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 the select devices, since the
14 lookup was a relatively expensive operation due to O(n) time
15 complexity. Since this is now O(log(n)), we decided to add
16 in this data for all devices where this is available, by
17 default. Note that the indexed database needs to be rebuilt
18 when new data files are installed. To achieve this you need
19 to update your packaging scripts to invoke "udevadm hwdb
20 --update" after installation of hwdb data files. For
21 RPM-based distributions we introduced the new
22 %udev_hwdb_update macro for this purpose.
24 * The Journal gained support for the "Message Catalog", an
25 indexed database to link up additional information with
26 journal entries. For further details please check:
28 http://www.freedesktop.org/wiki/Software/systemd/catalog
30 The indexed message catalog database also needs to be
31 rebuilt after installation of message catalog files. Use
32 "journalctl --update-catalog" for this. For RPM-based
33 distributions we introduced the %journal_catalog_update
34 macro for this purpose.
36 * The Python Journal bindings gained support for the standard
37 Python logging framework.
39 * The Journal API gained new functions for checking whether
40 the underlying file system of a journal file is capable of
41 properly reporting file change notifications, or whether
42 applications that want to reflect journal changes "live"
43 need to recheck journal files continously in appropriate
46 * It is now possible to set the "age" field for tmpfiles
47 entries to 0, indicating that files matching this entry
48 shall always be removed when the directories are cleaned up.
50 * coredumpctl gained a new "gdb" verb which invokes gdb
51 right-away on the selected coredump.
53 * There's now support for "hybrid sleep" on kernels that
54 support this, in addition to "suspend" and "hibernate". Use
55 "systemctl hybrid-sleep" to make use of this.
57 * logind's HandleSuspendKey= setting (and related settings)
58 now gained support for a new "lock" setting to simply
59 request the screen lock on all local sessions, instead of
60 actually executing a suspend or hibernation.
62 * systemd will now mount the EFI variables file system by
65 * Socket units now gained support for configuration of the
68 * timedatectl will now output the time of the last and next
69 daylight saving change.
71 * We dropped support for various legacy and distro-specific
72 concepts, such as insserv, early-boot SysV services
73 (i.e. those for non-standard runlevels such as 'b' or 'S')
74 or ArchLinux /etc/rc.conf support. We recommend the
75 distributions who still need support this to either continue
76 to maintain the necessary patches downstream, or find a
77 different solution. (Talk to us if you have questions!)
79 * Various systemd components will now bypass PolicyKit checks
80 for root and otherwise handle properly if PolicyKit is not
81 found to be around. This should fix most issues for
82 PolicyKit-less systems. Quite frankly this should have been
83 this way since day one. It is absolutely our intention to
84 make systemd work fine on PolicyKit-less systems, and we
85 consider it a bug if something doesn't work as it should if
86 PolicyKit is not around.
88 * For embedded systems it is now possible to build udev and
89 systemd without blkid and/or kmod support.
91 * "systemctl switch-root" is now capable of switching root
92 more than once. I.e. in addition to transitions from the
93 initrd to the host OS it is now possible to transition to
94 further OS images from the host. This is useful to implement
95 offline updating tools.
97 * Various other additions have been made to the RPM macros
98 shipped with systemd. Use %udev_rules_update() after
99 installing new udev rules files. %_udevhwdbdir,
100 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
101 %_sysctldir are now available which resolve to the right
102 directories for packages to place various data files in.
104 * journalctl gained the new --full switch (in addition to
105 --all, to disable ellipsation for long messages.
107 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
108 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
109 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
110 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
111 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
112 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
113 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
114 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
115 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
119 * journalctl gained new --since= and --until= switches to
120 filter by time. It also now supports nice filtering for
121 units via --unit=/-u.
123 * Type=oneshot services may use ExecReload= and do the
126 * The journal daemon now supports time-based rotation and
127 vacuuming, in addition to the usual disk-space based
130 * The journal will now index the available field values for
131 each field name. This enables clients to show pretty drop
132 downs of available match values when filtering. The bash
133 completion of journalctl has been updated
134 accordingly. journalctl gained a new switch -F to list all
135 values a certain field takes in the journal database.
137 * More service events are now written as structured messages
138 to the journal, and made recognizable via message IDs.
140 * The timedated, localed and hostnamed mini-services which
141 previously only provided support for changing time, locale
142 and hostname settings from graphical DEs such as GNOME now
143 also have a minimal (but very useful) text-based client
144 utility each. This is probably the nicest way to changing
145 these settings from the command line now, especially since
146 it lists available options and is fully integrated with bash
149 * There's now a new tool "systemd-coredumpctl" to list and
150 extract coredumps from the journal.
152 * We now install a README each in /var/log/ and
153 /etc/rc.d/init.d explaining where the system logs and init
154 scripts went. This hopefully should help folks who go to
155 that dirs and look into the otherwise now empty void and
158 * When user-services are invoked (by systemd --user) the
159 $MANAGERPID env var is set to the PID of systemd.
161 * SIGRTMIN+24 when sent to a --user instance will now result
162 in immediate termination of systemd.
164 * gatewayd received numerous feature additions such as a
165 "follow" mode, for live syncing and filtering.
167 * browse.html now allows filtering and showing detailed
168 information on specific entries. Keyboard navigation and
169 mouse screen support has been added.
171 * gatewayd/journalctl now supports HTML5/JSON
172 Server-Sent-Events as output.
174 * The SysV init script compatibility logic will now
175 heuristically determine whether a script supports the
176 "reload" verb, and only then make this available as
179 * "systemctl status --follow" has been removed, use "journalctl
182 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
183 have been removed since they are hardly useful to be
186 * And I'd like to take the opportunity to specifically mention
187 Zbigniew for his great contributions. Zbigniew, you rock!
189 Contributions from: Andrew Eikum, Christian Hesse, Colin
190 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
191 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
192 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
193 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
194 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
195 Jędrzejewski-Szmek, Сковорода Никита Андреевич
199 * If /etc/vconsole.conf is non-existent or empty we will no
200 longer load any console font or key map at boot by
201 default. Instead the kernel defaults will be left
202 intact. This is definitely the right thing to do, as no
203 configuration should mean no configuration, and hard-coding
204 font names that are different on all archs is probably a bad
205 idea. Also, the kernel default key map and font should be
206 good enough for most cases anyway, and mostly identical to
207 the userspace fonts/key maps we previously overloaded them
208 with. If distributions want to continue to default to a
209 non-kernel font or key map they should ship a default
210 /etc/vconsole.conf with the appropriate contents.
212 Contributions from: Colin Walters, Daniel J Walsh, Dave
213 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
214 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
218 * journalctl gained a new --cursor= switch to show entries
219 starting from the specified location in the journal.
221 * We now enforce a size limit on journal entry fields exported
222 with "-o json" in journalctl. Fields larger than 4K will be
223 assigned null. This can be turned off with --all.
225 * An (optional) journal gateway daemon is now available as
226 "systemd-journal-gatewayd.service". This service provides
227 access to the journal via HTTP and JSON. This functionality
228 will be used to implement live log synchronization in both
229 pull and push modes, but has various other users too, such
230 as easy log access for debugging of embedded devices. Right
231 now it is already useful to retrieve the journal via HTTP:
233 # systemctl start systemd-journal-gatewayd.service
234 # wget http://localhost:19531/entries
236 This will download the journal contents in a
237 /var/log/messages compatible format. The same as JSON:
239 # curl -H"Accept: application/json" http://localhost:19531/entries
241 This service is also accessible via a web browser where a
242 single static HTML5 app is served that uses the JSON logic
243 to enable the user to do some basic browsing of the
244 journal. This will be extended later on. Here's an example
245 screenshot of this app in its current state:
247 http://0pointer.de/public/journal-gatewayd
249 Contributions from: Kay Sievers, Lennart Poettering, Robert
250 Milasan, Tom Gundersen
254 * The bash completion logic is now available for journalctl
257 * We don't mount the "cpuset" controller anymore together with
258 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
259 started if no parameters are assigned to it. "cpuset" hence
260 broke code that assumed it it could create "cpu" groups and
263 * journalctl -f will now subscribe to terminal size changes,
264 and line break accordingly.
266 Contributions from: Dave Reisner, Kay Sievers, Lennart
267 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
271 * nspawn will now create a symlink /etc/localtime in the
272 container environment, copying the host's timezone
273 setting. Previously this has been done via a bind mount, but
274 since symlinks cannot be bind mounted this has now been
275 changed to create/update the appropriate symlink.
277 * journalctl -n's line number argument is now optional, and
278 will default to 10 if omitted.
280 * journald will now log the maximum size the journal files may
281 take up on disk. This is particularly useful if the default
282 built-in logic of determining this parameter from the file
283 system size is used. Use "systemctl status
284 systemd-journald.service" to see this information.
286 * The multi-seat X wrapper tool has been stripped down. As X
287 is now capable of enumerating graphics devices via udev in a
288 seat-aware way the wrapper is not strictly necessary
289 anymore. A stripped down temporary stop-gap is still shipped
290 until the upstream display managers have been updated to
291 fully support the new X logic. Expect this wrapper to be
292 removed entirely in one of the next releases.
294 * HandleSleepKey= in logind.conf has been split up into
295 HandleSuspendKey= and HandleHibernateKey=. The old setting
296 is not available anymore. X11 and the kernel are
297 distuingishing between these keys and we should too. This
298 also means the inhibition lock for these keys has been split
301 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
302 Poettering, Lukas Nykryn, Václav Pavlín
306 * Whenever a unit changes state we'll now log this to the
307 journal and show along the unit's own log output in
310 * ConditionPathIsMountPoint= can now properly detect bind
311 mount points too. (Previously, a bind mount of one file
312 system to another place in the same file system could not be
313 detected as mount, since they shared struct stat's st_dev
316 * We will now mount the cgroup controllers cpu, cpuacct,
317 cpuset and the controllers net_cls, net_prio together by
320 * nspawn containers will now have a virtualized boot
321 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
322 over with a randomized ID at container initialization). This
323 has the effect of making "journalctl -b" do the right thing
326 * The JSON output journal serialization has been updated not
327 to generate "endless" list objects anymore, but rather one
328 JSON object per line. This is more in line how most JSON
329 parsers expect JSON objects. The new output mode
330 "json-pretty" has been added to provide similar output, but
331 neatly aligned for readability by humans.
333 * We dropped all explicit sync() invocations in the shutdown
334 code. The kernel does this implicitly anyway in the kernel
335 reboot() syscall. halt(8)'s -n option is now a compatibility
338 * We now support virtualized reboot() in containers, as
339 supported by newer kernels. We will fall back to exit() if
340 CAP_SYS_REBOOT is not available to the container. Also,
341 nspawn makes use of this now and will actually reboot the
342 container if the containerized OS asks for that.
344 * journalctl will only show local log output by default
345 now. Use --merge (-m) to show remote log output, too.
347 * libsystemd-journal gained the new sd_journal_get_usage()
348 call to determine the current disk usage of all journal
349 files. This is exposed in the new "journalctl --disk-usage"
352 * journald gained a new configuration setting SplitMode= in
353 journald.conf which may be used to control how user journals
354 are split off. See journald.conf(5) for details.
356 * A new condition type ConditionFileNotEmpty= has been added.
358 * tmpfiles' "w" lines now support file globbing, to write
359 multiple files at once.
361 * We added Python bindings for the journal submission
362 APIs. More Python APIs for a number of selected APIs will
363 likely follow. Note that we intend to add native bindings
364 only for the Python language, as we consider it common
365 enough to deserve bindings shipped within systemd. There are
366 various projects outside of systemd that provide bindings
367 for languages such as PHP or Lua.
369 * Many conditions will now resolve specifiers such as %i. In
370 addition, PathChanged= and related directives of .path units
371 now support specifiers as well.
373 * There's now a new RPM macro definition for the system preset
376 * journald will now warn if it can't foward a message to the
377 syslog daemon because it's socket is full.
379 * timedated will no longer write or process /etc/timezone,
380 except on Debian. As we do not support late mounted /usr
381 anymore /etc/localtime always being a symlink is now safe,
382 and hence the information in /etc/timezone is not necessary
385 * logind will now always reserve one VT for a text getty (VT6
386 by default). Previously if more than 6 X sessions where
387 started they took up all the VTs with auto-spawned gettys,
388 so that no text gettys were available anymore.
390 * udev will now automatically inform the btrfs kernel logic
391 about btrfs RAID components showing up. This should make
392 simple hotplug based btrfs RAID assembly work.
394 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
395 (but not for its children which will stay at the kernel
396 default). This should allow setups with a lot more listening
399 * systemd will now always pass the configured timezone to the
400 kernel at boot. timedated will do the same when the timezone
403 * logind's inhibition logic has been updated. By default,
404 logind will now handle the lid switch, the power and sleep
405 keys all the time, even in graphical sessions. If DEs want
406 to handle these events on their own they should take the new
407 handle-power-key, handle-sleep-key and handle-lid-switch
408 inhibitors during their runtime. A simple way to achiveve
409 that is to invoke the DE wrapped in an invocation of:
411 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
413 * Access to unit operations is now checked via SELinux taking
414 the unit file label and client process label into account.
416 * systemd will now notify the administrator in the journal
417 when he over-mounts a non-empty directory.
419 * There are new specifiers that are resolved in unit files,
420 for the host name (%H), the machine ID (%m) and the boot ID
423 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
424 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
425 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
426 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
427 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
428 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
429 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
433 * Support for reading structured kernel messages from
434 /dev/kmsg has now been added and is enabled by default.
436 * Support for reading kernel messages from /proc/kmsg has now
437 been removed. If you want kernel messages in the journal
438 make sure to run a recent kernel (>= 3.5) that supports
439 reading structured messages from /dev/kmsg (see
440 above). /proc/kmsg is now exclusive property of classic
441 syslog daemons again.
443 * The libudev API gained the new
444 udev_device_new_from_device_id() call.
446 * The logic for file system namespace (ReadOnlyDirectory=,
447 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
448 require pivot_root() anymore. This means fewer temporary
449 directories are created below /tmp for this feature.
451 * nspawn containers will now see and receive all submounts
452 made on the host OS below the root file system of the
455 * Forward Secure Sealing is now supported for Journal files,
456 which provide cryptographical sealing of journal files so
457 that attackers cannot alter log history anymore without this
458 being detectable. Lennart will soon post a blog story about
459 this explaining it in more detail.
461 * There are two new service settings RestartPreventExitStatus=
462 and SuccessExitStatus= which allow configuration of exit
463 status (exit code or signal) which will be excepted from the
464 restart logic, resp. consider successful.
466 * journalctl gained the new --verify switch that can be used
467 to check the integrity of the structure of journal files and
468 (if Forward Secure Sealing is enabled) the contents of
471 * nspawn containers will now be run with /dev/stdin, /dev/fd/
472 and similar symlinks pre-created. This makes running shells
473 as container init process a lot more fun.
475 * The fstab support can now handle PARTUUID= and PARTLABEL=
478 * A new ConditionHost= condition has been added to match
479 against the hostname (with globs) and machine ID. This is
480 useful for clusters where a single OS image is used to
481 provision a large number of hosts which shall run slightly
482 different sets of services.
484 * Services which hit the restart limit will now be placed in a
487 Contributions from: Bertram Poettering, Dave Reisner, Huang
488 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
489 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
493 * When running in --user mode systemd will now become a
494 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
495 tree a lot more organized.
497 * A new PartOf= unit dependency type has been introduced that
498 may be used to group services in a natural way.
500 * "systemctl enable" may now be used to enable instances of
503 * journalctl now prints error log levels in red, and
504 warning/notice log levels in bright white. It also supports
505 filtering by log level now.
507 * cgtop gained a new -n switch (similar to top), to configure
508 the maximum number of iterations to run for. It also gained
509 -b, to run in batch mode (accepting no input).
511 * The suffix ".service" may now be ommited on most systemctl
512 command lines involving service unit names.
514 * There's a new bus call in logind to lock all sessions, as
515 well as a loginctl verb for it "lock-sessions".
517 * libsystemd-logind.so gained a new call sd_journal_perror()
518 that works similar to libc perror() but logs to the journal
519 and encodes structured information about the error number.
521 * /etc/crypttab entries now understand the new keyfile-size=
524 * shutdown(8) now can send a (configurable) wall message when
525 a shutdown is cancelled.
527 * The mount propagation mode for the root file system will now
528 default to "shared", which is useful to make containers work
529 nicely out-of-the-box so that they receive new mounts from
530 the host. This can be undone locally by running "mount
531 --make-rprivate /" if needed.
533 * The prefdm.service file has been removed. Distributions
534 should maintain this unit downstream if they intend to keep
535 it around. However, we recommend writing normal unit files
536 for display managers instead.
538 * Since systemd is a crucial part of the OS we will now
539 default to a number of compiler switches that improve
540 security (hardening) such as read-only relocations, stack
541 protection, and suchlike.
543 * The TimeoutSec= setting for services is now split into
544 TimeoutStartSec= and TimeoutStopSec= to allow configuration
545 of individual time outs for the start and the stop phase of
548 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
549 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
550 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
551 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
552 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
553 Gundersen, Zbigniew Jędrzejewski-Szmek
557 * The journal and id128 C APIs are now fully documented as man
560 * Extra safety checks have been added when transitioning from
561 the initial RAM disk to the main system to avoid accidental
564 * /etc/crypttab entries now understand the new keyfile-offset=
567 * systemctl -t can now be used to filter by unit load state.
569 * The journal C API gained the new sd_journal_wait() call to
570 make writing synchronous journal clients easier.
572 * journalctl gained the new -D switch to show journals from a
575 * journalctl now displays a special marker between log
576 messages of two different boots.
578 * The journal is now explicitly flushed to /var via a service
579 systemd-journal-flush.service, rather than implicitly simply
580 by seeing /var/log/journal to be writable.
582 * journalctl (and the journal C APIs) can now match for much
583 more complex expressions, with alternatives and
586 * When transitioning from the initial RAM disk to the main
587 system we will now kill all processes in a killing spree to
588 ensure no processes stay around by accident.
590 * Three new specifiers may be used in unit files: %u, %h, %s
591 resolve to the user name, user home directory resp. user
592 shell. This is useful for running systemd user instances.
594 * We now automatically rotate journal files if their data
595 object hash table gets a fill level > 75%. We also size the
596 hash table based on the configured maximum file size. This
597 together should lower hash collisions drastically and thus
598 speed things up a bit.
600 * journalctl gained the new "--header" switch to introspect
601 header data of journal files.
603 * A new setting SystemCallFilters= has been added to services
604 which may be used to apply blacklists or whitelists to
605 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
607 * nspawn gained a new --link-journal= switch (and quicker: -j)
608 to link the container journal with the host. This makes it
609 very easy to centralize log viewing on the host for all
610 guests while still keeping the journal files separated.
612 * Many bugfixes and optimizations
614 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
615 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
616 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
621 * Several tools now understand kernel command line arguments,
622 which are only read when run in an initial RAM disk. They
623 usually follow closely their normal counterparts, but are
626 * There's a new tool to analyze the readahead files that are
627 automatically generated at boot. Use:
629 /usr/lib/systemd/systemd-readahead analyze /.readahead
631 * We now provide an early debug shell on tty9 if this enabled. Use:
633 systemctl enable debug-shell.service
635 * All plymouth related units have been moved into the Plymouth
636 package. Please make sure to upgrade your Plymouth version
639 * systemd-tmpfiles now supports getting passed the basename of
640 a configuration file only, in which case it will look for it
641 in all appropriate directories automatically.
643 * udevadm info now takes a /dev or /sys path as argument, and
644 does the right thing. Example:
646 udevadm info /dev/sda
647 udevadm info /sys/class/block/sda
649 * systemctl now prints a warning if a unit is stopped but a
650 unit that might trigger it continues to run. Example: a
651 service is stopped but the socket that activates it is left
654 * "systemctl status" will now mention if the log output was
655 shortened due to rotation since a service has been started.
657 * The journal API now exposes functions to determine the
658 "cutoff" times due to rotation.
660 * journald now understands SIGUSR1 and SIGUSR2 for triggering
661 immediately flushing of runtime logs to /var if possible,
662 resp. for triggering immediate rotation of the journal
665 * It is now considered an error if a service is attempted to
666 be stopped that is not loaded.
668 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
670 * systemd-analyze now supports Python 3
672 * tmpfiles now supports cleaning up directories via aging
673 where the first level dirs are always kept around but
674 directories beneath it automatically aged. This is enabled
675 by prefixing the age field with '~'.
677 * Seat objects now expose CanGraphical, CanTTY properties
678 which is required to deal with very fast bootups where the
679 display manager might be running before the graphics drivers
680 completed initialization.
682 * Seat objects now expose a State property.
684 * We now include RPM macros for service enabling/disabling
685 based on the preset logic. We recommend RPM based
686 distributions to make use of these macros if possible. This
687 makes it simpler to reuse RPM spec files across
690 * We now make sure that the collected systemd unit name is
691 always valid when services log to the journal via
694 * There's a new man page kernel-command-line(7) detailing all
695 command line options we understand.
697 * The fstab generator may now be disabled at boot by passing
698 fstab=0 on the kernel command line.
700 * A new kernel command line option modules-load= is now understood
701 to load a specific kernel module statically, early at boot.
703 * Unit names specified on the systemctl command line are now
704 automatically escaped as needed. Also, if file system or
705 device paths are specified they are automatically turned
706 into the appropriate mount or device unit names. Example:
708 systemctl status /home
709 systemctl status /dev/sda
711 * The SysVConsole= configuration option has been removed from
714 * The SysV search path is no longer exported on the D-Bus
717 * The Names= option is been removed from unit file parsing.
719 * There's a new man page bootup(7) detailing the boot process.
721 * Every unit and every generator we ship with systemd now
722 comes with full documentation. The self-explanatory boot is
725 * A couple of services gained "systemd-" prefixes in their
726 name if they wrap systemd code, rather than only external
727 code. Among them fsck@.service which is now
728 systemd-fsck@.service.
730 * The HaveWatchdog property has been removed from the D-Bus
733 * systemd.confirm_spawn= on the kernel command line should now
736 * There's a new man page crypttab(5) which details all options
737 we actually understand.
739 * systemd-nspawn gained a new --capability= switch to pass
740 additional capabilities to the container.
742 * timedated will now read known NTP implementation unit names
743 from /usr/lib/systemd/ntp-units.d/*.list,
744 systemd-timedated-ntp.target has been removed.
746 * journalctl gained a new switch "-b" that lists log data of
747 the current boot only.
749 * The notify socket is in the abstract namespace again, in
750 order to support daemons which chroot() at start-up.
752 * There is a new Storage= configuration option for journald
753 which allows configuration of where log data should go. This
754 also provides a way to disable journal logging entirely, so
755 that data collected is only forwarded to the console, the
756 kernel log buffer or another syslog implementation.
758 * Many bugfixes and optimizations
760 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
761 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
762 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
763 Shawn Landden, Tom Gundersen
767 * "systemctl help <unit>" now shows the man page if one is
770 * Several new man pages have been added.
772 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
773 MaxLevelConsole= can now be specified in
774 journald.conf. These options allow reducing the amount of
775 data stored on disk or forwarded by the log level.
777 * TimerSlackNSec= can now be specified in system.conf for
778 PID1. This allows system-wide power savings.
780 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
781 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
786 * logind is now capable of (optionally) handling power and
787 sleep keys as well as the lid switch.
789 * journalctl now understands the syntax "journalctl
790 /usr/bin/avahi-daemon" to get all log output of a specific
793 * CapabilityBoundingSet= in system.conf now also influences
794 the capability bound set of usermode helpers of the kernel.
796 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
797 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
798 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
799 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
803 * Note that we skipped 139 releases here in order to set the
804 new version to something that is greater than both udev's
805 and systemd's most recent version number.
807 * udev: all udev sources are merged into the systemd source tree now.
808 All future udev development will happen in the systemd tree. It
809 is still fully supported to use the udev daemon and tools without
810 systemd running, like in initramfs or other init systems. Building
811 udev though, will require the *build* of the systemd tree, but
812 udev can be properly *run* without systemd.
814 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
815 should be used to create dead device nodes as workarounds for broken
818 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
819 no longer supported. udev_monitor_new_from_netlink() needs to be
820 used to subscribe to events.
822 * udev: when udevd is started by systemd, processes which are left
823 behind by forking them off of udev rules, are unconditionally cleaned
824 up and killed now after the event handling has finished. Services or
825 daemons must be started as systemd services. Services can be
826 pulled-in by udev to get started, but they can no longer be directly
827 forked by udev rules.
829 * udev: the daemon binary is called systemd-udevd now and installed
830 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
831 to adapt to that, create symlink, or rename the binary after building
834 * libudev no longer provides these symbols:
835 udev_monitor_from_socket()
836 udev_queue_get_failed_list_entry()
837 udev_get_{dev,sys,run}_path()
838 The versions number was bumped and symbol versioning introduced.
840 * systemd-loginctl and systemd-journalctl have been renamed
841 to loginctl and journalctl to match systemctl.
843 * The config files: /etc/systemd/systemd-logind.conf and
844 /etc/systemd/systemd-journald.conf have been renamed to
845 logind.conf and journald.conf. Package updates should rename
846 the files to the new names on upgrade.
848 * For almost all files the license is now LGPL2.1+, changed
849 from the previous GPL2.0+. Exceptions are some minor stuff
850 of udev (which will be changed to LGPL2.1 eventually, too),
851 and the MIT licensed sd-daemon.[ch] library that is suitable
852 to be used as drop-in files.
854 * systemd and logind now handle system sleep states, in
855 particular suspending and hibernating.
857 * logind now implements a sleep/shutdown/idle inhibiting logic
858 suitable for a variety of uses. Soonishly Lennart will blog
859 about this in more detail.
861 * var-run.mount and var-lock.mount are no longer provided
862 (which prevously bind mounted these directories to their new
863 places). Distributions which have not converted these
864 directories to symlinks should consider stealing these files
865 from git history and add them downstream.
867 * We introduced the Documentation= field for units and added
868 this to all our shipped units. This is useful to make it
869 easier to explore the boot and the purpose of the various
872 * All smaller setup units (such as
873 systemd-vconsole-setup.service) now detect properly if they
874 are run in a container and are skipped when
875 appropriate. This guarantees an entirely noise-free boot in
876 Linux container environments such as systemd-nspawn.
878 * A framework for implementing offline system updates is now
879 integrated, for details see:
880 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
882 * A new service type Type=idle is available now which helps us
883 avoiding ugly interleaving of getty output and boot status
886 * There's now a system-wide CapabilityBoundingSet= option to
887 globally reduce the set of capabilities for the
888 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
889 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
890 even CAP_NET_ADMIN system-wide for secure systems.
892 * There are now system-wide DefaultLimitXXX= options to
893 globally change the defaults of the various resource limits
894 for all units started by PID 1.
896 * Harald Hoyer's systemd test suite has been integrated into
897 systemd which allows easy testing of systemd builds in qemu
898 and nspawn. (This is really awesome! Ask us for details!)
900 * The fstab parser is now implemented as generator, not inside
903 * systemctl will now warn you if .mount units generated from
904 /etc/fstab are out of date due to changes in fstab that
905 haven't been read by systemd yet.
907 * systemd is now suitable for usage in initrds. Dracut has
908 already been updated to make use of this. With this in place
909 initrds get a slight bit faster but primarily are much
910 easier to introspect and debug since "systemctl status" in
911 the host system can be used to introspect initrd services,
912 and the journal from the initrd is kept around too.
914 * systemd-delta has been added, a tool to explore differences
915 between user/admin configuration and vendor defaults.
917 * PrivateTmp= now affects both /tmp and /var/tmp.
919 * Boot time status messages are now much prettier and feature
920 proper english language. Booting up systemd has never been
923 * Read-ahead pack files now include the inode number of all
924 files to pre-cache. When the inode changes the pre-caching
925 is not attempted. This should be nicer to deal with updated
926 packages which might result in changes of read-ahead
929 * We now temporaritly lower the kernel's read_ahead_kb variable
930 when collecting read-ahead data to ensure the kernel's
931 built-in read-ahead does not add noise to our measurements
932 of necessary blocks to pre-cache.
934 * There's now RequiresMountsFor= to add automatic dependencies
935 for all mounts necessary for a specific file system path.
937 * MountAuto= and SwapAuto= have been removed from
938 system.conf. Mounting file systems at boot has to take place
941 * nspawn now learned a new switch --uuid= to set the machine
942 ID on the command line.
944 * nspawn now learned the -b switch to automatically search
947 * vt102 is now the default TERM for serial TTYs, upgraded from
950 * systemd-logind now works on VT-less systems.
952 * The build tree has been reorganized. The individual
953 components now have directories of their own.
955 * A new condition type ConditionPathIsReadWrite= is now available.
957 * nspawn learned the new -C switch to create cgroups for the
958 container in other hierarchies.
960 * We now have support for hardware watchdogs, configurable in
963 * The scheduled shutdown logic now has a public API.
965 * We now mount /tmp as tmpfs by default, but this can be
966 masked and /etc/fstab can override it.
968 * Since udisks doesn't make use of /media anymore we are not
969 mounting a tmpfs on it anymore.
971 * journalctl gained a new --local switch to only interleave
972 locally generated journal files.
974 * We can now load the IMA policy at boot automatically.
976 * The GTK tools have been split off into a systemd-ui.
978 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
979 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
980 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
981 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
982 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
983 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
984 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
985 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
986 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
991 * This is mostly a bugfix release
993 * Support optional initialization of the machine ID from the
994 KVM or container configured UUID.
996 * Support immediate reboots with "systemctl reboot -ff"
998 * Show /etc/os-release data in systemd-analyze output
1000 * Many bugfixes for the journal, including endianess fixes and
1001 ensuring that disk space enforcement works
1003 * sd-login.h is C++ comptaible again
1005 * Extend the /etc/os-release format on request of the Debian
1008 * We now refuse non-UTF8 strings used in various configuration
1009 and unit files. This is done to ensure we don't pass invalid
1010 data over D-Bus or expose it elsewhere.
1012 * Register Mimo USB Screens as suitable for automatic seat
1015 * Read SELinux client context from journal clients in a race
1018 * Reorder configuration file lookup order. /etc now always
1019 overrides /run in order to allow the administrator to always
1020 and unconditionally override vendor supplied or
1021 automatically generated data.
1023 * The various user visible bits of the journal now have man
1024 pages. We still lack man pages for the journal API calls
1027 * We now ship all man pages in HTML format again in the
1030 Contributions from: Dave Reisner, Dirk Eibach, Frederic
1031 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
1032 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
1037 * This is mostly a bugfix release
1039 * systems lacking /etc/os-release are no longer supported.
1041 * Various functionality updates to libsystemd-login.so
1043 * Track class of PAM logins to distuingish greeters from
1046 Contributions from: Kay Sievers, Lennart Poettering, Michael
1051 * This is an important bugfix release for v41.
1053 * Building man pages is now optional which should be useful
1054 for those building systemd from git but unwilling to install
1057 * Watchdog support for supervising services is now usable. In
1058 a future release support for hardware watchdogs
1059 (i.e. /dev/watchdog) will be added building on this.
1061 * Service start rate limiting is now configurable and can be
1062 turned off per service. When a start rate limit is hit a
1063 reboot can automatically be triggered.
1065 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
1067 Contributions from: Benjamin Franzke, Bill Nottingham,
1068 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
1069 Schmidt, Michał Górny, Piotr Drąg
1073 * The systemd binary is installed /usr/lib/systemd/systemd now;
1074 An existing /sbin/init symlink needs to be adapted with the
1077 * The code that loads kernel modules has been ported to invoke
1078 libkmod directly, instead of modprobe. This means we do not
1079 support systems with module-init-tools anymore.
1081 * Watchdog support is now already useful, but still not
1084 * A new kernel command line option systemd.setenv= is
1085 understood to set system wide environment variables
1086 dynamically at boot.
1088 * We now limit the set of capabilities of systemd-journald.
1090 * We now set SIGPIPE to ignore by default, since it only is
1091 useful in shell pipelines, and has little use in general
1092 code. This can be disabled with IgnoreSIPIPE=no in unit
1095 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
1096 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
1101 * This is mostly a bugfix release
1103 * We now expose the reason why a service failed in the
1104 "Result" D-Bus property.
1106 * Rudimentary service watchdog support (will be completed over
1107 the next few releases.)
1109 * When systemd forks off in order execute some service we will
1110 now immediately changes its argv[0] to reflect which process
1111 it will execute. This is useful to minimize the time window
1112 with a generic argv[0], which makes bootcharts more useful
1114 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
1115 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
1116 Mike Kazantsev, Ray Strode
1120 * This is mostly a test release, but incorporates many
1123 * New systemd-cgtop tool to show control groups by their
1126 * Linking against libacl for ACLs is optional again. If
1127 disabled, support tracking device access for active logins
1128 goes becomes unavailable, and so does access to the user
1129 journals by the respective users.
1131 * If a group "adm" exists, journal files are automatically
1132 owned by them, thus allow members of this group full access
1133 to the system journal as well as all user journals.
1135 * The journal now stores the SELinux context of the logging
1136 client for all entries.
1138 * Add C++ inclusion guards to all public headers
1140 * New output mode "cat" in the journal to print only text
1141 messages, without any meta data like date or time.
1143 * Include tiny X server wrapper as a temporary stop-gap to
1144 teach XOrg udev display enumeration. This is used by display
1145 managers such as gdm, and will go away as soon as XOrg
1146 learned native udev hotplugging for display devices.
1148 * Add new systemd-cat tool for executing arbitrary programs
1149 with STDERR/STDOUT connected to the journal. Can also act as
1150 BSD logger replacement, and does so by default.
1152 * Optionally store all locally generated coredumps in the
1153 journal along with meta data.
1155 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
1156 writing short strings to files (for usage for /sys), and for
1157 creating symlinks, character and block device nodes.
1159 * New unit file option ControlGroupPersistent= to make cgroups
1160 persistent, following the mechanisms outlined in
1161 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
1163 * Support multiple local RTCs in a sane way
1165 * No longer monopolize IO when replaying readahead data on
1166 rotating disks, since we might starve non-file-system IO to
1167 death, since fanotify() will not see accesses done by blkid,
1170 * Don't show kernel threads in systemd-cgls anymore, unless
1171 requested with new -k switch.
1173 Contributions from: Dan Horák, Kay Sievers, Lennart
1174 Poettering, Michal Schmidt
1178 * This is mostly a test release, but incorporates many
1181 * The git repository moved to:
1182 git://anongit.freedesktop.org/systemd/systemd
1183 ssh://git.freedesktop.org/git/systemd/systemd
1185 * First release with the journal
1186 http://0pointer.de/blog/projects/the-journal.html
1188 * The journal replaces both systemd-kmsg-syslogd and
1189 systemd-stdout-bridge.
1191 * New sd_pid_get_unit() API call in libsystemd-logind
1193 * Many systemadm clean-ups
1195 * Introduce remote-fs-pre.target which is ordered before all
1196 remote mounts and may be used to start services before all
1199 * Added Mageia support
1201 * Add bash completion for systemd-loginctl
1203 * Actively monitor PID file creation for daemons which exit in
1204 the parent process before having finished writing the PID
1205 file in the daemon process. Daemons which do this need to be
1206 fixed (i.e. PID file creation must have finished before the
1207 parent exits), but we now react a bit more gracefully to them.
1209 * Add colourful boot output, mimicking the well-known output
1210 of existing distributions.
1212 * New option PassCredentials= for socket units, for
1213 compatibility with a recent kernel ABI breakage.
1215 * /etc/rc.local is now hooked in via a generator binary, and
1216 thus will no longer act as synchronization point during
1219 * systemctl list-unit-files now supports --root=.
1221 * systemd-tmpfiles now understands two new commands: z, Z for
1222 relabelling files according to the SELinux database. This is
1223 useful to apply SELinux labels to specific files in /sys,
1226 * Output of SysV services is now forwarded to both the console
1227 and the journal by default, not only just the console.
1229 * New man pages for all APIs from libsystemd-login.
1231 * The build tree got reorganized and a the build system is a
1232 lot more modular allowing embedded setups to specifically
1233 select the components of systemd they are interested in.
1235 * Support for Linux systems lacking the kernel VT subsystem is
1238 * configure's --with-rootdir= got renamed to
1239 --with-rootprefix= to follow the naming used by udev and
1242 * Unless specified otherwise we'll now install to /usr instead
1243 of /usr/local by default.
1245 * Processes with '@' in argv[0][0] are now excluded from the
1246 final shut-down killing spree, following the logic explained
1248 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
1250 * All processes remaining in a service cgroup when we enter
1251 the START or START_PRE states are now killed with
1252 SIGKILL. That means it is no longer possible to spawn
1253 background processes from ExecStart= lines (which was never
1254 supported anyway, and bad style).
1256 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
1257 reloading of units together.
1259 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
1260 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
1261 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
1262 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
1263 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek