1 systemd System and Service Manager
5 * udev gained support for loading additional device properties
6 from an indexed database that is keyed by vendor/product IDs
7 and similar device identifiers. For the beginning this
8 "hwdb" is populated with data from the well-known PCI and
9 USB database, but also includes PNP, ACPI and OID data. In
10 the longer run this indexed database shall grow into
11 becoming the one central database for non-essential
12 userspace device metadata. Previously, data from the PCI/USB
13 database was only attached to select devices, since the
14 lookup was a relatively expensive operation due to O(n) time
15 complexity (with n being the number of entries in the
16 database). Since this is now O(1), we decided to add in this
17 data for all devices where this is available, by
18 default. Note that the indexed database needs to be rebuilt
19 when new data files are installed. To achieve this you need
20 to update your packaging scripts to invoke "udevadm hwdb
21 --update" after installation of hwdb data files. For
22 RPM-based distributions we introduced the new
23 %udev_hwdb_update macro for this purpose.
25 * The Journal gained support for the "Message Catalog", an
26 indexed database to link up additional information with
27 journal entries. For further details please check:
29 http://www.freedesktop.org/wiki/Software/systemd/catalog
31 The indexed message catalog database also needs to be
32 rebuilt after installation of message catalog files. Use
33 "journalctl --update-catalog" for this. For RPM-based
34 distributions we introduced the %journal_catalog_update
35 macro for this purpose.
37 * The Python Journal bindings gained support for the standard
38 Python logging framework.
40 * The Journal API gained new functions for checking whether
41 the underlying file system of a journal file is capable of
42 properly reporting file change notifications, or whether
43 applications that want to reflect journal changes "live"
44 need to recheck journal files continously in appropriate
47 * It is now possible to set the "age" field for tmpfiles
48 entries to 0, indicating that files matching this entry
49 shall always be removed when the directories are cleaned up.
51 * coredumpctl gained a new "gdb" verb which invokes gdb
52 right-away on the selected coredump.
54 * There's now support for "hybrid sleep" on kernels that
55 support this, in addition to "suspend" and "hibernate". Use
56 "systemctl hybrid-sleep" to make use of this.
58 * logind's HandleSuspendKey= setting (and related settings)
59 now gained support for a new "lock" setting to simply
60 request the screen lock on all local sessions, instead of
61 actually executing a suspend or hibernation.
63 * systemd will now mount the EFI variables file system by
66 * Socket units now gained support for configuration of the
69 * timedatectl will now output the time of the last and next
70 daylight saving change.
72 * We dropped support for various legacy and distro-specific
73 concepts, such as insserv, early-boot SysV services
74 (i.e. those for non-standard runlevels such as 'b' or 'S')
75 or ArchLinux /etc/rc.conf support. We recommend the
76 distributions who still need support this to either continue
77 to maintain the necessary patches downstream, or find a
78 different solution. (Talk to us if you have questions!)
80 * Various systemd components will now bypass PolicyKit checks
81 for root and otherwise handle properly if PolicyKit is not
82 found to be around. This should fix most issues for
83 PolicyKit-less systems. Quite frankly this should have been
84 this way since day one. It is absolutely our intention to
85 make systemd work fine on PolicyKit-less systems, and we
86 consider it a bug if something doesn't work as it should if
87 PolicyKit is not around.
89 * For embedded systems it is now possible to build udev and
90 systemd without blkid and/or kmod support.
92 * "systemctl switch-root" is now capable of switching root
93 more than once. I.e. in addition to transitions from the
94 initrd to the host OS it is now possible to transition to
95 further OS images from the host. This is useful to implement
96 offline updating tools.
98 * Various other additions have been made to the RPM macros
99 shipped with systemd. Use %udev_rules_update() after
100 installing new udev rules files. %_udevhwdbdir,
101 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
102 %_sysctldir are now available which resolve to the right
103 directories for packages to place various data files in.
105 * journalctl gained the new --full switch (in addition to
106 --all, to disable ellipsation for long messages.
108 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
109 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
110 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
111 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
112 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
113 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
114 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
115 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
116 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
120 * journalctl gained new --since= and --until= switches to
121 filter by time. It also now supports nice filtering for
122 units via --unit=/-u.
124 * Type=oneshot services may use ExecReload= and do the
127 * The journal daemon now supports time-based rotation and
128 vacuuming, in addition to the usual disk-space based
131 * The journal will now index the available field values for
132 each field name. This enables clients to show pretty drop
133 downs of available match values when filtering. The bash
134 completion of journalctl has been updated
135 accordingly. journalctl gained a new switch -F to list all
136 values a certain field takes in the journal database.
138 * More service events are now written as structured messages
139 to the journal, and made recognizable via message IDs.
141 * The timedated, localed and hostnamed mini-services which
142 previously only provided support for changing time, locale
143 and hostname settings from graphical DEs such as GNOME now
144 also have a minimal (but very useful) text-based client
145 utility each. This is probably the nicest way to changing
146 these settings from the command line now, especially since
147 it lists available options and is fully integrated with bash
150 * There's now a new tool "systemd-coredumpctl" to list and
151 extract coredumps from the journal.
153 * We now install a README each in /var/log/ and
154 /etc/rc.d/init.d explaining where the system logs and init
155 scripts went. This hopefully should help folks who go to
156 that dirs and look into the otherwise now empty void and
159 * When user-services are invoked (by systemd --user) the
160 $MANAGERPID env var is set to the PID of systemd.
162 * SIGRTMIN+24 when sent to a --user instance will now result
163 in immediate termination of systemd.
165 * gatewayd received numerous feature additions such as a
166 "follow" mode, for live syncing and filtering.
168 * browse.html now allows filtering and showing detailed
169 information on specific entries. Keyboard navigation and
170 mouse screen support has been added.
172 * gatewayd/journalctl now supports HTML5/JSON
173 Server-Sent-Events as output.
175 * The SysV init script compatibility logic will now
176 heuristically determine whether a script supports the
177 "reload" verb, and only then make this available as
180 * "systemctl status --follow" has been removed, use "journalctl
183 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
184 have been removed since they are hardly useful to be
187 * And I'd like to take the opportunity to specifically mention
188 Zbigniew for his great contributions. Zbigniew, you rock!
190 Contributions from: Andrew Eikum, Christian Hesse, Colin
191 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
192 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
193 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
194 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
195 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
196 Jędrzejewski-Szmek, Сковорода Никита Андреевич
200 * If /etc/vconsole.conf is non-existent or empty we will no
201 longer load any console font or key map at boot by
202 default. Instead the kernel defaults will be left
203 intact. This is definitely the right thing to do, as no
204 configuration should mean no configuration, and hard-coding
205 font names that are different on all archs is probably a bad
206 idea. Also, the kernel default key map and font should be
207 good enough for most cases anyway, and mostly identical to
208 the userspace fonts/key maps we previously overloaded them
209 with. If distributions want to continue to default to a
210 non-kernel font or key map they should ship a default
211 /etc/vconsole.conf with the appropriate contents.
213 Contributions from: Colin Walters, Daniel J Walsh, Dave
214 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
215 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
219 * journalctl gained a new --cursor= switch to show entries
220 starting from the specified location in the journal.
222 * We now enforce a size limit on journal entry fields exported
223 with "-o json" in journalctl. Fields larger than 4K will be
224 assigned null. This can be turned off with --all.
226 * An (optional) journal gateway daemon is now available as
227 "systemd-journal-gatewayd.service". This service provides
228 access to the journal via HTTP and JSON. This functionality
229 will be used to implement live log synchronization in both
230 pull and push modes, but has various other users too, such
231 as easy log access for debugging of embedded devices. Right
232 now it is already useful to retrieve the journal via HTTP:
234 # systemctl start systemd-journal-gatewayd.service
235 # wget http://localhost:19531/entries
237 This will download the journal contents in a
238 /var/log/messages compatible format. The same as JSON:
240 # curl -H"Accept: application/json" http://localhost:19531/entries
242 This service is also accessible via a web browser where a
243 single static HTML5 app is served that uses the JSON logic
244 to enable the user to do some basic browsing of the
245 journal. This will be extended later on. Here's an example
246 screenshot of this app in its current state:
248 http://0pointer.de/public/journal-gatewayd
250 Contributions from: Kay Sievers, Lennart Poettering, Robert
251 Milasan, Tom Gundersen
255 * The bash completion logic is now available for journalctl
258 * We don't mount the "cpuset" controller anymore together with
259 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
260 started if no parameters are assigned to it. "cpuset" hence
261 broke code that assumed it it could create "cpu" groups and
264 * journalctl -f will now subscribe to terminal size changes,
265 and line break accordingly.
267 Contributions from: Dave Reisner, Kay Sievers, Lennart
268 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
272 * nspawn will now create a symlink /etc/localtime in the
273 container environment, copying the host's timezone
274 setting. Previously this has been done via a bind mount, but
275 since symlinks cannot be bind mounted this has now been
276 changed to create/update the appropriate symlink.
278 * journalctl -n's line number argument is now optional, and
279 will default to 10 if omitted.
281 * journald will now log the maximum size the journal files may
282 take up on disk. This is particularly useful if the default
283 built-in logic of determining this parameter from the file
284 system size is used. Use "systemctl status
285 systemd-journald.service" to see this information.
287 * The multi-seat X wrapper tool has been stripped down. As X
288 is now capable of enumerating graphics devices via udev in a
289 seat-aware way the wrapper is not strictly necessary
290 anymore. A stripped down temporary stop-gap is still shipped
291 until the upstream display managers have been updated to
292 fully support the new X logic. Expect this wrapper to be
293 removed entirely in one of the next releases.
295 * HandleSleepKey= in logind.conf has been split up into
296 HandleSuspendKey= and HandleHibernateKey=. The old setting
297 is not available anymore. X11 and the kernel are
298 distuingishing between these keys and we should too. This
299 also means the inhibition lock for these keys has been split
302 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
303 Poettering, Lukas Nykryn, Václav Pavlín
307 * Whenever a unit changes state we'll now log this to the
308 journal and show along the unit's own log output in
311 * ConditionPathIsMountPoint= can now properly detect bind
312 mount points too. (Previously, a bind mount of one file
313 system to another place in the same file system could not be
314 detected as mount, since they shared struct stat's st_dev
317 * We will now mount the cgroup controllers cpu, cpuacct,
318 cpuset and the controllers net_cls, net_prio together by
321 * nspawn containers will now have a virtualized boot
322 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
323 over with a randomized ID at container initialization). This
324 has the effect of making "journalctl -b" do the right thing
327 * The JSON output journal serialization has been updated not
328 to generate "endless" list objects anymore, but rather one
329 JSON object per line. This is more in line how most JSON
330 parsers expect JSON objects. The new output mode
331 "json-pretty" has been added to provide similar output, but
332 neatly aligned for readability by humans.
334 * We dropped all explicit sync() invocations in the shutdown
335 code. The kernel does this implicitly anyway in the kernel
336 reboot() syscall. halt(8)'s -n option is now a compatibility
339 * We now support virtualized reboot() in containers, as
340 supported by newer kernels. We will fall back to exit() if
341 CAP_SYS_REBOOT is not available to the container. Also,
342 nspawn makes use of this now and will actually reboot the
343 container if the containerized OS asks for that.
345 * journalctl will only show local log output by default
346 now. Use --merge (-m) to show remote log output, too.
348 * libsystemd-journal gained the new sd_journal_get_usage()
349 call to determine the current disk usage of all journal
350 files. This is exposed in the new "journalctl --disk-usage"
353 * journald gained a new configuration setting SplitMode= in
354 journald.conf which may be used to control how user journals
355 are split off. See journald.conf(5) for details.
357 * A new condition type ConditionFileNotEmpty= has been added.
359 * tmpfiles' "w" lines now support file globbing, to write
360 multiple files at once.
362 * We added Python bindings for the journal submission
363 APIs. More Python APIs for a number of selected APIs will
364 likely follow. Note that we intend to add native bindings
365 only for the Python language, as we consider it common
366 enough to deserve bindings shipped within systemd. There are
367 various projects outside of systemd that provide bindings
368 for languages such as PHP or Lua.
370 * Many conditions will now resolve specifiers such as %i. In
371 addition, PathChanged= and related directives of .path units
372 now support specifiers as well.
374 * There's now a new RPM macro definition for the system preset
377 * journald will now warn if it can't foward a message to the
378 syslog daemon because it's socket is full.
380 * timedated will no longer write or process /etc/timezone,
381 except on Debian. As we do not support late mounted /usr
382 anymore /etc/localtime always being a symlink is now safe,
383 and hence the information in /etc/timezone is not necessary
386 * logind will now always reserve one VT for a text getty (VT6
387 by default). Previously if more than 6 X sessions where
388 started they took up all the VTs with auto-spawned gettys,
389 so that no text gettys were available anymore.
391 * udev will now automatically inform the btrfs kernel logic
392 about btrfs RAID components showing up. This should make
393 simple hotplug based btrfs RAID assembly work.
395 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
396 (but not for its children which will stay at the kernel
397 default). This should allow setups with a lot more listening
400 * systemd will now always pass the configured timezone to the
401 kernel at boot. timedated will do the same when the timezone
404 * logind's inhibition logic has been updated. By default,
405 logind will now handle the lid switch, the power and sleep
406 keys all the time, even in graphical sessions. If DEs want
407 to handle these events on their own they should take the new
408 handle-power-key, handle-sleep-key and handle-lid-switch
409 inhibitors during their runtime. A simple way to achiveve
410 that is to invoke the DE wrapped in an invocation of:
412 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
414 * Access to unit operations is now checked via SELinux taking
415 the unit file label and client process label into account.
417 * systemd will now notify the administrator in the journal
418 when he over-mounts a non-empty directory.
420 * There are new specifiers that are resolved in unit files,
421 for the host name (%H), the machine ID (%m) and the boot ID
424 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
425 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
426 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
427 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
428 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
429 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
430 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
434 * Support for reading structured kernel messages from
435 /dev/kmsg has now been added and is enabled by default.
437 * Support for reading kernel messages from /proc/kmsg has now
438 been removed. If you want kernel messages in the journal
439 make sure to run a recent kernel (>= 3.5) that supports
440 reading structured messages from /dev/kmsg (see
441 above). /proc/kmsg is now exclusive property of classic
442 syslog daemons again.
444 * The libudev API gained the new
445 udev_device_new_from_device_id() call.
447 * The logic for file system namespace (ReadOnlyDirectory=,
448 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
449 require pivot_root() anymore. This means fewer temporary
450 directories are created below /tmp for this feature.
452 * nspawn containers will now see and receive all submounts
453 made on the host OS below the root file system of the
456 * Forward Secure Sealing is now supported for Journal files,
457 which provide cryptographical sealing of journal files so
458 that attackers cannot alter log history anymore without this
459 being detectable. Lennart will soon post a blog story about
460 this explaining it in more detail.
462 * There are two new service settings RestartPreventExitStatus=
463 and SuccessExitStatus= which allow configuration of exit
464 status (exit code or signal) which will be excepted from the
465 restart logic, resp. consider successful.
467 * journalctl gained the new --verify switch that can be used
468 to check the integrity of the structure of journal files and
469 (if Forward Secure Sealing is enabled) the contents of
472 * nspawn containers will now be run with /dev/stdin, /dev/fd/
473 and similar symlinks pre-created. This makes running shells
474 as container init process a lot more fun.
476 * The fstab support can now handle PARTUUID= and PARTLABEL=
479 * A new ConditionHost= condition has been added to match
480 against the hostname (with globs) and machine ID. This is
481 useful for clusters where a single OS image is used to
482 provision a large number of hosts which shall run slightly
483 different sets of services.
485 * Services which hit the restart limit will now be placed in a
488 Contributions from: Bertram Poettering, Dave Reisner, Huang
489 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
490 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
494 * When running in --user mode systemd will now become a
495 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
496 tree a lot more organized.
498 * A new PartOf= unit dependency type has been introduced that
499 may be used to group services in a natural way.
501 * "systemctl enable" may now be used to enable instances of
504 * journalctl now prints error log levels in red, and
505 warning/notice log levels in bright white. It also supports
506 filtering by log level now.
508 * cgtop gained a new -n switch (similar to top), to configure
509 the maximum number of iterations to run for. It also gained
510 -b, to run in batch mode (accepting no input).
512 * The suffix ".service" may now be ommited on most systemctl
513 command lines involving service unit names.
515 * There's a new bus call in logind to lock all sessions, as
516 well as a loginctl verb for it "lock-sessions".
518 * libsystemd-logind.so gained a new call sd_journal_perror()
519 that works similar to libc perror() but logs to the journal
520 and encodes structured information about the error number.
522 * /etc/crypttab entries now understand the new keyfile-size=
525 * shutdown(8) now can send a (configurable) wall message when
526 a shutdown is cancelled.
528 * The mount propagation mode for the root file system will now
529 default to "shared", which is useful to make containers work
530 nicely out-of-the-box so that they receive new mounts from
531 the host. This can be undone locally by running "mount
532 --make-rprivate /" if needed.
534 * The prefdm.service file has been removed. Distributions
535 should maintain this unit downstream if they intend to keep
536 it around. However, we recommend writing normal unit files
537 for display managers instead.
539 * Since systemd is a crucial part of the OS we will now
540 default to a number of compiler switches that improve
541 security (hardening) such as read-only relocations, stack
542 protection, and suchlike.
544 * The TimeoutSec= setting for services is now split into
545 TimeoutStartSec= and TimeoutStopSec= to allow configuration
546 of individual time outs for the start and the stop phase of
549 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
550 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
551 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
552 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
553 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
554 Gundersen, Zbigniew Jędrzejewski-Szmek
558 * The journal and id128 C APIs are now fully documented as man
561 * Extra safety checks have been added when transitioning from
562 the initial RAM disk to the main system to avoid accidental
565 * /etc/crypttab entries now understand the new keyfile-offset=
568 * systemctl -t can now be used to filter by unit load state.
570 * The journal C API gained the new sd_journal_wait() call to
571 make writing synchronous journal clients easier.
573 * journalctl gained the new -D switch to show journals from a
576 * journalctl now displays a special marker between log
577 messages of two different boots.
579 * The journal is now explicitly flushed to /var via a service
580 systemd-journal-flush.service, rather than implicitly simply
581 by seeing /var/log/journal to be writable.
583 * journalctl (and the journal C APIs) can now match for much
584 more complex expressions, with alternatives and
587 * When transitioning from the initial RAM disk to the main
588 system we will now kill all processes in a killing spree to
589 ensure no processes stay around by accident.
591 * Three new specifiers may be used in unit files: %u, %h, %s
592 resolve to the user name, user home directory resp. user
593 shell. This is useful for running systemd user instances.
595 * We now automatically rotate journal files if their data
596 object hash table gets a fill level > 75%. We also size the
597 hash table based on the configured maximum file size. This
598 together should lower hash collisions drastically and thus
599 speed things up a bit.
601 * journalctl gained the new "--header" switch to introspect
602 header data of journal files.
604 * A new setting SystemCallFilters= has been added to services
605 which may be used to apply blacklists or whitelists to
606 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
608 * nspawn gained a new --link-journal= switch (and quicker: -j)
609 to link the container journal with the host. This makes it
610 very easy to centralize log viewing on the host for all
611 guests while still keeping the journal files separated.
613 * Many bugfixes and optimizations
615 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
616 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
617 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
622 * Several tools now understand kernel command line arguments,
623 which are only read when run in an initial RAM disk. They
624 usually follow closely their normal counterparts, but are
627 * There's a new tool to analyze the readahead files that are
628 automatically generated at boot. Use:
630 /usr/lib/systemd/systemd-readahead analyze /.readahead
632 * We now provide an early debug shell on tty9 if this enabled. Use:
634 systemctl enable debug-shell.service
636 * All plymouth related units have been moved into the Plymouth
637 package. Please make sure to upgrade your Plymouth version
640 * systemd-tmpfiles now supports getting passed the basename of
641 a configuration file only, in which case it will look for it
642 in all appropriate directories automatically.
644 * udevadm info now takes a /dev or /sys path as argument, and
645 does the right thing. Example:
647 udevadm info /dev/sda
648 udevadm info /sys/class/block/sda
650 * systemctl now prints a warning if a unit is stopped but a
651 unit that might trigger it continues to run. Example: a
652 service is stopped but the socket that activates it is left
655 * "systemctl status" will now mention if the log output was
656 shortened due to rotation since a service has been started.
658 * The journal API now exposes functions to determine the
659 "cutoff" times due to rotation.
661 * journald now understands SIGUSR1 and SIGUSR2 for triggering
662 immediately flushing of runtime logs to /var if possible,
663 resp. for triggering immediate rotation of the journal
666 * It is now considered an error if a service is attempted to
667 be stopped that is not loaded.
669 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
671 * systemd-analyze now supports Python 3
673 * tmpfiles now supports cleaning up directories via aging
674 where the first level dirs are always kept around but
675 directories beneath it automatically aged. This is enabled
676 by prefixing the age field with '~'.
678 * Seat objects now expose CanGraphical, CanTTY properties
679 which is required to deal with very fast bootups where the
680 display manager might be running before the graphics drivers
681 completed initialization.
683 * Seat objects now expose a State property.
685 * We now include RPM macros for service enabling/disabling
686 based on the preset logic. We recommend RPM based
687 distributions to make use of these macros if possible. This
688 makes it simpler to reuse RPM spec files across
691 * We now make sure that the collected systemd unit name is
692 always valid when services log to the journal via
695 * There's a new man page kernel-command-line(7) detailing all
696 command line options we understand.
698 * The fstab generator may now be disabled at boot by passing
699 fstab=0 on the kernel command line.
701 * A new kernel command line option modules-load= is now understood
702 to load a specific kernel module statically, early at boot.
704 * Unit names specified on the systemctl command line are now
705 automatically escaped as needed. Also, if file system or
706 device paths are specified they are automatically turned
707 into the appropriate mount or device unit names. Example:
709 systemctl status /home
710 systemctl status /dev/sda
712 * The SysVConsole= configuration option has been removed from
715 * The SysV search path is no longer exported on the D-Bus
718 * The Names= option is been removed from unit file parsing.
720 * There's a new man page bootup(7) detailing the boot process.
722 * Every unit and every generator we ship with systemd now
723 comes with full documentation. The self-explanatory boot is
726 * A couple of services gained "systemd-" prefixes in their
727 name if they wrap systemd code, rather than only external
728 code. Among them fsck@.service which is now
729 systemd-fsck@.service.
731 * The HaveWatchdog property has been removed from the D-Bus
734 * systemd.confirm_spawn= on the kernel command line should now
737 * There's a new man page crypttab(5) which details all options
738 we actually understand.
740 * systemd-nspawn gained a new --capability= switch to pass
741 additional capabilities to the container.
743 * timedated will now read known NTP implementation unit names
744 from /usr/lib/systemd/ntp-units.d/*.list,
745 systemd-timedated-ntp.target has been removed.
747 * journalctl gained a new switch "-b" that lists log data of
748 the current boot only.
750 * The notify socket is in the abstract namespace again, in
751 order to support daemons which chroot() at start-up.
753 * There is a new Storage= configuration option for journald
754 which allows configuration of where log data should go. This
755 also provides a way to disable journal logging entirely, so
756 that data collected is only forwarded to the console, the
757 kernel log buffer or another syslog implementation.
759 * Many bugfixes and optimizations
761 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
762 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
763 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
764 Shawn Landden, Tom Gundersen
768 * "systemctl help <unit>" now shows the man page if one is
771 * Several new man pages have been added.
773 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
774 MaxLevelConsole= can now be specified in
775 journald.conf. These options allow reducing the amount of
776 data stored on disk or forwarded by the log level.
778 * TimerSlackNSec= can now be specified in system.conf for
779 PID1. This allows system-wide power savings.
781 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
782 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
787 * logind is now capable of (optionally) handling power and
788 sleep keys as well as the lid switch.
790 * journalctl now understands the syntax "journalctl
791 /usr/bin/avahi-daemon" to get all log output of a specific
794 * CapabilityBoundingSet= in system.conf now also influences
795 the capability bound set of usermode helpers of the kernel.
797 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
798 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
799 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
800 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
804 * Note that we skipped 139 releases here in order to set the
805 new version to something that is greater than both udev's
806 and systemd's most recent version number.
808 * udev: all udev sources are merged into the systemd source tree now.
809 All future udev development will happen in the systemd tree. It
810 is still fully supported to use the udev daemon and tools without
811 systemd running, like in initramfs or other init systems. Building
812 udev though, will require the *build* of the systemd tree, but
813 udev can be properly *run* without systemd.
815 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
816 should be used to create dead device nodes as workarounds for broken
819 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
820 no longer supported. udev_monitor_new_from_netlink() needs to be
821 used to subscribe to events.
823 * udev: when udevd is started by systemd, processes which are left
824 behind by forking them off of udev rules, are unconditionally cleaned
825 up and killed now after the event handling has finished. Services or
826 daemons must be started as systemd services. Services can be
827 pulled-in by udev to get started, but they can no longer be directly
828 forked by udev rules.
830 * udev: the daemon binary is called systemd-udevd now and installed
831 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
832 to adapt to that, create symlink, or rename the binary after building
835 * libudev no longer provides these symbols:
836 udev_monitor_from_socket()
837 udev_queue_get_failed_list_entry()
838 udev_get_{dev,sys,run}_path()
839 The versions number was bumped and symbol versioning introduced.
841 * systemd-loginctl and systemd-journalctl have been renamed
842 to loginctl and journalctl to match systemctl.
844 * The config files: /etc/systemd/systemd-logind.conf and
845 /etc/systemd/systemd-journald.conf have been renamed to
846 logind.conf and journald.conf. Package updates should rename
847 the files to the new names on upgrade.
849 * For almost all files the license is now LGPL2.1+, changed
850 from the previous GPL2.0+. Exceptions are some minor stuff
851 of udev (which will be changed to LGPL2.1 eventually, too),
852 and the MIT licensed sd-daemon.[ch] library that is suitable
853 to be used as drop-in files.
855 * systemd and logind now handle system sleep states, in
856 particular suspending and hibernating.
858 * logind now implements a sleep/shutdown/idle inhibiting logic
859 suitable for a variety of uses. Soonishly Lennart will blog
860 about this in more detail.
862 * var-run.mount and var-lock.mount are no longer provided
863 (which prevously bind mounted these directories to their new
864 places). Distributions which have not converted these
865 directories to symlinks should consider stealing these files
866 from git history and add them downstream.
868 * We introduced the Documentation= field for units and added
869 this to all our shipped units. This is useful to make it
870 easier to explore the boot and the purpose of the various
873 * All smaller setup units (such as
874 systemd-vconsole-setup.service) now detect properly if they
875 are run in a container and are skipped when
876 appropriate. This guarantees an entirely noise-free boot in
877 Linux container environments such as systemd-nspawn.
879 * A framework for implementing offline system updates is now
880 integrated, for details see:
881 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
883 * A new service type Type=idle is available now which helps us
884 avoiding ugly interleaving of getty output and boot status
887 * There's now a system-wide CapabilityBoundingSet= option to
888 globally reduce the set of capabilities for the
889 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
890 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
891 even CAP_NET_ADMIN system-wide for secure systems.
893 * There are now system-wide DefaultLimitXXX= options to
894 globally change the defaults of the various resource limits
895 for all units started by PID 1.
897 * Harald Hoyer's systemd test suite has been integrated into
898 systemd which allows easy testing of systemd builds in qemu
899 and nspawn. (This is really awesome! Ask us for details!)
901 * The fstab parser is now implemented as generator, not inside
904 * systemctl will now warn you if .mount units generated from
905 /etc/fstab are out of date due to changes in fstab that
906 haven't been read by systemd yet.
908 * systemd is now suitable for usage in initrds. Dracut has
909 already been updated to make use of this. With this in place
910 initrds get a slight bit faster but primarily are much
911 easier to introspect and debug since "systemctl status" in
912 the host system can be used to introspect initrd services,
913 and the journal from the initrd is kept around too.
915 * systemd-delta has been added, a tool to explore differences
916 between user/admin configuration and vendor defaults.
918 * PrivateTmp= now affects both /tmp and /var/tmp.
920 * Boot time status messages are now much prettier and feature
921 proper english language. Booting up systemd has never been
924 * Read-ahead pack files now include the inode number of all
925 files to pre-cache. When the inode changes the pre-caching
926 is not attempted. This should be nicer to deal with updated
927 packages which might result in changes of read-ahead
930 * We now temporaritly lower the kernel's read_ahead_kb variable
931 when collecting read-ahead data to ensure the kernel's
932 built-in read-ahead does not add noise to our measurements
933 of necessary blocks to pre-cache.
935 * There's now RequiresMountsFor= to add automatic dependencies
936 for all mounts necessary for a specific file system path.
938 * MountAuto= and SwapAuto= have been removed from
939 system.conf. Mounting file systems at boot has to take place
942 * nspawn now learned a new switch --uuid= to set the machine
943 ID on the command line.
945 * nspawn now learned the -b switch to automatically search
948 * vt102 is now the default TERM for serial TTYs, upgraded from
951 * systemd-logind now works on VT-less systems.
953 * The build tree has been reorganized. The individual
954 components now have directories of their own.
956 * A new condition type ConditionPathIsReadWrite= is now available.
958 * nspawn learned the new -C switch to create cgroups for the
959 container in other hierarchies.
961 * We now have support for hardware watchdogs, configurable in
964 * The scheduled shutdown logic now has a public API.
966 * We now mount /tmp as tmpfs by default, but this can be
967 masked and /etc/fstab can override it.
969 * Since udisks doesn't make use of /media anymore we are not
970 mounting a tmpfs on it anymore.
972 * journalctl gained a new --local switch to only interleave
973 locally generated journal files.
975 * We can now load the IMA policy at boot automatically.
977 * The GTK tools have been split off into a systemd-ui.
979 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
980 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
981 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
982 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
983 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
984 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
985 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
986 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
987 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
992 * This is mostly a bugfix release
994 * Support optional initialization of the machine ID from the
995 KVM or container configured UUID.
997 * Support immediate reboots with "systemctl reboot -ff"
999 * Show /etc/os-release data in systemd-analyze output
1001 * Many bugfixes for the journal, including endianess fixes and
1002 ensuring that disk space enforcement works
1004 * sd-login.h is C++ comptaible again
1006 * Extend the /etc/os-release format on request of the Debian
1009 * We now refuse non-UTF8 strings used in various configuration
1010 and unit files. This is done to ensure we don't pass invalid
1011 data over D-Bus or expose it elsewhere.
1013 * Register Mimo USB Screens as suitable for automatic seat
1016 * Read SELinux client context from journal clients in a race
1019 * Reorder configuration file lookup order. /etc now always
1020 overrides /run in order to allow the administrator to always
1021 and unconditionally override vendor supplied or
1022 automatically generated data.
1024 * The various user visible bits of the journal now have man
1025 pages. We still lack man pages for the journal API calls
1028 * We now ship all man pages in HTML format again in the
1031 Contributions from: Dave Reisner, Dirk Eibach, Frederic
1032 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
1033 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
1038 * This is mostly a bugfix release
1040 * systems lacking /etc/os-release are no longer supported.
1042 * Various functionality updates to libsystemd-login.so
1044 * Track class of PAM logins to distuingish greeters from
1047 Contributions from: Kay Sievers, Lennart Poettering, Michael
1052 * This is an important bugfix release for v41.
1054 * Building man pages is now optional which should be useful
1055 for those building systemd from git but unwilling to install
1058 * Watchdog support for supervising services is now usable. In
1059 a future release support for hardware watchdogs
1060 (i.e. /dev/watchdog) will be added building on this.
1062 * Service start rate limiting is now configurable and can be
1063 turned off per service. When a start rate limit is hit a
1064 reboot can automatically be triggered.
1066 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
1068 Contributions from: Benjamin Franzke, Bill Nottingham,
1069 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
1070 Schmidt, Michał Górny, Piotr Drąg
1074 * The systemd binary is installed /usr/lib/systemd/systemd now;
1075 An existing /sbin/init symlink needs to be adapted with the
1078 * The code that loads kernel modules has been ported to invoke
1079 libkmod directly, instead of modprobe. This means we do not
1080 support systems with module-init-tools anymore.
1082 * Watchdog support is now already useful, but still not
1085 * A new kernel command line option systemd.setenv= is
1086 understood to set system wide environment variables
1087 dynamically at boot.
1089 * We now limit the set of capabilities of systemd-journald.
1091 * We now set SIGPIPE to ignore by default, since it only is
1092 useful in shell pipelines, and has little use in general
1093 code. This can be disabled with IgnoreSIPIPE=no in unit
1096 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
1097 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
1102 * This is mostly a bugfix release
1104 * We now expose the reason why a service failed in the
1105 "Result" D-Bus property.
1107 * Rudimentary service watchdog support (will be completed over
1108 the next few releases.)
1110 * When systemd forks off in order execute some service we will
1111 now immediately changes its argv[0] to reflect which process
1112 it will execute. This is useful to minimize the time window
1113 with a generic argv[0], which makes bootcharts more useful
1115 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
1116 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
1117 Mike Kazantsev, Ray Strode
1121 * This is mostly a test release, but incorporates many
1124 * New systemd-cgtop tool to show control groups by their
1127 * Linking against libacl for ACLs is optional again. If
1128 disabled, support tracking device access for active logins
1129 goes becomes unavailable, and so does access to the user
1130 journals by the respective users.
1132 * If a group "adm" exists, journal files are automatically
1133 owned by them, thus allow members of this group full access
1134 to the system journal as well as all user journals.
1136 * The journal now stores the SELinux context of the logging
1137 client for all entries.
1139 * Add C++ inclusion guards to all public headers
1141 * New output mode "cat" in the journal to print only text
1142 messages, without any meta data like date or time.
1144 * Include tiny X server wrapper as a temporary stop-gap to
1145 teach XOrg udev display enumeration. This is used by display
1146 managers such as gdm, and will go away as soon as XOrg
1147 learned native udev hotplugging for display devices.
1149 * Add new systemd-cat tool for executing arbitrary programs
1150 with STDERR/STDOUT connected to the journal. Can also act as
1151 BSD logger replacement, and does so by default.
1153 * Optionally store all locally generated coredumps in the
1154 journal along with meta data.
1156 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
1157 writing short strings to files (for usage for /sys), and for
1158 creating symlinks, character and block device nodes.
1160 * New unit file option ControlGroupPersistent= to make cgroups
1161 persistent, following the mechanisms outlined in
1162 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
1164 * Support multiple local RTCs in a sane way
1166 * No longer monopolize IO when replaying readahead data on
1167 rotating disks, since we might starve non-file-system IO to
1168 death, since fanotify() will not see accesses done by blkid,
1171 * Don't show kernel threads in systemd-cgls anymore, unless
1172 requested with new -k switch.
1174 Contributions from: Dan Horák, Kay Sievers, Lennart
1175 Poettering, Michal Schmidt
1179 * This is mostly a test release, but incorporates many
1182 * The git repository moved to:
1183 git://anongit.freedesktop.org/systemd/systemd
1184 ssh://git.freedesktop.org/git/systemd/systemd
1186 * First release with the journal
1187 http://0pointer.de/blog/projects/the-journal.html
1189 * The journal replaces both systemd-kmsg-syslogd and
1190 systemd-stdout-bridge.
1192 * New sd_pid_get_unit() API call in libsystemd-logind
1194 * Many systemadm clean-ups
1196 * Introduce remote-fs-pre.target which is ordered before all
1197 remote mounts and may be used to start services before all
1200 * Added Mageia support
1202 * Add bash completion for systemd-loginctl
1204 * Actively monitor PID file creation for daemons which exit in
1205 the parent process before having finished writing the PID
1206 file in the daemon process. Daemons which do this need to be
1207 fixed (i.e. PID file creation must have finished before the
1208 parent exits), but we now react a bit more gracefully to them.
1210 * Add colourful boot output, mimicking the well-known output
1211 of existing distributions.
1213 * New option PassCredentials= for socket units, for
1214 compatibility with a recent kernel ABI breakage.
1216 * /etc/rc.local is now hooked in via a generator binary, and
1217 thus will no longer act as synchronization point during
1220 * systemctl list-unit-files now supports --root=.
1222 * systemd-tmpfiles now understands two new commands: z, Z for
1223 relabelling files according to the SELinux database. This is
1224 useful to apply SELinux labels to specific files in /sys,
1227 * Output of SysV services is now forwarded to both the console
1228 and the journal by default, not only just the console.
1230 * New man pages for all APIs from libsystemd-login.
1232 * The build tree got reorganized and a the build system is a
1233 lot more modular allowing embedded setups to specifically
1234 select the components of systemd they are interested in.
1236 * Support for Linux systems lacking the kernel VT subsystem is
1239 * configure's --with-rootdir= got renamed to
1240 --with-rootprefix= to follow the naming used by udev and
1243 * Unless specified otherwise we'll now install to /usr instead
1244 of /usr/local by default.
1246 * Processes with '@' in argv[0][0] are now excluded from the
1247 final shut-down killing spree, following the logic explained
1249 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
1251 * All processes remaining in a service cgroup when we enter
1252 the START or START_PRE states are now killed with
1253 SIGKILL. That means it is no longer possible to spawn
1254 background processes from ExecStart= lines (which was never
1255 supported anyway, and bad style).
1257 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
1258 reloading of units together.
1260 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
1261 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
1262 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
1263 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
1264 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek