1 systemd System and Service Manager
5 * journalctl gained option -t/--identifier to match on the
6 syslog identifier and --utc option to show timestamps in UTC
7 timezone. journalctl now also accepts -n/--lines=all to
8 disable line cap under a pager.
10 * Services can notify the manager before they start a reload
11 (by sending RELOADING=1) or shutdown (by sending
12 STOPPING=1). This allows the manager to show the
13 internal state of daemons and closes a race condition
14 when the process is still running but has closed its
17 * Services with Type=oneshot do not have to have any
20 * User units are now loaded also from
21 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
22 /run/systemd/user directory that was already previously
23 supported, but is under the control of the user.
25 * A timeout for the start of the system can be configured. The
26 system can be configured to reboot or poweroff if the basic
27 system default target is not reached before the timeout (new
28 StartTimeoutSec=, StartTimeoutAction=,
29 StartTimeoutRebootArgument= options).
31 * systemd-logind can be configured to also handle lid switch
32 events even when the machine is docked or multiple displays
33 are attached (HandleLidSwitchDocked= option).
35 * A helper binary and a service have been added which can be
36 used to resume from hibernation in the initramfs. A
37 generator will parse the resume= option on the kernel
38 command-line to trigger resume.
40 * systemd-evcat input debugging tool has been added. It will
41 print keys read for the input system and show modifier
44 * systemd-modeset graphics debugging tool will show undulating
45 rainbows on all connected displays.
47 * A user console daemon systemd-consoled has been added. It is
48 a preview, and will so far open a single terminal on each
49 session of the user marked as Desktop=SYSTEMD-CONSOLE.
51 * Route metrics can be specified for DHCP routes added by
54 * SELinux context of socket-actived services can be set from
55 the information provided by the remote peer
56 (SELinuxContextFromNet= option).
58 * Userspace firmware loading support has been removed and
59 the minimum supported kernel version is thus bumped to 3.7.
61 * Timeout for udev workers has been increased from 1 to 3
62 minutes, but a warning will be printed after 1 minute to
63 help diagnose kernel modules that take a long time to load.
65 * Udev rules can now remove tags with TAG-="foobar".
67 * Readhead implementation has been removed. In many
68 circumstatances it didn't give expected benefits even for
69 rotational disk drives and was becoming less relevant in the
72 * Swap units can use Discard= to specify discard options.
73 Discard options specified for swaps in /etc/fstab are now
76 * Docker containers are now detected as a separate type of
79 * The Password Agent protocol gained support for queries where
80 the user input is shown, useful e.g. for usernames.
81 systemd-ask-password gained a new --echo option to turn
86 * timedated no longer reads NTP implementation unit names from
87 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
88 implementations should add a
90 Conflicts=systemd-timesyncd.service
92 to their unit files to take over and replace systemd's NTP
93 default functionality.
95 * systemd-sysusers gained a new line type "r" for configuring
96 which UID/GID ranges to allocate system users/groups
97 from. Lines of type "u" may now add an additional column
98 that specifies the home directory for the system user to be
99 created. Also, systemd-sysusers may now optionally read user
100 information from STDIN instead of a file. This is useful for
101 invoking it from RPM preinst scriptlets that need to create
102 users before the first RPM file is installed since these
103 files might need to be owned by them. A new
104 %sysusers_create_inline RPM macro has been introduced to do
105 just that. systemd-sysusers now updates the shadow files as
106 well as the user/group databases, which should enhance
107 compatibility with certain tools like grpck.
109 * A number of bus APIs of PID 1 now optionally consult
110 PolicyKit to permit access for otherwise unprivileged
111 clients under certain conditions. Note that this currently
112 doesn't support interactive authentication yet, but this is
113 expected to be added eventually, too.
115 * /etc/machine-info now has new fields for configuring the
116 deployment environment of the machine, as well as the
117 location of the machine. hostnamectl has been updated with
118 new command to update these fields.
120 * systemd-timesyncd has been updated to automatically acquire
121 NTP server information from systemd-networkd, which might
122 have been discovered via DHCP.
124 * systemd-resolved now includes a caching DNS stub resolver
125 and a complete LLMNR name resolution implementation. A new
126 NSS module "nss-resolve" has been added which can be used
127 instead of glibc's own "nss-dns" to resolve hostnames via
128 systemd-resolved. Hostnames, addresses and arbitrary RRs may
129 be resolved via systemd-resolved D-Bus APIs. In contrast to
130 the glibc internal resolver systemd-resolved is aware of
131 multi-homed system, and keeps DNS server and caches separate
132 and per-interface. Queries are sent simultaneously on all
133 interfaces that have DNS servers configured, in order to
134 properly handle VPNs and local LANs which might resolve
135 separate sets of domain names. systemd-resolved may acquire
136 DNS server information from systemd-networkd automatically,
137 which in turn might have discovered them via DHCP. A tool
138 "systemd-resolve-host" has been added that may be used to
139 query the DNS logic in resolved. systemd-resolved implements
140 IDNA and automatically uses IDNA or UTF-8 encoding depending
141 on whether classic DNS or LLMNR is used as transport. In the
142 next releases we intend to add a DNSSEC and mDNS/DNS-SD
143 implementation to systemd-resolved.
145 * A new NSS module nss-mymachines has been added, that
146 automatically resolves the names of all local registered
147 containers to their respective IP addresses.
149 * A new client tool "networkctl" for systemd-networkd has been
150 added. It currently is entirely passive and will query
151 networking configuration from udev, rtnetlink and networkd,
152 and present it to the user in a very friendly
153 way. Eventually, we hope to extend it to become a full
154 control utility for networkd.
156 * .socket units gained a new DeferAcceptSec= setting that
157 controls the kernels' TCP_DEFER_ACCEPT sockopt for
158 TCP. Similar, support for controlling TCP keep-alive
159 settings has been added (KeepAliveTimeSec=,
160 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
161 turning off Nagle's algorithm on TCP has been added
164 * logind learned a new session type "web", for use in projects
165 like Cockpit which register web clients as PAM sessions.
167 * timer units with at least one OnCalendar= setting will now
168 be started only after timer-sync.target has been
169 reached. This way they will not elapse before the system
170 clock has been corrected by a local NTP client or
171 similar. This is particular useful on RTC-less embedded
172 machines, that come up with an invalid system clock.
174 * systemd-nspawn's --network-veth= switch should now result in
175 stable MAC addresses for both the outer and the inner side
178 * systemd-nspawn gained a new --volatile= switch for running
179 container instances with /etc or /var unpopulated.
181 * The kdbus client code has been updated to use the new Linux
182 3.17 memfd subsystem instead of the old kdbus-specific one.
184 * systemd-networkd's DHCP client and server now support
185 FORCERENEW. There are also new configuration options to
186 configure the vendor client identifier and broadcast mode
189 * systemd will no longer inform the kernel about the current
190 timezone, as this is necessarily incorrect and racy as the
191 kernel has no understanding of DST and similar
192 concepts. This hence means FAT timestamps will be always
193 considered UTC, similar to what Android is already
194 doing. Also, when the RTC is configured to the local time
195 (rather than UTC) systemd will never synchronize back to it,
196 as this might confuse Windows at a later boot.
198 * systemd-analyze gained a new command "verify" for offline
199 validation of unit files.
201 * systemd-networkd gained support for a couple of additional
202 settings for bonding networking setups. Also, the metric for
203 statically configured routes may now be configured. For
204 network interfaces where this is appropriate the peer IP
205 address may now be configured.
207 * systemd-networkd's DHCP client will no longer request
208 broadcasting by default, as this tripped up some networks.
209 For hardware where broadcast is required the feature should
210 be switched back on using RequestBroadcast=yes.
212 * systemd-networkd will now set up IPv4LL addresses (when
213 enabled) even if DHCP is configured successfully.
215 * udev will now default to respect network device names given
216 by the kernel when the kernel indicates that these are
217 predictable. This behavior can be tweaked by changing
218 NamePolicy= in the relevant .link file.
220 * A new library systemd-terminal has been added that
221 implements full TTY stream parsing and rendering. This
222 library is supposed to be used later on for implementing a
223 full userspace VT subsystem, replacing the current kernel
226 * A new tool systemd-journal-upload has been added to push
227 journal data to a remote system running
228 systemd-journal-remote.
230 * journald will no longer forward all local data to another
231 running syslog daemon. This change has been made because
232 rsyslog (which appears to be the most commonly used syslog
233 implementation these days) no longer makes use of this, and
234 instead pulls the data out of the journal on its own. Since
235 forwarding the messages to a non-existent syslog server is
236 more expensive than we assumed we have now turned this
237 off. If you run a syslog server that is not a recent rsyslog
238 version, you have to turn this option on again
239 (ForwardToSyslog= in journald.conf).
241 * journald now optionally supports the LZ4 compressor for
242 larger journal fields. This compressor should perform much
243 better than XZ which was the previous default.
245 * machinectl now shows the IP addresses of local containers,
246 if it knows them, plus the interface name of the container.
248 * A new tool "systemd-escape" has been added that makes it
249 easy to escape strings to build unit names and similar.
251 * sd_notify() messages may now include a new ERRNO= field
252 which is parsed and collected by systemd and shown among the
253 "systemctl status" output for a service.
255 * A new component "systemd-firstboot" has been added that
256 queries the most basic systemd information (timezone,
257 hostname, root password) interactively on first
258 boot. Alternatively it may also be used to provision these
259 things offline on OS images installed into directories.
261 * The default sysctl.d/ snippets will now set
263 net.ipv4.conf.default.promote_secondaries=1
265 This has the benefit of no flushing secondary IP addresses
266 when primary addresses are removed.
268 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
269 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
270 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
271 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
272 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
273 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
274 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
275 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
276 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
277 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
278 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
279 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
280 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
281 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
282 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
284 -- Berlin, 2014-08-19
288 * A new tool systemd-sysusers has been added. This tool
289 creates system users and groups in /etc/passwd and
290 /etc/group, based on static declarative system user/group
291 definitions in /usr/lib/sysusers.d/. This is useful to
292 enable factory resets and volatile systems that boot up with
293 an empty /etc directory, and thus need system users and
294 groups created during early boot. systemd now also ships
295 with two default sysusers.d/ files for the most basic
296 users and groups systemd and the core operating system
299 * A new tmpfiles snippet has been added that rebuilds the
300 essential files in /etc on boot, should they be missing.
302 * A directive for ensuring automatic clean-up of
303 /var/cache/man/ has been removed from the default
304 configuration. This line should now be shipped by the man
305 implementation. The necessary change has been made to the
306 man-db implementation. Note that you need to update your man
307 implementation to one that ships this line, otherwise no
308 automatic clean-up of /var/cache/man will take place.
310 * A new condition ConditionNeedsUpdate= has been added that
311 may conditionalize services to only run when /etc or /var
312 are "older" than the vendor operating system resources in
313 /usr. This is useful for reconstructing or updating /etc
314 after an offline update of /usr or a factory reset, on the
315 next reboot. Services that want to run once after such an
316 update or reset should use this condition and order
317 themselves before the new systemd-update-done.service, which
318 will mark the two directories as fully updated. A number of
319 service files have been added making use of this, to rebuild
320 the udev hardware database, the journald message catalog and
321 dynamic loader cache (ldconfig). The systemd-sysusers tool
322 described above also makes use of this now. With this in
323 place it is now possible to start up a minimal operating
324 system with /etc empty cleanly. For more information on the
325 concepts involved see this recent blog story:
327 http://0pointer.de/blog/projects/stateless.html
329 * A new system group "input" has been introduced, and all
330 input device nodes get this group assigned. This is useful
331 for system-level software to get access to input devices. It
332 complements what is already done for "audio" and "video".
334 * systemd-networkd learnt minimal DHCPv4 server support in
335 addition to the existing DHCPv4 client support. It also
336 learnt DHCPv6 client and IPv6 Router Solicitation client
337 support. The DHCPv4 client gained support for static routes
338 passed in from the server. Note that the [DHCPv4] section
339 known in older systemd-networkd versions has been renamed to
340 [DHCP] and is now also used by the DHCPv6 client. Existing
341 .network files using settings of this section should be
342 updated, though compatibility is maintained. Optionally, the
343 client hostname may now be sent to the DHCP server.
345 * networkd gained support for vxlan virtual networks as well
346 as tun/tap and dummy devices.
348 * networkd gained support for automatic allocation of address
349 ranges for interfaces from a system-wide pool of
350 addresses. This is useful for dynamically managing a large
351 number of interfaces with a single network configuration
352 file. In particular this is useful to easily assign
353 appropriate IP addresses to the veth links of a large number
356 * RPM macros for processing sysusers, sysctl and binfmt
357 drop-in snippets at package installation time have been
360 * The /etc/os-release file should now be placed in
361 /usr/lib/os-release. The old location is automatically
362 created as symlink. /usr/lib is the more appropriate
363 location of this file, since it shall actually describe the
364 vendor operating system shipped in /usr, and not the
365 configuration stored in /etc.
367 * .mount units gained a new boolean SloppyOptions= setting
368 that maps to mount(8)'s -s option which enables permissive
369 parsing of unknown mount options.
371 * tmpfiles learnt a new "L+" directive which creates a symlink
372 but (unlike "L") deletes a pre-existing file first, should
373 it already exist and not already be the correct
374 symlink. Similar, "b+", "c+" and "p+" directives have been
375 added as well, which create block and character devices, as
376 well as fifos in the filesystem, possibly removing any
377 pre-existing files of different types.
379 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
380 'argument' field (which so far specified the source to
381 symlink/copy the files from) is now optional. If omitted the
382 same file os copied from /usr/share/factory/ suffixed by the
383 full destination path. This is useful for populating /etc
384 with essential files, by copying them from vendor defaults
385 shipped in /usr/share/factory/etc.
387 * A new command "systemctl preset-all" has been added that
388 applies the service preset settings to all installed unit
389 files. A new switch --preset-mode= has been added that
390 controls whether only enable or only disable operations
393 * A new command "systemctl is-system-running" has been added
394 that allows checking the overall state of the system, for
395 example whether it is fully up and running.
397 * When the system boots up with an empty /etc, the equivalent
398 to "systemctl preset-all" is executed during early boot, to
399 make sure all default services are enabled after a factory
402 * systemd now contains a minimal preset file that enables the
403 most basic services systemd ships by default.
405 * Unit files' [Install] section gained a new DefaultInstance=
406 field for defining the default instance to create if a
407 template unit is enabled with no instance specified.
409 * A new passive target cryptsetup-pre.target has been added
410 that may be used by services that need to make they run and
411 finish before the first LUKS cryptographic device is set up.
413 * The /dev/loop-control and /dev/btrfs-control device nodes
414 are now owned by the "disk" group by default, opening up
415 access to this group.
417 * systemd-coredump will now automatically generate a
418 stack trace of all core dumps taking place on the system,
419 based on elfutils' libdw library. This stack trace is logged
422 * systemd-coredump may now optionally store coredumps directly
423 on disk (in /var/lib/systemd/coredump, possibly compressed),
424 instead of storing them unconditionally in the journal. This
425 mode is the new default. A new configuration file
426 /etc/systemd/coredump.conf has been added to configure this
427 and other parameters of systemd-coredump.
429 * coredumpctl gained a new "info" verb to show details about a
430 specific coredump. A new switch "-1" has also been added
431 that makes sure to only show information about the most
432 recent entry instead of all entries. Also, as the tool is
433 generally useful now the "systemd-" prefix of the binary
434 name has been removed. Distributions that want to maintain
435 compatibility with the old name should add a symlink from
436 the old name to the new name.
438 * journald's SplitMode= now defaults to "uid". This makes sure
439 that unprivileged users can access their own coredumps with
440 coredumpctl without restrictions.
442 * New kernel command line options "systemd.wants=" (for
443 pulling an additional unit during boot), "systemd.mask="
444 (for masking a specific unit for the boot), and
445 "systemd.debug-shell" (for enabling the debug shell on tty9)
446 have been added. This is implemented in the new generator
447 "systemd-debug-generator".
449 * systemd-nspawn will now by default filter a couple of
450 syscalls for containers, among them those required for
451 kernel module loading, direct x86 IO port access, swap
452 management, and kexec. Most importantly though
453 open_by_handle_at() is now prohibited for containers,
454 closing a hole similar to a recently discussed vulnerability
455 in docker regarding access to files on file hierarchies the
456 container should normally not have access to. Note that for
457 nspawn we generally make no security claims anyway (and
458 this is explicitly documented in the man page), so this is
459 just a fix for one of the most obvious problems.
461 * A new man page file-hierarchy(7) has been added that
462 contains a minimized, modernized version of the file system
463 layout systemd expects, similar in style to the FHS
464 specification or hier(5). A new tool systemd-path(1) has
465 been added to query many of these paths for the local
468 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
469 longer done. Since the directory now has a per-user size
470 limit, and is cleaned on logout this appears unnecessary,
471 in particular since this now brings the lifecycle of this
472 directory closer in line with how IPC objects are handled.
474 * systemd.pc now exports a number of additional directories,
475 including $libdir (which is useful to identify the library
476 path for the primary architecture of the system), and a
477 couple of drop-in directories.
479 * udev's predictable network interface names now use the dev_port
480 sysfs attribute, introduced in linux 3.15 instead of dev_id to
481 distinguish between ports of the same PCI function. dev_id should
482 only be used for ports using the same HW address, hence the need
485 * machined has been updated to export the OS version of a
486 container (read from /etc/os-release and
487 /usr/lib/os-release) on the bus. This is now shown in
488 "machinectl status" for a machine.
490 * A new service setting RestartForceExitStatus= has been
491 added. If configured to a set of exit signals or process
492 return values, the service will be restarted when the main
493 daemon process exits with any of them, regardless of the
496 * systemctl's -H switch for connecting to remote systemd
497 machines has been extended so that it may be used to
498 directly connect to a specific container on the
499 host. "systemctl -H root@foobar:waldi" will now connect as
500 user "root" to host "foobar", and then proceed directly to
501 the container named "waldi". Note that currently you have to
502 authenticate as user "root" for this to work, as entering
503 containers is a privileged operation.
505 Contributions from: Andreas Henriksson, Benjamin Steinwender,
506 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
507 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
508 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
509 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
510 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
511 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
512 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
513 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
514 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
515 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
516 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
518 -- Berlin, 2014-07-03
522 * As an experimental feature, udev now tries to lock the
523 disk device node (flock(LOCK_SH|LOCK_NB)) while it
524 executes events for the disk or any of its partitions.
525 Applications like partitioning programs can lock the
526 disk device node (flock(LOCK_EX)) and claim temporary
527 device ownership that way; udev will entirely skip all event
528 handling for this disk and its partitions. If the disk
529 was opened for writing, the close will trigger a partition
530 table rescan in udev's "watch" facility, and if needed
531 synthesize "change" events for the disk and all its partitions.
532 This is now unconditionally enabled, and if it turns out to
533 cause major problems, we might turn it on only for specific
534 devices, or might need to disable it entirely. Device Mapper
535 devices are excluded from this logic.
537 * We temporarily dropped the "-l" switch for fsck invocations,
538 since they collide with the flock() logic above. util-linux
539 upstream has been changed already to avoid this conflict,
540 and we will readd "-l" as soon as util-linux with this
541 change has been released.
543 * The dependency on libattr has been removed. Since a long
544 time, the extended attribute calls have moved to glibc, and
545 libattr is thus unnecessary.
547 * Virtualization detection works without priviliges now. This
548 means the systemd-detect-virt binary no longer requires
549 CAP_SYS_PTRACE file capabilities, and our daemons can run
550 with fewer privileges.
552 * systemd-networkd now runs under its own "systemd-network"
553 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
554 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
555 loses the ability to write to files owned by root this way.
557 * Similar, systemd-resolved now runs under its own
558 "systemd-resolve" user with no capabilities remaining.
560 * Similar, systemd-bus-proxyd now runs under its own
561 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
563 * systemd-networkd gained support for setting up "veth"
564 virtual ethernet devices for container connectivity, as well
565 as GRE and VTI tunnels.
567 * systemd-networkd will no longer automatically attempt to
568 manually load kernel modules necessary for certain tunnel
569 transports. Instead, it is assumed the kernel loads them
570 automatically when required. This only works correctly on
571 very new kernels. On older kernels, please consider adding
572 the kernel modules to /etc/modules-load.d/ as a work-around.
574 * The resolv.conf file systemd-resolved generates has been
575 moved to /run/systemd/resolve/. If you have a symlink from
576 /etc/resolv.conf, it might be necessary to correct it.
578 * Two new service settings, ProtectHome= and ProtectSystem=,
579 have been added. When enabled, they will make the user data
580 (such as /home) inaccessible or read-only and the system
581 (such as /usr) read-only, for specific services. This allows
582 very light-weight per-service sandboxing to avoid
583 modifications of user data or system files from
584 services. These two new switches have been enabled for all
585 of systemd's long-running services, where appropriate.
587 * Socket units gained new SocketUser= and SocketGroup=
588 settings to set the owner user and group of AF_UNIX sockets
589 and FIFOs in the file system.
591 * Socket units gained a new RemoveOnStop= setting. If enabled,
592 all FIFOS and sockets in the file system will be removed
593 when the specific socket unit is stopped.
595 * Socket units gained a new Symlinks= setting. It takes a list
596 of symlinks to create to file system sockets or FIFOs
597 created by the specific Unix sockets. This is useful to
598 manage symlinks to socket nodes with the same life-cycle as
601 * The /dev/log socket and /dev/initctl FIFO have been moved to
602 /run, and have been replaced by symlinks. This allows
603 connecting to these facilities even if PrivateDevices=yes is
604 used for a service (which makes /dev/log itself unavailable,
605 but /run is left). This also has the benefit of ensuring
606 that /dev only contains device nodes, directories and
607 symlinks, and nothing else.
609 * sd-daemon gained two new calls sd_pid_notify() and
610 sd_pid_notifyf(). They are similar to sd_notify() and
611 sd_notifyf(), but allow overriding of the source PID of
612 notification messages if permissions permit this. This is
613 useful to send notify messages on behalf of a different
614 process (for example, the parent process). The
615 systemd-notify tool has been updated to make use of this
616 when sending messages (so that notification messages now
617 originate from the shell script invoking systemd-notify and
618 not the systemd-notify process itself. This should minimize
619 a race where systemd fails to associate notification
620 messages to services when the originating process already
623 * A new "on-abnormal" setting for Restart= has been added. If
624 set, it will result in automatic restarts on all "abnormal"
625 reasons for a process to exit, which includes unclean
626 signals, core dumps, timeouts and watchdog timeouts, but
627 does not include clean and unclean exit codes or clean
628 signals. Restart=on-abnormal is an alternative for
629 Restart=on-failure for services that shall be able to
630 terminate and avoid restarts on certain errors, by
631 indicating so with an unclean exit code. Restart=on-failure
632 or Restart=on-abnormal is now the recommended setting for
633 all long-running services.
635 * If the InaccessibleDirectories= service setting points to a
636 mount point (or if there are any submounts contained within
637 it), it is now attempted to completely unmount it, to make
638 the file systems truly unavailable for the respective
641 * The ReadOnlyDirectories= service setting and
642 systemd-nspawn's --read-only parameter are now recursively
643 applied to all submounts, too.
645 * Mount units may now be created transiently via the bus APIs.
647 * The support for SysV and LSB init scripts has been removed
648 from the systemd daemon itself. Instead, it is now
649 implemented as a generator that creates native systemd units
650 from these scripts when needed. This enables us to remove a
651 substantial amount of legacy code from PID 1, following the
652 fact that many distributions only ship a very small number
653 of LSB/SysV init scripts nowadays.
655 * Priviliged Xen (dom0) domains are not considered
656 virtualization anymore by the virtualization detection
657 logic. After all, they generally have unrestricted access to
658 the hardware and usually are used to manage the unprivileged
661 * systemd-tmpfiles gained a new "C" line type, for copying
662 files or entire directories.
664 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
665 lines. So far, they have been non-globbing versions of the
666 latter, and have thus been redundant. In future, it is
667 recommended to only use "z". "m" has hence been removed
668 from the documentation, even though it stays supported.
670 * A tmpfiles snippet to recreate the most basic structure in
671 /var has been added. This is enough to create the /var/run →
672 /run symlink and create a couple of structural
673 directories. This allows systems to boot up with an empty or
674 volatile /var. Of course, while with this change, the core OS
675 now is capable with dealing with a volatile /var, not all
676 user services are ready for it. However, we hope that sooner
677 or later, many service daemons will be changed upstream so
678 that they are able to automatically create their necessary
679 directories in /var at boot, should they be missing. This is
680 the first step to allow state-less systems that only require
681 the vendor image for /usr to boot.
683 * systemd-nspawn has gained a new --tmpfs= switch to mount an
684 empty tmpfs instance to a specific directory. This is
685 particularly useful for making use of the automatic
686 reconstruction of /var (see above), by passing --tmpfs=/var.
688 * Access modes specified in tmpfiles snippets may now be
689 prefixed with "~", which indicates that they shall be masked
690 by whether the existing file or directory is currently
691 writable, readable or executable at all. Also, if specified,
692 the sgid/suid/sticky bits will be masked for all
695 * A new passive target unit "network-pre.target" has been
696 added which is useful for services that shall run before any
697 network is configured, for example firewall scripts.
699 * The "floppy" group that previously owned the /dev/fd*
700 devices is no longer used. The "disk" group is now used
701 instead. Distributions should probably deprecate usage of
704 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
705 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
706 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
707 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
708 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
709 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
712 -- Berlin, 2014-06-11
716 * A new "systemd-timesyncd" daemon has been added for
717 synchronizing the system clock across the network. It
718 implements an SNTP client. In contrast to NTP
719 implementations such as chrony or the NTP reference server,
720 this only implements a client side, and does not bother with
721 the full NTP complexity, focusing only on querying time from
722 one remote server and synchronizing the local clock to
723 it. Unless you intend to serve NTP to networked clients or
724 want to connect to local hardware clocks, this simple NTP
725 client should be more than appropriate for most
726 installations. The daemon runs with minimal privileges, and
727 has been hooked up with networkd to only operate when
728 network connectivity is available. The daemon saves the
729 current clock to disk every time a new NTP sync has been
730 acquired, and uses this to possibly correct the system clock
731 early at bootup, in order to accommodate for systems that
732 lack an RTC such as the Raspberry Pi and embedded devices,
733 and to make sure that time monotonically progresses on these
734 systems, even if it is not always correct. To make use of
735 this daemon, a new system user and group "systemd-timesync"
736 needs to be created on installation of systemd.
738 * The queue "seqnum" interface of libudev has been disabled, as
739 it was generally incompatible with device namespacing as
740 sequence numbers of devices go "missing" if the devices are
741 part of a different namespace.
743 * "systemctl list-timers" and "systemctl list-sockets" gained
744 a --recursive switch for showing units of these types also
745 for all local containers, similar in style to the already
746 supported --recursive switch for "systemctl list-units".
748 * A new RebootArgument= setting has been added for service
749 units, which may be used to specify a kernel reboot argument
750 to use when triggering reboots with StartLimitAction=.
752 * A new FailureAction= setting has been added for service
753 units which may be used to specify an operation to trigger
754 when a service fails. This works similarly to
755 StartLimitAction=, but unlike it, controls what is done
756 immediately rather than only after several attempts to
757 restart the service in question.
759 * hostnamed got updated to also expose the kernel name,
760 release, and version on the bus. This is useful for
761 executing commands like hostnamectl with the -H switch.
762 systemd-analyze makes use of this to properly display
763 details when running non-locally.
765 * The bootchart tool can now show cgroup information in the
768 * The CFS CPU quota cgroup attribute is now exposed for
769 services. The new CPUQuota= switch has been added for this
770 which takes a percentage value. Setting this will have the
771 result that a service may never get more CPU time than the
772 specified percentage, even if the machine is otherwise idle.
774 * systemd-networkd learned IPIP and SIT tunnel support.
776 * LSB init scripts exposing a dependency on $network will now
777 get a dependency on network-online.target rather than simply
778 network.target. This should bring LSB handling closer to
779 what it was on SysV systems.
781 * A new fsck.repair= kernel option has been added to control
782 how fsck shall deal with unclean file systems at boot.
784 * The (.ini) configuration file parser will now silently
785 ignore sections whose name begins with "X-". This may be
786 used to maintain application-specific extension sections in unit
789 * machined gained a new API to query the IP addresses of
790 registered containers. "machinectl status" has been updated
791 to show these addresses in its output.
793 * A new call sd_uid_get_display() has been added to the
794 sd-login APIs for querying the "primary" session of a
795 user. The "primary" session of the user is elected from the
796 user's sessions and generally a graphical session is
797 preferred over a text one.
799 * A minimal systemd-resolved daemon has been added. It
800 currently simply acts as a companion to systemd-networkd and
801 manages resolv.conf based on per-interface DNS
802 configuration, possibly supplied via DHCP. In the long run
803 we hope to extend this into a local DNSSEC enabled DNS and
806 * The systemd-networkd-wait-online tool is now enabled by
807 default. It will delay network-online.target until a network
808 connection has been configured. The tool primarily integrates
809 with networkd, but will also make a best effort to make sense
810 of network configuration performed in some other way.
812 * Two new service options StartupCPUShares= and
813 StartupBlockIOWeight= have been added that work similarly to
814 CPUShares= and BlockIOWeight= however only apply during
815 system startup. This is useful to prioritize certain services
816 differently during bootup than during normal runtime.
818 * hostnamed has been changed to prefer the statically
819 configured hostname in /etc/hostname (unless set to
820 'localhost' or empty) over any dynamic one supplied by
821 dhcp. With this change, the rules for picking the hostname
822 match more closely the rules of other configuration settings
823 where the local administrator's configuration in /etc always
824 overrides any other settings.
826 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
827 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
828 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
829 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
830 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
831 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
832 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
833 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
834 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
835 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
836 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
837 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
838 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
839 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
840 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
841 Lindskog, WaLyong Cho, Will Woods, Zbigniew
844 -- Beijing, 2014-05-28
848 * When restoring the screen brightness at boot, stay away from
849 the darkest setting or from the lowest 5% of the available
850 range, depending on which is the larger value of both. This
851 should effectively protect the user from rebooting into a
852 black screen, should the brightness have been set to minimum
855 * sd-login gained a new sd_machine_get_class() call to
856 determine the class ("vm" or "container") of a machine
857 registered with machined.
859 * sd-login gained new calls
860 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
861 to query the identity of the peer of a local AF_UNIX
862 connection. They operate similarly to their sd_pid_get_xyz()
865 * PID 1 will now maintain a system-wide system state engine
866 with the states "starting", "running", "degraded",
867 "maintenance", "stopping". These states are bound to system
868 startup, normal runtime, runtime with at least one failed
869 service, rescue/emergency mode and system shutdown. This
870 state is shown in the "systemctl status" output when no unit
871 name is passed. It is useful to determine system state, in
872 particularly when doing so for many systems or containers at
875 * A new command "list-machines" has been added to "systemctl"
876 that lists all local OS containers and shows their system
877 state (see above), if systemd runs inside of them.
879 * systemctl gained a new "-r" switch to recursively enumerate
880 units on all local containers, when used with the
881 "list-unit" command (which is the default one that is
882 executed when no parameters are specified).
884 * The GPT automatic partition discovery logic will now honour
885 two GPT partition flags: one may be set on a partition to
886 cause it to be mounted read-only, and the other may be set
887 on a partition to ignore it during automatic discovery.
889 * Two new GPT type UUIDs have been added for automatic root
890 partition discovery, for 32-bit and 64-bit ARM. This is not
891 particularly useful for discovering the root directory on
892 these architectures during bare-metal boots (since UEFI is
893 not common there), but still very useful to allow booting of
894 ARM disk images in nspawn with the -i option.
896 * MAC addresses of interfaces created with nspawn's
897 --network-interface= switch will now be generated from the
898 machine name, and thus be stable between multiple invocations
901 * logind will now automatically remove all IPC objects owned
902 by a user if she or he fully logs out. This makes sure that
903 users who are logged out cannot continue to consume IPC
904 resources. This covers SysV memory, semaphores and message
905 queues as well as POSIX shared memory and message
906 queues. Traditionally, SysV and POSIX IPC had no life-cycle
907 limits. With this functionality, that is corrected. This may
908 be turned off by using the RemoveIPC= switch of logind.conf.
910 * The systemd-machine-id-setup and tmpfiles tools gained a
911 --root= switch to operate on a specific root directory,
914 * journald can now forward logged messages to the TTYs of all
915 logged in users ("wall"). This is the default for all
916 emergency messages now.
918 * A new tool systemd-journal-remote has been added to stream
919 journal log messages across the network.
921 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
922 controller trees are mounted into it. Note that the
923 directories mounted beneath it are not read-only. This is a
924 security measure and is particularly useful because glibc
925 actually includes a search logic to pick any tmpfs it can
926 find to implement shm_open() if /dev/shm is not available
927 (which it might very well be in namespaced setups).
929 * machinectl gained a new "poweroff" command to cleanly power
930 down a local OS container.
932 * The PrivateDevices= unit file setting will now also drop the
933 CAP_MKNOD capability from the capability bound set, and
934 imply DevicePolicy=closed.
936 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
937 comprehensively on all long-running systemd services where
940 * systemd-udevd will now run in a disassociated mount
941 namespace. To mount directories from udev rules, make sure to
942 pull in mount units via SYSTEMD_WANTS properties.
944 * The kdbus support gained support for uploading policy into
945 the kernel. sd-bus gained support for creating "monitoring"
946 connections that can eavesdrop into all bus communication
947 for debugging purposes.
949 * Timestamps may now be specified in seconds since the UNIX
950 epoch Jan 1st, 1970 by specifying "@" followed by the value
953 * Native tcpwrap support in systemd has been removed. tcpwrap
954 is old code, not really maintained anymore and has serious
955 shortcomings, and better options such as firewalls
956 exist. For setups that require tcpwrap usage, please
957 consider invoking your socket-activated service via tcpd,
958 like on traditional inetd.
960 * A new system.conf configuration option
961 DefaultTimerAccuracySec= has been added that controls the
962 default AccuracySec= setting of .timer units.
964 * Timer units gained a new WakeSystem= switch. If enabled,
965 timers configured this way will cause the system to resume
966 from system suspend (if the system supports that, which most
969 * Timer units gained a new Persistent= switch. If enabled,
970 timers configured this way will save to disk when they have
971 been last triggered. This information is then used on next
972 reboot to possible execute overdue timer events, that
973 could not take place because the system was powered off.
974 This enables simple anacron-like behaviour for timer units.
976 * systemctl's "list-timers" will now also list the time a
977 timer unit was last triggered in addition to the next time
978 it will be triggered.
980 * systemd-networkd will now assign predictable IPv4LL
981 addresses to its local interfaces.
983 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
984 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
985 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
986 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
987 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
988 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
989 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
990 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
993 -- Berlin, 2014-03-25
997 * A new unit file setting RestrictAddressFamilies= has been
998 added to restrict which socket address families unit
999 processes gain access to. This takes address family names
1000 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1001 attack surface of services via exotic protocol stacks. This
1002 is built on seccomp system call filters.
1004 * Two new unit file settings RuntimeDirectory= and
1005 RuntimeDirectoryMode= have been added that may be used to
1006 manage a per-daemon runtime directories below /run. This is
1007 an alternative for setting up directory permissions with
1008 tmpfiles snippets, and has the advantage that the runtime
1009 directory's lifetime is bound to the daemon runtime and that
1010 the daemon starts up with an empty directory each time. This
1011 is particularly useful when writing services that drop
1012 privileges using the User= or Group= setting.
1014 * The DeviceAllow= unit setting now supports globbing for
1015 matching against device group names.
1017 * The systemd configuration file system.conf gained new
1018 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1019 DefaultMemoryAccounting= to globally turn on/off accounting
1020 for specific resources (cgroups) for all units. These
1021 settings may still be overridden individually in each unit
1024 * systemd-gpt-auto-generator is now able to discover /srv and
1025 root partitions in addition to /home and swap partitions. It
1026 also supports LUKS-encrypted partitions now. With this in
1027 place, automatic discovery of partitions to mount following
1028 the Discoverable Partitions Specification
1029 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1030 is now a lot more complete. This allows booting without
1031 /etc/fstab and without root= on the kernel command line on
1032 systems prepared appropriately.
1034 * systemd-nspawn gained a new --image= switch which allows
1035 booting up disk images and Linux installations on any block
1036 device that follow the Discoverable Partitions Specification
1037 (see above). This means that installations made with
1038 appropriately updated installers may now be started and
1039 deployed using container managers, completely
1040 unmodified. (We hope that libvirt-lxc will add support for
1041 this feature soon, too.)
1043 * systemd-nspawn gained a new --network-macvlan= setting to
1044 set up a private macvlan interface for the
1045 container. Similarly, systemd-networkd gained a new
1046 Kind=macvlan setting in .netdev files.
1048 * systemd-networkd now supports configuring local addresses
1051 * A new tool systemd-network-wait-online has been added to
1052 synchronously wait for network connectivity using
1055 * The sd-bus.h bus API gained a new sd_bus_track object for
1056 tracking the life-cycle of bus peers. Note that sd-bus.h is
1057 still not a public API though (unless you specify
1058 --enable-kdbus on the configure command line, which however
1059 voids your warranty and you get no API stability guarantee).
1061 * The $XDG_RUNTIME_DIR runtime directories for each user are
1062 now individual tmpfs instances, which has the benefit of
1063 introducing separate pools for each user, with individual
1064 size limits, and thus making sure that unprivileged clients
1065 can no longer negatively impact the system or other users by
1066 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1067 RuntimeDirectorySize= has been introduced that allows
1068 controlling the default size limit for all users. It
1069 defaults to 10% of the available physical memory. This is no
1070 replacement for quotas on tmpfs though (which the kernel
1071 still does not support), as /dev/shm and /tmp are still
1072 shared resources used by both the system and unprivileged
1075 * logind will now automatically turn off automatic suspending
1076 on laptop lid close when more than one display is
1077 connected. This was previously expected to be implemented
1078 individually in desktop environments (such as GNOME),
1079 however has been added to logind now, in order to fix a
1080 boot-time race where a desktop environment might not have
1081 been started yet and thus not been able to take an inhibitor
1082 lock at the time where logind already suspends the system
1083 due to a closed lid.
1085 * logind will now wait at least 30s after each system
1086 suspend/resume cycle, and 3min after system boot before
1087 suspending the system due to a closed laptop lid. This
1088 should give USB docking stations and similar enough time to
1089 be probed and configured after system resume and boot in
1090 order to then act as suspend blocker.
1092 * systemd-run gained a new --property= setting which allows
1093 initialization of resource control properties (and others)
1094 for the created scope or service unit. Example: "systemd-run
1095 --property=BlockIOWeight=10 updatedb" may be used to run
1096 updatedb at a low block IO scheduling weight.
1098 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1099 now also work in --scope mode.
1101 * When systemd is compiled with kdbus support, basic support
1102 for enforced policies is now in place. (Note that enabling
1103 kdbus still voids your warranty and no API compatibility
1106 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1107 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1108 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1109 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1110 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1111 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1112 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1113 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1114 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1115 Zbigniew Jędrzejewski-Szmek
1117 -- Berlin, 2014-03-12
1121 * systemd will now relabel /dev after loading the SMACK policy
1122 according to SMACK rules.
1124 * A new unit file option AppArmorProfile= has been added to
1125 set the AppArmor profile for the processes of a unit.
1127 * A new condition check ConditionArchitecture= has been added
1128 to conditionalize units based on the system architecture, as
1129 reported by uname()'s "machine" field.
1131 * systemd-networkd now supports matching on the system
1132 virtualization, architecture, kernel command line, host name
1135 * logind is now a lot more aggressive when suspending the
1136 machine due to a closed laptop lid. Instead of acting only
1137 on the lid close action, it will continuously watch the lid
1138 status and act on it. This is useful for laptops where the
1139 power button is on the outside of the chassis so that it can
1140 be reached without opening the lid (such as the Lenovo
1141 Yoga). On those machines, logind will now immediately
1142 re-suspend the machine if the power button has been
1143 accidentally pressed while the laptop was suspended and in a
1144 backpack or similar.
1146 * logind will now watch SW_DOCK switches and inhibit reaction
1147 to the lid switch if it is pressed. This means that logind
1148 will not suspend the machine anymore if the lid is closed
1149 and the system is docked, if the laptop supports SW_DOCK
1150 notifications via the input layer. Note that ACPI docking
1151 stations do not generate this currently. Also note that this
1152 logic is usually not fully sufficient and Desktop
1153 Environments should take a lid switch inhibitor lock when an
1154 external display is connected, as systemd will not watch
1157 * nspawn will now make use of the devices cgroup controller by
1158 default, and only permit creation of and access to the usual
1159 API device nodes like /dev/null or /dev/random, as well as
1160 access to (but not creation of) the pty devices.
1162 * We will now ship a default .network file for
1163 systemd-networkd that automatically configures DHCP for
1164 network interfaces created by nspawn's --network-veth or
1165 --network-bridge= switches.
1167 * systemd will now understand the usual M, K, G, T suffixes
1168 according to SI conventions (i.e. to the base 1000) when
1169 referring to throughput and hardware metrics. It will stay
1170 with IEC conventions (i.e. to the base 1024) for software
1171 metrics, according to what is customary according to
1172 Wikipedia. We explicitly document which base applies for
1173 each configuration option.
1175 * The DeviceAllow= setting in unit files now supports a syntax
1176 to whitelist an entire group of devices node majors at once,
1177 based on the /proc/devices listing. For example, with the
1178 string "char-pts", it is now possible to whitelist all
1179 current and future pseudo-TTYs at once.
1181 * sd-event learned a new "post" event source. Event sources of
1182 this type are triggered by the dispatching of any event
1183 source of a type that is not "post". This is useful for
1184 implementing clean-up and check event sources that are
1185 triggered by other work being done in the program.
1187 * systemd-networkd is no longer statically enabled, but uses
1188 the usual [Install] sections so that it can be
1189 enabled/disabled using systemctl. It still is enabled by
1192 * When creating a veth interface pair with systemd-nspawn, the
1193 host side will now be prefixed with "vb-" if
1194 --network-bridge= is used, and with "ve-" if --network-veth
1195 is used. This way, it is easy to distinguish these cases on
1196 the host, for example to apply different configuration to
1197 them with systemd-networkd.
1199 * The compatibility libraries for libsystemd-journal.so,
1200 libsystem-id128.so, libsystemd-login.so and
1201 libsystemd-daemon.so do not make use of IFUNC
1202 anymore. Instead, we now build libsystemd.so multiple times
1203 under these alternative names. This means that the footprint
1204 is drastically increased, but given that these are
1205 transitional compatibility libraries, this should not matter
1206 much. This change has been made necessary to support the ARM
1207 platform for these compatibility libraries, as the ARM
1208 toolchain is not really at the same level as the toolchain
1209 for other architectures like x86 and does not support
1210 IFUNC. Please make sure to use --enable-compat-libs only
1211 during a transitional period!
1213 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1214 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1215 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1216 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1217 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1218 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1219 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1220 Zbigniew Jędrzejewski-Szmek
1222 -- Berlin, 2014-02-24
1226 * A new component "systemd-networkd" has been added that can
1227 be used to configure local network interfaces statically or
1228 via DHCP. It is capable of bringing up bridges, VLANs, and
1229 bonding. Currently, no hook-ups for interactive network
1230 configuration are provided. Use this for your initrd,
1231 container, embedded, or server setup if you need a simple,
1232 yet powerful, network configuration solution. This
1233 configuration subsystem is quite nifty, as it allows wildcard
1234 hotplug matching in interfaces. For example, with a single
1235 configuration snippet, you can configure that all Ethernet
1236 interfaces showing up are automatically added to a bridge,
1237 or similar. It supports link-sensing and more.
1239 * A new tool "systemd-socket-proxyd" has been added which can
1240 act as a bidirectional proxy for TCP sockets. This is
1241 useful for adding socket activation support to services that
1242 do not actually support socket activation, including virtual
1243 machines and the like.
1245 * Add a new tool to save/restore rfkill state on
1248 * Save/restore state of keyboard backlights in addition to
1249 display backlights on shutdown/boot.
1251 * udev learned a new SECLABEL{} construct to label device
1252 nodes with a specific security label when they appear. For
1253 now, only SECLABEL{selinux} is supported, but the syntax is
1254 prepared for additional security frameworks.
1256 * udev gained a new scheme to configure link-level attributes
1257 from files in /etc/systemd/network/*.link. These files can
1258 match against MAC address, device path, driver name and type,
1259 and will apply attributes like the naming policy, link speed,
1260 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1261 address assignment policy (randomized, ...).
1263 * The configuration of network interface naming rules for
1264 "permanent interface names" has changed: a new NamePolicy=
1265 setting in the [Link] section of .link files determines the
1266 priority of possible naming schemes (onboard, slot, mac,
1267 path). The default value of this setting is determined by
1268 /usr/lib/net/links/99-default.link. Old
1269 80-net-name-slot.rules udev configuration file has been
1270 removed, so local configuration overriding this file should
1271 be adapated to override 99-default.link instead.
1273 * When the User= switch is used in a unit file, also
1274 initialize $SHELL= based on the user database entry.
1276 * systemd no longer depends on libdbus. All communication is
1277 now done with sd-bus, systemd's low-level bus library
1280 * kdbus support has been added to PID 1 itself. When kdbus is
1281 enabled, this causes PID 1 to set up the system bus and
1282 enable support for a new ".busname" unit type that
1283 encapsulates bus name activation on kdbus. It works a little
1284 bit like ".socket" units, except for bus names. A new
1285 generator has been added that converts classic dbus1 service
1286 activation files automatically into native systemd .busname
1289 * sd-bus: add a light-weight vtable implementation that allows
1290 defining objects on the bus with a simple static const
1291 vtable array of its methods, signals and properties.
1293 * systemd will not generate or install static dbus
1294 introspection data anymore to /usr/share/dbus-1/interfaces,
1295 as the precise format of these files is unclear, and
1296 nothing makes use of it.
1298 * A proxy daemon is now provided to proxy clients connecting
1299 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1300 compatibility with classic D-Bus.
1302 * A bus driver implementation has been added that supports the
1303 classic D-Bus bus driver calls on kdbus, also for
1304 compatibility purposes.
1306 * A new API "sd-event.h" has been added that implements a
1307 minimal event loop API built around epoll. It provides a
1308 couple of features that direct epoll usage is lacking:
1309 prioritization of events, scales to large numbers of timer
1310 events, per-event timer slack (accuracy), system-wide
1311 coalescing of timer events, exit handlers, watchdog
1312 supervision support using systemd's sd_notify() API, child
1315 * A new API "sd-rntl.h" has been added that provides an API
1316 around the route netlink interface of the kernel, similar in
1317 style to "sd-bus.h".
1319 * A new API "sd-dhcp-client.h" has been added that provides a
1320 small DHCPv4 client-side implementation. This is used by
1323 * There is a new kernel command line option
1324 "systemd.restore_state=0|1". When set to "0", none of the
1325 systemd tools will restore saved runtime state to hardware
1326 devices. More specifically, the rfkill and backlight states
1329 * The FsckPassNo= compatibility option in mount/service units
1330 has been removed. The fstab generator will now add the
1331 necessary dependencies automatically, and does not require
1332 PID1's support for that anymore.
1334 * journalctl gained a new switch, --list-boots, that lists
1335 recent boots with their times and boot IDs.
1337 * The various tools like systemctl, loginctl, timedatectl,
1338 busctl, systemd-run, ... have gained a new switch "-M" to
1339 connect to a specific, local OS container (as direct
1340 connection, without requiring SSH). This works on any
1341 container that is registered with machined, such as those
1342 created by libvirt-lxc or nspawn.
1344 * systemd-run and systemd-analyze also gained support for "-H"
1345 to connect to remote hosts via SSH. This is particularly
1346 useful for systemd-run because it enables queuing of jobs
1347 onto remote systems.
1349 * machinectl gained a new command "login" to open a getty
1350 login in any local container. This works with any container
1351 that is registered with machined (such as those created by
1352 libvirt-lxc or nspawn), and which runs systemd inside.
1354 * machinectl gained a new "reboot" command that may be used to
1355 trigger a reboot on a specific container that is registered
1356 with machined. This works on any container that runs an init
1357 system of some kind.
1359 * systemctl gained a new "list-timers" command to print a nice
1360 listing of installed timer units with the times they elapse
1363 * Alternative reboot() parameters may now be specified on the
1364 "systemctl reboot" command line and are passed to the
1365 reboot() system call.
1367 * systemctl gained a new --job-mode= switch to configure the
1368 mode to queue a job with. This is a more generic version of
1369 --fail, --irreversible, and --ignore-dependencies, which are
1370 still available but not advertised anymore.
1372 * /etc/systemd/system.conf gained new settings to configure
1373 various default timeouts of units, as well as the default
1374 start limit interval and burst. These may still be overridden
1377 * PID1 will now export on the bus profile data of the security
1378 policy upload process (such as the SELinux policy upload to
1381 * journald: when forwarding logs to the console, include
1382 timestamps (following the setting in
1383 /sys/module/printk/parameters/time).
1385 * OnCalendar= in timer units now understands the special
1386 strings "yearly" and "annually". (Both are equivalent)
1388 * The accuracy of timer units is now configurable with the new
1389 AccuracySec= setting. It defaults to 1min.
1391 * A new dependency type JoinsNamespaceOf= has been added that
1392 allows running two services within the same /tmp and network
1393 namespace, if PrivateNetwork= or PrivateTmp= are used.
1395 * A new command "cat" has been added to systemctl. It outputs
1396 the original unit file of a unit, and concatenates the
1397 contents of additional "drop-in" unit file snippets, so that
1398 the full configuration is shown.
1400 * systemctl now supports globbing on the various "list-xyz"
1401 commands, like "list-units" or "list-sockets", as well as on
1402 those commands which take multiple unit names.
1404 * journalctl's --unit= switch gained support for globbing.
1406 * All systemd daemons now make use of the watchdog logic so
1407 that systemd automatically notices when they hang.
1409 * If the $container_ttys environment variable is set,
1410 getty-generator will automatically spawn a getty for each
1411 listed tty. This is useful for container managers to request
1412 login gettys to be spawned on as many ttys as needed.
1414 * %h, %s, %U specifier support is not available anymore when
1415 used in unit files for PID 1. This is because NSS calls are
1416 not safe from PID 1. They stay available for --user
1417 instances of systemd, and as special case for the root user.
1419 * loginctl gained a new "--no-legend" switch to turn off output
1422 * The "sd-login.h" API gained three new calls:
1423 sd_session_is_remote(), sd_session_get_remote_user(),
1424 sd_session_get_remote_host() to query information about
1427 * The udev hardware database now also carries vendor/product
1428 information of SDIO devices.
1430 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1431 determine whether watchdog notifications are requested by
1434 * Socket-activated per-connection services now include a
1435 short description of the connection parameters in the
1438 * tmpfiles gained a new "--boot" option. When this is not used,
1439 only lines where the command character is not suffixed with
1440 "!" are executed. When this option is specified, those
1441 options are executed too. This partitions tmpfiles
1442 directives into those that can be safely executed at any
1443 time, and those which should be run only at boot (for
1444 example, a line that creates /run/nologin).
1446 * A new API "sd-resolve.h" has been added which provides a simple
1447 asynchronous wrapper around glibc NSS host name resolution
1448 calls, such as getaddrinfo(). In contrast to glibc's
1449 getaddrinfo_a(), it does not use signals. In contrast to most
1450 other asynchronous name resolution libraries, this one does
1451 not reimplement DNS, but reuses NSS, so that alternate
1452 host name resolution systems continue to work, such as mDNS,
1453 LDAP, etc. This API is based on libasyncns, but it has been
1454 cleaned up for inclusion in systemd.
1456 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1457 "sd-daemon.h" are no longer found in individual libraries
1458 libsystemd-journal.so, libsystemd-login.so,
1459 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1460 merged them into a single library, libsystemd.so, which
1461 provides all symbols. The reason for this is cyclic
1462 dependencies, as these libraries tend to use each other's
1463 symbols. So far, we have managed to workaround that by linking
1464 a copy of a good part of our code into each of these
1465 libraries again and again, which, however, makes certain
1466 things hard to do, like sharing static variables. Also, it
1467 substantially increases footprint. With this change, there
1468 is only one library for the basic APIs systemd
1469 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1470 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1471 library as well, however are subject to the --enable-kdbus
1472 switch (see below). Note that "sd-dhcp-client.h" is not part
1473 of this library (this is because it only consumes, never
1474 provides, services of/to other APIs). To make the transition
1475 easy from the separate libraries to the unified one, we
1476 provide the --enable-compat-libs compile-time switch which
1477 will generate stub libraries that are compatible with the
1478 old ones but redirect all calls to the new one.
1480 * All of the kdbus logic and the new APIs "sd-bus.h",
1481 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1482 and "sd-utf8.h" are compile-time optional via the
1483 "--enable-kdbus" switch, and they are not compiled in by
1484 default. To make use of kdbus, you have to explicitly enable
1485 the switch. Note however, that neither the kernel nor the
1486 userspace API for all of this is considered stable yet. We
1487 want to maintain the freedom to still change the APIs for
1488 now. By specifying this build-time switch, you acknowledge
1489 that you are aware of the instability of the current
1492 * Also, note that while kdbus is pretty much complete,
1493 it lacks one thing: proper policy support. This means you
1494 can build a fully working system with all features; however,
1495 it will be highly insecure. Policy support will be added in
1496 one of the next releases, at the same time that we will
1497 declare the APIs stable.
1499 * When the kernel command-line argument "kdbus" is specified,
1500 systemd will automatically load the kdbus.ko kernel module. At
1501 this stage of development, it is only useful for testing kdbus
1502 and should not be used in production. Note: if "--enable-kdbus"
1503 is specified, and the kdbus.ko kernel module is available, and
1504 "kdbus" is added to the kernel command line, the entire system
1505 runs with kdbus instead of dbus-daemon, with the above mentioned
1506 problem of missing the system policy enforcement. Also a future
1507 version of kdbus.ko or a newer systemd will not be compatible with
1508 each other, and will unlikely be able to boot the machine if only
1509 one of them is updated.
1511 * systemctl gained a new "import-environment" command which
1512 uploads the caller's environment (or parts thereof) into the
1513 service manager so that it is inherited by services started
1514 by the manager. This is useful to upload variables like
1515 $DISPLAY into the user service manager.
1517 * A new PrivateDevices= switch has been added to service units
1518 which allows running a service with a namespaced /dev
1519 directory that does not contain any device nodes for
1520 physical devices. More specifically, it only includes devices
1521 such as /dev/null, /dev/urandom, and /dev/zero which are API
1524 * logind has been extended to support behaviour like VT
1525 switching on seats that do not support a VT. This makes
1526 multi-session available on seats that are not the first seat
1527 (seat0), and on systems where kernel support for VTs has
1528 been disabled at compile-time.
1530 * If a process holds a delay lock for system sleep or shutdown
1531 and fails to release it in time, we will now log its
1532 identity. This makes it easier to identify processes that
1533 cause slow suspends or power-offs.
1535 * When parsing /etc/crypttab, support for a new key-slot=
1536 option as supported by Debian is added. It allows indicating
1537 which LUKS slot to use on disk, speeding up key loading.
1539 * The sd_journald_sendv() API call has been checked and
1540 officially declared to be async-signal-safe so that it may
1541 be invoked from signal handlers for logging purposes.
1543 * Boot-time status output is now enabled automatically after a
1544 short timeout if boot does not progress, in order to give
1545 the user an indication what she or he is waiting for.
1547 * The boot-time output has been improved to show how much time
1548 remains until jobs expire.
1550 * The KillMode= switch in service units gained a new possible
1551 value "mixed". If set, and the unit is shut down, then the
1552 initial SIGTERM signal is sent only to the main daemon
1553 process, while the following SIGKILL signal is sent to
1554 all remaining processes of the service.
1556 * When a scope unit is registered, a new property "Controller"
1557 may be set. If set to a valid bus name, systemd will send a
1558 RequestStop() signal to this name when it would like to shut
1559 down the scope. This may be used to hook manager logic into
1560 the shutdown logic of scope units. Also, scope units may now
1561 be put in a special "abandoned" state, in which case the
1562 manager process which created them takes no further
1563 responsibilities for it.
1565 * When reading unit files, systemd will now verify
1566 the access mode of these files, and warn about certain
1567 suspicious combinations. This has been added to make it
1568 easier to track down packaging bugs where unit files are
1569 marked executable or world-writable.
1571 * systemd-nspawn gained a new "--setenv=" switch to set
1572 container-wide environment variables. The similar option in
1573 systemd-activate was renamed from "--environment=" to
1574 "--setenv=" for consistency.
1576 * systemd-nspawn has been updated to create a new kdbus domain
1577 for each container that is invoked, thus allowing each
1578 container to have its own set of system and user buses,
1579 independent of the host.
1581 * systemd-nspawn gained a new --drop-capability= switch to run
1582 the container with less capabilities than the default. Both
1583 --drop-capability= and --capability= now take the special
1584 string "all" for dropping or keeping all capabilities.
1586 * systemd-nspawn gained new switches for executing containers
1587 with specific SELinux labels set.
1589 * systemd-nspawn gained a new --quiet switch to not generate
1590 any additional output but the container's own console
1593 * systemd-nspawn gained a new --share-system switch to run a
1594 container without PID namespacing enabled.
1596 * systemd-nspawn gained a new --register= switch to control
1597 whether the container is registered with systemd-machined or
1598 not. This is useful for containers that do not run full
1599 OS images, but only specific apps.
1601 * systemd-nspawn gained a new --keep-unit which may be used
1602 when invoked as the only program from a service unit, and
1603 results in registration of the unit service itself in
1604 systemd-machined, instead of a newly opened scope unit.
1606 * systemd-nspawn gained a new --network-interface= switch for
1607 moving arbitrary interfaces to the container. The new
1608 --network-veth switch creates a virtual Ethernet connection
1609 between host and container. The new --network-bridge=
1610 switch then allows assigning the host side of this virtual
1611 Ethernet connection to a bridge device.
1613 * systemd-nspawn gained a new --personality= switch for
1614 setting the kernel personality for the container. This is
1615 useful when running a 32-bit container on a 64-bit host. A
1616 similar option Personality= is now also available for service
1619 * logind will now also track a "Desktop" identifier for each
1620 session which encodes the desktop environment of it. This is
1621 useful for desktop environments that want to identify
1622 multiple running sessions of itself easily.
1624 * A new SELinuxContext= setting for service units has been
1625 added that allows setting a specific SELinux execution
1626 context for a service.
1628 * Most systemd client tools will now honour $SYSTEMD_LESS for
1629 settings of the "less" pager. By default, these tools will
1630 override $LESS to allow certain operations to work, such as
1631 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1632 influence this logic.
1634 * systemd's "seccomp" hook-up has been changed to make use of
1635 the libseccomp library instead of using its own
1636 implementation. This has benefits for portability among
1639 * For usage together with SystemCallFilter=, a new
1640 SystemCallErrorNumber= setting has been introduced that
1641 allows configuration of a system error number to be returned
1642 on filtered system calls, instead of immediately killing the
1643 process. Also, SystemCallArchitectures= has been added to
1644 limit access to system calls of a particular architecture
1645 (in order to turn off support for unused secondary
1646 architectures). There is also a global
1647 SystemCallArchitectures= setting in system.conf now to turn
1648 off support for non-native system calls system-wide.
1650 * systemd requires a kernel with a working name_to_handle_at(),
1651 please see the kernel config requirements in the README file.
1653 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1654 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1655 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1656 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1657 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1658 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1659 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1660 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1661 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1662 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1663 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1664 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1665 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1666 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1667 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1668 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1669 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1670 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1671 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1672 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1673 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1674 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1675 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1676 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1678 -- Berlin, 2014-02-20
1682 * logind has gained support for facilitating privileged input
1683 and drm device access for unprivileged clients. This work is
1684 useful to allow Wayland display servers (and similar
1685 programs, such as kmscon) to run under the user's ID and
1686 access input and drm devices which are normally
1687 protected. When this is used (and the kernel is new enough)
1688 logind will "mute" IO on the file descriptors passed to
1689 Wayland as long as it is in the background and "unmute" it
1690 if it returns into the foreground. This allows secure
1691 session switching without allowing background sessions to
1692 eavesdrop on input and display data. This also introduces
1693 session switching support if VT support is turned off in the
1694 kernel, and on seats that are not seat0.
1696 * A new kernel command line option luks.options= is understood
1697 now which allows specifying LUKS options for usage for LUKS
1698 encrypted partitions specified with luks.uuid=.
1700 * tmpfiles.d(5) snippets may now use specifier expansion in
1701 path names. More specifically %m, %b, %H, %v, are now
1702 replaced by the local machine id, boot id, hostname, and
1703 kernel version number.
1705 * A new tmpfiles.d(5) command "m" has been introduced which
1706 may be used to change the owner/group/access mode of a file
1707 or directory if it exists, but do nothing if it does not.
1709 * This release removes high-level support for the
1710 MemorySoftLimit= cgroup setting. The underlying kernel
1711 cgroup attribute memory.soft_limit= is currently badly
1712 designed and likely to be removed from the kernel API in its
1713 current form, hence we should not expose it for now.
1715 * The memory.use_hierarchy cgroup attribute is now enabled for
1716 all cgroups systemd creates in the memory cgroup
1717 hierarchy. This option is likely to be come the built-in
1718 default in the kernel anyway, and the non-hierarchial mode
1719 never made much sense in the intrinsically hierarchial
1722 * A new field _SYSTEMD_SLICE= is logged along with all journal
1723 messages containing the slice a message was generated
1724 from. This is useful to allow easy per-customer filtering of
1725 logs among other things.
1727 * systemd-journald will no longer adjust the group of journal
1728 files it creates to the "systemd-journal" group. Instead we
1729 rely on the journal directory to be owned by the
1730 "systemd-journal" group, and its setgid bit set, so that the
1731 kernel file system layer will automatically enforce that
1732 journal files inherit this group assignment. The reason for
1733 this change is that we cannot allow NSS look-ups from
1734 journald which would be necessary to resolve
1735 "systemd-journal" to a numeric GID, because this might
1736 create deadlocks if NSS involves synchronous queries to
1737 other daemons (such as nscd, or sssd) which in turn are
1738 logging clients of journald and might block on it, which
1739 would then dead lock. A tmpfiles.d(5) snippet included in
1740 systemd will make sure the setgid bit and group are
1741 properly set on the journal directory if it exists on every
1742 boot. However, we recommend adjusting it manually after
1743 upgrades too (or from RPM scriptlets), so that the change is
1744 not delayed until next reboot.
1746 * Backlight and random seed files in /var/lib/ have moved into
1747 the /var/lib/systemd/ directory, in order to centralize all
1748 systemd generated files in one directory.
1750 * Boot time performance measurements (as displayed by
1751 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1752 performance information if that's available to determine how
1753 much time BIOS and boot loader initialization required. With
1754 a sufficiently new BIOS you hence no longer need to boot
1755 with Gummiboot to get access to such information.
1757 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1758 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1759 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1760 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1761 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1762 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1763 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1765 -- Berlin, 2013-10-02
1769 * The Restart= option for services now understands a new
1770 on-watchdog setting, which will restart the service
1771 automatically if the service stops sending out watchdog keep
1772 alive messages (as configured with WatchdogSec=).
1774 * The getty generator (which is responsible for bringing up a
1775 getty on configured serial consoles) will no longer only
1776 start a getty on the primary kernel console but on all
1777 others, too. This makes the order in which console= is
1778 specified on the kernel command line less important.
1780 * libsystemd-logind gained a new sd_session_get_vt() call to
1781 retrieve the VT number of a session.
1783 * If the option "tries=0" is set for an entry of /etc/crypttab
1784 its passphrase is queried indefinitely instead of any
1785 maximum number of tries.
1787 * If a service with a configure PID file terminates its PID
1788 file will now be removed automatically if it still exists
1789 afterwards. This should put an end to stale PID files.
1791 * systemd-run will now also take relative binary path names
1792 for execution and no longer insists on absolute paths.
1794 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1795 paths that are optionally prefixed with "-" to indicate that
1796 it should not be considered a failure if they do not exist.
1798 * journalctl -o (and similar commands) now understands a new
1799 output mode "short-precise", it is similar to "short" but
1800 shows timestamps with usec accuracy.
1802 * The option "discard" (as known from Debian) is now
1803 synonymous to "allow-discards" in /etc/crypttab. In fact,
1804 "discard" is preferred now (since it is easier to remember
1807 * Some licensing clean-ups were made, so that more code is now
1808 LGPL-2.1 licensed than before.
1810 * A minimal tool to save/restore the display backlight
1811 brightness across reboots has been added. It will store the
1812 backlight setting as late as possible at shutdown, and
1813 restore it as early as possible during reboot.
1815 * A logic to automatically discover and enable home and swap
1816 partitions on GPT disks has been added. With this in place
1817 /etc/fstab becomes optional for many setups as systemd can
1818 discover certain partitions located on the root disk
1819 automatically. Home partitions are recognized under their
1820 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1821 partitions are recognized under their GPT type ID
1822 0657fd6da4ab43c484e50933c84b4f4f.
1824 * systemd will no longer pass any environment from the kernel
1825 or initrd to system services. If you want to set an
1826 environment for all services, do so via the kernel command
1827 line systemd.setenv= assignment.
1829 * The systemd-sysctl tool no longer natively reads the file
1830 /etc/sysctl.conf. If desired, the file should be symlinked
1831 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1832 legacy support by a symlink rather than built-in code, it
1833 also makes the otherwise hidden order of application of the
1834 different files visible. (Note that this partly reverts to a
1835 pre-198 application order of sysctl knobs!)
1837 * The "systemctl set-log-level" and "systemctl dump" commands
1838 have been moved to systemd-analyze.
1840 * systemd-run learned the new --remain-after-exit switch,
1841 which causes the scope unit not to be cleaned up
1842 automatically after the process terminated.
1844 * tmpfiles learned a new --exclude-prefix= switch to exclude
1845 certain paths from operation.
1847 * journald will now automatically flush all messages to disk
1848 as soon as a message of the log priorities CRIT, ALERT or
1851 Contributions from: Andrew Cook, Brandon Philips, Christian
1852 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1853 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1854 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1855 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1856 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1857 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1858 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1859 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1860 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1861 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1862 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1863 William Giokas, Zbigniew Jędrzejewski-Szmek
1865 -- Berlin, 2013-09-13
1869 * The documentation has been updated to cover the various new
1870 concepts introduced with 205.
1872 * Unit files now understand the new %v specifier which
1873 resolves to the kernel version string as returned by "uname
1876 * systemctl now supports filtering the unit list output by
1877 load state, active state and sub state, using the new
1880 * "systemctl status" will now show the results of the
1881 condition checks (like ConditionPathExists= and similar) of
1882 the last start attempts of the unit. They are also logged to
1885 * "journalctl -b" may now be used to look for boot output of a
1886 specific boot. Try "journalctl -b -1" for the previous boot,
1887 but the syntax is substantially more powerful.
1889 * "journalctl --show-cursor" has been added which prints the
1890 cursor string the last shown log line. This may then be used
1891 with the new "journalctl --after-cursor=" switch to continue
1892 browsing logs from that point on.
1894 * "journalctl --force" may now be used to force regeneration
1897 * Creation of "dead" device nodes has been moved from udev
1898 into kmod and tmpfiles. Previously, udev would read the kmod
1899 databases to pre-generate dead device nodes based on meta
1900 information contained in kernel modules, so that these would
1901 be auto-loaded on access rather then at boot. As this
1902 does not really have much to do with the exposing actual
1903 kernel devices to userspace this has always been slightly
1904 alien in the udev codebase. Following the new scheme kmod
1905 will now generate a runtime snippet for tmpfiles from the
1906 module meta information and it now is tmpfiles' job to the
1907 create the nodes. This also allows overriding access and
1908 other parameters for the nodes using the usual tmpfiles
1909 facilities. As side effect this allows us to remove the
1910 CAP_SYS_MKNOD capability bit from udevd entirely.
1912 * logind's device ACLs may now be applied to these "dead"
1913 devices nodes too, thus finally allowing managed access to
1914 devices such as /dev/snd/sequencer whithout loading the
1915 backing module right-away.
1917 * A new RPM macro has been added that may be used to apply
1918 tmpfiles configuration during package installation.
1920 * systemd-detect-virt and ConditionVirtualization= now can
1921 detect User-Mode-Linux machines (UML).
1923 * journald will now implicitly log the effective capabilities
1924 set of processes in the message metadata.
1926 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1928 * The initrd interface has been simplified (more specifically,
1929 support for passing performance data via environment
1930 variables and fsck results via files in /run has been
1931 removed). These features were non-essential, and are
1932 nowadays available in a much nicer way by having systemd in
1933 the initrd serialize its state and have the hosts systemd
1934 deserialize it again.
1936 * The udev "keymap" data files and tools to apply keyboard
1937 specific mappings of scan to key codes, and force-release
1938 scan code lists have been entirely replaced by a udev
1939 "keyboard" builtin and a hwdb data file.
1941 * systemd will now honour the kernel's "quiet" command line
1942 argument also during late shutdown, resulting in a
1943 completely silent shutdown when used.
1945 * There's now an option to control the SO_REUSEPORT socket
1946 option in .socket units.
1948 * Instance units will now automatically get a per-template
1949 subslice of system.slice unless something else is explicitly
1950 configured. For example, instances of sshd@.service will now
1951 implicitly be placed in system-sshd.slice rather than
1952 system.slice as before.
1954 * Test coverage support may now be enabled at build time.
1956 Contributions from: Dave Reisner, Frederic Crozat, Harald
1957 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1958 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1959 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1960 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1961 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1962 Giokas, Zbigniew Jędrzejewski-Szmek
1964 -- Berlin, 2013-07-23
1968 * Two new unit types have been introduced:
1970 Scope units are very similar to service units, however, are
1971 created out of pre-existing processes -- instead of PID 1
1972 forking off the processes. By using scope units it is
1973 possible for system services and applications to group their
1974 own child processes (worker processes) in a powerful way
1975 which then maybe used to organize them, or kill them
1976 together, or apply resource limits on them.
1978 Slice units may be used to partition system resources in an
1979 hierarchial fashion and then assign other units to them. By
1980 default there are now three slices: system.slice (for all
1981 system services), user.slice (for all user sessions),
1982 machine.slice (for VMs and containers).
1984 Slices and scopes have been introduced primarily in
1985 context of the work to move cgroup handling to a
1986 single-writer scheme, where only PID 1
1987 creates/removes/manages cgroups.
1989 * There's a new concept of "transient" units. In contrast to
1990 normal units these units are created via an API at runtime,
1991 not from configuration from disk. More specifically this
1992 means it is now possible to run arbitrary programs as
1993 independent services, with all execution parameters passed
1994 in via bus APIs rather than read from disk. Transient units
1995 make systemd substantially more dynamic then it ever was,
1996 and useful as a general batch manager.
1998 * logind has been updated to make use of scope and slice units
1999 for managing user sessions. As a user logs in he will get
2000 his own private slice unit, to which all sessions are added
2001 as scope units. We also added support for automatically
2002 adding an instance of user@.service for the user into the
2003 slice. Effectively logind will no longer create cgroup
2004 hierarchies on its own now, it will defer entirely to PID 1
2005 for this by means of scope, service and slice units. Since
2006 user sessions this way become entities managed by PID 1
2007 the output of "systemctl" is now a lot more comprehensive.
2009 * A new mini-daemon "systemd-machined" has been added which
2010 may be used by virtualization managers to register local
2011 VMs/containers. nspawn has been updated accordingly, and
2012 libvirt will be updated shortly. machined will collect a bit
2013 of meta information about the VMs/containers, and assign
2014 them their own scope unit (see above). The collected
2015 meta-data is then made available via the "machinectl" tool,
2016 and exposed in "ps" and similar tools. machined/machinectl
2017 is compile-time optional.
2019 * As discussed earlier, the low-level cgroup configuration
2020 options ControlGroup=, ControlGroupModify=,
2021 ControlGroupPersistent=, ControlGroupAttribute= have been
2022 removed. Please use high-level attribute settings instead as
2023 well as slice units.
2025 * A new bus call SetUnitProperties() has been added to alter
2026 various runtime parameters of a unit. This is primarily
2027 useful to alter cgroup parameters dynamically in a nice way,
2028 but will be extended later on to make more properties
2029 modifiable at runtime. systemctl gained a new set-properties
2030 command that wraps this call.
2032 * A new tool "systemd-run" has been added which can be used to
2033 run arbitrary command lines as transient services or scopes,
2034 while configuring a number of settings via the command
2035 line. This tool is currently very basic, however already
2036 very useful. We plan to extend this tool to even allow
2037 queuing of execution jobs with time triggers from the
2038 command line, similar in fashion to "at".
2040 * nspawn will now inform the user explicitly that kernels with
2041 audit enabled break containers, and suggest the user to turn
2044 * Support for detecting the IMA and AppArmor security
2045 frameworks with ConditionSecurity= has been added.
2047 * journalctl gained a new "-k" switch for showing only kernel
2048 messages, mimicking dmesg output; in addition to "--user"
2049 and "--system" switches for showing only user's own logs
2052 * systemd-delta can now show information about drop-in
2053 snippets extending unit files.
2055 * libsystemd-bus has been substantially updated but is still
2056 not available as public API.
2058 * systemd will now look for the "debug" argument on the kernel
2059 command line and enable debug logging, similar to what
2060 "systemd.log_level=debug" already did before.
2062 * "systemctl set-default", "systemctl get-default" has been
2063 added to configure the default.target symlink, which
2064 controls what to boot into by default.
2066 * "systemctl set-log-level" has been added as a convenient
2067 way to raise and lower systemd logging threshold.
2069 * "systemd-analyze plot" will now show the time the various
2070 generators needed for execution, as well as information
2071 about the unit file loading.
2073 * libsystemd-journal gained a new sd_journal_open_files() call
2074 for opening specific journal files. journactl also gained a
2075 new switch to expose this new functionality. Previously we
2076 only supported opening all files from a directory, or all
2077 files from the system, as opening individual files only is
2078 racy due to journal file rotation.
2080 * systemd gained the new DefaultEnvironment= setting in
2081 /etc/systemd/system.conf to set environment variables for
2084 * If a privileged process logs a journal message with the
2085 OBJECT_PID= field set, then journald will automatically
2086 augment this with additional OBJECT_UID=, OBJECT_GID=,
2087 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2088 system services want to log events about specific client
2089 processes. journactl/systemctl has been updated to make use
2090 of this information if all log messages regarding a specific
2093 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2094 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2095 Reisner, David Coppa, David King, David Strauss, Eelco
2096 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2097 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2098 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2099 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2100 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2101 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2102 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2103 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2104 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2105 Łukasz Stelmach, 장동준
2109 * The Python bindings gained some minimal support for the APIs
2110 exposed by libsystemd-logind.
2112 * ConditionSecurity= gained support for detecting SMACK. Since
2113 this condition already supports SELinux and AppArmor we only
2114 miss IMA for this. Patches welcome!
2116 Contributions from: Karol Lewandowski, Lennart Poettering,
2117 Zbigniew Jędrzejewski-Szmek
2121 * systemd-nspawn will now create /etc/resolv.conf if
2122 necessary, before bind-mounting the host's file onto it.
2124 * systemd-nspawn will now store meta information about a
2125 container on the container's cgroup as extended attribute
2126 fields, including the root directory.
2128 * The cgroup hierarchy has been reworked in many ways. All
2129 objects any of the components systemd creates in the cgroup
2130 tree are now suffixed. More specifically, user sessions are
2131 now placed in cgroups suffixed with ".session", users in
2132 cgroups suffixed with ".user", and nspawn containers in
2133 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2134 names are now escaped in a simple scheme to avoid collision
2135 of userspace object names with kernel filenames. This work
2136 is preparation for making these objects relocatable in the
2137 cgroup tree, in order to allow easy resource partitioning of
2138 these objects without causing naming conflicts.
2140 * systemctl list-dependencies gained the new switches
2141 --plain, --reverse, --after and --before.
2143 * systemd-inhibit now shows the process name of processes that
2144 have taken an inhibitor lock.
2146 * nss-myhostname will now also resolve "localhost"
2147 implicitly. This makes /etc/hosts an optional file and
2148 nicely handles that on IPv6 ::1 maps to both "localhost" and
2151 * libsystemd-logind.so gained a new call
2152 sd_get_machine_names() to enumerate running containers and
2153 VMs (currently only supported by very new libvirt and
2154 nspawn). sd_login_monitor can now be used to watch
2155 VMs/containers coming and going.
2157 * .include is not allowed recursively anymore, and only in
2158 unit files. Usually it is better to use drop-in snippets in
2159 .d/*.conf anyway, as introduced with systemd 198.
2161 * systemd-analyze gained a new "critical-chain" command that
2162 determines the slowest chain of units run during system
2163 boot-up. It is very useful for tracking down where
2164 optimizing boot time is the most beneficial.
2166 * systemd will no longer allow manipulating service paths in
2167 the name=systemd:/system cgroup tree using ControlGroup= in
2168 units. (But is still fine with it in all other dirs.)
2170 * There's a new systemd-nspawn@.service service file that may
2171 be used to easily run nspawn containers as system
2172 services. With the container's root directory in
2173 /var/lib/container/foobar it is now sufficient to run
2174 "systemctl start systemd-nspawn@foobar.service" to boot it.
2176 * systemd-cgls gained a new parameter "--machine" to list only
2177 the processes within a certain container.
2179 * ConditionSecurity= now can check for "apparmor". We still
2180 are lacking checks for SMACK and IMA for this condition
2181 check though. Patches welcome!
2183 * A new configuration file /etc/systemd/sleep.conf has been
2184 added that may be used to configure which kernel operation
2185 systemd is supposed to execute when "suspend", "hibernate"
2186 or "hybrid-sleep" is requested. This makes the new kernel
2187 "freeze" state accessible to the user.
2189 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2190 the passed argument if applicable.
2192 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2193 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2194 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2195 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2196 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2197 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2198 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2203 * The output of 'systemctl list-jobs' got some polishing. The
2204 '--type=' argument may now be passed more than once. A new
2205 command 'systemctl list-sockets' has been added which shows
2206 a list of kernel sockets systemd is listening on with the
2207 socket units they belong to, plus the units these socket
2210 * The experimental libsystemd-bus library got substantial
2211 updates to work in conjunction with the (also experimental)
2212 kdbus kernel project. It works well enough to exchange
2213 messages with some sophistication. Note that kdbus is not
2214 ready yet, and the library is mostly an elaborate test case
2215 for now, and not installable.
2217 * systemd gained a new unit 'systemd-static-nodes.service'
2218 that generates static device nodes earlier during boot, and
2219 can run in conjunction with udev.
2221 * libsystemd-login gained a new call sd_pid_get_user_unit()
2222 to retrieve the user systemd unit a process is running
2223 in. This is useful for systems where systemd is used as
2226 * systemd-nspawn now places all containers in the new /machine
2227 top-level cgroup directory in the name=systemd
2228 hierarchy. libvirt will soon do the same, so that we get a
2229 uniform separation of /system, /user and /machine for system
2230 services, user processes and containers/virtual
2231 machines. This new cgroup hierarchy is also useful to stick
2232 stable names to specific container instances, which can be
2233 recognized later this way (this name may be controlled
2234 via systemd-nspawn's new -M switch). libsystemd-login also
2235 gained a new call sd_pid_get_machine_name() to retrieve the
2236 name of the container/VM a specific process belongs to.
2238 * bootchart can now store its data in the journal.
2240 * libsystemd-journal gained a new call
2241 sd_journal_add_conjunction() for AND expressions to the
2242 matching logic. This can be used to express more complex
2243 logical expressions.
2245 * journactl can now take multiple --unit= and --user-unit=
2248 * The cryptsetup logic now understands the "luks.key=" kernel
2249 command line switch for specifying a file to read the
2250 decryption key from. Also, if a configured key file is not
2251 found the tool will now automatically fall back to prompting
2254 * Python systemd.journal module was updated to wrap recently
2255 added functions from libsystemd-journal. The interface was
2256 changed to bring the low level interface in s.j._Reader
2257 closer to the C API, and the high level interface in
2258 s.j.Reader was updated to wrap and convert all data about
2261 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2262 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2263 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2264 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2265 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2266 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2270 * journalctl --update-catalog now understands a new --root=
2271 option to operate on catalogs found in a different root
2274 * During shutdown after systemd has terminated all running
2275 services a final killing loop kills all remaining left-over
2276 processes. We will now print the name of these processes
2277 when we send SIGKILL to them, since this usually indicates a
2280 * If /etc/crypttab refers to password files stored on
2281 configured mount points automatic dependencies will now be
2282 generated to ensure the specific mount is established first
2283 before the key file is attempted to be read.
2285 * 'systemctl status' will now show information about the
2286 network sockets a socket unit is listening on.
2288 * 'systemctl status' will also shown information about any
2289 drop-in configuration file for units. (Drop-In configuration
2290 files in this context are files such as
2291 /etc/systemd/systemd/foobar.service.d/*.conf)
2293 * systemd-cgtop now optionally shows summed up CPU times of
2294 cgroups. Press '%' while running cgtop to switch between
2295 percentage and absolute mode. This is useful to determine
2296 which cgroups use up the most CPU time over the entire
2297 runtime of the system. systemd-cgtop has also been updated
2298 to be 'pipeable' for processing with further shell tools.
2300 * 'hostnamectl set-hostname' will now allow setting of FQDN
2303 * The formatting and parsing of time span values has been
2304 changed. The parser now understands fractional expressions
2305 such as "5.5h". The formatter will now output fractional
2306 expressions for all time spans under 1min, i.e. "5.123456s"
2307 rather than "5s 123ms 456us". For time spans under 1s
2308 millisecond values are shown, for those under 1ms
2309 microsecond values are shown. This should greatly improve
2310 all time-related output of systemd.
2312 * libsystemd-login and libsystemd-journal gained new
2313 functions for querying the poll() events mask and poll()
2314 timeout value for integration into arbitrary event
2317 * localectl gained the ability to list available X11 keymaps
2318 (models, layouts, variants, options).
2320 * 'systemd-analyze dot' gained the ability to filter for
2321 specific units via shell-style globs, to create smaller,
2322 more useful graphs. I.e. it is now possible to create simple
2323 graphs of all the dependencies between only target units, or
2324 of all units that Avahi has dependencies with.
2326 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2327 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2328 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2329 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2330 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2331 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2332 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2336 * The boot-time readahead implementation for rotating media
2337 will now read the read-ahead data in multiple passes which
2338 consist of all read requests made in equidistant time
2339 intervals. This means instead of strictly reading read-ahead
2340 data in its physical order on disk we now try to find a
2341 middle ground between physical and access time order.
2343 * /etc/os-release files gained a new BUILD_ID= field for usage
2344 on operating systems that provide continuous builds of OS
2347 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2348 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2349 William Douglas, Zbigniew Jędrzejewski-Szmek
2353 * systemd-python gained an API exposing libsystemd-daemon.
2355 * The SMACK setup logic gained support for uploading CIPSO
2358 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2359 ReadOnlyDirectories= and InaccessibleDirectories= has
2360 changed. The private /tmp and /var/tmp directories are now
2361 shared by all processes of a service (which means
2362 ExecStartPre= may now leave data in /tmp that ExecStart= of
2363 the same service can still access). When a service is
2364 stopped its temporary directories are immediately deleted
2365 (normal clean-up with tmpfiles is still done in addition to
2368 * By default, systemd will now set a couple of sysctl
2369 variables in the kernel: the safe sysrq options are turned
2370 on, IP route verification is turned on, and source routing
2371 disabled. The recently added hardlink and softlink
2372 protection of the kernel is turned on. These settings should
2373 be reasonably safe, and good defaults for all new systems.
2375 * The predictable network naming logic may now be turned off
2376 with a new kernel command line switch: net.ifnames=0.
2378 * A new libsystemd-bus module has been added that implements a
2379 pretty complete D-Bus client library. For details see:
2381 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2383 * journald will now explicitly flush the journal files to disk
2384 at the latest 5min after each write. The file will then also
2385 be marked offline until the next write. This should increase
2386 reliability in case of a crash. The synchronization delay
2387 can be configured via SyncIntervalSec= in journald.conf.
2389 * There's a new remote-fs-setup.target unit that can be used
2390 to pull in specific services when at least one remote file
2391 system is to be mounted.
2393 * There are new targets timers.target and paths.target as
2394 canonical targets to pull user timer and path units in
2395 from. This complements sockets.target with a similar
2396 purpose for socket units.
2398 * libudev gained a new call udev_device_set_attribute_value()
2399 to set sysfs attributes of a device.
2401 * The udev daemon now sets the default number of worker
2402 processes executed in parallel based on the number of available
2403 CPUs instead of the amount of available RAM. This is supposed
2404 to provide a more reliable default and limit a too aggressive
2405 paralellism for setups with 1000s of devices connected.
2407 Contributions from: Auke Kok, Colin Walters, Cristian
2408 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2409 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2410 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2411 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2412 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2413 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2414 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2415 Zbigniew Jędrzejewski-Szmek
2419 * Configuration of unit files may now be extended via drop-in
2420 files without having to edit/override the unit files
2421 themselves. More specifically, if the administrator wants to
2422 change one value for a service file foobar.service he can
2423 now do so by dropping in a configuration snippet into
2424 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2425 will load all these snippets and apply them on top of the
2426 main unit configuration file, possibly extending or
2427 overriding its settings. Using these drop-in snippets is
2428 generally nicer than the two earlier options for changing
2429 unit files locally: copying the files from
2430 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2431 them there; or creating a new file in /etc/systemd/system/
2432 that incorporates the original one via ".include". Drop-in
2433 snippets into these .d/ directories can be placed in any
2434 directory systemd looks for units in, and the usual
2435 overriding semantics between /usr/lib, /etc and /run apply
2438 * Most unit file settings which take lists of items can now be
2439 reset by assigning the empty string to them. For example,
2440 normally, settings such as Environment=FOO=BAR append a new
2441 environment variable assignment to the environment block,
2442 each time they are used. By assigning Environment= the empty
2443 string the environment block can be reset to empty. This is
2444 particularly useful with the .d/*.conf drop-in snippets
2445 mentioned above, since this adds the ability to reset list
2446 settings from vendor unit files via these drop-ins.
2448 * systemctl gained a new "list-dependencies" command for
2449 listing the dependencies of a unit recursively.
2451 * Inhibitors are now honored and listed by "systemctl
2452 suspend", "systemctl poweroff" (and similar) too, not only
2453 GNOME. These commands will also list active sessions by
2456 * Resource limits (as exposed by the various control group
2457 controllers) can now be controlled dynamically at runtime
2458 for all units. More specifically, you can now use a command
2459 like "systemctl set-cgroup-attr foobar.service cpu.shares
2460 2000" to alter the CPU shares a specific service gets. These
2461 settings are stored persistently on disk, and thus allow the
2462 administrator to easily adjust the resource usage of
2463 services with a few simple commands. This dynamic resource
2464 management logic is also available to other programs via the
2465 bus. Almost any kernel cgroup attribute and controller is
2468 * systemd-vconsole-setup will now copy all font settings to
2469 all allocated VTs, where it previously applied them only to
2472 * libsystemd-login gained the new sd_session_get_tty() API
2475 * This release drops support for a few legacy or
2476 distribution-specific LSB facility names when parsing init
2477 scripts: $x-display-manager, $mail-transfer-agent,
2478 $mail-transport-agent, $mail-transfer-agent, $smtp,
2479 $null. Also, the mail-transfer-agent.target unit backing
2480 this has been removed. Distributions which want to retain
2481 compatibility with this should carry the burden for
2482 supporting this themselves and patch support for these back
2483 in, if they really need to. Also, the facilities $syslog and
2484 $local_fs are now ignored, since systemd does not support
2485 early-boot LSB init scripts anymore, and these facilities
2486 are implied anyway for normal services. syslog.target has
2489 * There are new bus calls on PID1's Manager object for
2490 cancelling jobs, and removing snapshot units. Previously,
2491 both calls were only available on the Job and Snapshot
2494 * systemd-journal-gatewayd gained SSL support.
2496 * The various "environment" files, such as /etc/locale.conf
2497 now support continuation lines with a backslash ("\") as
2498 last character in the line, similarly in style (but different)
2499 to how this is supported in shells.
2501 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2502 now implicitly appended to every log entry logged. systemctl
2503 has been updated to filter by this field when operating on a
2504 user systemd instance.
2506 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2507 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2508 the container. This makes it easier to boot unmodified
2509 Fedora systems in a container, which however still requires
2510 audit=0 to be passed on the kernel command line. Auditing in
2511 kernel and userspace is unfortunately still too broken in
2512 context of containers, hence we recommend compiling it out
2513 of the kernel or using audit=0. Hopefully this will be fixed
2514 one day for good in the kernel.
2516 * nspawn gained the new --bind= and --bind-ro= parameters to
2517 bind mount specific directories from the host into the
2520 * nspawn will now mount its own devpts file system instance
2521 into the container, in order not to leak pty devices from
2522 the host into the container.
2524 * systemd will now read the firmware boot time performance
2525 information from the EFI variables, if the used boot loader
2526 supports this, and takes it into account for boot performance
2527 analysis via "systemd-analyze". This is currently supported
2528 only in conjunction with Gummiboot, but could be supported
2529 by other boot loaders too. For details see:
2531 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2533 * A new generator has been added that automatically mounts the
2534 EFI System Partition (ESP) to /boot, if that directory
2535 exists, is empty, and no other file system has been
2536 configured to be mounted there.
2538 * logind will now send out PrepareForSleep(false) out
2539 unconditionally, after coming back from suspend. This may be
2540 used by applications as asynchronous notification for
2541 system resume events.
2543 * "systemctl unlock-sessions" has been added, that allows
2544 unlocking the screens of all user sessions at once, similar
2545 to how "systemctl lock-sessions" already locked all users
2546 sessions. This is backed by a new D-Bus call UnlockSessions().
2548 * "loginctl seat-status" will now show the master device of a
2549 seat. (i.e. the device of a seat that needs to be around for
2550 the seat to be considered available, usually the graphics
2553 * tmpfiles gained a new "X" line type, that allows
2554 configuration of files and directories (with wildcards) that
2555 shall be excluded from automatic cleanup ("aging").
2557 * udev default rules set the device node permissions now only
2558 at "add" events, and do not change them any longer with a
2559 later "change" event.
2561 * The log messages for lid events and power/sleep keypresses
2562 now carry a message ID.
2564 * We now have a substantially larger unit test suite, but this
2565 continues to be work in progress.
2567 * udevadm hwdb gained a new --root= parameter to change the
2568 root directory to operate relative to.
2570 * logind will now issue a background sync() request to the kernel
2571 early at shutdown, so that dirty buffers are flushed to disk early
2572 instead of at the last moment, in order to optimize shutdown
2575 * A new bootctl tool has been added that is an interface for
2576 certain boot loader operations. This is currently a preview
2577 and is likely to be extended into a small mechanism daemon
2578 like timedated, localed, hostnamed, and can be used by
2579 graphical UIs to enumerate available boot options, and
2580 request boot into firmware operations.
2582 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2583 the rest of the package. It also has been updated to work
2584 correctly in initrds.
2586 * Policykit previously has been runtime optional, and is now
2587 also compile time optional via a configure switch.
2589 * systemd-analyze has been reimplemented in C. Also "systemctl
2590 dot" has moved into systemd-analyze.
2592 * "systemctl status" with no further parameters will now print
2593 the status of all active or failed units.
2595 * Operations such as "systemctl start" can now be executed
2596 with a new mode "--irreversible" which may be used to queue
2597 operations that cannot accidentally be reversed by a later
2598 job queuing. This is by default used to make shutdown
2599 requests more robust.
2601 * The Python API of systemd now gained a new module for
2602 reading journal files.
2604 * A new tool kernel-install has been added that can install
2605 kernel images according to the Boot Loader Specification:
2607 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2609 * Boot time console output has been improved to provide
2610 animated boot time output for hanging jobs.
2612 * A new tool systemd-activate has been added which can be used
2613 to test socket activation with, directly from the command
2614 line. This should make it much easier to test and debug
2615 socket activation in daemons.
2617 * journalctl gained a new "--reverse" (or -r) option to show
2618 journal output in reverse order (i.e. newest line first).
2620 * journalctl gained a new "--pager-end" (or -e) option to jump
2621 to immediately jump to the end of the journal in the
2622 pager. This is only supported in conjunction with "less".
2624 * journalctl gained a new "--user-unit=" option, that works
2625 similarly to "--unit=" but filters for user units rather than
2628 * A number of unit files to ease adoption of systemd in
2629 initrds has been added. This moves some minimal logic from
2630 the various initrd implementations into systemd proper.
2632 * The journal files are now owned by a new group
2633 "systemd-journal", which exists specifically to allow access
2634 to the journal, and nothing else. Previously, we used the
2635 "adm" group for that, which however possibly covers more
2636 than just journal/log file access. This new group is now
2637 already used by systemd-journal-gatewayd to ensure this
2638 daemon gets access to the journal files and as little else
2639 as possible. Note that "make install" will also set FS ACLs
2640 up for /var/log/journal to give "adm" and "wheel" read
2641 access to it, in addition to "systemd-journal" which owns
2642 the journal files. We recommend that packaging scripts also
2643 add read access to "adm" + "wheel" to /var/log/journal, and
2644 all existing/future journal files. To normal users and
2645 administrators little changes, however packagers need to
2646 ensure to create the "systemd-journal" system group at
2647 package installation time.
2649 * The systemd-journal-gatewayd now runs as unprivileged user
2650 systemd-journal-gateway:systemd-journal-gateway. Packaging
2651 scripts need to create these system user/group at
2654 * timedated now exposes a new boolean property CanNTP that
2655 indicates whether a local NTP service is available or not.
2657 * systemd-detect-virt will now also detect xen PVs
2659 * The pstore file system is now mounted by default, if it is
2662 * In addition to the SELinux and IMA policies we will now also
2663 load SMACK policies at early boot.
2665 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2666 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2667 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2668 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2669 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2670 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2671 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2672 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2673 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2674 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2675 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2676 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2677 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2678 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2682 * Timer units now support calendar time events in addition to
2683 monotonic time events. That means you can now trigger a unit
2684 based on a calendar time specification such as "Thu,Fri
2685 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2686 or fifth day of any month of the year 2013, given that it is
2687 a thursday or friday. This brings timer event support
2688 considerably closer to cron's capabilities. For details on
2689 the supported calendar time specification language see
2692 * udev now supports a number of different naming policies for
2693 network interfaces for predictable names, and a combination
2694 of these policies is now the default. Please see this wiki
2695 document for details:
2697 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2699 * Auke Kok's bootchart implementation has been added to the
2700 systemd tree. It is an optional component that can graph the
2701 boot in quite some detail. It is one of the best bootchart
2702 implementations around and minimal in its code and
2705 * nss-myhostname has been integrated into the systemd source
2706 tree. nss-myhostname guarantees that the local hostname
2707 always stays resolvable via NSS. It has been a weak
2708 requirement of systemd-hostnamed since a long time, and
2709 since its code is actually trivial we decided to just
2710 include it in systemd's source tree. It can be turned off
2711 with a configure switch.
2713 * The read-ahead logic is now capable of properly detecting
2714 whether a btrfs file system is on SSD or rotating media, in
2715 order to optimize the read-ahead scheme. Previously, it was
2716 only capable of detecting this on traditional file systems
2719 * In udev, additional device properties are now read from the
2720 IAB in addition to the OUI database. Also, Bluetooth company
2721 identities are attached to the devices as well.
2723 * In service files %U may be used as specifier that is
2724 replaced by the configured user name of the service.
2726 * nspawn may now be invoked without a controlling TTY. This
2727 makes it suitable for invocation as its own service. This
2728 may be used to set up a simple containerized server system
2729 using only core OS tools.
2731 * systemd and nspawn can now accept socket file descriptors
2732 when they are started for socket activation. This enables
2733 implementation of socket activated nspawn
2734 containers. i.e. think about autospawning an entire OS image
2735 when the first SSH or HTTP connection is received. We expect
2736 that similar functionality will also be added to libvirt-lxc
2739 * journalctl will now suppress ANSI color codes when
2740 presenting log data.
2742 * systemctl will no longer show control group information for
2743 a unit if a the control group is empty anyway.
2745 * logind can now automatically suspend/hibernate/shutdown the
2748 * /etc/machine-info and hostnamed now also expose the chassis
2749 type of the system. This can be used to determine whether
2750 the local system is a laptop, desktop, handset or
2751 tablet. This information may either be configured by the
2752 user/vendor or is automatically determined from ACPI and DMI
2753 information if possible.
2755 * A number of PolicyKit actions are now bound together with
2756 "imply" rules. This should simplify creating UIs because
2757 many actions will now authenticate similar ones as well.
2759 * Unit files learnt a new condition ConditionACPower= which
2760 may be used to conditionalize a unit depending on whether an
2761 AC power source is connected or not, of whether the system
2762 is running on battery power.
2764 * systemctl gained a new "is-failed" verb that may be used in
2765 shell scripts and suchlike to check whether a specific unit
2766 is in the "failed" state.
2768 * The EnvironmentFile= setting in unit files now supports file
2769 globbing, and can hence be used to easily read a number of
2770 environment files at once.
2772 * systemd will no longer detect and recognize specific
2773 distributions. All distribution-specific #ifdeffery has been
2774 removed, systemd is now fully generic and
2775 distribution-agnostic. Effectively, not too much is lost as
2776 a lot of the code is still accessible via explicit configure
2777 switches. However, support for some distribution specific
2778 legacy configuration file formats has been dropped. We
2779 recommend distributions to simply adopt the configuration
2780 files everybody else uses now and convert the old
2781 configuration from packaging scripts. Most distributions
2782 already did that. If that's not possible or desirable,
2783 distributions are welcome to forward port the specific
2784 pieces of code locally from the git history.
2786 * When logging a message about a unit systemd will now always
2787 log the unit name in the message meta data.
2789 * localectl will now also discover system locale data that is
2790 not stored in locale archives, but directly unpacked.
2792 * logind will no longer unconditionally use framebuffer
2793 devices as seat masters, i.e. as devices that are required
2794 to be existing before a seat is considered preset. Instead,
2795 it will now look for all devices that are tagged as
2796 "seat-master" in udev. By default framebuffer devices will
2797 be marked as such, but depending on local systems other
2798 devices might be marked as well. This may be used to
2799 integrate graphics cards using closed source drivers (such
2800 as NVidia ones) more nicely into logind. Note however, that
2801 we recommend using the open source NVidia drivers instead,
2802 and no udev rules for the closed-source drivers will be
2803 shipped from us upstream.
2805 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2806 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2807 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2808 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2809 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2810 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2811 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2812 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2813 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2814 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2815 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2816 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2821 * udev gained support for loading additional device properties
2822 from an indexed database that is keyed by vendor/product IDs
2823 and similar device identifiers. For the beginning this
2824 "hwdb" is populated with data from the well-known PCI and
2825 USB database, but also includes PNP, ACPI and OID data. In
2826 the longer run this indexed database shall grow into
2827 becoming the one central database for non-essential
2828 userspace device metadata. Previously, data from the PCI/USB
2829 database was only attached to select devices, since the
2830 lookup was a relatively expensive operation due to O(n) time
2831 complexity (with n being the number of entries in the
2832 database). Since this is now O(1), we decided to add in this
2833 data for all devices where this is available, by
2834 default. Note that the indexed database needs to be rebuilt
2835 when new data files are installed. To achieve this you need
2836 to update your packaging scripts to invoke "udevadm hwdb
2837 --update" after installation of hwdb data files. For
2838 RPM-based distributions we introduced the new
2839 %udev_hwdb_update macro for this purpose.
2841 * The Journal gained support for the "Message Catalog", an
2842 indexed database to link up additional information with
2843 journal entries. For further details please check:
2845 http://www.freedesktop.org/wiki/Software/systemd/catalog
2847 The indexed message catalog database also needs to be
2848 rebuilt after installation of message catalog files. Use
2849 "journalctl --update-catalog" for this. For RPM-based
2850 distributions we introduced the %journal_catalog_update
2851 macro for this purpose.
2853 * The Python Journal bindings gained support for the standard
2854 Python logging framework.
2856 * The Journal API gained new functions for checking whether
2857 the underlying file system of a journal file is capable of
2858 properly reporting file change notifications, or whether
2859 applications that want to reflect journal changes "live"
2860 need to recheck journal files continuously in appropriate
2863 * It is now possible to set the "age" field for tmpfiles
2864 entries to 0, indicating that files matching this entry
2865 shall always be removed when the directories are cleaned up.
2867 * coredumpctl gained a new "gdb" verb which invokes gdb
2868 right-away on the selected coredump.
2870 * There's now support for "hybrid sleep" on kernels that
2871 support this, in addition to "suspend" and "hibernate". Use
2872 "systemctl hybrid-sleep" to make use of this.
2874 * logind's HandleSuspendKey= setting (and related settings)
2875 now gained support for a new "lock" setting to simply
2876 request the screen lock on all local sessions, instead of
2877 actually executing a suspend or hibernation.
2879 * systemd will now mount the EFI variables file system by
2882 * Socket units now gained support for configuration of the
2883 SMACK security label.
2885 * timedatectl will now output the time of the last and next
2886 daylight saving change.
2888 * We dropped support for various legacy and distro-specific
2889 concepts, such as insserv, early-boot SysV services
2890 (i.e. those for non-standard runlevels such as 'b' or 'S')
2891 or ArchLinux /etc/rc.conf support. We recommend the
2892 distributions who still need support this to either continue
2893 to maintain the necessary patches downstream, or find a
2894 different solution. (Talk to us if you have questions!)
2896 * Various systemd components will now bypass PolicyKit checks
2897 for root and otherwise handle properly if PolicyKit is not
2898 found to be around. This should fix most issues for
2899 PolicyKit-less systems. Quite frankly this should have been
2900 this way since day one. It is absolutely our intention to
2901 make systemd work fine on PolicyKit-less systems, and we
2902 consider it a bug if something does not work as it should if
2903 PolicyKit is not around.
2905 * For embedded systems it is now possible to build udev and
2906 systemd without blkid and/or kmod support.
2908 * "systemctl switch-root" is now capable of switching root
2909 more than once. I.e. in addition to transitions from the
2910 initrd to the host OS it is now possible to transition to
2911 further OS images from the host. This is useful to implement
2912 offline updating tools.
2914 * Various other additions have been made to the RPM macros
2915 shipped with systemd. Use %udev_rules_update() after
2916 installing new udev rules files. %_udevhwdbdir,
2917 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2918 %_sysctldir are now available which resolve to the right
2919 directories for packages to place various data files in.
2921 * journalctl gained the new --full switch (in addition to
2922 --all, to disable ellipsation for long messages.
2924 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2925 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2926 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2927 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2928 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2929 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2930 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2931 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2932 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2936 * journalctl gained new --since= and --until= switches to
2937 filter by time. It also now supports nice filtering for
2938 units via --unit=/-u.
2940 * Type=oneshot services may use ExecReload= and do the
2943 * The journal daemon now supports time-based rotation and
2944 vacuuming, in addition to the usual disk-space based
2947 * The journal will now index the available field values for
2948 each field name. This enables clients to show pretty drop
2949 downs of available match values when filtering. The bash
2950 completion of journalctl has been updated
2951 accordingly. journalctl gained a new switch -F to list all
2952 values a certain field takes in the journal database.
2954 * More service events are now written as structured messages
2955 to the journal, and made recognizable via message IDs.
2957 * The timedated, localed and hostnamed mini-services which
2958 previously only provided support for changing time, locale
2959 and hostname settings from graphical DEs such as GNOME now
2960 also have a minimal (but very useful) text-based client
2961 utility each. This is probably the nicest way to changing
2962 these settings from the command line now, especially since
2963 it lists available options and is fully integrated with bash
2966 * There's now a new tool "systemd-coredumpctl" to list and
2967 extract coredumps from the journal.
2969 * We now install a README each in /var/log/ and
2970 /etc/rc.d/init.d explaining where the system logs and init
2971 scripts went. This hopefully should help folks who go to
2972 that dirs and look into the otherwise now empty void and
2973 scratch their heads.
2975 * When user-services are invoked (by systemd --user) the
2976 $MANAGERPID env var is set to the PID of systemd.
2978 * SIGRTMIN+24 when sent to a --user instance will now result
2979 in immediate termination of systemd.
2981 * gatewayd received numerous feature additions such as a
2982 "follow" mode, for live syncing and filtering.
2984 * browse.html now allows filtering and showing detailed
2985 information on specific entries. Keyboard navigation and
2986 mouse screen support has been added.
2988 * gatewayd/journalctl now supports HTML5/JSON
2989 Server-Sent-Events as output.
2991 * The SysV init script compatibility logic will now
2992 heuristically determine whether a script supports the
2993 "reload" verb, and only then make this available as
2996 * "systemctl status --follow" has been removed, use "journalctl
2999 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3000 have been removed since they are hardly useful to be
3003 * And I'd like to take the opportunity to specifically mention
3004 Zbigniew for his great contributions. Zbigniew, you rock!
3006 Contributions from: Andrew Eikum, Christian Hesse, Colin
3007 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3008 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3009 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3010 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3011 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3012 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3016 * If /etc/vconsole.conf is non-existent or empty we will no
3017 longer load any console font or key map at boot by
3018 default. Instead the kernel defaults will be left
3019 intact. This is definitely the right thing to do, as no
3020 configuration should mean no configuration, and hard-coding
3021 font names that are different on all archs is probably a bad
3022 idea. Also, the kernel default key map and font should be
3023 good enough for most cases anyway, and mostly identical to
3024 the userspace fonts/key maps we previously overloaded them
3025 with. If distributions want to continue to default to a
3026 non-kernel font or key map they should ship a default
3027 /etc/vconsole.conf with the appropriate contents.
3029 Contributions from: Colin Walters, Daniel J Walsh, Dave
3030 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3031 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3035 * journalctl gained a new --cursor= switch to show entries
3036 starting from the specified location in the journal.
3038 * We now enforce a size limit on journal entry fields exported
3039 with "-o json" in journalctl. Fields larger than 4K will be
3040 assigned null. This can be turned off with --all.
3042 * An (optional) journal gateway daemon is now available as
3043 "systemd-journal-gatewayd.service". This service provides
3044 access to the journal via HTTP and JSON. This functionality
3045 will be used to implement live log synchronization in both
3046 pull and push modes, but has various other users too, such
3047 as easy log access for debugging of embedded devices. Right
3048 now it is already useful to retrieve the journal via HTTP:
3050 # systemctl start systemd-journal-gatewayd.service
3051 # wget http://localhost:19531/entries
3053 This will download the journal contents in a
3054 /var/log/messages compatible format. The same as JSON:
3056 # curl -H"Accept: application/json" http://localhost:19531/entries
3058 This service is also accessible via a web browser where a
3059 single static HTML5 app is served that uses the JSON logic
3060 to enable the user to do some basic browsing of the
3061 journal. This will be extended later on. Here's an example
3062 screenshot of this app in its current state:
3064 http://0pointer.de/public/journal-gatewayd
3066 Contributions from: Kay Sievers, Lennart Poettering, Robert
3067 Milasan, Tom Gundersen
3071 * The bash completion logic is now available for journalctl
3074 * We do not mount the "cpuset" controller anymore together with
3075 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3076 started if no parameters are assigned to it. "cpuset" hence
3077 broke code that assumed it it could create "cpu" groups and
3080 * journalctl -f will now subscribe to terminal size changes,
3081 and line break accordingly.
3083 Contributions from: Dave Reisner, Kay Sievers, Lennart
3084 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3088 * nspawn will now create a symlink /etc/localtime in the
3089 container environment, copying the host's timezone
3090 setting. Previously this has been done via a bind mount, but
3091 since symlinks cannot be bind mounted this has now been
3092 changed to create/update the appropriate symlink.
3094 * journalctl -n's line number argument is now optional, and
3095 will default to 10 if omitted.
3097 * journald will now log the maximum size the journal files may
3098 take up on disk. This is particularly useful if the default
3099 built-in logic of determining this parameter from the file
3100 system size is used. Use "systemctl status
3101 systemd-journald.service" to see this information.
3103 * The multi-seat X wrapper tool has been stripped down. As X
3104 is now capable of enumerating graphics devices via udev in a
3105 seat-aware way the wrapper is not strictly necessary
3106 anymore. A stripped down temporary stop-gap is still shipped
3107 until the upstream display managers have been updated to
3108 fully support the new X logic. Expect this wrapper to be
3109 removed entirely in one of the next releases.
3111 * HandleSleepKey= in logind.conf has been split up into
3112 HandleSuspendKey= and HandleHibernateKey=. The old setting
3113 is not available anymore. X11 and the kernel are
3114 distuingishing between these keys and we should too. This
3115 also means the inhibition lock for these keys has been split
3118 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3119 Poettering, Lukas Nykryn, Václav Pavlín
3123 * Whenever a unit changes state we will now log this to the
3124 journal and show along the unit's own log output in
3127 * ConditionPathIsMountPoint= can now properly detect bind
3128 mount points too. (Previously, a bind mount of one file
3129 system to another place in the same file system could not be
3130 detected as mount, since they shared struct stat's st_dev
3133 * We will now mount the cgroup controllers cpu, cpuacct,
3134 cpuset and the controllers net_cls, net_prio together by
3137 * nspawn containers will now have a virtualized boot
3138 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3139 over with a randomized ID at container initialization). This
3140 has the effect of making "journalctl -b" do the right thing
3143 * The JSON output journal serialization has been updated not
3144 to generate "endless" list objects anymore, but rather one
3145 JSON object per line. This is more in line how most JSON
3146 parsers expect JSON objects. The new output mode
3147 "json-pretty" has been added to provide similar output, but
3148 neatly aligned for readability by humans.
3150 * We dropped all explicit sync() invocations in the shutdown
3151 code. The kernel does this implicitly anyway in the kernel
3152 reboot() syscall. halt(8)'s -n option is now a compatibility
3155 * We now support virtualized reboot() in containers, as
3156 supported by newer kernels. We will fall back to exit() if
3157 CAP_SYS_REBOOT is not available to the container. Also,
3158 nspawn makes use of this now and will actually reboot the
3159 container if the containerized OS asks for that.
3161 * journalctl will only show local log output by default
3162 now. Use --merge (-m) to show remote log output, too.
3164 * libsystemd-journal gained the new sd_journal_get_usage()
3165 call to determine the current disk usage of all journal
3166 files. This is exposed in the new "journalctl --disk-usage"
3169 * journald gained a new configuration setting SplitMode= in
3170 journald.conf which may be used to control how user journals
3171 are split off. See journald.conf(5) for details.
3173 * A new condition type ConditionFileNotEmpty= has been added.
3175 * tmpfiles' "w" lines now support file globbing, to write
3176 multiple files at once.
3178 * We added Python bindings for the journal submission
3179 APIs. More Python APIs for a number of selected APIs will
3180 likely follow. Note that we intend to add native bindings
3181 only for the Python language, as we consider it common
3182 enough to deserve bindings shipped within systemd. There are
3183 various projects outside of systemd that provide bindings
3184 for languages such as PHP or Lua.
3186 * Many conditions will now resolve specifiers such as %i. In
3187 addition, PathChanged= and related directives of .path units
3188 now support specifiers as well.
3190 * There's now a new RPM macro definition for the system preset
3193 * journald will now warn if it ca not forward a message to the
3194 syslog daemon because its socket is full.
3196 * timedated will no longer write or process /etc/timezone,
3197 except on Debian. As we do not support late mounted /usr
3198 anymore /etc/localtime always being a symlink is now safe,
3199 and hence the information in /etc/timezone is not necessary
3202 * logind will now always reserve one VT for a text getty (VT6
3203 by default). Previously if more than 6 X sessions where
3204 started they took up all the VTs with auto-spawned gettys,
3205 so that no text gettys were available anymore.
3207 * udev will now automatically inform the btrfs kernel logic
3208 about btrfs RAID components showing up. This should make
3209 simple hotplug based btrfs RAID assembly work.
3211 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3212 (but not for its children which will stay at the kernel
3213 default). This should allow setups with a lot more listening
3216 * systemd will now always pass the configured timezone to the
3217 kernel at boot. timedated will do the same when the timezone
3220 * logind's inhibition logic has been updated. By default,
3221 logind will now handle the lid switch, the power and sleep
3222 keys all the time, even in graphical sessions. If DEs want
3223 to handle these events on their own they should take the new
3224 handle-power-key, handle-sleep-key and handle-lid-switch
3225 inhibitors during their runtime. A simple way to achiveve
3226 that is to invoke the DE wrapped in an invocation of:
3228 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3230 * Access to unit operations is now checked via SELinux taking
3231 the unit file label and client process label into account.
3233 * systemd will now notify the administrator in the journal
3234 when he over-mounts a non-empty directory.
3236 * There are new specifiers that are resolved in unit files,
3237 for the host name (%H), the machine ID (%m) and the boot ID
3240 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3241 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3242 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3243 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3244 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3245 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3246 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3250 * Support for reading structured kernel messages from
3251 /dev/kmsg has now been added and is enabled by default.
3253 * Support for reading kernel messages from /proc/kmsg has now
3254 been removed. If you want kernel messages in the journal
3255 make sure to run a recent kernel (>= 3.5) that supports
3256 reading structured messages from /dev/kmsg (see
3257 above). /proc/kmsg is now exclusive property of classic
3258 syslog daemons again.
3260 * The libudev API gained the new
3261 udev_device_new_from_device_id() call.
3263 * The logic for file system namespace (ReadOnlyDirectory=,
3264 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3265 require pivot_root() anymore. This means fewer temporary
3266 directories are created below /tmp for this feature.
3268 * nspawn containers will now see and receive all submounts
3269 made on the host OS below the root file system of the
3272 * Forward Secure Sealing is now supported for Journal files,
3273 which provide cryptographical sealing of journal files so
3274 that attackers cannot alter log history anymore without this
3275 being detectable. Lennart will soon post a blog story about
3276 this explaining it in more detail.
3278 * There are two new service settings RestartPreventExitStatus=
3279 and SuccessExitStatus= which allow configuration of exit
3280 status (exit code or signal) which will be excepted from the
3281 restart logic, resp. consider successful.
3283 * journalctl gained the new --verify switch that can be used
3284 to check the integrity of the structure of journal files and
3285 (if Forward Secure Sealing is enabled) the contents of
3288 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3289 and similar symlinks pre-created. This makes running shells
3290 as container init process a lot more fun.
3292 * The fstab support can now handle PARTUUID= and PARTLABEL=
3295 * A new ConditionHost= condition has been added to match
3296 against the hostname (with globs) and machine ID. This is
3297 useful for clusters where a single OS image is used to
3298 provision a large number of hosts which shall run slightly
3299 different sets of services.
3301 * Services which hit the restart limit will now be placed in a
3304 Contributions from: Bertram Poettering, Dave Reisner, Huang
3305 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3306 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3310 * When running in --user mode systemd will now become a
3311 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3312 tree a lot more organized.
3314 * A new PartOf= unit dependency type has been introduced that
3315 may be used to group services in a natural way.
3317 * "systemctl enable" may now be used to enable instances of
3320 * journalctl now prints error log levels in red, and
3321 warning/notice log levels in bright white. It also supports
3322 filtering by log level now.
3324 * cgtop gained a new -n switch (similar to top), to configure
3325 the maximum number of iterations to run for. It also gained
3326 -b, to run in batch mode (accepting no input).
3328 * The suffix ".service" may now be omitted on most systemctl
3329 command lines involving service unit names.
3331 * There's a new bus call in logind to lock all sessions, as
3332 well as a loginctl verb for it "lock-sessions".
3334 * libsystemd-logind.so gained a new call sd_journal_perror()
3335 that works similar to libc perror() but logs to the journal
3336 and encodes structured information about the error number.
3338 * /etc/crypttab entries now understand the new keyfile-size=
3341 * shutdown(8) now can send a (configurable) wall message when
3342 a shutdown is cancelled.
3344 * The mount propagation mode for the root file system will now
3345 default to "shared", which is useful to make containers work
3346 nicely out-of-the-box so that they receive new mounts from
3347 the host. This can be undone locally by running "mount
3348 --make-rprivate /" if needed.
3350 * The prefdm.service file has been removed. Distributions
3351 should maintain this unit downstream if they intend to keep
3352 it around. However, we recommend writing normal unit files
3353 for display managers instead.
3355 * Since systemd is a crucial part of the OS we will now
3356 default to a number of compiler switches that improve
3357 security (hardening) such as read-only relocations, stack
3358 protection, and suchlike.
3360 * The TimeoutSec= setting for services is now split into
3361 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3362 of individual time outs for the start and the stop phase of
3365 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3366 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3367 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3368 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3369 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3370 Gundersen, Zbigniew Jędrzejewski-Szmek
3374 * The journal and id128 C APIs are now fully documented as man
3377 * Extra safety checks have been added when transitioning from
3378 the initial RAM disk to the main system to avoid accidental
3381 * /etc/crypttab entries now understand the new keyfile-offset=
3384 * systemctl -t can now be used to filter by unit load state.
3386 * The journal C API gained the new sd_journal_wait() call to
3387 make writing synchronous journal clients easier.
3389 * journalctl gained the new -D switch to show journals from a
3392 * journalctl now displays a special marker between log
3393 messages of two different boots.
3395 * The journal is now explicitly flushed to /var via a service
3396 systemd-journal-flush.service, rather than implicitly simply
3397 by seeing /var/log/journal to be writable.
3399 * journalctl (and the journal C APIs) can now match for much
3400 more complex expressions, with alternatives and
3403 * When transitioning from the initial RAM disk to the main
3404 system we will now kill all processes in a killing spree to
3405 ensure no processes stay around by accident.
3407 * Three new specifiers may be used in unit files: %u, %h, %s
3408 resolve to the user name, user home directory resp. user
3409 shell. This is useful for running systemd user instances.
3411 * We now automatically rotate journal files if their data
3412 object hash table gets a fill level > 75%. We also size the
3413 hash table based on the configured maximum file size. This
3414 together should lower hash collisions drastically and thus
3415 speed things up a bit.
3417 * journalctl gained the new "--header" switch to introspect
3418 header data of journal files.
3420 * A new setting SystemCallFilters= has been added to services
3421 which may be used to apply blacklists or whitelists to
3422 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3424 * nspawn gained a new --link-journal= switch (and quicker: -j)
3425 to link the container journal with the host. This makes it
3426 very easy to centralize log viewing on the host for all
3427 guests while still keeping the journal files separated.
3429 * Many bugfixes and optimizations
3431 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3432 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3433 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3438 * Several tools now understand kernel command line arguments,
3439 which are only read when run in an initial RAM disk. They
3440 usually follow closely their normal counterparts, but are
3443 * There's a new tool to analyze the readahead files that are
3444 automatically generated at boot. Use:
3446 /usr/lib/systemd/systemd-readahead analyze /.readahead
3448 * We now provide an early debug shell on tty9 if this enabled. Use:
3450 systemctl enable debug-shell.service
3452 * All plymouth related units have been moved into the Plymouth
3453 package. Please make sure to upgrade your Plymouth version
3456 * systemd-tmpfiles now supports getting passed the basename of
3457 a configuration file only, in which case it will look for it
3458 in all appropriate directories automatically.
3460 * udevadm info now takes a /dev or /sys path as argument, and
3461 does the right thing. Example:
3463 udevadm info /dev/sda
3464 udevadm info /sys/class/block/sda
3466 * systemctl now prints a warning if a unit is stopped but a
3467 unit that might trigger it continues to run. Example: a
3468 service is stopped but the socket that activates it is left
3471 * "systemctl status" will now mention if the log output was
3472 shortened due to rotation since a service has been started.
3474 * The journal API now exposes functions to determine the
3475 "cutoff" times due to rotation.
3477 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3478 immediately flushing of runtime logs to /var if possible,
3479 resp. for triggering immediate rotation of the journal
3482 * It is now considered an error if a service is attempted to
3483 be stopped that is not loaded.
3485 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3487 * systemd-analyze now supports Python 3
3489 * tmpfiles now supports cleaning up directories via aging
3490 where the first level dirs are always kept around but
3491 directories beneath it automatically aged. This is enabled
3492 by prefixing the age field with '~'.
3494 * Seat objects now expose CanGraphical, CanTTY properties
3495 which is required to deal with very fast bootups where the
3496 display manager might be running before the graphics drivers
3497 completed initialization.
3499 * Seat objects now expose a State property.
3501 * We now include RPM macros for service enabling/disabling
3502 based on the preset logic. We recommend RPM based
3503 distributions to make use of these macros if possible. This
3504 makes it simpler to reuse RPM spec files across
3507 * We now make sure that the collected systemd unit name is
3508 always valid when services log to the journal via
3511 * There's a new man page kernel-command-line(7) detailing all
3512 command line options we understand.
3514 * The fstab generator may now be disabled at boot by passing
3515 fstab=0 on the kernel command line.
3517 * A new kernel command line option modules-load= is now understood
3518 to load a specific kernel module statically, early at boot.
3520 * Unit names specified on the systemctl command line are now
3521 automatically escaped as needed. Also, if file system or
3522 device paths are specified they are automatically turned
3523 into the appropriate mount or device unit names. Example:
3525 systemctl status /home
3526 systemctl status /dev/sda
3528 * The SysVConsole= configuration option has been removed from
3529 system.conf parsing.
3531 * The SysV search path is no longer exported on the D-Bus
3534 * The Names= option is been removed from unit file parsing.
3536 * There's a new man page bootup(7) detailing the boot process.
3538 * Every unit and every generator we ship with systemd now
3539 comes with full documentation. The self-explanatory boot is
3542 * A couple of services gained "systemd-" prefixes in their
3543 name if they wrap systemd code, rather than only external
3544 code. Among them fsck@.service which is now
3545 systemd-fsck@.service.
3547 * The HaveWatchdog property has been removed from the D-Bus
3550 * systemd.confirm_spawn= on the kernel command line should now
3553 * There's a new man page crypttab(5) which details all options
3554 we actually understand.
3556 * systemd-nspawn gained a new --capability= switch to pass
3557 additional capabilities to the container.
3559 * timedated will now read known NTP implementation unit names
3560 from /usr/lib/systemd/ntp-units.d/*.list,
3561 systemd-timedated-ntp.target has been removed.
3563 * journalctl gained a new switch "-b" that lists log data of
3564 the current boot only.
3566 * The notify socket is in the abstract namespace again, in
3567 order to support daemons which chroot() at start-up.
3569 * There is a new Storage= configuration option for journald
3570 which allows configuration of where log data should go. This
3571 also provides a way to disable journal logging entirely, so
3572 that data collected is only forwarded to the console, the
3573 kernel log buffer or another syslog implementation.
3575 * Many bugfixes and optimizations
3577 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3578 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3579 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3580 Shawn Landden, Tom Gundersen
3584 * "systemctl help <unit>" now shows the man page if one is
3587 * Several new man pages have been added.
3589 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3590 MaxLevelConsole= can now be specified in
3591 journald.conf. These options allow reducing the amount of
3592 data stored on disk or forwarded by the log level.
3594 * TimerSlackNSec= can now be specified in system.conf for
3595 PID1. This allows system-wide power savings.
3597 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3598 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3603 * logind is now capable of (optionally) handling power and
3604 sleep keys as well as the lid switch.
3606 * journalctl now understands the syntax "journalctl
3607 /usr/bin/avahi-daemon" to get all log output of a specific
3610 * CapabilityBoundingSet= in system.conf now also influences
3611 the capability bound set of usermode helpers of the kernel.
3613 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3614 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3615 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3616 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3620 * Note that we skipped 139 releases here in order to set the
3621 new version to something that is greater than both udev's
3622 and systemd's most recent version number.
3624 * udev: all udev sources are merged into the systemd source tree now.
3625 All future udev development will happen in the systemd tree. It
3626 is still fully supported to use the udev daemon and tools without
3627 systemd running, like in initramfs or other init systems. Building
3628 udev though, will require the *build* of the systemd tree, but
3629 udev can be properly *run* without systemd.
3631 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3632 should be used to create dead device nodes as workarounds for broken
3635 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3636 no longer supported. udev_monitor_new_from_netlink() needs to be
3637 used to subscribe to events.
3639 * udev: when udevd is started by systemd, processes which are left
3640 behind by forking them off of udev rules, are unconditionally cleaned
3641 up and killed now after the event handling has finished. Services or
3642 daemons must be started as systemd services. Services can be
3643 pulled-in by udev to get started, but they can no longer be directly
3644 forked by udev rules.
3646 * udev: the daemon binary is called systemd-udevd now and installed
3647 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3648 to adapt to that, create symlink, or rename the binary after building
3651 * libudev no longer provides these symbols:
3652 udev_monitor_from_socket()
3653 udev_queue_get_failed_list_entry()
3654 udev_get_{dev,sys,run}_path()
3655 The versions number was bumped and symbol versioning introduced.
3657 * systemd-loginctl and systemd-journalctl have been renamed
3658 to loginctl and journalctl to match systemctl.
3660 * The config files: /etc/systemd/systemd-logind.conf and
3661 /etc/systemd/systemd-journald.conf have been renamed to
3662 logind.conf and journald.conf. Package updates should rename
3663 the files to the new names on upgrade.
3665 * For almost all files the license is now LGPL2.1+, changed
3666 from the previous GPL2.0+. Exceptions are some minor stuff
3667 of udev (which will be changed to LGPL2.1 eventually, too),
3668 and the MIT licensed sd-daemon.[ch] library that is suitable
3669 to be used as drop-in files.
3671 * systemd and logind now handle system sleep states, in
3672 particular suspending and hibernating.
3674 * logind now implements a sleep/shutdown/idle inhibiting logic
3675 suitable for a variety of uses. Soonishly Lennart will blog
3676 about this in more detail.
3678 * var-run.mount and var-lock.mount are no longer provided
3679 (which prevously bind mounted these directories to their new
3680 places). Distributions which have not converted these
3681 directories to symlinks should consider stealing these files
3682 from git history and add them downstream.
3684 * We introduced the Documentation= field for units and added
3685 this to all our shipped units. This is useful to make it
3686 easier to explore the boot and the purpose of the various
3689 * All smaller setup units (such as
3690 systemd-vconsole-setup.service) now detect properly if they
3691 are run in a container and are skipped when
3692 appropriate. This guarantees an entirely noise-free boot in
3693 Linux container environments such as systemd-nspawn.
3695 * A framework for implementing offline system updates is now
3696 integrated, for details see:
3697 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3699 * A new service type Type=idle is available now which helps us
3700 avoiding ugly interleaving of getty output and boot status
3703 * There's now a system-wide CapabilityBoundingSet= option to
3704 globally reduce the set of capabilities for the
3705 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3706 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3707 even CAP_NET_ADMIN system-wide for secure systems.
3709 * There are now system-wide DefaultLimitXXX= options to
3710 globally change the defaults of the various resource limits
3711 for all units started by PID 1.
3713 * Harald Hoyer's systemd test suite has been integrated into
3714 systemd which allows easy testing of systemd builds in qemu
3715 and nspawn. (This is really awesome! Ask us for details!)
3717 * The fstab parser is now implemented as generator, not inside
3720 * systemctl will now warn you if .mount units generated from
3721 /etc/fstab are out of date due to changes in fstab that
3722 have not been read by systemd yet.
3724 * systemd is now suitable for usage in initrds. Dracut has
3725 already been updated to make use of this. With this in place
3726 initrds get a slight bit faster but primarily are much
3727 easier to introspect and debug since "systemctl status" in
3728 the host system can be used to introspect initrd services,
3729 and the journal from the initrd is kept around too.
3731 * systemd-delta has been added, a tool to explore differences
3732 between user/admin configuration and vendor defaults.
3734 * PrivateTmp= now affects both /tmp and /var/tmp.
3736 * Boot time status messages are now much prettier and feature
3737 proper english language. Booting up systemd has never been
3740 * Read-ahead pack files now include the inode number of all
3741 files to pre-cache. When the inode changes the pre-caching
3742 is not attempted. This should be nicer to deal with updated
3743 packages which might result in changes of read-ahead
3746 * We now temporaritly lower the kernel's read_ahead_kb variable
3747 when collecting read-ahead data to ensure the kernel's
3748 built-in read-ahead does not add noise to our measurements
3749 of necessary blocks to pre-cache.
3751 * There's now RequiresMountsFor= to add automatic dependencies
3752 for all mounts necessary for a specific file system path.
3754 * MountAuto= and SwapAuto= have been removed from
3755 system.conf. Mounting file systems at boot has to take place
3758 * nspawn now learned a new switch --uuid= to set the machine
3759 ID on the command line.
3761 * nspawn now learned the -b switch to automatically search
3764 * vt102 is now the default TERM for serial TTYs, upgraded from
3767 * systemd-logind now works on VT-less systems.
3769 * The build tree has been reorganized. The individual
3770 components now have directories of their own.
3772 * A new condition type ConditionPathIsReadWrite= is now available.
3774 * nspawn learned the new -C switch to create cgroups for the
3775 container in other hierarchies.
3777 * We now have support for hardware watchdogs, configurable in
3780 * The scheduled shutdown logic now has a public API.
3782 * We now mount /tmp as tmpfs by default, but this can be
3783 masked and /etc/fstab can override it.
3785 * Since udisks does not make use of /media anymore we are not
3786 mounting a tmpfs on it anymore.
3788 * journalctl gained a new --local switch to only interleave
3789 locally generated journal files.
3791 * We can now load the IMA policy at boot automatically.
3793 * The GTK tools have been split off into a systemd-ui.
3795 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3796 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3797 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3798 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3799 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3800 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3801 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3802 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3803 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3808 * This is mostly a bugfix release
3810 * Support optional initialization of the machine ID from the
3811 KVM or container configured UUID.
3813 * Support immediate reboots with "systemctl reboot -ff"
3815 * Show /etc/os-release data in systemd-analyze output
3817 * Many bugfixes for the journal, including endianness fixes and
3818 ensuring that disk space enforcement works
3820 * sd-login.h is C++ comptaible again
3822 * Extend the /etc/os-release format on request of the Debian
3825 * We now refuse non-UTF8 strings used in various configuration
3826 and unit files. This is done to ensure we do not pass invalid
3827 data over D-Bus or expose it elsewhere.
3829 * Register Mimo USB Screens as suitable for automatic seat
3832 * Read SELinux client context from journal clients in a race
3835 * Reorder configuration file lookup order. /etc now always
3836 overrides /run in order to allow the administrator to always
3837 and unconditionally override vendor supplied or
3838 automatically generated data.
3840 * The various user visible bits of the journal now have man
3841 pages. We still lack man pages for the journal API calls
3844 * We now ship all man pages in HTML format again in the
3847 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3848 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3849 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3854 * This is mostly a bugfix release
3856 * systems lacking /etc/os-release are no longer supported.
3858 * Various functionality updates to libsystemd-login.so
3860 * Track class of PAM logins to distuingish greeters from
3863 Contributions from: Kay Sievers, Lennart Poettering, Michael
3868 * This is an important bugfix release for v41.
3870 * Building man pages is now optional which should be useful
3871 for those building systemd from git but unwilling to install
3874 * Watchdog support for supervising services is now usable. In
3875 a future release support for hardware watchdogs
3876 (i.e. /dev/watchdog) will be added building on this.
3878 * Service start rate limiting is now configurable and can be
3879 turned off per service. When a start rate limit is hit a
3880 reboot can automatically be triggered.
3882 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3884 Contributions from: Benjamin Franzke, Bill Nottingham,
3885 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3886 Schmidt, Michał Górny, Piotr Drąg
3890 * The systemd binary is installed /usr/lib/systemd/systemd now;
3891 An existing /sbin/init symlink needs to be adapted with the
3894 * The code that loads kernel modules has been ported to invoke
3895 libkmod directly, instead of modprobe. This means we do not
3896 support systems with module-init-tools anymore.
3898 * Watchdog support is now already useful, but still not
3901 * A new kernel command line option systemd.setenv= is
3902 understood to set system wide environment variables
3903 dynamically at boot.
3905 * We now limit the set of capabilities of systemd-journald.
3907 * We now set SIGPIPE to ignore by default, since it only is
3908 useful in shell pipelines, and has little use in general
3909 code. This can be disabled with IgnoreSIPIPE=no in unit
3912 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3913 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3918 * This is mostly a bugfix release
3920 * We now expose the reason why a service failed in the
3921 "Result" D-Bus property.
3923 * Rudimentary service watchdog support (will be completed over
3924 the next few releases.)
3926 * When systemd forks off in order execute some service we will
3927 now immediately changes its argv[0] to reflect which process
3928 it will execute. This is useful to minimize the time window
3929 with a generic argv[0], which makes bootcharts more useful
3931 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3932 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3933 Mike Kazantsev, Ray Strode
3937 * This is mostly a test release, but incorporates many
3940 * New systemd-cgtop tool to show control groups by their
3943 * Linking against libacl for ACLs is optional again. If
3944 disabled, support tracking device access for active logins
3945 goes becomes unavailable, and so does access to the user
3946 journals by the respective users.
3948 * If a group "adm" exists, journal files are automatically
3949 owned by them, thus allow members of this group full access
3950 to the system journal as well as all user journals.
3952 * The journal now stores the SELinux context of the logging
3953 client for all entries.
3955 * Add C++ inclusion guards to all public headers
3957 * New output mode "cat" in the journal to print only text
3958 messages, without any meta data like date or time.
3960 * Include tiny X server wrapper as a temporary stop-gap to
3961 teach XOrg udev display enumeration. This is used by display
3962 managers such as gdm, and will go away as soon as XOrg
3963 learned native udev hotplugging for display devices.
3965 * Add new systemd-cat tool for executing arbitrary programs
3966 with STDERR/STDOUT connected to the journal. Can also act as
3967 BSD logger replacement, and does so by default.
3969 * Optionally store all locally generated coredumps in the
3970 journal along with meta data.
3972 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3973 writing short strings to files (for usage for /sys), and for
3974 creating symlinks, character and block device nodes.
3976 * New unit file option ControlGroupPersistent= to make cgroups
3977 persistent, following the mechanisms outlined in
3978 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3980 * Support multiple local RTCs in a sane way
3982 * No longer monopolize IO when replaying readahead data on
3983 rotating disks, since we might starve non-file-system IO to
3984 death, since fanotify() will not see accesses done by blkid,
3987 * Do not show kernel threads in systemd-cgls anymore, unless
3988 requested with new -k switch.
3990 Contributions from: Dan Horák, Kay Sievers, Lennart
3991 Poettering, Michal Schmidt
3995 * This is mostly a test release, but incorporates many
3998 * The git repository moved to:
3999 git://anongit.freedesktop.org/systemd/systemd
4000 ssh://git.freedesktop.org/git/systemd/systemd
4002 * First release with the journal
4003 http://0pointer.de/blog/projects/the-journal.html
4005 * The journal replaces both systemd-kmsg-syslogd and
4006 systemd-stdout-bridge.
4008 * New sd_pid_get_unit() API call in libsystemd-logind
4010 * Many systemadm clean-ups
4012 * Introduce remote-fs-pre.target which is ordered before all
4013 remote mounts and may be used to start services before all
4016 * Added Mageia support
4018 * Add bash completion for systemd-loginctl
4020 * Actively monitor PID file creation for daemons which exit in
4021 the parent process before having finished writing the PID
4022 file in the daemon process. Daemons which do this need to be
4023 fixed (i.e. PID file creation must have finished before the
4024 parent exits), but we now react a bit more gracefully to them.
4026 * Add colourful boot output, mimicking the well-known output
4027 of existing distributions.
4029 * New option PassCredentials= for socket units, for
4030 compatibility with a recent kernel ABI breakage.
4032 * /etc/rc.local is now hooked in via a generator binary, and
4033 thus will no longer act as synchronization point during
4036 * systemctl list-unit-files now supports --root=.
4038 * systemd-tmpfiles now understands two new commands: z, Z for
4039 relabelling files according to the SELinux database. This is
4040 useful to apply SELinux labels to specific files in /sys,
4043 * Output of SysV services is now forwarded to both the console
4044 and the journal by default, not only just the console.
4046 * New man pages for all APIs from libsystemd-login.
4048 * The build tree got reorganized and a the build system is a
4049 lot more modular allowing embedded setups to specifically
4050 select the components of systemd they are interested in.
4052 * Support for Linux systems lacking the kernel VT subsystem is
4055 * configure's --with-rootdir= got renamed to
4056 --with-rootprefix= to follow the naming used by udev and
4059 * Unless specified otherwise we will now install to /usr instead
4060 of /usr/local by default.
4062 * Processes with '@' in argv[0][0] are now excluded from the
4063 final shut-down killing spree, following the logic explained
4065 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4067 * All processes remaining in a service cgroup when we enter
4068 the START or START_PRE states are now killed with
4069 SIGKILL. That means it is no longer possible to spawn
4070 background processes from ExecStart= lines (which was never
4071 supported anyway, and bad style).
4073 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4074 reloading of units together.
4076 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4077 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4078 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4079 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4080 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek