. | |
2015 | ||
2016 | Previous defaults can be restored at compile time by the | |
2017 | --without-kill-user-processes option to "configure". | |
2018 | ||
2019 | * systemd-logind gained new configuration settings SessionsMax= and | |
2020 | InhibitorsMax=, both with a default of 8192. It will not register new | |
2021 | user sessions or inhibitors above this limit. | |
2022 | ||
2023 | * systemd-logind will now reload configuration on SIGHUP. | |
2024 | ||
2025 | * The unified cgroup hierarchy added in Linux 4.5 is now supported. | |
2026 | Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to | |
2027 | enable. Also, support for the "io" cgroup controller in the unified | |
2028 | hierarchy has been added, so that the "memory", "pids" and "io" are | |
2029 | now the controllers that are supported on the unified hierarchy. | |
2030 | ||
2031 | WARNING: it is not possible to use previous systemd versions with | |
2032 | systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it | |
2033 | is necessary to also update systemd in the initramfs if using the | |
2034 | unified hierarchy. An updated SELinux policy is also required. | |
2035 | ||
2036 | * LLDP support has been extended, and both passive (receive-only) and | |
2037 | active (sender) modes are supported. Passive mode ("routers-only") is | |
2038 | enabled by default in systemd-networkd. Active LLDP mode is enabled | |
2039 | by default for containers on the internal network. The "networkctl | |
2040 | lldp" command may be used to list information gathered. "networkctl | |
2041 | status" will also show basic LLDP information on connected peers now. | |
2042 | ||
2043 | * The IAID and DUID unique identifier sent in DHCP requests may now be | |
2044 | configured for the system and each .network file managed by | |
2045 | systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options. | |
2046 | ||
2047 | * systemd-networkd gained support for configuring proxy ARP support for | |
2048 | each interface, via the ProxyArp= setting in .network files. It also | |
2049 | gained support for configuring the multicast querier feature of | |
2050 | bridge devices, via the new MulticastQuerier= setting in .netdev | |
2051 | files. Similarly, snooping on the IGMP traffic can be controlled | |
2052 | via the new setting MulticastSnooping=. | |
2053 | ||
2054 | A new setting PreferredLifetime= has been added for addresses | |
2055 | configured in .network file to configure the lifetime intended for an | |
2056 | address. | |
2057 | ||
2058 | The systemd-networkd DHCP server gained the option EmitRouter=, which | |
2059 | defaults to yes, to configure whether the DHCP Option 3 (Router) | |
2060 | should be emitted. | |
2061 | ||
2062 | * The testing tool /usr/lib/systemd/systemd-activate is renamed to | |
2063 | systemd-socket-activate and installed into /usr/bin. It is now fully | |
2064 | supported. | |
2065 | ||
2066 | * systemd-journald now uses separate threads to flush changes to disk | |
2067 | when closing journal files, thus reducing impact of slow disk I/O on | |
2068 | logging performance. | |
2069 | ||
2070 | * The sd-journal API gained two new calls | |
2071 | sd_journal_open_directory_fd() and sd_journal_open_files_fd() which | |
2072 | can be used to open journal files using file descriptors instead of | |
2073 | file or directory paths. sd_journal_open_container() has been | |
2074 | deprecated, sd_journal_open_directory_fd() should be used instead | |
2075 | with the flag SD_JOURNAL_OS_ROOT. | |
2076 | ||
2077 | * journalctl learned a new output mode "-o short-unix" that outputs log | |
2078 | lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970 | |
2079 | UTC). It also gained support for a new --no-hostname setting to | |
2080 | suppress the hostname column in the family of "short" output modes. | |
2081 | ||
2082 | * systemd-ask-password now optionally skips printing of the password to | |
2083 | stdout with --no-output which can be useful in scripts. | |
2084 | ||
2085 | * Framebuffer devices (/dev/fb*) and 3D printers and scanners | |
2086 | (devices tagged with ID_MAKER_TOOL) are now tagged with | |
2087 | "uaccess" and are available to logged in users. | |
2088 | ||
2089 | * The DeviceAllow= unit setting now supports specifiers (with "%"). | |
2090 | ||
2091 | * "systemctl show" gained a new --value switch, which allows print a | |
2092 | only the contents of a specific unit property, without also printing | |
2093 | the property's name. Similar support was added to "show*" verbs | |
2094 | of loginctl and machinectl that output "key=value" lists. | |
2095 | ||
2096 | * A new unit type "generated" was added for files dynamically generated | |
2097 | by generator tools. Similarly, a new unit type "transient" is used | |
2098 | for unit files created using the runtime API. "systemctl enable" will | |
2099 | refuse to operate on such files. | |
2100 | ||
2101 | * A new command "systemctl revert" has been added that may be used to | |
2102 | revert to the vendor version of a unit file, in case local changes | |
2103 | have been made by adding drop-ins or overriding the unit file. | |
2104 | ||
2105 | * "machinectl clean" gained a new verb to automatically remove all or | |
2106 | just hidden container images. | |
2107 | ||
2108 | * systemd-tmpfiles gained support for a new line type "e" for emptying | |
2109 | directories, if they exist, without creating them if they don't. | |
2110 | ||
2111 | * systemd-nspawn gained support for automatically patching the UID/GIDs | |
2112 | of the owners and the ACLs of all files and directories in a | |
2113 | container tree to match the UID/GID user namespacing range selected | |
2114 | for the container invocation. This mode is enabled via the new | |
2115 | --private-users-chown switch. It also gained support for | |
2116 | automatically choosing a free, previously unused UID/GID range when | |
2117 | starting a container, via the new --private-users=pick setting (which | |
2118 | implies --private-users-chown). Together, these options for the first | |
2119 | time make user namespacing for nspawn containers fully automatic and | |
2120 | thus deployable. The systemd-nspawn@.service template unit file has | |
2121 | been changed to use this functionality by default. | |
2122 | ||
2123 | * systemd-nspawn gained a new --network-zone= switch, that allows | |
2124 | creating ad-hoc virtual Ethernet links between multiple containers, | |
2125 | that only exist as long as at least one container referencing them is | |
2126 | running. This allows easy connecting of multiple containers with a | |
2127 | common link that implements an Ethernet broadcast domain. Each of | |
2128 | these network "zones" may be named relatively freely by the user, and | |
2129 | may be referenced by any number of containers, but each container may | |
2130 | only reference one of these "zones". On the lower level, this is | |
2131 | implemented by an automatically managed bridge network interface for | |
2132 | each zone, that is created when the first container referencing its | |
2133 | zone is created and removed when the last one referencing its zone | |
2134 | terminates. | |
2135 | ||
2136 | * The default start timeout may now be configured on the kernel command | |
2137 | line via systemd.default_timeout_start_sec=. It was already | |
2138 | configurable via the DefaultTimeoutStartSec= option in | |
2139 | /etc/systemd/system.conf. | |
2140 | ||
2141 | * Socket units gained a new TriggerLimitIntervalSec= and | |
2142 | TriggerLimitBurst= setting to configure a limit on the activation | |
2143 | rate of the socket unit. | |
2144 | ||
2145 | * The LimitNICE= setting now optionally takes normal UNIX nice values | |
2146 | in addition to the raw integer limit value. If the specified | |
2147 | parameter is prefixed with "+" or "-" and is in the range -20..19 the | |
2148 | value is understood as UNIX nice value. If not prefixed like this it | |
2149 | is understood as raw RLIMIT_NICE limit. | |
2150 | ||
2151 | * Note that the effect of the PrivateDevices= unit file setting changed | |
2152 | slightly with this release: the per-device /dev file system will be | |
2153 | mounted read-only from this version on, and will have "noexec" | |
2154 | set. This (minor) change of behavior might cause some (exceptional) | |
2155 | legacy software to break, when PrivateDevices=yes is set for its | |
2156 | service. Please leave PrivateDevices= off if you run into problems | |
2157 | with this. | |
2158 | ||
2159 | * systemd-bootchart has been split out to a separate repository: | |
2160 | https://github.com/systemd/systemd-bootchart | |
2161 | ||
2162 | * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be | |
2163 | merged into the kernel in its current form. | |
2164 | ||
2165 | * The compatibility libraries libsystemd-daemon.so, | |
2166 | libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so | |
2167 | which have been deprecated since systemd-209 have been removed along | |
2168 | with the corresponding pkg-config files. All symbols provided by | |
2169 | those libraries are provided by libsystemd.so. | |
2170 | ||
2171 | * The Capabilities= unit file setting has been removed (it is ignored | |
2172 | for backwards compatibility). AmbientCapabilities= and | |
2173 | CapabilityBoundingSet= should be used instead. | |
2174 | ||
2175 | * A new special target has been added, initrd-root-device.target, | |
2176 | which creates a synchronization point for dependencies of the root | |
2177 | device in early userspace. Initramfs builders must ensure that this | |
2178 | target is now included in early userspace. | |
2179 | ||
2180 | Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov, | |
2181 | Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin | |
2182 | Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens | |
2183 | Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh, | |
2184 | Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David | |
2185 | R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny | |
2186 | Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck | |
2187 | Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik | |
2188 | Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo | |
2189 | Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth, | |
2190 | John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos | |
2191 | Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir | |
2192 | Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt, | |
2193 | Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný, | |
2194 | Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin, | |
2195 | mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween, | |
2196 | Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern, | |
2197 | Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert | |
2198 | Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan | |
2199 | Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain | |
2200 | Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller, | |
2201 | Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen, | |
2202 | Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso, | |
2203 | Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev, | |
2204 | Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew | |
2205 | Jędrzejewski-Szmek | |
2206 | ||
2207 | — Fairfax, 2016-05-21 | |
2208 | ||
2209 | CHANGES WITH 229: | |
2210 | ||
2211 | * The systemd-resolved DNS resolver service has gained a substantial | |
2212 | set of new features, most prominently it may now act as a DNSSEC | |
2213 | validating stub resolver. DNSSEC mode is currently turned off by | |
2214 | default, but is expected to be turned on by default in one of the | |
2215 | next releases. For now, we invite everybody to test the DNSSEC logic | |
2216 | by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The | |
2217 | service also gained a full set of D-Bus interfaces, including calls | |
2218 | to configure DNS and DNSSEC settings per link (for use by external | |
2219 | network management software). systemd-resolved and systemd-networkd | |
2220 | now distinguish between "search" and "routing" domains. The former | |
2221 | are used to qualify single-label names, the latter are used purely | |
2222 | for routing lookups within certain domains to specific links. | |
2223 | resolved now also synthesizes RRs for all entries from /etc/hosts. | |
2224 | ||
2225 | * The systemd-resolve tool (which is a client utility for | |
2226 | systemd-resolved) has been improved considerably and is now fully | |
2227 | supported and documented. Hence it has moved from /usr/lib/systemd to | |
2228 | /usr/bin. | |
2229 | ||
2230 | * /dev/disk/by-path/ symlink support has been (re-)added for virtio | |
2231 | devices. | |
2232 | ||
2233 | * The coredump collection logic has been reworked: when a coredump is | |
2234 | collected it is now written to disk, compressed and processed | |
2235 | (including stacktrace extraction) from a new instantiated service | |
2236 | systemd-coredump@.service, instead of directly from the | |
2237 | /proc/sys/kernel/core_pattern hook we provide. This is beneficial as | |
2238 | processing large coredumps can take up a substantial amount of | |
2239 | resources and time, and this previously happened entirely outside of | |
2240 | systemd's service supervision. With the new logic the core_pattern | |
2241 | hook only does minimal metadata collection before passing off control | |
2242 | to the new instantiated service, which is configured with a time | |
2243 | limit, a nice level and other settings to minimize negative impact on | |
2244 | the rest of the system. Also note that the new logic will honour the | |
2245 | RLIMIT_CORE setting of the crashed process, which now allows users | |
2246 | and processes to turn off coredumping for their processes by setting | |
2247 | this limit. | |
2248 | ||
2249 | * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1 | |
2250 | and all forked processes by default. Previously, PID 1 would leave | |
2251 | the setting at "0" for all processes, as set by the kernel. Note that | |
2252 | the resource limit traditionally has no effect on the generated | |
2253 | coredumps on the system if the /proc/sys/kernel/core_pattern hook | |
2254 | logic is used. Since the limit is now honoured (see above) its | |
2255 | default has been changed so that the coredumping logic is enabled by | |
2256 | default for all processes, while allowing specific opt-out. | |
2257 | ||
2258 | * When the stacktrace is extracted from processes of system users, this | |
2259 | is now done as "systemd-coredump" user, in order to sandbox this | |
2260 | potentially security sensitive parsing operation. (Note that when | |
2261 | processing coredumps of normal users this is done under the user ID | |
2262 | of process that crashed, as before.) Packagers should take notice | |
2263 | that it is now necessary to create the "systemd-coredump" system user | |
2264 | and group at package installation time. | |
2265 | ||
2266 | * The systemd-activate socket activation testing tool gained support | |
2267 | for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram | |
2268 | and --seqpacket switches. It also has been extended to support both | |
2269 | new-style and inetd-style file descriptor passing. Use the new | |
2270 | --inetd switch to request inetd-style file descriptor passing. | |
2271 | ||
2272 | * Most systemd tools now honor a new $SYSTEMD_COLORS environment | |
2273 | variable, which takes a boolean value. If set to false, ANSI color | |
2274 | output is disabled in the tools even when run on a terminal that | |
2275 | supports it. | |
2276 | ||
2277 | * The VXLAN support in networkd now supports two new settings | |
2278 | DestinationPort= and PortRange=. | |
2279 | ||
2280 | * A new systemd.machine_id= kernel command line switch has been added, | |
2281 | that may be used to set the machine ID in /etc/machine-id if it is | |
2282 | not initialized yet. This command line option has no effect if the | |
2283 | file is already initialized. | |
2284 | ||
2285 | * systemd-nspawn gained a new --as-pid2 switch that invokes any | |
2286 | specified command line as PID 2 rather than PID 1 in the | |
2287 | container. In this mode PID 1 is a minimal stub init process that | |
2288 | implements the special POSIX and Linux semantics of PID 1 regarding | |
2289 | signal and child process management. Note that this stub init process | |
2290 | is implemented in nspawn itself and requires no support from the | |
2291 | container image. This new logic is useful to support running | |
2292 | arbitrary commands in the container, as normal processes are | |
2293 | generally not prepared to run as PID 1. | |
2294 | ||
2295 | * systemd-nspawn gained a new --chdir= switch for setting the current | |
2296 | working directory for the process started in the container. | |
2297 | ||
2298 | * "journalctl /dev/sda" will now output all kernel log messages for | |
2299 | specified device from the current boot, in addition to all devices | |
2300 | that are parents of it. This should make log output about devices | |
2301 | pretty useful, as long as kernel drivers attach enough metadata to | |
2302 | the log messages. (The usual SATA drivers do.) | |
2303 | ||
2304 | * The sd-journal API gained two new calls | |
2305 | sd_journal_has_runtime_files() and sd_journal_has_persistent_files() | |
2306 | that report whether log data from /run or /var has been found. | |
2307 | ||
2308 | * journalctl gained a new switch "--fields" that prints all journal | |
2309 | record field names currently in use in the journal. This is backed | |
2310 | by two new sd-journal API calls sd_journal_enumerate_fields() and | |
2311 | sd_journal_restart_fields(). | |
2312 | ||
2313 | * Most configurable timeouts in systemd now expect an argument of | |
2314 | "infinity" to turn them off, instead of "0" as before. The semantics | |
2315 | from now on is that a timeout of "0" means "now", and "infinity" | |
2316 | means "never". To maintain backwards compatibility, "0" continues to | |
2317 | turn off previously existing timeout settings. | |
2318 | ||
2319 | * "systemctl reload-or-try-restart" has been renamed to "systemctl | |
2320 | try-reload-or-restart" to clarify what it actually does: the "try" | |
2321 | logic applies to both reloading and restarting, not just restarting. | |
2322 | The old name continues to be accepted for compatibility. | |
2323 | ||
2324 | * On boot-up, when PID 1 detects that the system clock is behind the | |
2325 | release date of the systemd version in use, the clock is now set | |
2326 | to the latter. Previously, this was already done in timesyncd, in order | |
2327 | to avoid running with clocks set to the various clock epochs such as | |
2328 | 1902, 1938 or 1970. With this change the logic is now done in PID 1 | |
2329 | in addition to timesyncd during early boot-up, so that it is enforced | |
2330 | before the first process is spawned by systemd. Note that the logic | |
2331 | in timesyncd remains, as it is more comprehensive and ensures | |
2332 | clock monotonicity by maintaining a persistent timestamp file in | |
2333 | /var. Since /var is generally not available in earliest boot or the | |
2334 | initrd, this part of the logic remains in timesyncd, and is not done | |
2335 | by PID 1. | |
2336 | ||
2337 | * Support for tweaking details in net_cls.class_id through the | |
2338 | NetClass= configuration directive has been removed, as the kernel | |
2339 | people have decided to deprecate that controller in cgroup v2. | |
2340 | Userspace tools such as nftables are moving over to setting rules | |
2341 | that are specific to the full cgroup path of a task, which obsoletes | |
2342 | these controllers anyway. The NetClass= directive is kept around for | |
2343 | legacy compatibility reasons. For a more in-depth description of the | |
2344 | kernel change, please refer to the respective upstream commit: | |
2345 | ||
2346 | https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671 | |
2347 | ||
2348 | * A new service setting RuntimeMaxSec= has been added that may be used | |
2349 | to specify a maximum runtime for a service. If the timeout is hit, the | |
2350 | service is terminated and put into a failure state. | |
2351 | ||
2352 | * A new service setting AmbientCapabilities= has been added. It allows | |
2353 | configuration of additional Linux process capabilities that are | |
2354 | passed to the activated processes. This is only available on very | |
2355 | recent kernels. | |
2356 | ||
2357 | * The process resource limit settings in service units may now be used | |
2358 | to configure hard and soft limits individually. | |
2359 | ||
2360 | * The various libsystemd APIs such as sd-bus or sd-event now publicly | |
2361 | expose support for gcc's __attribute__((cleanup())) C extension. | |
2362 | Specifically, for many object destructor functions alternative | |
2363 | versions have been added that have names suffixed with "p" and take a | |
2364 | pointer to a pointer to the object to destroy, instead of just a | |
2365 | pointer to the object itself. This is useful because these destructor | |
2366 | functions may be used directly as parameters to the cleanup | |
2367 | construct. Internally, systemd has been a heavy user of this GCC | |
2368 | extension for a long time, and with this change similar support is | |
2369 | now available to consumers of the library outside of systemd. Note | |
2370 | that by using this extension in your sources compatibility with old | |
2371 | and strictly ANSI compatible C compilers is lost. However, all gcc or | |
2372 | LLVM versions of recent years support this extension. | |
2373 | ||
2374 | * Timer units gained support for a new setting RandomizedDelaySec= that | |
2375 | allows configuring some additional randomized delay to the configured | |
2376 | time. This is useful to spread out timer events to avoid load peaks in | |
2377 | clusters or larger setups. | |
2378 | ||
2379 | * Calendar time specifications now support sub-second accuracy. | |
2380 | ||
2381 | * Socket units now support listening on SCTP and UDP-lite protocol | |
2382 | sockets. | |
2383 | ||
2384 | * The sd-event API now comes with a full set of man pages. | |
2385 | ||
2386 | * Older versions of systemd contained experimental support for | |
2387 | compressing journal files and coredumps with the LZ4 compressor that | |
2388 | was not compatible with the lz4 binary (due to API limitations of the | |
2389 | lz4 library). This support has been removed; only support for files | |
2390 | compatible with the lz4 binary remains. This LZ4 logic is now | |
2391 | officially supported and no longer considered experimental. | |
2392 | ||
2393 | * The dkr image import logic has been removed again from importd. dkr's | |
2394 | micro-services focus doesn't fit into the machine image focus of | |
2395 | importd, and quickly got out of date with the upstream dkr API. | |
2396 | ||
2397 | * Creation of the /run/lock/lockdev/ directory was dropped from | |
2398 | tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have | |
2399 | been available for many years. If you still need this, you need to | |
2400 | create your own tmpfiles.d config file with: | |
2401 | ||
2402 | d /run/lock/lockdev 0775 root lock - | |
2403 | ||
2404 | * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction= | |
2405 | and RebootArgument= have been moved from the [Service] section of | |
2406 | unit files to [Unit], and they are now supported on all unit types, | |
2407 | not just service units. Of course, systemd will continue to | |
2408 | understand these settings also at the old location, in order to | |
2409 | maintain compatibility. | |
2410 | ||
2411 | Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander | |
2412 | Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov, | |
2413 | Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler, | |
2414 | Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan | |
2415 | Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack, | |
2416 | David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman, | |
2417 | Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen, | |
2418 | Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen, | |
2419 | Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub | |
2420 | Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek, | |
2421 | Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos, | |
2422 | lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel | |
2423 | Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer, | |
2424 | Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils | |
2425 | Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz, | |
2426 | Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant | |
2427 | Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel | |
2428 | Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito | |
2429 | Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek | |
2430 | ||
2431 | — Berlin, 2016-02-11 | |
2432 | ||
2433 | CHANGES WITH 228: | |
2434 | ||
2435 | * A number of properties previously only settable in unit | |
2436 | files are now also available as properties to set when | |
2437 | creating transient units programmatically via the bus, as it | |
2438 | is exposed with systemd-run's --property= | |
2439 | setting. Specifically, these are: SyslogIdentifier=, | |
2440 | SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=, | |
2441 | EnvironmentFile=, ReadWriteDirectories=, | |
2442 | ReadOnlyDirectories=, InaccessibleDirectories=, | |
2443 | ProtectSystem=, ProtectHome=, RuntimeDirectory=. | |
2444 | ||
2445 | * When creating transient services via the bus API it is now | |
2446 | possible to pass in a set of file descriptors to use as | |
2447 | STDIN/STDOUT/STDERR for the invoked process. | |
2448 | ||
2449 | * Slice units may now be created transiently via the bus APIs, | |
2450 | similar to the way service and scope units may already be | |
2451 | created transiently. | |
2452 | ||
2453 | * Wherever systemd expects a calendar timestamp specification | |
2454 | (like in journalctl's --since= and --until= switches) UTC | |
2455 | timestamps are now supported. Timestamps suffixed with "UTC" | |
2456 | are now considered to be in Universal Time Coordinated | |
2457 | instead of the local timezone. Also, timestamps may now | |
2458 | optionally be specified with sub-second accuracy. Both of | |
2459 | these additions also apply to recurring calendar event | |
2460 | specification, such as OnCalendar= in timer units. | |
2461 | ||
2462 | * journalctl gained a new "--sync" switch that asks the | |
2463 | journal daemon to write all so far unwritten log messages to | |
2464 | disk and sync the files, before returning. | |
2465 | ||
2466 | * systemd-tmpfiles learned two new line types "q" and "Q" that | |
2467 | operate like "v", but also set up a basic btrfs quota | |
2468 | hierarchy when used on a btrfs file system with quota | |
2469 | enabled. | |
2470 | ||
2471 | * tmpfiles' "v", "q" and "Q" will now create a plain directory | |
2472 | instead of a subvolume (even on a btrfs file system) if the | |
2473 | root directory is a plain directory, and not a | |
2474 | subvolume. This should simplify things with certain chroot() | |
2475 | environments which are not aware of the concept of btrfs | |
2476 | subvolumes. | |
2477 | ||
2478 | * systemd-detect-virt gained a new --chroot switch to detect | |
2479 | whether execution takes place in a chroot() environment. | |
2480 | ||
2481 | * CPUAffinity= now takes CPU index ranges in addition to | |
2482 | individual indexes. | |
2483 | ||
2484 | * The various memory-related resource limit settings (such as | |
2485 | LimitAS=) now understand the usual K, M, G, ... suffixes to | |
2486 | the base of 1024 (IEC). Similar, the time-related resource | |
2487 | limit settings understand the usual min, h, day, ... | |
2488 | suffixes now. | |
2489 | ||
2490 | * There's a new system.conf setting DefaultTasksMax= to | |
2491 | control the default TasksMax= setting for services and | |
2492 | scopes running on the system. (TasksMax= is the primary | |
2493 | setting that exposes the "pids" cgroup controller on systemd | |
2494 | and was introduced in the previous systemd release.) The | |
2495 | setting now defaults to 512, which means services that are | |
2496 | not explicitly configured otherwise will only be able to | |
2497 | create 512 processes or threads at maximum, from this | |
2498 | version on. Note that this means that thread- or | |
2499 | process-heavy services might need to be reconfigured to set | |
2500 | TasksMax= to a higher value. It is sufficient to set | |
2501 | TasksMax= in these specific unit files to a higher value, or | |
2502 | even "infinity". Similar, there's now a logind.conf setting | |
2503 | UserTasksMax= that defaults to 4096 and limits the total | |
2504 | number of processes or tasks each user may own | |
2505 | concurrently. nspawn containers also have the TasksMax= | |
2506 | value set by default now, to 8192. Note that all of this | |
2507 | only has an effect if the "pids" cgroup controller is | |
2508 | enabled in the kernel. The general benefit of these changes | |
2509 | should be a more robust and safer system, that provides a | |
2510 | certain amount of per-service fork() bomb protection. | |
2511 | ||
2512 | * systemd-nspawn gained the new --network-veth-extra= switch | |
2513 | to define additional and arbitrarily-named virtual Ethernet | |
2514 | links between the host and the container. | |
2515 | ||
2516 | * A new service execution setting PassEnvironment= has been | |
2517 | added that allows importing select environment variables | |
2518 | from PID1's environment block into the environment block of | |
2519 | the service. | |
2520 | ||
2521 | * Timer units gained support for a new RemainAfterElapse= | |
2522 | setting which takes a boolean argument. It defaults to on, | |
2523 | exposing behaviour unchanged to previous releases. If set to | |
2524 | off, timer units are unloaded after they elapsed if they | |
2525 | cannot elapse again. This is particularly useful for | |
2526 | transient timer units, which shall not stay around longer | |
2527 | than until they first elapse. | |
2528 | ||
2529 | * systemd will now bump the net.unix.max_dgram_qlen to 512 by | |
2530 | default now (the kernel default is 16). This is beneficial | |
2531 | for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it | |
2532 | allows substantially larger numbers of queued | |
2533 | datagrams. This should increase the capability of systemd to | |
2534 | parallelize boot-up, as logging and sd_notify() are unlikely | |
2535 | to stall execution anymore. If you need to change the value | |
2536 | from the new defaults, use the usual sysctl.d/ snippets. | |
2537 | ||
2538 | * The compression framing format used by the journal or | |
2539 | coredump processing has changed to be in line with what the | |
2540 | official LZ4 tools generate. LZ4 compression support in | |
2541 | systemd was considered unsupported previously, as the format | |
2542 | was not compatible with the normal tools. With this release | |
2543 | this has changed now, and it is hence safe for downstream | |
2544 | distributions to turn it on. While not compressing as well | |
2545 | as the XZ, LZ4 is substantially faster, which makes | |
2546 | it a good default choice for the compression logic in the | |
2547 | journal and in coredump handling. | |
2548 | ||
2549 | * Any reference to /etc/mtab has been dropped from | |
2550 | systemd. The file has been obsolete since a while, but | |
2551 | systemd refused to work on systems where it was incorrectly | |
2552 | set up (it should be a symlink or non-existent). Please make | |
2553 | sure to update to util-linux 2.27.1 or newer in conjunction | |
2554 | with this systemd release, which also drops any reference to | |
2555 | /etc/mtab. If you maintain a distribution make sure that no | |
2556 | software you package still references it, as this is a | |
2557 | likely source of bugs. There's also a glibc bug pending, | |
2558 | asking for removal of any reference to this obsolete file: | |
2559 | ||
2560 | https://sourceware.org/bugzilla/show_bug.cgi?id=19108 | |
2561 | ||
2562 | Note that only util-linux versions built with | |
2563 | --enable-libmount-force-mountinfo are supported. | |
2564 | ||
2565 | * Support for the ".snapshot" unit type has been removed. This | |
2566 | feature turned out to be little useful and little used, and | |
2567 | has now been removed from the core and from systemctl. | |
2568 | ||
2569 | * The dependency types RequiresOverridable= and | |
2570 | RequisiteOverridable= have been removed from systemd. They | |
2571 | have been used only very sparingly to our knowledge and | |
2572 | other options that provide a similar effect (such as | |
2573 | systemctl --mode=ignore-dependencies) are much more useful | |
2574 | and commonly used. Moreover, they were only half-way | |
2575 | implemented as the option to control behaviour regarding | |
2576 | these dependencies was never added to systemctl. By removing | |
2577 | these dependency types the execution engine becomes a bit | |
2578 | simpler. Unit files that use these dependencies should be | |
2579 | changed to use the non-Overridable dependency types | |
2580 | instead. In fact, when parsing unit files with these | |
2581 | options, that's what systemd will automatically convert them | |
2582 | too, but it will also warn, asking users to fix the unit | |
2583 | files accordingly. Removal of these dependency types should | |
2584 | only affect a negligible number of unit files in the wild. | |
2585 | ||
2586 | * Behaviour of networkd's IPForward= option changed | |
2587 | (again). It will no longer maintain a per-interface setting, | |
2588 | but propagate one way from interfaces where this is enabled | |
2589 | to the global kernel setting. The global setting will be | |
2590 | enabled when requested by a network that is set up, but | |
2591 | never be disabled again. This change was made to make sure | |
2592 | IPv4 and IPv6 behaviour regarding packet forwarding is | |
2593 | similar (as the Linux IPv6 stack does not support | |
2594 | per-interface control of this setting) and to minimize | |
2595 | surprises. | |
2596 | ||
2597 | * In unit files the behaviour of %u, %U, %h, %s has | |
2598 | changed. These specifiers will now unconditionally resolve | |
2599 | to the various user database fields of the user that the | |
2600 | systemd instance is running as, instead of the user | |
2601 | configured in the specific unit via User=. Note that this | |
2602 | effectively doesn't change much, as resolving of these | |
2603 | specifiers was already turned off in the --system instance | |
2604 | of systemd, as we cannot do NSS lookups from PID 1. In the | |
2605 | --user instance of systemd these specifiers where correctly | |
2606 | resolved, but hardly made any sense, since the user instance | |
2607 | lacks privileges to do user switches anyway, and User= is | |
2608 | hence useless. Moreover, even in the --user instance of | |
2609 | systemd behaviour was awkward as it would only take settings | |
2610 | from User= assignment placed before the specifier into | |
2611 | account. In order to unify and simplify the logic around | |
2612 | this the specifiers will now always resolve to the | |
2613 | credentials of the user invoking the manager (which in case | |
2614 | of PID 1 is the root user). | |
2615 | ||
2616 | Contributions from: Andrew Jones, Beniamino Galvani, Boyuan | |
2617 | Yang, Daniel Machon, Daniel Mack, David Herrmann, David | |
2618 | Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin, | |
2619 | Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo | |
2620 | Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan | |
2621 | Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers, | |
2622 | Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel | |
2623 | Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark | |
2624 | Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich, | |
2625 | Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens, | |
2626 | Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer, | |
2627 | Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden, | |
2628 | Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen, | |
2629 | Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew | |
2630 | Jędrzejewski-Szmek | |
2631 | ||
2632 | — Berlin, 2015-11-18 | |
2633 | ||
2634 | CHANGES WITH 227: | |
2635 | ||
2636 | * systemd now depends on util-linux v2.27. More specifically, | |
2637 | the newly added mount monitor feature in libmount now | |
2638 | replaces systemd's former own implementation. | |
2639 | ||
2640 | * libmount mandates /etc/mtab not to be regular file, and | |
2641 | systemd now enforces this condition at early boot. | |
2642 | /etc/mtab has been deprecated and warned about for a very | |
2643 | long time, so systems running systemd should already have | |
2644 | stopped having this file around as anything else than a | |
2645 | symlink to /proc/self/mounts. | |
2646 | ||
2647 | * Support for the "pids" cgroup controller has been added. It | |
2648 | allows accounting the number of tasks in a cgroup and | |
2649 | enforcing limits on it. This adds two new setting | |
2650 | TasksAccounting= and TasksMax= to each unit, as well as a | |
2651 | global option DefaultTasksAccounting=. | |
2652 | ||
2653 | * Support for the "net_cls" cgroup controller has been added. | |
2654 | It allows assigning a net class ID to each task in the | |
2655 | cgroup, which can then be used in firewall rules and traffic | |
2656 | shaping configurations. Note that the kernel netfilter net | |
2657 | class code does not currently work reliably for ingress | |
2658 | packets on unestablished sockets. | |
2659 | ||
2660 | This adds a new config directive called NetClass= to CGroup | |
2661 | enabled units. Allowed values are positive numbers for fixed | |
2662 | assignments and "auto" for picking a free value | |
2663 | automatically. | |
2664 | ||
2665 | * 'systemctl is-system-running' now returns 'offline' if the | |
2666 | system is not booted with systemd. This command can now be | |
2667 | used as a substitute for 'systemd-notify --booted'. | |
2668 | ||
2669 | * Watchdog timeouts have been increased to 3 minutes for all | |
2670 | in-tree service files. Apparently, disk IO issues are more | |
2671 | frequent than we hoped, and user reported >1 minute waiting | |
2672 | for disk IO. | |
2673 | ||
2674 | * 'machine-id-commit' functionality has been merged into | |
2675 | 'machine-id-setup --commit'. The separate binary has been | |
2676 | removed. | |
2677 | ||
2678 | * The WorkingDirectory= directive in unit files may now be set | |
2679 | to the special value '~'. In this case, the working | |
2680 | directory is set to the home directory of the user | |
2681 | configured in User=. | |
2682 | ||
2683 | * "machinectl shell" will now open the shell in the home | |
2684 | directory of the selected user by default. | |
2685 | ||
2686 | * The CrashChVT= configuration file setting is renamed to | |
2687 | CrashChangeVT=, following our usual logic of not | |
2688 | abbreviating unnecessarily. The old directive is still | |
2689 | supported for compat reasons. Also, this directive now takes | |
2690 | an integer value between 1 and 63, or a boolean value. The | |
2691 | formerly supported '-1' value for disabling stays around for | |
2692 | compat reasons. | |
2693 | ||
2694 | * The PrivateTmp=, PrivateDevices=, PrivateNetwork=, | |
2695 | NoNewPrivileges=, TTYPath=, WorkingDirectory= and | |
2696 | RootDirectory= properties can now be set for transient | |
2697 | units. | |
2698 | ||
2699 | * The systemd-analyze tool gained a new "set-log-target" verb | |
2700 | to change the logging target the system manager logs to | |
2701 | dynamically during runtime. This is similar to how | |
2702 | "systemd-analyze set-log-level" already changes the log | |
2703 | level. | |
2704 | ||
2705 | * In nspawn /sys is now mounted as tmpfs, with only a selected | |
2706 | set of subdirectories mounted in from the real sysfs. This | |
2707 | enhances security slightly, and is useful for ensuring user | |
2708 | namespaces work correctly. | |
2709 | ||
2710 | * Support for USB FunctionFS activation has been added. This | |
2711 | allows implementation of USB gadget services that are | |
2712 | activated as soon as they are requested, so that they don't | |
2713 | have to run continuously, similar to classic socket | |
2714 | activation. | |
2715 | ||
2716 | * The "systemctl exit" command now optionally takes an | |
2717 | additional parameter that sets the exit code to return from | |
2718 | the systemd manager when exiting. This is only relevant when | |
2719 | running the systemd user instance, or when running the | |
2720 | system instance in a container. | |
2721 | ||
2722 | * sd-bus gained the new API calls sd_bus_path_encode_many() | |
2723 | and sd_bus_path_decode_many() that allow easy encoding and | |
2724 | decoding of multiple identifier strings inside a D-Bus | |
2725 | object path. Another new call sd_bus_default_flush_close() | |
2726 | has been added to flush and close per-thread default | |
2727 | connections. | |
2728 | ||
2729 | * systemd-cgtop gained support for a -M/--machine= switch to | |
2730 | show the control groups within a certain container only. | |
2731 | ||
2732 | * "systemctl kill" gained support for an optional --fail | |
2733 | switch. If specified the requested operation will fail of no | |
2734 | processes have been killed, because the unit had no | |
2735 | processes attached, or similar. | |
2736 | ||
2737 | * A new systemd.crash_reboot=1 kernel command line option has | |
2738 | been added that triggers a reboot after crashing. This can | |
2739 | also be set through CrashReboot= in systemd.conf. | |
2740 | ||
2741 | * The RuntimeDirectory= setting now understands unit | |
2742 | specifiers like %i or %f. | |
2743 | ||
2744 | * A new (still internal) library API sd-ipv4acd has been added, | |
2745 | that implements address conflict detection for IPv4. It's | |
2746 | based on code from sd-ipv4ll, and will be useful for | |
2747 | detecting DHCP address conflicts. | |
2748 | ||
2749 | * File descriptors passed during socket activation may now be | |
2750 | named. A new API sd_listen_fds_with_names() is added to | |
2751 | access the names. The default names may be overridden, | |
2752 | either in the .socket file using the FileDescriptorName= | |
2753 | parameter, or by passing FDNAME= when storing the file | |
2754 | descriptors using sd_notify(). | |
2755 | ||
2756 | * systemd-networkd gained support for: | |
2757 | ||
2758 | - Setting the IPv6 Router Advertisement settings via | |
2759 | IPv6AcceptRouterAdvertisements= in .network files. | |
2760 | ||
2761 | - Configuring the HelloTimeSec=, MaxAgeSec= and | |
2762 | ForwardDelaySec= bridge parameters in .netdev files. | |
2763 | ||
2764 | - Configuring PreferredSource= for static routes in | |
2765 | .network files. | |
2766 | ||
2767 | * The "ask-password" framework used to query for LUKS harddisk | |
2768 | passwords or SSL passwords during boot gained support for | |
2769 | caching passwords in the kernel keyring, if it is | |
2770 | available. This makes sure that the user only has to type in | |
2771 | a passphrase once if there are multiple objects to unlock | |
2772 | with the same one. Previously, such password caching was | |
2773 | available only when Plymouth was used; this moves the | |
2774 | caching logic into the systemd codebase itself. The | |
2775 | "systemd-ask-password" utility gained a new --keyname= | |
2776 | switch to control which kernel keyring key to use for | |
2777 | caching a password in. This functionality is also useful for | |
2778 | enabling display managers such as gdm to automatically | |
2779 | unlock the user's GNOME keyring if its passphrase, the | |
2780 | user's password and the harddisk password are the same, if | |
2781 | gdm-autologin is used. | |
2782 | ||
2783 | * When downloading tar or raw images using "machinectl | |
2784 | pull-tar" or "machinectl pull-raw", a matching ".nspawn" | |
2785 | file is now also downloaded, if it is available and stored | |
2786 | next to the image file. | |
2787 | ||
2788 | * Units of type ".socket" gained a new boolean setting | |
2789 | Writable= which is only useful in conjunction with | |
2790 | ListenSpecial=. If true, enables opening the specified | |
2791 | special file in O_RDWR mode rather than O_RDONLY mode. | |
2792 | ||
2793 | * systemd-rfkill has been reworked to become a singleton | |
2794 | service that is activated through /dev/rfkill on each rfkill | |
2795 | state change and saves the settings to disk. This way, | |
2796 | systemd-rfkill is now compatible with devices that exist | |
2797 | only intermittendly, and even restores state if the previous | |
2798 | system shutdown was abrupt rather than clean. | |
2799 | ||
2800 | * The journal daemon gained support for vacuuming old journal | |
2801 | files controlled by the number of files that shall remain, | |
2802 | in addition to the already existing control by size and by | |
2803 | date. This is useful as journal interleaving performance | |
2804 | degrades with too many separate journal files, and allows | |
2805 | putting an effective limit on them. The new setting defaults | |
2806 | to 100, but this may be changed by setting SystemMaxFiles= | |
2807 | and RuntimeMaxFiles= in journald.conf. Also, the | |
2808 | "journalctl" tool gained the new --vacuum-files= switch to | |
2809 | manually vacuum journal files to leave only the specified | |
2810 | number of files in place. | |
2811 | ||
2812 | * udev will now create /dev/disk/by-path links for ATA devices | |
2813 | on kernels where that is supported. | |
2814 | ||
2815 | * Galician, Serbian, Turkish and Korean translations were added. | |
2816 | ||
2817 | Contributions from: Aaro Koskinen, Alban Crequy, Beniamino | |
2818 | Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao | |
2819 | (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David | |
2820 | Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin, | |
2821 | Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel | |
2822 | de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner, | |
2823 | Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen, | |
2824 | Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir | |
2825 | Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart | |
2826 | Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, | |
2827 | Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl, | |
2828 | Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike | |
2829 | Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma, | |
2830 | Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer, | |
2831 | Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani, | |
2832 | Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel | |
2833 | Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich, | |
2834 | Zbigniew Jędrzejewski-Szmek, Марко М. Костић | |
2835 | ||
2836 | — Berlin, 2015-10-07 | |
2837 | ||
2838 | CHANGES WITH 226: | |
2839 | ||
2840 | * The DHCP implementation of systemd-networkd gained a set of | |
2841 | new features: | |
2842 | ||
2843 | - The DHCP server now supports emitting DNS and NTP | |
2844 | information. It may be enabled and configured via | |
2845 | EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS | |
2846 | and NTP information is enabled, but no servers are | |
2847 | configured, the corresponding uplink information (if there | |
2848 | is any) is propagated. | |
2849 | ||
2850 | - Server and client now support transmission and reception | |
2851 | of timezone information. It can be configured via the | |
2852 | newly introduced network options UseTimezone=, | |
2853 | EmitTimezone=, and Timezone=. Transmission of timezone | |
2854 | information is enabled between host and containers by | |
2855 | default now: the container will change its local timezone | |
2856 | to what the host has set. | |
2857 | ||
2858 | - Lease timeouts can now be configured via | |
2859 | MaxLeaseTimeSec= and DefaultLeaseTimeSec=. | |
2860 | ||
2861 | - The DHCP server improved on the stability of | |
2862 | leases. Clients are more likely to get the same lease | |
2863 | information back, even if the server loses state. | |
2864 | ||
2865 | - The DHCP server supports two new configuration options to | |
2866 | control the lease address pool metrics, PoolOffset= and | |
2867 | PoolSize=. | |
2868 | ||
2869 | * The encapsulation limit of tunnels in systemd-networkd may | |
2870 | now be configured via 'EncapsulationLimit='. It allows | |
2871 | modifying the maximum additional levels of encapsulation | |
2872 | that are permitted to be prepended to a packet. | |
2873 | ||
2874 | * systemd now supports the concept of user buses replacing | |
2875 | session buses, if used with dbus-1.10 (and enabled via dbus | |
2876 | --enable-user-session). It previously only supported this on | |
2877 | kdbus-enabled systems, and this release expands this to | |
2878 | 'dbus-daemon' systems. | |
2879 | ||
2880 | * systemd-networkd now supports predictable interface names | |
2881 | for virtio devices. | |
2882 | ||
2883 | * systemd now optionally supports the new Linux kernel | |
2884 | "unified" control group hierarchy. If enabled via the kernel | |
2885 | command-line option 'systemd.unified_cgroup_hierarchy=1', | |
2886 | systemd will try to mount the unified cgroup hierarchy | |
2887 | directly on /sys/fs/cgroup. If not enabled, or not | |
2888 | available, systemd will fall back to the legacy cgroup | |
2889 | hierarchy setup, as before. Host system and containers can | |
2890 | mix and match legacy and unified hierarchies as they | |
2891 | wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY | |
2892 | environment variable to individually select the hierarchy to | |
2893 | use for executed containers. By default, nspawn will use the | |
2894 | unified hierarchy for the containers if the host uses the | |
2895 | unified hierarchy, and the legacy hierarchy otherwise. | |
2896 | Please note that at this point the unified hierarchy is an | |
2897 | experimental kernel feature and is likely to change in one | |
2898 | of the next kernel releases. Therefore, it should not be | |
2899 | enabled by default in downstream distributions yet. The | |
2900 | minimum required kernel version for the unified hierarchy to | |
2901 | work is 4.2. Note that when the unified hierarchy is used | |
2902 | for the first time delegated access to controllers is | |
2903 | safe. Because of this systemd-nspawn containers will get | |
2904 | access to controllers now, as will systemd user | |
2905 | sessions. This means containers and user sessions may now | |
2906 | manage their own resources, partitioning up what the system | |
2907 | grants them. | |
2908 | ||
2909 | * A new special scope unit "init.scope" has been introduced | |
2910 | that encapsulates PID 1 of the system. It may be used to | |
2911 | determine resource usage and enforce resource limits on PID | |
2912 | 1 itself. PID 1 hence moved out of the root of the control | |
2913 | group tree. | |
2914 | ||
2915 | * The cgtop tool gained support for filtering out kernel | |
2916 | threads when counting tasks in a control group. Also, the | |
2917 | count of processes is now recursively summed up by | |
2918 | default. Two options -k and --recursive= have been added to | |
2919 | revert to old behaviour. The tool has also been updated to | |
2920 | work correctly in containers now. | |
2921 | ||
2922 | * systemd-nspawn's --bind= and --bind-ro= options have been | |
2923 | extended to allow creation of non-recursive bind mounts. | |
2924 | ||
2925 | * libsystemd gained two new calls sd_pid_get_cgroup() and | |
2926 | sd_peer_get_cgroup() which return the control group path of | |
2927 | a process or peer of a connected AF_UNIX socket. This | |
2928 | function call is particularly useful when implementing | |
2929 | delegated subtrees support in the control group hierarchy. | |
2930 | ||
2931 | * The "sd-event" event loop API of libsystemd now supports | |
2932 | correct dequeuing of real-time signals, without losing | |
2933 | signal events. | |
2934 | ||
2935 | * When systemd requests a PolicyKit decision when managing | |
2936 | units it will now add additional fields to the request, | |
2937 | including unit name and desired operation. This enables more | |
2938 | powerful PolicyKit policies, that make decisions depending | |
2939 | on these parameters. | |
2940 | ||
2941 | * nspawn learnt support for .nspawn settings files, that may | |
2942 | accompany the image files or directories of containers, and | |
2943 | may contain additional settings for the container. This is | |
2944 | an alternative to configuring container parameters via the | |
2945 | nspawn command line. | |
2946 | ||
2947 | Contributions from: Cristian Rodríguez, Daniel Mack, David | |
2948 | Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe | |
2949 | Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan | |
2950 | Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel | |
2951 | Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal | |
2952 | Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin | |
2953 | Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel | |
2954 | Andersen, Tom Gundersen, Torstein Husebø | |
2955 | ||
2956 | — Berlin, 2015-09-08 | |
2957 | ||
2958 | CHANGES WITH 225: | |
2959 | ||
2960 | * machinectl gained a new verb 'shell' which opens a fresh | |
2961 | shell on the target container or the host. It is similar to | |
2962 | the existing 'login' command of machinectl, but spawns the | |
2963 | shell directly without prompting for username or | |
2964 | password. The pseudo machine '.host' now refers to the local | |
2965 | host and is used by default. Hence, 'machinectl shell' can | |
2966 | be used as replacement for 'su -' which spawns a session as | |
2967 | a fresh systemd unit in a way that is fully isolated from | |
2968 | the originating session. | |
2969 | ||
2970 | * systemd-networkd learned to cope with private-zone DHCP | |
2971 | options and allows other programs to query the values. | |
2972 | ||
2973 | * SELinux access control when enabling/disabling units is no | |
2974 | longer enforced with this release. The previous | |
2975 | implementation was incorrect, and a new corrected | |
2976 | implementation is not yet available. As unit file operations | |
2977 | are still protected via PolicyKit and D-Bus policy this is | |
2978 | not a security problem. Yet, distributions which care about | |
2979 | optimal SELinux support should probably not stabilize on | |
2980 | this release. | |
2981 | ||
2982 | * sd-bus gained support for matches of type "arg0has=", that | |
2983 | test for membership of strings in string arrays sent in bus | |
2984 | messages. | |
2985 | ||
2986 | * systemd-resolved now dumps the contents of its DNS and LLMNR | |
2987 | caches to the logs on reception of the SIGUSR1 signal. This | |
2988 | is useful to debug DNS behaviour. | |
2989 | ||
2990 | * The coredumpctl tool gained a new --directory= option to | |
2991 | operate on journal files in a specific directory. | |
2992 | ||
2993 | * "systemctl reboot" and related commands gained a new | |
2994 | "--message=" option which may be used to set a free-text | |
2995 | wall message when shutting down or rebooting the | |
2996 | system. This message is also logged, which is useful for | |
2997 | figuring out the reason for a reboot or shutdown a | |
2998 | posteriori. | |
2999 | ||
3000 | * The "systemd-resolve-host" tool's -i switch now takes | |
3001 | network interface numbers as alternative to interface names. | |
3002 | ||
3003 | * A new unit file setting for services has been introduced: | |
3004 | UtmpMode= allows configuration of how precisely systemd | |
3005 | handles utmp and wtmp entries for the service if this is | |
3006 | enabled. This allows writing services that appear similar to | |
3007 | user sessions in the output of the "w", "who", "last" and | |
3008 | "lastlog" tools. | |
3009 | ||
3010 | * systemd-resolved will now locally synthesize DNS resource | |
3011 | records for the "localhost" and "gateway" domains as well as | |
3012 | the local hostname. This should ensure that clients querying | |
3013 | RRs via resolved will get similar results as those going via | |
3014 | NSS, if nss-myhostname is enabled. | |
3015 | ||
3016 | Contributions from: Alastair Hughes, Alex Crawford, Daniel | |
3017 | Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski, | |
3018 | Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan | |
3019 | Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers, | |
3020 | Kefeng Wang, Lennart Poettering, Major Hayden, Marcel | |
3021 | Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt | |
3022 | Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim, | |
3023 | Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer, | |
3024 | reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings, | |
3025 | Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe | |
3026 | Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts, | |
3027 | WaLyong Cho, Zbigniew Jędrzejewski-Szmek | |
3028 | ||
3029 | — Berlin, 2015-08-27 | |
3030 | ||
3031 | CHANGES WITH 224: | |
3032 | ||
3033 | * The systemd-efi-boot-generator functionality was merged into | |
3034 | systemd-gpt-auto-generator. | |
3035 | ||
3036 | * systemd-networkd now supports Group Policy for vxlan | |
3037 | devices. It can be enabled via the new boolean configuration | |
3038 | option called 'GroupPolicyExtension='. | |
3039 | ||
3040 | Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David | |
3041 | Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart | |
3042 | Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen | |
3043 | ||
3044 | — Berlin, 2015-07-31 | |
3045 | ||
3046 | CHANGES WITH 223: | |
3047 | ||
3048 | * The python-systemd code has been removed from the systemd repository. | |
3049 | A new repository has been created which accommodates the code from | |
3050 | now on, and we kindly ask distributions to create a separate package | |
3051 | for this: https://github.com/systemd/python-systemd | |
3052 | ||
3053 | * The systemd daemon will now reload its main configuration | |
3054 | (/etc/systemd/system.conf) on daemon-reload. | |
3055 | ||
3056 | * sd-dhcp now exposes vendor specific extensions via | |
3057 | sd_dhcp_lease_get_vendor_specific(). | |
3058 | ||
3059 | * systemd-networkd gained a number of new configuration options. | |
3060 | ||
3061 | - A new boolean configuration option for TAP devices called | |
3062 | 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the | |
3063 | device, thus allowing to send and receive GSO packets. | |
3064 | ||
3065 | - A new tunnel configuration option called 'CopyDSCP='. | |
3066 | If enabled, the DSCP field of ip6 tunnels is copied into the | |
3067 | decapsulated packet. | |
3068 | ||
3069 | - A set of boolean bridge configuration options were added. | |
3070 | 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=', | |
3071 | and 'UnicastFlood=' are now parsed by networkd and applied to the | |
3072 | respective bridge link device via the respective IFLA_BRPORT_* | |
3073 | netlink attribute. | |
3074 | ||
3075 | - A new string configuration option to override the hostname sent | |
3076 | to a DHCP server, called 'Hostname='. If set and 'SendHostname=' | |
3077 | is true, networkd will use the configured hostname instead of the | |
3078 | system hostname when sending DHCP requests. | |
3079 | ||
3080 | - A new tunnel configuration option called 'IPv6FlowLabel='. If set, | |
3081 | networkd will configure the IPv6 flow-label of the tunnel device | |
3082 | according to RFC2460. | |
3083 | ||
3084 | - The 'macvtap' virtual network devices are now supported, similar to | |
3085 | the already supported 'macvlan' devices. | |
3086 | ||
3087 | * systemd-resolved now implements RFC5452 to improve resilience against | |
3088 | cache poisoning. Additionally, source port randomization is enabled | |
3089 | by default to further protect against DNS spoofing attacks. | |
3090 | ||
3091 | * nss-mymachines now supports translating UIDs and GIDs of running | |
3092 | containers with user-namespaces enabled. If a container 'foo' | |
3093 | translates a host uid 'UID' to the container uid 'TUID', then | |
3094 | nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID' | |
3095 | (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are | |
3096 | mapped as 'vg-foo-TGID'. | |
3097 | ||
3098 | Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel | |
3099 | Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov, | |
3100 | HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig), | |
3101 | Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers, | |
3102 | Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael | |
3103 | Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim, | |
3104 | Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo, | |
3105 | Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom | |
3106 | Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo, | |
3107 | Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek | |
3108 | ||
3109 | — Berlin, 2015-07-29 | |
3110 | ||
3111 | CHANGES WITH 222: | |
3112 | ||
3113 | * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules. | |
3114 | There are no known issues with current sysfs, and udev does not need | |
3115 | or should be used to work around such bugs. | |
3116 | ||
3117 | * udev does no longer enable USB HID power management. Several reports | |
3118 | indicate, that some devices cannot handle that setting. | |
3119 | ||
3120 | * The udev accelerometer helper was removed. The functionality | |
3121 | is now fully included in iio-sensor-proxy. But this means, | |
3122 | older iio-sensor-proxy versions will no longer provide | |
3123 | accelerometer/orientation data with this systemd version. | |
3124 | Please upgrade iio-sensor-proxy to version 1.0. | |
3125 | ||
3126 | * networkd gained a new configuration option IPv6PrivacyExtensions= | |
3127 | which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions | |
3128 | for Stateless Address") on selected networks. | |
3129 | ||
3130 | * For the sake of fewer build-time dependencies and less code in the | |
3131 | main repository, the python bindings are about to be removed in the | |
3132 | next release. A new repository has been created which accommodates | |
3133 | the code from now on, and we kindly ask distributions to create a | |
3134 | separate package for this. The removal will take place in v223. | |
3135 | ||
3136 | https://github.com/systemd/python-systemd | |
3137 | ||
3138 | Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera, | |
3139 | Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack, | |
3140 | daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric | |
3141 | Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario, | |
3142 | Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens | |
3143 | (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering, | |
3144 | Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal | |
3145 | Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne, | |
3146 | Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein | |
3147 | Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek | |
3148 | ||
3149 | — Berlin, 2015-07-07 | |
3150 | ||
3151 | CHANGES WITH 221: | |
3152 | ||
3153 | * The sd-bus.h and sd-event.h APIs have now been declared | |
3154 | stable and have been added to the official interface of | |
3155 | libsystemd.so. sd-bus implements an alternative D-Bus client | |
3156 | library, that is relatively easy to use, very efficient and | |
3157 | supports both classic D-Bus as well as kdbus as transport | |
3158 | backend. sd-event is a generic event loop abstraction that | |
3159 | is built around Linux epoll, but adds features such as event | |
3160 | prioritization or efficient timer handling. Both APIs are good | |
3161 | choices for C programs looking for a bus and/or event loop | |
3162 | implementation that is minimal and does not have to be | |
3163 | portable to other kernels. | |
3164 | ||
3165 | * kdbus support is no longer compile-time optional. It is now | |
3166 | always built-in. However, it can still be disabled at | |
3167 | runtime using the kdbus=0 kernel command line setting, and | |
3168 | that setting may be changed to default to off, by specifying | |
3169 | --disable-kdbus at build-time. Note though that the kernel | |
3170 | command line setting has no effect if the kdbus.ko kernel | |
3171 | module is not installed, in which case kdbus is (obviously) | |
3172 | also disabled. We encourage all downstream distributions to | |
3173 | begin testing kdbus by adding it to the kernel images in the | |
3174 | development distributions, and leaving kdbus support in | |
3175 | systemd enabled. | |
3176 | ||
3177 | * The minimal required util-linux version has been bumped to | |
3178 | 2.26. | |
3179 | ||
3180 | * Support for chkconfig (--enable-chkconfig) was removed in | |
3181 | favor of calling an abstraction tool | |
3182 | /lib/systemd/systemd-sysv-install. This needs to be | |
3183 | implemented for your distribution. See "SYSV INIT.D SCRIPTS" | |
3184 | in README for details. | |
3185 | ||
3186 | * If there's a systemd unit and a SysV init script for the | |
3187 | same service name, and the user executes "systemctl enable" | |
3188 | for it (or a related call), then this will now enable both | |
3189 | (or execute the related operation on both), not just the | |
3190 | unit. | |
3191 | ||
3192 | * The libudev API documentation has been converted from gtkdoc | |
3193 | into man pages. | |
3194 | ||
3195 | * gudev has been removed from the systemd tree, it is now an | |
3196 | external project. | |
3197 | ||
3198 | * The systemd-cgtop tool learnt a new --raw switch to generate | |
3199 | "raw" (machine parsable) output. | |
3200 | ||
3201 | * networkd's IPForwarding= .network file setting learnt the | |
3202 | new setting "kernel", which ensures that networkd does not | |
3203 | change the IP forwarding sysctl from the default kernel | |
3204 | state. | |
3205 | ||
3206 | * The systemd-logind bus API now exposes a new boolean | |
3207 | property "Docked" that reports whether logind considers the | |
3208 | system "docked", i.e. connected to a docking station or not. | |
3209 | ||
3210 | Contributions from: Alex Crawford, Andreas Pokorny, Andrei | |
3211 | Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez, | |
3212 | Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann, | |
3213 | David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed | |
3214 | Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario, | |
3215 | Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek, | |
3216 | Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang, | |
3217 | Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart | |
3218 | Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario | |
3219 | Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich, | |
3220 | Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes, | |
3221 | Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip | |
3222 | Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani, | |
3223 | Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein | |
3224 | Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner | |
3225 | Fink, Zbigniew Jędrzejewski-Szmek | |
3226 | ||
3227 | — Berlin, 2015-06-19 | |
3228 | ||
3229 | CHANGES WITH 220: | |
3230 | ||
3231 | * The gudev library has been extracted into a separate repository | |
3232 | available at: https://git.gnome.org/browse/libgudev/ | |
3233 | It is now managed as part of the Gnome project. Distributions | |
3234 | are recommended to pass --disable-gudev to systemd and use | |
3235 | gudev from the Gnome project instead. gudev is still included | |
3236 | in systemd, for now. It will be removed soon, though. Please | |
3237 | also see the announcement-thread on systemd-devel: | |
3238 | https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html | |
3239 | ||
3240 | * systemd now exposes a CPUUsageNSec= property for each | |
3241 | service unit on the bus, that contains the overall consumed | |
3242 | CPU time of a service (the sum of what each process of the | |
3243 | service consumed). This value is only available if | |
3244 | CPUAccounting= is turned on for a service, and is then shown | |
3245 | in the "systemctl status" output. | |
3246 | ||
3247 | * Support for configuring alternative mappings of the old SysV | |
3248 | runlevels to systemd targets has been removed. They are now | |
3249 | hardcoded in a way that runlevels 2, 3, 4 all map to | |
3250 | multi-user.target and 5 to graphical.target (which | |
3251 | previously was already the default behaviour). | |
3252 | ||
3253 | * The auto-mounter logic gained support for mount point | |
3254 | expiry, using a new TimeoutIdleSec= setting in .automount | |
3255 | units. (Also available as x-systemd.idle-timeout= in /etc/fstab). | |
3256 | ||
3257 | * The EFI System Partition (ESP) as mounted to /boot by | |
3258 | systemd-efi-boot-generator will now be unmounted | |
3259 | automatically after 2 minutes of not being used. This should | |
3260 | minimize the risk of ESP corruptions. | |
3261 | ||
3262 | * New /etc/fstab options x-systemd.requires= and | |
3263 | x-systemd.requires-mounts-for= are now supported to express | |
3264 | additional dependencies for mounts. This is useful for | |
3265 | journalling file systems that support external journal | |
3266 | devices or overlay file systems that require underlying file | |
3267 | systems to be mounted. | |
3268 | ||
3269 | * systemd does not support direct live-upgrades (via systemctl | |
3270 | daemon-reexec) from versions older than v44 anymore. As no | |
3271 | distribution we are aware of shipped such old versions in a | |
3272 | stable release this should not be problematic. | |
3273 | ||
3274 | * When systemd forks off a new per-connection service instance | |
3275 | it will now set the $REMOTE_ADDR environment variable to the | |
3276 | remote IP address, and $REMOTE_PORT environment variable to | |
3277 | the remote IP port. This behaviour is similar to the | |
3278 | corresponding environment variables defined by CGI. | |
3279 | ||
3280 | * systemd-networkd gained support for uplink failure | |
3281 | detection. The BindCarrier= option allows binding interface | |
3282 | configuration dynamically to the link sense of other | |
3283 | interfaces. This is useful to achieve behaviour like in | |
3284 | network switches. | |
3285 | ||
3286 | * systemd-networkd gained support for configuring the DHCP | |
3287 | client identifier to use when requesting leases. | |
3288 | ||
3289 | * systemd-networkd now has a per-network UseNTP= option to | |
3290 | configure whether NTP server information acquired via DHCP | |
3291 | is passed on to services like systemd-timesyncd. | |
3292 | ||
3293 | * systemd-networkd gained support for vti6 tunnels. | |
3294 | ||
3295 | * Note that systemd-networkd manages the sysctl variable | |
3296 | /proc/sys/net/ipv[46]/conf/*/forwarding for each interface | |
3297 | it is configured for since v219. The variable controls IP | |
3298 | forwarding, and is a per-interface alternative to the global | |
3299 | /proc/sys/net/ipv[46]/ip_forward. This setting is | |
3300 | configurable in the IPForward= option, which defaults to | |
3301 | "no". This means if networkd is used for an interface it is | |
3302 | no longer sufficient to set the global sysctl option to turn | |
3303 | on IP forwarding! Instead, the .network file option | |
3304 | IPForward= needs to be turned on! Note that the | |
3305 | implementation of this behaviour was broken in v219 and has | |
3306 | been fixed in v220. | |
3307 | ||
3308 | * Many bonding and vxlan options are now configurable in | |
3309 | systemd-networkd. | |
3310 | ||
3311 | * systemd-nspawn gained a new --property= setting to set unit | |
3312 | properties for the container scope. This is useful for | |
3313 | setting resource parameters (e.g. "CPUShares=500") on | |
3314 | containers started from the command line. | |
3315 | ||
3316 | * systemd-nspawn gained a new --private-users= switch to make | |
3317 | use of user namespacing available on recent Linux kernels. | |
3318 | ||
3319 | * systemd-nspawn may now be called as part of a shell pipeline | |
3320 | in which case the pipes used for stdin and stdout are passed | |
3321 | directly to the process invoked in the container, without | |
3322 | indirection via a pseudo tty. | |
3323 | ||
3324 | * systemd-nspawn gained a new switch to control the UNIX | |
3325 | signal to use when killing the init process of the container | |
3326 | when shutting down. | |
3327 | ||
3328 | * systemd-nspawn gained a new --overlay= switch for mounting | |
3329 | overlay file systems into the container using the new kernel | |
3330 | overlayfs support. | |
3331 | ||
3332 | * When a container image is imported via systemd-importd and | |
3333 | the host file system is not btrfs, a loopback block device | |
3334 | file is created in /var/lib/machines.raw with a btrfs file | |
3335 | system inside. It is then mounted to /var/lib/machines to | |
3336 | enable btrfs features for container management. The loopback | |
3337 | file and btrfs file system is grown as needed when container | |
3338 | images are imported via systemd-importd. | |
3339 | ||
3340 | * systemd-machined/systemd-importd gained support for btrfs | |
3341 | quota, to enforce container disk space limits on disk. This | |
3342 | is exposed in "machinectl set-limit". | |
3343 | ||
3344 | * systemd-importd now can import containers from local .tar, | |
3345 | .raw and .qcow2 images, and export them to .tar and .raw. It | |
3346 | can also import dkr v2 images now from the network (on top | |
3347 | of v1 as before). | |
3348 | ||
3349 | * systemd-importd gained support for verifying downloaded | |
3350 | images with gpg2 (previously only gpg1 was supported). | |
3351 | ||
3352 | * systemd-machined, systemd-logind, systemd: most bus calls | |
3353 | are now accessible to unprivileged processes via | |
3354 | PolicyKit. Also, systemd-logind will now allow users to kill | |
3355 | their own sessions without further privileges or | |
3356 | authorization. | |
3357 | ||
3358 | * systemd-shutdownd has been removed. This service was | |
3359 | previously responsible for implementing scheduled shutdowns | |
3360 | as exposed in /usr/bin/shutdown's time parameter. This | |
3361 | functionality has now been moved into systemd-logind and is | |
3362 | accessible via a bus interface. | |
3363 | ||
3364 | * "systemctl reboot" gained a new switch --firmware-setup that | |
3365 | can be used to reboot into the EFI firmware setup, if that | |
3366 | is available. systemd-logind now exposes an API on the bus | |
3367 | to trigger such reboots, in case graphical desktop UIs want | |
3368 | to cover this functionality. | |
3369 | ||
3370 | * "systemctl enable", "systemctl disable" and "systemctl mask" | |
3371 | now support a new "--now" switch. If specified the units | |
3372 | that are enabled will also be started, and the ones | |
3373 | disabled/masked also stopped. | |
3374 | ||
3375 | * The Gummiboot EFI boot loader tool has been merged into | |
3376 | systemd, and renamed to "systemd-boot". The bootctl tool has been | |
3377 | updated to support systemd-boot. | |
3378 | ||
3379 | * An EFI kernel stub has been added that may be used to create | |
3380 | kernel EFI binaries that contain not only the actual kernel, | |
3381 | but also an initrd, boot splash, command line and OS release | |
3382 | information. This combined binary can then be signed as a | |
3383 | single image, so that the firmware can verify it all in one | |
3384 | step. systemd-boot has special support for EFI binaries created | |
3385 | like this and can extract OS release information from them | |
3386 | and show them in the boot menu. This functionality is useful | |
3387 | to implement cryptographically verified boot schemes. | |
3388 | ||
3389 | * Optional support has been added to systemd-fsck to pass | |
3390 | fsck's progress report to an AF_UNIX socket in the file | |
3391 | system. | |
3392 | ||
3393 | * udev will no longer create device symlinks for all block | |
3394 | devices by default. A blacklist for excluding special block | |
3395 | devices from this logic has been turned into a whitelist | |
3396 | that requires picking block devices explicitly that require | |
3397 | device symlinks. | |
3398 | ||
3399 | * A new (currently still internal) API sd-device.h has been | |
3400 | added to libsystemd. This modernized API is supposed to | |
3401 | replace libudev eventually. In fact, already much of libudev | |
3402 | is now just a wrapper around sd-device.h. | |
3403 | ||
3404 | * A new hwdb database for storing metadata about pointing | |
3405 | stick devices has been added. | |
3406 | ||
3407 | * systemd-tmpfiles gained support for setting file attributes | |
3408 | similar to the "chattr" tool with new 'h' and 'H' lines. | |
3409 | ||
3410 | * systemd-journald will no longer unconditionally set the | |
3411 | btrfs NOCOW flag on new journal files. This is instead done | |
3412 | with tmpfiles snippet using the new 'h' line type. This | |
3413 | allows easy disabling of this logic, by masking the | |
3414 | journal-nocow.conf tmpfiles file. | |
3415 | ||
3416 | * systemd-journald will now translate audit message types to | |
3417 | human readable identifiers when writing them to the | |
3418 | journal. This should improve readability of audit messages. | |
3419 | ||
3420 | * The LUKS logic gained support for the offset= and skip= | |
3421 | options in /etc/crypttab, as previously implemented by | |
3422 | Debian. | |
3423 | ||
3424 | * /usr/lib/os-release gained a new optional field VARIANT= for | |
3425 | distributions that support multiple variants (such as a | |
3426 | desktop edition, a server edition, ...) | |
3427 | ||
3428 | Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy, | |
3429 | Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin | |
3430 | Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel, | |
3431 | Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž | |
3432 | Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian | |
3433 | Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel | |
3434 | Mack, Daniel Mustieles, daurnimator, Davide Bettio, David | |
3435 | Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov, | |
3436 | Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke, | |
3437 | Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López | |
3438 | Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan | |
3439 | Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John | |
3440 | Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay | |
3441 | Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas | |
3442 | De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz | |
3443 | Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel | |
3444 | Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett, | |
3445 | Michael Biebl, Michael Marineau, Michael Olbrich, Michal | |
3446 | Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik | |
3447 | Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter | |
3448 | Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny | |
3449 | Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick, | |
3450 | Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker, | |
3451 | Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas | |
3452 | Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom | |
3453 | Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will | |
3454 | Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek | |
3455 | ||
3456 | — Berlin, 2015-05-22 | |
3457 | ||
3458 | CHANGES WITH 219: | |
3459 | ||
3460 | * Introduce a new API "sd-hwdb.h" for querying the hardware | |
3461 | metadata database. With this minimal interface one can query | |
3462 | and enumerate the udev hwdb, decoupled from the old libudev | |
3463 | library. libudev's interface for this is now only a wrapper | |
3464 | around sd-hwdb. A new tool systemd-hwdb has been added to | |
3465 | interface with and update the database. | |
3466 | ||
3467 | * When any of systemd's tools copies files (for example due to | |
3468 | tmpfiles' C lines) a btrfs reflink will attempted first, | |
3469 | before bytewise copying is done. | |
3470 | ||
3471 | * systemd-nspawn gained a new --ephemeral switch. When | |
3472 | specified a btrfs snapshot is taken of the container's root | |
3473 | directory, and immediately removed when the container | |
3474 | terminates again. Thus, a container can be started whose | |
3475 | changes never alter the container's root directory, and are | |
3476 | lost on container termination. This switch can also be used | |
3477 | for starting a container off the root file system of the | |
3478 | host without affecting the host OS. This switch is only | |
3479 | available on btrfs file systems. | |
3480 | ||
3481 | * systemd-nspawn gained a new --template= switch. It takes the | |
3482 | path to a container tree to use as template for the tree | |
3483 | specified via --directory=, should that directory be | |
3484 | missing. This allows instantiating containers dynamically, | |
3485 | on first run. This switch is only available on btrfs file | |
3486 | systems. | |
3487 | ||
3488 | * When a .mount unit refers to a mount point on which multiple | |
3489 | mounts are stacked, and the .mount unit is stopped all of | |
3490 | the stacked mount points will now be unmounted until no | |
3491 | mount point remains. | |
3492 | ||
3493 | * systemd now has an explicit notion of supported and | |
3494 | unsupported unit types. Jobs enqueued for unsupported unit | |
3495 | types will now fail with an "unsupported" error code. More | |
3496 | specifically .swap, .automount and .device units are not | |
3497 | supported in containers, .busname units are not supported on | |
3498 | non-kdbus systems. .swap and .automount are also not | |
3499 | supported if their respective kernel compile time options | |
3500 | are disabled. | |
3501 | ||
3502 | * machinectl gained support for two new "copy-from" and | |
3503 | "copy-to" commands for copying files from a running | |
3504 | container to the host or vice versa. | |
3505 | ||
3506 | * machinectl gained support for a new "bind" command to bind | |
3507 | mount host directories into local containers. This is | |
3508 | currently only supported for nspawn containers. | |
3509 | ||
3510 | * networkd gained support for configuring bridge forwarding | |
3511 | database entries (fdb) from .network files. | |
3512 | ||
3513 | * A new tiny daemon "systemd-importd" has been added that can | |
3514 | download container images in tar, raw, qcow2 or dkr formats, | |
3515 | and make them available locally in /var/lib/machines, so | |
3516 | that they can run as nspawn containers. The daemon can GPG | |
3517 | verify the downloads (not supported for dkr, since it has no | |
3518 | provisions for verifying downloads). It will transparently | |
3519 | decompress bz2, xz, gzip compressed downloads if necessary, | |
3520 | and restore sparse files on disk. The daemon uses privilege | |
3521 | separation to ensure the actual download logic runs with | |
3522 | fewer privileges than the daemon itself. machinectl has | |
3523 | gained new commands "pull-tar", "pull-raw" and "pull-dkr" to | |
3524 | make the functionality of importd available to the | |
3525 | user. With this in place the Fedora and Ubuntu "Cloud" | |
3526 | images can be downloaded and booted as containers unmodified | |
3527 | (the Fedora images lack the appropriate GPG signature files | |
3528 | currently, so they cannot be verified, but this will change | |
3529 | soon, hopefully). Note that downloading images is currently | |
3530 | only fully supported on btrfs. | |
3531 | ||
3532 | * machinectl is now able to list container images found in | |
3533 | /var/lib/machines, along with some metadata about sizes of | |
3534 | disk and similar. If the directory is located on btrfs and | |
3535 | quota is enabled, this includes quota display. A new command | |
3536 | "image-status" has been added that shows additional | |
3537 | information about images. | |
3538 | ||
3539 | * machinectl is now able to clone container images | |
3540 | efficiently, if the underlying file system (btrfs) supports | |
3541 | it, with the new "machinectl clone" command. It also | |
3542 | gained commands for renaming and removing images, as well as | |
3543 | marking them read-only or read-write (supported also on | |
3544 | legacy file systems). | |
3545 | ||
3546 | * networkd gained support for collecting LLDP network | |
3547 | announcements, from hardware that supports this. This is | |
3548 | shown in networkctl output. | |
3549 | ||
3550 | * systemd-run gained support for a new -t (--pty) switch for | |
3551 | invoking a binary on a pty whose input and output is | |
3552 | connected to the invoking terminal. This allows executing | |
3553 | processes as system services while interactively | |
3554 | communicating with them via the terminal. Most interestingly | |
3555 | this is supported across container boundaries. Invoking | |
3556 | "systemd-run -t /bin/bash" is an alternative to running a | |
3557 | full login session, the difference being that the former | |
3558 | will not register a session, nor go through the PAM session | |
3559 | setup. | |
3560 | ||
3561 | * tmpfiles gained support for a new "v" line type for creating | |
3562 | btrfs subvolumes. If the underlying file system is a legacy | |
3563 | file system, this automatically degrades to creating a | |
3564 | normal directory. Among others /var/lib/machines is now | |
3565 | created like this at boot, should it be missing. | |
3566 | ||
3567 | * The directory /var/lib/containers/ has been deprecated and | |
3568 | been replaced by /var/lib/machines. The term "machines" has | |
3569 | been used in the systemd context as generic term for both | |
3570 | VMs and containers, and hence appears more appropriate for | |
3571 | this, as the directory can also contain raw images bootable | |
3572 | via qemu/kvm. | |
3573 | ||
3574 | * systemd-nspawn when invoked with -M but without --directory= | |
3575 | or --image= is now capable of searching for the container | |
3576 | root directory, subvolume or disk image automatically, in | |
3577 | /var/lib/machines. systemd-nspawn@.service has been updated | |
3578 | to make use of this, thus allowing it to be used for raw | |
3579 | disk images, too. | |
3580 | ||
3581 | * A new machines.target unit has been introduced that is | |
3582 | supposed to group all containers/VMs invoked as services on | |
3583 | the system. systemd-nspawn@.service has been updated to | |
3584 | integrate with that. | |
3585 | ||
3586 | * machinectl gained a new "start" command, for invoking a | |
3587 | container as a service. "machinectl start foo" is mostly | |
3588 | equivalent to "systemctl start systemd-nspawn@foo.service", | |
3589 | but handles escaping in a nicer way. | |
3590 | ||
3591 | * systemd-nspawn will now mount most of the cgroupfs tree | |
3592 | read-only into each container, with the exception of the | |
3593 | container's own subtree in the name=systemd hierarchy. | |
3594 | ||
3595 | * journald now sets the special FS_NOCOW file flag for its | |
3596 | journal files. This should improve performance on btrfs, by | |
3597 | avoiding heavy fragmentation when journald's write-pattern | |
3598 | is used on COW file systems. It degrades btrfs' data | |
3599 | integrity guarantees for the files to the same levels as for | |
3600 | ext3/ext4 however. This should be OK though as journald does | |
3601 | its own data integrity checks and all its objects are | |
3602 | checksummed on disk. Also, journald should handle btrfs disk | |
3603 | full events a lot more gracefully now, by processing SIGBUS | |
3604 | errors, and not relying on fallocate() anymore. | |
3605 | ||
3606 | * When journald detects that journal files it is writing to | |
3607 | have been deleted it will immediately start new journal | |
3608 | files. | |
3609 | ||
3610 | * systemd now provides a way to store file descriptors | |
3611 | per-service in PID 1. This is useful for daemons to ensure | |
3612 | that fds they require are not lost during a daemon | |
3613 | restart. The fds are passed to the daemon on the next | |
3614 | invocation in the same way socket activation fds are | |
3615 | passed. This is now used by journald to ensure that the | |
3616 | various sockets connected to all the system's stdout/stderr | |
3617 | are not lost when journald is restarted. File descriptors | |
3618 | may be stored in PID 1 via the sd_pid_notify_with_fds() API, | |
3619 | an extension to sd_notify(). Note that a limit is enforced | |
3620 | on the number of fds a service can store in PID 1, and it | |
3621 | defaults to 0, so that no fds may be stored, unless this is | |
3622 | explicitly turned on. | |
3623 | ||
3624 | * The default TERM variable to use for units connected to a | |
3625 | terminal, when no other value is explicitly is set is now | |
3626 | vt220 rather than vt102. This should be fairly safe still, | |
3627 | but allows PgUp/PgDn work. | |
3628 | ||
3629 | * The /etc/crypttab option header= as known from Debian is now | |
3630 | supported. | |
3631 | ||
3632 | * "loginctl user-status" and "loginctl session-status" will | |
3633 | now show the last 10 lines of log messages of the | |
3634 | user/session following the status output. Similar, | |
3635 | "machinectl status" will show the last 10 log lines | |
3636 | associated with a virtual machine or container | |
3637 | service. (Note that this is usually not the log messages | |
3638 | done in the VM/container itself, but simply what the | |
3639 | container manager logs. For nspawn this includes all console | |
3640 | output however.) | |
3641 | ||
3642 | * "loginctl session-status" without further argument will now | |
3643 | show the status of the session of the caller. Similar, | |
3644 | "lock-session", "unlock-session", "activate", | |
3645 | "enable-linger", "disable-linger" may now be called without | |
3646 | session/user parameter in which case they apply to the | |
3647 | caller's session/user. | |
3648 | ||
3649 | * An X11 session scriptlet is now shipped that uploads | |
3650 | $DISPLAY and $XAUTHORITY into the environment of the systemd | |
3651 | --user daemon if a session begins. This should improve | |
3652 | compatibility with X11 enabled applications run as systemd | |
3653 | user services. | |
3654 | ||
3655 | * Generators are now subject to masking via /etc and /run, the | |
3656 | same way as unit files. | |
3657 | ||
3658 | * networkd .network files gained support for configuring | |
3659 | per-link IPv4/IPv6 packet forwarding as well as IPv4 | |
3660 | masquerading. This is by default turned on for veth links to | |
3661 | containers, as registered by systemd-nspawn. This means that | |
3662 | nspawn containers run with --network-veth will now get | |
3663 | automatic routed access to the host's networks without any | |
3664 | further configuration or setup, as long as networkd runs on | |
3665 | the host. | |
3666 | ||
3667 | * systemd-nspawn gained the --port= (-p) switch to expose TCP | |
3668 | or UDP posts of a container on the host. With this in place | |
3669 | it is possible to run containers with private veth links | |
3670 | (--network-veth), and have their functionality exposed on | |
3671 | the host as if their services were running directly on the | |
3672 | host. | |
3673 | ||
3674 | * systemd-nspawn's --network-veth switch now gained a short | |
3675 | version "-n", since with the changes above it is now truly | |
3676 | useful out-of-the-box. The systemd-nspawn@.service has been | |
3677 | updated to make use of it too by default. | |
3678 | ||
3679 | * systemd-nspawn will now maintain a per-image R/W lock, to | |
3680 | ensure that the same image is not started more than once | |
3681 | writable. (It's OK to run an image multiple times | |
3682 | simultaneously in read-only mode.) | |
3683 | ||
3684 | * systemd-nspawn's --image= option is now capable of | |
3685 | dissecting and booting MBR and GPT disk images that contain | |
3686 | only a single active Linux partition. Previously it | |
3687 | supported only GPT disk images with proper GPT type | |
3688 | IDs. This allows running cloud images from major | |
3689 | distributions directly with systemd-nspawn, without | |
3690 | modification. | |
3691 | ||
3692 | * In addition to collecting mouse dpi data in the udev | |
3693 | hardware database, there's now support for collecting angle | |
3694 | information for mouse scroll wheels. The database is | |
3695 | supposed to guarantee similar scrolling behavior on mice | |
3696 | that it knows about. There's also support for collecting | |
3697 | information about Touchpad types. | |
3698 | ||
3699 | * udev's input_id built-in will now also collect touch screen | |
3700 | dimension data and attach it to probed devices. | |
3701 | ||
3702 | * /etc/os-release gained support for a Distribution Privacy | |
3703 | Policy link field. | |
3704 | ||
3705 | * networkd gained support for creating "ipvlan", "gretap", | |
3706 | "ip6gre", "ip6gretap" and "ip6tnl" network devices. | |
3707 | ||
3708 | * systemd-tmpfiles gained support for "a" lines for setting | |
3709 | ACLs on files. | |
3710 | ||
3711 | * systemd-nspawn will now mount /tmp in the container to | |
3712 | tmpfs, automatically. | |
3713 | ||
3714 | * systemd now exposes the memory.usage_in_bytes cgroup | |
3715 | attribute and shows it for each service in the "systemctl | |
3716 | status" output, if available. | |
3717 | ||
3718 | * When the user presses Ctrl-Alt-Del more than 7x within 2s an | |
3719 | immediate reboot is triggered. This useful if shutdown is | |
3720 | hung and is unable to complete, to expedite the | |
3721 | operation. Note that this kind of reboot will still unmount | |
3722 | all file systems, and hence should not result in fsck being | |
3723 | run on next reboot. | |
3724 | ||
3725 | * A .device unit for an optical block device will now be | |
3726 | considered active only when a medium is in the drive. Also, | |
3727 | mount units are now bound to their backing devices thus | |
3728 | triggering automatic unmounting when devices become | |
3729 | unavailable. With this in place systemd will now | |
3730 | automatically unmount left-over mounts when a CD-ROM is | |
3731 | ejected or an USB stick is yanked from the system. | |
3732 | ||
3733 | * networkd-wait-online now has support for waiting for | |
3734 | specific interfaces only (with globbing), and for giving up | |
3735 | after a configurable timeout. | |
3736 | ||
3737 | * networkd now exits when idle. It will be automatically | |
3738 | restarted as soon as interfaces show up, are removed or | |
3739 | change state. networkd will stay around as long as there is | |
3740 | at least one DHCP state machine or similar around, that keep | |
3741 | it non-idle. | |
3742 | ||
3743 | * networkd may now configure IPv6 link-local addressing in | |
3744 | addition to IPv4 link-local addressing. | |
3745 | ||
3746 | * The IPv6 "token" for use in SLAAC may now be configured for | |
3747 | each .network interface in networkd. | |
3748 | ||
3749 | * Routes configured with networkd may now be assigned a scope | |
3750 | in .network files. | |
3751 | ||
3752 | * networkd's [Match] sections now support globbing and lists | |
3753 | of multiple space-separated matches per item. | |
3754 | ||
3755 | Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser, | |
3756 | Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos | |
3757 | Morata Castillo, Chris Atkinson, Chris J. Arges, Christian | |
3758 | Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie, | |
3759 | Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack, | |
3760 | Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald, | |
3761 | Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de | |
3762 | Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan | |
3763 | Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas | |
3764 | Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken | |
3765 | Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian, | |
3766 | Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas, | |
3767 | Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko | |
3768 | Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl, | |
3769 | Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas | |
3770 | Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul | |
3771 | Martin, Peter Hutterer, Peter Mattern, Philippe De Swert, | |
3772 | Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny | |
3773 | Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick, | |
3774 | Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain | |
3775 | Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom | |
3776 | Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar | |
3777 | Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland | |
3778 | Hoffmann, Zbigniew Jędrzejewski-Szmek | |
3779 | ||
3780 | — Berlin, 2015-02-16 | |
3781 | ||
3782 | CHANGES WITH 218: | |
3783 | ||
3784 | * When querying unit file enablement status (for example via | |
3785 | "systemctl is-enabled"), a new state "indirect" is now known | |
3786 | which indicates that a unit might not be enabled itself, but | |
3787 | another unit listed in its Also= setting might be. | |
3788 | ||
3789 | * Similar to the various existing ConditionXYZ= settings for | |
3790 | units, there are now matching AssertXYZ= settings. While | |
3791 | failing conditions cause a unit to be skipped, but its job | |
3792 | to succeed, failing assertions declared like this will cause | |
3793 | a unit start operation and its job to fail. | |
3794 | ||
3795 | * hostnamed now knows a new chassis type "embedded". | |
3796 | ||
3797 | * systemctl gained a new "edit" command. When used on a unit | |
3798 | file, this allows extending unit files with .d/ drop-in | |
3799 | configuration snippets or editing the full file (after | |
3800 | copying it from /usr/lib to /etc). This will invoke the | |
3801 | user's editor (as configured with $EDITOR), and reload the | |
3802 | modified configuration after editing. | |
3803 | ||
3804 | * "systemctl status" now shows the suggested enablement state | |
3805 | for a unit, as declared in the (usually vendor-supplied) | |
3806 | system preset files. | |
3807 | ||
3808 | * nss-myhostname will now resolve the single-label host name | |
3809 | "gateway" to the locally configured default IP routing | |
3810 | gateways, ordered by their metrics. This assigns a stable | |
3811 | name to the used gateways, regardless which ones are | |
3812 | currently configured. Note that the name will only be | |
3813 | resolved after all other name sources (if nss-myhostname is | |
3814 | configured properly) and should hence not negatively impact | |
3815 | systems that use the single-label host name "gateway" in | |
3816 | other contexts. | |
3817 | ||
3818 | * systemd-inhibit now allows filtering by mode when listing | |
3819 | inhibitors. | |
3820 | ||
3821 | * Scope and service units gained a new "Delegate" boolean | |
3822 | property, which, when set, allows processes running inside the | |
3823 | unit to further partition resources. This is primarily | |
3824 | useful for systemd user instances as well as container | |
3825 | managers. | |
3826 | ||
3827 | * journald will now pick up audit messages directly from | |
3828 | the kernel, and log them like any other log message. The | |
3829 | audit fields are split up and fully indexed. This means that | |
3830 | journalctl in many ways is now a (nicer!) alternative to | |
3831 | ausearch, the traditional audit client. Note that this | |
3832 | implements only a minimal audit client. If you want the | |
3833 | special audit modes like reboot-on-log-overflow, please use | |
3834 | the traditional auditd instead, which can be used in | |
3835 | parallel to journald. | |
3836 | ||
3837 | * The ConditionSecurity= unit file option now understands the | |
3838 | special string "audit" to check whether auditing is | |
3839 | available. | |
3840 | ||
3841 | * journalctl gained two new commands --vacuum-size= and | |
3842 | --vacuum-time= to delete old journal files until the | |
3843 | remaining ones take up no more than the specified size on disk, | |
3844 | or are not older than the specified time. | |
3845 | ||
3846 | * A new, native PPPoE library has been added to sd-network, | |
3847 | systemd's library of light-weight networking protocols. This | |
3848 | library will be used in a future version of networkd to | |
3849 | enable PPPoE communication without an external pppd daemon. | |
3850 | ||
3851 | * The busctl tool now understands a new "capture" verb that | |
3852 | works similar to "monitor", but writes a packet capture | |
3853 | trace to STDOUT that can be redirected to a file which is | |
3854 | compatible with libcap's capture file format. This can then | |
3855 | be loaded in Wireshark and similar tools to inspect bus | |
3856 | communication. | |
3857 | ||
3858 | * The busctl tool now understands a new "tree" verb that shows | |
3859 | the object trees of a specific service on the bus, or of all | |
3860 | services. | |
3861 | ||
3862 | * The busctl tool now understands a new "introspect" verb that | |
3863 | shows all interfaces and members of objects on the bus, | |
3864 | including their signature and values. This is particularly | |
3865 | useful to get more information about bus objects shown by | |
3866 | the new "busctl tree" command. | |
3867 | ||
3868 | * The busctl tool now understands new verbs "call", | |
3869 | "set-property" and "get-property" for invoking bus method | |
3870 | calls, setting and getting bus object properties in a | |
3871 | friendly way. | |
3872 | ||
3873 | * busctl gained a new --augment-creds= argument that controls | |
3874 | whether the tool shall augment credential information it | |
3875 | gets from the bus with data from /proc, in a possibly | |
3876 | race-ful way. | |
3877 | ||
3878 | * nspawn's --link-journal= switch gained two new values | |
3879 | "try-guest" and "try-host" that work like "guest" and | |
3880 | "host", but do not fail if the host has no persistent | |
3881 | journalling enabled. -j is now equivalent to | |
3882 | --link-journal=try-guest. | |
3883 | ||
3884 | * macvlan network devices created by nspawn will now have | |
3885 | stable MAC addresses. | |
3886 | ||
3887 | * A new SmackProcessLabel= unit setting has been added, which | |
3888 | controls the SMACK security label processes forked off by | |
3889 | the respective unit shall use. | |
3890 | ||
3891 | * If compiled with --enable-xkbcommon, systemd-localed will | |
3892 | verify x11 keymap settings by compiling the given keymap. It | |
3893 | will spew out warnings if the compilation fails. This | |
3894 | requires libxkbcommon to be installed. | |
3895 | ||
3896 | * When a coredump is collected, a larger number of metadata | |
3897 | fields is now collected and included in the journal records | |
3898 | created for it. More specifically, control group membership, | |
3899 | environment variables, memory maps, working directory, | |
3900 | chroot directory, /proc/$PID/status, and a list of open file | |
3901 | descriptors is now stored in the log entry. | |
3902 | ||
3903 | * The udev hwdb now contains DPI information for mice. For | |
3904 | details see: | |
3905 | ||
3906 | http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html | |
3907 | ||
3908 | * All systemd programs that read standalone configuration | |
3909 | files in /etc now also support a corresponding series of | |
3910 | .conf.d configuration directories in /etc/, /run/, | |
3911 | /usr/local/lib/, /usr/lib/, and (if configured with | |
3912 | --enable-split-usr) /lib/. In particular, the following | |
3913 | configuration files now have corresponding configuration | |
3914 | directories: system.conf user.conf, logind.conf, | |
3915 | journald.conf, sleep.conf, bootchart.conf, coredump.conf, | |
3916 | resolved.conf, timesyncd.conf, journal-remote.conf, and | |
3917 | journal-upload.conf. Note that distributions should use the | |
3918 | configuration directories in /usr/lib/; the directories in | |
3919 | /etc/ are reserved for the system administrator. | |
3920 | ||
3921 | * systemd-rfkill will no longer take the rfkill device name | |
3922 | into account when storing rfkill state on disk, as the name | |
3923 | might be dynamically assigned and not stable. Instead, the | |
3924 | ID_PATH udev variable combined with the rfkill type (wlan, | |
3925 | bluetooth, ...) is used. | |
3926 | ||
3927 | * A new service systemd-machine-id-commit.service has been | |
3928 | added. When used on systems where /etc is read-only during | |
3929 | boot, and /etc/machine-id is not initialized (but an empty | |
3930 | file), this service will copy the temporary machine ID | |
3931 | created as replacement into /etc after the system is fully | |
3932 | booted up. This is useful for systems that are freshly | |
3933 | installed with a non-initialized machine ID, but should get | |
3934 | a fixed machine ID for subsequent boots. | |
3935 | ||
3936 | * networkd's .netdev files now provide a large set of | |
3937 | configuration parameters for VXLAN devices. Similarly, the | |
3938 | bridge port cost parameter is now configurable in .network | |
3939 | files. There's also new support for configuring IP source | |
3940 | routing. networkd .link files gained support for a new | |
3941 | OriginalName= match that is useful to match against the | |
3942 | original interface name the kernel assigned. .network files | |
3943 | may include MTU= and MACAddress= fields for altering the MTU | |
3944 | and MAC address while being connected to a specific network | |
3945 | interface. | |
3946 | ||
3947 | * The LUKS logic gained supported for configuring | |
3948 | UUID-specific key files. There's also new support for naming | |
3949 | LUKS device from the kernel command line, using the new | |
3950 | luks.name= argument. | |
3951 | ||
3952 | * Timer units may now be transiently created via the bus API | |
3953 | (this was previously already available for scope and service | |
3954 | units). In addition it is now possible to create multiple | |
3955 | transient units at the same time with a single bus call. The | |
3956 | "systemd-run" tool has been updated to make use of this for | |
3957 | running commands on a specified time, in at(1)-style. | |
3958 | ||
3959 | * tmpfiles gained support for "t" lines, for assigning | |
3960 | extended attributes to files. Among other uses this may be | |
3961 | used to assign SMACK labels to files. | |
3962 | ||
3963 | Contributions from: Alin Rauta, Alison Chaiken, Andrej | |
3964 | Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris | |
3965 | Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez, | |
3966 | Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave | |
3967 | Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin | |
3968 | Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan | |
3969 | Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe | |
3970 | Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering, | |
3971 | Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas | |
3972 | Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi, | |
3973 | Michael Biebl, Michael Chapman, Michael Marineau, Michal | |
3974 | Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter | |
3975 | Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode, | |
3976 | Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross | |
3977 | Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani, | |
3978 | Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen, | |
3979 | Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert | |
3980 | Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek | |
3981 | ||
3982 | — Berlin, 2014-12-10 | |
3983 | ||
3984 | CHANGES WITH 217: | |
3985 | ||
3986 | * journalctl gained the new options -t/--identifier= to match | |
3987 | on the syslog identifier (aka "tag"), as well as --utc to | |
3988 | show log timestamps in the UTC timezone. journalctl now also | |
3989 | accepts -n/--lines=all to disable line capping in a pager. | |
3990 | ||
3991 | * journalctl gained a new switch, --flush, that synchronously | |
3992 | flushes logs from /run/log/journal to /var/log/journal if | |
3993 | persistent storage is enabled. systemd-journal-flush.service | |
3994 | now waits until the operation is complete. | |
3995 | ||
3996 | * Services can notify the manager before they start a reload | |
3997 | (by sending RELOADING=1) or shutdown (by sending | |
3998 | STOPPING=1). This allows the manager to track and show the | |
3999 | internal state of daemons and closes a race condition when | |
4000 | the process is still running but has closed its D-Bus | |
4001 | connection. | |
4002 | ||
4003 | * Services with Type=oneshot do not have to have any ExecStart | |
4004 | commands anymore. | |
4005 | ||
4006 | * User units are now loaded also from | |
4007 | $XDG_RUNTIME_DIR/systemd/user/. This is similar to the | |
4008 | /run/systemd/user directory that was already previously | |
4009 | supported, but is under the control of the user. | |
4010 | ||
4011 | * Job timeouts (i.e. time-outs on the time a job that is | |
4012 | queued stays in the run queue) can now optionally result in | |
4013 | immediate reboot or power-off actions (JobTimeoutAction= and | |
4014 | JobTimeoutRebootArgument=). This is useful on ".target" | |
4015 | units, to limit the maximum time a target remains | |
4016 | undispatched in the run queue, and to trigger an emergency | |
4017 | operation in such a case. This is now used by default to | |
4018 | turn off the system if boot-up (as defined by everything in | |
4019 | basic.target) hangs and does not complete for at least | |
4020 | 15min. Also, if power-off or reboot hang for at least 30min | |
4021 | an immediate power-off/reboot operation is triggered. This | |
4022 | functionality is particularly useful to increase reliability | |
4023 | on embedded devices, but also on laptops which might | |
4024 | accidentally get powered on when carried in a backpack and | |
4025 | whose boot stays stuck in a hard disk encryption passphrase | |
4026 | question. | |
4027 | ||
4028 | * systemd-logind can be configured to also handle lid switch | |
4029 | events even when the machine is docked or multiple displays | |
4030 | are attached (HandleLidSwitchDocked= option). | |
4031 | ||
4032 | * A helper binary and a service have been added which can be | |
4033 | used to resume from hibernation in the initramfs. A | |
4034 | generator will parse the resume= option on the kernel | |
4035 | command line to trigger resume. | |
4036 | ||
4037 | * A user console daemon systemd-consoled has been | |
4038 | added. Currently, it is a preview, and will so far open a | |
4039 | single terminal on each session of the user marked as | |
4040 | Desktop=systemd-console. | |
4041 | ||
4042 | * Route metrics can be specified for DHCP routes added by | |
4043 | systemd-networkd. | |
4044 | ||
4045 | * The SELinux context of socket-activated services can be set | |
4046 | from the information provided by the networking stack | |
4047 | (SELinuxContextFromNet= option). | |
4048 | ||
4049 | * Userspace firmware loading support has been removed and | |
4050 | the minimum supported kernel version is thus bumped to 3.7. | |
4051 | ||
4052 | * Timeout for udev workers has been increased from 1 to 3 | |
4053 | minutes, but a warning will be printed after 1 minute to | |
4054 | help diagnose kernel modules that take a long time to load. | |
4055 | ||
4056 | * Udev rules can now remove tags on devices with TAG-="foobar". | |
4057 | ||
4058 | * systemd's readahead implementation has been removed. In many | |
4059 | circumstances it didn't give expected benefits even for | |
4060 | rotational disk drives and was becoming less relevant in the | |
4061 | age of SSDs. As none of the developers has been using | |
4062 | rotating media anymore, and nobody stepped up to actively | |
4063 | maintain this component of systemd it has now been removed. | |
4064 | ||
4065 | * Swap units can use Options= to specify discard options. | |
4066 | Discard options specified for swaps in /etc/fstab are now | |
4067 | respected. | |
4068 | ||
4069 | * Docker containers are now detected as a separate type of | |
4070 | virtualization. | |
4071 | ||
4072 | * The Password Agent protocol gained support for queries where | |
4073 | the user input is shown, useful e.g. for user names. | |
4074 | systemd-ask-password gained a new --echo option to turn that | |
4075 | on. | |
4076 | ||
4077 | * The default sysctl.d/ snippets will now set: | |
4078 | ||
4079 | net.core.default_qdisc = fq_codel | |
4080 | ||
4081 | This selects Fair Queuing Controlled Delay as the default | |
4082 | queuing discipline for network interfaces. fq_codel helps | |
4083 | fight the network bufferbloat problem. It is believed to be | |
4084 | a good default with no tuning required for most workloads. | |
4085 | Downstream distributions may override this choice. On 10Gbit | |
4086 | servers that do not do forwarding, "fq" may perform better. | |
4087 | Systems without a good clocksource should use "pfifo_fast". | |
4088 | ||
4089 | * If kdbus is enabled during build a new option BusPolicy= is | |
4090 | available for service units, that allows locking all service | |
4091 | processes into a stricter bus policy, in order to limit | |
4092 | access to various bus services, or even hide most of them | |
4093 | from the service's view entirely. | |
4094 | ||
4095 | * networkctl will now show the .network and .link file | |
4096 | networkd has applied to a specific interface. | |
4097 | ||
4098 | * sd-login gained a new API call sd_session_get_desktop() to | |
4099 | query which desktop environment has been selected for a | |
4100 | session. | |
4101 | ||
4102 | * UNIX utmp support is now compile-time optional to support | |
4103 | legacy-free systems. | |
4104 | ||
4105 | * systemctl gained two new commands "add-wants" and | |
4106 | "add-requires" for pulling in units from specific targets | |
4107 | easily. | |
4108 | ||
4109 | * If the word "rescue" is specified on the kernel command line | |
4110 | the system will now boot into rescue mode (aka | |
4111 | rescue.target), which was previously available only by | |
4112 | specifying "1" or "systemd.unit=rescue.target" on the kernel | |
4113 | command line. This new kernel command line option nicely | |
4114 | mirrors the already existing "emergency" kernel command line | |
4115 | option. | |
4116 | ||
4117 | * New kernel command line options mount.usr=, mount.usrflags=, | |
4118 | mount.usrfstype= have been added that match root=, rootflags=, | |
4119 | rootfstype= but allow mounting a specific file system to | |
4120 | /usr. | |
4121 | ||
4122 | * The $NOTIFY_SOCKET is now also passed to control processes of | |
4123 | services, not only the main process. | |
4124 | ||
4125 | * This version reenables support for fsck's -l switch. This | |
4126 | means at least version v2.25 of util-linux is required for | |
4127 | operation, otherwise dead-locks on device nodes may | |
4128 | occur. Again: you need to update util-linux to at least | |
4129 | v2.25 when updating systemd to v217. | |
4130 | ||
4131 | * The "multi-seat-x" tool has been removed from systemd, as | |
4132 | its functionality has been integrated into X servers 1.16, | |
4133 | and the tool is hence redundant. It is recommended to update | |
4134 | display managers invoking this tool to simply invoke X | |
4135 | directly from now on, again. | |
4136 | ||
4137 | * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus | |
4138 | message flag has been added for all of systemd's PolicyKit | |
4139 | authenticated method calls has been added. In particular | |
4140 | this now allows optional interactive authorization via | |
4141 | PolicyKit for many of PID1's privileged operations such as | |
4142 | unit file enabling and disabling. | |
4143 | ||
4144 | * "udevadm hwdb --update" learnt a new switch "--usr" for | |
4145 | placing the rebuilt hardware database in /usr instead of | |
4146 | /etc. When used only hardware database entries stored in | |
4147 | /usr will be used, and any user database entries in /etc are | |
4148 | ignored. This functionality is useful for vendors to ship a | |
4149 | pre-built database on systems where local configuration is | |
4150 | unnecessary or unlikely. | |
4151 | ||
4152 | * Calendar time specifications in .timer units now also | |
4153 | understand the strings "semi-annually", "quarterly" and | |
4154 | "minutely" as shortcuts (in addition to the preexisting | |
4155 | "anually", "hourly", ...). | |
4156 | ||
4157 | * systemd-tmpfiles will now correctly create files in /dev | |
4158 | at boot which are marked for creation only at boot. It is | |
4159 | recommended to always create static device nodes with 'c!' | |
4160 | and 'b!', so that they are created only at boot and not | |
4161 | overwritten at runtime. | |
4162 | ||
4163 | * When the watchdog logic is used for a service (WatchdogSec=) | |
4164 | and the watchdog timeout is hit the service will now be | |
4165 | terminated with SIGABRT (instead of just SIGTERM), in order | |
4166 | to make sure a proper coredump and backtrace is | |
4167 | generated. This ensures that hanging services will result in | |
4168 | similar coredump/backtrace behaviour as services that hit a | |
4169 | segmentation fault. | |
4170 | ||
4171 | Contributions from: Andreas Henriksson, Andrei Borzenkov, | |
4172 | Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L. | |
4173 | Black, Christian Hesse, Cristian Rodríguez, Daniel Buch, | |
4174 | Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David | |
4175 | Herrmann, David Sommerseth, David Strauss, Emil Renner | |
4176 | Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger, | |
4177 | Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo | |
4178 | Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan | |
4179 | Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus | |
4180 | Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz | |
4181 | Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann, | |
4182 | Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl, | |
4183 | Michael Marineau, Michael Olbrich, Michael Scherer, Michal | |
4184 | Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt, | |
4185 | Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard | |
4186 | Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof, | |
4187 | Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd | |
4188 | Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant | |
4189 | Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen, | |
4190 | Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein | |
4191 | Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew | |
4192 | Jędrzejewski-Szmek | |
4193 | ||
4194 | — Berlin, 2014-10-28 | |
4195 | ||
4196 | CHANGES WITH 216: | |
4197 | ||
4198 | * timedated no longer reads NTP implementation unit names from | |
4199 | /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP | |
4200 | implementations should add a | |
4201 | ||
4202 | Conflicts=systemd-timesyncd.service | |
4203 | ||
4204 | to their unit files to take over and replace systemd's NTP | |
4205 | default functionality. | |
4206 | ||
4207 | * systemd-sysusers gained a new line type "r" for configuring | |
4208 | which UID/GID ranges to allocate system users/groups | |
4209 | from. Lines of type "u" may now add an additional column | |
4210 | that specifies the home directory for the system user to be | |
4211 | created. Also, systemd-sysusers may now optionally read user | |
4212 | information from STDIN instead of a file. This is useful for | |
4213 | invoking it from RPM preinst scriptlets that need to create | |
4214 | users before the first RPM file is installed since these | |
4215 | files might need to be owned by them. A new | |
4216 | %sysusers_create_inline RPM macro has been introduced to do | |
4217 | just that. systemd-sysusers now updates the shadow files as | |
4218 | well as the user/group databases, which should enhance | |
4219 | compatibility with certain tools like grpck. | |
4220 | ||
4221 | * A number of bus APIs of PID 1 now optionally consult | |
4222 | PolicyKit to permit access for otherwise unprivileged | |
4223 | clients under certain conditions. Note that this currently | |
4224 | doesn't support interactive authentication yet, but this is | |
4225 | expected to be added eventually, too. | |
4226 | ||
4227 | * /etc/machine-info now has new fields for configuring the | |
4228 | deployment environment of the machine, as well as the | |
4229 | location of the machine. hostnamectl has been updated with | |
4230 | new command to update these fields. | |
4231 | ||
4232 | * systemd-timesyncd has been updated to automatically acquire | |
4233 | NTP server information from systemd-networkd, which might | |
4234 | have been discovered via DHCP. | |
4235 | ||
4236 | * systemd-resolved now includes a caching DNS stub resolver | |
4237 | and a complete LLMNR name resolution implementation. A new | |
4238 | NSS module "nss-resolve" has been added which can be used | |
4239 | instead of glibc's own "nss-dns" to resolve hostnames via | |
4240 | systemd-resolved. Hostnames, addresses and arbitrary RRs may | |
4241 | be resolved via systemd-resolved D-Bus APIs. In contrast to | |
4242 | the glibc internal resolver systemd-resolved is aware of | |
4243 | multi-homed system, and keeps DNS server and caches separate | |
4244 | and per-interface. Queries are sent simultaneously on all | |
4245 | interfaces that have DNS servers configured, in order to | |
4246 | properly handle VPNs and local LANs which might resolve | |
4247 | separate sets of domain names. systemd-resolved may acquire | |
4248 | DNS server information from systemd-networkd automatically, | |
4249 | which in turn might have discovered them via DHCP. A tool | |
4250 | "systemd-resolve-host" has been added that may be used to | |
4251 | query the DNS logic in resolved. systemd-resolved implements | |
4252 | IDNA and automatically uses IDNA or UTF-8 encoding depending | |
4253 | on whether classic DNS or LLMNR is used as transport. In the | |
4254 | next releases we intend to add a DNSSEC and mDNS/DNS-SD | |
4255 | implementation to systemd-resolved. | |
4256 | ||
4257 | * A new NSS module nss-mymachines has been added, that | |
4258 | automatically resolves the names of all local registered | |
4259 | containers to their respective IP addresses. | |
4260 | ||
4261 | * A new client tool "networkctl" for systemd-networkd has been | |
4262 | added. It currently is entirely passive and will query | |
4263 | networking configuration from udev, rtnetlink and networkd, | |
4264 | and present it to the user in a very friendly | |
4265 | way. Eventually, we hope to extend it to become a full | |
4266 | control utility for networkd. | |
4267 | ||
4268 | * .socket units gained a new DeferAcceptSec= setting that | |
4269 | controls the kernels' TCP_DEFER_ACCEPT sockopt for | |
4270 | TCP. Similarly, support for controlling TCP keep-alive | |
4271 | settings has been added (KeepAliveTimeSec=, | |
4272 | KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for | |
4273 | turning off Nagle's algorithm on TCP has been added | |
4274 | (NoDelay=). | |
4275 | ||
4276 | * logind learned a new session type "web", for use in projects | |
4277 | like Cockpit which register web clients as PAM sessions. | |
4278 | ||
4279 | * timer units with at least one OnCalendar= setting will now | |
4280 | be started only after time-sync.target has been | |
4281 | reached. This way they will not elapse before the system | |
4282 | clock has been corrected by a local NTP client or | |
4283 | similar. This is particular useful on RTC-less embedded | |
4284 | machines, that come up with an invalid system clock. | |
4285 | ||
4286 | * systemd-nspawn's --network-veth= switch should now result in | |
4287 | stable MAC addresses for both the outer and the inner side | |
4288 | of the link. | |
4289 | ||
4290 | * systemd-nspawn gained a new --volatile= switch for running | |
4291 | container instances with /etc or /var unpopulated. | |
4292 | ||
4293 | * The kdbus client code has been updated to use the new Linux | |
4294 | 3.17 memfd subsystem instead of the old kdbus-specific one. | |
4295 | ||
4296 | * systemd-networkd's DHCP client and server now support | |
4297 | FORCERENEW. There are also new configuration options to | |
4298 | configure the vendor client identifier and broadcast mode | |
4299 | for DHCP. | |
4300 | ||
4301 | * systemd will no longer inform the kernel about the current | |
4302 | timezone, as this is necessarily incorrect and racy as the | |
4303 | kernel has no understanding of DST and similar | |
4304 | concepts. This hence means FAT timestamps will be always | |
4305 | considered UTC, similar to what Android is already | |
4306 | doing. Also, when the RTC is configured to the local time | |
4307 | (rather than UTC) systemd will never synchronize back to it, | |
4308 | as this might confuse Windows at a later boot. | |
4309 | ||
4310 | * systemd-analyze gained a new command "verify" for offline | |
4311 | validation of unit files. | |
4312 | ||
4313 | * systemd-networkd gained support for a couple of additional | |
4314 | settings for bonding networking setups. Also, the metric for | |
4315 | statically configured routes may now be configured. For | |
4316 | network interfaces where this is appropriate the peer IP | |
4317 | address may now be configured. | |
4318 | ||
4319 | * systemd-networkd's DHCP client will no longer request | |
4320 | broadcasting by default, as this tripped up some networks. | |
4321 | For hardware where broadcast is required the feature should | |
4322 | be switched back on using RequestBroadcast=yes. | |
4323 | ||
4324 | * systemd-networkd will now set up IPv4LL addresses (when | |
4325 | enabled) even if DHCP is configured successfully. | |
4326 | ||
4327 | * udev will now default to respect network device names given | |
4328 | by the kernel when the kernel indicates that these are | |
4329 | predictable. This behavior can be tweaked by changing | |
4330 | NamePolicy= in the relevant .link file. | |
4331 | ||
4332 | * A new library systemd-terminal has been added that | |
4333 | implements full TTY stream parsing and rendering. This | |
4334 | library is supposed to be used later on for implementing a | |
4335 | full userspace VT subsystem, replacing the current kernel | |
4336 | implementation. | |
4337 | ||
4338 | * A new tool systemd-journal-upload has been added to push | |
4339 | journal data to a remote system running | |
4340 | systemd-journal-remote. | |
4341 | ||
4342 | * journald will no longer forward all local data to another | |
4343 | running syslog daemon. This change has been made because | |
4344 | rsyslog (which appears to be the most commonly used syslog | |
4345 | implementation these days) no longer makes use of this, and | |
4346 | instead pulls the data out of the journal on its own. Since | |
4347 | forwarding the messages to a non-existent syslog server is | |
4348 | more expensive than we assumed we have now turned this | |
4349 | off. If you run a syslog server that is not a recent rsyslog | |
4350 | version, you have to turn this option on again | |
4351 | (ForwardToSyslog= in journald.conf). | |
4352 | ||
4353 | * journald now optionally supports the LZ4 compressor for | |
4354 | larger journal fields. This compressor should perform much | |
4355 | better than XZ which was the previous default. | |
4356 | ||
4357 | * machinectl now shows the IP addresses of local containers, | |
4358 | if it knows them, plus the interface name of the container. | |
4359 | ||
4360 | * A new tool "systemd-escape" has been added that makes it | |
4361 | easy to escape strings to build unit names and similar. | |
4362 | ||
4363 | * sd_notify() messages may now include a new ERRNO= field | |
4364 | which is parsed and collected by systemd and shown among the | |
4365 | "systemctl status" output for a service. | |
4366 | ||
4367 | * A new component "systemd-firstboot" has been added that | |
4368 | queries the most basic systemd information (timezone, | |
4369 | hostname, root password) interactively on first | |
4370 | boot. Alternatively it may also be used to provision these | |
4371 | things offline on OS images installed into directories. | |
4372 | ||
4373 | * The default sysctl.d/ snippets will now set | |
4374 | ||
4375 | net.ipv4.conf.default.promote_secondaries=1 | |
4376 | ||
4377 | This has the benefit of no flushing secondary IP addresses | |
4378 | when primary addresses are removed. | |
4379 | ||
4380 | Contributions from: Ansgar Burchardt, Bastien Nocera, Colin | |
4381 | Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel | |
4382 | Mack, Dan Williams, Dave Reisner, David Herrmann, Denis | |
4383 | Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald | |
4384 | Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann | |
4385 | B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin | |
4386 | Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, | |
4387 | Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael | |
4388 | Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar, | |
4389 | Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert | |
4390 | Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef | |
4391 | Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas | |
4392 | Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets, | |
4393 | Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut | |
4394 | Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek | |
4395 | ||
4396 | — Berlin, 2014-08-19 | |
4397 | ||
4398 | CHANGES WITH 215: | |
4399 | ||
4400 | * A new tool systemd-sysusers has been added. This tool | |
4401 | creates system users and groups in /etc/passwd and | |
4402 | /etc/group, based on static declarative system user/group | |
4403 | definitions in /usr/lib/sysusers.d/. This is useful to | |
4404 | enable factory resets and volatile systems that boot up with | |
4405 | an empty /etc directory, and thus need system users and | |
4406 | groups created during early boot. systemd now also ships | |
4407 | with two default sysusers.d/ files for the most basic | |
4408 | users and groups systemd and the core operating system | |
4409 | require. | |
4410 | ||
4411 | * A new tmpfiles snippet has been added that rebuilds the | |
4412 | essential files in /etc on boot, should they be missing. | |
4413 | ||
4414 | * A directive for ensuring automatic clean-up of | |
4415 | /var/cache/man/ has been removed from the default | |
4416 | configuration. This line should now be shipped by the man | |
4417 | implementation. The necessary change has been made to the | |
4418 | man-db implementation. Note that you need to update your man | |
4419 | implementation to one that ships this line, otherwise no | |
4420 | automatic clean-up of /var/cache/man will take place. | |
4421 | ||
4422 | * A new condition ConditionNeedsUpdate= has been added that | |
4423 | may conditionalize services to only run when /etc or /var | |
4424 | are "older" than the vendor operating system resources in | |
4425 | /usr. This is useful for reconstructing or updating /etc | |
4426 | after an offline update of /usr or a factory reset, on the | |
4427 | next reboot. Services that want to run once after such an | |
4428 | update or reset should use this condition and order | |
4429 | themselves before the new systemd-update-done.service, which | |
4430 | will mark the two directories as fully updated. A number of | |
4431 | service files have been added making use of this, to rebuild | |
4432 | the udev hardware database, the journald message catalog and | |
4433 | dynamic loader cache (ldconfig). The systemd-sysusers tool | |
4434 | described above also makes use of this now. With this in | |
4435 | place it is now possible to start up a minimal operating | |
4436 | system with /etc empty cleanly. For more information on the | |
4437 | concepts involved see this recent blog story: | |
4438 | ||
4439 | http://0pointer.de/blog/projects/stateless.html | |
4440 | ||
4441 | * A new system group "input" has been introduced, and all | |
4442 | input device nodes get this group assigned. This is useful | |
4443 | for system-level software to get access to input devices. It | |
4444 | complements what is already done for "audio" and "video". | |
4445 | ||
4446 | * systemd-networkd learnt minimal DHCPv4 server support in | |
4447 | addition to the existing DHCPv4 client support. It also | |
4448 | learnt DHCPv6 client and IPv6 Router Solicitation client | |
4449 | support. The DHCPv4 client gained support for static routes | |
4450 | passed in from the server. Note that the [DHCPv4] section | |
4451 | known in older systemd-networkd versions has been renamed to | |
4452 | [DHCP] and is now also used by the DHCPv6 client. Existing | |
4453 | .network files using settings of this section should be | |
4454 | updated, though compatibility is maintained. Optionally, the | |
4455 | client hostname may now be sent to the DHCP server. | |
4456 | ||
4457 | * networkd gained support for vxlan virtual networks as well | |
4458 | as tun/tap and dummy devices. | |
4459 | ||
4460 | * networkd gained support for automatic allocation of address | |
4461 | ranges for interfaces from a system-wide pool of | |
4462 | addresses. This is useful for dynamically managing a large | |
4463 | number of interfaces with a single network configuration | |
4464 | file. In particular this is useful to easily assign | |
4465 | appropriate IP addresses to the veth links of a large number | |
4466 | of nspawn instances. | |
4467 | ||
4468 | * RPM macros for processing sysusers, sysctl and binfmt | |
4469 | drop-in snippets at package installation time have been | |
4470 | added. | |
4471 | ||
4472 | * The /etc/os-release file should now be placed in | |
4473 | /usr/lib/os-release. The old location is automatically | |
4474 | created as symlink. /usr/lib is the more appropriate | |
4475 | location of this file, since it shall actually describe the | |
4476 | vendor operating system shipped in /usr, and not the | |
4477 | configuration stored in /etc. | |
4478 | ||
4479 | * .mount units gained a new boolean SloppyOptions= setting | |
4480 | that maps to mount(8)'s -s option which enables permissive | |
4481 | parsing of unknown mount options. | |
4482 | ||
4483 | * tmpfiles learnt a new "L+" directive which creates a symlink | |
4484 | but (unlike "L") deletes a pre-existing file first, should | |
4485 | it already exist and not already be the correct | |
4486 | symlink. Similarly, "b+", "c+" and "p+" directives have been | |
4487 | added as well, which create block and character devices, as | |
4488 | well as fifos in the filesystem, possibly removing any | |
4489 | pre-existing files of different types. | |
4490 | ||
4491 | * For tmpfiles' "L", "L+", "C" and "C+" directives the final | |
4492 | 'argument' field (which so far specified the source to | |
4493 | symlink/copy the files from) is now optional. If omitted the | |
4494 | same file os copied from /usr/share/factory/ suffixed by the | |
4495 | full destination path. This is useful for populating /etc | |
4496 | with essential files, by copying them from vendor defaults | |
4497 | shipped in /usr/share/factory/etc. | |
4498 | ||
4499 | * A new command "systemctl preset-all" has been added that | |
4500 | applies the service preset settings to all installed unit | |
4501 | files. A new switch --preset-mode= has been added that | |
4502 | controls whether only enable or only disable operations | |
4503 | shall be executed. | |
4504 | ||
4505 | * A new command "systemctl is-system-running" has been added | |
4506 | that allows checking the overall state of the system, for | |
4507 | example whether it is fully up and running. | |
4508 | ||
4509 | * When the system boots up with an empty /etc, the equivalent | |
4510 | to "systemctl preset-all" is executed during early boot, to | |
4511 | make sure all default services are enabled after a factory | |
4512 | reset. | |
4513 | ||
4514 | * systemd now contains a minimal preset file that enables the | |
4515 | most basic services systemd ships by default. | |
4516 | ||
4517 | * Unit files' [Install] section gained a new DefaultInstance= | |
4518 | field for defining the default instance to create if a | |
4519 | template unit is enabled with no instance specified. | |
4520 | ||
4521 | * A new passive target cryptsetup-pre.target has been added | |
4522 | that may be used by services that need to make they run and | |
4523 | finish before the first LUKS cryptographic device is set up. | |
4524 | ||
4525 | * The /dev/loop-control and /dev/btrfs-control device nodes | |
4526 | are now owned by the "disk" group by default, opening up | |
4527 | access to this group. | |
4528 | ||
4529 | * systemd-coredump will now automatically generate a | |
4530 | stack trace of all core dumps taking place on the system, | |
4531 | based on elfutils' libdw library. This stack trace is logged | |
4532 | to the journal. | |
4533 | ||
4534 | * systemd-coredump may now optionally store coredumps directly | |
4535 | on disk (in /var/lib/systemd/coredump, possibly compressed), | |
4536 | instead of storing them unconditionally in the journal. This | |
4537 | mode is the new default. A new configuration file | |
4538 | /etc/systemd/coredump.conf has been added to configure this | |
4539 | and other parameters of systemd-coredump. | |
4540 | ||
4541 | * coredumpctl gained a new "info" verb to show details about a | |
4542 | specific coredump. A new switch "-1" has also been added | |
4543 | that makes sure to only show information about the most | |
4544 | recent entry instead of all entries. Also, as the tool is | |
4545 | generally useful now the "systemd-" prefix of the binary | |
4546 | name has been removed. Distributions that want to maintain | |
4547 | compatibility with the old name should add a symlink from | |
4548 | the old name to the new name. | |
4549 | ||
4550 | * journald's SplitMode= now defaults to "uid". This makes sure | |
4551 | that unprivileged users can access their own coredumps with | |
4552 | coredumpctl without restrictions. | |
4553 | ||
4554 | * New kernel command line options "systemd.wants=" (for | |
4555 | pulling an additional unit during boot), "systemd.mask=" | |
4556 | (for masking a specific unit for the boot), and | |
4557 | "systemd.debug-shell" (for enabling the debug shell on tty9) | |
4558 | have been added. This is implemented in the new generator | |
4559 | "systemd-debug-generator". | |
4560 | ||
4561 | * systemd-nspawn will now by default filter a couple of | |
4562 | syscalls for containers, among them those required for | |
4563 | kernel module loading, direct x86 IO port access, swap | |
4564 | management, and kexec. Most importantly though | |
4565 | open_by_handle_at() is now prohibited for containers, | |
4566 | closing a hole similar to a recently discussed vulnerability | |
4567 | in docker regarding access to files on file hierarchies the | |
4568 | container should normally not have access to. Note that, for | |
4569 | nspawn, we generally make no security claims anyway (and | |
4570 | this is explicitly documented in the man page), so this is | |
4571 | just a fix for one of the most obvious problems. | |
4572 | ||
4573 | * A new man page file-hierarchy(7) has been added that | |
4574 | contains a minimized, modernized version of the file system | |
4575 | layout systemd expects, similar in style to the FHS | |
4576 | specification or hier(5). A new tool systemd-path(1) has | |
4577 | been added to query many of these paths for the local | |
4578 | machine and user. | |
4579 | ||
4580 | * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no | |
4581 | longer done. Since the directory now has a per-user size | |
4582 | limit, and is cleaned on logout this appears unnecessary, | |
4583 | in particular since this now brings the lifecycle of this | |
4584 | directory closer in line with how IPC objects are handled. | |
4585 | ||
4586 | * systemd.pc now exports a number of additional directories, | |
4587 | including $libdir (which is useful to identify the library | |
4588 | path for the primary architecture of the system), and a | |
4589 | couple of drop-in directories. | |
4590 | ||
4591 | * udev's predictable network interface names now use the dev_port | |
4592 | sysfs attribute, introduced in linux 3.15 instead of dev_id to | |
4593 | distinguish between ports of the same PCI function. dev_id should | |
4594 | only be used for ports using the same HW address, hence the need | |
4595 | for dev_port. | |
4596 | ||
4597 | * machined has been updated to export the OS version of a | |
4598 | container (read from /etc/os-release and | |
4599 | /usr/lib/os-release) on the bus. This is now shown in | |
4600 | "machinectl status" for a machine. | |
4601 | ||
4602 | * A new service setting RestartForceExitStatus= has been | |
4603 | added. If configured to a set of exit signals or process | |
4604 | return values, the service will be restarted when the main | |
4605 | daemon process exits with any of them, regardless of the | |
4606 | Restart= setting. | |
4607 | ||
4608 | * systemctl's -H switch for connecting to remote systemd | |
4609 | machines has been extended so that it may be used to | |
4610 | directly connect to a specific container on the | |
4611 | host. "systemctl -H root@foobar:waldi" will now connect as | |
4612 | user "root" to host "foobar", and then proceed directly to | |
4613 | the container named "waldi". Note that currently you have to | |
4614 | authenticate as user "root" for this to work, as entering | |
4615 | containers is a privileged operation. | |
4616 | ||
4617 | Contributions from: Andreas Henriksson, Benjamin Steinwender, | |
4618 | Carl Schaefer, Christian Hesse, Colin Ian King, Cristian | |
4619 | Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene | |
4620 | Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo | |
4621 | Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart | |
4622 | Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine | |
4623 | Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich, | |
4624 | Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le | |
4625 | Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan, | |
4626 | Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe | |
4627 | Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar | |
4628 | Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek | |
4629 | ||
4630 | — Berlin, 2014-07-03 | |
4631 | ||
4632 | CHANGES WITH 214: | |
4633 | ||
4634 | * As an experimental feature, udev now tries to lock the | |
4635 | disk device node (flock(LOCK_SH|LOCK_NB)) while it | |
4636 | executes events for the disk or any of its partitions. | |
4637 | Applications like partitioning programs can lock the | |
4638 | disk device node (flock(LOCK_EX)) and claim temporary | |
4639 | device ownership that way; udev will entirely skip all event | |
4640 | handling for this disk and its partitions. If the disk | |
4641 | was opened for writing, the close will trigger a partition | |
4642 | table rescan in udev's "watch" facility, and if needed | |
4643 | synthesize "change" events for the disk and all its partitions. | |
4644 | This is now unconditionally enabled, and if it turns out to | |
4645 | cause major problems, we might turn it on only for specific | |
4646 | devices, or might need to disable it entirely. Device Mapper | |
4647 | devices are excluded from this logic. | |
4648 | ||
4649 | * We temporarily dropped the "-l" switch for fsck invocations, | |
4650 | since they collide with the flock() logic above. util-linux | |
4651 | upstream has been changed already to avoid this conflict, | |
4652 | and we will readd "-l" as soon as util-linux with this | |
4653 | change has been released. | |
4654 | ||
4655 | * The dependency on libattr has been removed. Since a long | |
4656 | time, the extended attribute calls have moved to glibc, and | |
4657 | libattr is thus unnecessary. | |
4658 | ||
4659 | * Virtualization detection works without privileges now. This | |
4660 | means the systemd-detect-virt binary no longer requires | |
4661 | CAP_SYS_PTRACE file capabilities, and our daemons can run | |
4662 | with fewer privileges. | |
4663 | ||
4664 | * systemd-networkd now runs under its own "systemd-network" | |
4665 | user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE, | |
4666 | CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but | |
4667 | loses the ability to write to files owned by root this way. | |
4668 | ||
4669 | * Similarly, systemd-resolved now runs under its own | |
4670 | "systemd-resolve" user with no capabilities remaining. | |
4671 | ||
4672 | * Similarly, systemd-bus-proxyd now runs under its own | |
4673 | "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining. | |
4674 | ||
4675 | * systemd-networkd gained support for setting up "veth" | |
4676 | virtual Ethernet devices for container connectivity, as well | |
4677 | as GRE and VTI tunnels. | |
4678 | ||
4679 | * systemd-networkd will no longer automatically attempt to | |
4680 | manually load kernel modules necessary for certain tunnel | |
4681 | transports. Instead, it is assumed the kernel loads them | |
4682 | automatically when required. This only works correctly on | |
4683 | very new kernels. On older kernels, please consider adding | |
4684 | the kernel modules to /etc/modules-load.d/ as a work-around. | |
4685 | ||
4686 | * The resolv.conf file systemd-resolved generates has been | |
4687 | moved to /run/systemd/resolve/. If you have a symlink from | |
4688 | /etc/resolv.conf, it might be necessary to correct it. | |
4689 | ||
4690 | * Two new service settings, ProtectHome= and ProtectSystem=, | |
4691 | have been added. When enabled, they will make the user data | |
4692 | (such as /home) inaccessible or read-only and the system | |
4693 | (such as /usr) read-only, for specific services. This allows | |
4694 | very light-weight per-service sandboxing to avoid | |
4695 | modifications of user data or system files from | |
4696 | services. These two new switches have been enabled for all | |
4697 | of systemd's long-running services, where appropriate. | |
4698 | ||
4699 | * Socket units gained new SocketUser= and SocketGroup= | |
4700 | settings to set the owner user and group of AF_UNIX sockets | |
4701 | and FIFOs in the file system. | |
4702 | ||
4703 | * Socket units gained a new RemoveOnStop= setting. If enabled, | |
4704 | all FIFOS and sockets in the file system will be removed | |
4705 | when the specific socket unit is stopped. | |
4706 | ||
4707 | * Socket units gained a new Symlinks= setting. It takes a list | |
4708 | of symlinks to create to file system sockets or FIFOs | |
4709 | created by the specific Unix sockets. This is useful to | |
4710 | manage symlinks to socket nodes with the same life-cycle as | |
4711 | the socket itself. | |
4712 | ||
4713 | * The /dev/log socket and /dev/initctl FIFO have been moved to | |
4714 | /run, and have been replaced by symlinks. This allows | |
4715 | connecting to these facilities even if PrivateDevices=yes is | |
4716 | used for a service (which makes /dev/log itself unavailable, | |
4717 | but /run is left). This also has the benefit of ensuring | |
4718 | that /dev only contains device nodes, directories and | |
4719 | symlinks, and nothing else. | |
4720 | ||
4721 | * sd-daemon gained two new calls sd_pid_notify() and | |
4722 | sd_pid_notifyf(). They are similar to sd_notify() and | |
4723 | sd_notifyf(), but allow overriding of the source PID of | |
4724 | notification messages if permissions permit this. This is | |
4725 | useful to send notify messages on behalf of a different | |
4726 | process (for example, the parent process). The | |
4727 | systemd-notify tool has been updated to make use of this | |
4728 | when sending messages (so that notification messages now | |
4729 | originate from the shell script invoking systemd-notify and | |
4730 | not the systemd-notify process itself. This should minimize | |
4731 | a race where systemd fails to associate notification | |
4732 | messages to services when the originating process already | |
4733 | vanished. | |
4734 | ||
4735 | * A new "on-abnormal" setting for Restart= has been added. If | |
4736 | set, it will result in automatic restarts on all "abnormal" | |
4737 | reasons for a process to exit, which includes unclean | |
4738 | signals, core dumps, timeouts and watchdog timeouts, but | |
4739 | does not include clean and unclean exit codes or clean | |
4740 | signals. Restart=on-abnormal is an alternative for | |
4741 | Restart=on-failure for services that shall be able to | |
4742 | terminate and avoid restarts on certain errors, by | |
4743 | indicating so with an unclean exit code. Restart=on-failure | |
4744 | or Restart=on-abnormal is now the recommended setting for | |
4745 | all long-running services. | |
4746 | ||
4747 | * If the InaccessibleDirectories= service setting points to a | |
4748 | mount point (or if there are any submounts contained within | |
4749 | it), it is now attempted to completely unmount it, to make | |
4750 | the file systems truly unavailable for the respective | |
4751 | service. | |
4752 | ||
4753 | * The ReadOnlyDirectories= service setting and | |
4754 | systemd-nspawn's --read-only parameter are now recursively | |
4755 | applied to all submounts, too. | |
4756 | ||
4757 | * Mount units may now be created transiently via the bus APIs. | |
4758 | ||
4759 | * The support for SysV and LSB init scripts has been removed | |
4760 | from the systemd daemon itself. Instead, it is now | |
4761 | implemented as a generator that creates native systemd units | |
4762 | from these scripts when needed. This enables us to remove a | |
4763 | substantial amount of legacy code from PID 1, following the | |
4764 | fact that many distributions only ship a very small number | |
4765 | of LSB/SysV init scripts nowadays. | |
4766 | ||
4767 | * Privileged Xen (dom0) domains are not considered | |
4768 | virtualization anymore by the virtualization detection | |
4769 | logic. After all, they generally have unrestricted access to | |
4770 | the hardware and usually are used to manage the unprivileged | |
4771 | (domU) domains. | |
4772 | ||
4773 | * systemd-tmpfiles gained a new "C" line type, for copying | |
4774 | files or entire directories. | |
4775 | ||
4776 | * systemd-tmpfiles "m" lines are now fully equivalent to "z" | |
4777 | lines. So far, they have been non-globbing versions of the | |
4778 | latter, and have thus been redundant. In future, it is | |
4779 | recommended to only use "z". "m" has hence been removed | |
4780 | from the documentation, even though it stays supported. | |
4781 | ||
4782 | * A tmpfiles snippet to recreate the most basic structure in | |
4783 | /var has been added. This is enough to create the /var/run → | |
4784 | /run symlink and create a couple of structural | |
4785 | directories. This allows systems to boot up with an empty or | |
4786 | volatile /var. Of course, while with this change, the core OS | |
4787 | now is capable with dealing with a volatile /var, not all | |
4788 | user services are ready for it. However, we hope that sooner | |
4789 | or later, many service daemons will be changed upstream so | |
4790 | that they are able to automatically create their necessary | |
4791 | directories in /var at boot, should they be missing. This is | |
4792 | the first step to allow state-less systems that only require | |
4793 | the vendor image for /usr to boot. | |
4794 | ||
4795 | * systemd-nspawn has gained a new --tmpfs= switch to mount an | |
4796 | empty tmpfs instance to a specific directory. This is | |
4797 | particularly useful for making use of the automatic | |
4798 | reconstruction of /var (see above), by passing --tmpfs=/var. | |
4799 | ||
4800 | * Access modes specified in tmpfiles snippets may now be | |
4801 | prefixed with "~", which indicates that they shall be masked | |
4802 | by whether the existing file or directory is currently | |
4803 | writable, readable or executable at all. Also, if specified, | |
4804 | the sgid/suid/sticky bits will be masked for all | |
4805 | non-directories. | |
4806 | ||
4807 | * A new passive target unit "network-pre.target" has been | |
4808 | added which is useful for services that shall run before any | |
4809 | network is configured, for example firewall scripts. | |
4810 | ||
4811 | * The "floppy" group that previously owned the /dev/fd* | |
4812 | devices is no longer used. The "disk" group is now used | |
4813 | instead. Distributions should probably deprecate usage of | |
4814 | this group. | |
4815 | ||
4816 | Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian | |
4817 | King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David | |
4818 | Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers, | |
4819 | Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny | |
4820 | Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel | |
4821 | Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew | |
4822 | Jędrzejewski-Szmek | |
4823 | ||
4824 | — Berlin, 2014-06-11 | |
4825 | ||
4826 | CHANGES WITH 213: | |
4827 | ||
4828 | * A new "systemd-timesyncd" daemon has been added for | |
4829 | synchronizing the system clock across the network. It | |
4830 | implements an SNTP client. In contrast to NTP | |
4831 | implementations such as chrony or the NTP reference server, | |
4832 | this only implements a client side, and does not bother with | |
4833 | the full NTP complexity, focusing only on querying time from | |
4834 | one remote server and synchronizing the local clock to | |
4835 | it. Unless you intend to serve NTP to networked clients or | |
4836 | want to connect to local hardware clocks, this simple NTP | |
4837 | client should be more than appropriate for most | |
4838 | installations. The daemon runs with minimal privileges, and | |
4839 | has been hooked up with networkd to only operate when | |
4840 | network connectivity is available. The daemon saves the | |
4841 | current clock to disk every time a new NTP sync has been | |
4842 | acquired, and uses this to possibly correct the system clock | |
4843 | early at bootup, in order to accommodate for systems that | |
4844 | lack an RTC such as the Raspberry Pi and embedded devices, | |
4845 | and to make sure that time monotonically progresses on these | |
4846 | systems, even if it is not always correct. To make use of | |
4847 | this daemon, a new system user and group "systemd-timesync" | |
4848 | needs to be created on installation of systemd. | |
4849 | ||
4850 | * The queue "seqnum" interface of libudev has been disabled, as | |
4851 | it was generally incompatible with device namespacing as | |
4852 | sequence numbers of devices go "missing" if the devices are | |
4853 | part of a different namespace. | |
4854 | ||
4855 | * "systemctl list-timers" and "systemctl list-sockets" gained | |
4856 | a --recursive switch for showing units of these types also | |
4857 | for all local containers, similar in style to the already | |
4858 | supported --recursive switch for "systemctl list-units". | |
4859 | ||
4860 | * A new RebootArgument= setting has been added for service | |
4861 | units, which may be used to specify a kernel reboot argument | |
4862 | to use when triggering reboots with StartLimitAction=. | |
4863 | ||
4864 | * A new FailureAction= setting has been added for service | |
4865 | units which may be used to specify an operation to trigger | |
4866 | when a service fails. This works similarly to | |
4867 | StartLimitAction=, but unlike it, controls what is done | |
4868 | immediately rather than only after several attempts to | |
4869 | restart the service in question. | |
4870 | ||
4871 | * hostnamed got updated to also expose the kernel name, | |
4872 | release, and version on the bus. This is useful for | |
4873 | executing commands like hostnamectl with the -H switch. | |
4874 | systemd-analyze makes use of this to properly display | |
4875 | details when running non-locally. | |
4876 | ||
4877 | * The bootchart tool can now show cgroup information in the | |
4878 | graphs it generates. | |
4879 | ||
4880 | * The CFS CPU quota cgroup attribute is now exposed for | |
4881 | services. The new CPUQuota= switch has been added for this | |
4882 | which takes a percentage value. Setting this will have the | |
4883 | result that a service may never get more CPU time than the | |
4884 | specified percentage, even if the machine is otherwise idle. | |
4885 | ||
4886 | * systemd-networkd learned IPIP and SIT tunnel support. | |
4887 | ||
4888 | * LSB init scripts exposing a dependency on $network will now | |
4889 | get a dependency on network-online.target rather than simply | |
4890 | network.target. This should bring LSB handling closer to | |
4891 | what it was on SysV systems. | |
4892 | ||
4893 | * A new fsck.repair= kernel option has been added to control | |
4894 | how fsck shall deal with unclean file systems at boot. | |
4895 | ||
4896 | * The (.ini) configuration file parser will now silently | |
4897 | ignore sections whose name begins with "X-". This may be | |
4898 | used to maintain application-specific extension sections in unit | |
4899 | files. | |
4900 | ||
4901 | * machined gained a new API to query the IP addresses of | |
4902 | registered containers. "machinectl status" has been updated | |
4903 | to show these addresses in its output. | |
4904 | ||
4905 | * A new call sd_uid_get_display() has been added to the | |
4906 | sd-login APIs for querying the "primary" session of a | |
4907 | user. The "primary" session of the user is elected from the | |
4908 | user's sessions and generally a graphical session is | |
4909 | preferred over a text one. | |
4910 | ||
4911 | * A minimal systemd-resolved daemon has been added. It | |
4912 | currently simply acts as a companion to systemd-networkd and | |
4913 | manages resolv.conf based on per-interface DNS | |
4914 | configuration, possibly supplied via DHCP. In the long run | |
4915 | we hope to extend this into a local DNSSEC enabled DNS and | |
4916 | mDNS cache. | |
4917 | ||
4918 | * The systemd-networkd-wait-online tool is now enabled by | |
4919 | default. It will delay network-online.target until a network | |
4920 | connection has been configured. The tool primarily integrates | |
4921 | with networkd, but will also make a best effort to make sense | |
4922 | of network configuration performed in some other way. | |
4923 | ||
4924 | * Two new service options StartupCPUShares= and | |
4925 | StartupBlockIOWeight= have been added that work similarly to | |
4926 | CPUShares= and BlockIOWeight= however only apply during | |
4927 | system startup. This is useful to prioritize certain services | |
4928 | differently during bootup than during normal runtime. | |
4929 | ||
4930 | * hostnamed has been changed to prefer the statically | |
4931 | configured hostname in /etc/hostname (unless set to | |
4932 | 'localhost' or empty) over any dynamic one supplied by | |
4933 | dhcp. With this change, the rules for picking the hostname | |
4934 | match more closely the rules of other configuration settings | |
4935 | where the local administrator's configuration in /etc always | |
4936 | overrides any other settings. | |
4937 | ||
4938 | Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van | |
4939 | den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch, | |
4940 | Dan Kilman, Dave Reisner, David Härdeman, David Herrmann, | |
4941 | David Strauss, Dimitris Spingos, Djalal Harouni, Eelco | |
4942 | Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg | |
4943 | Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan | |
4944 | Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark, | |
4945 | Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas | |
4946 | Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas, | |
4947 | Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael | |
4948 | Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis | |
4949 | Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode, | |
4950 | Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter, | |
4951 | Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler, | |
4952 | Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar | |
4953 | Lindskog, WaLyong Cho, Will Woods, Zbigniew | |
4954 | Jędrzejewski-Szmek | |
4955 | ||
4956 | — Beijing, 2014-05-28 | |
4957 | ||
4958 | CHANGES WITH 212: | |
4959 | ||
4960 | * When restoring the screen brightness at boot, stay away from | |
4961 | the darkest setting or from the lowest 5% of the available | |
4962 | range, depending on which is the larger value of both. This | |
4963 | should effectively protect the user from rebooting into a | |
4964 | black screen, should the brightness have been set to minimum | |
4965 | by accident. | |
4966 | ||
4967 | * sd-login gained a new sd_machine_get_class() call to | |
4968 | determine the class ("vm" or "container") of a machine | |
4969 | registered with machined. | |
4970 | ||
4971 | * sd-login gained new calls | |
4972 | sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(), | |
4973 | to query the identity of the peer of a local AF_UNIX | |
4974 | connection. They operate similarly to their sd_pid_get_xyz() | |
4975 | counterparts. | |
4976 | ||
4977 | * PID 1 will now maintain a system-wide system state engine | |
4978 | with the states "starting", "running", "degraded", | |
4979 | "maintenance", "stopping". These states are bound to system | |
4980 | startup, normal runtime, runtime with at least one failed | |
4981 | service, rescue/emergency mode and system shutdown. This | |
4982 | state is shown in the "systemctl status" output when no unit | |
4983 | name is passed. It is useful to determine system state, in | |
4984 | particularly when doing so for many systems or containers at | |
4985 | once. | |
4986 | ||
4987 | * A new command "list-machines" has been added to "systemctl" | |
4988 | that lists all local OS containers and shows their system | |
4989 | state (see above), if systemd runs inside of them. | |
4990 | ||
4991 | * systemctl gained a new "-r" switch to recursively enumerate | |
4992 | units on all local containers, when used with the | |
4993 | "list-unit" command (which is the default one that is | |
4994 | executed when no parameters are specified). | |
4995 | ||
4996 | * The GPT automatic partition discovery logic will now honour | |
4997 | two GPT partition flags: one may be set on a partition to | |
4998 | cause it to be mounted read-only, and the other may be set | |
4999 | on a partition to ignore it during automatic discovery. | |
5000 | ||
5001 | * Two new GPT type UUIDs have been added for automatic root | |
5002 | partition discovery, for 32-bit and 64-bit ARM. This is not | |
5003 | particularly useful for discovering the root directory on | |
5004 | these architectures during bare-metal boots (since UEFI is | |
5005 | not common there), but still very useful to allow booting of | |
5006 | ARM disk images in nspawn with the -i option. | |
5007 | ||
5008 | * MAC addresses of interfaces created with nspawn's | |
5009 | --network-interface= switch will now be generated from the | |
5010 | machine name, and thus be stable between multiple invocations | |
5011 | of the container. | |
5012 | ||
5013 | * logind will now automatically remove all IPC objects owned | |
5014 | by a user if she or he fully logs out. This makes sure that | |
5015 | users who are logged out cannot continue to consume IPC | |
5016 | resources. This covers SysV memory, semaphores and message | |
5017 | queues as well as POSIX shared memory and message | |
5018 | queues. Traditionally, SysV and POSIX IPC had no life-cycle | |
5019 | limits. With this functionality, that is corrected. This may | |
5020 | be turned off by using the RemoveIPC= switch of logind.conf. | |
5021 | ||
5022 | * The systemd-machine-id-setup and tmpfiles tools gained a | |
5023 | --root= switch to operate on a specific root directory, | |
5024 | instead of /. | |
5025 | ||
5026 | * journald can now forward logged messages to the TTYs of all | |
5027 | logged in users ("wall"). This is the default for all | |
5028 | emergency messages now. | |
5029 | ||
5030 | * A new tool systemd-journal-remote has been added to stream | |
5031 | journal log messages across the network. | |
5032 | ||
5033 | * /sys/fs/cgroup/ is now mounted read-only after all cgroup | |
5034 | controller trees are mounted into it. Note that the | |
5035 | directories mounted beneath it are not read-only. This is a | |
5036 | security measure and is particularly useful because glibc | |
5037 | actually includes a search logic to pick any tmpfs it can | |
5038 | find to implement shm_open() if /dev/shm is not available | |
5039 | (which it might very well be in namespaced setups). | |
5040 | ||
5041 | * machinectl gained a new "poweroff" command to cleanly power | |
5042 | down a local OS container. | |
5043 | ||
5044 | * The PrivateDevices= unit file setting will now also drop the | |
5045 | CAP_MKNOD capability from the capability bound set, and | |
5046 | imply DevicePolicy=closed. | |
5047 | ||
5048 | * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used | |
5049 | comprehensively on all long-running systemd services where | |
5050 | this is appropriate. | |
5051 | ||
5052 | * systemd-udevd will now run in a disassociated mount | |
5053 | namespace. To mount directories from udev rules, make sure to | |
5054 | pull in mount units via SYSTEMD_WANTS properties. | |
5055 | ||
5056 | * The kdbus support gained support for uploading policy into | |
5057 | the kernel. sd-bus gained support for creating "monitoring" | |
5058 | connections that can eavesdrop into all bus communication | |
5059 | for debugging purposes. | |
5060 | ||
5061 | * Timestamps may now be specified in seconds since the UNIX | |
5062 | epoch Jan 1st, 1970 by specifying "@" followed by the value | |
5063 | in seconds. | |
5064 | ||
5065 | * Native tcpwrap support in systemd has been removed. tcpwrap | |
5066 | is old code, not really maintained anymore and has serious | |
5067 | shortcomings, and better options such as firewalls | |
5068 | exist. For setups that require tcpwrap usage, please | |
5069 | consider invoking your socket-activated service via tcpd, | |
5070 | like on traditional inetd. | |
5071 | ||
5072 | * A new system.conf configuration option | |
5073 | DefaultTimerAccuracySec= has been added that controls the | |
5074 | default AccuracySec= setting of .timer units. | |
5075 | ||
5076 | * Timer units gained a new WakeSystem= switch. If enabled, | |
5077 | timers configured this way will cause the system to resume | |
5078 | from system suspend (if the system supports that, which most | |
5079 | do these days). | |
5080 | ||
5081 | * Timer units gained a new Persistent= switch. If enabled, | |
5082 | timers configured this way will save to disk when they have | |
5083 | been last triggered. This information is then used on next | |
5084 | reboot to possible execute overdue timer events, that | |
5085 | could not take place because the system was powered off. | |
5086 | This enables simple anacron-like behaviour for timer units. | |
5087 | ||
5088 | * systemctl's "list-timers" will now also list the time a | |
5089 | timer unit was last triggered in addition to the next time | |
5090 | it will be triggered. | |
5091 | ||
5092 | * systemd-networkd will now assign predictable IPv4LL | |
5093 | addresses to its local interfaces. | |
5094 | ||
5095 | Contributions from: Brandon Philips, Daniel Buch, Daniel Mack, | |
5096 | Dave Reisner, David Herrmann, Gerd Hoffmann, Greg | |
5097 | Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh | |
5098 | Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine | |
5099 | Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna, | |
5100 | Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler, | |
5101 | Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen, | |
5102 | Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew | |
5103 | Jędrzejewski-Szmek | |
5104 | ||
5105 | — Berlin, 2014-03-25 | |
5106 | ||
5107 | CHANGES WITH 211: | |
5108 | ||
5109 | * A new unit file setting RestrictAddressFamilies= has been | |
5110 | added to restrict which socket address families unit | |
5111 | processes gain access to. This takes address family names | |
5112 | like "AF_INET" or "AF_UNIX", and is useful to minimize the | |
5113 | attack surface of services via exotic protocol stacks. This | |
5114 | is built on seccomp system call filters. | |
5115 | ||
5116 | * Two new unit file settings RuntimeDirectory= and | |
5117 | RuntimeDirectoryMode= have been added that may be used to | |
5118 | manage a per-daemon runtime directories below /run. This is | |
5119 | an alternative for setting up directory permissions with | |
5120 | tmpfiles snippets, and has the advantage that the runtime | |
5121 | directory's lifetime is bound to the daemon runtime and that | |
5122 | the daemon starts up with an empty directory each time. This | |
5123 | is particularly useful when writing services that drop | |
5124 | privileges using the User= or Group= setting. | |
5125 | ||
5126 | * The DeviceAllow= unit setting now supports globbing for | |
5127 | matching against device group names. | |
5128 | ||
5129 | * The systemd configuration file system.conf gained new | |
5130 | settings DefaultCPUAccounting=, DefaultBlockIOAccounting=, | |
5131 | DefaultMemoryAccounting= to globally turn on/off accounting | |
5132 | for specific resources (cgroups) for all units. These | |
5133 | settings may still be overridden individually in each unit | |
5134 | though. | |
5135 | ||
5136 | * systemd-gpt-auto-generator is now able to discover /srv and | |
5137 | root partitions in addition to /home and swap partitions. It | |
5138 | also supports LUKS-encrypted partitions now. With this in | |
5139 | place, automatic discovery of partitions to mount following | |
5140 | the Discoverable Partitions Specification | |
5141 | (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec) | |
5142 | is now a lot more complete. This allows booting without | |
5143 | /etc/fstab and without root= on the kernel command line on | |
5144 | systems prepared appropriately. | |
5145 | ||
5146 | * systemd-nspawn gained a new --image= switch which allows | |
5147 | booting up disk images and Linux installations on any block | |
5148 | device that follow the Discoverable Partitions Specification | |
5149 | (see above). This means that installations made with | |
5150 | appropriately updated installers may now be started and | |
5151 | deployed using container managers, completely | |
5152 | unmodified. (We hope that libvirt-lxc will add support for | |
5153 | this feature soon, too.) | |
5154 | ||
5155 | * systemd-nspawn gained a new --network-macvlan= setting to | |
5156 | set up a private macvlan interface for the | |
5157 | container. Similarly, systemd-networkd gained a new | |
5158 | Kind=macvlan setting in .netdev files. | |
5159 | ||
5160 | * systemd-networkd now supports configuring local addresses | |
5161 | using IPv4LL. | |
5162 | ||
5163 | * A new tool systemd-network-wait-online has been added to | |
5164 | synchronously wait for network connectivity using | |
5165 | systemd-networkd. | |
5166 | ||
5167 | * The sd-bus.h bus API gained a new sd_bus_track object for | |
5168 | tracking the life-cycle of bus peers. Note that sd-bus.h is | |
5169 | still not a public API though (unless you specify | |
5170 | --enable-kdbus on the configure command line, which however | |
5171 | voids your warranty and you get no API stability guarantee). | |
5172 | ||
5173 | * The $XDG_RUNTIME_DIR runtime directories for each user are | |
5174 | now individual tmpfs instances, which has the benefit of | |
5175 | introducing separate pools for each user, with individual | |
5176 | size limits, and thus making sure that unprivileged clients | |
5177 | can no longer negatively impact the system or other users by | |
5178 | filling up their $XDG_RUNTIME_DIR. A new logind.conf setting | |
5179 | RuntimeDirectorySize= has been introduced that allows | |
5180 | controlling the default size limit for all users. It | |
5181 | defaults to 10% of the available physical memory. This is no | |
5182 | replacement for quotas on tmpfs though (which the kernel | |
5183 | still does not support), as /dev/shm and /tmp are still | |
5184 | shared resources used by both the system and unprivileged | |
5185 | users. | |
5186 | ||
5187 | * logind will now automatically turn off automatic suspending | |
5188 | on laptop lid close when more than one display is | |
5189 | connected. This was previously expected to be implemented | |
5190 | individually in desktop environments (such as GNOME), | |
5191 | however has been added to logind now, in order to fix a | |
5192 | boot-time race where a desktop environment might not have | |
5193 | been started yet and thus not been able to take an inhibitor | |
5194 | lock at the time where logind already suspends the system | |
5195 | due to a closed lid. | |
5196 | ||
5197 | * logind will now wait at least 30s after each system | |
5198 | suspend/resume cycle, and 3min after system boot before | |
5199 | suspending the system due to a closed laptop lid. This | |
5200 | should give USB docking stations and similar enough time to | |
5201 | be probed and configured after system resume and boot in | |
5202 | order to then act as suspend blocker. | |
5203 | ||
5204 | * systemd-run gained a new --property= setting which allows | |
5205 | initialization of resource control properties (and others) | |
5206 | for the created scope or service unit. Example: "systemd-run | |
5207 | --property=BlockIOWeight=10 updatedb" may be used to run | |
5208 | updatedb at a low block IO scheduling weight. | |
5209 | ||
5210 | * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches | |
5211 | now also work in --scope mode. | |
5212 | ||
5213 | * When systemd is compiled with kdbus support, basic support | |
5214 | for enforced policies is now in place. (Note that enabling | |
5215 | kdbus still voids your warranty and no API compatibility | |
5216 | promises are made.) | |
5217 | ||
5218 | Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin | |
5219 | K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni, | |
5220 | Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay | |
5221 | Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn, | |
5222 | Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt, | |
5223 | Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef | |
5224 | Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas | |
5225 | Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom | |
5226 | Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook, | |
5227 | Zbigniew Jędrzejewski-Szmek | |
5228 | ||
5229 | — Berlin, 2014-03-12 | |
5230 | ||
5231 | CHANGES WITH 210: | |
5232 | ||
5233 | * systemd will now relabel /dev after loading the SMACK policy | |
5234 | according to SMACK rules. | |
5235 | ||
5236 | * A new unit file option AppArmorProfile= has been added to | |
5237 | set the AppArmor profile for the processes of a unit. | |
5238 | ||
5239 | * A new condition check ConditionArchitecture= has been added | |
5240 | to conditionalize units based on the system architecture, as | |
5241 | reported by uname()'s "machine" field. | |
5242 | ||
5243 | * systemd-networkd now supports matching on the system | |
5244 | virtualization, architecture, kernel command line, host name | |
5245 | and machine ID. | |
5246 | ||
5247 | * logind is now a lot more aggressive when suspending the | |
5248 | machine due to a closed laptop lid. Instead of acting only | |
5249 | on the lid close action, it will continuously watch the lid | |
5250 | status and act on it. This is useful for laptops where the | |
5251 | power button is on the outside of the chassis so that it can | |
5252 | be reached without opening the lid (such as the Lenovo | |
5253 | Yoga). On those machines, logind will now immediately | |
5254 | re-suspend the machine if the power button has been | |
5255 | accidentally pressed while the laptop was suspended and in a | |
5256 | backpack or similar. | |
5257 | ||
5258 | * logind will now watch SW_DOCK switches and inhibit reaction | |
5259 | to the lid switch if it is pressed. This means that logind | |
5260 | will not suspend the machine anymore if the lid is closed | |
5261 | and the system is docked, if the laptop supports SW_DOCK | |
5262 | notifications via the input layer. Note that ACPI docking | |
5263 | stations do not generate this currently. Also note that this | |
5264 | logic is usually not fully sufficient and Desktop | |
5265 | Environments should take a lid switch inhibitor lock when an | |
5266 | external display is connected, as systemd will not watch | |
5267 | this on its own. | |
5268 | ||
5269 | * nspawn will now make use of the devices cgroup controller by | |
5270 | default, and only permit creation of and access to the usual | |
5271 | API device nodes like /dev/null or /dev/random, as well as | |
5272 | access to (but not creation of) the pty devices. | |
5273 | ||
5274 | * We will now ship a default .network file for | |
5275 | systemd-networkd that automatically configures DHCP for | |
5276 | network interfaces created by nspawn's --network-veth or | |
5277 | --network-bridge= switches. | |
5278 | ||
5279 | * systemd will now understand the usual M, K, G, T suffixes | |
5280 | according to SI conventions (i.e. to the base 1000) when | |
5281 | referring to throughput and hardware metrics. It will stay | |
5282 | with IEC conventions (i.e. to the base 1024) for software | |
5283 | metrics, according to what is customary according to | |
5284 | Wikipedia. We explicitly document which base applies for | |
5285 | each configuration option. | |
5286 | ||
5287 | * The DeviceAllow= setting in unit files now supports a syntax | |
5288 | to whitelist an entire group of devices node majors at once, | |
5289 | based on the /proc/devices listing. For example, with the | |
5290 | string "char-pts", it is now possible to whitelist all | |
5291 | current and future pseudo-TTYs at once. | |
5292 | ||
5293 | * sd-event learned a new "post" event source. Event sources of | |
5294 | this type are triggered by the dispatching of any event | |
5295 | source of a type that is not "post". This is useful for | |
5296 | implementing clean-up and check event sources that are | |
5297 | triggered by other work being done in the program. | |
5298 | ||
5299 | * systemd-networkd is no longer statically enabled, but uses | |
5300 | the usual [Install] sections so that it can be | |
5301 | enabled/disabled using systemctl. It still is enabled by | |
5302 | default however. | |
5303 | ||
5304 | * When creating a veth interface pair with systemd-nspawn, the | |
5305 | host side will now be prefixed with "vb-" if | |
5306 | --network-bridge= is used, and with "ve-" if --network-veth | |
5307 | is used. This way, it is easy to distinguish these cases on | |
5308 | the host, for example to apply different configuration to | |
5309 | them with systemd-networkd. | |
5310 | ||
5311 | * The compatibility libraries for libsystemd-journal.so, | |
5312 | libsystem-id128.so, libsystemd-login.so and | |
5313 | libsystemd-daemon.so do not make use of IFUNC | |
5314 | anymore. Instead, we now build libsystemd.so multiple times | |
5315 | under these alternative names. This means that the footprint | |
5316 | is drastically increased, but given that these are | |
5317 | transitional compatibility libraries, this should not matter | |
5318 | much. This change has been made necessary to support the ARM | |
5319 | platform for these compatibility libraries, as the ARM | |
5320 | toolchain is not really at the same level as the toolchain | |
5321 | for other architectures like x86 and does not support | |
5322 | IFUNC. Please make sure to use --enable-compat-libs only | |
5323 | during a transitional period! | |
5324 | ||
5325 | Contributions from: Andreas Fuchs, Armin K., Colin Walters, | |
5326 | Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni, | |
5327 | Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper | |
5328 | St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach, | |
5329 | Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike | |
5330 | Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe | |
5331 | Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog, | |
5332 | Zbigniew Jędrzejewski-Szmek | |
5333 | ||
5334 | — Berlin, 2014-02-24 | |
5335 | ||
5336 | CHANGES WITH 209: | |
5337 | ||
5338 | * A new component "systemd-networkd" has been added that can | |
5339 | be used to configure local network interfaces statically or | |
5340 | via DHCP. It is capable of bringing up bridges, VLANs, and | |
5341 | bonding. Currently, no hook-ups for interactive network | |
5342 | configuration are provided. Use this for your initrd, | |
5343 | container, embedded, or server setup if you need a simple, | |
5344 | yet powerful, network configuration solution. This | |
5345 | configuration subsystem is quite nifty, as it allows wildcard | |
5346 | hotplug matching in interfaces. For example, with a single | |
5347 | configuration snippet, you can configure that all Ethernet | |
5348 | interfaces showing up are automatically added to a bridge, | |
5349 | or similar. It supports link-sensing and more. | |
5350 | ||
5351 | * A new tool "systemd-socket-proxyd" has been added which can | |
5352 | act as a bidirectional proxy for TCP sockets. This is | |
5353 | useful for adding socket activation support to services that | |
5354 | do not actually support socket activation, including virtual | |
5355 | machines and the like. | |
5356 | ||
5357 | * Add a new tool to save/restore rfkill state on | |
5358 | shutdown/boot. | |
5359 | ||
5360 | * Save/restore state of keyboard backlights in addition to | |
5361 | display backlights on shutdown/boot. | |
5362 | ||
5363 | * udev learned a new SECLABEL{} construct to label device | |
5364 | nodes with a specific security label when they appear. For | |
5365 | now, only SECLABEL{selinux} is supported, but the syntax is | |
5366 | prepared for additional security frameworks. | |
5367 | ||
5368 | * udev gained a new scheme to configure link-level attributes | |
5369 | from files in /etc/systemd/network/*.link. These files can | |
5370 | match against MAC address, device path, driver name and type, | |
5371 | and will apply attributes like the naming policy, link speed, | |
5372 | MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC | |
5373 | address assignment policy (randomized, ...). | |
5374 | ||
5375 | * The configuration of network interface naming rules for | |
5376 | "permanent interface names" has changed: a new NamePolicy= | |
5377 | setting in the [Link] section of .link files determines the | |
5378 | priority of possible naming schemes (onboard, slot, MAC, | |
5379 | path). The default value of this setting is determined by | |
5380 | /usr/lib/net/links/99-default.link. Old | |
5381 | 80-net-name-slot.rules udev configuration file has been | |
5382 | removed, so local configuration overriding this file should | |
5383 | be adapted to override 99-default.link instead. | |
5384 | ||
5385 | * When the User= switch is used in a unit file, also | |
5386 | initialize $SHELL= based on the user database entry. | |
5387 | ||
5388 | * systemd no longer depends on libdbus. All communication is | |
5389 | now done with sd-bus, systemd's low-level bus library | |
5390 | implementation. | |
5391 | ||
5392 | * kdbus support has been added to PID 1 itself. When kdbus is | |
5393 | enabled, this causes PID 1 to set up the system bus and | |
5394 | enable support for a new ".busname" unit type that | |
5395 | encapsulates bus name activation on kdbus. It works a little | |
5396 | bit like ".socket" units, except for bus names. A new | |
5397 | generator has been added that converts classic dbus1 service | |
5398 | activation files automatically into native systemd .busname | |
5399 | and .service units. | |
5400 | ||
5401 | * sd-bus: add a light-weight vtable implementation that allows | |
5402 | defining objects on the bus with a simple static const | |
5403 | vtable array of its methods, signals and properties. | |
5404 | ||
5405 | * systemd will not generate or install static dbus | |
5406 | introspection data anymore to /usr/share/dbus-1/interfaces, | |
5407 | as the precise format of these files is unclear, and | |
5408 | nothing makes use of it. | |
5409 | ||
5410 | * A proxy daemon is now provided to proxy clients connecting | |
5411 | via classic D-Bus AF_UNIX sockets to kdbus, to provide full | |
5412 | compatibility with classic D-Bus. | |
5413 | ||
5414 | * A bus driver implementation has been added that supports the | |
5415 | classic D-Bus bus driver calls on kdbus, also for | |
5416 | compatibility purposes. | |
5417 | ||
5418 | * A new API "sd-event.h" has been added that implements a | |
5419 | minimal event loop API built around epoll. It provides a | |
5420 | couple of features that direct epoll usage is lacking: | |
5421 | prioritization of events, scales to large numbers of timer | |
5422 | events, per-event timer slack (accuracy), system-wide | |
5423 | coalescing of timer events, exit handlers, watchdog | |
5424 | supervision support using systemd's sd_notify() API, child | |
5425 | process handling. | |
5426 | ||
5427 | * A new API "sd-rntl.h" has been added that provides an API | |
5428 | around the route netlink interface of the kernel, similar in | |
5429 | style to "sd-bus.h". | |
5430 | ||
5431 | * A new API "sd-dhcp-client.h" has been added that provides a | |
5432 | small DHCPv4 client-side implementation. This is used by | |
5433 | "systemd-networkd". | |
5434 | ||
5435 | * There is a new kernel command line option | |
5436 | "systemd.restore_state=0|1". When set to "0", none of the | |
5437 | systemd tools will restore saved runtime state to hardware | |
5438 | devices. More specifically, the rfkill and backlight states | |
5439 | are not restored. | |
5440 | ||
5441 | * The FsckPassNo= compatibility option in mount/service units | |
5442 | has been removed. The fstab generator will now add the | |
5443 | necessary dependencies automatically, and does not require | |
5444 | PID1's support for that anymore. | |
5445 | ||
5446 | * journalctl gained a new switch, --list-boots, that lists | |
5447 | recent boots with their times and boot IDs. | |
5448 | ||
5449 | * The various tools like systemctl, loginctl, timedatectl, | |
5450 | busctl, systemd-run, ... have gained a new switch "-M" to | |
5451 | connect to a specific, local OS container (as direct | |
5452 | connection, without requiring SSH). This works on any | |
5453 | container that is registered with machined, such as those | |
5454 | created by libvirt-lxc or nspawn. | |
5455 | ||
5456 | * systemd-run and systemd-analyze also gained support for "-H" | |
5457 | to connect to remote hosts via SSH. This is particularly | |
5458 | useful for systemd-run because it enables queuing of jobs | |
5459 | onto remote systems. | |
5460 | ||
5461 | * machinectl gained a new command "login" to open a getty | |
5462 | login in any local container. This works with any container | |
5463 | that is registered with machined (such as those created by | |
5464 | libvirt-lxc or nspawn), and which runs systemd inside. | |
5465 | ||
5466 | * machinectl gained a new "reboot" command that may be used to | |
5467 | trigger a reboot on a specific container that is registered | |
5468 | with machined. This works on any container that runs an init | |
5469 | system of some kind. | |
5470 | ||
5471 | * systemctl gained a new "list-timers" command to print a nice | |
5472 | listing of installed timer units with the times they elapse | |
5473 | next. | |
5474 | ||
5475 | * Alternative reboot() parameters may now be specified on the | |
5476 | "systemctl reboot" command line and are passed to the | |
5477 | reboot() system call. | |
5478 | ||
5479 | * systemctl gained a new --job-mode= switch to configure the | |
5480 | mode to queue a job with. This is a more generic version of | |
5481 | --fail, --irreversible, and --ignore-dependencies, which are | |
5482 | still available but not advertised anymore. | |
5483 | ||
5484 | * /etc/systemd/system.conf gained new settings to configure | |
5485 | various default timeouts of units, as well as the default | |
5486 | start limit interval and burst. These may still be overridden | |
5487 | within each Unit. | |
5488 | ||
5489 | * PID1 will now export on the bus profile data of the security | |
5490 | policy upload process (such as the SELinux policy upload to | |
5491 | the kernel). | |
5492 | ||
5493 | * journald: when forwarding logs to the console, include | |
5494 | timestamps (following the setting in | |
5495 | /sys/module/printk/parameters/time). | |
5496 | ||
5497 | * OnCalendar= in timer units now understands the special | |
5498 | strings "yearly" and "annually". (Both are equivalent) | |
5499 | ||
5500 | * The accuracy of timer units is now configurable with the new | |
5501 | AccuracySec= setting. It defaults to 1min. | |
5502 | ||
5503 | * A new dependency type JoinsNamespaceOf= has been added that | |
5504 | allows running two services within the same /tmp and network | |
5505 | namespace, if PrivateNetwork= or PrivateTmp= are used. | |
5506 | ||
5507 | * A new command "cat" has been added to systemctl. It outputs | |
5508 | the original unit file of a unit, and concatenates the | |
5509 | contents of additional "drop-in" unit file snippets, so that | |
5510 | the full configuration is shown. | |
5511 | ||
5512 | * systemctl now supports globbing on the various "list-xyz" | |
5513 | commands, like "list-units" or "list-sockets", as well as on | |
5514 | those commands which take multiple unit names. | |
5515 | ||
5516 | * journalctl's --unit= switch gained support for globbing. | |
5517 | ||
5518 | * All systemd daemons now make use of the watchdog logic so | |
5519 | that systemd automatically notices when they hang. | |
5520 | ||
5521 | * If the $container_ttys environment variable is set, | |
5522 | getty-generator will automatically spawn a getty for each | |
5523 | listed tty. This is useful for container managers to request | |
5524 | login gettys to be spawned on as many ttys as needed. | |
5525 | ||
5526 | * %h, %s, %U specifier support is not available anymore when | |
5527 | used in unit files for PID 1. This is because NSS calls are | |
5528 | not safe from PID 1. They stay available for --user | |
5529 | instances of systemd, and as special case for the root user. | |
5530 | ||
5531 | * loginctl gained a new "--no-legend" switch to turn off output | |
5532 | of the legend text. | |
5533 | ||
5534 | * The "sd-login.h" API gained three new calls: | |
5535 | sd_session_is_remote(), sd_session_get_remote_user(), | |
5536 | sd_session_get_remote_host() to query information about | |
5537 | remote sessions. | |
5538 | ||
5539 | * The udev hardware database now also carries vendor/product | |
5540 | information of SDIO devices. | |
5541 | ||
5542 | * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to | |
5543 | determine whether watchdog notifications are requested by | |
5544 | the system manager. | |
5545 | ||
5546 | * Socket-activated per-connection services now include a | |
5547 | short description of the connection parameters in the | |
5548 | description. | |
5549 | ||
5550 | * tmpfiles gained a new "--boot" option. When this is not used, | |
5551 | only lines where the command character is not suffixed with | |
5552 | "!" are executed. When this option is specified, those | |
5553 | options are executed too. This partitions tmpfiles | |
5554 | directives into those that can be safely executed at any | |
5555 | time, and those which should be run only at boot (for | |
5556 | example, a line that creates /run/nologin). | |
5557 | ||
5558 | * A new API "sd-resolve.h" has been added which provides a simple | |
5559 | asynchronous wrapper around glibc NSS host name resolution | |
5560 | calls, such as getaddrinfo(). In contrast to glibc's | |
5561 | getaddrinfo_a(), it does not use signals. In contrast to most | |
5562 | other asynchronous name resolution libraries, this one does | |
5563 | not reimplement DNS, but reuses NSS, so that alternate | |
5564 | host name resolution systems continue to work, such as mDNS, | |
5565 | LDAP, etc. This API is based on libasyncns, but it has been | |
5566 | cleaned up for inclusion in systemd. | |
5567 | ||
5568 | * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h", | |
5569 | "sd-daemon.h" are no longer found in individual libraries | |
5570 | libsystemd-journal.so, libsystemd-login.so, | |
5571 | libsystemd-id128.so, libsystemd-daemon.so. Instead, we have | |
5572 | merged them into a single library, libsystemd.so, which | |
5573 | provides all symbols. The reason for this is cyclic | |
5574 | dependencies, as these libraries tend to use each other's | |
5575 | symbols. So far, we have managed to workaround that by linking | |
5576 | a copy of a good part of our code into each of these | |
5577 | libraries again and again, which, however, makes certain | |
5578 | things hard to do, like sharing static variables. Also, it | |
5579 | substantially increases footprint. With this change, there | |
5580 | is only one library for the basic APIs systemd | |
5581 | provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h", | |
5582 | "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this | |
5583 | library as well, however are subject to the --enable-kdbus | |
5584 | switch (see below). Note that "sd-dhcp-client.h" is not part | |
5585 | of this library (this is because it only consumes, never | |
5586 | provides, services of/to other APIs). To make the transition | |
5587 | easy from the separate libraries to the unified one, we | |
5588 | provide the --enable-compat-libs compile-time switch which | |
5589 | will generate stub libraries that are compatible with the | |
5590 | old ones but redirect all calls to the new one. | |
5591 | ||
5592 | * All of the kdbus logic and the new APIs "sd-bus.h", | |
5593 | "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h", | |
5594 | and "sd-utf8.h" are compile-time optional via the | |
5595 | "--enable-kdbus" switch, and they are not compiled in by | |
5596 | default. To make use of kdbus, you have to explicitly enable | |
5597 | the switch. Note however, that neither the kernel nor the | |
5598 | userspace API for all of this is considered stable yet. We | |
5599 | want to maintain the freedom to still change the APIs for | |
5600 | now. By specifying this build-time switch, you acknowledge | |
5601 | that you are aware of the instability of the current | |
5602 | APIs. | |
5603 | ||
5604 | * Also, note that while kdbus is pretty much complete, | |
5605 | it lacks one thing: proper policy support. This means you | |
5606 | can build a fully working system with all features; however, | |
5607 | it will be highly insecure. Policy support will be added in | |
5608 | one of the next releases, at the same time that we will | |
5609 | declare the APIs stable. | |
5610 | ||
5611 | * When the kernel command line argument "kdbus" is specified, | |
5612 | systemd will automatically load the kdbus.ko kernel module. At | |
5613 | this stage of development, it is only useful for testing kdbus | |
5614 | and should not be used in production. Note: if "--enable-kdbus" | |
5615 | is specified, and the kdbus.ko kernel module is available, and | |
5616 | "kdbus" is added to the kernel command line, the entire system | |
5617 | runs with kdbus instead of dbus-daemon, with the above mentioned | |
5618 | problem of missing the system policy enforcement. Also a future | |
5619 | version of kdbus.ko or a newer systemd will not be compatible with | |
5620 | each other, and will unlikely be able to boot the machine if only | |
5621 | one of them is updated. | |
5622 | ||
5623 | * systemctl gained a new "import-environment" command which | |
5624 | uploads the caller's environment (or parts thereof) into the | |
5625 | service manager so that it is inherited by services started | |
5626 | by the manager. This is useful to upload variables like | |
5627 | $DISPLAY into the user service manager. | |
5628 | ||
5629 | * A new PrivateDevices= switch has been added to service units | |
5630 | which allows running a service with a namespaced /dev | |
5631 | directory that does not contain any device nodes for | |
5632 | physical devices. More specifically, it only includes devices | |
5633 | such as /dev/null, /dev/urandom, and /dev/zero which are API | |
5634 | entry points. | |
5635 | ||
5636 | * logind has been extended to support behaviour like VT | |
5637 | switching on seats that do not support a VT. This makes | |
5638 | multi-session available on seats that are not the first seat | |
5639 | (seat0), and on systems where kernel support for VTs has | |
5640 | been disabled at compile-time. | |
5641 | ||
5642 | * If a process holds a delay lock for system sleep or shutdown | |
5643 | and fails to release it in time, we will now log its | |
5644 | identity. This makes it easier to identify processes that | |
5645 | cause slow suspends or power-offs. | |
5646 | ||
5647 | * When parsing /etc/crypttab, support for a new key-slot= | |
5648 | option as supported by Debian is added. It allows indicating | |
5649 | which LUKS slot to use on disk, speeding up key loading. | |
5650 | ||
5651 | * The sd_journal_sendv() API call has been checked and | |
5652 | officially declared to be async-signal-safe so that it may | |
5653 | be invoked from signal handlers for logging purposes. | |
5654 | ||
5655 | * Boot-time status output is now enabled automatically after a | |
5656 | short timeout if boot does not progress, in order to give | |
5657 | the user an indication what she or he is waiting for. | |
5658 | ||
5659 | * The boot-time output has been improved to show how much time | |
5660 | remains until jobs expire. | |
5661 | ||
5662 | * The KillMode= switch in service units gained a new possible | |
5663 | value "mixed". If set, and the unit is shut down, then the | |
5664 | initial SIGTERM signal is sent only to the main daemon | |
5665 | process, while the following SIGKILL signal is sent to | |
5666 | all remaining processes of the service. | |
5667 | ||
5668 | * When a scope unit is registered, a new property "Controller" | |
5669 | may be set. If set to a valid bus name, systemd will send a | |
5670 | RequestStop() signal to this name when it would like to shut | |
5671 | down the scope. This may be used to hook manager logic into | |
5672 | the shutdown logic of scope units. Also, scope units may now | |
5673 | be put in a special "abandoned" state, in which case the | |
5674 | manager process which created them takes no further | |
5675 | responsibilities for it. | |
5676 | ||
5677 | * When reading unit files, systemd will now verify | |
5678 | the access mode of these files, and warn about certain | |
5679 | suspicious combinations. This has been added to make it | |
5680 | easier to track down packaging bugs where unit files are | |
5681 | marked executable or world-writable. | |
5682 | ||
5683 | * systemd-nspawn gained a new "--setenv=" switch to set | |
5684 | container-wide environment variables. The similar option in | |
5685 | systemd-activate was renamed from "--environment=" to | |
5686 | "--setenv=" for consistency. | |
5687 | ||
5688 | * systemd-nspawn has been updated to create a new kdbus domain | |
5689 | for each container that is invoked, thus allowing each | |
5690 | container to have its own set of system and user buses, | |
5691 | independent of the host. | |
5692 | ||
5693 | * systemd-nspawn gained a new --drop-capability= switch to run | |
5694 | the container with less capabilities than the default. Both | |
5695 | --drop-capability= and --capability= now take the special | |
5696 | string "all" for dropping or keeping all capabilities. | |
5697 | ||
5698 | * systemd-nspawn gained new switches for executing containers | |
5699 | with specific SELinux labels set. | |
5700 | ||
5701 | * systemd-nspawn gained a new --quiet switch to not generate | |
5702 | any additional output but the container's own console | |
5703 | output. | |
5704 | ||
5705 | * systemd-nspawn gained a new --share-system switch to run a | |
5706 | container without PID namespacing enabled. | |
5707 | ||
5708 | * systemd-nspawn gained a new --register= switch to control | |
5709 | whether the container is registered with systemd-machined or | |
5710 | not. This is useful for containers that do not run full | |
5711 | OS images, but only specific apps. | |
5712 | ||
5713 | * systemd-nspawn gained a new --keep-unit which may be used | |
5714 | when invoked as the only program from a service unit, and | |
5715 | results in registration of the unit service itself in | |
5716 | systemd-machined, instead of a newly opened scope unit. | |
5717 | ||
5718 | * systemd-nspawn gained a new --network-interface= switch for | |
5719 | moving arbitrary interfaces to the container. The new | |
5720 | --network-veth switch creates a virtual Ethernet connection | |
5721 | between host and container. The new --network-bridge= | |
5722 | switch then allows assigning the host side of this virtual | |
5723 | Ethernet connection to a bridge device. | |
5724 | ||
5725 | * systemd-nspawn gained a new --personality= switch for | |
5726 | setting the kernel personality for the container. This is | |
5727 | useful when running a 32-bit container on a 64-bit host. A | |
5728 | similar option Personality= is now also available for service | |
5729 | units to use. | |
5730 | ||
5731 | * logind will now also track a "Desktop" identifier for each | |
5732 | session which encodes the desktop environment of it. This is | |
5733 | useful for desktop environments that want to identify | |
5734 | multiple running sessions of itself easily. | |
5735 | ||
5736 | * A new SELinuxContext= setting for service units has been | |
5737 | added that allows setting a specific SELinux execution | |
5738 | context for a service. | |
5739 | ||
5740 | * Most systemd client tools will now honour $SYSTEMD_LESS for | |
5741 | settings of the "less" pager. By default, these tools will | |
5742 | override $LESS to allow certain operations to work, such as | |
5743 | jump-to-the-end. With $SYSTEMD_LESS, it is possible to | |
5744 | influence this logic. | |
5745 | ||
5746 | * systemd's "seccomp" hook-up has been changed to make use of | |
5747 | the libseccomp library instead of using its own | |
5748 | implementation. This has benefits for portability among | |
5749 | other things. | |
5750 | ||
5751 | * For usage together with SystemCallFilter=, a new | |
5752 | SystemCallErrorNumber= setting has been introduced that | |
5753 | allows configuration of a system error number to be returned | |
5754 | on filtered system calls, instead of immediately killing the | |
5755 | process. Also, SystemCallArchitectures= has been added to | |
5756 | limit access to system calls of a particular architecture | |
5757 | (in order to turn off support for unused secondary | |
5758 | architectures). There is also a global | |
5759 | SystemCallArchitectures= setting in system.conf now to turn | |
5760 | off support for non-native system calls system-wide. | |
5761 | ||
5762 | * systemd requires a kernel with a working name_to_handle_at(), | |
5763 | please see the kernel config requirements in the README file. | |
5764 | ||
5765 | Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov, | |
5766 | Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera, | |
5767 | Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters, | |
5768 | Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J | |
5769 | Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa, | |
5770 | David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov, | |
5771 | Elia Pinto, Florian Weimer, George McCollister, Goffredo | |
5772 | Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor | |
5773 | Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld, | |
5774 | Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose | |
5775 | Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg, | |
5776 | Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz | |
5777 | Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas, | |
5778 | Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de | |
5779 | Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael | |
5780 | Marineau, Michael Scherer, Michał Górny, Michal Sekletar, | |
5781 | Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt, | |
5782 | Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien | |
5783 | Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters, | |
5784 | Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else, | |
5785 | Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen, | |
5786 | Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav | |
5787 | Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang | |
5788 | Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek | |
5789 | ||
5790 | — Berlin, 2014-02-20 | |
5791 | ||
5792 | CHANGES WITH 208: | |
5793 | ||
5794 | * logind has gained support for facilitating privileged input | |
5795 | and drm device access for unprivileged clients. This work is | |
5796 | useful to allow Wayland display servers (and similar | |
5797 | programs, such as kmscon) to run under the user's ID and | |
5798 | access input and drm devices which are normally | |
5799 | protected. When this is used (and the kernel is new enough) | |
5800 | logind will "mute" IO on the file descriptors passed to | |
5801 | Wayland as long as it is in the background and "unmute" it | |
5802 | if it returns into the foreground. This allows secure | |
5803 | session switching without allowing background sessions to | |
5804 | eavesdrop on input and display data. This also introduces | |
5805 | session switching support if VT support is turned off in the | |
5806 | kernel, and on seats that are not seat0. | |
5807 | ||
5808 | * A new kernel command line option luks.options= is understood | |
5809 | now which allows specifying LUKS options for usage for LUKS | |
5810 | encrypted partitions specified with luks.uuid=. | |
5811 | ||
5812 | * tmpfiles.d(5) snippets may now use specifier expansion in | |
5813 | path names. More specifically %m, %b, %H, %v, are now | |
5814 | replaced by the local machine id, boot id, hostname, and | |
5815 | kernel version number. | |
5816 | ||
5817 | * A new tmpfiles.d(5) command "m" has been introduced which | |
5818 | may be used to change the owner/group/access mode of a file | |
5819 | or directory if it exists, but do nothing if it does not. | |
5820 | ||
5821 | * This release removes high-level support for the | |
5822 | MemorySoftLimit= cgroup setting. The underlying kernel | |
5823 | cgroup attribute memory.soft_limit= is currently badly | |
5824 | designed and likely to be removed from the kernel API in its | |
5825 | current form, hence we should not expose it for now. | |
5826 | ||
5827 | * The memory.use_hierarchy cgroup attribute is now enabled for | |
5828 | all cgroups systemd creates in the memory cgroup | |
5829 | hierarchy. This option is likely to be come the built-in | |
5830 | default in the kernel anyway, and the non-hierarchical mode | |
5831 | never made much sense in the intrinsically hierarchical | |
5832 | cgroup system. | |
5833 | ||
5834 | * A new field _SYSTEMD_SLICE= is logged along with all journal | |
5835 | messages containing the slice a message was generated | |
5836 | from. This is useful to allow easy per-customer filtering of | |
5837 | logs among other things. | |
5838 | ||
5839 | * systemd-journald will no longer adjust the group of journal | |
5840 | files it creates to the "systemd-journal" group. Instead we | |
5841 | rely on the journal directory to be owned by the | |
5842 | "systemd-journal" group, and its setgid bit set, so that the | |
5843 | kernel file system layer will automatically enforce that | |
5844 | journal files inherit this group assignment. The reason for | |
5845 | this change is that we cannot allow NSS look-ups from | |
5846 | journald which would be necessary to resolve | |
5847 | "systemd-journal" to a numeric GID, because this might | |
5848 | create deadlocks if NSS involves synchronous queries to | |
5849 | other daemons (such as nscd, or sssd) which in turn are | |
5850 | logging clients of journald and might block on it, which | |
5851 | would then dead lock. A tmpfiles.d(5) snippet included in | |
5852 | systemd will make sure the setgid bit and group are | |
5853 | properly set on the journal directory if it exists on every | |
5854 | boot. However, we recommend adjusting it manually after | |
5855 | upgrades too (or from RPM scriptlets), so that the change is | |
5856 | not delayed until next reboot. | |
5857 | ||
5858 | * Backlight and random seed files in /var/lib/ have moved into | |
5859 | the /var/lib/systemd/ directory, in order to centralize all | |
5860 | systemd generated files in one directory. | |
5861 | ||
5862 | * Boot time performance measurements (as displayed by | |
5863 | "systemd-analyze" for example) will now read ACPI 5.0 FPDT | |
5864 | performance information if that's available to determine how | |
5865 | much time BIOS and boot loader initialization required. With | |
5866 | a sufficiently new BIOS you hence no longer need to boot | |
5867 | with Gummiboot to get access to such information. | |
5868 | ||
5869 | Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters, | |
5870 | Cristian Rodríguez, Dave Reisner, David Herrmann, David | |
5871 | Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao | |
5872 | feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart | |
5873 | Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt, | |
5874 | Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty, | |
5875 | Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek | |
5876 | ||
5877 | — Berlin, 2013-10-02 | |
5878 | ||
5879 | CHANGES WITH 207: | |
5880 | ||
5881 | * The Restart= option for services now understands a new | |
5882 | on-watchdog setting, which will restart the service | |
5883 | automatically if the service stops sending out watchdog keep | |
5884 | alive messages (as configured with WatchdogSec=). | |
5885 | ||
5886 | * The getty generator (which is responsible for bringing up a | |
5887 | getty on configured serial consoles) will no longer only | |
5888 | start a getty on the primary kernel console but on all | |
5889 | others, too. This makes the order in which console= is | |
5890 | specified on the kernel command line less important. | |
5891 | ||
5892 | * libsystemd-logind gained a new sd_session_get_vt() call to | |
5893 | retrieve the VT number of a session. | |
5894 | ||
5895 | * If the option "tries=0" is set for an entry of /etc/crypttab | |
5896 | its passphrase is queried indefinitely instead of any | |
5897 | maximum number of tries. | |
5898 | ||
5899 | * If a service with a configure PID file terminates its PID | |
5900 | file will now be removed automatically if it still exists | |
5901 | afterwards. This should put an end to stale PID files. | |
5902 | ||
5903 | * systemd-run will now also take relative binary path names | |
5904 | for execution and no longer insists on absolute paths. | |
5905 | ||
5906 | * InaccessibleDirectories= and ReadOnlyDirectories= now take | |
5907 | paths that are optionally prefixed with "-" to indicate that | |
5908 | it should not be considered a failure if they do not exist. | |
5909 | ||
5910 | * journalctl -o (and similar commands) now understands a new | |
5911 | output mode "short-precise", it is similar to "short" but | |
5912 | shows timestamps with usec accuracy. | |
5913 | ||
5914 | * The option "discard" (as known from Debian) is now | |
5915 | synonymous to "allow-discards" in /etc/crypttab. In fact, | |
5916 | "discard" is preferred now (since it is easier to remember | |
5917 | and type). | |
5918 | ||
5919 | * Some licensing clean-ups were made, so that more code is now | |
5920 | LGPL-2.1 licensed than before. | |
5921 | ||
5922 | * A minimal tool to save/restore the display backlight | |
5923 | brightness across reboots has been added. It will store the | |
5924 | backlight setting as late as possible at shutdown, and | |
5925 | restore it as early as possible during reboot. | |
5926 | ||
5927 | * A logic to automatically discover and enable home and swap | |
5928 | partitions on GPT disks has been added. With this in place | |
5929 | /etc/fstab becomes optional for many setups as systemd can | |
5930 | discover certain partitions located on the root disk | |
5931 | automatically. Home partitions are recognized under their | |
5932 | GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap | |
5933 | partitions are recognized under their GPT type ID | |
5934 | 0657fd6da4ab43c484e50933c84b4f4f. | |
5935 | ||
5936 | * systemd will no longer pass any environment from the kernel | |
5937 | or initrd to system services. If you want to set an | |
5938 | environment for all services, do so via the kernel command | |
5939 | line systemd.setenv= assignment. | |
5940 | ||
5941 | * The systemd-sysctl tool no longer natively reads the file | |
5942 | /etc/sysctl.conf. If desired, the file should be symlinked | |
5943 | from /etc/sysctl.d/99-sysctl.conf. Apart from providing | |
5944 | legacy support by a symlink rather than built-in code, it | |
5945 | also makes the otherwise hidden order of application of the | |
5946 | different files visible. (Note that this partly reverts to a | |
5947 | pre-198 application order of sysctl knobs!) | |
5948 | ||
5949 | * The "systemctl set-log-level" and "systemctl dump" commands | |
5950 | have been moved to systemd-analyze. | |
5951 | ||
5952 | * systemd-run learned the new --remain-after-exit switch, | |
5953 | which causes the scope unit not to be cleaned up | |
5954 | automatically after the process terminated. | |
5955 | ||
5956 | * tmpfiles learned a new --exclude-prefix= switch to exclude | |
5957 | certain paths from operation. | |
5958 | ||
5959 | * journald will now automatically flush all messages to disk | |
5960 | as soon as a message at the log level CRIT, ALERT or EMERG | |
5961 | is received. | |
5962 | ||
5963 | Contributions from: Andrew Cook, Brandon Philips, Christian | |
5964 | Hesse, Christoph Junghans, Colin Walters, Daniel Schaal, | |
5965 | Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George | |
5966 | McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer, | |
5967 | Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt, | |
5968 | Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering, | |
5969 | Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel | |
5970 | Holtmann, Martin Pitt, Michael Biebl, Michael Marineau, | |
5971 | Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał | |
5972 | Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn | |
5973 | Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe | |
5974 | Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao, | |
5975 | William Giokas, Zbigniew Jędrzejewski-Szmek | |
5976 | ||
5977 | — Berlin, 2013-09-13 | |
5978 | ||
5979 | CHANGES WITH 206: | |
5980 | ||
5981 | * The documentation has been updated to cover the various new | |
5982 | concepts introduced with 205. | |
5983 | ||
5984 | * Unit files now understand the new %v specifier which | |
5985 | resolves to the kernel version string as returned by "uname | |
5986 | -r". | |
5987 | ||
5988 | * systemctl now supports filtering the unit list output by | |
5989 | load state, active state and sub state, using the new | |
5990 | --state= parameter. | |
5991 | ||
5992 | * "systemctl status" will now show the results of the | |
5993 | condition checks (like ConditionPathExists= and similar) of | |
5994 | the last start attempts of the unit. They are also logged to | |
5995 | the journal. | |
5996 | ||
5997 | * "journalctl -b" may now be used to look for boot output of a | |
5998 | specific boot. Try "journalctl -b -1" for the previous boot, | |
5999 | but the syntax is substantially more powerful. | |
6000 | ||
6001 | * "journalctl --show-cursor" has been added which prints the | |
6002 | cursor string the last shown log line. This may then be used | |
6003 | with the new "journalctl --after-cursor=" switch to continue | |
6004 | browsing logs from that point on. | |
6005 | ||
6006 | * "journalctl --force" may now be used to force regeneration | |
6007 | of an FSS key. | |
6008 | ||
6009 | * Creation of "dead" device nodes has been moved from udev | |
6010 | into kmod and tmpfiles. Previously, udev would read the kmod | |
6011 | databases to pre-generate dead device nodes based on meta | |
6012 | information contained in kernel modules, so that these would | |
6013 | be auto-loaded on access rather then at boot. As this | |
6014 | does not really have much to do with the exposing actual | |
6015 | kernel devices to userspace this has always been slightly | |
6016 | alien in the udev codebase. Following the new scheme kmod | |
6017 | will now generate a runtime snippet for tmpfiles from the | |
6018 | module meta information and it now is tmpfiles' job to the | |
6019 | create the nodes. This also allows overriding access and | |
6020 | other parameters for the nodes using the usual tmpfiles | |
6021 | facilities. As side effect this allows us to remove the | |
6022 | CAP_SYS_MKNOD capability bit from udevd entirely. | |
6023 | ||
6024 | * logind's device ACLs may now be applied to these "dead" | |
6025 | devices nodes too, thus finally allowing managed access to | |
6026 | devices such as /dev/snd/sequencer without loading the | |
6027 | backing module right-away. | |
6028 | ||
6029 | * A new RPM macro has been added that may be used to apply | |
6030 | tmpfiles configuration during package installation. | |
6031 | ||
6032 | * systemd-detect-virt and ConditionVirtualization= now can | |
6033 | detect User-Mode-Linux machines (UML). | |
6034 | ||
6035 | * journald will now implicitly log the effective capabilities | |
6036 | set of processes in the message metadata. | |
6037 | ||
6038 | * systemd-cryptsetup has gained support for TrueCrypt volumes. | |
6039 | ||
6040 | * The initrd interface has been simplified (more specifically, | |
6041 | support for passing performance data via environment | |
6042 | variables and fsck results via files in /run has been | |
6043 | removed). These features were non-essential, and are | |
6044 | nowadays available in a much nicer way by having systemd in | |
6045 | the initrd serialize its state and have the hosts systemd | |
6046 | deserialize it again. | |
6047 | ||
6048 | * The udev "keymap" data files and tools to apply keyboard | |
6049 | specific mappings of scan to key codes, and force-release | |
6050 | scan code lists have been entirely replaced by a udev | |
6051 | "keyboard" builtin and a hwdb data file. | |
6052 | ||
6053 | * systemd will now honour the kernel's "quiet" command line | |
6054 | argument also during late shutdown, resulting in a | |
6055 | completely silent shutdown when used. | |
6056 | ||
6057 | * There's now an option to control the SO_REUSEPORT socket | |
6058 | option in .socket units. | |
6059 | ||
6060 | * Instance units will now automatically get a per-template | |
6061 | subslice of system.slice unless something else is explicitly | |
6062 | configured. For example, instances of sshd@.service will now | |
6063 | implicitly be placed in system-sshd.slice rather than | |
6064 | system.slice as before. | |
6065 | ||
6066 | * Test coverage support may now be enabled at build time. | |
6067 | ||
6068 | Contributions from: Dave Reisner, Frederic Crozat, Harald | |
6069 | Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan | |
6070 | Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart | |
6071 | Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael | |
6072 | Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden, | |
6073 | Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William | |
6074 | Giokas, Zbigniew Jędrzejewski-Szmek | |
6075 | ||
6076 | — Berlin, 2013-07-23 | |
6077 | ||
6078 | CHANGES WITH 205: | |
6079 | ||
6080 | * Two new unit types have been introduced: | |
6081 | ||
6082 | Scope units are very similar to service units, however, are | |
6083 | created out of pre-existing processes — instead of PID 1 | |
6084 | forking off the processes. By using scope units it is | |
6085 | possible for system services and applications to group their | |
6086 | own child processes (worker processes) in a powerful way | |
6087 | which then maybe used to organize them, or kill them | |
6088 | together, or apply resource limits on them. | |
6089 | ||
6090 | Slice units may be used to partition system resources in an | |
6091 | hierarchical fashion and then assign other units to them. By | |
6092 | default there are now three slices: system.slice (for all | |
6093 | system services), user.slice (for all user sessions), | |
6094 | machine.slice (for VMs and containers). | |
6095 | ||
6096 | Slices and scopes have been introduced primarily in | |
6097 | context of the work to move cgroup handling to a | |
6098 | single-writer scheme, where only PID 1 | |
6099 | creates/removes/manages cgroups. | |
6100 | ||
6101 | * There's a new concept of "transient" units. In contrast to | |
6102 | normal units these units are created via an API at runtime, | |
6103 | not from configuration from disk. More specifically this | |
6104 | means it is now possible to run arbitrary programs as | |
6105 | independent services, with all execution parameters passed | |
6106 | in via bus APIs rather than read from disk. Transient units | |
6107 | make systemd substantially more dynamic then it ever was, | |
6108 | and useful as a general batch manager. | |
6109 | ||
6110 | * logind has been updated to make use of scope and slice units | |
6111 | for managing user sessions. As a user logs in he will get | |
6112 | his own private slice unit, to which all sessions are added | |
6113 | as scope units. We also added support for automatically | |
6114 | adding an instance of user@.service for the user into the | |
6115 | slice. Effectively logind will no longer create cgroup | |
6116 | hierarchies on its own now, it will defer entirely to PID 1 | |
6117 | for this by means of scope, service and slice units. Since | |
6118 | user sessions this way become entities managed by PID 1 | |
6119 | the output of "systemctl" is now a lot more comprehensive. | |
6120 | ||
6121 | * A new mini-daemon "systemd-machined" has been added which | |
6122 | may be used by virtualization managers to register local | |
6123 | VMs/containers. nspawn has been updated accordingly, and | |
6124 | libvirt will be updated shortly. machined will collect a bit | |
6125 | of meta information about the VMs/containers, and assign | |
6126 | them their own scope unit (see above). The collected | |
6127 | meta-data is then made available via the "machinectl" tool, | |
6128 | and exposed in "ps" and similar tools. machined/machinectl | |
6129 | is compile-time optional. | |
6130 | ||
6131 | * As discussed earlier, the low-level cgroup configuration | |
6132 | options ControlGroup=, ControlGroupModify=, | |
6133 | ControlGroupPersistent=, ControlGroupAttribute= have been | |
6134 | removed. Please use high-level attribute settings instead as | |
6135 | well as slice units. | |
6136 | ||
6137 | * A new bus call SetUnitProperties() has been added to alter | |
6138 | various runtime parameters of a unit. This is primarily | |
6139 | useful to alter cgroup parameters dynamically in a nice way, | |
6140 | but will be extended later on to make more properties | |
6141 | modifiable at runtime. systemctl gained a new set-properties | |
6142 | command that wraps this call. | |
6143 | ||
6144 | * A new tool "systemd-run" has been added which can be used to | |
6145 | run arbitrary command lines as transient services or scopes, | |
6146 | while configuring a number of settings via the command | |
6147 | line. This tool is currently very basic, however already | |
6148 | very useful. We plan to extend this tool to even allow | |
6149 | queuing of execution jobs with time triggers from the | |
6150 | command line, similar in fashion to "at". | |
6151 | ||
6152 | * nspawn will now inform the user explicitly that kernels with | |
6153 | audit enabled break containers, and suggest the user to turn | |
6154 | off audit. | |
6155 | ||
6156 | * Support for detecting the IMA and AppArmor security | |
6157 | frameworks with ConditionSecurity= has been added. | |
6158 | ||
6159 | * journalctl gained a new "-k" switch for showing only kernel | |
6160 | messages, mimicking dmesg output; in addition to "--user" | |
6161 | and "--system" switches for showing only user's own logs | |
6162 | and system logs. | |
6163 | ||
6164 | * systemd-delta can now show information about drop-in | |
6165 | snippets extending unit files. | |
6166 | ||
6167 | * libsystemd-bus has been substantially updated but is still | |
6168 | not available as public API. | |
6169 | ||
6170 | * systemd will now look for the "debug" argument on the kernel | |
6171 | command line and enable debug logging, similar to what | |
6172 | "systemd.log_level=debug" already did before. | |
6173 | ||
6174 | * "systemctl set-default", "systemctl get-default" has been | |
6175 | added to configure the default.target symlink, which | |
6176 | controls what to boot into by default. | |
6177 | ||
6178 | * "systemctl set-log-level" has been added as a convenient | |
6179 | way to raise and lower systemd logging threshold. | |
6180 | ||
6181 | * "systemd-analyze plot" will now show the time the various | |
6182 | generators needed for execution, as well as information | |
6183 | about the unit file loading. | |
6184 | ||
6185 | * libsystemd-journal gained a new sd_journal_open_files() call | |
6186 | for opening specific journal files. journactl also gained a | |
6187 | new switch to expose this new functionality. Previously we | |
6188 | only supported opening all files from a directory, or all | |
6189 | files from the system, as opening individual files only is | |
6190 | racy due to journal file rotation. | |
6191 | ||
6192 | * systemd gained the new DefaultEnvironment= setting in | |
6193 | /etc/systemd/system.conf to set environment variables for | |
6194 | all services. | |
6195 | ||
6196 | * If a privileged process logs a journal message with the | |
6197 | OBJECT_PID= field set, then journald will automatically | |
6198 | augment this with additional OBJECT_UID=, OBJECT_GID=, | |
6199 | OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if | |
6200 | system services want to log events about specific client | |
6201 | processes. journactl/systemctl has been updated to make use | |
6202 | of this information if all log messages regarding a specific | |
6203 | unit is requested. | |
6204 | ||
6205 | Contributions from: Auke Kok, Chengwei Yang, Colin Walters, | |
6206 | Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave | |
6207 | Reisner, David Coppa, David King, David Strauss, Eelco | |
6208 | Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander | |
6209 | Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan | |
6210 | Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart | |
6211 | Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer, | |
6212 | Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer, | |
6213 | Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan, | |
6214 | Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern, | |
6215 | Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar, | |
6216 | Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek, | |
6217 | Łukasz Stelmach, 장동준 | |
6218 | ||
6219 | CHANGES WITH 204: | |
6220 | ||
6221 | * The Python bindings gained some minimal support for the APIs | |
6222 | exposed by libsystemd-logind. | |
6223 | ||
6224 | * ConditionSecurity= gained support for detecting SMACK. Since | |
6225 | this condition already supports SELinux and AppArmor we only | |
6226 | miss IMA for this. Patches welcome! | |
6227 | ||
6228 | Contributions from: Karol Lewandowski, Lennart Poettering, | |
6229 | Zbigniew Jędrzejewski-Szmek | |
6230 | ||
6231 | CHANGES WITH 203: | |
6232 | ||
6233 | * systemd-nspawn will now create /etc/resolv.conf if | |
6234 | necessary, before bind-mounting the host's file onto it. | |
6235 | ||
6236 | * systemd-nspawn will now store meta information about a | |
6237 | container on the container's cgroup as extended attribute | |
6238 | fields, including the root directory. | |
6239 | ||
6240 | * The cgroup hierarchy has been reworked in many ways. All | |
6241 | objects any of the components systemd creates in the cgroup | |
6242 | tree are now suffixed. More specifically, user sessions are | |
6243 | now placed in cgroups suffixed with ".session", users in | |
6244 | cgroups suffixed with ".user", and nspawn containers in | |
6245 | cgroups suffixed with ".nspawn". Furthermore, all cgroup | |
6246 | names are now escaped in a simple scheme to avoid collision | |
6247 | of userspace object names with kernel filenames. This work | |
6248 | is preparation for making these objects relocatable in the | |
6249 | cgroup tree, in order to allow easy resource partitioning of | |
6250 | these objects without causing naming conflicts. | |
6251 | ||
6252 | * systemctl list-dependencies gained the new switches | |
6253 | --plain, --reverse, --after and --before. | |
6254 | ||
6255 | * systemd-inhibit now shows the process name of processes that | |
6256 | have taken an inhibitor lock. | |
6257 | ||
6258 | * nss-myhostname will now also resolve "localhost" | |
6259 | implicitly. This makes /etc/hosts an optional file and | |
6260 | nicely handles that on IPv6 ::1 maps to both "localhost" and | |
6261 | the local hostname. | |
6262 | ||
6263 | * libsystemd-logind.so gained a new call | |
6264 | sd_get_machine_names() to enumerate running containers and | |
6265 | VMs (currently only supported by very new libvirt and | |
6266 | nspawn). sd_login_monitor can now be used to watch | |
6267 | VMs/containers coming and going. | |
6268 | ||
6269 | * .include is not allowed recursively anymore, and only in | |
6270 | unit files. Usually it is better to use drop-in snippets in | |
6271 | .d/*.conf anyway, as introduced with systemd 198. | |
6272 | ||
6273 | * systemd-analyze gained a new "critical-chain" command that | |
6274 | determines the slowest chain of units run during system | |
6275 | boot-up. It is very useful for tracking down where | |
6276 | optimizing boot time is the most beneficial. | |
6277 | ||
6278 | * systemd will no longer allow manipulating service paths in | |
6279 | the name=systemd:/system cgroup tree using ControlGroup= in | |
6280 | units. (But is still fine with it in all other dirs.) | |
6281 | ||
6282 | * There's a new systemd-nspawn@.service service file that may | |
6283 | be used to easily run nspawn containers as system | |
6284 | services. With the container's root directory in | |
6285 | /var/lib/container/foobar it is now sufficient to run | |
6286 | "systemctl start systemd-nspawn@foobar.service" to boot it. | |
6287 | ||
6288 | * systemd-cgls gained a new parameter "--machine" to list only | |
6289 | the processes within a certain container. | |
6290 | ||
6291 | * ConditionSecurity= now can check for "apparmor". We still | |
6292 | are lacking checks for SMACK and IMA for this condition | |
6293 | check though. Patches welcome! | |
6294 | ||
6295 | * A new configuration file /etc/systemd/sleep.conf has been | |
6296 | added that may be used to configure which kernel operation | |
6297 | systemd is supposed to execute when "suspend", "hibernate" | |
6298 | or "hybrid-sleep" is requested. This makes the new kernel | |
6299 | "freeze" state accessible to the user. | |
6300 | ||
6301 | * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape | |
6302 | the passed argument if applicable. | |
6303 | ||
6304 | Contributions from: Auke Kok, Colin Guthrie, Colin Walters, | |
6305 | Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner, | |
6306 | Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh | |
6307 | Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn, | |
6308 | MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel | |
6309 | Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom | |
6310 | Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew | |
6311 | Jędrzejewski-Szmek | |
6312 | ||
6313 | CHANGES WITH 202: | |
6314 | ||
6315 | * The output of 'systemctl list-jobs' got some polishing. The | |
6316 | '--type=' argument may now be passed more than once. A new | |
6317 | command 'systemctl list-sockets' has been added which shows | |
6318 | a list of kernel sockets systemd is listening on with the | |
6319 | socket units they belong to, plus the units these socket | |
6320 | units activate. | |
6321 | ||
6322 | * The experimental libsystemd-bus library got substantial | |
6323 | updates to work in conjunction with the (also experimental) | |
6324 | kdbus kernel project. It works well enough to exchange | |
6325 | messages with some sophistication. Note that kdbus is not | |
6326 | ready yet, and the library is mostly an elaborate test case | |
6327 | for now, and not installable. | |
6328 | ||
6329 | * systemd gained a new unit 'systemd-static-nodes.service' | |
6330 | that generates static device nodes earlier during boot, and | |
6331 | can run in conjunction with udev. | |
6332 | ||
6333 | * libsystemd-login gained a new call sd_pid_get_user_unit() | |
6334 | to retrieve the user systemd unit a process is running | |
6335 | in. This is useful for systems where systemd is used as | |
6336 | session manager. | |
6337 | ||
6338 | * systemd-nspawn now places all containers in the new /machine | |
6339 | top-level cgroup directory in the name=systemd | |
6340 | hierarchy. libvirt will soon do the same, so that we get a | |
6341 | uniform separation of /system, /user and /machine for system | |
6342 | services, user processes and containers/virtual | |
6343 | machines. This new cgroup hierarchy is also useful to stick | |
6344 | stable names to specific container instances, which can be | |
6345 | recognized later this way (this name may be controlled | |
6346 | via systemd-nspawn's new -M switch). libsystemd-login also | |
6347 | gained a new call sd_pid_get_machine_name() to retrieve the | |
6348 | name of the container/VM a specific process belongs to. | |
6349 | ||
6350 | * bootchart can now store its data in the journal. | |
6351 | ||
6352 | * libsystemd-journal gained a new call | |
6353 | sd_journal_add_conjunction() for AND expressions to the | |
6354 | matching logic. This can be used to express more complex | |
6355 | logical expressions. | |
6356 | ||
6357 | * journactl can now take multiple --unit= and --user-unit= | |
6358 | switches. | |
6359 | ||
6360 | * The cryptsetup logic now understands the "luks.key=" kernel | |
6361 | command line switch for specifying a file to read the | |
6362 | decryption key from. Also, if a configured key file is not | |
6363 | found the tool will now automatically fall back to prompting | |
6364 | the user. | |
6365 | ||
6366 | * Python systemd.journal module was updated to wrap recently | |
6367 | added functions from libsystemd-journal. The interface was | |
6368 | changed to bring the low level interface in s.j._Reader | |
6369 | closer to the C API, and the high level interface in | |
6370 | s.j.Reader was updated to wrap and convert all data about | |
6371 | an entry. | |
6372 | ||
6373 | Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer, | |
6374 | Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart | |
6375 | Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer, | |
6376 | Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt, | |
6377 | Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks, | |
6378 | Tom Gundersen, Zbigniew Jędrzejewski-Szmek | |
6379 | ||
6380 | CHANGES WITH 201: | |
6381 | ||
6382 | * journalctl --update-catalog now understands a new --root= | |
6383 | option to operate on catalogs found in a different root | |
6384 | directory. | |
6385 | ||
6386 | * During shutdown after systemd has terminated all running | |
6387 | services a final killing loop kills all remaining left-over | |
6388 | processes. We will now print the name of these processes | |
6389 | when we send SIGKILL to them, since this usually indicates a | |
6390 | problem. | |
6391 | ||
6392 | * If /etc/crypttab refers to password files stored on | |
6393 | configured mount points automatic dependencies will now be | |
6394 | generated to ensure the specific mount is established first | |
6395 | before the key file is attempted to be read. | |
6396 | ||
6397 | * 'systemctl status' will now show information about the | |
6398 | network sockets a socket unit is listening on. | |
6399 | ||
6400 | * 'systemctl status' will also shown information about any | |
6401 | drop-in configuration file for units. (Drop-In configuration | |
6402 | files in this context are files such as | |
6403 | /etc/systemd/systemd/foobar.service.d/*.conf) | |
6404 | ||
6405 | * systemd-cgtop now optionally shows summed up CPU times of | |
6406 | cgroups. Press '%' while running cgtop to switch between | |
6407 | percentage and absolute mode. This is useful to determine | |
6408 | which cgroups use up the most CPU time over the entire | |
6409 | runtime of the system. systemd-cgtop has also been updated | |
6410 | to be 'pipeable' for processing with further shell tools. | |
6411 | ||
6412 | * 'hostnamectl set-hostname' will now allow setting of FQDN | |
6413 | hostnames. | |
6414 | ||
6415 | * The formatting and parsing of time span values has been | |
6416 | changed. The parser now understands fractional expressions | |
6417 | such as "5.5h". The formatter will now output fractional | |
6418 | expressions for all time spans under 1min, i.e. "5.123456s" | |
6419 | rather than "5s 123ms 456us". For time spans under 1s | |
6420 | millisecond values are shown, for those under 1ms | |
6421 | microsecond values are shown. This should greatly improve | |
6422 | all time-related output of systemd. | |
6423 | ||
6424 | * libsystemd-login and libsystemd-journal gained new | |
6425 | functions for querying the poll() events mask and poll() | |
6426 | timeout value for integration into arbitrary event | |
6427 | loops. | |
6428 | ||
6429 | * localectl gained the ability to list available X11 keymaps | |
6430 | (models, layouts, variants, options). | |
6431 | ||
6432 | * 'systemd-analyze dot' gained the ability to filter for | |
6433 | specific units via shell-style globs, to create smaller, | |
6434 | more useful graphs. I.e. it is now possible to create simple | |
6435 | graphs of all the dependencies between only target units, or | |
6436 | of all units that Avahi has dependencies with. | |
6437 | ||
6438 | Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck, | |
6439 | Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly | |
6440 | Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau, | |
6441 | Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal | |
6442 | Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie, | |
6443 | Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav | |
6444 | Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach | |
6445 | ||
6446 | CHANGES WITH 200: | |
6447 | ||
6448 | * The boot-time readahead implementation for rotating media | |
6449 | will now read the read-ahead data in multiple passes which | |
6450 | consist of all read requests made in equidistant time | |
6451 | intervals. This means instead of strictly reading read-ahead | |
6452 | data in its physical order on disk we now try to find a | |
6453 | middle ground between physical and access time order. | |
6454 | ||
6455 | * /etc/os-release files gained a new BUILD_ID= field for usage | |
6456 | on operating systems that provide continuous builds of OS | |
6457 | images. | |
6458 | ||
6459 | Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers, | |
6460 | Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín | |
6461 | William Douglas, Zbigniew Jędrzejewski-Szmek | |
6462 | ||
6463 | CHANGES WITH 199: | |
6464 | ||
6465 | * systemd-python gained an API exposing libsystemd-daemon. | |
6466 | ||
6467 | * The SMACK setup logic gained support for uploading CIPSO | |
6468 | security policy. | |
6469 | ||
6470 | * Behaviour of PrivateTmp=, ReadWriteDirectories=, | |
6471 | ReadOnlyDirectories= and InaccessibleDirectories= has | |
6472 | changed. The private /tmp and /var/tmp directories are now | |
6473 | shared by all processes of a service (which means | |
6474 | ExecStartPre= may now leave data in /tmp that ExecStart= of | |
6475 | the same service can still access). When a service is | |
6476 | stopped its temporary directories are immediately deleted | |
6477 | (normal clean-up with tmpfiles is still done in addition to | |
6478 | this though). | |
6479 | ||
6480 | * By default, systemd will now set a couple of sysctl | |
6481 | variables in the kernel: the safe sysrq options are turned | |
6482 | on, IP route verification is turned on, and source routing | |
6483 | disabled. The recently added hardlink and softlink | |
6484 | protection of the kernel is turned on. These settings should | |
6485 | be reasonably safe, and good defaults for all new systems. | |
6486 | ||
6487 | * The predictable network naming logic may now be turned off | |
6488 | with a new kernel command line switch: net.ifnames=0. | |
6489 | ||
6490 | * A new libsystemd-bus module has been added that implements a | |
6491 | pretty complete D-Bus client library. For details see: | |
6492 | ||
6493 | https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html | |
6494 | ||
6495 | * journald will now explicitly flush the journal files to disk | |
6496 | at the latest 5min after each write. The file will then also | |
6497 | be marked offline until the next write. This should increase | |
6498 | reliability in case of a crash. The synchronization delay | |
6499 | can be configured via SyncIntervalSec= in journald.conf. | |
6500 | ||
6501 | * There's a new remote-fs-setup.target unit that can be used | |
6502 | to pull in specific services when at least one remote file | |
6503 | system is to be mounted. | |
6504 | ||
6505 | * There are new targets timers.target and paths.target as | |
6506 | canonical targets to pull user timer and path units in | |
6507 | from. This complements sockets.target with a similar | |
6508 | purpose for socket units. | |
6509 | ||
6510 | * libudev gained a new call udev_device_set_attribute_value() | |
6511 | to set sysfs attributes of a device. | |
6512 | ||
6513 | * The udev daemon now sets the default number of worker | |
6514 | processes executed in parallel based on the number of available | |
6515 | CPUs instead of the amount of available RAM. This is supposed | |
6516 | to provide a more reliable default and limit a too aggressive | |
6517 | parallelism for setups with 1000s of devices connected. | |
6518 | ||
6519 | Contributions from: Auke Kok, Colin Walters, Cristian | |
6520 | Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes | |
6521 | Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan | |
6522 | Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering, | |
6523 | Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl, | |
6524 | Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen, | |
6525 | Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel | |
6526 | Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, | |
6527 | Zbigniew Jędrzejewski-Szmek | |
6528 | ||
6529 | CHANGES WITH 198: | |
6530 | ||
6531 | * Configuration of unit files may now be extended via drop-in | |
6532 | files without having to edit/override the unit files | |
6533 | themselves. More specifically, if the administrator wants to | |
6534 | change one value for a service file foobar.service he can | |
6535 | now do so by dropping in a configuration snippet into | |
6536 | /etc/systemd/system/foobar.service.d/*.conf. The unit logic | |
6537 | will load all these snippets and apply them on top of the | |
6538 | main unit configuration file, possibly extending or | |
6539 | overriding its settings. Using these drop-in snippets is | |
6540 | generally nicer than the two earlier options for changing | |
6541 | unit files locally: copying the files from | |
6542 | /usr/lib/systemd/system/ to /etc/systemd/system/ and editing | |
6543 | them there; or creating a new file in /etc/systemd/system/ | |
6544 | that incorporates the original one via ".include". Drop-in | |
6545 | snippets into these .d/ directories can be placed in any | |
6546 | directory systemd looks for units in, and the usual | |
6547 | overriding semantics between /usr/lib, /etc and /run apply | |
6548 | for them too. | |
6549 | ||
6550 | * Most unit file settings which take lists of items can now be | |
6551 | reset by assigning the empty string to them. For example, | |
6552 | normally, settings such as Environment=FOO=BAR append a new | |
6553 | environment variable assignment to the environment block, | |
6554 | each time they are used. By assigning Environment= the empty | |
6555 | string the environment block can be reset to empty. This is | |
6556 | particularly useful with the .d/*.conf drop-in snippets | |
6557 | mentioned above, since this adds the ability to reset list | |
6558 | settings from vendor unit files via these drop-ins. | |
6559 | ||
6560 | * systemctl gained a new "list-dependencies" command for | |
6561 | listing the dependencies of a unit recursively. | |
6562 | ||
6563 | * Inhibitors are now honored and listed by "systemctl | |
6564 | suspend", "systemctl poweroff" (and similar) too, not only | |
6565 | GNOME. These commands will also list active sessions by | |
6566 | other users. | |
6567 | ||
6568 | * Resource limits (as exposed by the various control group | |
6569 | controllers) can now be controlled dynamically at runtime | |
6570 | for all units. More specifically, you can now use a command | |
6571 | like "systemctl set-cgroup-attr foobar.service cpu.shares | |
6572 | 2000" to alter the CPU shares a specific service gets. These | |
6573 | settings are stored persistently on disk, and thus allow the | |
6574 | administrator to easily adjust the resource usage of | |
6575 | services with a few simple commands. This dynamic resource | |
6576 | management logic is also available to other programs via the | |
6577 | bus. Almost any kernel cgroup attribute and controller is | |
6578 | supported. | |
6579 | ||
6580 | * systemd-vconsole-setup will now copy all font settings to | |
6581 | all allocated VTs, where it previously applied them only to | |
6582 | the foreground VT. | |
6583 | ||
6584 | * libsystemd-login gained the new sd_session_get_tty() API | |
6585 | call. | |
6586 | ||
6587 | * This release drops support for a few legacy or | |
6588 | distribution-specific LSB facility names when parsing init | |
6589 | scripts: $x-display-manager, $mail-transfer-agent, | |
6590 | $mail-transport-agent, $mail-transfer-agent, $smtp, | |
6591 | $null. Also, the mail-transfer-agent.target unit backing | |
6592 | this has been removed. Distributions which want to retain | |
6593 | compatibility with this should carry the burden for | |
6594 | supporting this themselves and patch support for these back | |
6595 | in, if they really need to. Also, the facilities $syslog and | |
6596 | $local_fs are now ignored, since systemd does not support | |
6597 | early-boot LSB init scripts anymore, and these facilities | |
6598 | are implied anyway for normal services. syslog.target has | |
6599 | also been removed. | |
6600 | ||
6601 | * There are new bus calls on PID1's Manager object for | |
6602 | cancelling jobs, and removing snapshot units. Previously, | |
6603 | both calls were only available on the Job and Snapshot | |
6604 | objects themselves. | |
6605 | ||
6606 | * systemd-journal-gatewayd gained SSL support. | |
6607 | ||
6608 | * The various "environment" files, such as /etc/locale.conf | |
6609 | now support continuation lines with a backslash ("\") as | |
6610 | last character in the line, similarly in style (but different) | |
6611 | to how this is supported in shells. | |
6612 | ||
6613 | * For normal user processes the _SYSTEMD_USER_UNIT= field is | |
6614 | now implicitly appended to every log entry logged. systemctl | |
6615 | has been updated to filter by this field when operating on a | |
6616 | user systemd instance. | |
6617 | ||
6618 | * nspawn will now implicitly add the CAP_AUDIT_WRITE and | |
6619 | CAP_AUDIT_CONTROL capabilities to the capabilities set for | |
6620 | the container. This makes it easier to boot unmodified | |
6621 | Fedora systems in a container, which however still requires | |
6622 | audit=0 to be passed on the kernel command line. Auditing in | |
6623 | kernel and userspace is unfortunately still too broken in | |
6624 | context of containers, hence we recommend compiling it out | |
6625 | of the kernel or using audit=0. Hopefully this will be fixed | |
6626 | one day for good in the kernel. | |
6627 | ||
6628 | * nspawn gained the new --bind= and --bind-ro= parameters to | |
6629 | bind mount specific directories from the host into the | |
6630 | container. | |
6631 | ||
6632 | * nspawn will now mount its own devpts file system instance | |
6633 | into the container, in order not to leak pty devices from | |
6634 | the host into the container. | |
6635 | ||
6636 | * systemd will now read the firmware boot time performance | |
6637 | information from the EFI variables, if the used boot loader | |
6638 | supports this, and takes it into account for boot performance | |
6639 | analysis via "systemd-analyze". This is currently supported | |
6640 | only in conjunction with Gummiboot, but could be supported | |
6641 | by other boot loaders too. For details see: | |
6642 | ||
6643 | https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface | |
6644 | ||
6645 | * A new generator has been added that automatically mounts the | |
6646 | EFI System Partition (ESP) to /boot, if that directory | |
6647 | exists, is empty, and no other file system has been | |
6648 | configured to be mounted there. | |
6649 | ||
6650 | * logind will now send out PrepareForSleep(false) out | |
6651 | unconditionally, after coming back from suspend. This may be | |
6652 | used by applications as asynchronous notification for | |
6653 | system resume events. | |
6654 | ||
6655 | * "systemctl unlock-sessions" has been added, that allows | |
6656 | unlocking the screens of all user sessions at once, similar | |
6657 | to how "systemctl lock-sessions" already locked all users | |
6658 | sessions. This is backed by a new D-Bus call UnlockSessions(). | |
6659 | ||
6660 | * "loginctl seat-status" will now show the master device of a | |
6661 | seat. (i.e. the device of a seat that needs to be around for | |
6662 | the seat to be considered available, usually the graphics | |
6663 | card). | |
6664 | ||
6665 | * tmpfiles gained a new "X" line type, that allows | |
6666 | configuration of files and directories (with wildcards) that | |
6667 | shall be excluded from automatic cleanup ("aging"). | |
6668 | ||
6669 | * udev default rules set the device node permissions now only | |
6670 | at "add" events, and do not change them any longer with a | |
6671 | later "change" event. | |
6672 | ||
6673 | * The log messages for lid events and power/sleep keypresses | |
6674 | now carry a message ID. | |
6675 | ||
6676 | * We now have a substantially larger unit test suite, but this | |
6677 | continues to be work in progress. | |
6678 | ||
6679 | * udevadm hwdb gained a new --root= parameter to change the | |
6680 | root directory to operate relative to. | |
6681 | ||
6682 | * logind will now issue a background sync() request to the kernel | |
6683 | early at shutdown, so that dirty buffers are flushed to disk early | |
6684 | instead of at the last moment, in order to optimize shutdown | |
6685 | times a little. | |
6686 | ||
6687 | * A new bootctl tool has been added that is an interface for | |
6688 | certain boot loader operations. This is currently a preview | |
6689 | and is likely to be extended into a small mechanism daemon | |
6690 | like timedated, localed, hostnamed, and can be used by | |
6691 | graphical UIs to enumerate available boot options, and | |
6692 | request boot into firmware operations. | |
6693 | ||
6694 | * systemd-bootchart has been relicensed to LGPLv2.1+ to match | |
6695 | the rest of the package. It also has been updated to work | |
6696 | correctly in initrds. | |
6697 | ||
6698 | * Policykit previously has been runtime optional, and is now | |
6699 | also compile time optional via a configure switch. | |
6700 | ||
6701 | * systemd-analyze has been reimplemented in C. Also "systemctl | |
6702 | dot" has moved into systemd-analyze. | |
6703 | ||
6704 | * "systemctl status" with no further parameters will now print | |
6705 | the status of all active or failed units. | |
6706 | ||
6707 | * Operations such as "systemctl start" can now be executed | |
6708 | with a new mode "--irreversible" which may be used to queue | |
6709 | operations that cannot accidentally be reversed by a later | |
6710 | job queuing. This is by default used to make shutdown | |
6711 | requests more robust. | |
6712 | ||
6713 | * The Python API of systemd now gained a new module for | |
6714 | reading journal files. | |
6715 | ||
6716 | * A new tool kernel-install has been added that can install | |
6717 | kernel images according to the Boot Loader Specification: | |
6718 | ||
6719 | https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec | |
6720 | ||
6721 | * Boot time console output has been improved to provide | |
6722 | animated boot time output for hanging jobs. | |
6723 | ||
6724 | * A new tool systemd-activate has been added which can be used | |
6725 | to test socket activation with, directly from the command | |
6726 | line. This should make it much easier to test and debug | |
6727 | socket activation in daemons. | |
6728 | ||
6729 | * journalctl gained a new "--reverse" (or -r) option to show | |
6730 | journal output in reverse order (i.e. newest line first). | |
6731 | ||
6732 | * journalctl gained a new "--pager-end" (or -e) option to jump | |
6733 | to immediately jump to the end of the journal in the | |
6734 | pager. This is only supported in conjunction with "less". | |
6735 | ||
6736 | * journalctl gained a new "--user-unit=" option, that works | |
6737 | similarly to "--unit=" but filters for user units rather than | |
6738 | system units. | |
6739 | ||
6740 | * A number of unit files to ease adoption of systemd in | |
6741 | initrds has been added. This moves some minimal logic from | |
6742 | the various initrd implementations into systemd proper. | |
6743 | ||
6744 | * The journal files are now owned by a new group | |
6745 | "systemd-journal", which exists specifically to allow access | |
6746 | to the journal, and nothing else. Previously, we used the | |
6747 | "adm" group for that, which however possibly covers more | |
6748 | than just journal/log file access. This new group is now | |
6749 | already used by systemd-journal-gatewayd to ensure this | |
6750 | daemon gets access to the journal files and as little else | |
6751 | as possible. Note that "make install" will also set FS ACLs | |
6752 | up for /var/log/journal to give "adm" and "wheel" read | |
6753 | access to it, in addition to "systemd-journal" which owns | |
6754 | the journal files. We recommend that packaging scripts also | |
6755 | add read access to "adm" + "wheel" to /var/log/journal, and | |
6756 | all existing/future journal files. To normal users and | |
6757 | administrators little changes, however packagers need to | |
6758 | ensure to create the "systemd-journal" system group at | |
6759 | package installation time. | |
6760 | ||
6761 | * The systemd-journal-gatewayd now runs as unprivileged user | |
6762 | systemd-journal-gateway:systemd-journal-gateway. Packaging | |
6763 | scripts need to create these system user/group at | |
6764 | installation time. | |
6765 | ||
6766 | * timedated now exposes a new boolean property CanNTP that | |
6767 | indicates whether a local NTP service is available or not. | |
6768 | ||
6769 | * systemd-detect-virt will now also detect xen PVs | |
6770 | ||
6771 | * The pstore file system is now mounted by default, if it is | |
6772 | available. | |
6773 | ||
6774 | * In addition to the SELinux and IMA policies we will now also | |
6775 | load SMACK policies at early boot. | |
6776 | ||
6777 | Contributions from: Adel Gadllah, Aleksander Morgado, Auke | |
6778 | Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch, | |
6779 | Daniel Wallace, Dave Reisner, David Herrmann, David Strauss, | |
6780 | Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer, | |
6781 | Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering, | |
6782 | Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin | |
6783 | Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael | |
6784 | Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil, | |
6785 | Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor | |
6786 | Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob | |
6787 | Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven | |
6788 | Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom | |
6789 | Gundersen, Umut Tezduyar, William Giokas, Zbigniew | |
6790 | Jędrzejewski-Szmek, Zeeshan Ali (Khattak) | |
6791 | ||
6792 | CHANGES WITH 197: | |
6793 | ||
6794 | * Timer units now support calendar time events in addition to | |
6795 | monotonic time events. That means you can now trigger a unit | |
6796 | based on a calendar time specification such as "Thu,Fri | |
6797 | 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first | |
6798 | or fifth day of any month of the year 2013, given that it is | |
6799 | a thursday or friday. This brings timer event support | |
6800 | considerably closer to cron's capabilities. For details on | |
6801 | the supported calendar time specification language see | |
6802 | systemd.time(7). | |
6803 | ||
6804 | * udev now supports a number of different naming policies for | |
6805 | network interfaces for predictable names, and a combination | |
6806 | of these policies is now the default. Please see this wiki | |
6807 | document for details: | |
6808 | ||
6809 | https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames | |
6810 | ||
6811 | * Auke Kok's bootchart implementation has been added to the | |
6812 | systemd tree. It is an optional component that can graph the | |
6813 | boot in quite some detail. It is one of the best bootchart | |
6814 | implementations around and minimal in its code and | |
6815 | dependencies. | |
6816 | ||
6817 | * nss-myhostname has been integrated into the systemd source | |
6818 | tree. nss-myhostname guarantees that the local hostname | |
6819 | always stays resolvable via NSS. It has been a weak | |
6820 | requirement of systemd-hostnamed since a long time, and | |
6821 | since its code is actually trivial we decided to just | |
6822 | include it in systemd's source tree. It can be turned off | |
6823 | with a configure switch. | |
6824 | ||
6825 | * The read-ahead logic is now capable of properly detecting | |
6826 | whether a btrfs file system is on SSD or rotating media, in | |
6827 | order to optimize the read-ahead scheme. Previously, it was | |
6828 | only capable of detecting this on traditional file systems | |
6829 | such as ext4. | |
6830 | ||
6831 | * In udev, additional device properties are now read from the | |
6832 | IAB in addition to the OUI database. Also, Bluetooth company | |
6833 | identities are attached to the devices as well. | |
6834 | ||
6835 | * In service files %U may be used as specifier that is | |
6836 | replaced by the configured user name of the service. | |
6837 | ||
6838 | * nspawn may now be invoked without a controlling TTY. This | |
6839 | makes it suitable for invocation as its own service. This | |
6840 | may be used to set up a simple containerized server system | |
6841 | using only core OS tools. | |
6842 | ||
6843 | * systemd and nspawn can now accept socket file descriptors | |
6844 | when they are started for socket activation. This enables | |
6845 | implementation of socket activated nspawn | |
6846 | containers. i.e. think about autospawning an entire OS image | |
6847 | when the first SSH or HTTP connection is received. We expect | |
6848 | that similar functionality will also be added to libvirt-lxc | |
6849 | eventually. | |
6850 | ||
6851 | * journalctl will now suppress ANSI color codes when | |
6852 | presenting log data. | |
6853 | ||
6854 | * systemctl will no longer show control group information for | |
6855 | a unit if the control group is empty anyway. | |
6856 | ||
6857 | * logind can now automatically suspend/hibernate/shutdown the | |
6858 | system on idle. | |
6859 | ||
6860 | * /etc/machine-info and hostnamed now also expose the chassis | |
6861 | type of the system. This can be used to determine whether | |
6862 | the local system is a laptop, desktop, handset or | |
6863 | tablet. This information may either be configured by the | |
6864 | user/vendor or is automatically determined from ACPI and DMI | |
6865 | information if possible. | |
6866 | ||
6867 | * A number of PolicyKit actions are now bound together with | |
6868 | "imply" rules. This should simplify creating UIs because | |
6869 | many actions will now authenticate similar ones as well. | |
6870 | ||
6871 | * Unit files learnt a new condition ConditionACPower= which | |
6872 | may be used to conditionalize a unit depending on whether an | |
6873 | AC power source is connected or not, of whether the system | |
6874 | is running on battery power. | |
6875 | ||
6876 | * systemctl gained a new "is-failed" verb that may be used in | |
6877 | shell scripts and suchlike to check whether a specific unit | |
6878 | is in the "failed" state. | |
6879 | ||
6880 | * The EnvironmentFile= setting in unit files now supports file | |
6881 | globbing, and can hence be used to easily read a number of | |
6882 | environment files at once. | |
6883 | ||
6884 | * systemd will no longer detect and recognize specific | |
6885 | distributions. All distribution-specific #ifdeffery has been | |
6886 | removed, systemd is now fully generic and | |
6887 | distribution-agnostic. Effectively, not too much is lost as | |
6888 | a lot of the code is still accessible via explicit configure | |
6889 | switches. However, support for some distribution specific | |
6890 | legacy configuration file formats has been dropped. We | |
6891 | recommend distributions to simply adopt the configuration | |
6892 | files everybody else uses now and convert the old | |
6893 | configuration from packaging scripts. Most distributions | |
6894 | already did that. If that's not possible or desirable, | |
6895 | distributions are welcome to forward port the specific | |
6896 | pieces of code locally from the git history. | |
6897 | ||
6898 | * When logging a message about a unit systemd will now always | |
6899 | log the unit name in the message meta data. | |
6900 | ||
6901 | * localectl will now also discover system locale data that is | |
6902 | not stored in locale archives, but directly unpacked. | |
6903 | ||
6904 | * logind will no longer unconditionally use framebuffer | |
6905 | devices as seat masters, i.e. as devices that are required | |
6906 | to be existing before a seat is considered preset. Instead, | |
6907 | it will now look for all devices that are tagged as | |
6908 | "seat-master" in udev. By default, framebuffer devices will | |
6909 | be marked as such, but depending on local systems, other | |
6910 | devices might be marked as well. This may be used to | |
6911 | integrate graphics cards using closed source drivers (such | |
6912 | as NVidia ones) more nicely into logind. Note however, that | |
6913 | we recommend using the open source NVidia drivers instead, | |
6914 | and no udev rules for the closed-source drivers will be | |
6915 | shipped from us upstream. | |
6916 | ||
6917 | Contributions from: Adam Williamson, Alessandro Crismani, Auke | |
6918 | Kok, Colin Walters, Daniel Wallace, Dave Reisner, David | |
6919 | Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra, | |
6920 | Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik | |
6921 | Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart | |
6922 | Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann, | |
6923 | Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry, | |
6924 | Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg | |
6925 | Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar | |
6926 | Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn | |
6927 | Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch, | |
6928 | Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew | |
6929 | Jędrzejewski-Szmek | |
6930 | ||
6931 | CHANGES WITH 196: | |
6932 | ||
6933 | * udev gained support for loading additional device properties | |
6934 | from an indexed database that is keyed by vendor/product IDs | |
6935 | and similar device identifiers. For the beginning this | |
6936 | "hwdb" is populated with data from the well-known PCI and | |
6937 | USB database, but also includes PNP, ACPI and OID data. In | |
6938 | the longer run this indexed database shall grow into | |
6939 | becoming the one central database for non-essential | |
6940 | userspace device metadata. Previously, data from the PCI/USB | |
6941 | database was only attached to select devices, since the | |
6942 | lookup was a relatively expensive operation due to O(n) time | |
6943 | complexity (with n being the number of entries in the | |
6944 | database). Since this is now O(1), we decided to add in this | |
6945 | data for all devices where this is available, by | |
6946 | default. Note that the indexed database needs to be rebuilt | |
6947 | when new data files are installed. To achieve this you need | |
6948 | to update your packaging scripts to invoke "udevadm hwdb | |
6949 | --update" after installation of hwdb data files. For | |
6950 | RPM-based distributions we introduced the new | |
6951 | %udev_hwdb_update macro for this purpose. | |
6952 | ||
6953 | * The Journal gained support for the "Message Catalog", an | |
6954 | indexed database to link up additional information with | |
6955 | journal entries. For further details please check: | |
6956 | ||
6957 | https://www.freedesktop.org/wiki/Software/systemd/catalog | |
6958 | ||
6959 | The indexed message catalog database also needs to be | |
6960 | rebuilt after installation of message catalog files. Use | |
6961 | "journalctl --update-catalog" for this. For RPM-based | |
6962 | distributions we introduced the %journal_catalog_update | |
6963 | macro for this purpose. | |
6964 | ||
6965 | * The Python Journal bindings gained support for the standard | |
6966 | Python logging framework. | |
6967 | ||
6968 | * The Journal API gained new functions for checking whether | |
6969 | the underlying file system of a journal file is capable of | |
6970 | properly reporting file change notifications, or whether | |
6971 | applications that want to reflect journal changes "live" | |
6972 | need to recheck journal files continuously in appropriate | |
6973 | time intervals. | |
6974 | ||
6975 | * It is now possible to set the "age" field for tmpfiles | |
6976 | entries to 0, indicating that files matching this entry | |
6977 | shall always be removed when the directories are cleaned up. | |
6978 | ||
6979 | * coredumpctl gained a new "gdb" verb which invokes gdb | |
6980 | right-away on the selected coredump. | |
6981 | ||
6982 | * There's now support for "hybrid sleep" on kernels that | |
6983 | support this, in addition to "suspend" and "hibernate". Use | |
6984 | "systemctl hybrid-sleep" to make use of this. | |
6985 | ||
6986 | * logind's HandleSuspendKey= setting (and related settings) | |
6987 | now gained support for a new "lock" setting to simply | |
6988 | request the screen lock on all local sessions, instead of | |
6989 | actually executing a suspend or hibernation. | |
6990 | ||
6991 | * systemd will now mount the EFI variables file system by | |
6992 | default. | |
6993 | ||
6994 | * Socket units now gained support for configuration of the | |
6995 | SMACK security label. | |
6996 | ||
6997 | * timedatectl will now output the time of the last and next | |
6998 | daylight saving change. | |
6999 | ||
7000 | * We dropped support for various legacy and distro-specific | |
7001 | concepts, such as insserv, early-boot SysV services | |
7002 | (i.e. those for non-standard runlevels such as 'b' or 'S') | |
7003 | or ArchLinux /etc/rc.conf support. We recommend the | |
7004 | distributions who still need support this to either continue | |
7005 | to maintain the necessary patches downstream, or find a | |
7006 | different solution. (Talk to us if you have questions!) | |
7007 | ||
7008 | * Various systemd components will now bypass PolicyKit checks | |
7009 | for root and otherwise handle properly if PolicyKit is not | |
7010 | found to be around. This should fix most issues for | |
7011 | PolicyKit-less systems. Quite frankly this should have been | |
7012 | this way since day one. It is absolutely our intention to | |
7013 | make systemd work fine on PolicyKit-less systems, and we | |
7014 | consider it a bug if something does not work as it should if | |
7015 | PolicyKit is not around. | |
7016 | ||
7017 | * For embedded systems it is now possible to build udev and | |
7018 | systemd without blkid and/or kmod support. | |
7019 | ||
7020 | * "systemctl switch-root" is now capable of switching root | |
7021 | more than once. I.e. in addition to transitions from the | |
7022 | initrd to the host OS it is now possible to transition to | |
7023 | further OS images from the host. This is useful to implement | |
7024 | offline updating tools. | |
7025 | ||
7026 | * Various other additions have been made to the RPM macros | |
7027 | shipped with systemd. Use %udev_rules_update() after | |
7028 | installing new udev rules files. %_udevhwdbdir, | |
7029 | %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir, | |
7030 | %_sysctldir are now available which resolve to the right | |
7031 | directories for packages to place various data files in. | |
7032 | ||
7033 | * journalctl gained the new --full switch (in addition to | |
7034 | --all, to disable ellipsation for long messages. | |
7035 | ||
7036 | Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel, | |
7037 | Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner, | |
7038 | Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers, | |
7039 | Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas, | |
7040 | Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl, | |
7041 | Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen, | |
7042 | Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas | |
7043 | Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony | |
7044 | Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek | |
7045 | ||
7046 | CHANGES WITH 195: | |
7047 | ||
7048 | * journalctl gained new --since= and --until= switches to | |
7049 | filter by time. It also now supports nice filtering for | |
7050 | units via --unit=/-u. | |
7051 | ||
7052 | * Type=oneshot services may use ExecReload= and do the | |
7053 | right thing. | |
7054 | ||
7055 | * The journal daemon now supports time-based rotation and | |
7056 | vacuuming, in addition to the usual disk-space based | |
7057 | rotation. | |
7058 | ||
7059 | * The journal will now index the available field values for | |
7060 | each field name. This enables clients to show pretty drop | |
7061 | downs of available match values when filtering. The bash | |
7062 | completion of journalctl has been updated | |
7063 | accordingly. journalctl gained a new switch -F to list all | |
7064 | values a certain field takes in the journal database. | |
7065 | ||
7066 | * More service events are now written as structured messages | |
7067 | to the journal, and made recognizable via message IDs. | |
7068 | ||
7069 | * The timedated, localed and hostnamed mini-services which | |
7070 | previously only provided support for changing time, locale | |
7071 | and hostname settings from graphical DEs such as GNOME now | |
7072 | also have a minimal (but very useful) text-based client | |
7073 | utility each. This is probably the nicest way to changing | |
7074 | these settings from the command line now, especially since | |
7075 | it lists available options and is fully integrated with bash | |
7076 | completion. | |
7077 | ||
7078 | * There's now a new tool "systemd-coredumpctl" to list and | |
7079 | extract coredumps from the journal. | |
7080 | ||
7081 | * We now install a README each in /var/log/ and | |
7082 | /etc/rc.d/init.d explaining where the system logs and init | |
7083 | scripts went. This hopefully should help folks who go to | |
7084 | that dirs and look into the otherwise now empty void and | |
7085 | scratch their heads. | |
7086 | ||
7087 | * When user-services are invoked (by systemd --user) the | |
7088 | $MANAGERPID env var is set to the PID of systemd. | |
7089 | ||
7090 | * SIGRTMIN+24 when sent to a --user instance will now result | |
7091 | in immediate termination of systemd. | |
7092 | ||
7093 | * gatewayd received numerous feature additions such as a | |
7094 | "follow" mode, for live syncing and filtering. | |
7095 | ||
7096 | * browse.html now allows filtering and showing detailed | |
7097 | information on specific entries. Keyboard navigation and | |
7098 | mouse screen support has been added. | |
7099 | ||
7100 | * gatewayd/journalctl now supports HTML5/JSON | |
7101 | Server-Sent-Events as output. | |
7102 | ||
7103 | * The SysV init script compatibility logic will now | |
7104 | heuristically determine whether a script supports the | |
7105 | "reload" verb, and only then make this available as | |
7106 | "systemctl reload". | |
7107 | ||
7108 | * "systemctl status --follow" has been removed, use "journalctl | |
7109 | -u" instead. | |
7110 | ||
7111 | * journald.conf's RuntimeMinSize=, PersistentMinSize= settings | |
7112 | have been removed since they are hardly useful to be | |
7113 | configured. | |
7114 | ||
7115 | * And I'd like to take the opportunity to specifically mention | |
7116 | Zbigniew for his great contributions. Zbigniew, you rock! | |
7117 | ||
7118 | Contributions from: Andrew Eikum, Christian Hesse, Colin | |
7119 | Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc | |
7120 | Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas | |
7121 | Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich, | |
7122 | Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas | |
7123 | Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew | |
7124 | Jędrzejewski-Szmek, Сковорода Никита Андреевич | |
7125 | ||
7126 | CHANGES WITH 194: | |
7127 | ||
7128 | * If /etc/vconsole.conf is non-existent or empty we will no | |
7129 | longer load any console font or key map at boot by | |
7130 | default. Instead the kernel defaults will be left | |
7131 | intact. This is definitely the right thing to do, as no | |
7132 | configuration should mean no configuration, and hard-coding | |
7133 | font names that are different on all archs is probably a bad | |
7134 | idea. Also, the kernel default key map and font should be | |
7135 | good enough for most cases anyway, and mostly identical to | |
7136 | the userspace fonts/key maps we previously overloaded them | |
7137 | with. If distributions want to continue to default to a | |
7138 | non-kernel font or key map they should ship a default | |
7139 | /etc/vconsole.conf with the appropriate contents. | |
7140 | ||
7141 | Contributions from: Colin Walters, Daniel J Walsh, Dave | |
7142 | Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef | |
7143 | Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek | |
7144 | ||
7145 | CHANGES WITH 193: | |
7146 | ||
7147 | * journalctl gained a new --cursor= switch to show entries | |
7148 | starting from the specified location in the journal. | |
7149 | ||
7150 | * We now enforce a size limit on journal entry fields exported | |
7151 | with "-o json" in journalctl. Fields larger than 4K will be | |
7152 | assigned null. This can be turned off with --all. | |
7153 | ||
7154 | * An (optional) journal gateway daemon is now available as | |
7155 | "systemd-journal-gatewayd.service". This service provides | |
7156 | access to the journal via HTTP and JSON. This functionality | |
7157 | will be used to implement live log synchronization in both | |
7158 | pull and push modes, but has various other users too, such | |
7159 | as easy log access for debugging of embedded devices. Right | |
7160 | now it is already useful to retrieve the journal via HTTP: | |
7161 | ||
7162 | # systemctl start systemd-journal-gatewayd.service | |
7163 | # wget http://localhost:19531/entries | |
7164 | ||
7165 | This will download the journal contents in a | |
7166 | /var/log/messages compatible format. The same as JSON: | |
7167 | ||
7168 | # curl -H"Accept: application/json" http://localhost:19531/entries | |
7169 | ||
7170 | This service is also accessible via a web browser where a | |
7171 | single static HTML5 app is served that uses the JSON logic | |
7172 | to enable the user to do some basic browsing of the | |
7173 | journal. This will be extended later on. Here's an example | |
7174 | screenshot of this app in its current state: | |
7175 | ||
7176 | http://0pointer.de/public/journal-gatewayd | |
7177 | ||
7178 | Contributions from: Kay Sievers, Lennart Poettering, Robert | |
7179 | Milasan, Tom Gundersen | |
7180 | ||
7181 | CHANGES WITH 192: | |
7182 | ||
7183 | * The bash completion logic is now available for journalctl | |
7184 | too. | |
7185 | ||
7186 | * We do not mount the "cpuset" controller anymore together with | |
7187 | "cpu" and "cpuacct", as "cpuset" groups generally cannot be | |
7188 | started if no parameters are assigned to it. "cpuset" hence | |
7189 | broke code that assumed it could create "cpu" groups and | |
7190 | just start them. | |
7191 | ||
7192 | * journalctl -f will now subscribe to terminal size changes, | |
7193 | and line break accordingly. | |
7194 | ||
7195 | Contributions from: Dave Reisner, Kay Sievers, Lennart | |
7196 | Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín | |
7197 | ||
7198 | CHANGES WITH 191: | |
7199 | ||
7200 | * nspawn will now create a symlink /etc/localtime in the | |
7201 | container environment, copying the host's timezone | |
7202 | setting. Previously this has been done via a bind mount, but | |
7203 | since symlinks cannot be bind mounted this has now been | |
7204 | changed to create/update the appropriate symlink. | |
7205 | ||
7206 | * journalctl -n's line number argument is now optional, and | |
7207 | will default to 10 if omitted. | |
7208 | ||
7209 | * journald will now log the maximum size the journal files may | |
7210 | take up on disk. This is particularly useful if the default | |
7211 | built-in logic of determining this parameter from the file | |
7212 | system size is used. Use "systemctl status | |
7213 | systemd-journald.service" to see this information. | |
7214 | ||
7215 | * The multi-seat X wrapper tool has been stripped down. As X | |
7216 | is now capable of enumerating graphics devices via udev in a | |
7217 | seat-aware way the wrapper is not strictly necessary | |
7218 | anymore. A stripped down temporary stop-gap is still shipped | |
7219 | until the upstream display managers have been updated to | |
7220 | fully support the new X logic. Expect this wrapper to be | |
7221 | removed entirely in one of the next releases. | |
7222 | ||
7223 | * HandleSleepKey= in logind.conf has been split up into | |
7224 | HandleSuspendKey= and HandleHibernateKey=. The old setting | |
7225 | is not available anymore. X11 and the kernel are | |
7226 | distinguishing between these keys and we should too. This | |
7227 | also means the inhibition lock for these keys has been split | |
7228 | into two. | |
7229 | ||
7230 | Contributions from: Dave Airlie, Eelco Dolstra, Lennart | |
7231 | Poettering, Lukas Nykryn, Václav Pavlín | |
7232 | ||
7233 | CHANGES WITH 190: | |
7234 | ||
7235 | * Whenever a unit changes state we will now log this to the | |
7236 | journal and show along the unit's own log output in | |
7237 | "systemctl status". | |
7238 | ||
7239 | * ConditionPathIsMountPoint= can now properly detect bind | |
7240 | mount points too. (Previously, a bind mount of one file | |
7241 | system to another place in the same file system could not be | |
7242 | detected as mount, since they shared struct stat's st_dev | |
7243 | field.) | |
7244 | ||
7245 | * We will now mount the cgroup controllers cpu, cpuacct, | |
7246 | cpuset and the controllers net_cls, net_prio together by | |
7247 | default. | |
7248 | ||
7249 | * nspawn containers will now have a virtualized boot | |
7250 | ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted | |
7251 | over with a randomized ID at container initialization). This | |
7252 | has the effect of making "journalctl -b" do the right thing | |
7253 | in a container. | |
7254 | ||
7255 | * The JSON output journal serialization has been updated not | |
7256 | to generate "endless" list objects anymore, but rather one | |
7257 | JSON object per line. This is more in line how most JSON | |
7258 | parsers expect JSON objects. The new output mode | |
7259 | "json-pretty" has been added to provide similar output, but | |
7260 | neatly aligned for readability by humans. | |
7261 | ||
7262 | * We dropped all explicit sync() invocations in the shutdown | |
7263 | code. The kernel does this implicitly anyway in the kernel | |
7264 | reboot() syscall. halt(8)'s -n option is now a compatibility | |
7265 | no-op. | |
7266 | ||
7267 | * We now support virtualized reboot() in containers, as | |
7268 | supported by newer kernels. We will fall back to exit() if | |
7269 | CAP_SYS_REBOOT is not available to the container. Also, | |
7270 | nspawn makes use of this now and will actually reboot the | |
7271 | container if the containerized OS asks for that. | |
7272 | ||
7273 | * journalctl will only show local log output by default | |
7274 | now. Use --merge (-m) to show remote log output, too. | |
7275 | ||
7276 | * libsystemd-journal gained the new sd_journal_get_usage() | |
7277 | call to determine the current disk usage of all journal | |
7278 | files. This is exposed in the new "journalctl --disk-usage" | |
7279 | command. | |
7280 | ||
7281 | * journald gained a new configuration setting SplitMode= in | |
7282 | journald.conf which may be used to control how user journals | |
7283 | are split off. See journald.conf(5) for details. | |
7284 | ||
7285 | * A new condition type ConditionFileNotEmpty= has been added. | |
7286 | ||
7287 | * tmpfiles' "w" lines now support file globbing, to write | |
7288 | multiple files at once. | |
7289 | ||
7290 | * We added Python bindings for the journal submission | |
7291 | APIs. More Python APIs for a number of selected APIs will | |
7292 | likely follow. Note that we intend to add native bindings | |
7293 | only for the Python language, as we consider it common | |
7294 | enough to deserve bindings shipped within systemd. There are | |
7295 | various projects outside of systemd that provide bindings | |
7296 | for languages such as PHP or Lua. | |
7297 | ||
7298 | * Many conditions will now resolve specifiers such as %i. In | |
7299 | addition, PathChanged= and related directives of .path units | |
7300 | now support specifiers as well. | |
7301 | ||
7302 | * There's now a new RPM macro definition for the system preset | |
7303 | dir: %_presetdir. | |
7304 | ||
7305 | * journald will now warn if it ca not forward a message to the | |
7306 | syslog daemon because its socket is full. | |
7307 | ||
7308 | * timedated will no longer write or process /etc/timezone, | |
7309 | except on Debian. As we do not support late mounted /usr | |
7310 | anymore /etc/localtime always being a symlink is now safe, | |
7311 | and hence the information in /etc/timezone is not necessary | |
7312 | anymore. | |
7313 | ||
7314 | * logind will now always reserve one VT for a text getty (VT6 | |
7315 | by default). Previously if more than 6 X sessions where | |
7316 | started they took up all the VTs with auto-spawned gettys, | |
7317 | so that no text gettys were available anymore. | |
7318 | ||
7319 | * udev will now automatically inform the btrfs kernel logic | |
7320 | about btrfs RAID components showing up. This should make | |
7321 | simple hotplug based btrfs RAID assembly work. | |
7322 | ||
7323 | * PID 1 will now increase its RLIMIT_NOFILE to 64K by default | |
7324 | (but not for its children which will stay at the kernel | |
7325 | default). This should allow setups with a lot more listening | |
7326 | sockets. | |
7327 | ||
7328 | * systemd will now always pass the configured timezone to the | |
7329 | kernel at boot. timedated will do the same when the timezone | |
7330 | is changed. | |
7331 | ||
7332 | * logind's inhibition logic has been updated. By default, | |
7333 | logind will now handle the lid switch, the power and sleep | |
7334 | keys all the time, even in graphical sessions. If DEs want | |
7335 | to handle these events on their own they should take the new | |
7336 | handle-power-key, handle-sleep-key and handle-lid-switch | |
7337 | inhibitors during their runtime. A simple way to achieve | |
7338 | that is to invoke the DE wrapped in an invocation of: | |
7339 | ||
7340 | systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ... | |
7341 | ||
7342 | * Access to unit operations is now checked via SELinux taking | |
7343 | the unit file label and client process label into account. | |
7344 | ||
7345 | * systemd will now notify the administrator in the journal | |
7346 | when he over-mounts a non-empty directory. | |
7347 | ||
7348 | * There are new specifiers that are resolved in unit files, | |
7349 | for the host name (%H), the machine ID (%m) and the boot ID | |
7350 | (%b). | |
7351 | ||
7352 | Contributions from: Allin Cottrell, Auke Kok, Brandon Philips, | |
7353 | Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner, | |
7354 | Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart | |
7355 | Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas, | |
7356 | Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz, | |
7357 | Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen, | |
7358 | Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek | |
7359 | ||
7360 | CHANGES WITH 189: | |
7361 | ||
7362 | * Support for reading structured kernel messages from | |
7363 | /dev/kmsg has now been added and is enabled by default. | |
7364 | ||
7365 | * Support for reading kernel messages from /proc/kmsg has now | |
7366 | been removed. If you want kernel messages in the journal | |
7367 | make sure to run a recent kernel (>= 3.5) that supports | |
7368 | reading structured messages from /dev/kmsg (see | |
7369 | above). /proc/kmsg is now exclusive property of classic | |
7370 | syslog daemons again. | |
7371 | ||
7372 | * The libudev API gained the new | |
7373 | udev_device_new_from_device_id() call. | |
7374 | ||
7375 | * The logic for file system namespace (ReadOnlyDirectory=, | |
7376 | ReadWriteDirectoy=, PrivateTmp=) has been reworked not to | |
7377 | require pivot_root() anymore. This means fewer temporary | |
7378 | directories are created below /tmp for this feature. | |
7379 | ||
7380 | * nspawn containers will now see and receive all submounts | |
7381 | made on the host OS below the root file system of the | |
7382 | container. | |
7383 | ||
7384 | * Forward Secure Sealing is now supported for Journal files, | |
7385 | which provide cryptographical sealing of journal files so | |
7386 | that attackers cannot alter log history anymore without this | |
7387 | being detectable. Lennart will soon post a blog story about | |
7388 | this explaining it in more detail. | |
7389 | ||
7390 | * There are two new service settings RestartPreventExitStatus= | |
7391 | and SuccessExitStatus= which allow configuration of exit | |
7392 | status (exit code or signal) which will be excepted from the | |
7393 | restart logic, resp. consider successful. | |
7394 | ||
7395 | * journalctl gained the new --verify switch that can be used | |
7396 | to check the integrity of the structure of journal files and | |
7397 | (if Forward Secure Sealing is enabled) the contents of | |
7398 | journal files. | |
7399 | ||
7400 | * nspawn containers will now be run with /dev/stdin, /dev/fd/ | |
7401 | and similar symlinks pre-created. This makes running shells | |
7402 | as container init process a lot more fun. | |
7403 | ||
7404 | * The fstab support can now handle PARTUUID= and PARTLABEL= | |
7405 | entries. | |
7406 | ||
7407 | * A new ConditionHost= condition has been added to match | |
7408 | against the hostname (with globs) and machine ID. This is | |
7409 | useful for clusters where a single OS image is used to | |
7410 | provision a large number of hosts which shall run slightly | |
7411 | different sets of services. | |
7412 | ||
7413 | * Services which hit the restart limit will now be placed in a | |
7414 | failure state. | |
7415 | ||
7416 | Contributions from: Bertram Poettering, Dave Reisner, Huang | |
7417 | Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin | |
7418 | Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek | |
7419 | ||
7420 | CHANGES WITH 188: | |
7421 | ||
7422 | * When running in --user mode systemd will now become a | |
7423 | subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps | |
7424 | tree a lot more organized. | |
7425 | ||
7426 | * A new PartOf= unit dependency type has been introduced that | |
7427 | may be used to group services in a natural way. | |
7428 | ||
7429 | * "systemctl enable" may now be used to enable instances of | |
7430 | services. | |
7431 | ||
7432 | * journalctl now prints error log levels in red, and | |
7433 | warning/notice log levels in bright white. It also supports | |
7434 | filtering by log level now. | |
7435 | ||
7436 | * cgtop gained a new -n switch (similar to top), to configure | |
7437 | the maximum number of iterations to run for. It also gained | |
7438 | -b, to run in batch mode (accepting no input). | |
7439 | ||
7440 | * The suffix ".service" may now be omitted on most systemctl | |
7441 | command lines involving service unit names. | |
7442 | ||
7443 | * There's a new bus call in logind to lock all sessions, as | |
7444 | well as a loginctl verb for it "lock-sessions". | |
7445 | ||
7446 | * libsystemd-logind.so gained a new call sd_journal_perror() | |
7447 | that works similar to libc perror() but logs to the journal | |
7448 | and encodes structured information about the error number. | |
7449 | ||
7450 | * /etc/crypttab entries now understand the new keyfile-size= | |
7451 | option. | |
7452 | ||
7453 | * shutdown(8) now can send a (configurable) wall message when | |
7454 | a shutdown is cancelled. | |
7455 | ||
7456 | * The mount propagation mode for the root file system will now | |
7457 | default to "shared", which is useful to make containers work | |
7458 | nicely out-of-the-box so that they receive new mounts from | |
7459 | the host. This can be undone locally by running "mount | |
7460 | --make-rprivate /" if needed. | |
7461 | ||
7462 | * The prefdm.service file has been removed. Distributions | |
7463 | should maintain this unit downstream if they intend to keep | |
7464 | it around. However, we recommend writing normal unit files | |
7465 | for display managers instead. | |
7466 | ||
7467 | * Since systemd is a crucial part of the OS we will now | |
7468 | default to a number of compiler switches that improve | |
7469 | security (hardening) such as read-only relocations, stack | |
7470 | protection, and suchlike. | |
7471 | ||
7472 | * The TimeoutSec= setting for services is now split into | |
7473 | TimeoutStartSec= and TimeoutStopSec= to allow configuration | |
7474 | of individual time outs for the start and the stop phase of | |
7475 | the service. | |
7476 | ||
7477 | Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke | |
7478 | Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer, | |
7479 | Jim Meyering, Kay Sievers, Lennart Poettering, Mantas | |
7480 | Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter | |
7481 | Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom | |
7482 | Gundersen, Zbigniew Jędrzejewski-Szmek | |
7483 | ||
7484 | CHANGES WITH 187: | |
7485 | ||
7486 | * The journal and id128 C APIs are now fully documented as man | |
7487 | pages. | |
7488 | ||
7489 | * Extra safety checks have been added when transitioning from | |
7490 | the initial RAM disk to the main system to avoid accidental | |
7491 | data loss. | |
7492 | ||
7493 | * /etc/crypttab entries now understand the new keyfile-offset= | |
7494 | option. | |
7495 | ||
7496 | * systemctl -t can now be used to filter by unit load state. | |
7497 | ||
7498 | * The journal C API gained the new sd_journal_wait() call to | |
7499 | make writing synchronous journal clients easier. | |
7500 | ||
7501 | * journalctl gained the new -D switch to show journals from a | |
7502 | specific directory. | |
7503 | ||
7504 | * journalctl now displays a special marker between log | |
7505 | messages of two different boots. | |
7506 | ||
7507 | * The journal is now explicitly flushed to /var via a service | |
7508 | systemd-journal-flush.service, rather than implicitly simply | |
7509 | by seeing /var/log/journal to be writable. | |
7510 | ||
7511 | * journalctl (and the journal C APIs) can now match for much | |
7512 | more complex expressions, with alternatives and | |
7513 | disjunctions. | |
7514 | ||
7515 | * When transitioning from the initial RAM disk to the main | |
7516 | system we will now kill all processes in a killing spree to | |
7517 | ensure no processes stay around by accident. | |
7518 | ||
7519 | * Three new specifiers may be used in unit files: %u, %h, %s | |
7520 | resolve to the user name, user home directory resp. user | |
7521 | shell. This is useful for running systemd user instances. | |
7522 | ||
7523 | * We now automatically rotate journal files if their data | |
7524 | object hash table gets a fill level > 75%. We also size the | |
7525 | hash table based on the configured maximum file size. This | |
7526 | together should lower hash collisions drastically and thus | |
7527 | speed things up a bit. | |
7528 | ||
7529 | * journalctl gained the new "--header" switch to introspect | |
7530 | header data of journal files. | |
7531 | ||
7532 | * A new setting SystemCallFilters= has been added to services | |
7533 | which may be used to apply blacklists or whitelists to | |
7534 | system calls. This is based on SECCOMP Mode 2 of Linux 3.5. | |
7535 | ||
7536 | * nspawn gained a new --link-journal= switch (and quicker: -j) | |
7537 | to link the container journal with the host. This makes it | |
7538 | very easy to centralize log viewing on the host for all | |
7539 | guests while still keeping the journal files separated. | |
7540 | ||
7541 | * Many bugfixes and optimizations | |
7542 | ||
7543 | Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay | |
7544 | Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex | |
7545 | Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew | |
7546 | Jędrzejewski-Szmek | |
7547 | ||
7548 | CHANGES WITH 186: | |
7549 | ||
7550 | * Several tools now understand kernel command line arguments, | |
7551 | which are only read when run in an initial RAM disk. They | |
7552 | usually follow closely their normal counterparts, but are | |
7553 | prefixed with rd. | |
7554 | ||
7555 | * There's a new tool to analyze the readahead files that are | |
7556 | automatically generated at boot. Use: | |
7557 | ||
7558 | /usr/lib/systemd/systemd-readahead analyze /.readahead | |
7559 | ||
7560 | * We now provide an early debug shell on tty9 if this enabled. Use: | |
7561 | ||
7562 | systemctl enable debug-shell.service | |
7563 | ||
7564 | * All plymouth related units have been moved into the Plymouth | |
7565 | package. Please make sure to upgrade your Plymouth version | |
7566 | as well. | |
7567 | ||
7568 | * systemd-tmpfiles now supports getting passed the basename of | |
7569 | a configuration file only, in which case it will look for it | |
7570 | in all appropriate directories automatically. | |
7571 | ||
7572 | * udevadm info now takes a /dev or /sys path as argument, and | |
7573 | does the right thing. Example: | |
7574 | ||
7575 | udevadm info /dev/sda | |
7576 | udevadm info /sys/class/block/sda | |
7577 | ||
7578 | * systemctl now prints a warning if a unit is stopped but a | |
7579 | unit that might trigger it continues to run. Example: a | |
7580 | service is stopped but the socket that activates it is left | |
7581 | running. | |
7582 | ||
7583 | * "systemctl status" will now mention if the log output was | |
7584 | shortened due to rotation since a service has been started. | |
7585 | ||
7586 | * The journal API now exposes functions to determine the | |
7587 | "cutoff" times due to rotation. | |
7588 | ||
7589 | * journald now understands SIGUSR1 and SIGUSR2 for triggering | |
7590 | immediately flushing of runtime logs to /var if possible, | |
7591 | resp. for triggering immediate rotation of the journal | |
7592 | files. | |
7593 | ||
7594 | * It is now considered an error if a service is attempted to | |
7595 | be stopped that is not loaded. | |
7596 | ||
7597 | * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames. | |
7598 | ||
7599 | * systemd-analyze now supports Python 3 | |
7600 | ||
7601 | * tmpfiles now supports cleaning up directories via aging | |
7602 | where the first level dirs are always kept around but | |
7603 | directories beneath it automatically aged. This is enabled | |
7604 | by prefixing the age field with '~'. | |
7605 | ||
7606 | * Seat objects now expose CanGraphical, CanTTY properties | |
7607 | which is required to deal with very fast bootups where the | |
7608 | display manager might be running before the graphics drivers | |
7609 | completed initialization. | |
7610 | ||
7611 | * Seat objects now expose a State property. | |
7612 | ||
7613 | * We now include RPM macros for service enabling/disabling | |
7614 | based on the preset logic. We recommend RPM based | |
7615 | distributions to make use of these macros if possible. This | |
7616 | makes it simpler to reuse RPM spec files across | |
7617 | distributions. | |
7618 | ||
7619 | * We now make sure that the collected systemd unit name is | |
7620 | always valid when services log to the journal via | |
7621 | STDOUT/STDERR. | |
7622 | ||
7623 | * There's a new man page kernel-command-line(7) detailing all | |
7624 | command line options we understand. | |
7625 | ||
7626 | * The fstab generator may now be disabled at boot by passing | |
7627 | fstab=0 on the kernel command line. | |
7628 | ||
7629 | * A new kernel command line option modules-load= is now understood | |
7630 | to load a specific kernel module statically, early at boot. | |
7631 | ||
7632 | * Unit names specified on the systemctl command line are now | |
7633 | automatically escaped as needed. Also, if file system or | |
7634 | device paths are specified they are automatically turned | |
7635 | into the appropriate mount or device unit names. Example: | |
7636 | ||
7637 | systemctl status /home | |
7638 | systemctl status /dev/sda | |
7639 | ||
7640 | * The SysVConsole= configuration option has been removed from | |
7641 | system.conf parsing. | |
7642 | ||
7643 | * The SysV search path is no longer exported on the D-Bus | |
7644 | Manager object. | |
7645 | ||
7646 | * The Names= option has been removed from unit file parsing. | |
7647 | ||
7648 | * There's a new man page bootup(7) detailing the boot process. | |
7649 | ||
7650 | * Every unit and every generator we ship with systemd now | |
7651 | comes with full documentation. The self-explanatory boot is | |
7652 | complete. | |
7653 | ||
7654 | * A couple of services gained "systemd-" prefixes in their | |
7655 | name if they wrap systemd code, rather than only external | |
7656 | code. Among them fsck@.service which is now | |
7657 | systemd-fsck@.service. | |
7658 | ||
7659 | * The HaveWatchdog property has been removed from the D-Bus | |
7660 | Manager object. | |
7661 | ||
7662 | * systemd.confirm_spawn= on the kernel command line should now | |
7663 | work sensibly. | |
7664 | ||
7665 | * There's a new man page crypttab(5) which details all options | |
7666 | we actually understand. | |
7667 | ||
7668 | * systemd-nspawn gained a new --capability= switch to pass | |
7669 | additional capabilities to the container. | |
7670 | ||
7671 | * timedated will now read known NTP implementation unit names | |
7672 | from /usr/lib/systemd/ntp-units.d/*.list, | |
7673 | systemd-timedated-ntp.target has been removed. | |
7674 | ||
7675 | * journalctl gained a new switch "-b" that lists log data of | |
7676 | the current boot only. | |
7677 | ||
7678 | * The notify socket is in the abstract namespace again, in | |
7679 | order to support daemons which chroot() at start-up. | |
7680 | ||
7681 | * There is a new Storage= configuration option for journald | |
7682 | which allows configuration of where log data should go. This | |
7683 | also provides a way to disable journal logging entirely, so | |
7684 | that data collected is only forwarded to the console, the | |
7685 | kernel log buffer or another syslog implementation. | |
7686 | ||
7687 | * Many bugfixes and optimizations | |
7688 | ||
7689 | Contributions from: Auke Kok, Colin Guthrie, Dave Reisner, | |
7690 | David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering, | |
7691 | Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel, | |
7692 | Shawn Landden, Tom Gundersen | |
7693 | ||
7694 | CHANGES WITH 185: | |
7695 | ||
7696 | * "systemctl help <unit>" now shows the man page if one is | |
7697 | available. | |
7698 | ||
7699 | * Several new man pages have been added. | |
7700 | ||
7701 | * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=, | |
7702 | MaxLevelConsole= can now be specified in | |
7703 | journald.conf. These options allow reducing the amount of | |
7704 | data stored on disk or forwarded by the log level. | |
7705 | ||
7706 | * TimerSlackNSec= can now be specified in system.conf for | |
7707 | PID1. This allows system-wide power savings. | |
7708 | ||
7709 | Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen, | |
7710 | Lennart Poettering, Malte Starostik, Marc-Antoine Perennou, | |
7711 | Matthias Clasen | |
7712 | ||
7713 | CHANGES WITH 184: | |
7714 | ||
7715 | * logind is now capable of (optionally) handling power and | |
7716 | sleep keys as well as the lid switch. | |
7717 | ||
7718 | * journalctl now understands the syntax "journalctl | |
7719 | /usr/bin/avahi-daemon" to get all log output of a specific | |
7720 | daemon. | |
7721 | ||
7722 | * CapabilityBoundingSet= in system.conf now also influences | |
7723 | the capability bound set of usermode helpers of the kernel. | |
7724 | ||
7725 | Contributions from: Daniel Drake, Daniel J. Walsh, Gert | |
7726 | Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers, | |
7727 | Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul | |
7728 | Menzel, Shawn Landden, Tero Roponen, Tom Gundersen | |
7729 | ||
7730 | CHANGES WITH 183: | |
7731 | ||
7732 | * Note that we skipped 139 releases here in order to set the | |
7733 | new version to something that is greater than both udev's | |
7734 | and systemd's most recent version number. | |
7735 | ||
7736 | * udev: all udev sources are merged into the systemd source tree now. | |
7737 | All future udev development will happen in the systemd tree. It | |
7738 | is still fully supported to use the udev daemon and tools without | |
7739 | systemd running, like in initramfs or other init systems. Building | |
7740 | udev though, will require the *build* of the systemd tree, but | |
7741 | udev can be properly *run* without systemd. | |
7742 | ||
7743 | * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles | |
7744 | should be used to create dead device nodes as workarounds for broken | |
7745 | subsystems. | |
7746 | ||
7747 | * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is | |
7748 | no longer supported. udev_monitor_new_from_netlink() needs to be | |
7749 | used to subscribe to events. | |
7750 | ||
7751 | * udev: when udevd is started by systemd, processes which are left | |
7752 | behind by forking them off of udev rules, are unconditionally cleaned | |
7753 | up and killed now after the event handling has finished. Services or | |
7754 | daemons must be started as systemd services. Services can be | |
7755 | pulled-in by udev to get started, but they can no longer be directly | |
7756 | forked by udev rules. | |
7757 | ||
7758 | * udev: the daemon binary is called systemd-udevd now and installed | |
7759 | in /usr/lib/systemd/. Standalone builds or non-systemd systems need | |
7760 | to adapt to that, create symlink, or rename the binary after building | |
7761 | it. | |
7762 | ||
7763 | * libudev no longer provides these symbols: | |
7764 | udev_monitor_from_socket() | |
7765 | udev_queue_get_failed_list_entry() | |
7766 | udev_get_{dev,sys,run}_path() | |
7767 | The versions number was bumped and symbol versioning introduced. | |
7768 | ||
7769 | * systemd-loginctl and systemd-journalctl have been renamed | |
7770 | to loginctl and journalctl to match systemctl. | |
7771 | ||
7772 | * The config files: /etc/systemd/systemd-logind.conf and | |
7773 | /etc/systemd/systemd-journald.conf have been renamed to | |
7774 | logind.conf and journald.conf. Package updates should rename | |
7775 | the files to the new names on upgrade. | |
7776 | ||
7777 | * For almost all files the license is now LGPL2.1+, changed | |
7778 | from the previous GPL2.0+. Exceptions are some minor stuff | |
7779 | of udev (which will be changed to LGPL2.1 eventually, too), | |
7780 | and the MIT licensed sd-daemon.[ch] library that is suitable | |
7781 | to be used as drop-in files. | |
7782 | ||
7783 | * systemd and logind now handle system sleep states, in | |
7784 | particular suspending and hibernating. | |
7785 | ||
7786 | * logind now implements a sleep/shutdown/idle inhibiting logic | |
7787 | suitable for a variety of uses. Soonishly Lennart will blog | |
7788 | about this in more detail. | |
7789 | ||
7790 | * var-run.mount and var-lock.mount are no longer provided | |
7791 | (which previously bind mounted these directories to their new | |
7792 | places). Distributions which have not converted these | |
7793 | directories to symlinks should consider stealing these files | |
7794 | from git history and add them downstream. | |
7795 | ||
7796 | * We introduced the Documentation= field for units and added | |
7797 | this to all our shipped units. This is useful to make it | |
7798 | easier to explore the boot and the purpose of the various | |
7799 | units. | |
7800 | ||
7801 | * All smaller setup units (such as | |
7802 | systemd-vconsole-setup.service) now detect properly if they | |
7803 | are run in a container and are skipped when | |
7804 | appropriate. This guarantees an entirely noise-free boot in | |
7805 | Linux container environments such as systemd-nspawn. | |
7806 | ||
7807 | * A framework for implementing offline system updates is now | |
7808 | integrated, for details see: | |
7809 | https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates | |
7810 | ||
7811 | * A new service type Type=idle is available now which helps us | |
7812 | avoiding ugly interleaving of getty output and boot status | |
7813 | messages. | |
7814 | ||
7815 | * There's now a system-wide CapabilityBoundingSet= option to | |
7816 | globally reduce the set of capabilities for the | |
7817 | system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO, | |
7818 | CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or | |
7819 | even CAP_NET_ADMIN system-wide for secure systems. | |
7820 | ||
7821 | * There are now system-wide DefaultLimitXXX= options to | |
7822 | globally change the defaults of the various resource limits | |
7823 | for all units started by PID 1. | |
7824 | ||
7825 | * Harald Hoyer's systemd test suite has been integrated into | |
7826 | systemd which allows easy testing of systemd builds in qemu | |
7827 | and nspawn. (This is really awesome! Ask us for details!) | |
7828 | ||
7829 | * The fstab parser is now implemented as generator, not inside | |
7830 | of PID 1 anymore. | |
7831 | ||
7832 | * systemctl will now warn you if .mount units generated from | |
7833 | /etc/fstab are out of date due to changes in fstab that | |
7834 | have not been read by systemd yet. | |
7835 | ||
7836 | * systemd is now suitable for usage in initrds. Dracut has | |
7837 | already been updated to make use of this. With this in place | |
7838 | initrds get a slight bit faster but primarily are much | |
7839 | easier to introspect and debug since "systemctl status" in | |
7840 | the host system can be used to introspect initrd services, | |
7841 | and the journal from the initrd is kept around too. | |
7842 | ||
7843 | * systemd-delta has been added, a tool to explore differences | |
7844 | between user/admin configuration and vendor defaults. | |
7845 | ||
7846 | * PrivateTmp= now affects both /tmp and /var/tmp. | |
7847 | ||
7848 | * Boot time status messages are now much prettier and feature | |
7849 | proper english language. Booting up systemd has never been | |
7850 | so sexy. | |
7851 | ||
7852 | * Read-ahead pack files now include the inode number of all | |
7853 | files to pre-cache. When the inode changes the pre-caching | |
7854 | is not attempted. This should be nicer to deal with updated | |
7855 | packages which might result in changes of read-ahead | |
7856 | patterns. | |
7857 | ||
7858 | * We now temporaritly lower the kernel's read_ahead_kb variable | |
7859 | when collecting read-ahead data to ensure the kernel's | |
7860 | built-in read-ahead does not add noise to our measurements | |
7861 | of necessary blocks to pre-cache. | |
7862 | ||
7863 | * There's now RequiresMountsFor= to add automatic dependencies | |
7864 | for all mounts necessary for a specific file system path. | |
7865 | ||
7866 | * MountAuto= and SwapAuto= have been removed from | |
7867 | system.conf. Mounting file systems at boot has to take place | |
7868 | in systemd now. | |
7869 | ||
7870 | * nspawn now learned a new switch --uuid= to set the machine | |
7871 | ID on the command line. | |
7872 | ||
7873 | * nspawn now learned the -b switch to automatically search | |
7874 | for an init system. | |
7875 | ||
7876 | * vt102 is now the default TERM for serial TTYs, upgraded from | |
7877 | vt100. | |
7878 | ||
7879 | * systemd-logind now works on VT-less systems. | |
7880 | ||
7881 | * The build tree has been reorganized. The individual | |
7882 | components now have directories of their own. | |
7883 | ||
7884 | * A new condition type ConditionPathIsReadWrite= is now available. | |
7885 | ||
7886 | * nspawn learned the new -C switch to create cgroups for the | |
7887 | container in other hierarchies. | |
7888 | ||
7889 | * We now have support for hardware watchdogs, configurable in | |
7890 | system.conf. | |
7891 | ||
7892 | * The scheduled shutdown logic now has a public API. | |
7893 | ||
7894 | * We now mount /tmp as tmpfs by default, but this can be | |
7895 | masked and /etc/fstab can override it. | |
7896 | ||
7897 | * Since udisks does not make use of /media anymore we are not | |
7898 | mounting a tmpfs on it anymore. | |
7899 | ||
7900 | * journalctl gained a new --local switch to only interleave | |
7901 | locally generated journal files. | |
7902 | ||
7903 | * We can now load the IMA policy at boot automatically. | |
7904 | ||
7905 | * The GTK tools have been split off into a systemd-ui. | |
7906 | ||
7907 | Contributions from: Andreas Schwab, Auke Kok, Ayan George, | |
7908 | Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan | |
7909 | Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal, | |
7910 | Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers, | |
7911 | Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure, | |
7912 | Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim | |
7913 | A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal | |
7914 | Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn | |
7915 | Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom | |
7916 | Gundersen | |
7917 | ||
7918 | CHANGES WITH 44: | |
7919 | ||
7920 | * This is mostly a bugfix release | |
7921 | ||
7922 | * Support optional initialization of the machine ID from the | |
7923 | KVM or container configured UUID. | |
7924 | ||
7925 | * Support immediate reboots with "systemctl reboot -ff" | |
7926 | ||
7927 | * Show /etc/os-release data in systemd-analyze output | |
7928 | ||
7929 | * Many bugfixes for the journal, including endianness fixes and | |
7930 | ensuring that disk space enforcement works | |
7931 | ||
7932 | * sd-login.h is C++ compatible again | |
7933 | ||
7934 | * Extend the /etc/os-release format on request of the Debian | |
7935 | folks | |
7936 | ||
7937 | * We now refuse non-UTF8 strings used in various configuration | |
7938 | and unit files. This is done to ensure we do not pass invalid | |
7939 | data over D-Bus or expose it elsewhere. | |
7940 | ||
7941 | * Register Mimo USB Screens as suitable for automatic seat | |
7942 | configuration | |
7943 | ||
7944 | * Read SELinux client context from journal clients in a race | |
7945 | free fashion | |
7946 | ||
7947 | * Reorder configuration file lookup order. /etc now always | |
7948 | overrides /run in order to allow the administrator to always | |
7949 | and unconditionally override vendor-supplied or | |
7950 | automatically generated data. | |
7951 | ||
7952 | * The various user visible bits of the journal now have man | |
7953 | pages. We still lack man pages for the journal API calls | |
7954 | however. | |
7955 | ||
7956 | * We now ship all man pages in HTML format again in the | |
7957 | tarball. | |
7958 | ||
7959 | Contributions from: Dave Reisner, Dirk Eibach, Frederic | |
7960 | Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti | |
7961 | Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry | |
7962 | Reding | |
7963 | ||
7964 | CHANGES WITH 43: | |
7965 | ||
7966 | * This is mostly a bugfix release | |
7967 | ||
7968 | * systems lacking /etc/os-release are no longer supported. | |
7969 | ||
7970 | * Various functionality updates to libsystemd-login.so | |
7971 | ||
7972 | * Track class of PAM logins to distinguish greeters from | |
7973 | normal user logins. | |
7974 | ||
7975 | Contributions from: Kay Sievers, Lennart Poettering, Michael | |
7976 | Biebl | |
7977 | ||
7978 | CHANGES WITH 42: | |
7979 | ||
7980 | * This is an important bugfix release for v41. | |
7981 | ||
7982 | * Building man pages is now optional which should be useful | |
7983 | for those building systemd from git but unwilling to install | |
7984 | xsltproc. | |
7985 | ||
7986 | * Watchdog support for supervising services is now usable. In | |
7987 | a future release support for hardware watchdogs | |
7988 | (i.e. /dev/watchdog) will be added building on this. | |
7989 | ||
7990 | * Service start rate limiting is now configurable and can be | |
7991 | turned off per service. When a start rate limit is hit a | |
7992 | reboot can automatically be triggered. | |
7993 | ||
7994 | * New CanReboot(), CanPowerOff() bus calls in systemd-logind. | |
7995 | ||
7996 | Contributions from: Benjamin Franzke, Bill Nottingham, | |
7997 | Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal | |
7998 | Schmidt, Michał Górny, Piotr Drąg | |
7999 | ||
8000 | CHANGES WITH 41: | |
8001 | ||
8002 | * The systemd binary is installed /usr/lib/systemd/systemd now; | |
8003 | An existing /sbin/init symlink needs to be adapted with the | |
8004 | package update. | |
8005 | ||
8006 | * The code that loads kernel modules has been ported to invoke | |
8007 | libkmod directly, instead of modprobe. This means we do not | |
8008 | support systems with module-init-tools anymore. | |
8009 | ||
8010 | * Watchdog support is now already useful, but still not | |
8011 | complete. | |
8012 | ||
8013 | * A new kernel command line option systemd.setenv= is | |
8014 | understood to set system wide environment variables | |
8015 | dynamically at boot. | |
8016 | ||
8017 | * We now limit the set of capabilities of systemd-journald. | |
8018 | ||
8019 | * We now set SIGPIPE to ignore by default, since it only is | |
8020 | useful in shell pipelines, and has little use in general | |
8021 | code. This can be disabled with IgnoreSIPIPE=no in unit | |
8022 | files. | |
8023 | ||
8024 | Contributions from: Benjamin Franzke, Kay Sievers, Lennart | |
8025 | Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen, | |
8026 | William Douglas | |
8027 | ||
8028 | CHANGES WITH 40: | |
8029 | ||
8030 | * This is mostly a bugfix release | |
8031 | ||
8032 | * We now expose the reason why a service failed in the | |
8033 | "Result" D-Bus property. | |
8034 | ||
8035 | * Rudimentary service watchdog support (will be completed over | |
8036 | the next few releases.) | |
8037 | ||
8038 | * When systemd forks off in order execute some service we will | |
8039 | now immediately changes its argv[0] to reflect which process | |
8040 | it will execute. This is useful to minimize the time window | |
8041 | with a generic argv[0], which makes bootcharts more useful | |
8042 | ||
8043 | Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay | |
8044 | Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt, | |
8045 | Mike Kazantsev, Ray Strode | |
8046 | ||
8047 | CHANGES WITH 39: | |
8048 | ||
8049 | * This is mostly a test release, but incorporates many | |
8050 | bugfixes. | |
8051 | ||
8052 | * New systemd-cgtop tool to show control groups by their | |
8053 | resource usage. | |
8054 | ||
8055 | * Linking against libacl for ACLs is optional again. If | |
8056 | disabled, support tracking device access for active logins | |
8057 | goes becomes unavailable, and so does access to the user | |
8058 | journals by the respective users. | |
8059 | ||
8060 | * If a group "adm" exists, journal files are automatically | |
8061 | owned by them, thus allow members of this group full access | |
8062 | to the system journal as well as all user journals. | |
8063 | ||
8064 | * The journal now stores the SELinux context of the logging | |
8065 | client for all entries. | |
8066 | ||
8067 | * Add C++ inclusion guards to all public headers | |
8068 | ||
8069 | * New output mode "cat" in the journal to print only text | |
8070 | messages, without any meta data like date or time. | |
8071 | ||
8072 | * Include tiny X server wrapper as a temporary stop-gap to | |
8073 | teach XOrg udev display enumeration. This is used by display | |
8074 | managers such as gdm, and will go away as soon as XOrg | |
8075 | learned native udev hotplugging for display devices. | |
8076 | ||
8077 | * Add new systemd-cat tool for executing arbitrary programs | |
8078 | with STDERR/STDOUT connected to the journal. Can also act as | |
8079 | BSD logger replacement, and does so by default. | |
8080 | ||
8081 | * Optionally store all locally generated coredumps in the | |
8082 | journal along with meta data. | |
8083 | ||
8084 | * systemd-tmpfiles learnt four new commands: n, L, c, b, for | |
8085 | writing short strings to files (for usage for /sys), and for | |
8086 | creating symlinks, character and block device nodes. | |
8087 | ||
8088 | * New unit file option ControlGroupPersistent= to make cgroups | |
8089 | persistent, following the mechanisms outlined in | |
8090 | https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups | |
8091 | ||
8092 | * Support multiple local RTCs in a sane way | |
8093 | ||
8094 | * No longer monopolize IO when replaying readahead data on | |
8095 | rotating disks, since we might starve non-file-system IO to | |
8096 | death, since fanotify() will not see accesses done by blkid, | |
8097 | or fsck. | |
8098 | ||
8099 | * Do not show kernel threads in systemd-cgls anymore, unless | |
8100 | requested with new -k switch. | |
8101 | ||
8102 | Contributions from: Dan Horák, Kay Sievers, Lennart | |
8103 | Poettering, Michal Schmidt | |
8104 | ||
8105 | CHANGES WITH 38: | |
8106 | ||
8107 | * This is mostly a test release, but incorporates many | |
8108 | bugfixes. | |
8109 | ||
8110 | * The git repository moved to: | |
8111 | git://anongit.freedesktop.org/systemd/systemd | |
8112 | ssh://git.freedesktop.org/git/systemd/systemd | |
8113 | ||
8114 | * First release with the journal | |
8115 | http://0pointer.de/blog/projects/the-journal.html | |
8116 | ||
8117 | * The journal replaces both systemd-kmsg-syslogd and | |
8118 | systemd-stdout-bridge. | |
8119 | ||
8120 | * New sd_pid_get_unit() API call in libsystemd-logind | |
8121 | ||
8122 | * Many systemadm clean-ups | |
8123 | ||
8124 | * Introduce remote-fs-pre.target which is ordered before all | |
8125 | remote mounts and may be used to start services before all | |
8126 | remote mounts. | |
8127 | ||
8128 | * Added Mageia support | |
8129 | ||
8130 | * Add bash completion for systemd-loginctl | |
8131 | ||
8132 | * Actively monitor PID file creation for daemons which exit in | |
8133 | the parent process before having finished writing the PID | |
8134 | file in the daemon process. Daemons which do this need to be | |
8135 | fixed (i.e. PID file creation must have finished before the | |
8136 | parent exits), but we now react a bit more gracefully to them. | |
8137 | ||
8138 | * Add colourful boot output, mimicking the well-known output | |
8139 | of existing distributions. | |
8140 | ||
8141 | * New option PassCredentials= for socket units, for | |
8142 | compatibility with a recent kernel ABI breakage. | |
8143 | ||
8144 | * /etc/rc.local is now hooked in via a generator binary, and | |
8145 | thus will no longer act as synchronization point during | |
8146 | boot. | |
8147 | ||
8148 | * systemctl list-unit-files now supports --root=. | |
8149 | ||
8150 | * systemd-tmpfiles now understands two new commands: z, Z for | |
8151 | relabelling files according to the SELinux database. This is | |
8152 | useful to apply SELinux labels to specific files in /sys, | |
8153 | among other things. | |
8154 | ||
8155 | * Output of SysV services is now forwarded to both the console | |
8156 | and the journal by default, not only just the console. | |
8157 | ||
8158 | * New man pages for all APIs from libsystemd-login. | |
8159 | ||
8160 | * The build tree got reorganized and the build system is a | |
8161 | lot more modular allowing embedded setups to specifically | |
8162 | select the components of systemd they are interested in. | |
8163 | ||
8164 | * Support for Linux systems lacking the kernel VT subsystem is | |
8165 | restored. | |
8166 | ||
8167 | * configure's --with-rootdir= got renamed to | |
8168 | --with-rootprefix= to follow the naming used by udev and | |
8169 | kmod | |
8170 | ||
8171 | * Unless specified otherwise we will now install to /usr instead | |
8172 | of /usr/local by default. | |
8173 | ||
8174 | * Processes with '@' in argv[0][0] are now excluded from the | |
8175 | final shut-down killing spree, following the logic explained | |
8176 | in: | |
8177 | https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons | |
8178 | ||
8179 | * All processes remaining in a service cgroup when we enter | |
8180 | the START or START_PRE states are now killed with | |
8181 | SIGKILL. That means it is no longer possible to spawn | |
8182 | background processes from ExecStart= lines (which was never | |
8183 | supported anyway, and bad style). | |
8184 | ||
8185 | * New PropagateReloadTo=/PropagateReloadFrom= options to bind | |
8186 | reloading of units together. | |
8187 | ||
8188 | Contributions from: Bill Nottingham, Daniel J. Walsh, Dave | |
8189 | Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay | |
8190 | Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt, | |
8191 | Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef | |
8192 | Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek |