1 systemd System and Service Manager
5 * If compiled with --enable-xkbcommon, systemd-localed will
6 verify x11 keymap settings by compiling the given keymap. It
7 will spew out warnings if the compilation fails. This
8 requires libxkbcommon to be installed.
10 * All systemd programs that read standalone configuration files
11 in /etc now also support a corresponding series of
12 .conf.d configuration directories in /etc/, /run/,
13 /usr/local/lib/, /usr/lib/, and (if configured with
14 --enable-split-usr) /lib/. In particular, the following
15 configuration files now have corresponding configuration
16 directories: system.conf user.conf, logind.conf,
17 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
18 resolved.conf, timesyncd.conf, journal-remote.conf, and
19 journal-upload.conf. Note that distributions should use the
20 configuration directories in /usr/lib/; the directories in
21 /etc/ are reserved for the system administrator.
25 * journalctl gained the new options -t/--identifier= to match
26 on the syslog identifier (aka "tag"), as well as --utc to
27 show log timestamps in the UTC timezone. journalctl now also
28 accepts -n/--lines=all to disable line capping in a pager.
30 * journalctl gained a new switch, --flush, that synchronously
31 flushes logs from /run/log/journal to /var/log/journal if
32 persistent storage is enabled. systemd-journal-flush.service
33 now waits until the operation is complete.
35 * Services can notify the manager before they start a reload
36 (by sending RELOADING=1) or shutdown (by sending
37 STOPPING=1). This allows the manager to track and show the
38 internal state of daemons and closes a race condition when
39 the process is still running but has closed its D-Bus
42 * Services with Type=oneshot do not have to have any ExecStart
45 * User units are now loaded also from
46 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
47 /run/systemd/user directory that was already previously
48 supported, but is under the control of the user.
50 * Job timeouts (i.e. time-outs on the time a job that is
51 queued stays in the run queue) can now optionally result in
52 immediate reboot or power-off actions (JobTimeoutAction= and
53 JobTimeoutRebootArgument=). This is useful on ".target"
54 units, to limit the maximum time a target remains
55 undispatched in the run queue, and to trigger an emergency
56 operation in such a case. This is now used by default to
57 turn off the system if boot-up (as defined by everything in
58 basic.target) hangs and does not complete for at least
59 15min. Also, if power-off or reboot hang for at least 30min
60 an immediate power-off/reboot operation is triggered. This
61 functionality is particularly useful to increase reliability
62 on embedded devices, but also on laptops which might
63 accidentally get powered on when carried in a backpack and
64 whose boot stays stuck in a hard disk encryption passphrase
67 * systemd-logind can be configured to also handle lid switch
68 events even when the machine is docked or multiple displays
69 are attached (HandleLidSwitchDocked= option).
71 * A helper binary and a service have been added which can be
72 used to resume from hibernation in the initramfs. A
73 generator will parse the resume= option on the kernel
74 command line to trigger resume.
76 * A user console daemon systemd-consoled has been
77 added. Currently, it is a preview, and will so far open a
78 single terminal on each session of the user marked as
79 Desktop=systemd-console.
81 * Route metrics can be specified for DHCP routes added by
84 * The SELinux context of socket-activated services can be set
85 from the information provided by the networking stack
86 (SELinuxContextFromNet= option).
88 * Userspace firmware loading support has been removed and
89 the minimum supported kernel version is thus bumped to 3.7.
91 * Timeout for udev workers has been increased from 1 to 3
92 minutes, but a warning will be printed after 1 minute to
93 help diagnose kernel modules that take a long time to load.
95 * Udev rules can now remove tags on devices with TAG-="foobar".
97 * systemd's readahead implementation has been removed. In many
98 circumstances it didn't give expected benefits even for
99 rotational disk drives and was becoming less relevant in the
100 age of SSDs. As none of the developers has been using
101 rotating media anymore, and nobody stepped up to actively
102 maintain this component of systemd it has now been removed.
104 * Swap units can use Options= to specify discard options.
105 Discard options specified for swaps in /etc/fstab are now
108 * Docker containers are now detected as a separate type of
111 * The Password Agent protocol gained support for queries where
112 the user input is shown, useful e.g. for user names.
113 systemd-ask-password gained a new --echo option to turn that
116 * The default sysctl.d/ snippets will now set:
118 net.core.default_qdisc = fq_codel
120 This selects Fair Queuing Controlled Delay as the default
121 queuing discipline for network interfaces. fq_codel helps
122 fight the network bufferbloat problem. It is believed to be
123 a good default with no tuning required for most workloads.
124 Downstream distributions may override this choice. On 10Gbit
125 servers that do not do forwarding, "fq" may perform better.
126 Systems without a good clocksource should use "pfifo_fast".
128 * If kdbus is enabled during build a new option BusPolicy= is
129 available for service units, that allows locking all service
130 processes into a stricter bus policy, in order to limit
131 access to various bus services, or even hide most of them
132 from the service's view entirely.
134 * networkctl will now show the .network and .link file
135 networkd has applied to a specific interface.
137 * sd-login gained a new API call sd_session_get_desktop() to
138 query which desktop environment has been selected for a
141 * UNIX utmp support is now compile-time optional to support
144 * systemctl gained two new commands "add-wants" and
145 "add-requires" for pulling in units from specific targets
148 * If the word "rescue" is specified on the kernel command line
149 the system will now boot into rescue mode (aka
150 rescue.target), which was previously available only by
151 specifying "1" or "systemd.unit=rescue.target" on the kernel
152 command line. This new kernel command line option nicely
153 mirrors the already existing "emergency" kernel command line
156 * New kernel command line options mount.usr=, mount.usrflags=,
157 mount.usrfstype= have been added that match root=, rootflags=,
158 rootfstype= but allow mounting a specific file system to
161 * The $NOTIFY_SOCKET is now also passed to control processes of
162 services, not only the main process.
164 * This version reenables support for fsck's -l switch. This
165 means at least version v2.25 of util-linux is required for
166 operation, otherwise dead-locks on device nodes may
167 occur. Again: you need to update util-linux to at least
168 v2.25 when updating systemd to v217.
170 * The "multi-seat-x" tool has been removed from systemd, as
171 its functionality has been integrated into X servers 1.16,
172 and the tool is hence redundant. It is recommended to update
173 display managers invoking this tool to simply invoke X
174 directly from now on, again.
176 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
177 message flag has been added for all of systemd's PolicyKit
178 authenticated method calls has been added. In particular
179 this now allows optional interactive authorization via
180 PolicyKit for many of PID1's privileged operations such as
181 unit file enabling and disabling.
183 * "udevadm hwdb --update" learnt a new switch "--usr" for
184 placing the rebuilt hardware database in /usr instead of
185 /etc. When used only hardware database entries stored in
186 /usr will be used, and any user database entries in /etc are
187 ignored. This functionality is useful for vendors to ship a
188 pre-built database on systems where local configuration is
189 unnecessary or unlikely.
191 * Calendar time specifications in .timer units now also
192 understand the strings "semi-annually", "quarterly" and
193 "minutely" as shortcuts (in addition to the preexisting
194 "anually", "hourly", ...).
196 * systemd-tmpfiles will now correctly create files in /dev
197 at boot which are marked for creation only at boot. It is
198 recommended to always create static device nodes with 'c!'
199 and 'b!', so that they are created only at boot and not
200 overwritten at runtime.
202 * When the watchdog logic is used for a service (WatchdogSec=)
203 and the watchdog timeout is hit the service will now be
204 terminated with SIGABRT (instead of just SIGTERM), in order
205 to make sure a proper coredump and backtrace is
206 generated. This ensures that hanging services will result in
207 similar coredump/backtrace behaviour as services that hit a
210 Contributions from: Andreas Henriksson, Andrei Borzenkov,
211 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
212 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
213 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
214 Herrmann, David Sommerseth, David Strauss, Emil Renner
215 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
216 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
217 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
218 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
219 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
220 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
221 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
222 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
223 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
224 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
225 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
226 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
227 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
228 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
229 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
230 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
233 -- Berlin, 2014-10-28
237 * timedated no longer reads NTP implementation unit names from
238 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
239 implementations should add a
241 Conflicts=systemd-timesyncd.service
243 to their unit files to take over and replace systemd's NTP
244 default functionality.
246 * systemd-sysusers gained a new line type "r" for configuring
247 which UID/GID ranges to allocate system users/groups
248 from. Lines of type "u" may now add an additional column
249 that specifies the home directory for the system user to be
250 created. Also, systemd-sysusers may now optionally read user
251 information from STDIN instead of a file. This is useful for
252 invoking it from RPM preinst scriptlets that need to create
253 users before the first RPM file is installed since these
254 files might need to be owned by them. A new
255 %sysusers_create_inline RPM macro has been introduced to do
256 just that. systemd-sysusers now updates the shadow files as
257 well as the user/group databases, which should enhance
258 compatibility with certain tools like grpck.
260 * A number of bus APIs of PID 1 now optionally consult
261 PolicyKit to permit access for otherwise unprivileged
262 clients under certain conditions. Note that this currently
263 doesn't support interactive authentication yet, but this is
264 expected to be added eventually, too.
266 * /etc/machine-info now has new fields for configuring the
267 deployment environment of the machine, as well as the
268 location of the machine. hostnamectl has been updated with
269 new command to update these fields.
271 * systemd-timesyncd has been updated to automatically acquire
272 NTP server information from systemd-networkd, which might
273 have been discovered via DHCP.
275 * systemd-resolved now includes a caching DNS stub resolver
276 and a complete LLMNR name resolution implementation. A new
277 NSS module "nss-resolve" has been added which can be used
278 instead of glibc's own "nss-dns" to resolve hostnames via
279 systemd-resolved. Hostnames, addresses and arbitrary RRs may
280 be resolved via systemd-resolved D-Bus APIs. In contrast to
281 the glibc internal resolver systemd-resolved is aware of
282 multi-homed system, and keeps DNS server and caches separate
283 and per-interface. Queries are sent simultaneously on all
284 interfaces that have DNS servers configured, in order to
285 properly handle VPNs and local LANs which might resolve
286 separate sets of domain names. systemd-resolved may acquire
287 DNS server information from systemd-networkd automatically,
288 which in turn might have discovered them via DHCP. A tool
289 "systemd-resolve-host" has been added that may be used to
290 query the DNS logic in resolved. systemd-resolved implements
291 IDNA and automatically uses IDNA or UTF-8 encoding depending
292 on whether classic DNS or LLMNR is used as transport. In the
293 next releases we intend to add a DNSSEC and mDNS/DNS-SD
294 implementation to systemd-resolved.
296 * A new NSS module nss-mymachines has been added, that
297 automatically resolves the names of all local registered
298 containers to their respective IP addresses.
300 * A new client tool "networkctl" for systemd-networkd has been
301 added. It currently is entirely passive and will query
302 networking configuration from udev, rtnetlink and networkd,
303 and present it to the user in a very friendly
304 way. Eventually, we hope to extend it to become a full
305 control utility for networkd.
307 * .socket units gained a new DeferAcceptSec= setting that
308 controls the kernels' TCP_DEFER_ACCEPT sockopt for
309 TCP. Similar, support for controlling TCP keep-alive
310 settings has been added (KeepAliveTimeSec=,
311 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
312 turning off Nagle's algorithm on TCP has been added
315 * logind learned a new session type "web", for use in projects
316 like Cockpit which register web clients as PAM sessions.
318 * timer units with at least one OnCalendar= setting will now
319 be started only after timer-sync.target has been
320 reached. This way they will not elapse before the system
321 clock has been corrected by a local NTP client or
322 similar. This is particular useful on RTC-less embedded
323 machines, that come up with an invalid system clock.
325 * systemd-nspawn's --network-veth= switch should now result in
326 stable MAC addresses for both the outer and the inner side
329 * systemd-nspawn gained a new --volatile= switch for running
330 container instances with /etc or /var unpopulated.
332 * The kdbus client code has been updated to use the new Linux
333 3.17 memfd subsystem instead of the old kdbus-specific one.
335 * systemd-networkd's DHCP client and server now support
336 FORCERENEW. There are also new configuration options to
337 configure the vendor client identifier and broadcast mode
340 * systemd will no longer inform the kernel about the current
341 timezone, as this is necessarily incorrect and racy as the
342 kernel has no understanding of DST and similar
343 concepts. This hence means FAT timestamps will be always
344 considered UTC, similar to what Android is already
345 doing. Also, when the RTC is configured to the local time
346 (rather than UTC) systemd will never synchronize back to it,
347 as this might confuse Windows at a later boot.
349 * systemd-analyze gained a new command "verify" for offline
350 validation of unit files.
352 * systemd-networkd gained support for a couple of additional
353 settings for bonding networking setups. Also, the metric for
354 statically configured routes may now be configured. For
355 network interfaces where this is appropriate the peer IP
356 address may now be configured.
358 * systemd-networkd's DHCP client will no longer request
359 broadcasting by default, as this tripped up some networks.
360 For hardware where broadcast is required the feature should
361 be switched back on using RequestBroadcast=yes.
363 * systemd-networkd will now set up IPv4LL addresses (when
364 enabled) even if DHCP is configured successfully.
366 * udev will now default to respect network device names given
367 by the kernel when the kernel indicates that these are
368 predictable. This behavior can be tweaked by changing
369 NamePolicy= in the relevant .link file.
371 * A new library systemd-terminal has been added that
372 implements full TTY stream parsing and rendering. This
373 library is supposed to be used later on for implementing a
374 full userspace VT subsystem, replacing the current kernel
377 * A new tool systemd-journal-upload has been added to push
378 journal data to a remote system running
379 systemd-journal-remote.
381 * journald will no longer forward all local data to another
382 running syslog daemon. This change has been made because
383 rsyslog (which appears to be the most commonly used syslog
384 implementation these days) no longer makes use of this, and
385 instead pulls the data out of the journal on its own. Since
386 forwarding the messages to a non-existent syslog server is
387 more expensive than we assumed we have now turned this
388 off. If you run a syslog server that is not a recent rsyslog
389 version, you have to turn this option on again
390 (ForwardToSyslog= in journald.conf).
392 * journald now optionally supports the LZ4 compressor for
393 larger journal fields. This compressor should perform much
394 better than XZ which was the previous default.
396 * machinectl now shows the IP addresses of local containers,
397 if it knows them, plus the interface name of the container.
399 * A new tool "systemd-escape" has been added that makes it
400 easy to escape strings to build unit names and similar.
402 * sd_notify() messages may now include a new ERRNO= field
403 which is parsed and collected by systemd and shown among the
404 "systemctl status" output for a service.
406 * A new component "systemd-firstboot" has been added that
407 queries the most basic systemd information (timezone,
408 hostname, root password) interactively on first
409 boot. Alternatively it may also be used to provision these
410 things offline on OS images installed into directories.
412 * The default sysctl.d/ snippets will now set
414 net.ipv4.conf.default.promote_secondaries=1
416 This has the benefit of no flushing secondary IP addresses
417 when primary addresses are removed.
419 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
420 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
421 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
422 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
423 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
424 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
425 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
426 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
427 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
428 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
429 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
430 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
431 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
432 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
433 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
435 -- Berlin, 2014-08-19
439 * A new tool systemd-sysusers has been added. This tool
440 creates system users and groups in /etc/passwd and
441 /etc/group, based on static declarative system user/group
442 definitions in /usr/lib/sysusers.d/. This is useful to
443 enable factory resets and volatile systems that boot up with
444 an empty /etc directory, and thus need system users and
445 groups created during early boot. systemd now also ships
446 with two default sysusers.d/ files for the most basic
447 users and groups systemd and the core operating system
450 * A new tmpfiles snippet has been added that rebuilds the
451 essential files in /etc on boot, should they be missing.
453 * A directive for ensuring automatic clean-up of
454 /var/cache/man/ has been removed from the default
455 configuration. This line should now be shipped by the man
456 implementation. The necessary change has been made to the
457 man-db implementation. Note that you need to update your man
458 implementation to one that ships this line, otherwise no
459 automatic clean-up of /var/cache/man will take place.
461 * A new condition ConditionNeedsUpdate= has been added that
462 may conditionalize services to only run when /etc or /var
463 are "older" than the vendor operating system resources in
464 /usr. This is useful for reconstructing or updating /etc
465 after an offline update of /usr or a factory reset, on the
466 next reboot. Services that want to run once after such an
467 update or reset should use this condition and order
468 themselves before the new systemd-update-done.service, which
469 will mark the two directories as fully updated. A number of
470 service files have been added making use of this, to rebuild
471 the udev hardware database, the journald message catalog and
472 dynamic loader cache (ldconfig). The systemd-sysusers tool
473 described above also makes use of this now. With this in
474 place it is now possible to start up a minimal operating
475 system with /etc empty cleanly. For more information on the
476 concepts involved see this recent blog story:
478 http://0pointer.de/blog/projects/stateless.html
480 * A new system group "input" has been introduced, and all
481 input device nodes get this group assigned. This is useful
482 for system-level software to get access to input devices. It
483 complements what is already done for "audio" and "video".
485 * systemd-networkd learnt minimal DHCPv4 server support in
486 addition to the existing DHCPv4 client support. It also
487 learnt DHCPv6 client and IPv6 Router Solicitation client
488 support. The DHCPv4 client gained support for static routes
489 passed in from the server. Note that the [DHCPv4] section
490 known in older systemd-networkd versions has been renamed to
491 [DHCP] and is now also used by the DHCPv6 client. Existing
492 .network files using settings of this section should be
493 updated, though compatibility is maintained. Optionally, the
494 client hostname may now be sent to the DHCP server.
496 * networkd gained support for vxlan virtual networks as well
497 as tun/tap and dummy devices.
499 * networkd gained support for automatic allocation of address
500 ranges for interfaces from a system-wide pool of
501 addresses. This is useful for dynamically managing a large
502 number of interfaces with a single network configuration
503 file. In particular this is useful to easily assign
504 appropriate IP addresses to the veth links of a large number
507 * RPM macros for processing sysusers, sysctl and binfmt
508 drop-in snippets at package installation time have been
511 * The /etc/os-release file should now be placed in
512 /usr/lib/os-release. The old location is automatically
513 created as symlink. /usr/lib is the more appropriate
514 location of this file, since it shall actually describe the
515 vendor operating system shipped in /usr, and not the
516 configuration stored in /etc.
518 * .mount units gained a new boolean SloppyOptions= setting
519 that maps to mount(8)'s -s option which enables permissive
520 parsing of unknown mount options.
522 * tmpfiles learnt a new "L+" directive which creates a symlink
523 but (unlike "L") deletes a pre-existing file first, should
524 it already exist and not already be the correct
525 symlink. Similar, "b+", "c+" and "p+" directives have been
526 added as well, which create block and character devices, as
527 well as fifos in the filesystem, possibly removing any
528 pre-existing files of different types.
530 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
531 'argument' field (which so far specified the source to
532 symlink/copy the files from) is now optional. If omitted the
533 same file os copied from /usr/share/factory/ suffixed by the
534 full destination path. This is useful for populating /etc
535 with essential files, by copying them from vendor defaults
536 shipped in /usr/share/factory/etc.
538 * A new command "systemctl preset-all" has been added that
539 applies the service preset settings to all installed unit
540 files. A new switch --preset-mode= has been added that
541 controls whether only enable or only disable operations
544 * A new command "systemctl is-system-running" has been added
545 that allows checking the overall state of the system, for
546 example whether it is fully up and running.
548 * When the system boots up with an empty /etc, the equivalent
549 to "systemctl preset-all" is executed during early boot, to
550 make sure all default services are enabled after a factory
553 * systemd now contains a minimal preset file that enables the
554 most basic services systemd ships by default.
556 * Unit files' [Install] section gained a new DefaultInstance=
557 field for defining the default instance to create if a
558 template unit is enabled with no instance specified.
560 * A new passive target cryptsetup-pre.target has been added
561 that may be used by services that need to make they run and
562 finish before the first LUKS cryptographic device is set up.
564 * The /dev/loop-control and /dev/btrfs-control device nodes
565 are now owned by the "disk" group by default, opening up
566 access to this group.
568 * systemd-coredump will now automatically generate a
569 stack trace of all core dumps taking place on the system,
570 based on elfutils' libdw library. This stack trace is logged
573 * systemd-coredump may now optionally store coredumps directly
574 on disk (in /var/lib/systemd/coredump, possibly compressed),
575 instead of storing them unconditionally in the journal. This
576 mode is the new default. A new configuration file
577 /etc/systemd/coredump.conf has been added to configure this
578 and other parameters of systemd-coredump.
580 * coredumpctl gained a new "info" verb to show details about a
581 specific coredump. A new switch "-1" has also been added
582 that makes sure to only show information about the most
583 recent entry instead of all entries. Also, as the tool is
584 generally useful now the "systemd-" prefix of the binary
585 name has been removed. Distributions that want to maintain
586 compatibility with the old name should add a symlink from
587 the old name to the new name.
589 * journald's SplitMode= now defaults to "uid". This makes sure
590 that unprivileged users can access their own coredumps with
591 coredumpctl without restrictions.
593 * New kernel command line options "systemd.wants=" (for
594 pulling an additional unit during boot), "systemd.mask="
595 (for masking a specific unit for the boot), and
596 "systemd.debug-shell" (for enabling the debug shell on tty9)
597 have been added. This is implemented in the new generator
598 "systemd-debug-generator".
600 * systemd-nspawn will now by default filter a couple of
601 syscalls for containers, among them those required for
602 kernel module loading, direct x86 IO port access, swap
603 management, and kexec. Most importantly though
604 open_by_handle_at() is now prohibited for containers,
605 closing a hole similar to a recently discussed vulnerability
606 in docker regarding access to files on file hierarchies the
607 container should normally not have access to. Note that for
608 nspawn we generally make no security claims anyway (and
609 this is explicitly documented in the man page), so this is
610 just a fix for one of the most obvious problems.
612 * A new man page file-hierarchy(7) has been added that
613 contains a minimized, modernized version of the file system
614 layout systemd expects, similar in style to the FHS
615 specification or hier(5). A new tool systemd-path(1) has
616 been added to query many of these paths for the local
619 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
620 longer done. Since the directory now has a per-user size
621 limit, and is cleaned on logout this appears unnecessary,
622 in particular since this now brings the lifecycle of this
623 directory closer in line with how IPC objects are handled.
625 * systemd.pc now exports a number of additional directories,
626 including $libdir (which is useful to identify the library
627 path for the primary architecture of the system), and a
628 couple of drop-in directories.
630 * udev's predictable network interface names now use the dev_port
631 sysfs attribute, introduced in linux 3.15 instead of dev_id to
632 distinguish between ports of the same PCI function. dev_id should
633 only be used for ports using the same HW address, hence the need
636 * machined has been updated to export the OS version of a
637 container (read from /etc/os-release and
638 /usr/lib/os-release) on the bus. This is now shown in
639 "machinectl status" for a machine.
641 * A new service setting RestartForceExitStatus= has been
642 added. If configured to a set of exit signals or process
643 return values, the service will be restarted when the main
644 daemon process exits with any of them, regardless of the
647 * systemctl's -H switch for connecting to remote systemd
648 machines has been extended so that it may be used to
649 directly connect to a specific container on the
650 host. "systemctl -H root@foobar:waldi" will now connect as
651 user "root" to host "foobar", and then proceed directly to
652 the container named "waldi". Note that currently you have to
653 authenticate as user "root" for this to work, as entering
654 containers is a privileged operation.
656 Contributions from: Andreas Henriksson, Benjamin Steinwender,
657 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
658 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
659 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
660 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
661 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
662 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
663 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
664 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
665 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
666 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
667 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
669 -- Berlin, 2014-07-03
673 * As an experimental feature, udev now tries to lock the
674 disk device node (flock(LOCK_SH|LOCK_NB)) while it
675 executes events for the disk or any of its partitions.
676 Applications like partitioning programs can lock the
677 disk device node (flock(LOCK_EX)) and claim temporary
678 device ownership that way; udev will entirely skip all event
679 handling for this disk and its partitions. If the disk
680 was opened for writing, the close will trigger a partition
681 table rescan in udev's "watch" facility, and if needed
682 synthesize "change" events for the disk and all its partitions.
683 This is now unconditionally enabled, and if it turns out to
684 cause major problems, we might turn it on only for specific
685 devices, or might need to disable it entirely. Device Mapper
686 devices are excluded from this logic.
688 * We temporarily dropped the "-l" switch for fsck invocations,
689 since they collide with the flock() logic above. util-linux
690 upstream has been changed already to avoid this conflict,
691 and we will readd "-l" as soon as util-linux with this
692 change has been released.
694 * The dependency on libattr has been removed. Since a long
695 time, the extended attribute calls have moved to glibc, and
696 libattr is thus unnecessary.
698 * Virtualization detection works without priviliges now. This
699 means the systemd-detect-virt binary no longer requires
700 CAP_SYS_PTRACE file capabilities, and our daemons can run
701 with fewer privileges.
703 * systemd-networkd now runs under its own "systemd-network"
704 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
705 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
706 loses the ability to write to files owned by root this way.
708 * Similar, systemd-resolved now runs under its own
709 "systemd-resolve" user with no capabilities remaining.
711 * Similar, systemd-bus-proxyd now runs under its own
712 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
714 * systemd-networkd gained support for setting up "veth"
715 virtual ethernet devices for container connectivity, as well
716 as GRE and VTI tunnels.
718 * systemd-networkd will no longer automatically attempt to
719 manually load kernel modules necessary for certain tunnel
720 transports. Instead, it is assumed the kernel loads them
721 automatically when required. This only works correctly on
722 very new kernels. On older kernels, please consider adding
723 the kernel modules to /etc/modules-load.d/ as a work-around.
725 * The resolv.conf file systemd-resolved generates has been
726 moved to /run/systemd/resolve/. If you have a symlink from
727 /etc/resolv.conf, it might be necessary to correct it.
729 * Two new service settings, ProtectHome= and ProtectSystem=,
730 have been added. When enabled, they will make the user data
731 (such as /home) inaccessible or read-only and the system
732 (such as /usr) read-only, for specific services. This allows
733 very light-weight per-service sandboxing to avoid
734 modifications of user data or system files from
735 services. These two new switches have been enabled for all
736 of systemd's long-running services, where appropriate.
738 * Socket units gained new SocketUser= and SocketGroup=
739 settings to set the owner user and group of AF_UNIX sockets
740 and FIFOs in the file system.
742 * Socket units gained a new RemoveOnStop= setting. If enabled,
743 all FIFOS and sockets in the file system will be removed
744 when the specific socket unit is stopped.
746 * Socket units gained a new Symlinks= setting. It takes a list
747 of symlinks to create to file system sockets or FIFOs
748 created by the specific Unix sockets. This is useful to
749 manage symlinks to socket nodes with the same life-cycle as
752 * The /dev/log socket and /dev/initctl FIFO have been moved to
753 /run, and have been replaced by symlinks. This allows
754 connecting to these facilities even if PrivateDevices=yes is
755 used for a service (which makes /dev/log itself unavailable,
756 but /run is left). This also has the benefit of ensuring
757 that /dev only contains device nodes, directories and
758 symlinks, and nothing else.
760 * sd-daemon gained two new calls sd_pid_notify() and
761 sd_pid_notifyf(). They are similar to sd_notify() and
762 sd_notifyf(), but allow overriding of the source PID of
763 notification messages if permissions permit this. This is
764 useful to send notify messages on behalf of a different
765 process (for example, the parent process). The
766 systemd-notify tool has been updated to make use of this
767 when sending messages (so that notification messages now
768 originate from the shell script invoking systemd-notify and
769 not the systemd-notify process itself. This should minimize
770 a race where systemd fails to associate notification
771 messages to services when the originating process already
774 * A new "on-abnormal" setting for Restart= has been added. If
775 set, it will result in automatic restarts on all "abnormal"
776 reasons for a process to exit, which includes unclean
777 signals, core dumps, timeouts and watchdog timeouts, but
778 does not include clean and unclean exit codes or clean
779 signals. Restart=on-abnormal is an alternative for
780 Restart=on-failure for services that shall be able to
781 terminate and avoid restarts on certain errors, by
782 indicating so with an unclean exit code. Restart=on-failure
783 or Restart=on-abnormal is now the recommended setting for
784 all long-running services.
786 * If the InaccessibleDirectories= service setting points to a
787 mount point (or if there are any submounts contained within
788 it), it is now attempted to completely unmount it, to make
789 the file systems truly unavailable for the respective
792 * The ReadOnlyDirectories= service setting and
793 systemd-nspawn's --read-only parameter are now recursively
794 applied to all submounts, too.
796 * Mount units may now be created transiently via the bus APIs.
798 * The support for SysV and LSB init scripts has been removed
799 from the systemd daemon itself. Instead, it is now
800 implemented as a generator that creates native systemd units
801 from these scripts when needed. This enables us to remove a
802 substantial amount of legacy code from PID 1, following the
803 fact that many distributions only ship a very small number
804 of LSB/SysV init scripts nowadays.
806 * Priviliged Xen (dom0) domains are not considered
807 virtualization anymore by the virtualization detection
808 logic. After all, they generally have unrestricted access to
809 the hardware and usually are used to manage the unprivileged
812 * systemd-tmpfiles gained a new "C" line type, for copying
813 files or entire directories.
815 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
816 lines. So far, they have been non-globbing versions of the
817 latter, and have thus been redundant. In future, it is
818 recommended to only use "z". "m" has hence been removed
819 from the documentation, even though it stays supported.
821 * A tmpfiles snippet to recreate the most basic structure in
822 /var has been added. This is enough to create the /var/run →
823 /run symlink and create a couple of structural
824 directories. This allows systems to boot up with an empty or
825 volatile /var. Of course, while with this change, the core OS
826 now is capable with dealing with a volatile /var, not all
827 user services are ready for it. However, we hope that sooner
828 or later, many service daemons will be changed upstream so
829 that they are able to automatically create their necessary
830 directories in /var at boot, should they be missing. This is
831 the first step to allow state-less systems that only require
832 the vendor image for /usr to boot.
834 * systemd-nspawn has gained a new --tmpfs= switch to mount an
835 empty tmpfs instance to a specific directory. This is
836 particularly useful for making use of the automatic
837 reconstruction of /var (see above), by passing --tmpfs=/var.
839 * Access modes specified in tmpfiles snippets may now be
840 prefixed with "~", which indicates that they shall be masked
841 by whether the existing file or directory is currently
842 writable, readable or executable at all. Also, if specified,
843 the sgid/suid/sticky bits will be masked for all
846 * A new passive target unit "network-pre.target" has been
847 added which is useful for services that shall run before any
848 network is configured, for example firewall scripts.
850 * The "floppy" group that previously owned the /dev/fd*
851 devices is no longer used. The "disk" group is now used
852 instead. Distributions should probably deprecate usage of
855 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
856 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
857 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
858 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
859 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
860 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
863 -- Berlin, 2014-06-11
867 * A new "systemd-timesyncd" daemon has been added for
868 synchronizing the system clock across the network. It
869 implements an SNTP client. In contrast to NTP
870 implementations such as chrony or the NTP reference server,
871 this only implements a client side, and does not bother with
872 the full NTP complexity, focusing only on querying time from
873 one remote server and synchronizing the local clock to
874 it. Unless you intend to serve NTP to networked clients or
875 want to connect to local hardware clocks, this simple NTP
876 client should be more than appropriate for most
877 installations. The daemon runs with minimal privileges, and
878 has been hooked up with networkd to only operate when
879 network connectivity is available. The daemon saves the
880 current clock to disk every time a new NTP sync has been
881 acquired, and uses this to possibly correct the system clock
882 early at bootup, in order to accommodate for systems that
883 lack an RTC such as the Raspberry Pi and embedded devices,
884 and to make sure that time monotonically progresses on these
885 systems, even if it is not always correct. To make use of
886 this daemon, a new system user and group "systemd-timesync"
887 needs to be created on installation of systemd.
889 * The queue "seqnum" interface of libudev has been disabled, as
890 it was generally incompatible with device namespacing as
891 sequence numbers of devices go "missing" if the devices are
892 part of a different namespace.
894 * "systemctl list-timers" and "systemctl list-sockets" gained
895 a --recursive switch for showing units of these types also
896 for all local containers, similar in style to the already
897 supported --recursive switch for "systemctl list-units".
899 * A new RebootArgument= setting has been added for service
900 units, which may be used to specify a kernel reboot argument
901 to use when triggering reboots with StartLimitAction=.
903 * A new FailureAction= setting has been added for service
904 units which may be used to specify an operation to trigger
905 when a service fails. This works similarly to
906 StartLimitAction=, but unlike it, controls what is done
907 immediately rather than only after several attempts to
908 restart the service in question.
910 * hostnamed got updated to also expose the kernel name,
911 release, and version on the bus. This is useful for
912 executing commands like hostnamectl with the -H switch.
913 systemd-analyze makes use of this to properly display
914 details when running non-locally.
916 * The bootchart tool can now show cgroup information in the
919 * The CFS CPU quota cgroup attribute is now exposed for
920 services. The new CPUQuota= switch has been added for this
921 which takes a percentage value. Setting this will have the
922 result that a service may never get more CPU time than the
923 specified percentage, even if the machine is otherwise idle.
925 * systemd-networkd learned IPIP and SIT tunnel support.
927 * LSB init scripts exposing a dependency on $network will now
928 get a dependency on network-online.target rather than simply
929 network.target. This should bring LSB handling closer to
930 what it was on SysV systems.
932 * A new fsck.repair= kernel option has been added to control
933 how fsck shall deal with unclean file systems at boot.
935 * The (.ini) configuration file parser will now silently
936 ignore sections whose name begins with "X-". This may be
937 used to maintain application-specific extension sections in unit
940 * machined gained a new API to query the IP addresses of
941 registered containers. "machinectl status" has been updated
942 to show these addresses in its output.
944 * A new call sd_uid_get_display() has been added to the
945 sd-login APIs for querying the "primary" session of a
946 user. The "primary" session of the user is elected from the
947 user's sessions and generally a graphical session is
948 preferred over a text one.
950 * A minimal systemd-resolved daemon has been added. It
951 currently simply acts as a companion to systemd-networkd and
952 manages resolv.conf based on per-interface DNS
953 configuration, possibly supplied via DHCP. In the long run
954 we hope to extend this into a local DNSSEC enabled DNS and
957 * The systemd-networkd-wait-online tool is now enabled by
958 default. It will delay network-online.target until a network
959 connection has been configured. The tool primarily integrates
960 with networkd, but will also make a best effort to make sense
961 of network configuration performed in some other way.
963 * Two new service options StartupCPUShares= and
964 StartupBlockIOWeight= have been added that work similarly to
965 CPUShares= and BlockIOWeight= however only apply during
966 system startup. This is useful to prioritize certain services
967 differently during bootup than during normal runtime.
969 * hostnamed has been changed to prefer the statically
970 configured hostname in /etc/hostname (unless set to
971 'localhost' or empty) over any dynamic one supplied by
972 dhcp. With this change, the rules for picking the hostname
973 match more closely the rules of other configuration settings
974 where the local administrator's configuration in /etc always
975 overrides any other settings.
977 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
978 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
979 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
980 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
981 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
982 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
983 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
984 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
985 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
986 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
987 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
988 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
989 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
990 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
991 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
992 Lindskog, WaLyong Cho, Will Woods, Zbigniew
995 -- Beijing, 2014-05-28
999 * When restoring the screen brightness at boot, stay away from
1000 the darkest setting or from the lowest 5% of the available
1001 range, depending on which is the larger value of both. This
1002 should effectively protect the user from rebooting into a
1003 black screen, should the brightness have been set to minimum
1006 * sd-login gained a new sd_machine_get_class() call to
1007 determine the class ("vm" or "container") of a machine
1008 registered with machined.
1010 * sd-login gained new calls
1011 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
1012 to query the identity of the peer of a local AF_UNIX
1013 connection. They operate similarly to their sd_pid_get_xyz()
1016 * PID 1 will now maintain a system-wide system state engine
1017 with the states "starting", "running", "degraded",
1018 "maintenance", "stopping". These states are bound to system
1019 startup, normal runtime, runtime with at least one failed
1020 service, rescue/emergency mode and system shutdown. This
1021 state is shown in the "systemctl status" output when no unit
1022 name is passed. It is useful to determine system state, in
1023 particularly when doing so for many systems or containers at
1026 * A new command "list-machines" has been added to "systemctl"
1027 that lists all local OS containers and shows their system
1028 state (see above), if systemd runs inside of them.
1030 * systemctl gained a new "-r" switch to recursively enumerate
1031 units on all local containers, when used with the
1032 "list-unit" command (which is the default one that is
1033 executed when no parameters are specified).
1035 * The GPT automatic partition discovery logic will now honour
1036 two GPT partition flags: one may be set on a partition to
1037 cause it to be mounted read-only, and the other may be set
1038 on a partition to ignore it during automatic discovery.
1040 * Two new GPT type UUIDs have been added for automatic root
1041 partition discovery, for 32-bit and 64-bit ARM. This is not
1042 particularly useful for discovering the root directory on
1043 these architectures during bare-metal boots (since UEFI is
1044 not common there), but still very useful to allow booting of
1045 ARM disk images in nspawn with the -i option.
1047 * MAC addresses of interfaces created with nspawn's
1048 --network-interface= switch will now be generated from the
1049 machine name, and thus be stable between multiple invocations
1052 * logind will now automatically remove all IPC objects owned
1053 by a user if she or he fully logs out. This makes sure that
1054 users who are logged out cannot continue to consume IPC
1055 resources. This covers SysV memory, semaphores and message
1056 queues as well as POSIX shared memory and message
1057 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1058 limits. With this functionality, that is corrected. This may
1059 be turned off by using the RemoveIPC= switch of logind.conf.
1061 * The systemd-machine-id-setup and tmpfiles tools gained a
1062 --root= switch to operate on a specific root directory,
1065 * journald can now forward logged messages to the TTYs of all
1066 logged in users ("wall"). This is the default for all
1067 emergency messages now.
1069 * A new tool systemd-journal-remote has been added to stream
1070 journal log messages across the network.
1072 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1073 controller trees are mounted into it. Note that the
1074 directories mounted beneath it are not read-only. This is a
1075 security measure and is particularly useful because glibc
1076 actually includes a search logic to pick any tmpfs it can
1077 find to implement shm_open() if /dev/shm is not available
1078 (which it might very well be in namespaced setups).
1080 * machinectl gained a new "poweroff" command to cleanly power
1081 down a local OS container.
1083 * The PrivateDevices= unit file setting will now also drop the
1084 CAP_MKNOD capability from the capability bound set, and
1085 imply DevicePolicy=closed.
1087 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1088 comprehensively on all long-running systemd services where
1089 this is appropriate.
1091 * systemd-udevd will now run in a disassociated mount
1092 namespace. To mount directories from udev rules, make sure to
1093 pull in mount units via SYSTEMD_WANTS properties.
1095 * The kdbus support gained support for uploading policy into
1096 the kernel. sd-bus gained support for creating "monitoring"
1097 connections that can eavesdrop into all bus communication
1098 for debugging purposes.
1100 * Timestamps may now be specified in seconds since the UNIX
1101 epoch Jan 1st, 1970 by specifying "@" followed by the value
1104 * Native tcpwrap support in systemd has been removed. tcpwrap
1105 is old code, not really maintained anymore and has serious
1106 shortcomings, and better options such as firewalls
1107 exist. For setups that require tcpwrap usage, please
1108 consider invoking your socket-activated service via tcpd,
1109 like on traditional inetd.
1111 * A new system.conf configuration option
1112 DefaultTimerAccuracySec= has been added that controls the
1113 default AccuracySec= setting of .timer units.
1115 * Timer units gained a new WakeSystem= switch. If enabled,
1116 timers configured this way will cause the system to resume
1117 from system suspend (if the system supports that, which most
1120 * Timer units gained a new Persistent= switch. If enabled,
1121 timers configured this way will save to disk when they have
1122 been last triggered. This information is then used on next
1123 reboot to possible execute overdue timer events, that
1124 could not take place because the system was powered off.
1125 This enables simple anacron-like behaviour for timer units.
1127 * systemctl's "list-timers" will now also list the time a
1128 timer unit was last triggered in addition to the next time
1129 it will be triggered.
1131 * systemd-networkd will now assign predictable IPv4LL
1132 addresses to its local interfaces.
1134 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1135 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1136 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1137 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1138 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1139 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1140 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1141 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1144 -- Berlin, 2014-03-25
1148 * A new unit file setting RestrictAddressFamilies= has been
1149 added to restrict which socket address families unit
1150 processes gain access to. This takes address family names
1151 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1152 attack surface of services via exotic protocol stacks. This
1153 is built on seccomp system call filters.
1155 * Two new unit file settings RuntimeDirectory= and
1156 RuntimeDirectoryMode= have been added that may be used to
1157 manage a per-daemon runtime directories below /run. This is
1158 an alternative for setting up directory permissions with
1159 tmpfiles snippets, and has the advantage that the runtime
1160 directory's lifetime is bound to the daemon runtime and that
1161 the daemon starts up with an empty directory each time. This
1162 is particularly useful when writing services that drop
1163 privileges using the User= or Group= setting.
1165 * The DeviceAllow= unit setting now supports globbing for
1166 matching against device group names.
1168 * The systemd configuration file system.conf gained new
1169 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1170 DefaultMemoryAccounting= to globally turn on/off accounting
1171 for specific resources (cgroups) for all units. These
1172 settings may still be overridden individually in each unit
1175 * systemd-gpt-auto-generator is now able to discover /srv and
1176 root partitions in addition to /home and swap partitions. It
1177 also supports LUKS-encrypted partitions now. With this in
1178 place, automatic discovery of partitions to mount following
1179 the Discoverable Partitions Specification
1180 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1181 is now a lot more complete. This allows booting without
1182 /etc/fstab and without root= on the kernel command line on
1183 systems prepared appropriately.
1185 * systemd-nspawn gained a new --image= switch which allows
1186 booting up disk images and Linux installations on any block
1187 device that follow the Discoverable Partitions Specification
1188 (see above). This means that installations made with
1189 appropriately updated installers may now be started and
1190 deployed using container managers, completely
1191 unmodified. (We hope that libvirt-lxc will add support for
1192 this feature soon, too.)
1194 * systemd-nspawn gained a new --network-macvlan= setting to
1195 set up a private macvlan interface for the
1196 container. Similarly, systemd-networkd gained a new
1197 Kind=macvlan setting in .netdev files.
1199 * systemd-networkd now supports configuring local addresses
1202 * A new tool systemd-network-wait-online has been added to
1203 synchronously wait for network connectivity using
1206 * The sd-bus.h bus API gained a new sd_bus_track object for
1207 tracking the life-cycle of bus peers. Note that sd-bus.h is
1208 still not a public API though (unless you specify
1209 --enable-kdbus on the configure command line, which however
1210 voids your warranty and you get no API stability guarantee).
1212 * The $XDG_RUNTIME_DIR runtime directories for each user are
1213 now individual tmpfs instances, which has the benefit of
1214 introducing separate pools for each user, with individual
1215 size limits, and thus making sure that unprivileged clients
1216 can no longer negatively impact the system or other users by
1217 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1218 RuntimeDirectorySize= has been introduced that allows
1219 controlling the default size limit for all users. It
1220 defaults to 10% of the available physical memory. This is no
1221 replacement for quotas on tmpfs though (which the kernel
1222 still does not support), as /dev/shm and /tmp are still
1223 shared resources used by both the system and unprivileged
1226 * logind will now automatically turn off automatic suspending
1227 on laptop lid close when more than one display is
1228 connected. This was previously expected to be implemented
1229 individually in desktop environments (such as GNOME),
1230 however has been added to logind now, in order to fix a
1231 boot-time race where a desktop environment might not have
1232 been started yet and thus not been able to take an inhibitor
1233 lock at the time where logind already suspends the system
1234 due to a closed lid.
1236 * logind will now wait at least 30s after each system
1237 suspend/resume cycle, and 3min after system boot before
1238 suspending the system due to a closed laptop lid. This
1239 should give USB docking stations and similar enough time to
1240 be probed and configured after system resume and boot in
1241 order to then act as suspend blocker.
1243 * systemd-run gained a new --property= setting which allows
1244 initialization of resource control properties (and others)
1245 for the created scope or service unit. Example: "systemd-run
1246 --property=BlockIOWeight=10 updatedb" may be used to run
1247 updatedb at a low block IO scheduling weight.
1249 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1250 now also work in --scope mode.
1252 * When systemd is compiled with kdbus support, basic support
1253 for enforced policies is now in place. (Note that enabling
1254 kdbus still voids your warranty and no API compatibility
1257 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1258 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1259 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1260 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1261 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1262 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1263 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1264 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1265 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1266 Zbigniew Jędrzejewski-Szmek
1268 -- Berlin, 2014-03-12
1272 * systemd will now relabel /dev after loading the SMACK policy
1273 according to SMACK rules.
1275 * A new unit file option AppArmorProfile= has been added to
1276 set the AppArmor profile for the processes of a unit.
1278 * A new condition check ConditionArchitecture= has been added
1279 to conditionalize units based on the system architecture, as
1280 reported by uname()'s "machine" field.
1282 * systemd-networkd now supports matching on the system
1283 virtualization, architecture, kernel command line, host name
1286 * logind is now a lot more aggressive when suspending the
1287 machine due to a closed laptop lid. Instead of acting only
1288 on the lid close action, it will continuously watch the lid
1289 status and act on it. This is useful for laptops where the
1290 power button is on the outside of the chassis so that it can
1291 be reached without opening the lid (such as the Lenovo
1292 Yoga). On those machines, logind will now immediately
1293 re-suspend the machine if the power button has been
1294 accidentally pressed while the laptop was suspended and in a
1295 backpack or similar.
1297 * logind will now watch SW_DOCK switches and inhibit reaction
1298 to the lid switch if it is pressed. This means that logind
1299 will not suspend the machine anymore if the lid is closed
1300 and the system is docked, if the laptop supports SW_DOCK
1301 notifications via the input layer. Note that ACPI docking
1302 stations do not generate this currently. Also note that this
1303 logic is usually not fully sufficient and Desktop
1304 Environments should take a lid switch inhibitor lock when an
1305 external display is connected, as systemd will not watch
1308 * nspawn will now make use of the devices cgroup controller by
1309 default, and only permit creation of and access to the usual
1310 API device nodes like /dev/null or /dev/random, as well as
1311 access to (but not creation of) the pty devices.
1313 * We will now ship a default .network file for
1314 systemd-networkd that automatically configures DHCP for
1315 network interfaces created by nspawn's --network-veth or
1316 --network-bridge= switches.
1318 * systemd will now understand the usual M, K, G, T suffixes
1319 according to SI conventions (i.e. to the base 1000) when
1320 referring to throughput and hardware metrics. It will stay
1321 with IEC conventions (i.e. to the base 1024) for software
1322 metrics, according to what is customary according to
1323 Wikipedia. We explicitly document which base applies for
1324 each configuration option.
1326 * The DeviceAllow= setting in unit files now supports a syntax
1327 to whitelist an entire group of devices node majors at once,
1328 based on the /proc/devices listing. For example, with the
1329 string "char-pts", it is now possible to whitelist all
1330 current and future pseudo-TTYs at once.
1332 * sd-event learned a new "post" event source. Event sources of
1333 this type are triggered by the dispatching of any event
1334 source of a type that is not "post". This is useful for
1335 implementing clean-up and check event sources that are
1336 triggered by other work being done in the program.
1338 * systemd-networkd is no longer statically enabled, but uses
1339 the usual [Install] sections so that it can be
1340 enabled/disabled using systemctl. It still is enabled by
1343 * When creating a veth interface pair with systemd-nspawn, the
1344 host side will now be prefixed with "vb-" if
1345 --network-bridge= is used, and with "ve-" if --network-veth
1346 is used. This way, it is easy to distinguish these cases on
1347 the host, for example to apply different configuration to
1348 them with systemd-networkd.
1350 * The compatibility libraries for libsystemd-journal.so,
1351 libsystem-id128.so, libsystemd-login.so and
1352 libsystemd-daemon.so do not make use of IFUNC
1353 anymore. Instead, we now build libsystemd.so multiple times
1354 under these alternative names. This means that the footprint
1355 is drastically increased, but given that these are
1356 transitional compatibility libraries, this should not matter
1357 much. This change has been made necessary to support the ARM
1358 platform for these compatibility libraries, as the ARM
1359 toolchain is not really at the same level as the toolchain
1360 for other architectures like x86 and does not support
1361 IFUNC. Please make sure to use --enable-compat-libs only
1362 during a transitional period!
1364 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1365 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1366 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1367 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1368 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1369 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1370 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1371 Zbigniew Jędrzejewski-Szmek
1373 -- Berlin, 2014-02-24
1377 * A new component "systemd-networkd" has been added that can
1378 be used to configure local network interfaces statically or
1379 via DHCP. It is capable of bringing up bridges, VLANs, and
1380 bonding. Currently, no hook-ups for interactive network
1381 configuration are provided. Use this for your initrd,
1382 container, embedded, or server setup if you need a simple,
1383 yet powerful, network configuration solution. This
1384 configuration subsystem is quite nifty, as it allows wildcard
1385 hotplug matching in interfaces. For example, with a single
1386 configuration snippet, you can configure that all Ethernet
1387 interfaces showing up are automatically added to a bridge,
1388 or similar. It supports link-sensing and more.
1390 * A new tool "systemd-socket-proxyd" has been added which can
1391 act as a bidirectional proxy for TCP sockets. This is
1392 useful for adding socket activation support to services that
1393 do not actually support socket activation, including virtual
1394 machines and the like.
1396 * Add a new tool to save/restore rfkill state on
1399 * Save/restore state of keyboard backlights in addition to
1400 display backlights on shutdown/boot.
1402 * udev learned a new SECLABEL{} construct to label device
1403 nodes with a specific security label when they appear. For
1404 now, only SECLABEL{selinux} is supported, but the syntax is
1405 prepared for additional security frameworks.
1407 * udev gained a new scheme to configure link-level attributes
1408 from files in /etc/systemd/network/*.link. These files can
1409 match against MAC address, device path, driver name and type,
1410 and will apply attributes like the naming policy, link speed,
1411 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1412 address assignment policy (randomized, ...).
1414 * The configuration of network interface naming rules for
1415 "permanent interface names" has changed: a new NamePolicy=
1416 setting in the [Link] section of .link files determines the
1417 priority of possible naming schemes (onboard, slot, mac,
1418 path). The default value of this setting is determined by
1419 /usr/lib/net/links/99-default.link. Old
1420 80-net-name-slot.rules udev configuration file has been
1421 removed, so local configuration overriding this file should
1422 be adapated to override 99-default.link instead.
1424 * When the User= switch is used in a unit file, also
1425 initialize $SHELL= based on the user database entry.
1427 * systemd no longer depends on libdbus. All communication is
1428 now done with sd-bus, systemd's low-level bus library
1431 * kdbus support has been added to PID 1 itself. When kdbus is
1432 enabled, this causes PID 1 to set up the system bus and
1433 enable support for a new ".busname" unit type that
1434 encapsulates bus name activation on kdbus. It works a little
1435 bit like ".socket" units, except for bus names. A new
1436 generator has been added that converts classic dbus1 service
1437 activation files automatically into native systemd .busname
1440 * sd-bus: add a light-weight vtable implementation that allows
1441 defining objects on the bus with a simple static const
1442 vtable array of its methods, signals and properties.
1444 * systemd will not generate or install static dbus
1445 introspection data anymore to /usr/share/dbus-1/interfaces,
1446 as the precise format of these files is unclear, and
1447 nothing makes use of it.
1449 * A proxy daemon is now provided to proxy clients connecting
1450 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1451 compatibility with classic D-Bus.
1453 * A bus driver implementation has been added that supports the
1454 classic D-Bus bus driver calls on kdbus, also for
1455 compatibility purposes.
1457 * A new API "sd-event.h" has been added that implements a
1458 minimal event loop API built around epoll. It provides a
1459 couple of features that direct epoll usage is lacking:
1460 prioritization of events, scales to large numbers of timer
1461 events, per-event timer slack (accuracy), system-wide
1462 coalescing of timer events, exit handlers, watchdog
1463 supervision support using systemd's sd_notify() API, child
1466 * A new API "sd-rntl.h" has been added that provides an API
1467 around the route netlink interface of the kernel, similar in
1468 style to "sd-bus.h".
1470 * A new API "sd-dhcp-client.h" has been added that provides a
1471 small DHCPv4 client-side implementation. This is used by
1474 * There is a new kernel command line option
1475 "systemd.restore_state=0|1". When set to "0", none of the
1476 systemd tools will restore saved runtime state to hardware
1477 devices. More specifically, the rfkill and backlight states
1480 * The FsckPassNo= compatibility option in mount/service units
1481 has been removed. The fstab generator will now add the
1482 necessary dependencies automatically, and does not require
1483 PID1's support for that anymore.
1485 * journalctl gained a new switch, --list-boots, that lists
1486 recent boots with their times and boot IDs.
1488 * The various tools like systemctl, loginctl, timedatectl,
1489 busctl, systemd-run, ... have gained a new switch "-M" to
1490 connect to a specific, local OS container (as direct
1491 connection, without requiring SSH). This works on any
1492 container that is registered with machined, such as those
1493 created by libvirt-lxc or nspawn.
1495 * systemd-run and systemd-analyze also gained support for "-H"
1496 to connect to remote hosts via SSH. This is particularly
1497 useful for systemd-run because it enables queuing of jobs
1498 onto remote systems.
1500 * machinectl gained a new command "login" to open a getty
1501 login in any local container. This works with any container
1502 that is registered with machined (such as those created by
1503 libvirt-lxc or nspawn), and which runs systemd inside.
1505 * machinectl gained a new "reboot" command that may be used to
1506 trigger a reboot on a specific container that is registered
1507 with machined. This works on any container that runs an init
1508 system of some kind.
1510 * systemctl gained a new "list-timers" command to print a nice
1511 listing of installed timer units with the times they elapse
1514 * Alternative reboot() parameters may now be specified on the
1515 "systemctl reboot" command line and are passed to the
1516 reboot() system call.
1518 * systemctl gained a new --job-mode= switch to configure the
1519 mode to queue a job with. This is a more generic version of
1520 --fail, --irreversible, and --ignore-dependencies, which are
1521 still available but not advertised anymore.
1523 * /etc/systemd/system.conf gained new settings to configure
1524 various default timeouts of units, as well as the default
1525 start limit interval and burst. These may still be overridden
1528 * PID1 will now export on the bus profile data of the security
1529 policy upload process (such as the SELinux policy upload to
1532 * journald: when forwarding logs to the console, include
1533 timestamps (following the setting in
1534 /sys/module/printk/parameters/time).
1536 * OnCalendar= in timer units now understands the special
1537 strings "yearly" and "annually". (Both are equivalent)
1539 * The accuracy of timer units is now configurable with the new
1540 AccuracySec= setting. It defaults to 1min.
1542 * A new dependency type JoinsNamespaceOf= has been added that
1543 allows running two services within the same /tmp and network
1544 namespace, if PrivateNetwork= or PrivateTmp= are used.
1546 * A new command "cat" has been added to systemctl. It outputs
1547 the original unit file of a unit, and concatenates the
1548 contents of additional "drop-in" unit file snippets, so that
1549 the full configuration is shown.
1551 * systemctl now supports globbing on the various "list-xyz"
1552 commands, like "list-units" or "list-sockets", as well as on
1553 those commands which take multiple unit names.
1555 * journalctl's --unit= switch gained support for globbing.
1557 * All systemd daemons now make use of the watchdog logic so
1558 that systemd automatically notices when they hang.
1560 * If the $container_ttys environment variable is set,
1561 getty-generator will automatically spawn a getty for each
1562 listed tty. This is useful for container managers to request
1563 login gettys to be spawned on as many ttys as needed.
1565 * %h, %s, %U specifier support is not available anymore when
1566 used in unit files for PID 1. This is because NSS calls are
1567 not safe from PID 1. They stay available for --user
1568 instances of systemd, and as special case for the root user.
1570 * loginctl gained a new "--no-legend" switch to turn off output
1573 * The "sd-login.h" API gained three new calls:
1574 sd_session_is_remote(), sd_session_get_remote_user(),
1575 sd_session_get_remote_host() to query information about
1578 * The udev hardware database now also carries vendor/product
1579 information of SDIO devices.
1581 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1582 determine whether watchdog notifications are requested by
1585 * Socket-activated per-connection services now include a
1586 short description of the connection parameters in the
1589 * tmpfiles gained a new "--boot" option. When this is not used,
1590 only lines where the command character is not suffixed with
1591 "!" are executed. When this option is specified, those
1592 options are executed too. This partitions tmpfiles
1593 directives into those that can be safely executed at any
1594 time, and those which should be run only at boot (for
1595 example, a line that creates /run/nologin).
1597 * A new API "sd-resolve.h" has been added which provides a simple
1598 asynchronous wrapper around glibc NSS host name resolution
1599 calls, such as getaddrinfo(). In contrast to glibc's
1600 getaddrinfo_a(), it does not use signals. In contrast to most
1601 other asynchronous name resolution libraries, this one does
1602 not reimplement DNS, but reuses NSS, so that alternate
1603 host name resolution systems continue to work, such as mDNS,
1604 LDAP, etc. This API is based on libasyncns, but it has been
1605 cleaned up for inclusion in systemd.
1607 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1608 "sd-daemon.h" are no longer found in individual libraries
1609 libsystemd-journal.so, libsystemd-login.so,
1610 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1611 merged them into a single library, libsystemd.so, which
1612 provides all symbols. The reason for this is cyclic
1613 dependencies, as these libraries tend to use each other's
1614 symbols. So far, we have managed to workaround that by linking
1615 a copy of a good part of our code into each of these
1616 libraries again and again, which, however, makes certain
1617 things hard to do, like sharing static variables. Also, it
1618 substantially increases footprint. With this change, there
1619 is only one library for the basic APIs systemd
1620 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1621 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1622 library as well, however are subject to the --enable-kdbus
1623 switch (see below). Note that "sd-dhcp-client.h" is not part
1624 of this library (this is because it only consumes, never
1625 provides, services of/to other APIs). To make the transition
1626 easy from the separate libraries to the unified one, we
1627 provide the --enable-compat-libs compile-time switch which
1628 will generate stub libraries that are compatible with the
1629 old ones but redirect all calls to the new one.
1631 * All of the kdbus logic and the new APIs "sd-bus.h",
1632 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1633 and "sd-utf8.h" are compile-time optional via the
1634 "--enable-kdbus" switch, and they are not compiled in by
1635 default. To make use of kdbus, you have to explicitly enable
1636 the switch. Note however, that neither the kernel nor the
1637 userspace API for all of this is considered stable yet. We
1638 want to maintain the freedom to still change the APIs for
1639 now. By specifying this build-time switch, you acknowledge
1640 that you are aware of the instability of the current
1643 * Also, note that while kdbus is pretty much complete,
1644 it lacks one thing: proper policy support. This means you
1645 can build a fully working system with all features; however,
1646 it will be highly insecure. Policy support will be added in
1647 one of the next releases, at the same time that we will
1648 declare the APIs stable.
1650 * When the kernel command line argument "kdbus" is specified,
1651 systemd will automatically load the kdbus.ko kernel module. At
1652 this stage of development, it is only useful for testing kdbus
1653 and should not be used in production. Note: if "--enable-kdbus"
1654 is specified, and the kdbus.ko kernel module is available, and
1655 "kdbus" is added to the kernel command line, the entire system
1656 runs with kdbus instead of dbus-daemon, with the above mentioned
1657 problem of missing the system policy enforcement. Also a future
1658 version of kdbus.ko or a newer systemd will not be compatible with
1659 each other, and will unlikely be able to boot the machine if only
1660 one of them is updated.
1662 * systemctl gained a new "import-environment" command which
1663 uploads the caller's environment (or parts thereof) into the
1664 service manager so that it is inherited by services started
1665 by the manager. This is useful to upload variables like
1666 $DISPLAY into the user service manager.
1668 * A new PrivateDevices= switch has been added to service units
1669 which allows running a service with a namespaced /dev
1670 directory that does not contain any device nodes for
1671 physical devices. More specifically, it only includes devices
1672 such as /dev/null, /dev/urandom, and /dev/zero which are API
1675 * logind has been extended to support behaviour like VT
1676 switching on seats that do not support a VT. This makes
1677 multi-session available on seats that are not the first seat
1678 (seat0), and on systems where kernel support for VTs has
1679 been disabled at compile-time.
1681 * If a process holds a delay lock for system sleep or shutdown
1682 and fails to release it in time, we will now log its
1683 identity. This makes it easier to identify processes that
1684 cause slow suspends or power-offs.
1686 * When parsing /etc/crypttab, support for a new key-slot=
1687 option as supported by Debian is added. It allows indicating
1688 which LUKS slot to use on disk, speeding up key loading.
1690 * The sd_journald_sendv() API call has been checked and
1691 officially declared to be async-signal-safe so that it may
1692 be invoked from signal handlers for logging purposes.
1694 * Boot-time status output is now enabled automatically after a
1695 short timeout if boot does not progress, in order to give
1696 the user an indication what she or he is waiting for.
1698 * The boot-time output has been improved to show how much time
1699 remains until jobs expire.
1701 * The KillMode= switch in service units gained a new possible
1702 value "mixed". If set, and the unit is shut down, then the
1703 initial SIGTERM signal is sent only to the main daemon
1704 process, while the following SIGKILL signal is sent to
1705 all remaining processes of the service.
1707 * When a scope unit is registered, a new property "Controller"
1708 may be set. If set to a valid bus name, systemd will send a
1709 RequestStop() signal to this name when it would like to shut
1710 down the scope. This may be used to hook manager logic into
1711 the shutdown logic of scope units. Also, scope units may now
1712 be put in a special "abandoned" state, in which case the
1713 manager process which created them takes no further
1714 responsibilities for it.
1716 * When reading unit files, systemd will now verify
1717 the access mode of these files, and warn about certain
1718 suspicious combinations. This has been added to make it
1719 easier to track down packaging bugs where unit files are
1720 marked executable or world-writable.
1722 * systemd-nspawn gained a new "--setenv=" switch to set
1723 container-wide environment variables. The similar option in
1724 systemd-activate was renamed from "--environment=" to
1725 "--setenv=" for consistency.
1727 * systemd-nspawn has been updated to create a new kdbus domain
1728 for each container that is invoked, thus allowing each
1729 container to have its own set of system and user buses,
1730 independent of the host.
1732 * systemd-nspawn gained a new --drop-capability= switch to run
1733 the container with less capabilities than the default. Both
1734 --drop-capability= and --capability= now take the special
1735 string "all" for dropping or keeping all capabilities.
1737 * systemd-nspawn gained new switches for executing containers
1738 with specific SELinux labels set.
1740 * systemd-nspawn gained a new --quiet switch to not generate
1741 any additional output but the container's own console
1744 * systemd-nspawn gained a new --share-system switch to run a
1745 container without PID namespacing enabled.
1747 * systemd-nspawn gained a new --register= switch to control
1748 whether the container is registered with systemd-machined or
1749 not. This is useful for containers that do not run full
1750 OS images, but only specific apps.
1752 * systemd-nspawn gained a new --keep-unit which may be used
1753 when invoked as the only program from a service unit, and
1754 results in registration of the unit service itself in
1755 systemd-machined, instead of a newly opened scope unit.
1757 * systemd-nspawn gained a new --network-interface= switch for
1758 moving arbitrary interfaces to the container. The new
1759 --network-veth switch creates a virtual Ethernet connection
1760 between host and container. The new --network-bridge=
1761 switch then allows assigning the host side of this virtual
1762 Ethernet connection to a bridge device.
1764 * systemd-nspawn gained a new --personality= switch for
1765 setting the kernel personality for the container. This is
1766 useful when running a 32-bit container on a 64-bit host. A
1767 similar option Personality= is now also available for service
1770 * logind will now also track a "Desktop" identifier for each
1771 session which encodes the desktop environment of it. This is
1772 useful for desktop environments that want to identify
1773 multiple running sessions of itself easily.
1775 * A new SELinuxContext= setting for service units has been
1776 added that allows setting a specific SELinux execution
1777 context for a service.
1779 * Most systemd client tools will now honour $SYSTEMD_LESS for
1780 settings of the "less" pager. By default, these tools will
1781 override $LESS to allow certain operations to work, such as
1782 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1783 influence this logic.
1785 * systemd's "seccomp" hook-up has been changed to make use of
1786 the libseccomp library instead of using its own
1787 implementation. This has benefits for portability among
1790 * For usage together with SystemCallFilter=, a new
1791 SystemCallErrorNumber= setting has been introduced that
1792 allows configuration of a system error number to be returned
1793 on filtered system calls, instead of immediately killing the
1794 process. Also, SystemCallArchitectures= has been added to
1795 limit access to system calls of a particular architecture
1796 (in order to turn off support for unused secondary
1797 architectures). There is also a global
1798 SystemCallArchitectures= setting in system.conf now to turn
1799 off support for non-native system calls system-wide.
1801 * systemd requires a kernel with a working name_to_handle_at(),
1802 please see the kernel config requirements in the README file.
1804 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1805 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1806 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1807 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1808 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1809 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1810 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1811 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1812 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1813 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1814 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1815 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1816 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1817 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1818 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1819 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1820 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1821 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1822 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1823 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1824 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1825 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1826 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1827 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1829 -- Berlin, 2014-02-20
1833 * logind has gained support for facilitating privileged input
1834 and drm device access for unprivileged clients. This work is
1835 useful to allow Wayland display servers (and similar
1836 programs, such as kmscon) to run under the user's ID and
1837 access input and drm devices which are normally
1838 protected. When this is used (and the kernel is new enough)
1839 logind will "mute" IO on the file descriptors passed to
1840 Wayland as long as it is in the background and "unmute" it
1841 if it returns into the foreground. This allows secure
1842 session switching without allowing background sessions to
1843 eavesdrop on input and display data. This also introduces
1844 session switching support if VT support is turned off in the
1845 kernel, and on seats that are not seat0.
1847 * A new kernel command line option luks.options= is understood
1848 now which allows specifying LUKS options for usage for LUKS
1849 encrypted partitions specified with luks.uuid=.
1851 * tmpfiles.d(5) snippets may now use specifier expansion in
1852 path names. More specifically %m, %b, %H, %v, are now
1853 replaced by the local machine id, boot id, hostname, and
1854 kernel version number.
1856 * A new tmpfiles.d(5) command "m" has been introduced which
1857 may be used to change the owner/group/access mode of a file
1858 or directory if it exists, but do nothing if it does not.
1860 * This release removes high-level support for the
1861 MemorySoftLimit= cgroup setting. The underlying kernel
1862 cgroup attribute memory.soft_limit= is currently badly
1863 designed and likely to be removed from the kernel API in its
1864 current form, hence we should not expose it for now.
1866 * The memory.use_hierarchy cgroup attribute is now enabled for
1867 all cgroups systemd creates in the memory cgroup
1868 hierarchy. This option is likely to be come the built-in
1869 default in the kernel anyway, and the non-hierarchial mode
1870 never made much sense in the intrinsically hierarchial
1873 * A new field _SYSTEMD_SLICE= is logged along with all journal
1874 messages containing the slice a message was generated
1875 from. This is useful to allow easy per-customer filtering of
1876 logs among other things.
1878 * systemd-journald will no longer adjust the group of journal
1879 files it creates to the "systemd-journal" group. Instead we
1880 rely on the journal directory to be owned by the
1881 "systemd-journal" group, and its setgid bit set, so that the
1882 kernel file system layer will automatically enforce that
1883 journal files inherit this group assignment. The reason for
1884 this change is that we cannot allow NSS look-ups from
1885 journald which would be necessary to resolve
1886 "systemd-journal" to a numeric GID, because this might
1887 create deadlocks if NSS involves synchronous queries to
1888 other daemons (such as nscd, or sssd) which in turn are
1889 logging clients of journald and might block on it, which
1890 would then dead lock. A tmpfiles.d(5) snippet included in
1891 systemd will make sure the setgid bit and group are
1892 properly set on the journal directory if it exists on every
1893 boot. However, we recommend adjusting it manually after
1894 upgrades too (or from RPM scriptlets), so that the change is
1895 not delayed until next reboot.
1897 * Backlight and random seed files in /var/lib/ have moved into
1898 the /var/lib/systemd/ directory, in order to centralize all
1899 systemd generated files in one directory.
1901 * Boot time performance measurements (as displayed by
1902 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1903 performance information if that's available to determine how
1904 much time BIOS and boot loader initialization required. With
1905 a sufficiently new BIOS you hence no longer need to boot
1906 with Gummiboot to get access to such information.
1908 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1909 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1910 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1911 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1912 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1913 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1914 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1916 -- Berlin, 2013-10-02
1920 * The Restart= option for services now understands a new
1921 on-watchdog setting, which will restart the service
1922 automatically if the service stops sending out watchdog keep
1923 alive messages (as configured with WatchdogSec=).
1925 * The getty generator (which is responsible for bringing up a
1926 getty on configured serial consoles) will no longer only
1927 start a getty on the primary kernel console but on all
1928 others, too. This makes the order in which console= is
1929 specified on the kernel command line less important.
1931 * libsystemd-logind gained a new sd_session_get_vt() call to
1932 retrieve the VT number of a session.
1934 * If the option "tries=0" is set for an entry of /etc/crypttab
1935 its passphrase is queried indefinitely instead of any
1936 maximum number of tries.
1938 * If a service with a configure PID file terminates its PID
1939 file will now be removed automatically if it still exists
1940 afterwards. This should put an end to stale PID files.
1942 * systemd-run will now also take relative binary path names
1943 for execution and no longer insists on absolute paths.
1945 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1946 paths that are optionally prefixed with "-" to indicate that
1947 it should not be considered a failure if they do not exist.
1949 * journalctl -o (and similar commands) now understands a new
1950 output mode "short-precise", it is similar to "short" but
1951 shows timestamps with usec accuracy.
1953 * The option "discard" (as known from Debian) is now
1954 synonymous to "allow-discards" in /etc/crypttab. In fact,
1955 "discard" is preferred now (since it is easier to remember
1958 * Some licensing clean-ups were made, so that more code is now
1959 LGPL-2.1 licensed than before.
1961 * A minimal tool to save/restore the display backlight
1962 brightness across reboots has been added. It will store the
1963 backlight setting as late as possible at shutdown, and
1964 restore it as early as possible during reboot.
1966 * A logic to automatically discover and enable home and swap
1967 partitions on GPT disks has been added. With this in place
1968 /etc/fstab becomes optional for many setups as systemd can
1969 discover certain partitions located on the root disk
1970 automatically. Home partitions are recognized under their
1971 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1972 partitions are recognized under their GPT type ID
1973 0657fd6da4ab43c484e50933c84b4f4f.
1975 * systemd will no longer pass any environment from the kernel
1976 or initrd to system services. If you want to set an
1977 environment for all services, do so via the kernel command
1978 line systemd.setenv= assignment.
1980 * The systemd-sysctl tool no longer natively reads the file
1981 /etc/sysctl.conf. If desired, the file should be symlinked
1982 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1983 legacy support by a symlink rather than built-in code, it
1984 also makes the otherwise hidden order of application of the
1985 different files visible. (Note that this partly reverts to a
1986 pre-198 application order of sysctl knobs!)
1988 * The "systemctl set-log-level" and "systemctl dump" commands
1989 have been moved to systemd-analyze.
1991 * systemd-run learned the new --remain-after-exit switch,
1992 which causes the scope unit not to be cleaned up
1993 automatically after the process terminated.
1995 * tmpfiles learned a new --exclude-prefix= switch to exclude
1996 certain paths from operation.
1998 * journald will now automatically flush all messages to disk
1999 as soon as a message at the log level CRIT, ALERT or EMERG
2002 Contributions from: Andrew Cook, Brandon Philips, Christian
2003 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
2004 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
2005 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
2006 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
2007 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
2008 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
2009 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
2010 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
2011 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
2012 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
2013 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
2014 William Giokas, Zbigniew Jędrzejewski-Szmek
2016 -- Berlin, 2013-09-13
2020 * The documentation has been updated to cover the various new
2021 concepts introduced with 205.
2023 * Unit files now understand the new %v specifier which
2024 resolves to the kernel version string as returned by "uname
2027 * systemctl now supports filtering the unit list output by
2028 load state, active state and sub state, using the new
2031 * "systemctl status" will now show the results of the
2032 condition checks (like ConditionPathExists= and similar) of
2033 the last start attempts of the unit. They are also logged to
2036 * "journalctl -b" may now be used to look for boot output of a
2037 specific boot. Try "journalctl -b -1" for the previous boot,
2038 but the syntax is substantially more powerful.
2040 * "journalctl --show-cursor" has been added which prints the
2041 cursor string the last shown log line. This may then be used
2042 with the new "journalctl --after-cursor=" switch to continue
2043 browsing logs from that point on.
2045 * "journalctl --force" may now be used to force regeneration
2048 * Creation of "dead" device nodes has been moved from udev
2049 into kmod and tmpfiles. Previously, udev would read the kmod
2050 databases to pre-generate dead device nodes based on meta
2051 information contained in kernel modules, so that these would
2052 be auto-loaded on access rather then at boot. As this
2053 does not really have much to do with the exposing actual
2054 kernel devices to userspace this has always been slightly
2055 alien in the udev codebase. Following the new scheme kmod
2056 will now generate a runtime snippet for tmpfiles from the
2057 module meta information and it now is tmpfiles' job to the
2058 create the nodes. This also allows overriding access and
2059 other parameters for the nodes using the usual tmpfiles
2060 facilities. As side effect this allows us to remove the
2061 CAP_SYS_MKNOD capability bit from udevd entirely.
2063 * logind's device ACLs may now be applied to these "dead"
2064 devices nodes too, thus finally allowing managed access to
2065 devices such as /dev/snd/sequencer whithout loading the
2066 backing module right-away.
2068 * A new RPM macro has been added that may be used to apply
2069 tmpfiles configuration during package installation.
2071 * systemd-detect-virt and ConditionVirtualization= now can
2072 detect User-Mode-Linux machines (UML).
2074 * journald will now implicitly log the effective capabilities
2075 set of processes in the message metadata.
2077 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2079 * The initrd interface has been simplified (more specifically,
2080 support for passing performance data via environment
2081 variables and fsck results via files in /run has been
2082 removed). These features were non-essential, and are
2083 nowadays available in a much nicer way by having systemd in
2084 the initrd serialize its state and have the hosts systemd
2085 deserialize it again.
2087 * The udev "keymap" data files and tools to apply keyboard
2088 specific mappings of scan to key codes, and force-release
2089 scan code lists have been entirely replaced by a udev
2090 "keyboard" builtin and a hwdb data file.
2092 * systemd will now honour the kernel's "quiet" command line
2093 argument also during late shutdown, resulting in a
2094 completely silent shutdown when used.
2096 * There's now an option to control the SO_REUSEPORT socket
2097 option in .socket units.
2099 * Instance units will now automatically get a per-template
2100 subslice of system.slice unless something else is explicitly
2101 configured. For example, instances of sshd@.service will now
2102 implicitly be placed in system-sshd.slice rather than
2103 system.slice as before.
2105 * Test coverage support may now be enabled at build time.
2107 Contributions from: Dave Reisner, Frederic Crozat, Harald
2108 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2109 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2110 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2111 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2112 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2113 Giokas, Zbigniew Jędrzejewski-Szmek
2115 -- Berlin, 2013-07-23
2119 * Two new unit types have been introduced:
2121 Scope units are very similar to service units, however, are
2122 created out of pre-existing processes -- instead of PID 1
2123 forking off the processes. By using scope units it is
2124 possible for system services and applications to group their
2125 own child processes (worker processes) in a powerful way
2126 which then maybe used to organize them, or kill them
2127 together, or apply resource limits on them.
2129 Slice units may be used to partition system resources in an
2130 hierarchial fashion and then assign other units to them. By
2131 default there are now three slices: system.slice (for all
2132 system services), user.slice (for all user sessions),
2133 machine.slice (for VMs and containers).
2135 Slices and scopes have been introduced primarily in
2136 context of the work to move cgroup handling to a
2137 single-writer scheme, where only PID 1
2138 creates/removes/manages cgroups.
2140 * There's a new concept of "transient" units. In contrast to
2141 normal units these units are created via an API at runtime,
2142 not from configuration from disk. More specifically this
2143 means it is now possible to run arbitrary programs as
2144 independent services, with all execution parameters passed
2145 in via bus APIs rather than read from disk. Transient units
2146 make systemd substantially more dynamic then it ever was,
2147 and useful as a general batch manager.
2149 * logind has been updated to make use of scope and slice units
2150 for managing user sessions. As a user logs in he will get
2151 his own private slice unit, to which all sessions are added
2152 as scope units. We also added support for automatically
2153 adding an instance of user@.service for the user into the
2154 slice. Effectively logind will no longer create cgroup
2155 hierarchies on its own now, it will defer entirely to PID 1
2156 for this by means of scope, service and slice units. Since
2157 user sessions this way become entities managed by PID 1
2158 the output of "systemctl" is now a lot more comprehensive.
2160 * A new mini-daemon "systemd-machined" has been added which
2161 may be used by virtualization managers to register local
2162 VMs/containers. nspawn has been updated accordingly, and
2163 libvirt will be updated shortly. machined will collect a bit
2164 of meta information about the VMs/containers, and assign
2165 them their own scope unit (see above). The collected
2166 meta-data is then made available via the "machinectl" tool,
2167 and exposed in "ps" and similar tools. machined/machinectl
2168 is compile-time optional.
2170 * As discussed earlier, the low-level cgroup configuration
2171 options ControlGroup=, ControlGroupModify=,
2172 ControlGroupPersistent=, ControlGroupAttribute= have been
2173 removed. Please use high-level attribute settings instead as
2174 well as slice units.
2176 * A new bus call SetUnitProperties() has been added to alter
2177 various runtime parameters of a unit. This is primarily
2178 useful to alter cgroup parameters dynamically in a nice way,
2179 but will be extended later on to make more properties
2180 modifiable at runtime. systemctl gained a new set-properties
2181 command that wraps this call.
2183 * A new tool "systemd-run" has been added which can be used to
2184 run arbitrary command lines as transient services or scopes,
2185 while configuring a number of settings via the command
2186 line. This tool is currently very basic, however already
2187 very useful. We plan to extend this tool to even allow
2188 queuing of execution jobs with time triggers from the
2189 command line, similar in fashion to "at".
2191 * nspawn will now inform the user explicitly that kernels with
2192 audit enabled break containers, and suggest the user to turn
2195 * Support for detecting the IMA and AppArmor security
2196 frameworks with ConditionSecurity= has been added.
2198 * journalctl gained a new "-k" switch for showing only kernel
2199 messages, mimicking dmesg output; in addition to "--user"
2200 and "--system" switches for showing only user's own logs
2203 * systemd-delta can now show information about drop-in
2204 snippets extending unit files.
2206 * libsystemd-bus has been substantially updated but is still
2207 not available as public API.
2209 * systemd will now look for the "debug" argument on the kernel
2210 command line and enable debug logging, similar to what
2211 "systemd.log_level=debug" already did before.
2213 * "systemctl set-default", "systemctl get-default" has been
2214 added to configure the default.target symlink, which
2215 controls what to boot into by default.
2217 * "systemctl set-log-level" has been added as a convenient
2218 way to raise and lower systemd logging threshold.
2220 * "systemd-analyze plot" will now show the time the various
2221 generators needed for execution, as well as information
2222 about the unit file loading.
2224 * libsystemd-journal gained a new sd_journal_open_files() call
2225 for opening specific journal files. journactl also gained a
2226 new switch to expose this new functionality. Previously we
2227 only supported opening all files from a directory, or all
2228 files from the system, as opening individual files only is
2229 racy due to journal file rotation.
2231 * systemd gained the new DefaultEnvironment= setting in
2232 /etc/systemd/system.conf to set environment variables for
2235 * If a privileged process logs a journal message with the
2236 OBJECT_PID= field set, then journald will automatically
2237 augment this with additional OBJECT_UID=, OBJECT_GID=,
2238 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2239 system services want to log events about specific client
2240 processes. journactl/systemctl has been updated to make use
2241 of this information if all log messages regarding a specific
2244 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2245 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2246 Reisner, David Coppa, David King, David Strauss, Eelco
2247 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2248 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2249 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2250 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2251 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2252 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2253 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2254 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2255 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2256 Łukasz Stelmach, 장동준
2260 * The Python bindings gained some minimal support for the APIs
2261 exposed by libsystemd-logind.
2263 * ConditionSecurity= gained support for detecting SMACK. Since
2264 this condition already supports SELinux and AppArmor we only
2265 miss IMA for this. Patches welcome!
2267 Contributions from: Karol Lewandowski, Lennart Poettering,
2268 Zbigniew Jędrzejewski-Szmek
2272 * systemd-nspawn will now create /etc/resolv.conf if
2273 necessary, before bind-mounting the host's file onto it.
2275 * systemd-nspawn will now store meta information about a
2276 container on the container's cgroup as extended attribute
2277 fields, including the root directory.
2279 * The cgroup hierarchy has been reworked in many ways. All
2280 objects any of the components systemd creates in the cgroup
2281 tree are now suffixed. More specifically, user sessions are
2282 now placed in cgroups suffixed with ".session", users in
2283 cgroups suffixed with ".user", and nspawn containers in
2284 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2285 names are now escaped in a simple scheme to avoid collision
2286 of userspace object names with kernel filenames. This work
2287 is preparation for making these objects relocatable in the
2288 cgroup tree, in order to allow easy resource partitioning of
2289 these objects without causing naming conflicts.
2291 * systemctl list-dependencies gained the new switches
2292 --plain, --reverse, --after and --before.
2294 * systemd-inhibit now shows the process name of processes that
2295 have taken an inhibitor lock.
2297 * nss-myhostname will now also resolve "localhost"
2298 implicitly. This makes /etc/hosts an optional file and
2299 nicely handles that on IPv6 ::1 maps to both "localhost" and
2302 * libsystemd-logind.so gained a new call
2303 sd_get_machine_names() to enumerate running containers and
2304 VMs (currently only supported by very new libvirt and
2305 nspawn). sd_login_monitor can now be used to watch
2306 VMs/containers coming and going.
2308 * .include is not allowed recursively anymore, and only in
2309 unit files. Usually it is better to use drop-in snippets in
2310 .d/*.conf anyway, as introduced with systemd 198.
2312 * systemd-analyze gained a new "critical-chain" command that
2313 determines the slowest chain of units run during system
2314 boot-up. It is very useful for tracking down where
2315 optimizing boot time is the most beneficial.
2317 * systemd will no longer allow manipulating service paths in
2318 the name=systemd:/system cgroup tree using ControlGroup= in
2319 units. (But is still fine with it in all other dirs.)
2321 * There's a new systemd-nspawn@.service service file that may
2322 be used to easily run nspawn containers as system
2323 services. With the container's root directory in
2324 /var/lib/container/foobar it is now sufficient to run
2325 "systemctl start systemd-nspawn@foobar.service" to boot it.
2327 * systemd-cgls gained a new parameter "--machine" to list only
2328 the processes within a certain container.
2330 * ConditionSecurity= now can check for "apparmor". We still
2331 are lacking checks for SMACK and IMA for this condition
2332 check though. Patches welcome!
2334 * A new configuration file /etc/systemd/sleep.conf has been
2335 added that may be used to configure which kernel operation
2336 systemd is supposed to execute when "suspend", "hibernate"
2337 or "hybrid-sleep" is requested. This makes the new kernel
2338 "freeze" state accessible to the user.
2340 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2341 the passed argument if applicable.
2343 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2344 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2345 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2346 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2347 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2348 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2349 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2354 * The output of 'systemctl list-jobs' got some polishing. The
2355 '--type=' argument may now be passed more than once. A new
2356 command 'systemctl list-sockets' has been added which shows
2357 a list of kernel sockets systemd is listening on with the
2358 socket units they belong to, plus the units these socket
2361 * The experimental libsystemd-bus library got substantial
2362 updates to work in conjunction with the (also experimental)
2363 kdbus kernel project. It works well enough to exchange
2364 messages with some sophistication. Note that kdbus is not
2365 ready yet, and the library is mostly an elaborate test case
2366 for now, and not installable.
2368 * systemd gained a new unit 'systemd-static-nodes.service'
2369 that generates static device nodes earlier during boot, and
2370 can run in conjunction with udev.
2372 * libsystemd-login gained a new call sd_pid_get_user_unit()
2373 to retrieve the user systemd unit a process is running
2374 in. This is useful for systems where systemd is used as
2377 * systemd-nspawn now places all containers in the new /machine
2378 top-level cgroup directory in the name=systemd
2379 hierarchy. libvirt will soon do the same, so that we get a
2380 uniform separation of /system, /user and /machine for system
2381 services, user processes and containers/virtual
2382 machines. This new cgroup hierarchy is also useful to stick
2383 stable names to specific container instances, which can be
2384 recognized later this way (this name may be controlled
2385 via systemd-nspawn's new -M switch). libsystemd-login also
2386 gained a new call sd_pid_get_machine_name() to retrieve the
2387 name of the container/VM a specific process belongs to.
2389 * bootchart can now store its data in the journal.
2391 * libsystemd-journal gained a new call
2392 sd_journal_add_conjunction() for AND expressions to the
2393 matching logic. This can be used to express more complex
2394 logical expressions.
2396 * journactl can now take multiple --unit= and --user-unit=
2399 * The cryptsetup logic now understands the "luks.key=" kernel
2400 command line switch for specifying a file to read the
2401 decryption key from. Also, if a configured key file is not
2402 found the tool will now automatically fall back to prompting
2405 * Python systemd.journal module was updated to wrap recently
2406 added functions from libsystemd-journal. The interface was
2407 changed to bring the low level interface in s.j._Reader
2408 closer to the C API, and the high level interface in
2409 s.j.Reader was updated to wrap and convert all data about
2412 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2413 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2414 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2415 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2416 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2417 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2421 * journalctl --update-catalog now understands a new --root=
2422 option to operate on catalogs found in a different root
2425 * During shutdown after systemd has terminated all running
2426 services a final killing loop kills all remaining left-over
2427 processes. We will now print the name of these processes
2428 when we send SIGKILL to them, since this usually indicates a
2431 * If /etc/crypttab refers to password files stored on
2432 configured mount points automatic dependencies will now be
2433 generated to ensure the specific mount is established first
2434 before the key file is attempted to be read.
2436 * 'systemctl status' will now show information about the
2437 network sockets a socket unit is listening on.
2439 * 'systemctl status' will also shown information about any
2440 drop-in configuration file for units. (Drop-In configuration
2441 files in this context are files such as
2442 /etc/systemd/systemd/foobar.service.d/*.conf)
2444 * systemd-cgtop now optionally shows summed up CPU times of
2445 cgroups. Press '%' while running cgtop to switch between
2446 percentage and absolute mode. This is useful to determine
2447 which cgroups use up the most CPU time over the entire
2448 runtime of the system. systemd-cgtop has also been updated
2449 to be 'pipeable' for processing with further shell tools.
2451 * 'hostnamectl set-hostname' will now allow setting of FQDN
2454 * The formatting and parsing of time span values has been
2455 changed. The parser now understands fractional expressions
2456 such as "5.5h". The formatter will now output fractional
2457 expressions for all time spans under 1min, i.e. "5.123456s"
2458 rather than "5s 123ms 456us". For time spans under 1s
2459 millisecond values are shown, for those under 1ms
2460 microsecond values are shown. This should greatly improve
2461 all time-related output of systemd.
2463 * libsystemd-login and libsystemd-journal gained new
2464 functions for querying the poll() events mask and poll()
2465 timeout value for integration into arbitrary event
2468 * localectl gained the ability to list available X11 keymaps
2469 (models, layouts, variants, options).
2471 * 'systemd-analyze dot' gained the ability to filter for
2472 specific units via shell-style globs, to create smaller,
2473 more useful graphs. I.e. it is now possible to create simple
2474 graphs of all the dependencies between only target units, or
2475 of all units that Avahi has dependencies with.
2477 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2478 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2479 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2480 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2481 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2482 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2483 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2487 * The boot-time readahead implementation for rotating media
2488 will now read the read-ahead data in multiple passes which
2489 consist of all read requests made in equidistant time
2490 intervals. This means instead of strictly reading read-ahead
2491 data in its physical order on disk we now try to find a
2492 middle ground between physical and access time order.
2494 * /etc/os-release files gained a new BUILD_ID= field for usage
2495 on operating systems that provide continuous builds of OS
2498 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2499 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2500 William Douglas, Zbigniew Jędrzejewski-Szmek
2504 * systemd-python gained an API exposing libsystemd-daemon.
2506 * The SMACK setup logic gained support for uploading CIPSO
2509 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2510 ReadOnlyDirectories= and InaccessibleDirectories= has
2511 changed. The private /tmp and /var/tmp directories are now
2512 shared by all processes of a service (which means
2513 ExecStartPre= may now leave data in /tmp that ExecStart= of
2514 the same service can still access). When a service is
2515 stopped its temporary directories are immediately deleted
2516 (normal clean-up with tmpfiles is still done in addition to
2519 * By default, systemd will now set a couple of sysctl
2520 variables in the kernel: the safe sysrq options are turned
2521 on, IP route verification is turned on, and source routing
2522 disabled. The recently added hardlink and softlink
2523 protection of the kernel is turned on. These settings should
2524 be reasonably safe, and good defaults for all new systems.
2526 * The predictable network naming logic may now be turned off
2527 with a new kernel command line switch: net.ifnames=0.
2529 * A new libsystemd-bus module has been added that implements a
2530 pretty complete D-Bus client library. For details see:
2532 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2534 * journald will now explicitly flush the journal files to disk
2535 at the latest 5min after each write. The file will then also
2536 be marked offline until the next write. This should increase
2537 reliability in case of a crash. The synchronization delay
2538 can be configured via SyncIntervalSec= in journald.conf.
2540 * There's a new remote-fs-setup.target unit that can be used
2541 to pull in specific services when at least one remote file
2542 system is to be mounted.
2544 * There are new targets timers.target and paths.target as
2545 canonical targets to pull user timer and path units in
2546 from. This complements sockets.target with a similar
2547 purpose for socket units.
2549 * libudev gained a new call udev_device_set_attribute_value()
2550 to set sysfs attributes of a device.
2552 * The udev daemon now sets the default number of worker
2553 processes executed in parallel based on the number of available
2554 CPUs instead of the amount of available RAM. This is supposed
2555 to provide a more reliable default and limit a too aggressive
2556 paralellism for setups with 1000s of devices connected.
2558 Contributions from: Auke Kok, Colin Walters, Cristian
2559 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2560 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2561 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2562 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2563 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2564 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2565 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2566 Zbigniew Jędrzejewski-Szmek
2570 * Configuration of unit files may now be extended via drop-in
2571 files without having to edit/override the unit files
2572 themselves. More specifically, if the administrator wants to
2573 change one value for a service file foobar.service he can
2574 now do so by dropping in a configuration snippet into
2575 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2576 will load all these snippets and apply them on top of the
2577 main unit configuration file, possibly extending or
2578 overriding its settings. Using these drop-in snippets is
2579 generally nicer than the two earlier options for changing
2580 unit files locally: copying the files from
2581 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2582 them there; or creating a new file in /etc/systemd/system/
2583 that incorporates the original one via ".include". Drop-in
2584 snippets into these .d/ directories can be placed in any
2585 directory systemd looks for units in, and the usual
2586 overriding semantics between /usr/lib, /etc and /run apply
2589 * Most unit file settings which take lists of items can now be
2590 reset by assigning the empty string to them. For example,
2591 normally, settings such as Environment=FOO=BAR append a new
2592 environment variable assignment to the environment block,
2593 each time they are used. By assigning Environment= the empty
2594 string the environment block can be reset to empty. This is
2595 particularly useful with the .d/*.conf drop-in snippets
2596 mentioned above, since this adds the ability to reset list
2597 settings from vendor unit files via these drop-ins.
2599 * systemctl gained a new "list-dependencies" command for
2600 listing the dependencies of a unit recursively.
2602 * Inhibitors are now honored and listed by "systemctl
2603 suspend", "systemctl poweroff" (and similar) too, not only
2604 GNOME. These commands will also list active sessions by
2607 * Resource limits (as exposed by the various control group
2608 controllers) can now be controlled dynamically at runtime
2609 for all units. More specifically, you can now use a command
2610 like "systemctl set-cgroup-attr foobar.service cpu.shares
2611 2000" to alter the CPU shares a specific service gets. These
2612 settings are stored persistently on disk, and thus allow the
2613 administrator to easily adjust the resource usage of
2614 services with a few simple commands. This dynamic resource
2615 management logic is also available to other programs via the
2616 bus. Almost any kernel cgroup attribute and controller is
2619 * systemd-vconsole-setup will now copy all font settings to
2620 all allocated VTs, where it previously applied them only to
2623 * libsystemd-login gained the new sd_session_get_tty() API
2626 * This release drops support for a few legacy or
2627 distribution-specific LSB facility names when parsing init
2628 scripts: $x-display-manager, $mail-transfer-agent,
2629 $mail-transport-agent, $mail-transfer-agent, $smtp,
2630 $null. Also, the mail-transfer-agent.target unit backing
2631 this has been removed. Distributions which want to retain
2632 compatibility with this should carry the burden for
2633 supporting this themselves and patch support for these back
2634 in, if they really need to. Also, the facilities $syslog and
2635 $local_fs are now ignored, since systemd does not support
2636 early-boot LSB init scripts anymore, and these facilities
2637 are implied anyway for normal services. syslog.target has
2640 * There are new bus calls on PID1's Manager object for
2641 cancelling jobs, and removing snapshot units. Previously,
2642 both calls were only available on the Job and Snapshot
2645 * systemd-journal-gatewayd gained SSL support.
2647 * The various "environment" files, such as /etc/locale.conf
2648 now support continuation lines with a backslash ("\") as
2649 last character in the line, similarly in style (but different)
2650 to how this is supported in shells.
2652 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2653 now implicitly appended to every log entry logged. systemctl
2654 has been updated to filter by this field when operating on a
2655 user systemd instance.
2657 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2658 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2659 the container. This makes it easier to boot unmodified
2660 Fedora systems in a container, which however still requires
2661 audit=0 to be passed on the kernel command line. Auditing in
2662 kernel and userspace is unfortunately still too broken in
2663 context of containers, hence we recommend compiling it out
2664 of the kernel or using audit=0. Hopefully this will be fixed
2665 one day for good in the kernel.
2667 * nspawn gained the new --bind= and --bind-ro= parameters to
2668 bind mount specific directories from the host into the
2671 * nspawn will now mount its own devpts file system instance
2672 into the container, in order not to leak pty devices from
2673 the host into the container.
2675 * systemd will now read the firmware boot time performance
2676 information from the EFI variables, if the used boot loader
2677 supports this, and takes it into account for boot performance
2678 analysis via "systemd-analyze". This is currently supported
2679 only in conjunction with Gummiboot, but could be supported
2680 by other boot loaders too. For details see:
2682 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2684 * A new generator has been added that automatically mounts the
2685 EFI System Partition (ESP) to /boot, if that directory
2686 exists, is empty, and no other file system has been
2687 configured to be mounted there.
2689 * logind will now send out PrepareForSleep(false) out
2690 unconditionally, after coming back from suspend. This may be
2691 used by applications as asynchronous notification for
2692 system resume events.
2694 * "systemctl unlock-sessions" has been added, that allows
2695 unlocking the screens of all user sessions at once, similar
2696 to how "systemctl lock-sessions" already locked all users
2697 sessions. This is backed by a new D-Bus call UnlockSessions().
2699 * "loginctl seat-status" will now show the master device of a
2700 seat. (i.e. the device of a seat that needs to be around for
2701 the seat to be considered available, usually the graphics
2704 * tmpfiles gained a new "X" line type, that allows
2705 configuration of files and directories (with wildcards) that
2706 shall be excluded from automatic cleanup ("aging").
2708 * udev default rules set the device node permissions now only
2709 at "add" events, and do not change them any longer with a
2710 later "change" event.
2712 * The log messages for lid events and power/sleep keypresses
2713 now carry a message ID.
2715 * We now have a substantially larger unit test suite, but this
2716 continues to be work in progress.
2718 * udevadm hwdb gained a new --root= parameter to change the
2719 root directory to operate relative to.
2721 * logind will now issue a background sync() request to the kernel
2722 early at shutdown, so that dirty buffers are flushed to disk early
2723 instead of at the last moment, in order to optimize shutdown
2726 * A new bootctl tool has been added that is an interface for
2727 certain boot loader operations. This is currently a preview
2728 and is likely to be extended into a small mechanism daemon
2729 like timedated, localed, hostnamed, and can be used by
2730 graphical UIs to enumerate available boot options, and
2731 request boot into firmware operations.
2733 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2734 the rest of the package. It also has been updated to work
2735 correctly in initrds.
2737 * Policykit previously has been runtime optional, and is now
2738 also compile time optional via a configure switch.
2740 * systemd-analyze has been reimplemented in C. Also "systemctl
2741 dot" has moved into systemd-analyze.
2743 * "systemctl status" with no further parameters will now print
2744 the status of all active or failed units.
2746 * Operations such as "systemctl start" can now be executed
2747 with a new mode "--irreversible" which may be used to queue
2748 operations that cannot accidentally be reversed by a later
2749 job queuing. This is by default used to make shutdown
2750 requests more robust.
2752 * The Python API of systemd now gained a new module for
2753 reading journal files.
2755 * A new tool kernel-install has been added that can install
2756 kernel images according to the Boot Loader Specification:
2758 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2760 * Boot time console output has been improved to provide
2761 animated boot time output for hanging jobs.
2763 * A new tool systemd-activate has been added which can be used
2764 to test socket activation with, directly from the command
2765 line. This should make it much easier to test and debug
2766 socket activation in daemons.
2768 * journalctl gained a new "--reverse" (or -r) option to show
2769 journal output in reverse order (i.e. newest line first).
2771 * journalctl gained a new "--pager-end" (or -e) option to jump
2772 to immediately jump to the end of the journal in the
2773 pager. This is only supported in conjunction with "less".
2775 * journalctl gained a new "--user-unit=" option, that works
2776 similarly to "--unit=" but filters for user units rather than
2779 * A number of unit files to ease adoption of systemd in
2780 initrds has been added. This moves some minimal logic from
2781 the various initrd implementations into systemd proper.
2783 * The journal files are now owned by a new group
2784 "systemd-journal", which exists specifically to allow access
2785 to the journal, and nothing else. Previously, we used the
2786 "adm" group for that, which however possibly covers more
2787 than just journal/log file access. This new group is now
2788 already used by systemd-journal-gatewayd to ensure this
2789 daemon gets access to the journal files and as little else
2790 as possible. Note that "make install" will also set FS ACLs
2791 up for /var/log/journal to give "adm" and "wheel" read
2792 access to it, in addition to "systemd-journal" which owns
2793 the journal files. We recommend that packaging scripts also
2794 add read access to "adm" + "wheel" to /var/log/journal, and
2795 all existing/future journal files. To normal users and
2796 administrators little changes, however packagers need to
2797 ensure to create the "systemd-journal" system group at
2798 package installation time.
2800 * The systemd-journal-gatewayd now runs as unprivileged user
2801 systemd-journal-gateway:systemd-journal-gateway. Packaging
2802 scripts need to create these system user/group at
2805 * timedated now exposes a new boolean property CanNTP that
2806 indicates whether a local NTP service is available or not.
2808 * systemd-detect-virt will now also detect xen PVs
2810 * The pstore file system is now mounted by default, if it is
2813 * In addition to the SELinux and IMA policies we will now also
2814 load SMACK policies at early boot.
2816 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2817 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2818 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2819 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2820 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2821 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2822 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2823 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2824 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2825 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2826 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2827 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2828 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2829 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2833 * Timer units now support calendar time events in addition to
2834 monotonic time events. That means you can now trigger a unit
2835 based on a calendar time specification such as "Thu,Fri
2836 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2837 or fifth day of any month of the year 2013, given that it is
2838 a thursday or friday. This brings timer event support
2839 considerably closer to cron's capabilities. For details on
2840 the supported calendar time specification language see
2843 * udev now supports a number of different naming policies for
2844 network interfaces for predictable names, and a combination
2845 of these policies is now the default. Please see this wiki
2846 document for details:
2848 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2850 * Auke Kok's bootchart implementation has been added to the
2851 systemd tree. It is an optional component that can graph the
2852 boot in quite some detail. It is one of the best bootchart
2853 implementations around and minimal in its code and
2856 * nss-myhostname has been integrated into the systemd source
2857 tree. nss-myhostname guarantees that the local hostname
2858 always stays resolvable via NSS. It has been a weak
2859 requirement of systemd-hostnamed since a long time, and
2860 since its code is actually trivial we decided to just
2861 include it in systemd's source tree. It can be turned off
2862 with a configure switch.
2864 * The read-ahead logic is now capable of properly detecting
2865 whether a btrfs file system is on SSD or rotating media, in
2866 order to optimize the read-ahead scheme. Previously, it was
2867 only capable of detecting this on traditional file systems
2870 * In udev, additional device properties are now read from the
2871 IAB in addition to the OUI database. Also, Bluetooth company
2872 identities are attached to the devices as well.
2874 * In service files %U may be used as specifier that is
2875 replaced by the configured user name of the service.
2877 * nspawn may now be invoked without a controlling TTY. This
2878 makes it suitable for invocation as its own service. This
2879 may be used to set up a simple containerized server system
2880 using only core OS tools.
2882 * systemd and nspawn can now accept socket file descriptors
2883 when they are started for socket activation. This enables
2884 implementation of socket activated nspawn
2885 containers. i.e. think about autospawning an entire OS image
2886 when the first SSH or HTTP connection is received. We expect
2887 that similar functionality will also be added to libvirt-lxc
2890 * journalctl will now suppress ANSI color codes when
2891 presenting log data.
2893 * systemctl will no longer show control group information for
2894 a unit if a the control group is empty anyway.
2896 * logind can now automatically suspend/hibernate/shutdown the
2899 * /etc/machine-info and hostnamed now also expose the chassis
2900 type of the system. This can be used to determine whether
2901 the local system is a laptop, desktop, handset or
2902 tablet. This information may either be configured by the
2903 user/vendor or is automatically determined from ACPI and DMI
2904 information if possible.
2906 * A number of PolicyKit actions are now bound together with
2907 "imply" rules. This should simplify creating UIs because
2908 many actions will now authenticate similar ones as well.
2910 * Unit files learnt a new condition ConditionACPower= which
2911 may be used to conditionalize a unit depending on whether an
2912 AC power source is connected or not, of whether the system
2913 is running on battery power.
2915 * systemctl gained a new "is-failed" verb that may be used in
2916 shell scripts and suchlike to check whether a specific unit
2917 is in the "failed" state.
2919 * The EnvironmentFile= setting in unit files now supports file
2920 globbing, and can hence be used to easily read a number of
2921 environment files at once.
2923 * systemd will no longer detect and recognize specific
2924 distributions. All distribution-specific #ifdeffery has been
2925 removed, systemd is now fully generic and
2926 distribution-agnostic. Effectively, not too much is lost as
2927 a lot of the code is still accessible via explicit configure
2928 switches. However, support for some distribution specific
2929 legacy configuration file formats has been dropped. We
2930 recommend distributions to simply adopt the configuration
2931 files everybody else uses now and convert the old
2932 configuration from packaging scripts. Most distributions
2933 already did that. If that's not possible or desirable,
2934 distributions are welcome to forward port the specific
2935 pieces of code locally from the git history.
2937 * When logging a message about a unit systemd will now always
2938 log the unit name in the message meta data.
2940 * localectl will now also discover system locale data that is
2941 not stored in locale archives, but directly unpacked.
2943 * logind will no longer unconditionally use framebuffer
2944 devices as seat masters, i.e. as devices that are required
2945 to be existing before a seat is considered preset. Instead,
2946 it will now look for all devices that are tagged as
2947 "seat-master" in udev. By default framebuffer devices will
2948 be marked as such, but depending on local systems other
2949 devices might be marked as well. This may be used to
2950 integrate graphics cards using closed source drivers (such
2951 as NVidia ones) more nicely into logind. Note however, that
2952 we recommend using the open source NVidia drivers instead,
2953 and no udev rules for the closed-source drivers will be
2954 shipped from us upstream.
2956 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2957 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2958 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2959 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2960 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2961 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2962 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2963 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2964 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2965 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2966 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2967 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2972 * udev gained support for loading additional device properties
2973 from an indexed database that is keyed by vendor/product IDs
2974 and similar device identifiers. For the beginning this
2975 "hwdb" is populated with data from the well-known PCI and
2976 USB database, but also includes PNP, ACPI and OID data. In
2977 the longer run this indexed database shall grow into
2978 becoming the one central database for non-essential
2979 userspace device metadata. Previously, data from the PCI/USB
2980 database was only attached to select devices, since the
2981 lookup was a relatively expensive operation due to O(n) time
2982 complexity (with n being the number of entries in the
2983 database). Since this is now O(1), we decided to add in this
2984 data for all devices where this is available, by
2985 default. Note that the indexed database needs to be rebuilt
2986 when new data files are installed. To achieve this you need
2987 to update your packaging scripts to invoke "udevadm hwdb
2988 --update" after installation of hwdb data files. For
2989 RPM-based distributions we introduced the new
2990 %udev_hwdb_update macro for this purpose.
2992 * The Journal gained support for the "Message Catalog", an
2993 indexed database to link up additional information with
2994 journal entries. For further details please check:
2996 http://www.freedesktop.org/wiki/Software/systemd/catalog
2998 The indexed message catalog database also needs to be
2999 rebuilt after installation of message catalog files. Use
3000 "journalctl --update-catalog" for this. For RPM-based
3001 distributions we introduced the %journal_catalog_update
3002 macro for this purpose.
3004 * The Python Journal bindings gained support for the standard
3005 Python logging framework.
3007 * The Journal API gained new functions for checking whether
3008 the underlying file system of a journal file is capable of
3009 properly reporting file change notifications, or whether
3010 applications that want to reflect journal changes "live"
3011 need to recheck journal files continuously in appropriate
3014 * It is now possible to set the "age" field for tmpfiles
3015 entries to 0, indicating that files matching this entry
3016 shall always be removed when the directories are cleaned up.
3018 * coredumpctl gained a new "gdb" verb which invokes gdb
3019 right-away on the selected coredump.
3021 * There's now support for "hybrid sleep" on kernels that
3022 support this, in addition to "suspend" and "hibernate". Use
3023 "systemctl hybrid-sleep" to make use of this.
3025 * logind's HandleSuspendKey= setting (and related settings)
3026 now gained support for a new "lock" setting to simply
3027 request the screen lock on all local sessions, instead of
3028 actually executing a suspend or hibernation.
3030 * systemd will now mount the EFI variables file system by
3033 * Socket units now gained support for configuration of the
3034 SMACK security label.
3036 * timedatectl will now output the time of the last and next
3037 daylight saving change.
3039 * We dropped support for various legacy and distro-specific
3040 concepts, such as insserv, early-boot SysV services
3041 (i.e. those for non-standard runlevels such as 'b' or 'S')
3042 or ArchLinux /etc/rc.conf support. We recommend the
3043 distributions who still need support this to either continue
3044 to maintain the necessary patches downstream, or find a
3045 different solution. (Talk to us if you have questions!)
3047 * Various systemd components will now bypass PolicyKit checks
3048 for root and otherwise handle properly if PolicyKit is not
3049 found to be around. This should fix most issues for
3050 PolicyKit-less systems. Quite frankly this should have been
3051 this way since day one. It is absolutely our intention to
3052 make systemd work fine on PolicyKit-less systems, and we
3053 consider it a bug if something does not work as it should if
3054 PolicyKit is not around.
3056 * For embedded systems it is now possible to build udev and
3057 systemd without blkid and/or kmod support.
3059 * "systemctl switch-root" is now capable of switching root
3060 more than once. I.e. in addition to transitions from the
3061 initrd to the host OS it is now possible to transition to
3062 further OS images from the host. This is useful to implement
3063 offline updating tools.
3065 * Various other additions have been made to the RPM macros
3066 shipped with systemd. Use %udev_rules_update() after
3067 installing new udev rules files. %_udevhwdbdir,
3068 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3069 %_sysctldir are now available which resolve to the right
3070 directories for packages to place various data files in.
3072 * journalctl gained the new --full switch (in addition to
3073 --all, to disable ellipsation for long messages.
3075 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3076 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3077 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3078 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3079 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3080 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3081 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3082 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3083 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3087 * journalctl gained new --since= and --until= switches to
3088 filter by time. It also now supports nice filtering for
3089 units via --unit=/-u.
3091 * Type=oneshot services may use ExecReload= and do the
3094 * The journal daemon now supports time-based rotation and
3095 vacuuming, in addition to the usual disk-space based
3098 * The journal will now index the available field values for
3099 each field name. This enables clients to show pretty drop
3100 downs of available match values when filtering. The bash
3101 completion of journalctl has been updated
3102 accordingly. journalctl gained a new switch -F to list all
3103 values a certain field takes in the journal database.
3105 * More service events are now written as structured messages
3106 to the journal, and made recognizable via message IDs.
3108 * The timedated, localed and hostnamed mini-services which
3109 previously only provided support for changing time, locale
3110 and hostname settings from graphical DEs such as GNOME now
3111 also have a minimal (but very useful) text-based client
3112 utility each. This is probably the nicest way to changing
3113 these settings from the command line now, especially since
3114 it lists available options and is fully integrated with bash
3117 * There's now a new tool "systemd-coredumpctl" to list and
3118 extract coredumps from the journal.
3120 * We now install a README each in /var/log/ and
3121 /etc/rc.d/init.d explaining where the system logs and init
3122 scripts went. This hopefully should help folks who go to
3123 that dirs and look into the otherwise now empty void and
3124 scratch their heads.
3126 * When user-services are invoked (by systemd --user) the
3127 $MANAGERPID env var is set to the PID of systemd.
3129 * SIGRTMIN+24 when sent to a --user instance will now result
3130 in immediate termination of systemd.
3132 * gatewayd received numerous feature additions such as a
3133 "follow" mode, for live syncing and filtering.
3135 * browse.html now allows filtering and showing detailed
3136 information on specific entries. Keyboard navigation and
3137 mouse screen support has been added.
3139 * gatewayd/journalctl now supports HTML5/JSON
3140 Server-Sent-Events as output.
3142 * The SysV init script compatibility logic will now
3143 heuristically determine whether a script supports the
3144 "reload" verb, and only then make this available as
3147 * "systemctl status --follow" has been removed, use "journalctl
3150 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3151 have been removed since they are hardly useful to be
3154 * And I'd like to take the opportunity to specifically mention
3155 Zbigniew for his great contributions. Zbigniew, you rock!
3157 Contributions from: Andrew Eikum, Christian Hesse, Colin
3158 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3159 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3160 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3161 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3162 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3163 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3167 * If /etc/vconsole.conf is non-existent or empty we will no
3168 longer load any console font or key map at boot by
3169 default. Instead the kernel defaults will be left
3170 intact. This is definitely the right thing to do, as no
3171 configuration should mean no configuration, and hard-coding
3172 font names that are different on all archs is probably a bad
3173 idea. Also, the kernel default key map and font should be
3174 good enough for most cases anyway, and mostly identical to
3175 the userspace fonts/key maps we previously overloaded them
3176 with. If distributions want to continue to default to a
3177 non-kernel font or key map they should ship a default
3178 /etc/vconsole.conf with the appropriate contents.
3180 Contributions from: Colin Walters, Daniel J Walsh, Dave
3181 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3182 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3186 * journalctl gained a new --cursor= switch to show entries
3187 starting from the specified location in the journal.
3189 * We now enforce a size limit on journal entry fields exported
3190 with "-o json" in journalctl. Fields larger than 4K will be
3191 assigned null. This can be turned off with --all.
3193 * An (optional) journal gateway daemon is now available as
3194 "systemd-journal-gatewayd.service". This service provides
3195 access to the journal via HTTP and JSON. This functionality
3196 will be used to implement live log synchronization in both
3197 pull and push modes, but has various other users too, such
3198 as easy log access for debugging of embedded devices. Right
3199 now it is already useful to retrieve the journal via HTTP:
3201 # systemctl start systemd-journal-gatewayd.service
3202 # wget http://localhost:19531/entries
3204 This will download the journal contents in a
3205 /var/log/messages compatible format. The same as JSON:
3207 # curl -H"Accept: application/json" http://localhost:19531/entries
3209 This service is also accessible via a web browser where a
3210 single static HTML5 app is served that uses the JSON logic
3211 to enable the user to do some basic browsing of the
3212 journal. This will be extended later on. Here's an example
3213 screenshot of this app in its current state:
3215 http://0pointer.de/public/journal-gatewayd
3217 Contributions from: Kay Sievers, Lennart Poettering, Robert
3218 Milasan, Tom Gundersen
3222 * The bash completion logic is now available for journalctl
3225 * We do not mount the "cpuset" controller anymore together with
3226 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3227 started if no parameters are assigned to it. "cpuset" hence
3228 broke code that assumed it it could create "cpu" groups and
3231 * journalctl -f will now subscribe to terminal size changes,
3232 and line break accordingly.
3234 Contributions from: Dave Reisner, Kay Sievers, Lennart
3235 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3239 * nspawn will now create a symlink /etc/localtime in the
3240 container environment, copying the host's timezone
3241 setting. Previously this has been done via a bind mount, but
3242 since symlinks cannot be bind mounted this has now been
3243 changed to create/update the appropriate symlink.
3245 * journalctl -n's line number argument is now optional, and
3246 will default to 10 if omitted.
3248 * journald will now log the maximum size the journal files may
3249 take up on disk. This is particularly useful if the default
3250 built-in logic of determining this parameter from the file
3251 system size is used. Use "systemctl status
3252 systemd-journald.service" to see this information.
3254 * The multi-seat X wrapper tool has been stripped down. As X
3255 is now capable of enumerating graphics devices via udev in a
3256 seat-aware way the wrapper is not strictly necessary
3257 anymore. A stripped down temporary stop-gap is still shipped
3258 until the upstream display managers have been updated to
3259 fully support the new X logic. Expect this wrapper to be
3260 removed entirely in one of the next releases.
3262 * HandleSleepKey= in logind.conf has been split up into
3263 HandleSuspendKey= and HandleHibernateKey=. The old setting
3264 is not available anymore. X11 and the kernel are
3265 distuingishing between these keys and we should too. This
3266 also means the inhibition lock for these keys has been split
3269 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3270 Poettering, Lukas Nykryn, Václav Pavlín
3274 * Whenever a unit changes state we will now log this to the
3275 journal and show along the unit's own log output in
3278 * ConditionPathIsMountPoint= can now properly detect bind
3279 mount points too. (Previously, a bind mount of one file
3280 system to another place in the same file system could not be
3281 detected as mount, since they shared struct stat's st_dev
3284 * We will now mount the cgroup controllers cpu, cpuacct,
3285 cpuset and the controllers net_cls, net_prio together by
3288 * nspawn containers will now have a virtualized boot
3289 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3290 over with a randomized ID at container initialization). This
3291 has the effect of making "journalctl -b" do the right thing
3294 * The JSON output journal serialization has been updated not
3295 to generate "endless" list objects anymore, but rather one
3296 JSON object per line. This is more in line how most JSON
3297 parsers expect JSON objects. The new output mode
3298 "json-pretty" has been added to provide similar output, but
3299 neatly aligned for readability by humans.
3301 * We dropped all explicit sync() invocations in the shutdown
3302 code. The kernel does this implicitly anyway in the kernel
3303 reboot() syscall. halt(8)'s -n option is now a compatibility
3306 * We now support virtualized reboot() in containers, as
3307 supported by newer kernels. We will fall back to exit() if
3308 CAP_SYS_REBOOT is not available to the container. Also,
3309 nspawn makes use of this now and will actually reboot the
3310 container if the containerized OS asks for that.
3312 * journalctl will only show local log output by default
3313 now. Use --merge (-m) to show remote log output, too.
3315 * libsystemd-journal gained the new sd_journal_get_usage()
3316 call to determine the current disk usage of all journal
3317 files. This is exposed in the new "journalctl --disk-usage"
3320 * journald gained a new configuration setting SplitMode= in
3321 journald.conf which may be used to control how user journals
3322 are split off. See journald.conf(5) for details.
3324 * A new condition type ConditionFileNotEmpty= has been added.
3326 * tmpfiles' "w" lines now support file globbing, to write
3327 multiple files at once.
3329 * We added Python bindings for the journal submission
3330 APIs. More Python APIs for a number of selected APIs will
3331 likely follow. Note that we intend to add native bindings
3332 only for the Python language, as we consider it common
3333 enough to deserve bindings shipped within systemd. There are
3334 various projects outside of systemd that provide bindings
3335 for languages such as PHP or Lua.
3337 * Many conditions will now resolve specifiers such as %i. In
3338 addition, PathChanged= and related directives of .path units
3339 now support specifiers as well.
3341 * There's now a new RPM macro definition for the system preset
3344 * journald will now warn if it ca not forward a message to the
3345 syslog daemon because its socket is full.
3347 * timedated will no longer write or process /etc/timezone,
3348 except on Debian. As we do not support late mounted /usr
3349 anymore /etc/localtime always being a symlink is now safe,
3350 and hence the information in /etc/timezone is not necessary
3353 * logind will now always reserve one VT for a text getty (VT6
3354 by default). Previously if more than 6 X sessions where
3355 started they took up all the VTs with auto-spawned gettys,
3356 so that no text gettys were available anymore.
3358 * udev will now automatically inform the btrfs kernel logic
3359 about btrfs RAID components showing up. This should make
3360 simple hotplug based btrfs RAID assembly work.
3362 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3363 (but not for its children which will stay at the kernel
3364 default). This should allow setups with a lot more listening
3367 * systemd will now always pass the configured timezone to the
3368 kernel at boot. timedated will do the same when the timezone
3371 * logind's inhibition logic has been updated. By default,
3372 logind will now handle the lid switch, the power and sleep
3373 keys all the time, even in graphical sessions. If DEs want
3374 to handle these events on their own they should take the new
3375 handle-power-key, handle-sleep-key and handle-lid-switch
3376 inhibitors during their runtime. A simple way to achiveve
3377 that is to invoke the DE wrapped in an invocation of:
3379 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3381 * Access to unit operations is now checked via SELinux taking
3382 the unit file label and client process label into account.
3384 * systemd will now notify the administrator in the journal
3385 when he over-mounts a non-empty directory.
3387 * There are new specifiers that are resolved in unit files,
3388 for the host name (%H), the machine ID (%m) and the boot ID
3391 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3392 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3393 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3394 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3395 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3396 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3397 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3401 * Support for reading structured kernel messages from
3402 /dev/kmsg has now been added and is enabled by default.
3404 * Support for reading kernel messages from /proc/kmsg has now
3405 been removed. If you want kernel messages in the journal
3406 make sure to run a recent kernel (>= 3.5) that supports
3407 reading structured messages from /dev/kmsg (see
3408 above). /proc/kmsg is now exclusive property of classic
3409 syslog daemons again.
3411 * The libudev API gained the new
3412 udev_device_new_from_device_id() call.
3414 * The logic for file system namespace (ReadOnlyDirectory=,
3415 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3416 require pivot_root() anymore. This means fewer temporary
3417 directories are created below /tmp for this feature.
3419 * nspawn containers will now see and receive all submounts
3420 made on the host OS below the root file system of the
3423 * Forward Secure Sealing is now supported for Journal files,
3424 which provide cryptographical sealing of journal files so
3425 that attackers cannot alter log history anymore without this
3426 being detectable. Lennart will soon post a blog story about
3427 this explaining it in more detail.
3429 * There are two new service settings RestartPreventExitStatus=
3430 and SuccessExitStatus= which allow configuration of exit
3431 status (exit code or signal) which will be excepted from the
3432 restart logic, resp. consider successful.
3434 * journalctl gained the new --verify switch that can be used
3435 to check the integrity of the structure of journal files and
3436 (if Forward Secure Sealing is enabled) the contents of
3439 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3440 and similar symlinks pre-created. This makes running shells
3441 as container init process a lot more fun.
3443 * The fstab support can now handle PARTUUID= and PARTLABEL=
3446 * A new ConditionHost= condition has been added to match
3447 against the hostname (with globs) and machine ID. This is
3448 useful for clusters where a single OS image is used to
3449 provision a large number of hosts which shall run slightly
3450 different sets of services.
3452 * Services which hit the restart limit will now be placed in a
3455 Contributions from: Bertram Poettering, Dave Reisner, Huang
3456 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3457 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3461 * When running in --user mode systemd will now become a
3462 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3463 tree a lot more organized.
3465 * A new PartOf= unit dependency type has been introduced that
3466 may be used to group services in a natural way.
3468 * "systemctl enable" may now be used to enable instances of
3471 * journalctl now prints error log levels in red, and
3472 warning/notice log levels in bright white. It also supports
3473 filtering by log level now.
3475 * cgtop gained a new -n switch (similar to top), to configure
3476 the maximum number of iterations to run for. It also gained
3477 -b, to run in batch mode (accepting no input).
3479 * The suffix ".service" may now be omitted on most systemctl
3480 command lines involving service unit names.
3482 * There's a new bus call in logind to lock all sessions, as
3483 well as a loginctl verb for it "lock-sessions".
3485 * libsystemd-logind.so gained a new call sd_journal_perror()
3486 that works similar to libc perror() but logs to the journal
3487 and encodes structured information about the error number.
3489 * /etc/crypttab entries now understand the new keyfile-size=
3492 * shutdown(8) now can send a (configurable) wall message when
3493 a shutdown is cancelled.
3495 * The mount propagation mode for the root file system will now
3496 default to "shared", which is useful to make containers work
3497 nicely out-of-the-box so that they receive new mounts from
3498 the host. This can be undone locally by running "mount
3499 --make-rprivate /" if needed.
3501 * The prefdm.service file has been removed. Distributions
3502 should maintain this unit downstream if they intend to keep
3503 it around. However, we recommend writing normal unit files
3504 for display managers instead.
3506 * Since systemd is a crucial part of the OS we will now
3507 default to a number of compiler switches that improve
3508 security (hardening) such as read-only relocations, stack
3509 protection, and suchlike.
3511 * The TimeoutSec= setting for services is now split into
3512 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3513 of individual time outs for the start and the stop phase of
3516 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3517 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3518 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3519 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3520 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3521 Gundersen, Zbigniew Jędrzejewski-Szmek
3525 * The journal and id128 C APIs are now fully documented as man
3528 * Extra safety checks have been added when transitioning from
3529 the initial RAM disk to the main system to avoid accidental
3532 * /etc/crypttab entries now understand the new keyfile-offset=
3535 * systemctl -t can now be used to filter by unit load state.
3537 * The journal C API gained the new sd_journal_wait() call to
3538 make writing synchronous journal clients easier.
3540 * journalctl gained the new -D switch to show journals from a
3543 * journalctl now displays a special marker between log
3544 messages of two different boots.
3546 * The journal is now explicitly flushed to /var via a service
3547 systemd-journal-flush.service, rather than implicitly simply
3548 by seeing /var/log/journal to be writable.
3550 * journalctl (and the journal C APIs) can now match for much
3551 more complex expressions, with alternatives and
3554 * When transitioning from the initial RAM disk to the main
3555 system we will now kill all processes in a killing spree to
3556 ensure no processes stay around by accident.
3558 * Three new specifiers may be used in unit files: %u, %h, %s
3559 resolve to the user name, user home directory resp. user
3560 shell. This is useful for running systemd user instances.
3562 * We now automatically rotate journal files if their data
3563 object hash table gets a fill level > 75%. We also size the
3564 hash table based on the configured maximum file size. This
3565 together should lower hash collisions drastically and thus
3566 speed things up a bit.
3568 * journalctl gained the new "--header" switch to introspect
3569 header data of journal files.
3571 * A new setting SystemCallFilters= has been added to services
3572 which may be used to apply blacklists or whitelists to
3573 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3575 * nspawn gained a new --link-journal= switch (and quicker: -j)
3576 to link the container journal with the host. This makes it
3577 very easy to centralize log viewing on the host for all
3578 guests while still keeping the journal files separated.
3580 * Many bugfixes and optimizations
3582 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3583 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3584 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3589 * Several tools now understand kernel command line arguments,
3590 which are only read when run in an initial RAM disk. They
3591 usually follow closely their normal counterparts, but are
3594 * There's a new tool to analyze the readahead files that are
3595 automatically generated at boot. Use:
3597 /usr/lib/systemd/systemd-readahead analyze /.readahead
3599 * We now provide an early debug shell on tty9 if this enabled. Use:
3601 systemctl enable debug-shell.service
3603 * All plymouth related units have been moved into the Plymouth
3604 package. Please make sure to upgrade your Plymouth version
3607 * systemd-tmpfiles now supports getting passed the basename of
3608 a configuration file only, in which case it will look for it
3609 in all appropriate directories automatically.
3611 * udevadm info now takes a /dev or /sys path as argument, and
3612 does the right thing. Example:
3614 udevadm info /dev/sda
3615 udevadm info /sys/class/block/sda
3617 * systemctl now prints a warning if a unit is stopped but a
3618 unit that might trigger it continues to run. Example: a
3619 service is stopped but the socket that activates it is left
3622 * "systemctl status" will now mention if the log output was
3623 shortened due to rotation since a service has been started.
3625 * The journal API now exposes functions to determine the
3626 "cutoff" times due to rotation.
3628 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3629 immediately flushing of runtime logs to /var if possible,
3630 resp. for triggering immediate rotation of the journal
3633 * It is now considered an error if a service is attempted to
3634 be stopped that is not loaded.
3636 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3638 * systemd-analyze now supports Python 3
3640 * tmpfiles now supports cleaning up directories via aging
3641 where the first level dirs are always kept around but
3642 directories beneath it automatically aged. This is enabled
3643 by prefixing the age field with '~'.
3645 * Seat objects now expose CanGraphical, CanTTY properties
3646 which is required to deal with very fast bootups where the
3647 display manager might be running before the graphics drivers
3648 completed initialization.
3650 * Seat objects now expose a State property.
3652 * We now include RPM macros for service enabling/disabling
3653 based on the preset logic. We recommend RPM based
3654 distributions to make use of these macros if possible. This
3655 makes it simpler to reuse RPM spec files across
3658 * We now make sure that the collected systemd unit name is
3659 always valid when services log to the journal via
3662 * There's a new man page kernel-command-line(7) detailing all
3663 command line options we understand.
3665 * The fstab generator may now be disabled at boot by passing
3666 fstab=0 on the kernel command line.
3668 * A new kernel command line option modules-load= is now understood
3669 to load a specific kernel module statically, early at boot.
3671 * Unit names specified on the systemctl command line are now
3672 automatically escaped as needed. Also, if file system or
3673 device paths are specified they are automatically turned
3674 into the appropriate mount or device unit names. Example:
3676 systemctl status /home
3677 systemctl status /dev/sda
3679 * The SysVConsole= configuration option has been removed from
3680 system.conf parsing.
3682 * The SysV search path is no longer exported on the D-Bus
3685 * The Names= option is been removed from unit file parsing.
3687 * There's a new man page bootup(7) detailing the boot process.
3689 * Every unit and every generator we ship with systemd now
3690 comes with full documentation. The self-explanatory boot is
3693 * A couple of services gained "systemd-" prefixes in their
3694 name if they wrap systemd code, rather than only external
3695 code. Among them fsck@.service which is now
3696 systemd-fsck@.service.
3698 * The HaveWatchdog property has been removed from the D-Bus
3701 * systemd.confirm_spawn= on the kernel command line should now
3704 * There's a new man page crypttab(5) which details all options
3705 we actually understand.
3707 * systemd-nspawn gained a new --capability= switch to pass
3708 additional capabilities to the container.
3710 * timedated will now read known NTP implementation unit names
3711 from /usr/lib/systemd/ntp-units.d/*.list,
3712 systemd-timedated-ntp.target has been removed.
3714 * journalctl gained a new switch "-b" that lists log data of
3715 the current boot only.
3717 * The notify socket is in the abstract namespace again, in
3718 order to support daemons which chroot() at start-up.
3720 * There is a new Storage= configuration option for journald
3721 which allows configuration of where log data should go. This
3722 also provides a way to disable journal logging entirely, so
3723 that data collected is only forwarded to the console, the
3724 kernel log buffer or another syslog implementation.
3726 * Many bugfixes and optimizations
3728 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3729 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3730 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3731 Shawn Landden, Tom Gundersen
3735 * "systemctl help <unit>" now shows the man page if one is
3738 * Several new man pages have been added.
3740 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3741 MaxLevelConsole= can now be specified in
3742 journald.conf. These options allow reducing the amount of
3743 data stored on disk or forwarded by the log level.
3745 * TimerSlackNSec= can now be specified in system.conf for
3746 PID1. This allows system-wide power savings.
3748 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3749 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3754 * logind is now capable of (optionally) handling power and
3755 sleep keys as well as the lid switch.
3757 * journalctl now understands the syntax "journalctl
3758 /usr/bin/avahi-daemon" to get all log output of a specific
3761 * CapabilityBoundingSet= in system.conf now also influences
3762 the capability bound set of usermode helpers of the kernel.
3764 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3765 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3766 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3767 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3771 * Note that we skipped 139 releases here in order to set the
3772 new version to something that is greater than both udev's
3773 and systemd's most recent version number.
3775 * udev: all udev sources are merged into the systemd source tree now.
3776 All future udev development will happen in the systemd tree. It
3777 is still fully supported to use the udev daemon and tools without
3778 systemd running, like in initramfs or other init systems. Building
3779 udev though, will require the *build* of the systemd tree, but
3780 udev can be properly *run* without systemd.
3782 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3783 should be used to create dead device nodes as workarounds for broken
3786 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3787 no longer supported. udev_monitor_new_from_netlink() needs to be
3788 used to subscribe to events.
3790 * udev: when udevd is started by systemd, processes which are left
3791 behind by forking them off of udev rules, are unconditionally cleaned
3792 up and killed now after the event handling has finished. Services or
3793 daemons must be started as systemd services. Services can be
3794 pulled-in by udev to get started, but they can no longer be directly
3795 forked by udev rules.
3797 * udev: the daemon binary is called systemd-udevd now and installed
3798 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3799 to adapt to that, create symlink, or rename the binary after building
3802 * libudev no longer provides these symbols:
3803 udev_monitor_from_socket()
3804 udev_queue_get_failed_list_entry()
3805 udev_get_{dev,sys,run}_path()
3806 The versions number was bumped and symbol versioning introduced.
3808 * systemd-loginctl and systemd-journalctl have been renamed
3809 to loginctl and journalctl to match systemctl.
3811 * The config files: /etc/systemd/systemd-logind.conf and
3812 /etc/systemd/systemd-journald.conf have been renamed to
3813 logind.conf and journald.conf. Package updates should rename
3814 the files to the new names on upgrade.
3816 * For almost all files the license is now LGPL2.1+, changed
3817 from the previous GPL2.0+. Exceptions are some minor stuff
3818 of udev (which will be changed to LGPL2.1 eventually, too),
3819 and the MIT licensed sd-daemon.[ch] library that is suitable
3820 to be used as drop-in files.
3822 * systemd and logind now handle system sleep states, in
3823 particular suspending and hibernating.
3825 * logind now implements a sleep/shutdown/idle inhibiting logic
3826 suitable for a variety of uses. Soonishly Lennart will blog
3827 about this in more detail.
3829 * var-run.mount and var-lock.mount are no longer provided
3830 (which prevously bind mounted these directories to their new
3831 places). Distributions which have not converted these
3832 directories to symlinks should consider stealing these files
3833 from git history and add them downstream.
3835 * We introduced the Documentation= field for units and added
3836 this to all our shipped units. This is useful to make it
3837 easier to explore the boot and the purpose of the various
3840 * All smaller setup units (such as
3841 systemd-vconsole-setup.service) now detect properly if they
3842 are run in a container and are skipped when
3843 appropriate. This guarantees an entirely noise-free boot in
3844 Linux container environments such as systemd-nspawn.
3846 * A framework for implementing offline system updates is now
3847 integrated, for details see:
3848 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3850 * A new service type Type=idle is available now which helps us
3851 avoiding ugly interleaving of getty output and boot status
3854 * There's now a system-wide CapabilityBoundingSet= option to
3855 globally reduce the set of capabilities for the
3856 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3857 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3858 even CAP_NET_ADMIN system-wide for secure systems.
3860 * There are now system-wide DefaultLimitXXX= options to
3861 globally change the defaults of the various resource limits
3862 for all units started by PID 1.
3864 * Harald Hoyer's systemd test suite has been integrated into
3865 systemd which allows easy testing of systemd builds in qemu
3866 and nspawn. (This is really awesome! Ask us for details!)
3868 * The fstab parser is now implemented as generator, not inside
3871 * systemctl will now warn you if .mount units generated from
3872 /etc/fstab are out of date due to changes in fstab that
3873 have not been read by systemd yet.
3875 * systemd is now suitable for usage in initrds. Dracut has
3876 already been updated to make use of this. With this in place
3877 initrds get a slight bit faster but primarily are much
3878 easier to introspect and debug since "systemctl status" in
3879 the host system can be used to introspect initrd services,
3880 and the journal from the initrd is kept around too.
3882 * systemd-delta has been added, a tool to explore differences
3883 between user/admin configuration and vendor defaults.
3885 * PrivateTmp= now affects both /tmp and /var/tmp.
3887 * Boot time status messages are now much prettier and feature
3888 proper english language. Booting up systemd has never been
3891 * Read-ahead pack files now include the inode number of all
3892 files to pre-cache. When the inode changes the pre-caching
3893 is not attempted. This should be nicer to deal with updated
3894 packages which might result in changes of read-ahead
3897 * We now temporaritly lower the kernel's read_ahead_kb variable
3898 when collecting read-ahead data to ensure the kernel's
3899 built-in read-ahead does not add noise to our measurements
3900 of necessary blocks to pre-cache.
3902 * There's now RequiresMountsFor= to add automatic dependencies
3903 for all mounts necessary for a specific file system path.
3905 * MountAuto= and SwapAuto= have been removed from
3906 system.conf. Mounting file systems at boot has to take place
3909 * nspawn now learned a new switch --uuid= to set the machine
3910 ID on the command line.
3912 * nspawn now learned the -b switch to automatically search
3915 * vt102 is now the default TERM for serial TTYs, upgraded from
3918 * systemd-logind now works on VT-less systems.
3920 * The build tree has been reorganized. The individual
3921 components now have directories of their own.
3923 * A new condition type ConditionPathIsReadWrite= is now available.
3925 * nspawn learned the new -C switch to create cgroups for the
3926 container in other hierarchies.
3928 * We now have support for hardware watchdogs, configurable in
3931 * The scheduled shutdown logic now has a public API.
3933 * We now mount /tmp as tmpfs by default, but this can be
3934 masked and /etc/fstab can override it.
3936 * Since udisks does not make use of /media anymore we are not
3937 mounting a tmpfs on it anymore.
3939 * journalctl gained a new --local switch to only interleave
3940 locally generated journal files.
3942 * We can now load the IMA policy at boot automatically.
3944 * The GTK tools have been split off into a systemd-ui.
3946 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3947 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3948 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3949 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3950 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3951 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3952 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3953 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3954 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3959 * This is mostly a bugfix release
3961 * Support optional initialization of the machine ID from the
3962 KVM or container configured UUID.
3964 * Support immediate reboots with "systemctl reboot -ff"
3966 * Show /etc/os-release data in systemd-analyze output
3968 * Many bugfixes for the journal, including endianness fixes and
3969 ensuring that disk space enforcement works
3971 * sd-login.h is C++ comptaible again
3973 * Extend the /etc/os-release format on request of the Debian
3976 * We now refuse non-UTF8 strings used in various configuration
3977 and unit files. This is done to ensure we do not pass invalid
3978 data over D-Bus or expose it elsewhere.
3980 * Register Mimo USB Screens as suitable for automatic seat
3983 * Read SELinux client context from journal clients in a race
3986 * Reorder configuration file lookup order. /etc now always
3987 overrides /run in order to allow the administrator to always
3988 and unconditionally override vendor supplied or
3989 automatically generated data.
3991 * The various user visible bits of the journal now have man
3992 pages. We still lack man pages for the journal API calls
3995 * We now ship all man pages in HTML format again in the
3998 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3999 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
4000 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
4005 * This is mostly a bugfix release
4007 * systems lacking /etc/os-release are no longer supported.
4009 * Various functionality updates to libsystemd-login.so
4011 * Track class of PAM logins to distuingish greeters from
4014 Contributions from: Kay Sievers, Lennart Poettering, Michael
4019 * This is an important bugfix release for v41.
4021 * Building man pages is now optional which should be useful
4022 for those building systemd from git but unwilling to install
4025 * Watchdog support for supervising services is now usable. In
4026 a future release support for hardware watchdogs
4027 (i.e. /dev/watchdog) will be added building on this.
4029 * Service start rate limiting is now configurable and can be
4030 turned off per service. When a start rate limit is hit a
4031 reboot can automatically be triggered.
4033 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4035 Contributions from: Benjamin Franzke, Bill Nottingham,
4036 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4037 Schmidt, Michał Górny, Piotr Drąg
4041 * The systemd binary is installed /usr/lib/systemd/systemd now;
4042 An existing /sbin/init symlink needs to be adapted with the
4045 * The code that loads kernel modules has been ported to invoke
4046 libkmod directly, instead of modprobe. This means we do not
4047 support systems with module-init-tools anymore.
4049 * Watchdog support is now already useful, but still not
4052 * A new kernel command line option systemd.setenv= is
4053 understood to set system wide environment variables
4054 dynamically at boot.
4056 * We now limit the set of capabilities of systemd-journald.
4058 * We now set SIGPIPE to ignore by default, since it only is
4059 useful in shell pipelines, and has little use in general
4060 code. This can be disabled with IgnoreSIPIPE=no in unit
4063 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4064 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4069 * This is mostly a bugfix release
4071 * We now expose the reason why a service failed in the
4072 "Result" D-Bus property.
4074 * Rudimentary service watchdog support (will be completed over
4075 the next few releases.)
4077 * When systemd forks off in order execute some service we will
4078 now immediately changes its argv[0] to reflect which process
4079 it will execute. This is useful to minimize the time window
4080 with a generic argv[0], which makes bootcharts more useful
4082 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4083 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4084 Mike Kazantsev, Ray Strode
4088 * This is mostly a test release, but incorporates many
4091 * New systemd-cgtop tool to show control groups by their
4094 * Linking against libacl for ACLs is optional again. If
4095 disabled, support tracking device access for active logins
4096 goes becomes unavailable, and so does access to the user
4097 journals by the respective users.
4099 * If a group "adm" exists, journal files are automatically
4100 owned by them, thus allow members of this group full access
4101 to the system journal as well as all user journals.
4103 * The journal now stores the SELinux context of the logging
4104 client for all entries.
4106 * Add C++ inclusion guards to all public headers
4108 * New output mode "cat" in the journal to print only text
4109 messages, without any meta data like date or time.
4111 * Include tiny X server wrapper as a temporary stop-gap to
4112 teach XOrg udev display enumeration. This is used by display
4113 managers such as gdm, and will go away as soon as XOrg
4114 learned native udev hotplugging for display devices.
4116 * Add new systemd-cat tool for executing arbitrary programs
4117 with STDERR/STDOUT connected to the journal. Can also act as
4118 BSD logger replacement, and does so by default.
4120 * Optionally store all locally generated coredumps in the
4121 journal along with meta data.
4123 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4124 writing short strings to files (for usage for /sys), and for
4125 creating symlinks, character and block device nodes.
4127 * New unit file option ControlGroupPersistent= to make cgroups
4128 persistent, following the mechanisms outlined in
4129 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4131 * Support multiple local RTCs in a sane way
4133 * No longer monopolize IO when replaying readahead data on
4134 rotating disks, since we might starve non-file-system IO to
4135 death, since fanotify() will not see accesses done by blkid,
4138 * Do not show kernel threads in systemd-cgls anymore, unless
4139 requested with new -k switch.
4141 Contributions from: Dan Horák, Kay Sievers, Lennart
4142 Poettering, Michal Schmidt
4146 * This is mostly a test release, but incorporates many
4149 * The git repository moved to:
4150 git://anongit.freedesktop.org/systemd/systemd
4151 ssh://git.freedesktop.org/git/systemd/systemd
4153 * First release with the journal
4154 http://0pointer.de/blog/projects/the-journal.html
4156 * The journal replaces both systemd-kmsg-syslogd and
4157 systemd-stdout-bridge.
4159 * New sd_pid_get_unit() API call in libsystemd-logind
4161 * Many systemadm clean-ups
4163 * Introduce remote-fs-pre.target which is ordered before all
4164 remote mounts and may be used to start services before all
4167 * Added Mageia support
4169 * Add bash completion for systemd-loginctl
4171 * Actively monitor PID file creation for daemons which exit in
4172 the parent process before having finished writing the PID
4173 file in the daemon process. Daemons which do this need to be
4174 fixed (i.e. PID file creation must have finished before the
4175 parent exits), but we now react a bit more gracefully to them.
4177 * Add colourful boot output, mimicking the well-known output
4178 of existing distributions.
4180 * New option PassCredentials= for socket units, for
4181 compatibility with a recent kernel ABI breakage.
4183 * /etc/rc.local is now hooked in via a generator binary, and
4184 thus will no longer act as synchronization point during
4187 * systemctl list-unit-files now supports --root=.
4189 * systemd-tmpfiles now understands two new commands: z, Z for
4190 relabelling files according to the SELinux database. This is
4191 useful to apply SELinux labels to specific files in /sys,
4194 * Output of SysV services is now forwarded to both the console
4195 and the journal by default, not only just the console.
4197 * New man pages for all APIs from libsystemd-login.
4199 * The build tree got reorganized and a the build system is a
4200 lot more modular allowing embedded setups to specifically
4201 select the components of systemd they are interested in.
4203 * Support for Linux systems lacking the kernel VT subsystem is
4206 * configure's --with-rootdir= got renamed to
4207 --with-rootprefix= to follow the naming used by udev and
4210 * Unless specified otherwise we will now install to /usr instead
4211 of /usr/local by default.
4213 * Processes with '@' in argv[0][0] are now excluded from the
4214 final shut-down killing spree, following the logic explained
4216 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4218 * All processes remaining in a service cgroup when we enter
4219 the START or START_PRE states are now killed with
4220 SIGKILL. That means it is no longer possible to spawn
4221 background processes from ExecStart= lines (which was never
4222 supported anyway, and bad style).
4224 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4225 reloading of units together.
4227 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4228 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4229 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4230 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4231 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob