1 systemd System and Service Manager
4 * timedated does no longer read NTP unit names from
5 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
6 implementations should add a:
7 Conflicts=systemd-timesyncd.service
8 to take over and replace systemd's NTP functionality.
12 * A new tool systemd-sysusers has been added. This tool
13 creates system users and groups in /etc/passwd and
14 /etc/group, based on static declarative system user/group
15 definitions in /usr/lib/sysusers.d/. This is useful to
16 enable factory resets and volatile systems that boot up with
17 an empty /etc directory, and thus need system users and
18 groups created during early boot. systemd now also ships
19 with two default sysusers.d/ files for the most basic
20 users and groups systemd and the core operating system
23 * A new tmpfiles snippet has been added that rebuilds the
24 essential files in /etc on boot, should they be missing.
26 * A directive for ensuring automatic clean-up of
27 /var/cache/man/ has been removed from the default
28 configuration. This line should now be shipped by the man
29 implementation. The necessary change has been made to the
30 man-db implementation. Note that you need to update your man
31 implementation to one that ships this line, otherwise no
32 automatic clean-up of /var/cache/man will take place.
34 * A new condition ConditionNeedsUpdate= has been added that
35 may conditionalize services to only run when /etc or /var
36 are "older" than the vendor operating system resources in
37 /usr. This is useful for reconstructing or updating /etc
38 after an offline update of /usr or a factory reset, on the
39 next reboot. Services that want to run once after such an
40 update or reset should use this condition and order
41 themselves before the new systemd-update-done.service, which
42 will mark the two directories as fully updated. A number of
43 service files have been added making use of this, to rebuild
44 the udev hardware database, the journald message catalog and
45 dynamic loader cache (ldconfig). The systemd-sysusers tool
46 described above also makes use of this now. With this in
47 place it is now possible to start up a minimal operating
48 system with /etc empty cleanly. For more information on the
49 concepts involved see this recent blog story:
51 http://0pointer.de/blog/projects/stateless.html
53 * A new system group "input" has been introduced, and all
54 input device nodes get this group assigned. This is useful
55 for system-level software to get access to input devices. It
56 complements what is already done for "audio" and "video".
58 * systemd-networkd learnt minimal DHCPv4 server support in
59 addition to the existing DHCPv4 client support. It also
60 learnt DHCPv6 client and IPv6 Router Solicitation client
61 support. The DHCPv4 client gained support for static routes
62 passed in from the server. Note that the [DHCPv4] section
63 known in older systemd-networkd versions has been renamed to
64 [DHCP] and is now also used by the DHCPv6 client. Existing
65 .network files using settings of this section should be
66 updated, though compatibility is maintained. Optionally, the
67 client hostname may now be sent to the DHCP server.
69 * networkd gained support for vxlan virtual networks as well
70 as tun/tap and dummy devices.
72 * networkd gained support for automatic allocation of address
73 ranges for interfaces from a system-wide pool of
74 addresses. This is useful for dynamically managing a large
75 number of interfaces with a single network configuration
76 file. In particular this is useful to easily assign
77 appropriate IP addresses to the veth links of a large number
80 * RPM macros for processing sysusers, sysctl and binfmt
81 drop-in snippets at package installation time have been
84 * The /etc/os-release file should now be placed in
85 /usr/lib/os-release. The old location is automatically
86 created as symlink. /usr/lib is the more appropriate
87 location of this file, since it shall actually describe the
88 vendor operating system shipped in /usr, and not the
89 configuration stored in /etc.
91 * .mount units gained a new boolean SloppyOptions= setting
92 that maps to mount(8)'s -s option which enables permissive
93 parsing of unknown mount options.
95 * tmpfiles learnt a new "L+" directive which creates a symlink
96 but (unlike "L") deletes a pre-existing file first, should
97 it already exist and not already be the correct
98 symlink. Similar, "b+", "c+" and "p+" directives have been
99 added as well, which create block and character devices, as
100 well as fifos in the filesystem, possibly removing any
101 pre-existing files of different types.
103 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
104 'argument' field (which so far specified the source to
105 symlink/copy the files from) is now optional. If omitted the
106 same file os copied from /usr/share/factory/ suffixed by the
107 full destination path. This is useful for populating /etc
108 with essential files, by copying them from vendor defaults
109 shipped in /usr/share/factory/etc.
111 * A new command "systemctl preset-all" has been added that
112 applies the service preset settings to all installed unit
113 files. A new switch --preset-mode= has been added that
114 controls whether only enable or only disable operations
117 * A new command "systemctl is-system-running" has been added
118 that allows checking the overall state of the system, for
119 example whether it is fully up and running.
121 * When the system boots up with an empty /etc, the equivalent
122 to "systemctl preset-all" is executed during early boot, to
123 make sure all default services are enabled after a factory
126 * systemd now contains a minimal preset file that enables the
127 most basic services systemd ships by default.
129 * Unit files' [Install] section gained a new DefaultInstance=
130 field for defining the default instance to create if a
131 template unit is enabled with no instance specified.
133 * A new passive target cryptsetup-pre.target has been added
134 that may be used by services that need to make they run and
135 finish before the first LUKS cryptographic device is set up.
137 * The /dev/loop-control and /dev/btrfs-control device nodes
138 are now owned by the "disk" group by default, opening up
139 access to this group.
141 * systemd-coredump will now automatically generate a
142 stack trace of all core dumps taking place on the system,
143 based on elfutils' libdw library. This stack trace is logged
146 * systemd-coredump may now optionally store coredumps directly
147 on disk (in /var/lib/systemd/coredump, possibly compressed),
148 instead of storing them unconditionally in the journal. This
149 mode is the new default. A new configuration file
150 /etc/systemd/coredump.conf has been added to configure this
151 and other parameters of systemd-coredump.
153 * coredumpctl gained a new "info" verb to show details about a
154 specific coredump. A new switch "-1" has also been added
155 that makes sure to only show information about the most
156 recent entry instead of all entries. Also, as the tool is
157 generally useful now the "systemd-" prefix of the binary
158 name has been removed. Distributions that want to maintain
159 compatibility with the old name should add a symlink from
160 the old name to the new name.
162 * journald's SplitMode= now defaults to "uid". This makes sure
163 that unprivileged users can access their own coredumps with
164 coredumpctl without restrictions.
166 * New kernel command line options "systemd.wants=" (for
167 pulling an additional unit during boot), "systemd.mask="
168 (for masking a specific unit for the boot), and
169 "systemd.debug-shell" (for enabling the debug shell on tty9)
170 have been added. This is implemented in the new generator
171 "systemd-debug-generator".
173 * systemd-nspawn will now by default filter a couple of
174 syscalls for containers, among them those required for
175 kernel module loading, direct x86 IO port access, swap
176 management, and kexec. Most importantly though
177 open_by_handle_at() is now prohibited for containers,
178 closing a hole similar to a recently discussed vulnerability
179 in docker regarding access to files on file hierarchies the
180 container should normally not have access to. Note that for
181 nspawn we generally make no security claims anyway (and
182 this is explicitly documented in the man page), so this is
183 just a fix for one of the most obvious problems.
185 * A new man page file-hierarchy(7) has been added that
186 contains a minimized, modernized version of the file system
187 layout systemd expects, similar in style to the FHS
188 specification or hier(5). A new tool systemd-path(1) has
189 been added to query many of these paths for the local
192 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
193 longer done. Since the directory now has a per-user size
194 limit, and is cleaned on logout this appears unnecessary,
195 in particular since this now brings the lifecycle of this
196 directory closer in line with how IPC objects are handled.
198 * systemd.pc now exports a number of additional directories,
199 including $libdir (which is useful to identify the library
200 path for the primary architecture of the system), and a
201 couple of drop-in directories.
203 * udev's predictable network interface names now use the dev_port
204 sysfs attribute, introduced in linux 3.15 instead of dev_id to
205 distinguish between ports of the same PCI function. dev_id should
206 only be used for ports using the same HW address, hence the need
209 * machined has been updated to export the OS version of a
210 container (read from /etc/os-release and
211 /usr/lib/os-release) on the bus. This is now shown in
212 "machinectl status" for a machine.
214 * A new service setting RestartForceExitStatus= has been
215 added. If configured to a set of exit signals or process
216 return values, the service will be restarted when the main
217 daemon process exits with any of them, regardless of the
220 * systemctl's -H switch for connecting to remote systemd
221 machines has been extended so that it may be used to
222 directly connect to a specific container on the
223 host. "systemctl -H root@foobar:waldi" will now connect as
224 user "root" to host "foobar", and then proceed directly to
225 the container named "waldi". Note that currently you have to
226 authenticate as user "root" for this to work, as entering
227 containers is a privileged operation.
229 Contributions from: Andreas Henriksson, Benjamin Steinwender,
230 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
231 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
232 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
233 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
234 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
235 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
236 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
237 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
238 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
239 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
240 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
242 -- Berlin, 2014-07-03
246 * As an experimental feature, udev now tries to lock the
247 disk device node (flock(LOCK_SH|LOCK_NB)) while it
248 executes events for the disk or any of its partitions.
249 Applications like partitioning programs can lock the
250 disk device node (flock(LOCK_EX)) and claim temporary
251 device ownership that way; udev will entirely skip all event
252 handling for this disk and its partitions. If the disk
253 was opened for writing, the close will trigger a partition
254 table rescan in udev's "watch" facility, and if needed
255 synthesize "change" events for the disk and all its partitions.
256 This is now unconditionally enabled, and if it turns out to
257 cause major problems, we might turn it on only for specific
258 devices, or might need to disable it entirely. Device Mapper
259 devices are excluded from this logic.
261 * We temporarily dropped the "-l" switch for fsck invocations,
262 since they collide with the flock() logic above. util-linux
263 upstream has been changed already to avoid this conflict,
264 and we will readd "-l" as soon as util-linux with this
265 change has been released.
267 * The dependency on libattr has been removed. Since a long
268 time, the extended attribute calls have moved to glibc, and
269 libattr is thus unnecessary.
271 * Virtualization detection works without priviliges now. This
272 means the systemd-detect-virt binary no longer requires
273 CAP_SYS_PTRACE file capabilities, and our daemons can run
274 with fewer privileges.
276 * systemd-networkd now runs under its own "systemd-network"
277 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
278 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
279 loses the ability to write to files owned by root this way.
281 * Similar, systemd-resolved now runs under its own
282 "systemd-resolve" user with no capabilities remaining.
284 * Similar, systemd-bus-proxyd now runs under its own
285 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
287 * systemd-networkd gained support for setting up "veth"
288 virtual ethernet devices for container connectivity, as well
289 as GRE and VTI tunnels.
291 * systemd-networkd will no longer automatically attempt to
292 manually load kernel modules necessary for certain tunnel
293 transports. Instead, it is assumed the kernel loads them
294 automatically when required. This only works correctly on
295 very new kernels. On older kernels, please consider adding
296 the kernel modules to /etc/modules-load.d/ as a work-around.
298 * The resolv.conf file systemd-resolved generates has been
299 moved to /run/systemd/resolve/. If you have a symlink from
300 /etc/resolv.conf, it might be necessary to correct it.
302 * Two new service settings, ProtectedHome= and ProtectedSystem=,
303 have been added. When enabled, they will make the user data
304 (such as /home) inaccessible or read-only and the system
305 (such as /usr) read-only, for specific services. This allows
306 very light-weight per-service sandboxing to avoid
307 modifications of user data or system files from
308 services. These two new switches have been enabled for all
309 of systemd's long-running services, where appropriate.
311 * Socket units gained new SocketUser= and SocketGroup=
312 settings to set the owner user and group of AF_UNIX sockets
313 and FIFOs in the file system.
315 * Socket units gained a new RemoveOnStop= setting. If enabled,
316 all FIFOS and sockets in the file system will be removed
317 when the specific socket unit is stopped.
319 * Socket units gained a new Symlinks= setting. It takes a list
320 of symlinks to create to file system sockets or FIFOs
321 created by the specific Unix sockets. This is useful to
322 manage symlinks to socket nodes with the same life-cycle as
325 * The /dev/log socket and /dev/initctl FIFO have been moved to
326 /run, and have been replaced by symlinks. This allows
327 connecting to these facilities even if PrivateDevices=yes is
328 used for a service (which makes /dev/log itself unavailable,
329 but /run is left). This also has the benefit of ensuring
330 that /dev only contains device nodes, directories and
331 symlinks, and nothing else.
333 * sd-daemon gained two new calls sd_pid_notify() and
334 sd_pid_notifyf(). They are similar to sd_notify() and
335 sd_notifyf(), but allow overriding of the source PID of
336 notification messages if permissions permit this. This is
337 useful to send notify messages on behalf of a different
338 process (for example, the parent process). The
339 systemd-notify tool has been updated to make use of this
340 when sending messages (so that notification messages now
341 originate from the shell script invoking systemd-notify and
342 not the systemd-notify process itself. This should minimize
343 a race where systemd fails to associate notification
344 messages to services when the originating process already
347 * A new "on-abnormal" setting for Restart= has been added. If
348 set, it will result in automatic restarts on all "abnormal"
349 reasons for a process to exit, which includes unclean
350 signals, core dumps, timeouts and watchdog timeouts, but
351 does not include clean and unclean exit codes or clean
352 signals. Restart=on-abnormal is an alternative for
353 Restart=on-failure for services that shall be able to
354 terminate and avoid restarts on certain errors, by
355 indicating so with an unclean exit code. Restart=on-failure
356 or Restart=on-abnormal is now the recommended setting for
357 all long-running services.
359 * If the InaccessibleDirectories= service setting points to a
360 mount point (or if there are any submounts contained within
361 it), it is now attempted to completely unmount it, to make
362 the file systems truly unavailable for the respective
365 * The ReadOnlyDirectories= service setting and
366 systemd-nspawn's --read-only parameter are now recursively
367 applied to all submounts, too.
369 * Mount units may now be created transiently via the bus APIs.
371 * The support for SysV and LSB init scripts has been removed
372 from the systemd daemon itself. Instead, it is now
373 implemented as a generator that creates native systemd units
374 from these scripts when needed. This enables us to remove a
375 substantial amount of legacy code from PID 1, following the
376 fact that many distributions only ship a very small number
377 of LSB/SysV init scripts nowadays.
379 * Priviliged Xen (dom0) domains are not considered
380 virtualization anymore by the virtualization detection
381 logic. After all, they generally have unrestricted access to
382 the hardware and usually are used to manage the unprivileged
385 * systemd-tmpfiles gained a new "C" line type, for copying
386 files or entire directories.
388 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
389 lines. So far, they have been non-globbing versions of the
390 latter, and have thus been redundant. In future, it is
391 recommended to only use "z". "m" has hence been removed
392 from the documentation, even though it stays supported.
394 * A tmpfiles snippet to recreate the most basic structure in
395 /var has been added. This is enough to create the /var/run →
396 /run symlink and create a couple of structural
397 directories. This allows systems to boot up with an empty or
398 volatile /var. Of course, while with this change, the core OS
399 now is capable with dealing with a volatile /var, not all
400 user services are ready for it. However, we hope that sooner
401 or later, many service daemons will be changed upstream so
402 that they are able to automatically create their necessary
403 directories in /var at boot, should they be missing. This is
404 the first step to allow state-less systems that only require
405 the vendor image for /usr to boot.
407 * systemd-nspawn has gained a new --tmpfs= switch to mount an
408 empty tmpfs instance to a specific directory. This is
409 particularly useful for making use of the automatic
410 reconstruction of /var (see above), by passing --tmpfs=/var.
412 * Access modes specified in tmpfiles snippets may now be
413 prefixed with "~", which indicates that they shall be masked
414 by whether the existing file or directly is currently
415 writable, readable or executable at all. Also, if specified,
416 the sgid/suid/sticky bits will be masked for all
419 * A new passive target unit "network-pre.target" has been
420 added which is useful for services that shall run before any
421 network is configured, for example firewall scripts.
423 * The "floppy" group that previously owned the /dev/fd*
424 devices is no longer used. The "disk" group is now used
425 instead. Distributions should probably deprecate usage of
428 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
429 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
430 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
431 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
432 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
433 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
436 -- Berlin, 2014-06-11
440 * A new "systemd-timesyncd" daemon has been added for
441 synchronizing the system clock across the network. It
442 implements an SNTP client. In contrast to NTP
443 implementations such as chrony or the NTP reference server,
444 this only implements a client side, and does not bother with
445 the full NTP complexity, focusing only on querying time from
446 one remote server and synchronizing the local clock to
447 it. Unless you intend to serve NTP to networked clients or
448 want to connect to local hardware clocks, this simple NTP
449 client should be more than appropriate for most
450 installations. The daemon runs with minimal privileges, and
451 has been hooked up with networkd to only operate when
452 network connectivity is available. The daemon saves the
453 current clock to disk every time a new NTP sync has been
454 acquired, and uses this to possibly correct the system clock
455 early at bootup, in order to accommodate for systems that
456 lack an RTC such as the Raspberry Pi and embedded devices,
457 and to make sure that time monotonically progresses on these
458 systems, even if it is not always correct. To make use of
459 this daemon, a new system user and group "systemd-timesync"
460 needs to be created on installation of systemd.
462 * The queue "seqnum" interface of libudev has been disabled, as
463 it was generally incompatible with device namespacing as
464 sequence numbers of devices go "missing" if the devices are
465 part of a different namespace.
467 * "systemctl list-timers" and "systemctl list-sockets" gained
468 a --recursive switch for showing units of these types also
469 for all local containers, similar in style to the already
470 supported --recursive switch for "systemctl list-units".
472 * A new RebootArgument= setting has been added for service
473 units, which may be used to specify a kernel reboot argument
474 to use when triggering reboots with StartLimitAction=.
476 * A new FailureAction= setting has been added for service
477 units which may be used to specify an operation to trigger
478 when a service fails. This works similarly to
479 StartLimitAction=, but unlike it, controls what is done
480 immediately rather than only after several attempts to
481 restart the service in question.
483 * hostnamed got updated to also expose the kernel name,
484 release, and version on the bus. This is useful for
485 executing commands like hostnamectl with the -H switch.
486 systemd-analyze makes use of this to properly display
487 details when running non-locally.
489 * The bootchart tool can now show cgroup information in the
492 * The CFS CPU quota cgroup attribute is now exposed for
493 services. The new CPUQuota= switch has been added for this
494 which takes a percentage value. Setting this will have the
495 result that a service may never get more CPU time than the
496 specified percentage, even if the machine is otherwise idle.
498 * systemd-networkd learned IPIP and SIT tunnel support.
500 * LSB init scripts exposing a dependency on $network will now
501 get a dependency on network-online.target rather than simply
502 network.target. This should bring LSB handling closer to
503 what it was on SysV systems.
505 * A new fsck.repair= kernel option has been added to control
506 how fsck shall deal with unclean file systems at boot.
508 * The (.ini) configuration file parser will now silently
509 ignore sections whose name begins with "X-". This may be
510 used to maintain application-specific extension sections in unit
513 * machined gained a new API to query the IP addresses of
514 registered containers. "machinectl status" has been updated
515 to show these addresses in its output.
517 * A new call sd_uid_get_display() has been added to the
518 sd-login APIs for querying the "primary" session of a
519 user. The "primary" session of the user is elected from the
520 user's sessions and generally a graphical session is
521 preferred over a text one.
523 * A minimal systemd-resolved daemon has been added. It
524 currently simply acts as a companion to systemd-networkd and
525 manages resolv.conf based on per-interface DNS
526 configuration, possibly supplied via DHCP. In the long run
527 we hope to extend this into a local DNSSEC enabled DNS and
530 * The systemd-networkd-wait-online tool is now enabled by
531 default. It will delay network-online.target until a network
532 connection has been configured. The tool primarily integrates
533 with networkd, but will also make a best effort to make sense
534 of network configuration performed in some other way.
536 * Two new service options StartupCPUShares= and
537 StartupBlockIOWeight= have been added that work similarly to
538 CPUShares= and BlockIOWeight= however only apply during
539 system startup. This is useful to prioritize certain services
540 differently during bootup than during normal runtime.
542 * hostnamed has been changed to prefer the statically
543 configured hostname in /etc/hostname (unless set to
544 'localhost' or empty) over any dynamic one supplied by
545 dhcp. With this change, the rules for picking the hostname
546 match more closely the rules of other configuration settings
547 where the local administrator's configuration in /etc always
548 overrides any other settings.
550 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
551 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
552 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
553 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
554 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
555 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
556 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
557 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
558 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
559 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
560 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
561 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
562 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
563 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
564 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
565 Lindskog, WaLyong Cho, Will Woods, Zbigniew
568 -- Beijing, 2014-05-28
572 * When restoring the screen brightness at boot, stay away from
573 the darkest setting or from the lowest 5% of the available
574 range, depending on which is the larger value of both. This
575 should effectively protect the user from rebooting into a
576 black screen, should the brightness have been set to minimum
579 * sd-login gained a new sd_machine_get_class() call to
580 determine the class ("vm" or "container") of a machine
581 registered with machined.
583 * sd-login gained new calls
584 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
585 to query the identity of the peer of a local AF_UNIX
586 connection. They operate similarly to their sd_pid_get_xyz()
589 * PID 1 will now maintain a system-wide system state engine
590 with the states "starting", "running", "degraded",
591 "maintenance", "stopping". These states are bound to system
592 startup, normal runtime, runtime with at least one failed
593 service, rescue/emergency mode and system shutdown. This
594 state is shown in the "systemctl status" output when no unit
595 name is passed. It is useful to determine system state, in
596 particularly when doing so for many systems or containers at
599 * A new command "list-machines" has been added to "systemctl"
600 that lists all local OS containers and shows their system
601 state (see above), if systemd runs inside of them.
603 * systemctl gained a new "-r" switch to recursively enumerate
604 units on all local containers, when used with the
605 "list-unit" command (which is the default one that is
606 executed when no parameters are specified).
608 * The GPT automatic partition discovery logic will now honour
609 two GPT partition flags: one may be set on a partition to
610 cause it to be mounted read-only, and the other may be set
611 on a partition to ignore it during automatic discovery.
613 * Two new GPT type UUIDs have been added for automatic root
614 partition discovery, for 32-bit and 64-bit ARM. This is not
615 particularly useful for discovering the root directory on
616 these architectures during bare-metal boots (since UEFI is
617 not common there), but still very useful to allow booting of
618 ARM disk images in nspawn with the -i option.
620 * MAC addresses of interfaces created with nspawn's
621 --network-interface= switch will now be generated from the
622 machine name, and thus be stable between multiple invocations
625 * logind will now automatically remove all IPC objects owned
626 by a user if she or he fully logs out. This makes sure that
627 users who are logged out cannot continue to consume IPC
628 resources. This covers SysV memory, semaphores and message
629 queues as well as POSIX shared memory and message
630 queues. Traditionally, SysV and POSIX IPC had no life-cycle
631 limits. With this functionality, that is corrected. This may
632 be turned off by using the RemoveIPC= switch of logind.conf.
634 * The systemd-machine-id-setup and tmpfiles tools gained a
635 --root= switch to operate on a specific root directory,
638 * journald can now forward logged messages to the TTYs of all
639 logged in users ("wall"). This is the default for all
640 emergency messages now.
642 * A new tool systemd-journal-remote has been added to stream
643 journal log messages across the network.
645 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
646 controller trees are mounted into it. Note that the
647 directories mounted beneath it are not read-only. This is a
648 security measure and is particularly useful because glibc
649 actually includes a search logic to pick any tmpfs it can
650 find to implement shm_open() if /dev/shm is not available
651 (which it might very well be in namespaced setups).
653 * machinectl gained a new "poweroff" command to cleanly power
654 down a local OS container.
656 * The PrivateDevices= unit file setting will now also drop the
657 CAP_MKNOD capability from the capability bound set, and
658 imply DevicePolicy=closed.
660 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
661 comprehensively on all long-running systemd services where
664 * systemd-udevd will now run in a disassociated mount
665 namespace. To mount directories from udev rules, make sure to
666 pull in mount units via SYSTEMD_WANTS properties.
668 * The kdbus support gained support for uploading policy into
669 the kernel. sd-bus gained support for creating "monitoring"
670 connections that can eavesdrop into all bus communication
671 for debugging purposes.
673 * Timestamps may now be specified in seconds since the UNIX
674 epoch Jan 1st, 1970 by specifying "@" followed by the value
677 * Native tcpwrap support in systemd has been removed. tcpwrap
678 is old code, not really maintained anymore and has serious
679 shortcomings, and better options such as firewalls
680 exist. For setups that require tcpwrap usage, please
681 consider invoking your socket-activated service via tcpd,
682 like on traditional inetd.
684 * A new system.conf configuration option
685 DefaultTimerAccuracySec= has been added that controls the
686 default AccuracySec= setting of .timer units.
688 * Timer units gained a new WakeSystem= switch. If enabled,
689 timers configured this way will cause the system to resume
690 from system suspend (if the system supports that, which most
693 * Timer units gained a new Persistent= switch. If enabled,
694 timers configured this way will save to disk when they have
695 been last triggered. This information is then used on next
696 reboot to possible execute overdue timer events, that
697 could not take place because the system was powered off.
698 This enables simple anacron-like behaviour for timer units.
700 * systemctl's "list-timers" will now also list the time a
701 timer unit was last triggered in addition to the next time
702 it will be triggered.
704 * systemd-networkd will now assign predictable IPv4LL
705 addresses to its local interfaces.
707 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
708 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
709 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
710 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
711 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
712 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
713 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
714 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
717 -- Berlin, 2014-03-25
721 * A new unit file setting RestrictAddressFamilies= has been
722 added to restrict which socket address families unit
723 processes gain access to. This takes address family names
724 like "AF_INET" or "AF_UNIX", and is useful to minimize the
725 attack surface of services via exotic protocol stacks. This
726 is built on seccomp system call filters.
728 * Two new unit file settings RuntimeDirectory= and
729 RuntimeDirectoryMode= have been added that may be used to
730 manage a per-daemon runtime directories below /run. This is
731 an alternative for setting up directory permissions with
732 tmpfiles snippets, and has the advantage that the runtime
733 directory's lifetime is bound to the daemon runtime and that
734 the daemon starts up with an empty directory each time. This
735 is particularly useful when writing services that drop
736 privileges using the User= or Group= setting.
738 * The DeviceAllow= unit setting now supports globbing for
739 matching against device group names.
741 * The systemd configuration file system.conf gained new
742 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
743 DefaultMemoryAccounting= to globally turn on/off accounting
744 for specific resources (cgroups) for all units. These
745 settings may still be overridden individually in each unit
748 * systemd-gpt-auto-generator is now able to discover /srv and
749 root partitions in addition to /home and swap partitions. It
750 also supports LUKS-encrypted partitions now. With this in
751 place, automatic discovery of partitions to mount following
752 the Discoverable Partitions Specification
753 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
754 is now a lot more complete. This allows booting without
755 /etc/fstab and without root= on the kernel command line on
756 systems prepared appropriately.
758 * systemd-nspawn gained a new --image= switch which allows
759 booting up disk images and Linux installations on any block
760 device that follow the Discoverable Partitions Specification
761 (see above). This means that installations made with
762 appropriately updated installers may now be started and
763 deployed using container managers, completely
764 unmodified. (We hope that libvirt-lxc will add support for
765 this feature soon, too.)
767 * systemd-nspawn gained a new --network-macvlan= setting to
768 set up a private macvlan interface for the
769 container. Similarly, systemd-networkd gained a new
770 Kind=macvlan setting in .netdev files.
772 * systemd-networkd now supports configuring local addresses
775 * A new tool systemd-network-wait-online has been added to
776 synchronously wait for network connectivity using
779 * The sd-bus.h bus API gained a new sd_bus_track object for
780 tracking the life-cycle of bus peers. Note that sd-bus.h is
781 still not a public API though (unless you specify
782 --enable-kdbus on the configure command line, which however
783 voids your warranty and you get no API stability guarantee).
785 * The $XDG_RUNTIME_DIR runtime directories for each user are
786 now individual tmpfs instances, which has the benefit of
787 introducing separate pools for each user, with individual
788 size limits, and thus making sure that unprivileged clients
789 can no longer negatively impact the system or other users by
790 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
791 RuntimeDirectorySize= has been introduced that allows
792 controlling the default size limit for all users. It
793 defaults to 10% of the available physical memory. This is no
794 replacement for quotas on tmpfs though (which the kernel
795 still does not support), as /dev/shm and /tmp are still
796 shared resources used by both the system and unprivileged
799 * logind will now automatically turn off automatic suspending
800 on laptop lid close when more than one display is
801 connected. This was previously expected to be implemented
802 individually in desktop environments (such as GNOME),
803 however has been added to logind now, in order to fix a
804 boot-time race where a desktop environment might not have
805 been started yet and thus not been able to take an inhibitor
806 lock at the time where logind already suspends the system
809 * logind will now wait at least 30s after each system
810 suspend/resume cycle, and 3min after system boot before
811 suspending the system due to a closed laptop lid. This
812 should give USB docking stations and similar enough time to
813 be probed and configured after system resume and boot in
814 order to then act as suspend blocker.
816 * systemd-run gained a new --property= setting which allows
817 initialization of resource control properties (and others)
818 for the created scope or service unit. Example: "systemd-run
819 --property=BlockIOWeight=10 updatedb" may be used to run
820 updatedb at a low block IO scheduling weight.
822 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
823 now also work in --scope mode.
825 * When systemd is compiled with kdbus support, basic support
826 for enforced policies is now in place. (Note that enabling
827 kdbus still voids your warranty and no API compatibility
830 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
831 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
832 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
833 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
834 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
835 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
836 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
837 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
838 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
839 Zbigniew Jędrzejewski-Szmek
841 -- Berlin, 2014-03-12
845 * systemd will now relabel /dev after loading the SMACK policy
846 according to SMACK rules.
848 * A new unit file option AppArmorProfile= has been added to
849 set the AppArmor profile for the processes of a unit.
851 * A new condition check ConditionArchitecture= has been added
852 to conditionalize units based on the system architecture, as
853 reported by uname()'s "machine" field.
855 * systemd-networkd now supports matching on the system
856 virtualization, architecture, kernel command line, host name
859 * logind is now a lot more aggressive when suspending the
860 machine due to a closed laptop lid. Instead of acting only
861 on the lid close action, it will continuously watch the lid
862 status and act on it. This is useful for laptops where the
863 power button is on the outside of the chassis so that it can
864 be reached without opening the lid (such as the Lenovo
865 Yoga). On those machines, logind will now immediately
866 re-suspend the machine if the power button has been
867 accidentally pressed while the laptop was suspended and in a
870 * logind will now watch SW_DOCK switches and inhibit reaction
871 to the lid switch if it is pressed. This means that logind
872 will not suspend the machine anymore if the lid is closed
873 and the system is docked, if the laptop supports SW_DOCK
874 notifications via the input layer. Note that ACPI docking
875 stations do not generate this currently. Also note that this
876 logic is usually not fully sufficient and Desktop
877 Environments should take a lid switch inhibitor lock when an
878 external display is connected, as systemd will not watch
881 * nspawn will now make use of the devices cgroup controller by
882 default, and only permit creation of and access to the usual
883 API device nodes like /dev/null or /dev/random, as well as
884 access to (but not creation of) the pty devices.
886 * We will now ship a default .network file for
887 systemd-networkd that automatically configures DHCP for
888 network interfaces created by nspawn's --network-veth or
889 --network-bridge= switches.
891 * systemd will now understand the usual M, K, G, T suffixes
892 according to SI conventions (i.e. to the base 1000) when
893 referring to throughput and hardware metrics. It will stay
894 with IEC conventions (i.e. to the base 1024) for software
895 metrics, according to what is customary according to
896 Wikipedia. We explicitly document which base applies for
897 each configuration option.
899 * The DeviceAllow= setting in unit files now supports a syntax
900 to whitelist an entire group of devices node majors at once,
901 based on the /proc/devices listing. For example, with the
902 string "char-pts", it is now possible to whitelist all
903 current and future pseudo-TTYs at once.
905 * sd-event learned a new "post" event source. Event sources of
906 this type are triggered by the dispatching of any event
907 source of a type that is not "post". This is useful for
908 implementing clean-up and check event sources that are
909 triggered by other work being done in the program.
911 * systemd-networkd is no longer statically enabled, but uses
912 the usual [Install] sections so that it can be
913 enabled/disabled using systemctl. It still is enabled by
916 * When creating a veth interface pair with systemd-nspawn, the
917 host side will now be prefixed with "vb-" if
918 --network-bridge= is used, and with "ve-" if --network-veth
919 is used. This way, it is easy to distinguish these cases on
920 the host, for example to apply different configuration to
921 them with systemd-networkd.
923 * The compatibility libraries for libsystemd-journal.so,
924 libsystem-id128.so, libsystemd-login.so and
925 libsystemd-daemon.so do not make use of IFUNC
926 anymore. Instead, we now build libsystemd.so multiple times
927 under these alternative names. This means that the footprint
928 is drastically increased, but given that these are
929 transitional compatibility libraries, this should not matter
930 much. This change has been made necessary to support the ARM
931 platform for these compatibility libraries, as the ARM
932 toolchain is not really at the same level as the toolchain
933 for other architectures like x86 and does not support
934 IFUNC. Please make sure to use --enable-compat-libs only
935 during a transitional period!
937 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
938 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
939 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
940 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
941 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
942 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
943 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
944 Zbigniew Jędrzejewski-Szmek
946 -- Berlin, 2014-02-24
950 * A new component "systemd-networkd" has been added that can
951 be used to configure local network interfaces statically or
952 via DHCP. It is capable of bringing up bridges, VLANs, and
953 bonding. Currently, no hook-ups for interactive network
954 configuration are provided. Use this for your initrd,
955 container, embedded, or server setup if you need a simple,
956 yet powerful, network configuration solution. This
957 configuration subsystem is quite nifty, as it allows wildcard
958 hotplug matching in interfaces. For example, with a single
959 configuration snippet, you can configure that all Ethernet
960 interfaces showing up are automatically added to a bridge,
961 or similar. It supports link-sensing and more.
963 * A new tool "systemd-socket-proxyd" has been added which can
964 act as a bidirectional proxy for TCP sockets. This is
965 useful for adding socket activation support to services that
966 do not actually support socket activation, including virtual
967 machines and the like.
969 * Add a new tool to save/restore rfkill state on
972 * Save/restore state of keyboard backlights in addition to
973 display backlights on shutdown/boot.
975 * udev learned a new SECLABEL{} construct to label device
976 nodes with a specific security label when they appear. For
977 now, only SECLABEL{selinux} is supported, but the syntax is
978 prepared for additional security frameworks.
980 * udev gained a new scheme to configure link-level attributes
981 from files in /etc/systemd/network/*.link. These files can
982 match against MAC address, device path, driver name and type,
983 and will apply attributes like the naming policy, link speed,
984 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
985 address assignment policy (randomized, ...).
987 * The configuration of network interface naming rules for
988 "permanent interface names" has changed: a new NamePolicy=
989 setting in the [Link] section of .link files determines the
990 priority of possible naming schemes (onboard, slot, mac,
991 path). The default value of this setting is determined by
992 /usr/lib/net/links/99-default.link. Old
993 80-net-name-slot.rules udev configuration file has been
994 removed, so local configuration overriding this file should
995 be adapated to override 99-default.link instead.
997 * When the User= switch is used in a unit file, also
998 initialize $SHELL= based on the user database entry.
1000 * systemd no longer depends on libdbus. All communication is
1001 now done with sd-bus, systemd's low-level bus library
1004 * kdbus support has been added to PID 1 itself. When kdbus is
1005 enabled, this causes PID 1 to set up the system bus and
1006 enable support for a new ".busname" unit type that
1007 encapsulates bus name activation on kdbus. It works a little
1008 bit like ".socket" units, except for bus names. A new
1009 generator has been added that converts classic dbus1 service
1010 activation files automatically into native systemd .busname
1013 * sd-bus: add a light-weight vtable implementation that allows
1014 defining objects on the bus with a simple static const
1015 vtable array of its methods, signals and properties.
1017 * systemd will not generate or install static dbus
1018 introspection data anymore to /usr/share/dbus-1/interfaces,
1019 as the precise format of these files is unclear, and
1020 nothing makes use of it.
1022 * A proxy daemon is now provided to proxy clients connecting
1023 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1024 compatibility with classic D-Bus.
1026 * A bus driver implementation has been added that supports the
1027 classic D-Bus bus driver calls on kdbus, also for
1028 compatibility purposes.
1030 * A new API "sd-event.h" has been added that implements a
1031 minimal event loop API built around epoll. It provides a
1032 couple of features that direct epoll usage is lacking:
1033 prioritization of events, scales to large numbers of timer
1034 events, per-event timer slack (accuracy), system-wide
1035 coalescing of timer events, exit handlers, watchdog
1036 supervision support using systemd's sd_notify() API, child
1039 * A new API "sd-rntl.h" has been added that provides an API
1040 around the route netlink interface of the kernel, similar in
1041 style to "sd-bus.h".
1043 * A new API "sd-dhcp-client.h" has been added that provides a
1044 small DHCPv4 client-side implementation. This is used by
1047 * There is a new kernel command line option
1048 "systemd.restore_state=0|1". When set to "0", none of the
1049 systemd tools will restore saved runtime state to hardware
1050 devices. More specifically, the rfkill and backlight states
1053 * The FsckPassNo= compatibility option in mount/service units
1054 has been removed. The fstab generator will now add the
1055 necessary dependencies automatically, and does not require
1056 PID1's support for that anymore.
1058 * journalctl gained a new switch, --list-boots, that lists
1059 recent boots with their times and boot IDs.
1061 * The various tools like systemctl, loginctl, timedatectl,
1062 busctl, systemd-run, ... have gained a new switch "-M" to
1063 connect to a specific, local OS container (as direct
1064 connection, without requiring SSH). This works on any
1065 container that is registered with machined, such as those
1066 created by libvirt-lxc or nspawn.
1068 * systemd-run and systemd-analyze also gained support for "-H"
1069 to connect to remote hosts via SSH. This is particularly
1070 useful for systemd-run because it enables queuing of jobs
1071 onto remote systems.
1073 * machinectl gained a new command "login" to open a getty
1074 login in any local container. This works with any container
1075 that is registered with machined (such as those created by
1076 libvirt-lxc or nspawn), and which runs systemd inside.
1078 * machinectl gained a new "reboot" command that may be used to
1079 trigger a reboot on a specific container that is registered
1080 with machined. This works on any container that runs an init
1081 system of some kind.
1083 * systemctl gained a new "list-timers" command to print a nice
1084 listing of installed timer units with the times they elapse
1087 * Alternative reboot() parameters may now be specified on the
1088 "systemctl reboot" command line and are passed to the
1089 reboot() system call.
1091 * systemctl gained a new --job-mode= switch to configure the
1092 mode to queue a job with. This is a more generic version of
1093 --fail, --irreversible, and --ignore-dependencies, which are
1094 still available but not advertised anymore.
1096 * /etc/systemd/system.conf gained new settings to configure
1097 various default timeouts of units, as well as the default
1098 start limit interval and burst. These may still be overridden
1101 * PID1 will now export on the bus profile data of the security
1102 policy upload process (such as the SELinux policy upload to
1105 * journald: when forwarding logs to the console, include
1106 timestamps (following the setting in
1107 /sys/module/printk/parameters/time).
1109 * OnCalendar= in timer units now understands the special
1110 strings "yearly" and "annually". (Both are equivalent)
1112 * The accuracy of timer units is now configurable with the new
1113 AccuracySec= setting. It defaults to 1min.
1115 * A new dependency type JoinsNamespaceOf= has been added that
1116 allows running two services within the same /tmp and network
1117 namespace, if PrivateNetwork= or PrivateTmp= are used.
1119 * A new command "cat" has been added to systemctl. It outputs
1120 the original unit file of a unit, and concatenates the
1121 contents of additional "drop-in" unit file snippets, so that
1122 the full configuration is shown.
1124 * systemctl now supports globbing on the various "list-xyz"
1125 commands, like "list-units" or "list-sockets", as well as on
1126 those commands which take multiple unit names.
1128 * journalctl's --unit= switch gained support for globbing.
1130 * All systemd daemons now make use of the watchdog logic so
1131 that systemd automatically notices when they hang.
1133 * If the $container_ttys environment variable is set,
1134 getty-generator will automatically spawn a getty for each
1135 listed tty. This is useful for container managers to request
1136 login gettys to be spawned on as many ttys as needed.
1138 * %h, %s, %U specifier support is not available anymore when
1139 used in unit files for PID 1. This is because NSS calls are
1140 not safe from PID 1. They stay available for --user
1141 instances of systemd, and as special case for the root user.
1143 * loginctl gained a new "--no-legend" switch to turn off output
1146 * The "sd-login.h" API gained three new calls:
1147 sd_session_is_remote(), sd_session_get_remote_user(),
1148 sd_session_get_remote_host() to query information about
1151 * The udev hardware database now also carries vendor/product
1152 information of SDIO devices.
1154 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1155 determine whether watchdog notifications are requested by
1158 * Socket-activated per-connection services now include a
1159 short description of the connection parameters in the
1162 * tmpfiles gained a new "--boot" option. When this is not used,
1163 only lines where the command character is not suffixed with
1164 "!" are executed. When this option is specified, those
1165 options are executed too. This partitions tmpfiles
1166 directives into those that can be safely executed at any
1167 time, and those which should be run only at boot (for
1168 example, a line that creates /run/nologin).
1170 * A new API "sd-resolve.h" has been added which provides a simple
1171 asynchronous wrapper around glibc NSS host name resolution
1172 calls, such as getaddrinfo(). In contrast to glibc's
1173 getaddrinfo_a(), it does not use signals. In contrast to most
1174 other asynchronous name resolution libraries, this one does
1175 not reimplement DNS, but reuses NSS, so that alternate
1176 host name resolution systems continue to work, such as mDNS,
1177 LDAP, etc. This API is based on libasyncns, but it has been
1178 cleaned up for inclusion in systemd.
1180 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1181 "sd-daemon.h" are no longer found in individual libraries
1182 libsystemd-journal.so, libsystemd-login.so,
1183 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1184 merged them into a single library, libsystemd.so, which
1185 provides all symbols. The reason for this is cyclic
1186 dependencies, as these libraries tend to use each other's
1187 symbols. So far, we have managed to workaround that by linking
1188 a copy of a good part of our code into each of these
1189 libraries again and again, which, however, makes certain
1190 things hard to do, like sharing static variables. Also, it
1191 substantially increases footprint. With this change, there
1192 is only one library for the basic APIs systemd
1193 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1194 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1195 library as well, however are subject to the --enable-kdbus
1196 switch (see below). Note that "sd-dhcp-client.h" is not part
1197 of this library (this is because it only consumes, never
1198 provides, services of/to other APIs). To make the transition
1199 easy from the separate libraries to the unified one, we
1200 provide the --enable-compat-libs compile-time switch which
1201 will generate stub libraries that are compatible with the
1202 old ones but redirect all calls to the new one.
1204 * All of the kdbus logic and the new APIs "sd-bus.h",
1205 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1206 and "sd-utf8.h" are compile-time optional via the
1207 "--enable-kdbus" switch, and they are not compiled in by
1208 default. To make use of kdbus, you have to explicitly enable
1209 the switch. Note however, that neither the kernel nor the
1210 userspace API for all of this is considered stable yet. We
1211 want to maintain the freedom to still change the APIs for
1212 now. By specifying this build-time switch, you acknowledge
1213 that you are aware of the instability of the current
1216 * Also, note that while kdbus is pretty much complete,
1217 it lacks one thing: proper policy support. This means you
1218 can build a fully working system with all features; however,
1219 it will be highly insecure. Policy support will be added in
1220 one of the next releases, at the same time that we will
1221 declare the APIs stable.
1223 * When the kernel command-line argument "kdbus" is specified,
1224 systemd will automatically load the kdbus.ko kernel module. At
1225 this stage of development, it is only useful for testing kdbus
1226 and should not be used in production. Note: if "--enable-kdbus"
1227 is specified, and the kdbus.ko kernel module is available, and
1228 "kdbus" is added to the kernel command line, the entire system
1229 runs with kdbus instead of dbus-daemon, with the above mentioned
1230 problem of missing the system policy enforcement. Also a future
1231 version of kdbus.ko or a newer systemd will not be compatible with
1232 each other, and will unlikely be able to boot the machine if only
1233 one of them is updated.
1235 * systemctl gained a new "import-environment" command which
1236 uploads the caller's environment (or parts thereof) into the
1237 service manager so that it is inherited by services started
1238 by the manager. This is useful to upload variables like
1239 $DISPLAY into the user service manager.
1241 * A new PrivateDevices= switch has been added to service units
1242 which allows running a service with a namespaced /dev
1243 directory that does not contain any device nodes for
1244 physical devices. More specifically, it only includes devices
1245 such as /dev/null, /dev/urandom, and /dev/zero which are API
1248 * logind has been extended to support behaviour like VT
1249 switching on seats that do not support a VT. This makes
1250 multi-session available on seats that are not the first seat
1251 (seat0), and on systems where kernel support for VTs has
1252 been disabled at compile-time.
1254 * If a process holds a delay lock for system sleep or shutdown
1255 and fails to release it in time, we will now log its
1256 identity. This makes it easier to identify processes that
1257 cause slow suspends or power-offs.
1259 * When parsing /etc/crypttab, support for a new key-slot=
1260 option as supported by Debian is added. It allows indicating
1261 which LUKS slot to use on disk, speeding up key loading.
1263 * The sd_journald_sendv() API call has been checked and
1264 officially declared to be async-signal-safe so that it may
1265 be invoked from signal handlers for logging purposes.
1267 * Boot-time status output is now enabled automatically after a
1268 short timeout if boot does not progress, in order to give
1269 the user an indication what she or he is waiting for.
1271 * The boot-time output has been improved to show how much time
1272 remains until jobs expire.
1274 * The KillMode= switch in service units gained a new possible
1275 value "mixed". If set, and the unit is shut down, then the
1276 initial SIGTERM signal is sent only to the main daemon
1277 process, while the following SIGKILL signal is sent to
1278 all remaining processes of the service.
1280 * When a scope unit is registered, a new property "Controller"
1281 may be set. If set to a valid bus name, systemd will send a
1282 RequestStop() signal to this name when it would like to shut
1283 down the scope. This may be used to hook manager logic into
1284 the shutdown logic of scope units. Also, scope units may now
1285 be put in a special "abandoned" state, in which case the
1286 manager process which created them takes no further
1287 responsibilities for it.
1289 * When reading unit files, systemd will now verify
1290 the access mode of these files, and warn about certain
1291 suspicious combinations. This has been added to make it
1292 easier to track down packaging bugs where unit files are
1293 marked executable or world-writable.
1295 * systemd-nspawn gained a new "--setenv=" switch to set
1296 container-wide environment variables. The similar option in
1297 systemd-activate was renamed from "--environment=" to
1298 "--setenv=" for consistency.
1300 * systemd-nspawn has been updated to create a new kdbus domain
1301 for each container that is invoked, thus allowing each
1302 container to have its own set of system and user buses,
1303 independent of the host.
1305 * systemd-nspawn gained a new --drop-capability= switch to run
1306 the container with less capabilities than the default. Both
1307 --drop-capability= and --capability= now take the special
1308 string "all" for dropping or keeping all capabilities.
1310 * systemd-nspawn gained new switches for executing containers
1311 with specific SELinux labels set.
1313 * systemd-nspawn gained a new --quiet switch to not generate
1314 any additional output but the container's own console
1317 * systemd-nspawn gained a new --share-system switch to run a
1318 container without PID namespacing enabled.
1320 * systemd-nspawn gained a new --register= switch to control
1321 whether the container is registered with systemd-machined or
1322 not. This is useful for containers that do not run full
1323 OS images, but only specific apps.
1325 * systemd-nspawn gained a new --keep-unit which may be used
1326 when invoked as the only program from a service unit, and
1327 results in registration of the unit service itself in
1328 systemd-machined, instead of a newly opened scope unit.
1330 * systemd-nspawn gained a new --network-interface= switch for
1331 moving arbitrary interfaces to the container. The new
1332 --network-veth switch creates a virtual Ethernet connection
1333 between host and container. The new --network-bridge=
1334 switch then allows assigning the host side of this virtual
1335 Ethernet connection to a bridge device.
1337 * systemd-nspawn gained a new --personality= switch for
1338 setting the kernel personality for the container. This is
1339 useful when running a 32-bit container on a 64-bit host. A
1340 similar option Personality= is now also available for service
1343 * logind will now also track a "Desktop" identifier for each
1344 session which encodes the desktop environment of it. This is
1345 useful for desktop environments that want to identify
1346 multiple running sessions of itself easily.
1348 * A new SELinuxContext= setting for service units has been
1349 added that allows setting a specific SELinux execution
1350 context for a service.
1352 * Most systemd client tools will now honour $SYSTEMD_LESS for
1353 settings of the "less" pager. By default, these tools will
1354 override $LESS to allow certain operations to work, such as
1355 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1356 influence this logic.
1358 * systemd's "seccomp" hook-up has been changed to make use of
1359 the libseccomp library instead of using its own
1360 implementation. This has benefits for portability among
1363 * For usage together with SystemCallFilter=, a new
1364 SystemCallErrorNumber= setting has been introduced that
1365 allows configuration of a system error number to be returned
1366 on filtered system calls, instead of immediately killing the
1367 process. Also, SystemCallArchitectures= has been added to
1368 limit access to system calls of a particular architecture
1369 (in order to turn off support for unused secondary
1370 architectures). There is also a global
1371 SystemCallArchitectures= setting in system.conf now to turn
1372 off support for non-native system calls system-wide.
1374 * systemd requires a kernel with a working name_to_handle_at(),
1375 please see the kernel config requirements in the README file.
1377 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1378 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1379 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1380 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1381 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1382 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1383 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1384 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1385 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1386 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1387 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1388 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1389 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1390 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1391 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1392 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1393 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1394 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1395 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1396 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1397 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1398 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1399 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1400 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1402 -- Berlin, 2014-02-20
1406 * logind has gained support for facilitating privileged input
1407 and drm device access for unprivileged clients. This work is
1408 useful to allow Wayland display servers (and similar
1409 programs, such as kmscon) to run under the user's ID and
1410 access input and drm devices which are normally
1411 protected. When this is used (and the kernel is new enough)
1412 logind will "mute" IO on the file descriptors passed to
1413 Wayland as long as it is in the background and "unmute" it
1414 if it returns into the foreground. This allows secure
1415 session switching without allowing background sessions to
1416 eavesdrop on input and display data. This also introduces
1417 session switching support if VT support is turned off in the
1418 kernel, and on seats that are not seat0.
1420 * A new kernel command line option luks.options= is understood
1421 now which allows specifiying LUKS options for usage for LUKS
1422 encrypted partitions specified with luks.uuid=.
1424 * tmpfiles.d(5) snippets may now use specifier expansion in
1425 path names. More specifically %m, %b, %H, %v, are now
1426 replaced by the local machine id, boot id, hostname, and
1427 kernel version number.
1429 * A new tmpfiles.d(5) command "m" has been introduced which
1430 may be used to change the owner/group/access mode of a file
1431 or directory if it exists, but do nothing if it does not.
1433 * This release removes high-level support for the
1434 MemorySoftLimit= cgroup setting. The underlying kernel
1435 cgroup attribute memory.soft_limit= is currently badly
1436 designed and likely to be removed from the kernel API in its
1437 current form, hence we should not expose it for now.
1439 * The memory.use_hierarchy cgroup attribute is now enabled for
1440 all cgroups systemd creates in the memory cgroup
1441 hierarchy. This option is likely to be come the built-in
1442 default in the kernel anyway, and the non-hierarchial mode
1443 never made much sense in the intrinsically hierarchial
1446 * A new field _SYSTEMD_SLICE= is logged along with all journal
1447 messages containing the slice a message was generated
1448 from. This is useful to allow easy per-customer filtering of
1449 logs among other things.
1451 * systemd-journald will no longer adjust the group of journal
1452 files it creates to the "systemd-journal" group. Instead we
1453 rely on the journal directory to be owned by the
1454 "systemd-journal" group, and its setgid bit set, so that the
1455 kernel file system layer will automatically enforce that
1456 journal files inherit this group assignment. The reason for
1457 this change is that we cannot allow NSS look-ups from
1458 journald which would be necessary to resolve
1459 "systemd-journal" to a numeric GID, because this might
1460 create deadlocks if NSS involves synchronous queries to
1461 other daemons (such as nscd, or sssd) which in turn are
1462 logging clients of journald and might block on it, which
1463 would then dead lock. A tmpfiles.d(5) snippet included in
1464 systemd will make sure the setgid bit and group are
1465 properly set on the journal directory if it exists on every
1466 boot. However, we recommend adjusting it manually after
1467 upgrades too (or from RPM scriptlets), so that the change is
1468 not delayed until next reboot.
1470 * Backlight and random seed files in /var/lib/ have moved into
1471 the /var/lib/systemd/ directory, in order to centralize all
1472 systemd generated files in one directory.
1474 * Boot time performance measurements (as displayed by
1475 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1476 performance information if that's available to determine how
1477 much time BIOS and boot loader initialization required. With
1478 a sufficiently new BIOS you hence no longer need to boot
1479 with Gummiboot to get access to such information.
1481 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1482 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1483 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1484 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1485 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1486 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1487 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1489 -- Berlin, 2013-10-02
1493 * The Restart= option for services now understands a new
1494 on-watchdog setting, which will restart the service
1495 automatically if the service stops sending out watchdog keep
1496 alive messages (as configured with WatchdogSec=).
1498 * The getty generator (which is responsible for bringing up a
1499 getty on configured serial consoles) will no longer only
1500 start a getty on the primary kernel console but on all
1501 others, too. This makes the order in which console= is
1502 specified on the kernel command line less important.
1504 * libsystemd-logind gained a new sd_session_get_vt() call to
1505 retrieve the VT number of a session.
1507 * If the option "tries=0" is set for an entry of /etc/crypttab
1508 its passphrase is queried indefinitely instead of any
1509 maximum number of tries.
1511 * If a service with a configure PID file terminates its PID
1512 file will now be removed automatically if it still exists
1513 afterwards. This should put an end to stale PID files.
1515 * systemd-run will now also take relative binary path names
1516 for execution and no longer insists on absolute paths.
1518 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1519 paths that are optionally prefixed with "-" to indicate that
1520 it should not be considered a failure if they do not exist.
1522 * journalctl -o (and similar commands) now understands a new
1523 output mode "short-precise", it is similar to "short" but
1524 shows timestamps with usec accuracy.
1526 * The option "discard" (as known from Debian) is now
1527 synonymous to "allow-discards" in /etc/crypttab. In fact,
1528 "discard" is preferred now (since it is easier to remember
1531 * Some licensing clean-ups were made, so that more code is now
1532 LGPL-2.1 licensed than before.
1534 * A minimal tool to save/restore the display backlight
1535 brightness across reboots has been added. It will store the
1536 backlight setting as late as possible at shutdown, and
1537 restore it as early as possible during reboot.
1539 * A logic to automatically discover and enable home and swap
1540 partitions on GPT disks has been added. With this in place
1541 /etc/fstab becomes optional for many setups as systemd can
1542 discover certain partitions located on the root disk
1543 automatically. Home partitions are recognized under their
1544 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1545 partitions are recognized under their GPT type ID
1546 0657fd6da4ab43c484e50933c84b4f4f.
1548 * systemd will no longer pass any environment from the kernel
1549 or initrd to system services. If you want to set an
1550 environment for all services, do so via the kernel command
1551 line systemd.setenv= assignment.
1553 * The systemd-sysctl tool no longer natively reads the file
1554 /etc/sysctl.conf. If desired, the file should be symlinked
1555 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1556 legacy support by a symlink rather than built-in code, it
1557 also makes the otherwise hidden order of application of the
1558 different files visible. (Note that this partly reverts to a
1559 pre-198 application order of sysctl knobs!)
1561 * The "systemctl set-log-level" and "systemctl dump" commands
1562 have been moved to systemd-analyze.
1564 * systemd-run learned the new --remain-after-exit switch,
1565 which causes the scope unit not to be cleaned up
1566 automatically after the process terminated.
1568 * tmpfiles learned a new --exclude-prefix= switch to exclude
1569 certain paths from operation.
1571 * journald will now automatically flush all messages to disk
1572 as soon as a message of the log priorities CRIT, ALERT or
1575 Contributions from: Andrew Cook, Brandon Philips, Christian
1576 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1577 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1578 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1579 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1580 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1581 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1582 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1583 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1584 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1585 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1586 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1587 William Giokas, Zbigniew Jędrzejewski-Szmek
1589 -- Berlin, 2013-09-13
1593 * The documentation has been updated to cover the various new
1594 concepts introduced with 205.
1596 * Unit files now understand the new %v specifier which
1597 resolves to the kernel version string as returned by "uname
1600 * systemctl now supports filtering the unit list output by
1601 load state, active state and sub state, using the new
1604 * "systemctl status" will now show the results of the
1605 condition checks (like ConditionPathExists= and similar) of
1606 the last start attempts of the unit. They are also logged to
1609 * "journalctl -b" may now be used to look for boot output of a
1610 specific boot. Try "journalctl -b -1" for the previous boot,
1611 but the syntax is substantially more powerful.
1613 * "journalctl --show-cursor" has been added which prints the
1614 cursor string the last shown log line. This may then be used
1615 with the new "journalctl --after-cursor=" switch to continue
1616 browsing logs from that point on.
1618 * "journalctl --force" may now be used to force regeneration
1621 * Creation of "dead" device nodes has been moved from udev
1622 into kmod and tmpfiles. Previously, udev would read the kmod
1623 databases to pre-generate dead device nodes based on meta
1624 information contained in kernel modules, so that these would
1625 be auto-loaded on access rather then at boot. As this
1626 does not really have much to do with the exposing actual
1627 kernel devices to userspace this has always been slightly
1628 alien in the udev codebase. Following the new scheme kmod
1629 will now generate a runtime snippet for tmpfiles from the
1630 module meta information and it now is tmpfiles' job to the
1631 create the nodes. This also allows overriding access and
1632 other parameters for the nodes using the usual tmpfiles
1633 facilities. As side effect this allows us to remove the
1634 CAP_SYS_MKNOD capability bit from udevd entirely.
1636 * logind's device ACLs may now be applied to these "dead"
1637 devices nodes too, thus finally allowing managed access to
1638 devices such as /dev/snd/sequencer whithout loading the
1639 backing module right-away.
1641 * A new RPM macro has been added that may be used to apply
1642 tmpfiles configuration during package installation.
1644 * systemd-detect-virt and ConditionVirtualization= now can
1645 detect User-Mode-Linux machines (UML).
1647 * journald will now implicitly log the effective capabilities
1648 set of processes in the message metadata.
1650 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1652 * The initrd interface has been simplified (more specifically,
1653 support for passing performance data via environment
1654 variables and fsck results via files in /run has been
1655 removed). These features were non-essential, and are
1656 nowadays available in a much nicer way by having systemd in
1657 the initrd serialize its state and have the hosts systemd
1658 deserialize it again.
1660 * The udev "keymap" data files and tools to apply keyboard
1661 specific mappings of scan to key codes, and force-release
1662 scan code lists have been entirely replaced by a udev
1663 "keyboard" builtin and a hwdb data file.
1665 * systemd will now honour the kernel's "quiet" command line
1666 argument also during late shutdown, resulting in a
1667 completely silent shutdown when used.
1669 * There's now an option to control the SO_REUSEPORT socket
1670 option in .socket units.
1672 * Instance units will now automatically get a per-template
1673 subslice of system.slice unless something else is explicitly
1674 configured. For example, instances of sshd@.service will now
1675 implicitly be placed in system-sshd.slice rather than
1676 system.slice as before.
1678 * Test coverage support may now be enabled at build time.
1680 Contributions from: Dave Reisner, Frederic Crozat, Harald
1681 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1682 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1683 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1684 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1685 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1686 Giokas, Zbigniew Jędrzejewski-Szmek
1688 -- Berlin, 2013-07-23
1692 * Two new unit types have been introduced:
1694 Scope units are very similar to service units, however, are
1695 created out of pre-existing processes -- instead of PID 1
1696 forking off the processes. By using scope units it is
1697 possible for system services and applications to group their
1698 own child processes (worker processes) in a powerful way
1699 which then maybe used to organize them, or kill them
1700 together, or apply resource limits on them.
1702 Slice units may be used to partition system resources in an
1703 hierarchial fashion and then assign other units to them. By
1704 default there are now three slices: system.slice (for all
1705 system services), user.slice (for all user sessions),
1706 machine.slice (for VMs and containers).
1708 Slices and scopes have been introduced primarily in
1709 context of the work to move cgroup handling to a
1710 single-writer scheme, where only PID 1
1711 creates/removes/manages cgroups.
1713 * There's a new concept of "transient" units. In contrast to
1714 normal units these units are created via an API at runtime,
1715 not from configuration from disk. More specifically this
1716 means it is now possible to run arbitrary programs as
1717 independent services, with all execution parameters passed
1718 in via bus APIs rather than read from disk. Transient units
1719 make systemd substantially more dynamic then it ever was,
1720 and useful as a general batch manager.
1722 * logind has been updated to make use of scope and slice units
1723 for managing user sessions. As a user logs in he will get
1724 his own private slice unit, to which all sessions are added
1725 as scope units. We also added support for automatically
1726 adding an instance of user@.service for the user into the
1727 slice. Effectively logind will no longer create cgroup
1728 hierarchies on its own now, it will defer entirely to PID 1
1729 for this by means of scope, service and slice units. Since
1730 user sessions this way become entities managed by PID 1
1731 the output of "systemctl" is now a lot more comprehensive.
1733 * A new mini-daemon "systemd-machined" has been added which
1734 may be used by virtualization managers to register local
1735 VMs/containers. nspawn has been updated accordingly, and
1736 libvirt will be updated shortly. machined will collect a bit
1737 of meta information about the VMs/containers, and assign
1738 them their own scope unit (see above). The collected
1739 meta-data is then made available via the "machinectl" tool,
1740 and exposed in "ps" and similar tools. machined/machinectl
1741 is compile-time optional.
1743 * As discussed earlier, the low-level cgroup configuration
1744 options ControlGroup=, ControlGroupModify=,
1745 ControlGroupPersistent=, ControlGroupAttribute= have been
1746 removed. Please use high-level attribute settings instead as
1747 well as slice units.
1749 * A new bus call SetUnitProperties() has been added to alter
1750 various runtime parameters of a unit. This is primarily
1751 useful to alter cgroup parameters dynamically in a nice way,
1752 but will be extended later on to make more properties
1753 modifiable at runtime. systemctl gained a new set-properties
1754 command that wraps this call.
1756 * A new tool "systemd-run" has been added which can be used to
1757 run arbitrary command lines as transient services or scopes,
1758 while configuring a number of settings via the command
1759 line. This tool is currently very basic, however already
1760 very useful. We plan to extend this tool to even allow
1761 queuing of execution jobs with time triggers from the
1762 command line, similar in fashion to "at".
1764 * nspawn will now inform the user explicitly that kernels with
1765 audit enabled break containers, and suggest the user to turn
1768 * Support for detecting the IMA and AppArmor security
1769 frameworks with ConditionSecurity= has been added.
1771 * journalctl gained a new "-k" switch for showing only kernel
1772 messages, mimicking dmesg output; in addition to "--user"
1773 and "--system" switches for showing only user's own logs
1776 * systemd-delta can now show information about drop-in
1777 snippets extending unit files.
1779 * libsystemd-bus has been substantially updated but is still
1780 not available as public API.
1782 * systemd will now look for the "debug" argument on the kernel
1783 command line and enable debug logging, similar to what
1784 "systemd.log_level=debug" already did before.
1786 * "systemctl set-default", "systemctl get-default" has been
1787 added to configure the default.target symlink, which
1788 controls what to boot into by default.
1790 * "systemctl set-log-level" has been added as a convenient
1791 way to raise and lower systemd logging threshold.
1793 * "systemd-analyze plot" will now show the time the various
1794 generators needed for execution, as well as information
1795 about the unit file loading.
1797 * libsystemd-journal gained a new sd_journal_open_files() call
1798 for opening specific journal files. journactl also gained a
1799 new switch to expose this new functionality. Previously we
1800 only supported opening all files from a directory, or all
1801 files from the system, as opening individual files only is
1802 racy due to journal file rotation.
1804 * systemd gained the new DefaultEnvironment= setting in
1805 /etc/systemd/system.conf to set environment variables for
1808 * If a privileged process logs a journal message with the
1809 OBJECT_PID= field set, then journald will automatically
1810 augment this with additional OBJECT_UID=, OBJECT_GID=,
1811 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
1812 system services want to log events about specific client
1813 processes. journactl/systemctl has been updated to make use
1814 of this information if all log messages regarding a specific
1817 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
1818 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
1819 Reisner, David Coppa, David King, David Strauss, Eelco
1820 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
1821 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
1822 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
1823 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
1824 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
1825 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
1826 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
1827 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
1828 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
1829 Łukasz Stelmach, 장동준
1833 * The Python bindings gained some minimal support for the APIs
1834 exposed by libsystemd-logind.
1836 * ConditionSecurity= gained support for detecting SMACK. Since
1837 this condition already supports SELinux and AppArmor we only
1838 miss IMA for this. Patches welcome!
1840 Contributions from: Karol Lewandowski, Lennart Poettering,
1841 Zbigniew Jędrzejewski-Szmek
1845 * systemd-nspawn will now create /etc/resolv.conf if
1846 necessary, before bind-mounting the host's file onto it.
1848 * systemd-nspawn will now store meta information about a
1849 container on the container's cgroup as extended attribute
1850 fields, including the root directory.
1852 * The cgroup hierarchy has been reworked in many ways. All
1853 objects any of the components systemd creates in the cgroup
1854 tree are now suffixed. More specifically, user sessions are
1855 now placed in cgroups suffixed with ".session", users in
1856 cgroups suffixed with ".user", and nspawn containers in
1857 cgroups suffixed with ".nspawn". Furthermore, all cgroup
1858 names are now escaped in a simple scheme to avoid collision
1859 of userspace object names with kernel filenames. This work
1860 is preparation for making these objects relocatable in the
1861 cgroup tree, in order to allow easy resource partitioning of
1862 these objects without causing naming conflicts.
1864 * systemctl list-dependencies gained the new switches
1865 --plain, --reverse, --after and --before.
1867 * systemd-inhibit now shows the process name of processes that
1868 have taken an inhibitor lock.
1870 * nss-myhostname will now also resolve "localhost"
1871 implicitly. This makes /etc/hosts an optional file and
1872 nicely handles that on IPv6 ::1 maps to both "localhost" and
1875 * libsystemd-logind.so gained a new call
1876 sd_get_machine_names() to enumerate running containers and
1877 VMs (currently only supported by very new libvirt and
1878 nspawn). sd_login_monitor can now be used to watch
1879 VMs/containers coming and going.
1881 * .include is not allowed recursively anymore, and only in
1882 unit files. Usually it is better to use drop-in snippets in
1883 .d/*.conf anyway, as introduced with systemd 198.
1885 * systemd-analyze gained a new "critical-chain" command that
1886 determines the slowest chain of units run during system
1887 boot-up. It is very useful for tracking down where
1888 optimizing boot time is the most beneficial.
1890 * systemd will no longer allow manipulating service paths in
1891 the name=systemd:/system cgroup tree using ControlGroup= in
1892 units. (But is still fine with it in all other dirs.)
1894 * There's a new systemd-nspawn@.service service file that may
1895 be used to easily run nspawn containers as system
1896 services. With the container's root directory in
1897 /var/lib/container/foobar it is now sufficient to run
1898 "systemctl start systemd-nspawn@foobar.service" to boot it.
1900 * systemd-cgls gained a new parameter "--machine" to list only
1901 the processes within a certain container.
1903 * ConditionSecurity= now can check for "apparmor". We still
1904 are lacking checks for SMACK and IMA for this condition
1905 check though. Patches welcome!
1907 * A new configuration file /etc/systemd/sleep.conf has been
1908 added that may be used to configure which kernel operation
1909 systemd is supposed to execute when "suspend", "hibernate"
1910 or "hybrid-sleep" is requested. This makes the new kernel
1911 "freeze" state accessible to the user.
1913 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
1914 the passed argument if applicable.
1916 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
1917 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
1918 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
1919 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
1920 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
1921 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
1922 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
1927 * The output of 'systemctl list-jobs' got some polishing. The
1928 '--type=' argument may now be passed more than once. A new
1929 command 'systemctl list-sockets' has been added which shows
1930 a list of kernel sockets systemd is listening on with the
1931 socket units they belong to, plus the units these socket
1934 * The experimental libsystemd-bus library got substantial
1935 updates to work in conjunction with the (also experimental)
1936 kdbus kernel project. It works well enough to exchange
1937 messages with some sophistication. Note that kdbus is not
1938 ready yet, and the library is mostly an elaborate test case
1939 for now, and not installable.
1941 * systemd gained a new unit 'systemd-static-nodes.service'
1942 that generates static device nodes earlier during boot, and
1943 can run in conjunction with udev.
1945 * libsystemd-login gained a new call sd_pid_get_user_unit()
1946 to retrieve the user systemd unit a process is running
1947 in. This is useful for systems where systemd is used as
1950 * systemd-nspawn now places all containers in the new /machine
1951 top-level cgroup directory in the name=systemd
1952 hierarchy. libvirt will soon do the same, so that we get a
1953 uniform separation of /system, /user and /machine for system
1954 services, user processes and containers/virtual
1955 machines. This new cgroup hierarchy is also useful to stick
1956 stable names to specific container instances, which can be
1957 recognized later this way (this name may be controlled
1958 via systemd-nspawn's new -M switch). libsystemd-login also
1959 gained a new call sd_pid_get_machine_name() to retrieve the
1960 name of the container/VM a specific process belongs to.
1962 * bootchart can now store its data in the journal.
1964 * libsystemd-journal gained a new call
1965 sd_journal_add_conjunction() for AND expressions to the
1966 matching logic. This can be used to express more complex
1967 logical expressions.
1969 * journactl can now take multiple --unit= and --user-unit=
1972 * The cryptsetup logic now understands the "luks.key=" kernel
1973 command line switch for specifying a file to read the
1974 decryption key from. Also, if a configured key file is not
1975 found the tool will now automatically fall back to prompting
1978 * Python systemd.journal module was updated to wrap recently
1979 added functions from libsystemd-journal. The interface was
1980 changed to bring the low level interface in s.j._Reader
1981 closer to the C API, and the high level interface in
1982 s.j.Reader was updated to wrap and convert all data about
1985 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
1986 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
1987 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
1988 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
1989 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
1990 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1994 * journalctl --update-catalog now understands a new --root=
1995 option to operate on catalogs found in a different root
1998 * During shutdown after systemd has terminated all running
1999 services a final killing loop kills all remaining left-over
2000 processes. We will now print the name of these processes
2001 when we send SIGKILL to them, since this usually indicates a
2004 * If /etc/crypttab refers to password files stored on
2005 configured mount points automatic dependencies will now be
2006 generated to ensure the specific mount is established first
2007 before the key file is attempted to be read.
2009 * 'systemctl status' will now show information about the
2010 network sockets a socket unit is listening on.
2012 * 'systemctl status' will also shown information about any
2013 drop-in configuration file for units. (Drop-In configuration
2014 files in this context are files such as
2015 /etc/systemd/systemd/foobar.service.d/*.conf)
2017 * systemd-cgtop now optionally shows summed up CPU times of
2018 cgroups. Press '%' while running cgtop to switch between
2019 percentage and absolute mode. This is useful to determine
2020 which cgroups use up the most CPU time over the entire
2021 runtime of the system. systemd-cgtop has also been updated
2022 to be 'pipeable' for processing with further shell tools.
2024 * 'hostnamectl set-hostname' will now allow setting of FQDN
2027 * The formatting and parsing of time span values has been
2028 changed. The parser now understands fractional expressions
2029 such as "5.5h". The formatter will now output fractional
2030 expressions for all time spans under 1min, i.e. "5.123456s"
2031 rather than "5s 123ms 456us". For time spans under 1s
2032 millisecond values are shown, for those under 1ms
2033 microsecond values are shown. This should greatly improve
2034 all time-related output of systemd.
2036 * libsystemd-login and libsystemd-journal gained new
2037 functions for querying the poll() events mask and poll()
2038 timeout value for integration into arbitrary event
2041 * localectl gained the ability to list available X11 keymaps
2042 (models, layouts, variants, options).
2044 * 'systemd-analyze dot' gained the ability to filter for
2045 specific units via shell-style globs, to create smaller,
2046 more useful graphs. I.e. it is now possible to create simple
2047 graphs of all the dependencies between only target units, or
2048 of all units that Avahi has dependencies with.
2050 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2051 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2052 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2053 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2054 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2055 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2056 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2060 * The boot-time readahead implementation for rotating media
2061 will now read the read-ahead data in multiple passes which
2062 consist of all read requests made in equidistant time
2063 intervals. This means instead of strictly reading read-ahead
2064 data in its physical order on disk we now try to find a
2065 middle ground between physical and access time order.
2067 * /etc/os-release files gained a new BUILD_ID= field for usage
2068 on operating systems that provide continuous builds of OS
2071 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2072 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2073 William Douglas, Zbigniew Jędrzejewski-Szmek
2077 * systemd-python gained an API exposing libsystemd-daemon.
2079 * The SMACK setup logic gained support for uploading CIPSO
2082 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2083 ReadOnlyDirectories= and InaccessibleDirectories= has
2084 changed. The private /tmp and /var/tmp directories are now
2085 shared by all processes of a service (which means
2086 ExecStartPre= may now leave data in /tmp that ExecStart= of
2087 the same service can still access). When a service is
2088 stopped its temporary directories are immediately deleted
2089 (normal clean-up with tmpfiles is still done in addition to
2092 * By default, systemd will now set a couple of sysctl
2093 variables in the kernel: the safe sysrq options are turned
2094 on, IP route verification is turned on, and source routing
2095 disabled. The recently added hardlink and softlink
2096 protection of the kernel is turned on. These settings should
2097 be reasonably safe, and good defaults for all new systems.
2099 * The predictable network naming logic may now be turned off
2100 with a new kernel command line switch: net.ifnames=0.
2102 * A new libsystemd-bus module has been added that implements a
2103 pretty complete D-Bus client library. For details see:
2105 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2107 * journald will now explicitly flush the journal files to disk
2108 at the latest 5min after each write. The file will then also
2109 be marked offline until the next write. This should increase
2110 reliability in case of a crash. The synchronization delay
2111 can be configured via SyncIntervalSec= in journald.conf.
2113 * There's a new remote-fs-setup.target unit that can be used
2114 to pull in specific services when at least one remote file
2115 system is to be mounted.
2117 * There are new targets timers.target and paths.target as
2118 canonical targets to pull user timer and path units in
2119 from. This complements sockets.target with a similar
2120 purpose for socket units.
2122 * libudev gained a new call udev_device_set_attribute_value()
2123 to set sysfs attributes of a device.
2125 * The udev daemon now sets the default number of worker
2126 processes executed in parallel based on the number of available
2127 CPUs instead of the amount of available RAM. This is supposed
2128 to provide a more reliable default and limit a too aggressive
2129 paralellism for setups with 1000s of devices connected.
2131 Contributions from: Auke Kok, Colin Walters, Cristian
2132 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2133 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2134 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2135 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2136 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2137 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2138 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2139 Zbigniew Jędrzejewski-Szmek
2143 * Configuration of unit files may now be extended via drop-in
2144 files without having to edit/override the unit files
2145 themselves. More specifically, if the administrator wants to
2146 change one value for a service file foobar.service he can
2147 now do so by dropping in a configuration snippet into
2148 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2149 will load all these snippets and apply them on top of the
2150 main unit configuration file, possibly extending or
2151 overriding its settings. Using these drop-in snippets is
2152 generally nicer than the two earlier options for changing
2153 unit files locally: copying the files from
2154 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2155 them there; or creating a new file in /etc/systemd/system/
2156 that incorporates the original one via ".include". Drop-in
2157 snippets into these .d/ directories can be placed in any
2158 directory systemd looks for units in, and the usual
2159 overriding semantics between /usr/lib, /etc and /run apply
2162 * Most unit file settings which take lists of items can now be
2163 reset by assigning the empty string to them. For example,
2164 normally, settings such as Environment=FOO=BAR append a new
2165 environment variable assignment to the environment block,
2166 each time they are used. By assigning Environment= the empty
2167 string the environment block can be reset to empty. This is
2168 particularly useful with the .d/*.conf drop-in snippets
2169 mentioned above, since this adds the ability to reset list
2170 settings from vendor unit files via these drop-ins.
2172 * systemctl gained a new "list-dependencies" command for
2173 listing the dependencies of a unit recursively.
2175 * Inhibitors are now honored and listed by "systemctl
2176 suspend", "systemctl poweroff" (and similar) too, not only
2177 GNOME. These commands will also list active sessions by
2180 * Resource limits (as exposed by the various control group
2181 controllers) can now be controlled dynamically at runtime
2182 for all units. More specifically, you can now use a command
2183 like "systemctl set-cgroup-attr foobar.service cpu.shares
2184 2000" to alter the CPU shares a specific service gets. These
2185 settings are stored persistently on disk, and thus allow the
2186 administrator to easily adjust the resource usage of
2187 services with a few simple commands. This dynamic resource
2188 management logic is also available to other programs via the
2189 bus. Almost any kernel cgroup attribute and controller is
2192 * systemd-vconsole-setup will now copy all font settings to
2193 all allocated VTs, where it previously applied them only to
2196 * libsystemd-login gained the new sd_session_get_tty() API
2199 * This release drops support for a few legacy or
2200 distribution-specific LSB facility names when parsing init
2201 scripts: $x-display-manager, $mail-transfer-agent,
2202 $mail-transport-agent, $mail-transfer-agent, $smtp,
2203 $null. Also, the mail-transfer-agent.target unit backing
2204 this has been removed. Distributions which want to retain
2205 compatibility with this should carry the burden for
2206 supporting this themselves and patch support for these back
2207 in, if they really need to. Also, the facilities $syslog and
2208 $local_fs are now ignored, since systemd does not support
2209 early-boot LSB init scripts anymore, and these facilities
2210 are implied anyway for normal services. syslog.target has
2213 * There are new bus calls on PID1's Manager object for
2214 cancelling jobs, and removing snapshot units. Previously,
2215 both calls were only available on the Job and Snapshot
2218 * systemd-journal-gatewayd gained SSL support.
2220 * The various "environment" files, such as /etc/locale.conf
2221 now support continuation lines with a backslash ("\") as
2222 last character in the line, similarly in style (but different)
2223 to how this is supported in shells.
2225 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2226 now implicitly appended to every log entry logged. systemctl
2227 has been updated to filter by this field when operating on a
2228 user systemd instance.
2230 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2231 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2232 the container. This makes it easier to boot unmodified
2233 Fedora systems in a container, which however still requires
2234 audit=0 to be passed on the kernel command line. Auditing in
2235 kernel and userspace is unfortunately still too broken in
2236 context of containers, hence we recommend compiling it out
2237 of the kernel or using audit=0. Hopefully this will be fixed
2238 one day for good in the kernel.
2240 * nspawn gained the new --bind= and --bind-ro= parameters to
2241 bind mount specific directories from the host into the
2244 * nspawn will now mount its own devpts file system instance
2245 into the container, in order not to leak pty devices from
2246 the host into the container.
2248 * systemd will now read the firmware boot time performance
2249 information from the EFI variables, if the used boot loader
2250 supports this, and takes it into account for boot performance
2251 analysis via "systemd-analyze". This is currently supported
2252 only in conjunction with Gummiboot, but could be supported
2253 by other boot loaders too. For details see:
2255 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2257 * A new generator has been added that automatically mounts the
2258 EFI System Partition (ESP) to /boot, if that directory
2259 exists, is empty, and no other file system has been
2260 configured to be mounted there.
2262 * logind will now send out PrepareForSleep(false) out
2263 unconditionally, after coming back from suspend. This may be
2264 used by applications as asynchronous notification for
2265 system resume events.
2267 * "systemctl unlock-sessions" has been added, that allows
2268 unlocking the screens of all user sessions at once, similar
2269 to how "systemctl lock-sessions" already locked all users
2270 sessions. This is backed by a new D-Bus call UnlockSessions().
2272 * "loginctl seat-status" will now show the master device of a
2273 seat. (i.e. the device of a seat that needs to be around for
2274 the seat to be considered available, usually the graphics
2277 * tmpfiles gained a new "X" line type, that allows
2278 configuration of files and directories (with wildcards) that
2279 shall be excluded from automatic cleanup ("aging").
2281 * udev default rules set the device node permissions now only
2282 at "add" events, and do not change them any longer with a
2283 later "change" event.
2285 * The log messages for lid events and power/sleep keypresses
2286 now carry a message ID.
2288 * We now have a substantially larger unit test suite, but this
2289 continues to be work in progress.
2291 * udevadm hwdb gained a new --root= parameter to change the
2292 root directory to operate relative to.
2294 * logind will now issue a background sync() request to the kernel
2295 early at shutdown, so that dirty buffers are flushed to disk early
2296 instead of at the last moment, in order to optimize shutdown
2299 * A new bootctl tool has been added that is an interface for
2300 certain boot loader operations. This is currently a preview
2301 and is likely to be extended into a small mechanism daemon
2302 like timedated, localed, hostnamed, and can be used by
2303 graphical UIs to enumerate available boot options, and
2304 request boot into firmware operations.
2306 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2307 the rest of the package. It also has been updated to work
2308 correctly in initrds.
2310 * Policykit previously has been runtime optional, and is now
2311 also compile time optional via a configure switch.
2313 * systemd-analyze has been reimplemented in C. Also "systemctl
2314 dot" has moved into systemd-analyze.
2316 * "systemctl status" with no further parameters will now print
2317 the status of all active or failed units.
2319 * Operations such as "systemctl start" can now be executed
2320 with a new mode "--irreversible" which may be used to queue
2321 operations that cannot accidentally be reversed by a later
2322 job queuing. This is by default used to make shutdown
2323 requests more robust.
2325 * The Python API of systemd now gained a new module for
2326 reading journal files.
2328 * A new tool kernel-install has been added that can install
2329 kernel images according to the Boot Loader Specification:
2331 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2333 * Boot time console output has been improved to provide
2334 animated boot time output for hanging jobs.
2336 * A new tool systemd-activate has been added which can be used
2337 to test socket activation with, directly from the command
2338 line. This should make it much easier to test and debug
2339 socket activation in daemons.
2341 * journalctl gained a new "--reverse" (or -r) option to show
2342 journal output in reverse order (i.e. newest line first).
2344 * journalctl gained a new "--pager-end" (or -e) option to jump
2345 to immediately jump to the end of the journal in the
2346 pager. This is only supported in conjunction with "less".
2348 * journalctl gained a new "--user-unit=" option, that works
2349 similarly to "--unit=" but filters for user units rather than
2352 * A number of unit files to ease adoption of systemd in
2353 initrds has been added. This moves some minimal logic from
2354 the various initrd implementations into systemd proper.
2356 * The journal files are now owned by a new group
2357 "systemd-journal", which exists specifically to allow access
2358 to the journal, and nothing else. Previously, we used the
2359 "adm" group for that, which however possibly covers more
2360 than just journal/log file access. This new group is now
2361 already used by systemd-journal-gatewayd to ensure this
2362 daemon gets access to the journal files and as little else
2363 as possible. Note that "make install" will also set FS ACLs
2364 up for /var/log/journal to give "adm" and "wheel" read
2365 access to it, in addition to "systemd-journal" which owns
2366 the journal files. We recommend that packaging scripts also
2367 add read access to "adm" + "wheel" to /var/log/journal, and
2368 all existing/future journal files. To normal users and
2369 administrators little changes, however packagers need to
2370 ensure to create the "systemd-journal" system group at
2371 package installation time.
2373 * The systemd-journal-gatewayd now runs as unprivileged user
2374 systemd-journal-gateway:systemd-journal-gateway. Packaging
2375 scripts need to create these system user/group at
2378 * timedated now exposes a new boolean property CanNTP that
2379 indicates whether a local NTP service is available or not.
2381 * systemd-detect-virt will now also detect xen PVs
2383 * The pstore file system is now mounted by default, if it is
2386 * In addition to the SELinux and IMA policies we will now also
2387 load SMACK policies at early boot.
2389 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2390 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2391 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2392 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2393 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2394 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2395 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2396 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2397 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2398 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2399 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2400 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2401 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2402 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2406 * Timer units now support calendar time events in addition to
2407 monotonic time events. That means you can now trigger a unit
2408 based on a calendar time specification such as "Thu,Fri
2409 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2410 or fifth day of any month of the year 2013, given that it is
2411 a thursday or friday. This brings timer event support
2412 considerably closer to cron's capabilities. For details on
2413 the supported calendar time specification language see
2416 * udev now supports a number of different naming policies for
2417 network interfaces for predictable names, and a combination
2418 of these policies is now the default. Please see this wiki
2419 document for details:
2421 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2423 * Auke Kok's bootchart implementation has been added to the
2424 systemd tree. It is an optional component that can graph the
2425 boot in quite some detail. It is one of the best bootchart
2426 implementations around and minimal in its code and
2429 * nss-myhostname has been integrated into the systemd source
2430 tree. nss-myhostname guarantees that the local hostname
2431 always stays resolvable via NSS. It has been a weak
2432 requirement of systemd-hostnamed since a long time, and
2433 since its code is actually trivial we decided to just
2434 include it in systemd's source tree. It can be turned off
2435 with a configure switch.
2437 * The read-ahead logic is now capable of properly detecting
2438 whether a btrfs file system is on SSD or rotating media, in
2439 order to optimize the read-ahead scheme. Previously, it was
2440 only capable of detecting this on traditional file systems
2443 * In udev, additional device properties are now read from the
2444 IAB in addition to the OUI database. Also, Bluetooth company
2445 identities are attached to the devices as well.
2447 * In service files %U may be used as specifier that is
2448 replaced by the configured user name of the service.
2450 * nspawn may now be invoked without a controlling TTY. This
2451 makes it suitable for invocation as its own service. This
2452 may be used to set up a simple containerized server system
2453 using only core OS tools.
2455 * systemd and nspawn can now accept socket file descriptors
2456 when they are started for socket activation. This enables
2457 implementation of socket activated nspawn
2458 containers. i.e. think about autospawning an entire OS image
2459 when the first SSH or HTTP connection is received. We expect
2460 that similar functionality will also be added to libvirt-lxc
2463 * journalctl will now suppress ANSI color codes when
2464 presenting log data.
2466 * systemctl will no longer show control group information for
2467 a unit if a the control group is empty anyway.
2469 * logind can now automatically suspend/hibernate/shutdown the
2472 * /etc/machine-info and hostnamed now also expose the chassis
2473 type of the system. This can be used to determine whether
2474 the local system is a laptop, desktop, handset or
2475 tablet. This information may either be configured by the
2476 user/vendor or is automatically determined from ACPI and DMI
2477 information if possible.
2479 * A number of PolicyKit actions are now bound together with
2480 "imply" rules. This should simplify creating UIs because
2481 many actions will now authenticate similar ones as well.
2483 * Unit files learnt a new condition ConditionACPower= which
2484 may be used to conditionalize a unit depending on whether an
2485 AC power source is connected or not, of whether the system
2486 is running on battery power.
2488 * systemctl gained a new "is-failed" verb that may be used in
2489 shell scripts and suchlike to check whether a specific unit
2490 is in the "failed" state.
2492 * The EnvironmentFile= setting in unit files now supports file
2493 globbing, and can hence be used to easily read a number of
2494 environment files at once.
2496 * systemd will no longer detect and recognize specific
2497 distributions. All distribution-specific #ifdeffery has been
2498 removed, systemd is now fully generic and
2499 distribution-agnostic. Effectively, not too much is lost as
2500 a lot of the code is still accessible via explicit configure
2501 switches. However, support for some distribution specific
2502 legacy configuration file formats has been dropped. We
2503 recommend distributions to simply adopt the configuration
2504 files everybody else uses now and convert the old
2505 configuration from packaging scripts. Most distributions
2506 already did that. If that's not possible or desirable,
2507 distributions are welcome to forward port the specific
2508 pieces of code locally from the git history.
2510 * When logging a message about a unit systemd will now always
2511 log the unit name in the message meta data.
2513 * localectl will now also discover system locale data that is
2514 not stored in locale archives, but directly unpacked.
2516 * logind will no longer unconditionally use framebuffer
2517 devices as seat masters, i.e. as devices that are required
2518 to be existing before a seat is considered preset. Instead,
2519 it will now look for all devices that are tagged as
2520 "seat-master" in udev. By default framebuffer devices will
2521 be marked as such, but depending on local systems other
2522 devices might be marked as well. This may be used to
2523 integrate graphics cards using closed source drivers (such
2524 as NVidia ones) more nicely into logind. Note however, that
2525 we recommend using the open source NVidia drivers instead,
2526 and no udev rules for the closed-source drivers will be
2527 shipped from us upstream.
2529 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2530 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2531 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2532 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2533 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2534 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2535 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2536 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2537 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2538 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2539 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2540 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2545 * udev gained support for loading additional device properties
2546 from an indexed database that is keyed by vendor/product IDs
2547 and similar device identifiers. For the beginning this
2548 "hwdb" is populated with data from the well-known PCI and
2549 USB database, but also includes PNP, ACPI and OID data. In
2550 the longer run this indexed database shall grow into
2551 becoming the one central database for non-essential
2552 userspace device metadata. Previously, data from the PCI/USB
2553 database was only attached to select devices, since the
2554 lookup was a relatively expensive operation due to O(n) time
2555 complexity (with n being the number of entries in the
2556 database). Since this is now O(1), we decided to add in this
2557 data for all devices where this is available, by
2558 default. Note that the indexed database needs to be rebuilt
2559 when new data files are installed. To achieve this you need
2560 to update your packaging scripts to invoke "udevadm hwdb
2561 --update" after installation of hwdb data files. For
2562 RPM-based distributions we introduced the new
2563 %udev_hwdb_update macro for this purpose.
2565 * The Journal gained support for the "Message Catalog", an
2566 indexed database to link up additional information with
2567 journal entries. For further details please check:
2569 http://www.freedesktop.org/wiki/Software/systemd/catalog
2571 The indexed message catalog database also needs to be
2572 rebuilt after installation of message catalog files. Use
2573 "journalctl --update-catalog" for this. For RPM-based
2574 distributions we introduced the %journal_catalog_update
2575 macro for this purpose.
2577 * The Python Journal bindings gained support for the standard
2578 Python logging framework.
2580 * The Journal API gained new functions for checking whether
2581 the underlying file system of a journal file is capable of
2582 properly reporting file change notifications, or whether
2583 applications that want to reflect journal changes "live"
2584 need to recheck journal files continuously in appropriate
2587 * It is now possible to set the "age" field for tmpfiles
2588 entries to 0, indicating that files matching this entry
2589 shall always be removed when the directories are cleaned up.
2591 * coredumpctl gained a new "gdb" verb which invokes gdb
2592 right-away on the selected coredump.
2594 * There's now support for "hybrid sleep" on kernels that
2595 support this, in addition to "suspend" and "hibernate". Use
2596 "systemctl hybrid-sleep" to make use of this.
2598 * logind's HandleSuspendKey= setting (and related settings)
2599 now gained support for a new "lock" setting to simply
2600 request the screen lock on all local sessions, instead of
2601 actually executing a suspend or hibernation.
2603 * systemd will now mount the EFI variables file system by
2606 * Socket units now gained support for configuration of the
2607 SMACK security label.
2609 * timedatectl will now output the time of the last and next
2610 daylight saving change.
2612 * We dropped support for various legacy and distro-specific
2613 concepts, such as insserv, early-boot SysV services
2614 (i.e. those for non-standard runlevels such as 'b' or 'S')
2615 or ArchLinux /etc/rc.conf support. We recommend the
2616 distributions who still need support this to either continue
2617 to maintain the necessary patches downstream, or find a
2618 different solution. (Talk to us if you have questions!)
2620 * Various systemd components will now bypass PolicyKit checks
2621 for root and otherwise handle properly if PolicyKit is not
2622 found to be around. This should fix most issues for
2623 PolicyKit-less systems. Quite frankly this should have been
2624 this way since day one. It is absolutely our intention to
2625 make systemd work fine on PolicyKit-less systems, and we
2626 consider it a bug if something does not work as it should if
2627 PolicyKit is not around.
2629 * For embedded systems it is now possible to build udev and
2630 systemd without blkid and/or kmod support.
2632 * "systemctl switch-root" is now capable of switching root
2633 more than once. I.e. in addition to transitions from the
2634 initrd to the host OS it is now possible to transition to
2635 further OS images from the host. This is useful to implement
2636 offline updating tools.
2638 * Various other additions have been made to the RPM macros
2639 shipped with systemd. Use %udev_rules_update() after
2640 installing new udev rules files. %_udevhwdbdir,
2641 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2642 %_sysctldir are now available which resolve to the right
2643 directories for packages to place various data files in.
2645 * journalctl gained the new --full switch (in addition to
2646 --all, to disable ellipsation for long messages.
2648 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2649 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2650 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2651 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2652 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2653 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2654 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2655 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2656 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2660 * journalctl gained new --since= and --until= switches to
2661 filter by time. It also now supports nice filtering for
2662 units via --unit=/-u.
2664 * Type=oneshot services may use ExecReload= and do the
2667 * The journal daemon now supports time-based rotation and
2668 vacuuming, in addition to the usual disk-space based
2671 * The journal will now index the available field values for
2672 each field name. This enables clients to show pretty drop
2673 downs of available match values when filtering. The bash
2674 completion of journalctl has been updated
2675 accordingly. journalctl gained a new switch -F to list all
2676 values a certain field takes in the journal database.
2678 * More service events are now written as structured messages
2679 to the journal, and made recognizable via message IDs.
2681 * The timedated, localed and hostnamed mini-services which
2682 previously only provided support for changing time, locale
2683 and hostname settings from graphical DEs such as GNOME now
2684 also have a minimal (but very useful) text-based client
2685 utility each. This is probably the nicest way to changing
2686 these settings from the command line now, especially since
2687 it lists available options and is fully integrated with bash
2690 * There's now a new tool "systemd-coredumpctl" to list and
2691 extract coredumps from the journal.
2693 * We now install a README each in /var/log/ and
2694 /etc/rc.d/init.d explaining where the system logs and init
2695 scripts went. This hopefully should help folks who go to
2696 that dirs and look into the otherwise now empty void and
2697 scratch their heads.
2699 * When user-services are invoked (by systemd --user) the
2700 $MANAGERPID env var is set to the PID of systemd.
2702 * SIGRTMIN+24 when sent to a --user instance will now result
2703 in immediate termination of systemd.
2705 * gatewayd received numerous feature additions such as a
2706 "follow" mode, for live syncing and filtering.
2708 * browse.html now allows filtering and showing detailed
2709 information on specific entries. Keyboard navigation and
2710 mouse screen support has been added.
2712 * gatewayd/journalctl now supports HTML5/JSON
2713 Server-Sent-Events as output.
2715 * The SysV init script compatibility logic will now
2716 heuristically determine whether a script supports the
2717 "reload" verb, and only then make this available as
2720 * "systemctl status --follow" has been removed, use "journalctl
2723 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
2724 have been removed since they are hardly useful to be
2727 * And I'd like to take the opportunity to specifically mention
2728 Zbigniew for his great contributions. Zbigniew, you rock!
2730 Contributions from: Andrew Eikum, Christian Hesse, Colin
2731 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
2732 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
2733 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
2734 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
2735 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
2736 Jędrzejewski-Szmek, Сковорода Никита Андреевич
2740 * If /etc/vconsole.conf is non-existent or empty we will no
2741 longer load any console font or key map at boot by
2742 default. Instead the kernel defaults will be left
2743 intact. This is definitely the right thing to do, as no
2744 configuration should mean no configuration, and hard-coding
2745 font names that are different on all archs is probably a bad
2746 idea. Also, the kernel default key map and font should be
2747 good enough for most cases anyway, and mostly identical to
2748 the userspace fonts/key maps we previously overloaded them
2749 with. If distributions want to continue to default to a
2750 non-kernel font or key map they should ship a default
2751 /etc/vconsole.conf with the appropriate contents.
2753 Contributions from: Colin Walters, Daniel J Walsh, Dave
2754 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
2755 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2759 * journalctl gained a new --cursor= switch to show entries
2760 starting from the specified location in the journal.
2762 * We now enforce a size limit on journal entry fields exported
2763 with "-o json" in journalctl. Fields larger than 4K will be
2764 assigned null. This can be turned off with --all.
2766 * An (optional) journal gateway daemon is now available as
2767 "systemd-journal-gatewayd.service". This service provides
2768 access to the journal via HTTP and JSON. This functionality
2769 will be used to implement live log synchronization in both
2770 pull and push modes, but has various other users too, such
2771 as easy log access for debugging of embedded devices. Right
2772 now it is already useful to retrieve the journal via HTTP:
2774 # systemctl start systemd-journal-gatewayd.service
2775 # wget http://localhost:19531/entries
2777 This will download the journal contents in a
2778 /var/log/messages compatible format. The same as JSON:
2780 # curl -H"Accept: application/json" http://localhost:19531/entries
2782 This service is also accessible via a web browser where a
2783 single static HTML5 app is served that uses the JSON logic
2784 to enable the user to do some basic browsing of the
2785 journal. This will be extended later on. Here's an example
2786 screenshot of this app in its current state:
2788 http://0pointer.de/public/journal-gatewayd
2790 Contributions from: Kay Sievers, Lennart Poettering, Robert
2791 Milasan, Tom Gundersen
2795 * The bash completion logic is now available for journalctl
2798 * We do not mount the "cpuset" controller anymore together with
2799 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
2800 started if no parameters are assigned to it. "cpuset" hence
2801 broke code that assumed it it could create "cpu" groups and
2804 * journalctl -f will now subscribe to terminal size changes,
2805 and line break accordingly.
2807 Contributions from: Dave Reisner, Kay Sievers, Lennart
2808 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
2812 * nspawn will now create a symlink /etc/localtime in the
2813 container environment, copying the host's timezone
2814 setting. Previously this has been done via a bind mount, but
2815 since symlinks cannot be bind mounted this has now been
2816 changed to create/update the appropriate symlink.
2818 * journalctl -n's line number argument is now optional, and
2819 will default to 10 if omitted.
2821 * journald will now log the maximum size the journal files may
2822 take up on disk. This is particularly useful if the default
2823 built-in logic of determining this parameter from the file
2824 system size is used. Use "systemctl status
2825 systemd-journald.service" to see this information.
2827 * The multi-seat X wrapper tool has been stripped down. As X
2828 is now capable of enumerating graphics devices via udev in a
2829 seat-aware way the wrapper is not strictly necessary
2830 anymore. A stripped down temporary stop-gap is still shipped
2831 until the upstream display managers have been updated to
2832 fully support the new X logic. Expect this wrapper to be
2833 removed entirely in one of the next releases.
2835 * HandleSleepKey= in logind.conf has been split up into
2836 HandleSuspendKey= and HandleHibernateKey=. The old setting
2837 is not available anymore. X11 and the kernel are
2838 distuingishing between these keys and we should too. This
2839 also means the inhibition lock for these keys has been split
2842 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
2843 Poettering, Lukas Nykryn, Václav Pavlín
2847 * Whenever a unit changes state we will now log this to the
2848 journal and show along the unit's own log output in
2851 * ConditionPathIsMountPoint= can now properly detect bind
2852 mount points too. (Previously, a bind mount of one file
2853 system to another place in the same file system could not be
2854 detected as mount, since they shared struct stat's st_dev
2857 * We will now mount the cgroup controllers cpu, cpuacct,
2858 cpuset and the controllers net_cls, net_prio together by
2861 * nspawn containers will now have a virtualized boot
2862 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
2863 over with a randomized ID at container initialization). This
2864 has the effect of making "journalctl -b" do the right thing
2867 * The JSON output journal serialization has been updated not
2868 to generate "endless" list objects anymore, but rather one
2869 JSON object per line. This is more in line how most JSON
2870 parsers expect JSON objects. The new output mode
2871 "json-pretty" has been added to provide similar output, but
2872 neatly aligned for readability by humans.
2874 * We dropped all explicit sync() invocations in the shutdown
2875 code. The kernel does this implicitly anyway in the kernel
2876 reboot() syscall. halt(8)'s -n option is now a compatibility
2879 * We now support virtualized reboot() in containers, as
2880 supported by newer kernels. We will fall back to exit() if
2881 CAP_SYS_REBOOT is not available to the container. Also,
2882 nspawn makes use of this now and will actually reboot the
2883 container if the containerized OS asks for that.
2885 * journalctl will only show local log output by default
2886 now. Use --merge (-m) to show remote log output, too.
2888 * libsystemd-journal gained the new sd_journal_get_usage()
2889 call to determine the current disk usage of all journal
2890 files. This is exposed in the new "journalctl --disk-usage"
2893 * journald gained a new configuration setting SplitMode= in
2894 journald.conf which may be used to control how user journals
2895 are split off. See journald.conf(5) for details.
2897 * A new condition type ConditionFileNotEmpty= has been added.
2899 * tmpfiles' "w" lines now support file globbing, to write
2900 multiple files at once.
2902 * We added Python bindings for the journal submission
2903 APIs. More Python APIs for a number of selected APIs will
2904 likely follow. Note that we intend to add native bindings
2905 only for the Python language, as we consider it common
2906 enough to deserve bindings shipped within systemd. There are
2907 various projects outside of systemd that provide bindings
2908 for languages such as PHP or Lua.
2910 * Many conditions will now resolve specifiers such as %i. In
2911 addition, PathChanged= and related directives of .path units
2912 now support specifiers as well.
2914 * There's now a new RPM macro definition for the system preset
2917 * journald will now warn if it ca not forward a message to the
2918 syslog daemon because its socket is full.
2920 * timedated will no longer write or process /etc/timezone,
2921 except on Debian. As we do not support late mounted /usr
2922 anymore /etc/localtime always being a symlink is now safe,
2923 and hence the information in /etc/timezone is not necessary
2926 * logind will now always reserve one VT for a text getty (VT6
2927 by default). Previously if more than 6 X sessions where
2928 started they took up all the VTs with auto-spawned gettys,
2929 so that no text gettys were available anymore.
2931 * udev will now automatically inform the btrfs kernel logic
2932 about btrfs RAID components showing up. This should make
2933 simple hotplug based btrfs RAID assembly work.
2935 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
2936 (but not for its children which will stay at the kernel
2937 default). This should allow setups with a lot more listening
2940 * systemd will now always pass the configured timezone to the
2941 kernel at boot. timedated will do the same when the timezone
2944 * logind's inhibition logic has been updated. By default,
2945 logind will now handle the lid switch, the power and sleep
2946 keys all the time, even in graphical sessions. If DEs want
2947 to handle these events on their own they should take the new
2948 handle-power-key, handle-sleep-key and handle-lid-switch
2949 inhibitors during their runtime. A simple way to achiveve
2950 that is to invoke the DE wrapped in an invocation of:
2952 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
2954 * Access to unit operations is now checked via SELinux taking
2955 the unit file label and client process label into account.
2957 * systemd will now notify the administrator in the journal
2958 when he over-mounts a non-empty directory.
2960 * There are new specifiers that are resolved in unit files,
2961 for the host name (%H), the machine ID (%m) and the boot ID
2964 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
2965 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
2966 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
2967 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
2968 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
2969 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2970 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
2974 * Support for reading structured kernel messages from
2975 /dev/kmsg has now been added and is enabled by default.
2977 * Support for reading kernel messages from /proc/kmsg has now
2978 been removed. If you want kernel messages in the journal
2979 make sure to run a recent kernel (>= 3.5) that supports
2980 reading structured messages from /dev/kmsg (see
2981 above). /proc/kmsg is now exclusive property of classic
2982 syslog daemons again.
2984 * The libudev API gained the new
2985 udev_device_new_from_device_id() call.
2987 * The logic for file system namespace (ReadOnlyDirectory=,
2988 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
2989 require pivot_root() anymore. This means fewer temporary
2990 directories are created below /tmp for this feature.
2992 * nspawn containers will now see and receive all submounts
2993 made on the host OS below the root file system of the
2996 * Forward Secure Sealing is now supported for Journal files,
2997 which provide cryptographical sealing of journal files so
2998 that attackers cannot alter log history anymore without this
2999 being detectable. Lennart will soon post a blog story about
3000 this explaining it in more detail.
3002 * There are two new service settings RestartPreventExitStatus=
3003 and SuccessExitStatus= which allow configuration of exit
3004 status (exit code or signal) which will be excepted from the
3005 restart logic, resp. consider successful.
3007 * journalctl gained the new --verify switch that can be used
3008 to check the integrity of the structure of journal files and
3009 (if Forward Secure Sealing is enabled) the contents of
3012 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3013 and similar symlinks pre-created. This makes running shells
3014 as container init process a lot more fun.
3016 * The fstab support can now handle PARTUUID= and PARTLABEL=
3019 * A new ConditionHost= condition has been added to match
3020 against the hostname (with globs) and machine ID. This is
3021 useful for clusters where a single OS image is used to
3022 provision a large number of hosts which shall run slightly
3023 different sets of services.
3025 * Services which hit the restart limit will now be placed in a
3028 Contributions from: Bertram Poettering, Dave Reisner, Huang
3029 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3030 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3034 * When running in --user mode systemd will now become a
3035 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3036 tree a lot more organized.
3038 * A new PartOf= unit dependency type has been introduced that
3039 may be used to group services in a natural way.
3041 * "systemctl enable" may now be used to enable instances of
3044 * journalctl now prints error log levels in red, and
3045 warning/notice log levels in bright white. It also supports
3046 filtering by log level now.
3048 * cgtop gained a new -n switch (similar to top), to configure
3049 the maximum number of iterations to run for. It also gained
3050 -b, to run in batch mode (accepting no input).
3052 * The suffix ".service" may now be omitted on most systemctl
3053 command lines involving service unit names.
3055 * There's a new bus call in logind to lock all sessions, as
3056 well as a loginctl verb for it "lock-sessions".
3058 * libsystemd-logind.so gained a new call sd_journal_perror()
3059 that works similar to libc perror() but logs to the journal
3060 and encodes structured information about the error number.
3062 * /etc/crypttab entries now understand the new keyfile-size=
3065 * shutdown(8) now can send a (configurable) wall message when
3066 a shutdown is cancelled.
3068 * The mount propagation mode for the root file system will now
3069 default to "shared", which is useful to make containers work
3070 nicely out-of-the-box so that they receive new mounts from
3071 the host. This can be undone locally by running "mount
3072 --make-rprivate /" if needed.
3074 * The prefdm.service file has been removed. Distributions
3075 should maintain this unit downstream if they intend to keep
3076 it around. However, we recommend writing normal unit files
3077 for display managers instead.
3079 * Since systemd is a crucial part of the OS we will now
3080 default to a number of compiler switches that improve
3081 security (hardening) such as read-only relocations, stack
3082 protection, and suchlike.
3084 * The TimeoutSec= setting for services is now split into
3085 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3086 of individual time outs for the start and the stop phase of
3089 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3090 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3091 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3092 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3093 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3094 Gundersen, Zbigniew Jędrzejewski-Szmek
3098 * The journal and id128 C APIs are now fully documented as man
3101 * Extra safety checks have been added when transitioning from
3102 the initial RAM disk to the main system to avoid accidental
3105 * /etc/crypttab entries now understand the new keyfile-offset=
3108 * systemctl -t can now be used to filter by unit load state.
3110 * The journal C API gained the new sd_journal_wait() call to
3111 make writing synchronous journal clients easier.
3113 * journalctl gained the new -D switch to show journals from a
3116 * journalctl now displays a special marker between log
3117 messages of two different boots.
3119 * The journal is now explicitly flushed to /var via a service
3120 systemd-journal-flush.service, rather than implicitly simply
3121 by seeing /var/log/journal to be writable.
3123 * journalctl (and the journal C APIs) can now match for much
3124 more complex expressions, with alternatives and
3127 * When transitioning from the initial RAM disk to the main
3128 system we will now kill all processes in a killing spree to
3129 ensure no processes stay around by accident.
3131 * Three new specifiers may be used in unit files: %u, %h, %s
3132 resolve to the user name, user home directory resp. user
3133 shell. This is useful for running systemd user instances.
3135 * We now automatically rotate journal files if their data
3136 object hash table gets a fill level > 75%. We also size the
3137 hash table based on the configured maximum file size. This
3138 together should lower hash collisions drastically and thus
3139 speed things up a bit.
3141 * journalctl gained the new "--header" switch to introspect
3142 header data of journal files.
3144 * A new setting SystemCallFilters= has been added to services
3145 which may be used to apply blacklists or whitelists to
3146 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3148 * nspawn gained a new --link-journal= switch (and quicker: -j)
3149 to link the container journal with the host. This makes it
3150 very easy to centralize log viewing on the host for all
3151 guests while still keeping the journal files separated.
3153 * Many bugfixes and optimizations
3155 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3156 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3157 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3162 * Several tools now understand kernel command line arguments,
3163 which are only read when run in an initial RAM disk. They
3164 usually follow closely their normal counterparts, but are
3167 * There's a new tool to analyze the readahead files that are
3168 automatically generated at boot. Use:
3170 /usr/lib/systemd/systemd-readahead analyze /.readahead
3172 * We now provide an early debug shell on tty9 if this enabled. Use:
3174 systemctl enable debug-shell.service
3176 * All plymouth related units have been moved into the Plymouth
3177 package. Please make sure to upgrade your Plymouth version
3180 * systemd-tmpfiles now supports getting passed the basename of
3181 a configuration file only, in which case it will look for it
3182 in all appropriate directories automatically.
3184 * udevadm info now takes a /dev or /sys path as argument, and
3185 does the right thing. Example:
3187 udevadm info /dev/sda
3188 udevadm info /sys/class/block/sda
3190 * systemctl now prints a warning if a unit is stopped but a
3191 unit that might trigger it continues to run. Example: a
3192 service is stopped but the socket that activates it is left
3195 * "systemctl status" will now mention if the log output was
3196 shortened due to rotation since a service has been started.
3198 * The journal API now exposes functions to determine the
3199 "cutoff" times due to rotation.
3201 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3202 immediately flushing of runtime logs to /var if possible,
3203 resp. for triggering immediate rotation of the journal
3206 * It is now considered an error if a service is attempted to
3207 be stopped that is not loaded.
3209 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3211 * systemd-analyze now supports Python 3
3213 * tmpfiles now supports cleaning up directories via aging
3214 where the first level dirs are always kept around but
3215 directories beneath it automatically aged. This is enabled
3216 by prefixing the age field with '~'.
3218 * Seat objects now expose CanGraphical, CanTTY properties
3219 which is required to deal with very fast bootups where the
3220 display manager might be running before the graphics drivers
3221 completed initialization.
3223 * Seat objects now expose a State property.
3225 * We now include RPM macros for service enabling/disabling
3226 based on the preset logic. We recommend RPM based
3227 distributions to make use of these macros if possible. This
3228 makes it simpler to reuse RPM spec files across
3231 * We now make sure that the collected systemd unit name is
3232 always valid when services log to the journal via
3235 * There's a new man page kernel-command-line(7) detailing all
3236 command line options we understand.
3238 * The fstab generator may now be disabled at boot by passing
3239 fstab=0 on the kernel command line.
3241 * A new kernel command line option modules-load= is now understood
3242 to load a specific kernel module statically, early at boot.
3244 * Unit names specified on the systemctl command line are now
3245 automatically escaped as needed. Also, if file system or
3246 device paths are specified they are automatically turned
3247 into the appropriate mount or device unit names. Example:
3249 systemctl status /home
3250 systemctl status /dev/sda
3252 * The SysVConsole= configuration option has been removed from
3253 system.conf parsing.
3255 * The SysV search path is no longer exported on the D-Bus
3258 * The Names= option is been removed from unit file parsing.
3260 * There's a new man page bootup(7) detailing the boot process.
3262 * Every unit and every generator we ship with systemd now
3263 comes with full documentation. The self-explanatory boot is
3266 * A couple of services gained "systemd-" prefixes in their
3267 name if they wrap systemd code, rather than only external
3268 code. Among them fsck@.service which is now
3269 systemd-fsck@.service.
3271 * The HaveWatchdog property has been removed from the D-Bus
3274 * systemd.confirm_spawn= on the kernel command line should now
3277 * There's a new man page crypttab(5) which details all options
3278 we actually understand.
3280 * systemd-nspawn gained a new --capability= switch to pass
3281 additional capabilities to the container.
3283 * timedated will now read known NTP implementation unit names
3284 from /usr/lib/systemd/ntp-units.d/*.list,
3285 systemd-timedated-ntp.target has been removed.
3287 * journalctl gained a new switch "-b" that lists log data of
3288 the current boot only.
3290 * The notify socket is in the abstract namespace again, in
3291 order to support daemons which chroot() at start-up.
3293 * There is a new Storage= configuration option for journald
3294 which allows configuration of where log data should go. This
3295 also provides a way to disable journal logging entirely, so
3296 that data collected is only forwarded to the console, the
3297 kernel log buffer or another syslog implementation.
3299 * Many bugfixes and optimizations
3301 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3302 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3303 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3304 Shawn Landden, Tom Gundersen
3308 * "systemctl help <unit>" now shows the man page if one is
3311 * Several new man pages have been added.
3313 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3314 MaxLevelConsole= can now be specified in
3315 journald.conf. These options allow reducing the amount of
3316 data stored on disk or forwarded by the log level.
3318 * TimerSlackNSec= can now be specified in system.conf for
3319 PID1. This allows system-wide power savings.
3321 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3322 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3327 * logind is now capable of (optionally) handling power and
3328 sleep keys as well as the lid switch.
3330 * journalctl now understands the syntax "journalctl
3331 /usr/bin/avahi-daemon" to get all log output of a specific
3334 * CapabilityBoundingSet= in system.conf now also influences
3335 the capability bound set of usermode helpers of the kernel.
3337 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3338 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3339 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3340 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3344 * Note that we skipped 139 releases here in order to set the
3345 new version to something that is greater than both udev's
3346 and systemd's most recent version number.
3348 * udev: all udev sources are merged into the systemd source tree now.
3349 All future udev development will happen in the systemd tree. It
3350 is still fully supported to use the udev daemon and tools without
3351 systemd running, like in initramfs or other init systems. Building
3352 udev though, will require the *build* of the systemd tree, but
3353 udev can be properly *run* without systemd.
3355 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3356 should be used to create dead device nodes as workarounds for broken
3359 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3360 no longer supported. udev_monitor_new_from_netlink() needs to be
3361 used to subscribe to events.
3363 * udev: when udevd is started by systemd, processes which are left
3364 behind by forking them off of udev rules, are unconditionally cleaned
3365 up and killed now after the event handling has finished. Services or
3366 daemons must be started as systemd services. Services can be
3367 pulled-in by udev to get started, but they can no longer be directly
3368 forked by udev rules.
3370 * udev: the daemon binary is called systemd-udevd now and installed
3371 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3372 to adapt to that, create symlink, or rename the binary after building
3375 * libudev no longer provides these symbols:
3376 udev_monitor_from_socket()
3377 udev_queue_get_failed_list_entry()
3378 udev_get_{dev,sys,run}_path()
3379 The versions number was bumped and symbol versioning introduced.
3381 * systemd-loginctl and systemd-journalctl have been renamed
3382 to loginctl and journalctl to match systemctl.
3384 * The config files: /etc/systemd/systemd-logind.conf and
3385 /etc/systemd/systemd-journald.conf have been renamed to
3386 logind.conf and journald.conf. Package updates should rename
3387 the files to the new names on upgrade.
3389 * For almost all files the license is now LGPL2.1+, changed
3390 from the previous GPL2.0+. Exceptions are some minor stuff
3391 of udev (which will be changed to LGPL2.1 eventually, too),
3392 and the MIT licensed sd-daemon.[ch] library that is suitable
3393 to be used as drop-in files.
3395 * systemd and logind now handle system sleep states, in
3396 particular suspending and hibernating.
3398 * logind now implements a sleep/shutdown/idle inhibiting logic
3399 suitable for a variety of uses. Soonishly Lennart will blog
3400 about this in more detail.
3402 * var-run.mount and var-lock.mount are no longer provided
3403 (which prevously bind mounted these directories to their new
3404 places). Distributions which have not converted these
3405 directories to symlinks should consider stealing these files
3406 from git history and add them downstream.
3408 * We introduced the Documentation= field for units and added
3409 this to all our shipped units. This is useful to make it
3410 easier to explore the boot and the purpose of the various
3413 * All smaller setup units (such as
3414 systemd-vconsole-setup.service) now detect properly if they
3415 are run in a container and are skipped when
3416 appropriate. This guarantees an entirely noise-free boot in
3417 Linux container environments such as systemd-nspawn.
3419 * A framework for implementing offline system updates is now
3420 integrated, for details see:
3421 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3423 * A new service type Type=idle is available now which helps us
3424 avoiding ugly interleaving of getty output and boot status
3427 * There's now a system-wide CapabilityBoundingSet= option to
3428 globally reduce the set of capabilities for the
3429 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3430 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3431 even CAP_NET_ADMIN system-wide for secure systems.
3433 * There are now system-wide DefaultLimitXXX= options to
3434 globally change the defaults of the various resource limits
3435 for all units started by PID 1.
3437 * Harald Hoyer's systemd test suite has been integrated into
3438 systemd which allows easy testing of systemd builds in qemu
3439 and nspawn. (This is really awesome! Ask us for details!)
3441 * The fstab parser is now implemented as generator, not inside
3444 * systemctl will now warn you if .mount units generated from
3445 /etc/fstab are out of date due to changes in fstab that
3446 have not been read by systemd yet.
3448 * systemd is now suitable for usage in initrds. Dracut has
3449 already been updated to make use of this. With this in place
3450 initrds get a slight bit faster but primarily are much
3451 easier to introspect and debug since "systemctl status" in
3452 the host system can be used to introspect initrd services,
3453 and the journal from the initrd is kept around too.
3455 * systemd-delta has been added, a tool to explore differences
3456 between user/admin configuration and vendor defaults.
3458 * PrivateTmp= now affects both /tmp and /var/tmp.
3460 * Boot time status messages are now much prettier and feature
3461 proper english language. Booting up systemd has never been
3464 * Read-ahead pack files now include the inode number of all
3465 files to pre-cache. When the inode changes the pre-caching
3466 is not attempted. This should be nicer to deal with updated
3467 packages which might result in changes of read-ahead
3470 * We now temporaritly lower the kernel's read_ahead_kb variable
3471 when collecting read-ahead data to ensure the kernel's
3472 built-in read-ahead does not add noise to our measurements
3473 of necessary blocks to pre-cache.
3475 * There's now RequiresMountsFor= to add automatic dependencies
3476 for all mounts necessary for a specific file system path.
3478 * MountAuto= and SwapAuto= have been removed from
3479 system.conf. Mounting file systems at boot has to take place
3482 * nspawn now learned a new switch --uuid= to set the machine
3483 ID on the command line.
3485 * nspawn now learned the -b switch to automatically search
3488 * vt102 is now the default TERM for serial TTYs, upgraded from
3491 * systemd-logind now works on VT-less systems.
3493 * The build tree has been reorganized. The individual
3494 components now have directories of their own.
3496 * A new condition type ConditionPathIsReadWrite= is now available.
3498 * nspawn learned the new -C switch to create cgroups for the
3499 container in other hierarchies.
3501 * We now have support for hardware watchdogs, configurable in
3504 * The scheduled shutdown logic now has a public API.
3506 * We now mount /tmp as tmpfs by default, but this can be
3507 masked and /etc/fstab can override it.
3509 * Since udisks does not make use of /media anymore we are not
3510 mounting a tmpfs on it anymore.
3512 * journalctl gained a new --local switch to only interleave
3513 locally generated journal files.
3515 * We can now load the IMA policy at boot automatically.
3517 * The GTK tools have been split off into a systemd-ui.
3519 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3520 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3521 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3522 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3523 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3524 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3525 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3526 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3527 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3532 * This is mostly a bugfix release
3534 * Support optional initialization of the machine ID from the
3535 KVM or container configured UUID.
3537 * Support immediate reboots with "systemctl reboot -ff"
3539 * Show /etc/os-release data in systemd-analyze output
3541 * Many bugfixes for the journal, including endianness fixes and
3542 ensuring that disk space enforcement works
3544 * sd-login.h is C++ comptaible again
3546 * Extend the /etc/os-release format on request of the Debian
3549 * We now refuse non-UTF8 strings used in various configuration
3550 and unit files. This is done to ensure we do not pass invalid
3551 data over D-Bus or expose it elsewhere.
3553 * Register Mimo USB Screens as suitable for automatic seat
3556 * Read SELinux client context from journal clients in a race
3559 * Reorder configuration file lookup order. /etc now always
3560 overrides /run in order to allow the administrator to always
3561 and unconditionally override vendor supplied or
3562 automatically generated data.
3564 * The various user visible bits of the journal now have man
3565 pages. We still lack man pages for the journal API calls
3568 * We now ship all man pages in HTML format again in the
3571 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3572 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3573 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3578 * This is mostly a bugfix release
3580 * systems lacking /etc/os-release are no longer supported.
3582 * Various functionality updates to libsystemd-login.so
3584 * Track class of PAM logins to distuingish greeters from
3587 Contributions from: Kay Sievers, Lennart Poettering, Michael
3592 * This is an important bugfix release for v41.
3594 * Building man pages is now optional which should be useful
3595 for those building systemd from git but unwilling to install
3598 * Watchdog support for supervising services is now usable. In
3599 a future release support for hardware watchdogs
3600 (i.e. /dev/watchdog) will be added building on this.
3602 * Service start rate limiting is now configurable and can be
3603 turned off per service. When a start rate limit is hit a
3604 reboot can automatically be triggered.
3606 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3608 Contributions from: Benjamin Franzke, Bill Nottingham,
3609 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3610 Schmidt, Michał Górny, Piotr Drąg
3614 * The systemd binary is installed /usr/lib/systemd/systemd now;
3615 An existing /sbin/init symlink needs to be adapted with the
3618 * The code that loads kernel modules has been ported to invoke
3619 libkmod directly, instead of modprobe. This means we do not
3620 support systems with module-init-tools anymore.
3622 * Watchdog support is now already useful, but still not
3625 * A new kernel command line option systemd.setenv= is
3626 understood to set system wide environment variables
3627 dynamically at boot.
3629 * We now limit the set of capabilities of systemd-journald.
3631 * We now set SIGPIPE to ignore by default, since it only is
3632 useful in shell pipelines, and has little use in general
3633 code. This can be disabled with IgnoreSIPIPE=no in unit
3636 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3637 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3642 * This is mostly a bugfix release
3644 * We now expose the reason why a service failed in the
3645 "Result" D-Bus property.
3647 * Rudimentary service watchdog support (will be completed over
3648 the next few releases.)
3650 * When systemd forks off in order execute some service we will
3651 now immediately changes its argv[0] to reflect which process
3652 it will execute. This is useful to minimize the time window
3653 with a generic argv[0], which makes bootcharts more useful
3655 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3656 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3657 Mike Kazantsev, Ray Strode
3661 * This is mostly a test release, but incorporates many
3664 * New systemd-cgtop tool to show control groups by their
3667 * Linking against libacl for ACLs is optional again. If
3668 disabled, support tracking device access for active logins
3669 goes becomes unavailable, and so does access to the user
3670 journals by the respective users.
3672 * If a group "adm" exists, journal files are automatically
3673 owned by them, thus allow members of this group full access
3674 to the system journal as well as all user journals.
3676 * The journal now stores the SELinux context of the logging
3677 client for all entries.
3679 * Add C++ inclusion guards to all public headers
3681 * New output mode "cat" in the journal to print only text
3682 messages, without any meta data like date or time.
3684 * Include tiny X server wrapper as a temporary stop-gap to
3685 teach XOrg udev display enumeration. This is used by display
3686 managers such as gdm, and will go away as soon as XOrg
3687 learned native udev hotplugging for display devices.
3689 * Add new systemd-cat tool for executing arbitrary programs
3690 with STDERR/STDOUT connected to the journal. Can also act as
3691 BSD logger replacement, and does so by default.
3693 * Optionally store all locally generated coredumps in the
3694 journal along with meta data.
3696 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3697 writing short strings to files (for usage for /sys), and for
3698 creating symlinks, character and block device nodes.
3700 * New unit file option ControlGroupPersistent= to make cgroups
3701 persistent, following the mechanisms outlined in
3702 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3704 * Support multiple local RTCs in a sane way
3706 * No longer monopolize IO when replaying readahead data on
3707 rotating disks, since we might starve non-file-system IO to
3708 death, since fanotify() will not see accesses done by blkid,
3711 * Do not show kernel threads in systemd-cgls anymore, unless
3712 requested with new -k switch.
3714 Contributions from: Dan Horák, Kay Sievers, Lennart
3715 Poettering, Michal Schmidt
3719 * This is mostly a test release, but incorporates many
3722 * The git repository moved to:
3723 git://anongit.freedesktop.org/systemd/systemd
3724 ssh://git.freedesktop.org/git/systemd/systemd
3726 * First release with the journal
3727 http://0pointer.de/blog/projects/the-journal.html
3729 * The journal replaces both systemd-kmsg-syslogd and
3730 systemd-stdout-bridge.
3732 * New sd_pid_get_unit() API call in libsystemd-logind
3734 * Many systemadm clean-ups
3736 * Introduce remote-fs-pre.target which is ordered before all
3737 remote mounts and may be used to start services before all
3740 * Added Mageia support
3742 * Add bash completion for systemd-loginctl
3744 * Actively monitor PID file creation for daemons which exit in
3745 the parent process before having finished writing the PID
3746 file in the daemon process. Daemons which do this need to be
3747 fixed (i.e. PID file creation must have finished before the
3748 parent exits), but we now react a bit more gracefully to them.
3750 * Add colourful boot output, mimicking the well-known output
3751 of existing distributions.
3753 * New option PassCredentials= for socket units, for
3754 compatibility with a recent kernel ABI breakage.
3756 * /etc/rc.local is now hooked in via a generator binary, and
3757 thus will no longer act as synchronization point during
3760 * systemctl list-unit-files now supports --root=.
3762 * systemd-tmpfiles now understands two new commands: z, Z for
3763 relabelling files according to the SELinux database. This is
3764 useful to apply SELinux labels to specific files in /sys,
3767 * Output of SysV services is now forwarded to both the console
3768 and the journal by default, not only just the console.
3770 * New man pages for all APIs from libsystemd-login.
3772 * The build tree got reorganized and a the build system is a
3773 lot more modular allowing embedded setups to specifically
3774 select the components of systemd they are interested in.
3776 * Support for Linux systems lacking the kernel VT subsystem is
3779 * configure's --with-rootdir= got renamed to
3780 --with-rootprefix= to follow the naming used by udev and
3783 * Unless specified otherwise we will now install to /usr instead
3784 of /usr/local by default.
3786 * Processes with '@' in argv[0][0] are now excluded from the
3787 final shut-down killing spree, following the logic explained
3789 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
3791 * All processes remaining in a service cgroup when we enter
3792 the START or START_PRE states are now killed with
3793 SIGKILL. That means it is no longer possible to spawn
3794 background processes from ExecStart= lines (which was never
3795 supported anyway, and bad style).
3797 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
3798 reloading of units together.
3800 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
3801 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
3802 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
3803 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
3804 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek