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 * A user console daemon systemd-consoled has been added. It is
41 a preview, and will so far open a single terminal on each
42 session of the user marked as Desktop=SYSTEMD-CONSOLE.
44 * Route metrics can be specified for DHCP routes added by
47 * SELinux context of socket-actived services can be set from
48 the information provided by the remote peer
49 (SELinuxContextFromNet= option).
51 * Userspace firmware loading support has been removed and
52 the minimum supported kernel version is thus bumped to 3.7.
54 * Timeout for udev workers has been increased from 1 to 3
55 minutes, but a warning will be printed after 1 minute to
56 help diagnose kernel modules that take a long time to load.
58 * Udev rules can now remove tags with TAG-="foobar".
60 * Readhead implementation has been removed. In many
61 circumstatances it didn't give expected benefits even for
62 rotational disk drives and was becoming less relevant in the
65 * Swap units can use Discard= to specify discard options.
66 Discard options specified for swaps in /etc/fstab are now
69 * Docker containers are now detected as a separate type of
72 * The Password Agent protocol gained support for queries where
73 the user input is shown, useful e.g. for usernames.
74 systemd-ask-password gained a new --echo option to turn
79 * timedated no longer reads NTP implementation unit names from
80 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
81 implementations should add a
83 Conflicts=systemd-timesyncd.service
85 to their unit files to take over and replace systemd's NTP
86 default functionality.
88 * systemd-sysusers gained a new line type "r" for configuring
89 which UID/GID ranges to allocate system users/groups
90 from. Lines of type "u" may now add an additional column
91 that specifies the home directory for the system user to be
92 created. Also, systemd-sysusers may now optionally read user
93 information from STDIN instead of a file. This is useful for
94 invoking it from RPM preinst scriptlets that need to create
95 users before the first RPM file is installed since these
96 files might need to be owned by them. A new
97 %sysusers_create_inline RPM macro has been introduced to do
98 just that. systemd-sysusers now updates the shadow files as
99 well as the user/group databases, which should enhance
100 compatibility with certain tools like grpck.
102 * A number of bus APIs of PID 1 now optionally consult
103 PolicyKit to permit access for otherwise unprivileged
104 clients under certain conditions. Note that this currently
105 doesn't support interactive authentication yet, but this is
106 expected to be added eventually, too.
108 * /etc/machine-info now has new fields for configuring the
109 deployment environment of the machine, as well as the
110 location of the machine. hostnamectl has been updated with
111 new command to update these fields.
113 * systemd-timesyncd has been updated to automatically acquire
114 NTP server information from systemd-networkd, which might
115 have been discovered via DHCP.
117 * systemd-resolved now includes a caching DNS stub resolver
118 and a complete LLMNR name resolution implementation. A new
119 NSS module "nss-resolve" has been added which can be used
120 instead of glibc's own "nss-dns" to resolve hostnames via
121 systemd-resolved. Hostnames, addresses and arbitrary RRs may
122 be resolved via systemd-resolved D-Bus APIs. In contrast to
123 the glibc internal resolver systemd-resolved is aware of
124 multi-homed system, and keeps DNS server and caches separate
125 and per-interface. Queries are sent simultaneously on all
126 interfaces that have DNS servers configured, in order to
127 properly handle VPNs and local LANs which might resolve
128 separate sets of domain names. systemd-resolved may acquire
129 DNS server information from systemd-networkd automatically,
130 which in turn might have discovered them via DHCP. A tool
131 "systemd-resolve-host" has been added that may be used to
132 query the DNS logic in resolved. systemd-resolved implements
133 IDNA and automatically uses IDNA or UTF-8 encoding depending
134 on whether classic DNS or LLMNR is used as transport. In the
135 next releases we intend to add a DNSSEC and mDNS/DNS-SD
136 implementation to systemd-resolved.
138 * A new NSS module nss-mymachines has been added, that
139 automatically resolves the names of all local registered
140 containers to their respective IP addresses.
142 * A new client tool "networkctl" for systemd-networkd has been
143 added. It currently is entirely passive and will query
144 networking configuration from udev, rtnetlink and networkd,
145 and present it to the user in a very friendly
146 way. Eventually, we hope to extend it to become a full
147 control utility for networkd.
149 * .socket units gained a new DeferAcceptSec= setting that
150 controls the kernels' TCP_DEFER_ACCEPT sockopt for
151 TCP. Similar, support for controlling TCP keep-alive
152 settings has been added (KeepAliveTimeSec=,
153 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
154 turning off Nagle's algorithm on TCP has been added
157 * logind learned a new session type "web", for use in projects
158 like Cockpit which register web clients as PAM sessions.
160 * timer units with at least one OnCalendar= setting will now
161 be started only after timer-sync.target has been
162 reached. This way they will not elapse before the system
163 clock has been corrected by a local NTP client or
164 similar. This is particular useful on RTC-less embedded
165 machines, that come up with an invalid system clock.
167 * systemd-nspawn's --network-veth= switch should now result in
168 stable MAC addresses for both the outer and the inner side
171 * systemd-nspawn gained a new --volatile= switch for running
172 container instances with /etc or /var unpopulated.
174 * The kdbus client code has been updated to use the new Linux
175 3.17 memfd subsystem instead of the old kdbus-specific one.
177 * systemd-networkd's DHCP client and server now support
178 FORCERENEW. There are also new configuration options to
179 configure the vendor client identifier and broadcast mode
182 * systemd will no longer inform the kernel about the current
183 timezone, as this is necessarily incorrect and racy as the
184 kernel has no understanding of DST and similar
185 concepts. This hence means FAT timestamps will be always
186 considered UTC, similar to what Android is already
187 doing. Also, when the RTC is configured to the local time
188 (rather than UTC) systemd will never synchronize back to it,
189 as this might confuse Windows at a later boot.
191 * systemd-analyze gained a new command "verify" for offline
192 validation of unit files.
194 * systemd-networkd gained support for a couple of additional
195 settings for bonding networking setups. Also, the metric for
196 statically configured routes may now be configured. For
197 network interfaces where this is appropriate the peer IP
198 address may now be configured.
200 * systemd-networkd's DHCP client will no longer request
201 broadcasting by default, as this tripped up some networks.
202 For hardware where broadcast is required the feature should
203 be switched back on using RequestBroadcast=yes.
205 * systemd-networkd will now set up IPv4LL addresses (when
206 enabled) even if DHCP is configured successfully.
208 * udev will now default to respect network device names given
209 by the kernel when the kernel indicates that these are
210 predictable. This behavior can be tweaked by changing
211 NamePolicy= in the relevant .link file.
213 * A new library systemd-terminal has been added that
214 implements full TTY stream parsing and rendering. This
215 library is supposed to be used later on for implementing a
216 full userspace VT subsystem, replacing the current kernel
219 * A new tool systemd-journal-upload has been added to push
220 journal data to a remote system running
221 systemd-journal-remote.
223 * journald will no longer forward all local data to another
224 running syslog daemon. This change has been made because
225 rsyslog (which appears to be the most commonly used syslog
226 implementation these days) no longer makes use of this, and
227 instead pulls the data out of the journal on its own. Since
228 forwarding the messages to a non-existent syslog server is
229 more expensive than we assumed we have now turned this
230 off. If you run a syslog server that is not a recent rsyslog
231 version, you have to turn this option on again
232 (ForwardToSyslog= in journald.conf).
234 * journald now optionally supports the LZ4 compressor for
235 larger journal fields. This compressor should perform much
236 better than XZ which was the previous default.
238 * machinectl now shows the IP addresses of local containers,
239 if it knows them, plus the interface name of the container.
241 * A new tool "systemd-escape" has been added that makes it
242 easy to escape strings to build unit names and similar.
244 * sd_notify() messages may now include a new ERRNO= field
245 which is parsed and collected by systemd and shown among the
246 "systemctl status" output for a service.
248 * A new component "systemd-firstboot" has been added that
249 queries the most basic systemd information (timezone,
250 hostname, root password) interactively on first
251 boot. Alternatively it may also be used to provision these
252 things offline on OS images installed into directories.
254 * The default sysctl.d/ snippets will now set
256 net.ipv4.conf.default.promote_secondaries=1
258 This has the benefit of no flushing secondary IP addresses
259 when primary addresses are removed.
261 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
262 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
263 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
264 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
265 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
266 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
267 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
268 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
269 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
270 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
271 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
272 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
273 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
274 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
275 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
277 -- Berlin, 2014-08-19
281 * A new tool systemd-sysusers has been added. This tool
282 creates system users and groups in /etc/passwd and
283 /etc/group, based on static declarative system user/group
284 definitions in /usr/lib/sysusers.d/. This is useful to
285 enable factory resets and volatile systems that boot up with
286 an empty /etc directory, and thus need system users and
287 groups created during early boot. systemd now also ships
288 with two default sysusers.d/ files for the most basic
289 users and groups systemd and the core operating system
292 * A new tmpfiles snippet has been added that rebuilds the
293 essential files in /etc on boot, should they be missing.
295 * A directive for ensuring automatic clean-up of
296 /var/cache/man/ has been removed from the default
297 configuration. This line should now be shipped by the man
298 implementation. The necessary change has been made to the
299 man-db implementation. Note that you need to update your man
300 implementation to one that ships this line, otherwise no
301 automatic clean-up of /var/cache/man will take place.
303 * A new condition ConditionNeedsUpdate= has been added that
304 may conditionalize services to only run when /etc or /var
305 are "older" than the vendor operating system resources in
306 /usr. This is useful for reconstructing or updating /etc
307 after an offline update of /usr or a factory reset, on the
308 next reboot. Services that want to run once after such an
309 update or reset should use this condition and order
310 themselves before the new systemd-update-done.service, which
311 will mark the two directories as fully updated. A number of
312 service files have been added making use of this, to rebuild
313 the udev hardware database, the journald message catalog and
314 dynamic loader cache (ldconfig). The systemd-sysusers tool
315 described above also makes use of this now. With this in
316 place it is now possible to start up a minimal operating
317 system with /etc empty cleanly. For more information on the
318 concepts involved see this recent blog story:
320 http://0pointer.de/blog/projects/stateless.html
322 * A new system group "input" has been introduced, and all
323 input device nodes get this group assigned. This is useful
324 for system-level software to get access to input devices. It
325 complements what is already done for "audio" and "video".
327 * systemd-networkd learnt minimal DHCPv4 server support in
328 addition to the existing DHCPv4 client support. It also
329 learnt DHCPv6 client and IPv6 Router Solicitation client
330 support. The DHCPv4 client gained support for static routes
331 passed in from the server. Note that the [DHCPv4] section
332 known in older systemd-networkd versions has been renamed to
333 [DHCP] and is now also used by the DHCPv6 client. Existing
334 .network files using settings of this section should be
335 updated, though compatibility is maintained. Optionally, the
336 client hostname may now be sent to the DHCP server.
338 * networkd gained support for vxlan virtual networks as well
339 as tun/tap and dummy devices.
341 * networkd gained support for automatic allocation of address
342 ranges for interfaces from a system-wide pool of
343 addresses. This is useful for dynamically managing a large
344 number of interfaces with a single network configuration
345 file. In particular this is useful to easily assign
346 appropriate IP addresses to the veth links of a large number
349 * RPM macros for processing sysusers, sysctl and binfmt
350 drop-in snippets at package installation time have been
353 * The /etc/os-release file should now be placed in
354 /usr/lib/os-release. The old location is automatically
355 created as symlink. /usr/lib is the more appropriate
356 location of this file, since it shall actually describe the
357 vendor operating system shipped in /usr, and not the
358 configuration stored in /etc.
360 * .mount units gained a new boolean SloppyOptions= setting
361 that maps to mount(8)'s -s option which enables permissive
362 parsing of unknown mount options.
364 * tmpfiles learnt a new "L+" directive which creates a symlink
365 but (unlike "L") deletes a pre-existing file first, should
366 it already exist and not already be the correct
367 symlink. Similar, "b+", "c+" and "p+" directives have been
368 added as well, which create block and character devices, as
369 well as fifos in the filesystem, possibly removing any
370 pre-existing files of different types.
372 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
373 'argument' field (which so far specified the source to
374 symlink/copy the files from) is now optional. If omitted the
375 same file os copied from /usr/share/factory/ suffixed by the
376 full destination path. This is useful for populating /etc
377 with essential files, by copying them from vendor defaults
378 shipped in /usr/share/factory/etc.
380 * A new command "systemctl preset-all" has been added that
381 applies the service preset settings to all installed unit
382 files. A new switch --preset-mode= has been added that
383 controls whether only enable or only disable operations
386 * A new command "systemctl is-system-running" has been added
387 that allows checking the overall state of the system, for
388 example whether it is fully up and running.
390 * When the system boots up with an empty /etc, the equivalent
391 to "systemctl preset-all" is executed during early boot, to
392 make sure all default services are enabled after a factory
395 * systemd now contains a minimal preset file that enables the
396 most basic services systemd ships by default.
398 * Unit files' [Install] section gained a new DefaultInstance=
399 field for defining the default instance to create if a
400 template unit is enabled with no instance specified.
402 * A new passive target cryptsetup-pre.target has been added
403 that may be used by services that need to make they run and
404 finish before the first LUKS cryptographic device is set up.
406 * The /dev/loop-control and /dev/btrfs-control device nodes
407 are now owned by the "disk" group by default, opening up
408 access to this group.
410 * systemd-coredump will now automatically generate a
411 stack trace of all core dumps taking place on the system,
412 based on elfutils' libdw library. This stack trace is logged
415 * systemd-coredump may now optionally store coredumps directly
416 on disk (in /var/lib/systemd/coredump, possibly compressed),
417 instead of storing them unconditionally in the journal. This
418 mode is the new default. A new configuration file
419 /etc/systemd/coredump.conf has been added to configure this
420 and other parameters of systemd-coredump.
422 * coredumpctl gained a new "info" verb to show details about a
423 specific coredump. A new switch "-1" has also been added
424 that makes sure to only show information about the most
425 recent entry instead of all entries. Also, as the tool is
426 generally useful now the "systemd-" prefix of the binary
427 name has been removed. Distributions that want to maintain
428 compatibility with the old name should add a symlink from
429 the old name to the new name.
431 * journald's SplitMode= now defaults to "uid". This makes sure
432 that unprivileged users can access their own coredumps with
433 coredumpctl without restrictions.
435 * New kernel command line options "systemd.wants=" (for
436 pulling an additional unit during boot), "systemd.mask="
437 (for masking a specific unit for the boot), and
438 "systemd.debug-shell" (for enabling the debug shell on tty9)
439 have been added. This is implemented in the new generator
440 "systemd-debug-generator".
442 * systemd-nspawn will now by default filter a couple of
443 syscalls for containers, among them those required for
444 kernel module loading, direct x86 IO port access, swap
445 management, and kexec. Most importantly though
446 open_by_handle_at() is now prohibited for containers,
447 closing a hole similar to a recently discussed vulnerability
448 in docker regarding access to files on file hierarchies the
449 container should normally not have access to. Note that for
450 nspawn we generally make no security claims anyway (and
451 this is explicitly documented in the man page), so this is
452 just a fix for one of the most obvious problems.
454 * A new man page file-hierarchy(7) has been added that
455 contains a minimized, modernized version of the file system
456 layout systemd expects, similar in style to the FHS
457 specification or hier(5). A new tool systemd-path(1) has
458 been added to query many of these paths for the local
461 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
462 longer done. Since the directory now has a per-user size
463 limit, and is cleaned on logout this appears unnecessary,
464 in particular since this now brings the lifecycle of this
465 directory closer in line with how IPC objects are handled.
467 * systemd.pc now exports a number of additional directories,
468 including $libdir (which is useful to identify the library
469 path for the primary architecture of the system), and a
470 couple of drop-in directories.
472 * udev's predictable network interface names now use the dev_port
473 sysfs attribute, introduced in linux 3.15 instead of dev_id to
474 distinguish between ports of the same PCI function. dev_id should
475 only be used for ports using the same HW address, hence the need
478 * machined has been updated to export the OS version of a
479 container (read from /etc/os-release and
480 /usr/lib/os-release) on the bus. This is now shown in
481 "machinectl status" for a machine.
483 * A new service setting RestartForceExitStatus= has been
484 added. If configured to a set of exit signals or process
485 return values, the service will be restarted when the main
486 daemon process exits with any of them, regardless of the
489 * systemctl's -H switch for connecting to remote systemd
490 machines has been extended so that it may be used to
491 directly connect to a specific container on the
492 host. "systemctl -H root@foobar:waldi" will now connect as
493 user "root" to host "foobar", and then proceed directly to
494 the container named "waldi". Note that currently you have to
495 authenticate as user "root" for this to work, as entering
496 containers is a privileged operation.
498 Contributions from: Andreas Henriksson, Benjamin Steinwender,
499 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
500 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
501 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
502 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
503 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
504 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
505 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
506 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
507 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
508 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
509 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
511 -- Berlin, 2014-07-03
515 * As an experimental feature, udev now tries to lock the
516 disk device node (flock(LOCK_SH|LOCK_NB)) while it
517 executes events for the disk or any of its partitions.
518 Applications like partitioning programs can lock the
519 disk device node (flock(LOCK_EX)) and claim temporary
520 device ownership that way; udev will entirely skip all event
521 handling for this disk and its partitions. If the disk
522 was opened for writing, the close will trigger a partition
523 table rescan in udev's "watch" facility, and if needed
524 synthesize "change" events for the disk and all its partitions.
525 This is now unconditionally enabled, and if it turns out to
526 cause major problems, we might turn it on only for specific
527 devices, or might need to disable it entirely. Device Mapper
528 devices are excluded from this logic.
530 * We temporarily dropped the "-l" switch for fsck invocations,
531 since they collide with the flock() logic above. util-linux
532 upstream has been changed already to avoid this conflict,
533 and we will readd "-l" as soon as util-linux with this
534 change has been released.
536 * The dependency on libattr has been removed. Since a long
537 time, the extended attribute calls have moved to glibc, and
538 libattr is thus unnecessary.
540 * Virtualization detection works without priviliges now. This
541 means the systemd-detect-virt binary no longer requires
542 CAP_SYS_PTRACE file capabilities, and our daemons can run
543 with fewer privileges.
545 * systemd-networkd now runs under its own "systemd-network"
546 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
547 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
548 loses the ability to write to files owned by root this way.
550 * Similar, systemd-resolved now runs under its own
551 "systemd-resolve" user with no capabilities remaining.
553 * Similar, systemd-bus-proxyd now runs under its own
554 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
556 * systemd-networkd gained support for setting up "veth"
557 virtual ethernet devices for container connectivity, as well
558 as GRE and VTI tunnels.
560 * systemd-networkd will no longer automatically attempt to
561 manually load kernel modules necessary for certain tunnel
562 transports. Instead, it is assumed the kernel loads them
563 automatically when required. This only works correctly on
564 very new kernels. On older kernels, please consider adding
565 the kernel modules to /etc/modules-load.d/ as a work-around.
567 * The resolv.conf file systemd-resolved generates has been
568 moved to /run/systemd/resolve/. If you have a symlink from
569 /etc/resolv.conf, it might be necessary to correct it.
571 * Two new service settings, ProtectHome= and ProtectSystem=,
572 have been added. When enabled, they will make the user data
573 (such as /home) inaccessible or read-only and the system
574 (such as /usr) read-only, for specific services. This allows
575 very light-weight per-service sandboxing to avoid
576 modifications of user data or system files from
577 services. These two new switches have been enabled for all
578 of systemd's long-running services, where appropriate.
580 * Socket units gained new SocketUser= and SocketGroup=
581 settings to set the owner user and group of AF_UNIX sockets
582 and FIFOs in the file system.
584 * Socket units gained a new RemoveOnStop= setting. If enabled,
585 all FIFOS and sockets in the file system will be removed
586 when the specific socket unit is stopped.
588 * Socket units gained a new Symlinks= setting. It takes a list
589 of symlinks to create to file system sockets or FIFOs
590 created by the specific Unix sockets. This is useful to
591 manage symlinks to socket nodes with the same life-cycle as
594 * The /dev/log socket and /dev/initctl FIFO have been moved to
595 /run, and have been replaced by symlinks. This allows
596 connecting to these facilities even if PrivateDevices=yes is
597 used for a service (which makes /dev/log itself unavailable,
598 but /run is left). This also has the benefit of ensuring
599 that /dev only contains device nodes, directories and
600 symlinks, and nothing else.
602 * sd-daemon gained two new calls sd_pid_notify() and
603 sd_pid_notifyf(). They are similar to sd_notify() and
604 sd_notifyf(), but allow overriding of the source PID of
605 notification messages if permissions permit this. This is
606 useful to send notify messages on behalf of a different
607 process (for example, the parent process). The
608 systemd-notify tool has been updated to make use of this
609 when sending messages (so that notification messages now
610 originate from the shell script invoking systemd-notify and
611 not the systemd-notify process itself. This should minimize
612 a race where systemd fails to associate notification
613 messages to services when the originating process already
616 * A new "on-abnormal" setting for Restart= has been added. If
617 set, it will result in automatic restarts on all "abnormal"
618 reasons for a process to exit, which includes unclean
619 signals, core dumps, timeouts and watchdog timeouts, but
620 does not include clean and unclean exit codes or clean
621 signals. Restart=on-abnormal is an alternative for
622 Restart=on-failure for services that shall be able to
623 terminate and avoid restarts on certain errors, by
624 indicating so with an unclean exit code. Restart=on-failure
625 or Restart=on-abnormal is now the recommended setting for
626 all long-running services.
628 * If the InaccessibleDirectories= service setting points to a
629 mount point (or if there are any submounts contained within
630 it), it is now attempted to completely unmount it, to make
631 the file systems truly unavailable for the respective
634 * The ReadOnlyDirectories= service setting and
635 systemd-nspawn's --read-only parameter are now recursively
636 applied to all submounts, too.
638 * Mount units may now be created transiently via the bus APIs.
640 * The support for SysV and LSB init scripts has been removed
641 from the systemd daemon itself. Instead, it is now
642 implemented as a generator that creates native systemd units
643 from these scripts when needed. This enables us to remove a
644 substantial amount of legacy code from PID 1, following the
645 fact that many distributions only ship a very small number
646 of LSB/SysV init scripts nowadays.
648 * Priviliged Xen (dom0) domains are not considered
649 virtualization anymore by the virtualization detection
650 logic. After all, they generally have unrestricted access to
651 the hardware and usually are used to manage the unprivileged
654 * systemd-tmpfiles gained a new "C" line type, for copying
655 files or entire directories.
657 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
658 lines. So far, they have been non-globbing versions of the
659 latter, and have thus been redundant. In future, it is
660 recommended to only use "z". "m" has hence been removed
661 from the documentation, even though it stays supported.
663 * A tmpfiles snippet to recreate the most basic structure in
664 /var has been added. This is enough to create the /var/run →
665 /run symlink and create a couple of structural
666 directories. This allows systems to boot up with an empty or
667 volatile /var. Of course, while with this change, the core OS
668 now is capable with dealing with a volatile /var, not all
669 user services are ready for it. However, we hope that sooner
670 or later, many service daemons will be changed upstream so
671 that they are able to automatically create their necessary
672 directories in /var at boot, should they be missing. This is
673 the first step to allow state-less systems that only require
674 the vendor image for /usr to boot.
676 * systemd-nspawn has gained a new --tmpfs= switch to mount an
677 empty tmpfs instance to a specific directory. This is
678 particularly useful for making use of the automatic
679 reconstruction of /var (see above), by passing --tmpfs=/var.
681 * Access modes specified in tmpfiles snippets may now be
682 prefixed with "~", which indicates that they shall be masked
683 by whether the existing file or directory is currently
684 writable, readable or executable at all. Also, if specified,
685 the sgid/suid/sticky bits will be masked for all
688 * A new passive target unit "network-pre.target" has been
689 added which is useful for services that shall run before any
690 network is configured, for example firewall scripts.
692 * The "floppy" group that previously owned the /dev/fd*
693 devices is no longer used. The "disk" group is now used
694 instead. Distributions should probably deprecate usage of
697 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
698 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
699 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
700 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
701 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
702 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
705 -- Berlin, 2014-06-11
709 * A new "systemd-timesyncd" daemon has been added for
710 synchronizing the system clock across the network. It
711 implements an SNTP client. In contrast to NTP
712 implementations such as chrony or the NTP reference server,
713 this only implements a client side, and does not bother with
714 the full NTP complexity, focusing only on querying time from
715 one remote server and synchronizing the local clock to
716 it. Unless you intend to serve NTP to networked clients or
717 want to connect to local hardware clocks, this simple NTP
718 client should be more than appropriate for most
719 installations. The daemon runs with minimal privileges, and
720 has been hooked up with networkd to only operate when
721 network connectivity is available. The daemon saves the
722 current clock to disk every time a new NTP sync has been
723 acquired, and uses this to possibly correct the system clock
724 early at bootup, in order to accommodate for systems that
725 lack an RTC such as the Raspberry Pi and embedded devices,
726 and to make sure that time monotonically progresses on these
727 systems, even if it is not always correct. To make use of
728 this daemon, a new system user and group "systemd-timesync"
729 needs to be created on installation of systemd.
731 * The queue "seqnum" interface of libudev has been disabled, as
732 it was generally incompatible with device namespacing as
733 sequence numbers of devices go "missing" if the devices are
734 part of a different namespace.
736 * "systemctl list-timers" and "systemctl list-sockets" gained
737 a --recursive switch for showing units of these types also
738 for all local containers, similar in style to the already
739 supported --recursive switch for "systemctl list-units".
741 * A new RebootArgument= setting has been added for service
742 units, which may be used to specify a kernel reboot argument
743 to use when triggering reboots with StartLimitAction=.
745 * A new FailureAction= setting has been added for service
746 units which may be used to specify an operation to trigger
747 when a service fails. This works similarly to
748 StartLimitAction=, but unlike it, controls what is done
749 immediately rather than only after several attempts to
750 restart the service in question.
752 * hostnamed got updated to also expose the kernel name,
753 release, and version on the bus. This is useful for
754 executing commands like hostnamectl with the -H switch.
755 systemd-analyze makes use of this to properly display
756 details when running non-locally.
758 * The bootchart tool can now show cgroup information in the
761 * The CFS CPU quota cgroup attribute is now exposed for
762 services. The new CPUQuota= switch has been added for this
763 which takes a percentage value. Setting this will have the
764 result that a service may never get more CPU time than the
765 specified percentage, even if the machine is otherwise idle.
767 * systemd-networkd learned IPIP and SIT tunnel support.
769 * LSB init scripts exposing a dependency on $network will now
770 get a dependency on network-online.target rather than simply
771 network.target. This should bring LSB handling closer to
772 what it was on SysV systems.
774 * A new fsck.repair= kernel option has been added to control
775 how fsck shall deal with unclean file systems at boot.
777 * The (.ini) configuration file parser will now silently
778 ignore sections whose name begins with "X-". This may be
779 used to maintain application-specific extension sections in unit
782 * machined gained a new API to query the IP addresses of
783 registered containers. "machinectl status" has been updated
784 to show these addresses in its output.
786 * A new call sd_uid_get_display() has been added to the
787 sd-login APIs for querying the "primary" session of a
788 user. The "primary" session of the user is elected from the
789 user's sessions and generally a graphical session is
790 preferred over a text one.
792 * A minimal systemd-resolved daemon has been added. It
793 currently simply acts as a companion to systemd-networkd and
794 manages resolv.conf based on per-interface DNS
795 configuration, possibly supplied via DHCP. In the long run
796 we hope to extend this into a local DNSSEC enabled DNS and
799 * The systemd-networkd-wait-online tool is now enabled by
800 default. It will delay network-online.target until a network
801 connection has been configured. The tool primarily integrates
802 with networkd, but will also make a best effort to make sense
803 of network configuration performed in some other way.
805 * Two new service options StartupCPUShares= and
806 StartupBlockIOWeight= have been added that work similarly to
807 CPUShares= and BlockIOWeight= however only apply during
808 system startup. This is useful to prioritize certain services
809 differently during bootup than during normal runtime.
811 * hostnamed has been changed to prefer the statically
812 configured hostname in /etc/hostname (unless set to
813 'localhost' or empty) over any dynamic one supplied by
814 dhcp. With this change, the rules for picking the hostname
815 match more closely the rules of other configuration settings
816 where the local administrator's configuration in /etc always
817 overrides any other settings.
819 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
820 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
821 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
822 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
823 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
824 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
825 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
826 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
827 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
828 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
829 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
830 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
831 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
832 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
833 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
834 Lindskog, WaLyong Cho, Will Woods, Zbigniew
837 -- Beijing, 2014-05-28
841 * When restoring the screen brightness at boot, stay away from
842 the darkest setting or from the lowest 5% of the available
843 range, depending on which is the larger value of both. This
844 should effectively protect the user from rebooting into a
845 black screen, should the brightness have been set to minimum
848 * sd-login gained a new sd_machine_get_class() call to
849 determine the class ("vm" or "container") of a machine
850 registered with machined.
852 * sd-login gained new calls
853 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
854 to query the identity of the peer of a local AF_UNIX
855 connection. They operate similarly to their sd_pid_get_xyz()
858 * PID 1 will now maintain a system-wide system state engine
859 with the states "starting", "running", "degraded",
860 "maintenance", "stopping". These states are bound to system
861 startup, normal runtime, runtime with at least one failed
862 service, rescue/emergency mode and system shutdown. This
863 state is shown in the "systemctl status" output when no unit
864 name is passed. It is useful to determine system state, in
865 particularly when doing so for many systems or containers at
868 * A new command "list-machines" has been added to "systemctl"
869 that lists all local OS containers and shows their system
870 state (see above), if systemd runs inside of them.
872 * systemctl gained a new "-r" switch to recursively enumerate
873 units on all local containers, when used with the
874 "list-unit" command (which is the default one that is
875 executed when no parameters are specified).
877 * The GPT automatic partition discovery logic will now honour
878 two GPT partition flags: one may be set on a partition to
879 cause it to be mounted read-only, and the other may be set
880 on a partition to ignore it during automatic discovery.
882 * Two new GPT type UUIDs have been added for automatic root
883 partition discovery, for 32-bit and 64-bit ARM. This is not
884 particularly useful for discovering the root directory on
885 these architectures during bare-metal boots (since UEFI is
886 not common there), but still very useful to allow booting of
887 ARM disk images in nspawn with the -i option.
889 * MAC addresses of interfaces created with nspawn's
890 --network-interface= switch will now be generated from the
891 machine name, and thus be stable between multiple invocations
894 * logind will now automatically remove all IPC objects owned
895 by a user if she or he fully logs out. This makes sure that
896 users who are logged out cannot continue to consume IPC
897 resources. This covers SysV memory, semaphores and message
898 queues as well as POSIX shared memory and message
899 queues. Traditionally, SysV and POSIX IPC had no life-cycle
900 limits. With this functionality, that is corrected. This may
901 be turned off by using the RemoveIPC= switch of logind.conf.
903 * The systemd-machine-id-setup and tmpfiles tools gained a
904 --root= switch to operate on a specific root directory,
907 * journald can now forward logged messages to the TTYs of all
908 logged in users ("wall"). This is the default for all
909 emergency messages now.
911 * A new tool systemd-journal-remote has been added to stream
912 journal log messages across the network.
914 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
915 controller trees are mounted into it. Note that the
916 directories mounted beneath it are not read-only. This is a
917 security measure and is particularly useful because glibc
918 actually includes a search logic to pick any tmpfs it can
919 find to implement shm_open() if /dev/shm is not available
920 (which it might very well be in namespaced setups).
922 * machinectl gained a new "poweroff" command to cleanly power
923 down a local OS container.
925 * The PrivateDevices= unit file setting will now also drop the
926 CAP_MKNOD capability from the capability bound set, and
927 imply DevicePolicy=closed.
929 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
930 comprehensively on all long-running systemd services where
933 * systemd-udevd will now run in a disassociated mount
934 namespace. To mount directories from udev rules, make sure to
935 pull in mount units via SYSTEMD_WANTS properties.
937 * The kdbus support gained support for uploading policy into
938 the kernel. sd-bus gained support for creating "monitoring"
939 connections that can eavesdrop into all bus communication
940 for debugging purposes.
942 * Timestamps may now be specified in seconds since the UNIX
943 epoch Jan 1st, 1970 by specifying "@" followed by the value
946 * Native tcpwrap support in systemd has been removed. tcpwrap
947 is old code, not really maintained anymore and has serious
948 shortcomings, and better options such as firewalls
949 exist. For setups that require tcpwrap usage, please
950 consider invoking your socket-activated service via tcpd,
951 like on traditional inetd.
953 * A new system.conf configuration option
954 DefaultTimerAccuracySec= has been added that controls the
955 default AccuracySec= setting of .timer units.
957 * Timer units gained a new WakeSystem= switch. If enabled,
958 timers configured this way will cause the system to resume
959 from system suspend (if the system supports that, which most
962 * Timer units gained a new Persistent= switch. If enabled,
963 timers configured this way will save to disk when they have
964 been last triggered. This information is then used on next
965 reboot to possible execute overdue timer events, that
966 could not take place because the system was powered off.
967 This enables simple anacron-like behaviour for timer units.
969 * systemctl's "list-timers" will now also list the time a
970 timer unit was last triggered in addition to the next time
971 it will be triggered.
973 * systemd-networkd will now assign predictable IPv4LL
974 addresses to its local interfaces.
976 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
977 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
978 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
979 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
980 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
981 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
982 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
983 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
986 -- Berlin, 2014-03-25
990 * A new unit file setting RestrictAddressFamilies= has been
991 added to restrict which socket address families unit
992 processes gain access to. This takes address family names
993 like "AF_INET" or "AF_UNIX", and is useful to minimize the
994 attack surface of services via exotic protocol stacks. This
995 is built on seccomp system call filters.
997 * Two new unit file settings RuntimeDirectory= and
998 RuntimeDirectoryMode= have been added that may be used to
999 manage a per-daemon runtime directories below /run. This is
1000 an alternative for setting up directory permissions with
1001 tmpfiles snippets, and has the advantage that the runtime
1002 directory's lifetime is bound to the daemon runtime and that
1003 the daemon starts up with an empty directory each time. This
1004 is particularly useful when writing services that drop
1005 privileges using the User= or Group= setting.
1007 * The DeviceAllow= unit setting now supports globbing for
1008 matching against device group names.
1010 * The systemd configuration file system.conf gained new
1011 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1012 DefaultMemoryAccounting= to globally turn on/off accounting
1013 for specific resources (cgroups) for all units. These
1014 settings may still be overridden individually in each unit
1017 * systemd-gpt-auto-generator is now able to discover /srv and
1018 root partitions in addition to /home and swap partitions. It
1019 also supports LUKS-encrypted partitions now. With this in
1020 place, automatic discovery of partitions to mount following
1021 the Discoverable Partitions Specification
1022 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1023 is now a lot more complete. This allows booting without
1024 /etc/fstab and without root= on the kernel command line on
1025 systems prepared appropriately.
1027 * systemd-nspawn gained a new --image= switch which allows
1028 booting up disk images and Linux installations on any block
1029 device that follow the Discoverable Partitions Specification
1030 (see above). This means that installations made with
1031 appropriately updated installers may now be started and
1032 deployed using container managers, completely
1033 unmodified. (We hope that libvirt-lxc will add support for
1034 this feature soon, too.)
1036 * systemd-nspawn gained a new --network-macvlan= setting to
1037 set up a private macvlan interface for the
1038 container. Similarly, systemd-networkd gained a new
1039 Kind=macvlan setting in .netdev files.
1041 * systemd-networkd now supports configuring local addresses
1044 * A new tool systemd-network-wait-online has been added to
1045 synchronously wait for network connectivity using
1048 * The sd-bus.h bus API gained a new sd_bus_track object for
1049 tracking the life-cycle of bus peers. Note that sd-bus.h is
1050 still not a public API though (unless you specify
1051 --enable-kdbus on the configure command line, which however
1052 voids your warranty and you get no API stability guarantee).
1054 * The $XDG_RUNTIME_DIR runtime directories for each user are
1055 now individual tmpfs instances, which has the benefit of
1056 introducing separate pools for each user, with individual
1057 size limits, and thus making sure that unprivileged clients
1058 can no longer negatively impact the system or other users by
1059 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1060 RuntimeDirectorySize= has been introduced that allows
1061 controlling the default size limit for all users. It
1062 defaults to 10% of the available physical memory. This is no
1063 replacement for quotas on tmpfs though (which the kernel
1064 still does not support), as /dev/shm and /tmp are still
1065 shared resources used by both the system and unprivileged
1068 * logind will now automatically turn off automatic suspending
1069 on laptop lid close when more than one display is
1070 connected. This was previously expected to be implemented
1071 individually in desktop environments (such as GNOME),
1072 however has been added to logind now, in order to fix a
1073 boot-time race where a desktop environment might not have
1074 been started yet and thus not been able to take an inhibitor
1075 lock at the time where logind already suspends the system
1076 due to a closed lid.
1078 * logind will now wait at least 30s after each system
1079 suspend/resume cycle, and 3min after system boot before
1080 suspending the system due to a closed laptop lid. This
1081 should give USB docking stations and similar enough time to
1082 be probed and configured after system resume and boot in
1083 order to then act as suspend blocker.
1085 * systemd-run gained a new --property= setting which allows
1086 initialization of resource control properties (and others)
1087 for the created scope or service unit. Example: "systemd-run
1088 --property=BlockIOWeight=10 updatedb" may be used to run
1089 updatedb at a low block IO scheduling weight.
1091 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1092 now also work in --scope mode.
1094 * When systemd is compiled with kdbus support, basic support
1095 for enforced policies is now in place. (Note that enabling
1096 kdbus still voids your warranty and no API compatibility
1099 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1100 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1101 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1102 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1103 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1104 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1105 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1106 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1107 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1108 Zbigniew Jędrzejewski-Szmek
1110 -- Berlin, 2014-03-12
1114 * systemd will now relabel /dev after loading the SMACK policy
1115 according to SMACK rules.
1117 * A new unit file option AppArmorProfile= has been added to
1118 set the AppArmor profile for the processes of a unit.
1120 * A new condition check ConditionArchitecture= has been added
1121 to conditionalize units based on the system architecture, as
1122 reported by uname()'s "machine" field.
1124 * systemd-networkd now supports matching on the system
1125 virtualization, architecture, kernel command line, host name
1128 * logind is now a lot more aggressive when suspending the
1129 machine due to a closed laptop lid. Instead of acting only
1130 on the lid close action, it will continuously watch the lid
1131 status and act on it. This is useful for laptops where the
1132 power button is on the outside of the chassis so that it can
1133 be reached without opening the lid (such as the Lenovo
1134 Yoga). On those machines, logind will now immediately
1135 re-suspend the machine if the power button has been
1136 accidentally pressed while the laptop was suspended and in a
1137 backpack or similar.
1139 * logind will now watch SW_DOCK switches and inhibit reaction
1140 to the lid switch if it is pressed. This means that logind
1141 will not suspend the machine anymore if the lid is closed
1142 and the system is docked, if the laptop supports SW_DOCK
1143 notifications via the input layer. Note that ACPI docking
1144 stations do not generate this currently. Also note that this
1145 logic is usually not fully sufficient and Desktop
1146 Environments should take a lid switch inhibitor lock when an
1147 external display is connected, as systemd will not watch
1150 * nspawn will now make use of the devices cgroup controller by
1151 default, and only permit creation of and access to the usual
1152 API device nodes like /dev/null or /dev/random, as well as
1153 access to (but not creation of) the pty devices.
1155 * We will now ship a default .network file for
1156 systemd-networkd that automatically configures DHCP for
1157 network interfaces created by nspawn's --network-veth or
1158 --network-bridge= switches.
1160 * systemd will now understand the usual M, K, G, T suffixes
1161 according to SI conventions (i.e. to the base 1000) when
1162 referring to throughput and hardware metrics. It will stay
1163 with IEC conventions (i.e. to the base 1024) for software
1164 metrics, according to what is customary according to
1165 Wikipedia. We explicitly document which base applies for
1166 each configuration option.
1168 * The DeviceAllow= setting in unit files now supports a syntax
1169 to whitelist an entire group of devices node majors at once,
1170 based on the /proc/devices listing. For example, with the
1171 string "char-pts", it is now possible to whitelist all
1172 current and future pseudo-TTYs at once.
1174 * sd-event learned a new "post" event source. Event sources of
1175 this type are triggered by the dispatching of any event
1176 source of a type that is not "post". This is useful for
1177 implementing clean-up and check event sources that are
1178 triggered by other work being done in the program.
1180 * systemd-networkd is no longer statically enabled, but uses
1181 the usual [Install] sections so that it can be
1182 enabled/disabled using systemctl. It still is enabled by
1185 * When creating a veth interface pair with systemd-nspawn, the
1186 host side will now be prefixed with "vb-" if
1187 --network-bridge= is used, and with "ve-" if --network-veth
1188 is used. This way, it is easy to distinguish these cases on
1189 the host, for example to apply different configuration to
1190 them with systemd-networkd.
1192 * The compatibility libraries for libsystemd-journal.so,
1193 libsystem-id128.so, libsystemd-login.so and
1194 libsystemd-daemon.so do not make use of IFUNC
1195 anymore. Instead, we now build libsystemd.so multiple times
1196 under these alternative names. This means that the footprint
1197 is drastically increased, but given that these are
1198 transitional compatibility libraries, this should not matter
1199 much. This change has been made necessary to support the ARM
1200 platform for these compatibility libraries, as the ARM
1201 toolchain is not really at the same level as the toolchain
1202 for other architectures like x86 and does not support
1203 IFUNC. Please make sure to use --enable-compat-libs only
1204 during a transitional period!
1206 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1207 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1208 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1209 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1210 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1211 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1212 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1213 Zbigniew Jędrzejewski-Szmek
1215 -- Berlin, 2014-02-24
1219 * A new component "systemd-networkd" has been added that can
1220 be used to configure local network interfaces statically or
1221 via DHCP. It is capable of bringing up bridges, VLANs, and
1222 bonding. Currently, no hook-ups for interactive network
1223 configuration are provided. Use this for your initrd,
1224 container, embedded, or server setup if you need a simple,
1225 yet powerful, network configuration solution. This
1226 configuration subsystem is quite nifty, as it allows wildcard
1227 hotplug matching in interfaces. For example, with a single
1228 configuration snippet, you can configure that all Ethernet
1229 interfaces showing up are automatically added to a bridge,
1230 or similar. It supports link-sensing and more.
1232 * A new tool "systemd-socket-proxyd" has been added which can
1233 act as a bidirectional proxy for TCP sockets. This is
1234 useful for adding socket activation support to services that
1235 do not actually support socket activation, including virtual
1236 machines and the like.
1238 * Add a new tool to save/restore rfkill state on
1241 * Save/restore state of keyboard backlights in addition to
1242 display backlights on shutdown/boot.
1244 * udev learned a new SECLABEL{} construct to label device
1245 nodes with a specific security label when they appear. For
1246 now, only SECLABEL{selinux} is supported, but the syntax is
1247 prepared for additional security frameworks.
1249 * udev gained a new scheme to configure link-level attributes
1250 from files in /etc/systemd/network/*.link. These files can
1251 match against MAC address, device path, driver name and type,
1252 and will apply attributes like the naming policy, link speed,
1253 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1254 address assignment policy (randomized, ...).
1256 * The configuration of network interface naming rules for
1257 "permanent interface names" has changed: a new NamePolicy=
1258 setting in the [Link] section of .link files determines the
1259 priority of possible naming schemes (onboard, slot, mac,
1260 path). The default value of this setting is determined by
1261 /usr/lib/net/links/99-default.link. Old
1262 80-net-name-slot.rules udev configuration file has been
1263 removed, so local configuration overriding this file should
1264 be adapated to override 99-default.link instead.
1266 * When the User= switch is used in a unit file, also
1267 initialize $SHELL= based on the user database entry.
1269 * systemd no longer depends on libdbus. All communication is
1270 now done with sd-bus, systemd's low-level bus library
1273 * kdbus support has been added to PID 1 itself. When kdbus is
1274 enabled, this causes PID 1 to set up the system bus and
1275 enable support for a new ".busname" unit type that
1276 encapsulates bus name activation on kdbus. It works a little
1277 bit like ".socket" units, except for bus names. A new
1278 generator has been added that converts classic dbus1 service
1279 activation files automatically into native systemd .busname
1282 * sd-bus: add a light-weight vtable implementation that allows
1283 defining objects on the bus with a simple static const
1284 vtable array of its methods, signals and properties.
1286 * systemd will not generate or install static dbus
1287 introspection data anymore to /usr/share/dbus-1/interfaces,
1288 as the precise format of these files is unclear, and
1289 nothing makes use of it.
1291 * A proxy daemon is now provided to proxy clients connecting
1292 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1293 compatibility with classic D-Bus.
1295 * A bus driver implementation has been added that supports the
1296 classic D-Bus bus driver calls on kdbus, also for
1297 compatibility purposes.
1299 * A new API "sd-event.h" has been added that implements a
1300 minimal event loop API built around epoll. It provides a
1301 couple of features that direct epoll usage is lacking:
1302 prioritization of events, scales to large numbers of timer
1303 events, per-event timer slack (accuracy), system-wide
1304 coalescing of timer events, exit handlers, watchdog
1305 supervision support using systemd's sd_notify() API, child
1308 * A new API "sd-rntl.h" has been added that provides an API
1309 around the route netlink interface of the kernel, similar in
1310 style to "sd-bus.h".
1312 * A new API "sd-dhcp-client.h" has been added that provides a
1313 small DHCPv4 client-side implementation. This is used by
1316 * There is a new kernel command line option
1317 "systemd.restore_state=0|1". When set to "0", none of the
1318 systemd tools will restore saved runtime state to hardware
1319 devices. More specifically, the rfkill and backlight states
1322 * The FsckPassNo= compatibility option in mount/service units
1323 has been removed. The fstab generator will now add the
1324 necessary dependencies automatically, and does not require
1325 PID1's support for that anymore.
1327 * journalctl gained a new switch, --list-boots, that lists
1328 recent boots with their times and boot IDs.
1330 * The various tools like systemctl, loginctl, timedatectl,
1331 busctl, systemd-run, ... have gained a new switch "-M" to
1332 connect to a specific, local OS container (as direct
1333 connection, without requiring SSH). This works on any
1334 container that is registered with machined, such as those
1335 created by libvirt-lxc or nspawn.
1337 * systemd-run and systemd-analyze also gained support for "-H"
1338 to connect to remote hosts via SSH. This is particularly
1339 useful for systemd-run because it enables queuing of jobs
1340 onto remote systems.
1342 * machinectl gained a new command "login" to open a getty
1343 login in any local container. This works with any container
1344 that is registered with machined (such as those created by
1345 libvirt-lxc or nspawn), and which runs systemd inside.
1347 * machinectl gained a new "reboot" command that may be used to
1348 trigger a reboot on a specific container that is registered
1349 with machined. This works on any container that runs an init
1350 system of some kind.
1352 * systemctl gained a new "list-timers" command to print a nice
1353 listing of installed timer units with the times they elapse
1356 * Alternative reboot() parameters may now be specified on the
1357 "systemctl reboot" command line and are passed to the
1358 reboot() system call.
1360 * systemctl gained a new --job-mode= switch to configure the
1361 mode to queue a job with. This is a more generic version of
1362 --fail, --irreversible, and --ignore-dependencies, which are
1363 still available but not advertised anymore.
1365 * /etc/systemd/system.conf gained new settings to configure
1366 various default timeouts of units, as well as the default
1367 start limit interval and burst. These may still be overridden
1370 * PID1 will now export on the bus profile data of the security
1371 policy upload process (such as the SELinux policy upload to
1374 * journald: when forwarding logs to the console, include
1375 timestamps (following the setting in
1376 /sys/module/printk/parameters/time).
1378 * OnCalendar= in timer units now understands the special
1379 strings "yearly" and "annually". (Both are equivalent)
1381 * The accuracy of timer units is now configurable with the new
1382 AccuracySec= setting. It defaults to 1min.
1384 * A new dependency type JoinsNamespaceOf= has been added that
1385 allows running two services within the same /tmp and network
1386 namespace, if PrivateNetwork= or PrivateTmp= are used.
1388 * A new command "cat" has been added to systemctl. It outputs
1389 the original unit file of a unit, and concatenates the
1390 contents of additional "drop-in" unit file snippets, so that
1391 the full configuration is shown.
1393 * systemctl now supports globbing on the various "list-xyz"
1394 commands, like "list-units" or "list-sockets", as well as on
1395 those commands which take multiple unit names.
1397 * journalctl's --unit= switch gained support for globbing.
1399 * All systemd daemons now make use of the watchdog logic so
1400 that systemd automatically notices when they hang.
1402 * If the $container_ttys environment variable is set,
1403 getty-generator will automatically spawn a getty for each
1404 listed tty. This is useful for container managers to request
1405 login gettys to be spawned on as many ttys as needed.
1407 * %h, %s, %U specifier support is not available anymore when
1408 used in unit files for PID 1. This is because NSS calls are
1409 not safe from PID 1. They stay available for --user
1410 instances of systemd, and as special case for the root user.
1412 * loginctl gained a new "--no-legend" switch to turn off output
1415 * The "sd-login.h" API gained three new calls:
1416 sd_session_is_remote(), sd_session_get_remote_user(),
1417 sd_session_get_remote_host() to query information about
1420 * The udev hardware database now also carries vendor/product
1421 information of SDIO devices.
1423 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1424 determine whether watchdog notifications are requested by
1427 * Socket-activated per-connection services now include a
1428 short description of the connection parameters in the
1431 * tmpfiles gained a new "--boot" option. When this is not used,
1432 only lines where the command character is not suffixed with
1433 "!" are executed. When this option is specified, those
1434 options are executed too. This partitions tmpfiles
1435 directives into those that can be safely executed at any
1436 time, and those which should be run only at boot (for
1437 example, a line that creates /run/nologin).
1439 * A new API "sd-resolve.h" has been added which provides a simple
1440 asynchronous wrapper around glibc NSS host name resolution
1441 calls, such as getaddrinfo(). In contrast to glibc's
1442 getaddrinfo_a(), it does not use signals. In contrast to most
1443 other asynchronous name resolution libraries, this one does
1444 not reimplement DNS, but reuses NSS, so that alternate
1445 host name resolution systems continue to work, such as mDNS,
1446 LDAP, etc. This API is based on libasyncns, but it has been
1447 cleaned up for inclusion in systemd.
1449 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1450 "sd-daemon.h" are no longer found in individual libraries
1451 libsystemd-journal.so, libsystemd-login.so,
1452 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1453 merged them into a single library, libsystemd.so, which
1454 provides all symbols. The reason for this is cyclic
1455 dependencies, as these libraries tend to use each other's
1456 symbols. So far, we have managed to workaround that by linking
1457 a copy of a good part of our code into each of these
1458 libraries again and again, which, however, makes certain
1459 things hard to do, like sharing static variables. Also, it
1460 substantially increases footprint. With this change, there
1461 is only one library for the basic APIs systemd
1462 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1463 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1464 library as well, however are subject to the --enable-kdbus
1465 switch (see below). Note that "sd-dhcp-client.h" is not part
1466 of this library (this is because it only consumes, never
1467 provides, services of/to other APIs). To make the transition
1468 easy from the separate libraries to the unified one, we
1469 provide the --enable-compat-libs compile-time switch which
1470 will generate stub libraries that are compatible with the
1471 old ones but redirect all calls to the new one.
1473 * All of the kdbus logic and the new APIs "sd-bus.h",
1474 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1475 and "sd-utf8.h" are compile-time optional via the
1476 "--enable-kdbus" switch, and they are not compiled in by
1477 default. To make use of kdbus, you have to explicitly enable
1478 the switch. Note however, that neither the kernel nor the
1479 userspace API for all of this is considered stable yet. We
1480 want to maintain the freedom to still change the APIs for
1481 now. By specifying this build-time switch, you acknowledge
1482 that you are aware of the instability of the current
1485 * Also, note that while kdbus is pretty much complete,
1486 it lacks one thing: proper policy support. This means you
1487 can build a fully working system with all features; however,
1488 it will be highly insecure. Policy support will be added in
1489 one of the next releases, at the same time that we will
1490 declare the APIs stable.
1492 * When the kernel command-line argument "kdbus" is specified,
1493 systemd will automatically load the kdbus.ko kernel module. At
1494 this stage of development, it is only useful for testing kdbus
1495 and should not be used in production. Note: if "--enable-kdbus"
1496 is specified, and the kdbus.ko kernel module is available, and
1497 "kdbus" is added to the kernel command line, the entire system
1498 runs with kdbus instead of dbus-daemon, with the above mentioned
1499 problem of missing the system policy enforcement. Also a future
1500 version of kdbus.ko or a newer systemd will not be compatible with
1501 each other, and will unlikely be able to boot the machine if only
1502 one of them is updated.
1504 * systemctl gained a new "import-environment" command which
1505 uploads the caller's environment (or parts thereof) into the
1506 service manager so that it is inherited by services started
1507 by the manager. This is useful to upload variables like
1508 $DISPLAY into the user service manager.
1510 * A new PrivateDevices= switch has been added to service units
1511 which allows running a service with a namespaced /dev
1512 directory that does not contain any device nodes for
1513 physical devices. More specifically, it only includes devices
1514 such as /dev/null, /dev/urandom, and /dev/zero which are API
1517 * logind has been extended to support behaviour like VT
1518 switching on seats that do not support a VT. This makes
1519 multi-session available on seats that are not the first seat
1520 (seat0), and on systems where kernel support for VTs has
1521 been disabled at compile-time.
1523 * If a process holds a delay lock for system sleep or shutdown
1524 and fails to release it in time, we will now log its
1525 identity. This makes it easier to identify processes that
1526 cause slow suspends or power-offs.
1528 * When parsing /etc/crypttab, support for a new key-slot=
1529 option as supported by Debian is added. It allows indicating
1530 which LUKS slot to use on disk, speeding up key loading.
1532 * The sd_journald_sendv() API call has been checked and
1533 officially declared to be async-signal-safe so that it may
1534 be invoked from signal handlers for logging purposes.
1536 * Boot-time status output is now enabled automatically after a
1537 short timeout if boot does not progress, in order to give
1538 the user an indication what she or he is waiting for.
1540 * The boot-time output has been improved to show how much time
1541 remains until jobs expire.
1543 * The KillMode= switch in service units gained a new possible
1544 value "mixed". If set, and the unit is shut down, then the
1545 initial SIGTERM signal is sent only to the main daemon
1546 process, while the following SIGKILL signal is sent to
1547 all remaining processes of the service.
1549 * When a scope unit is registered, a new property "Controller"
1550 may be set. If set to a valid bus name, systemd will send a
1551 RequestStop() signal to this name when it would like to shut
1552 down the scope. This may be used to hook manager logic into
1553 the shutdown logic of scope units. Also, scope units may now
1554 be put in a special "abandoned" state, in which case the
1555 manager process which created them takes no further
1556 responsibilities for it.
1558 * When reading unit files, systemd will now verify
1559 the access mode of these files, and warn about certain
1560 suspicious combinations. This has been added to make it
1561 easier to track down packaging bugs where unit files are
1562 marked executable or world-writable.
1564 * systemd-nspawn gained a new "--setenv=" switch to set
1565 container-wide environment variables. The similar option in
1566 systemd-activate was renamed from "--environment=" to
1567 "--setenv=" for consistency.
1569 * systemd-nspawn has been updated to create a new kdbus domain
1570 for each container that is invoked, thus allowing each
1571 container to have its own set of system and user buses,
1572 independent of the host.
1574 * systemd-nspawn gained a new --drop-capability= switch to run
1575 the container with less capabilities than the default. Both
1576 --drop-capability= and --capability= now take the special
1577 string "all" for dropping or keeping all capabilities.
1579 * systemd-nspawn gained new switches for executing containers
1580 with specific SELinux labels set.
1582 * systemd-nspawn gained a new --quiet switch to not generate
1583 any additional output but the container's own console
1586 * systemd-nspawn gained a new --share-system switch to run a
1587 container without PID namespacing enabled.
1589 * systemd-nspawn gained a new --register= switch to control
1590 whether the container is registered with systemd-machined or
1591 not. This is useful for containers that do not run full
1592 OS images, but only specific apps.
1594 * systemd-nspawn gained a new --keep-unit which may be used
1595 when invoked as the only program from a service unit, and
1596 results in registration of the unit service itself in
1597 systemd-machined, instead of a newly opened scope unit.
1599 * systemd-nspawn gained a new --network-interface= switch for
1600 moving arbitrary interfaces to the container. The new
1601 --network-veth switch creates a virtual Ethernet connection
1602 between host and container. The new --network-bridge=
1603 switch then allows assigning the host side of this virtual
1604 Ethernet connection to a bridge device.
1606 * systemd-nspawn gained a new --personality= switch for
1607 setting the kernel personality for the container. This is
1608 useful when running a 32-bit container on a 64-bit host. A
1609 similar option Personality= is now also available for service
1612 * logind will now also track a "Desktop" identifier for each
1613 session which encodes the desktop environment of it. This is
1614 useful for desktop environments that want to identify
1615 multiple running sessions of itself easily.
1617 * A new SELinuxContext= setting for service units has been
1618 added that allows setting a specific SELinux execution
1619 context for a service.
1621 * Most systemd client tools will now honour $SYSTEMD_LESS for
1622 settings of the "less" pager. By default, these tools will
1623 override $LESS to allow certain operations to work, such as
1624 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1625 influence this logic.
1627 * systemd's "seccomp" hook-up has been changed to make use of
1628 the libseccomp library instead of using its own
1629 implementation. This has benefits for portability among
1632 * For usage together with SystemCallFilter=, a new
1633 SystemCallErrorNumber= setting has been introduced that
1634 allows configuration of a system error number to be returned
1635 on filtered system calls, instead of immediately killing the
1636 process. Also, SystemCallArchitectures= has been added to
1637 limit access to system calls of a particular architecture
1638 (in order to turn off support for unused secondary
1639 architectures). There is also a global
1640 SystemCallArchitectures= setting in system.conf now to turn
1641 off support for non-native system calls system-wide.
1643 * systemd requires a kernel with a working name_to_handle_at(),
1644 please see the kernel config requirements in the README file.
1646 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1647 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1648 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1649 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1650 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1651 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1652 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1653 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1654 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1655 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1656 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1657 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1658 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1659 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1660 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1661 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1662 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1663 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1664 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1665 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1666 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1667 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1668 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1669 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1671 -- Berlin, 2014-02-20
1675 * logind has gained support for facilitating privileged input
1676 and drm device access for unprivileged clients. This work is
1677 useful to allow Wayland display servers (and similar
1678 programs, such as kmscon) to run under the user's ID and
1679 access input and drm devices which are normally
1680 protected. When this is used (and the kernel is new enough)
1681 logind will "mute" IO on the file descriptors passed to
1682 Wayland as long as it is in the background and "unmute" it
1683 if it returns into the foreground. This allows secure
1684 session switching without allowing background sessions to
1685 eavesdrop on input and display data. This also introduces
1686 session switching support if VT support is turned off in the
1687 kernel, and on seats that are not seat0.
1689 * A new kernel command line option luks.options= is understood
1690 now which allows specifying LUKS options for usage for LUKS
1691 encrypted partitions specified with luks.uuid=.
1693 * tmpfiles.d(5) snippets may now use specifier expansion in
1694 path names. More specifically %m, %b, %H, %v, are now
1695 replaced by the local machine id, boot id, hostname, and
1696 kernel version number.
1698 * A new tmpfiles.d(5) command "m" has been introduced which
1699 may be used to change the owner/group/access mode of a file
1700 or directory if it exists, but do nothing if it does not.
1702 * This release removes high-level support for the
1703 MemorySoftLimit= cgroup setting. The underlying kernel
1704 cgroup attribute memory.soft_limit= is currently badly
1705 designed and likely to be removed from the kernel API in its
1706 current form, hence we should not expose it for now.
1708 * The memory.use_hierarchy cgroup attribute is now enabled for
1709 all cgroups systemd creates in the memory cgroup
1710 hierarchy. This option is likely to be come the built-in
1711 default in the kernel anyway, and the non-hierarchial mode
1712 never made much sense in the intrinsically hierarchial
1715 * A new field _SYSTEMD_SLICE= is logged along with all journal
1716 messages containing the slice a message was generated
1717 from. This is useful to allow easy per-customer filtering of
1718 logs among other things.
1720 * systemd-journald will no longer adjust the group of journal
1721 files it creates to the "systemd-journal" group. Instead we
1722 rely on the journal directory to be owned by the
1723 "systemd-journal" group, and its setgid bit set, so that the
1724 kernel file system layer will automatically enforce that
1725 journal files inherit this group assignment. The reason for
1726 this change is that we cannot allow NSS look-ups from
1727 journald which would be necessary to resolve
1728 "systemd-journal" to a numeric GID, because this might
1729 create deadlocks if NSS involves synchronous queries to
1730 other daemons (such as nscd, or sssd) which in turn are
1731 logging clients of journald and might block on it, which
1732 would then dead lock. A tmpfiles.d(5) snippet included in
1733 systemd will make sure the setgid bit and group are
1734 properly set on the journal directory if it exists on every
1735 boot. However, we recommend adjusting it manually after
1736 upgrades too (or from RPM scriptlets), so that the change is
1737 not delayed until next reboot.
1739 * Backlight and random seed files in /var/lib/ have moved into
1740 the /var/lib/systemd/ directory, in order to centralize all
1741 systemd generated files in one directory.
1743 * Boot time performance measurements (as displayed by
1744 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1745 performance information if that's available to determine how
1746 much time BIOS and boot loader initialization required. With
1747 a sufficiently new BIOS you hence no longer need to boot
1748 with Gummiboot to get access to such information.
1750 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1751 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1752 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1753 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1754 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1755 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1756 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1758 -- Berlin, 2013-10-02
1762 * The Restart= option for services now understands a new
1763 on-watchdog setting, which will restart the service
1764 automatically if the service stops sending out watchdog keep
1765 alive messages (as configured with WatchdogSec=).
1767 * The getty generator (which is responsible for bringing up a
1768 getty on configured serial consoles) will no longer only
1769 start a getty on the primary kernel console but on all
1770 others, too. This makes the order in which console= is
1771 specified on the kernel command line less important.
1773 * libsystemd-logind gained a new sd_session_get_vt() call to
1774 retrieve the VT number of a session.
1776 * If the option "tries=0" is set for an entry of /etc/crypttab
1777 its passphrase is queried indefinitely instead of any
1778 maximum number of tries.
1780 * If a service with a configure PID file terminates its PID
1781 file will now be removed automatically if it still exists
1782 afterwards. This should put an end to stale PID files.
1784 * systemd-run will now also take relative binary path names
1785 for execution and no longer insists on absolute paths.
1787 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1788 paths that are optionally prefixed with "-" to indicate that
1789 it should not be considered a failure if they do not exist.
1791 * journalctl -o (and similar commands) now understands a new
1792 output mode "short-precise", it is similar to "short" but
1793 shows timestamps with usec accuracy.
1795 * The option "discard" (as known from Debian) is now
1796 synonymous to "allow-discards" in /etc/crypttab. In fact,
1797 "discard" is preferred now (since it is easier to remember
1800 * Some licensing clean-ups were made, so that more code is now
1801 LGPL-2.1 licensed than before.
1803 * A minimal tool to save/restore the display backlight
1804 brightness across reboots has been added. It will store the
1805 backlight setting as late as possible at shutdown, and
1806 restore it as early as possible during reboot.
1808 * A logic to automatically discover and enable home and swap
1809 partitions on GPT disks has been added. With this in place
1810 /etc/fstab becomes optional for many setups as systemd can
1811 discover certain partitions located on the root disk
1812 automatically. Home partitions are recognized under their
1813 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1814 partitions are recognized under their GPT type ID
1815 0657fd6da4ab43c484e50933c84b4f4f.
1817 * systemd will no longer pass any environment from the kernel
1818 or initrd to system services. If you want to set an
1819 environment for all services, do so via the kernel command
1820 line systemd.setenv= assignment.
1822 * The systemd-sysctl tool no longer natively reads the file
1823 /etc/sysctl.conf. If desired, the file should be symlinked
1824 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1825 legacy support by a symlink rather than built-in code, it
1826 also makes the otherwise hidden order of application of the
1827 different files visible. (Note that this partly reverts to a
1828 pre-198 application order of sysctl knobs!)
1830 * The "systemctl set-log-level" and "systemctl dump" commands
1831 have been moved to systemd-analyze.
1833 * systemd-run learned the new --remain-after-exit switch,
1834 which causes the scope unit not to be cleaned up
1835 automatically after the process terminated.
1837 * tmpfiles learned a new --exclude-prefix= switch to exclude
1838 certain paths from operation.
1840 * journald will now automatically flush all messages to disk
1841 as soon as a message of the log priorities CRIT, ALERT or
1844 Contributions from: Andrew Cook, Brandon Philips, Christian
1845 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1846 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1847 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1848 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1849 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1850 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1851 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1852 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1853 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1854 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1855 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1856 William Giokas, Zbigniew Jędrzejewski-Szmek
1858 -- Berlin, 2013-09-13
1862 * The documentation has been updated to cover the various new
1863 concepts introduced with 205.
1865 * Unit files now understand the new %v specifier which
1866 resolves to the kernel version string as returned by "uname
1869 * systemctl now supports filtering the unit list output by
1870 load state, active state and sub state, using the new
1873 * "systemctl status" will now show the results of the
1874 condition checks (like ConditionPathExists= and similar) of
1875 the last start attempts of the unit. They are also logged to
1878 * "journalctl -b" may now be used to look for boot output of a
1879 specific boot. Try "journalctl -b -1" for the previous boot,
1880 but the syntax is substantially more powerful.
1882 * "journalctl --show-cursor" has been added which prints the
1883 cursor string the last shown log line. This may then be used
1884 with the new "journalctl --after-cursor=" switch to continue
1885 browsing logs from that point on.
1887 * "journalctl --force" may now be used to force regeneration
1890 * Creation of "dead" device nodes has been moved from udev
1891 into kmod and tmpfiles. Previously, udev would read the kmod
1892 databases to pre-generate dead device nodes based on meta
1893 information contained in kernel modules, so that these would
1894 be auto-loaded on access rather then at boot. As this
1895 does not really have much to do with the exposing actual
1896 kernel devices to userspace this has always been slightly
1897 alien in the udev codebase. Following the new scheme kmod
1898 will now generate a runtime snippet for tmpfiles from the
1899 module meta information and it now is tmpfiles' job to the
1900 create the nodes. This also allows overriding access and
1901 other parameters for the nodes using the usual tmpfiles
1902 facilities. As side effect this allows us to remove the
1903 CAP_SYS_MKNOD capability bit from udevd entirely.
1905 * logind's device ACLs may now be applied to these "dead"
1906 devices nodes too, thus finally allowing managed access to
1907 devices such as /dev/snd/sequencer whithout loading the
1908 backing module right-away.
1910 * A new RPM macro has been added that may be used to apply
1911 tmpfiles configuration during package installation.
1913 * systemd-detect-virt and ConditionVirtualization= now can
1914 detect User-Mode-Linux machines (UML).
1916 * journald will now implicitly log the effective capabilities
1917 set of processes in the message metadata.
1919 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1921 * The initrd interface has been simplified (more specifically,
1922 support for passing performance data via environment
1923 variables and fsck results via files in /run has been
1924 removed). These features were non-essential, and are
1925 nowadays available in a much nicer way by having systemd in
1926 the initrd serialize its state and have the hosts systemd
1927 deserialize it again.
1929 * The udev "keymap" data files and tools to apply keyboard
1930 specific mappings of scan to key codes, and force-release
1931 scan code lists have been entirely replaced by a udev
1932 "keyboard" builtin and a hwdb data file.
1934 * systemd will now honour the kernel's "quiet" command line
1935 argument also during late shutdown, resulting in a
1936 completely silent shutdown when used.
1938 * There's now an option to control the SO_REUSEPORT socket
1939 option in .socket units.
1941 * Instance units will now automatically get a per-template
1942 subslice of system.slice unless something else is explicitly
1943 configured. For example, instances of sshd@.service will now
1944 implicitly be placed in system-sshd.slice rather than
1945 system.slice as before.
1947 * Test coverage support may now be enabled at build time.
1949 Contributions from: Dave Reisner, Frederic Crozat, Harald
1950 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1951 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1952 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1953 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1954 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1955 Giokas, Zbigniew Jędrzejewski-Szmek
1957 -- Berlin, 2013-07-23
1961 * Two new unit types have been introduced:
1963 Scope units are very similar to service units, however, are
1964 created out of pre-existing processes -- instead of PID 1
1965 forking off the processes. By using scope units it is
1966 possible for system services and applications to group their
1967 own child processes (worker processes) in a powerful way
1968 which then maybe used to organize them, or kill them
1969 together, or apply resource limits on them.
1971 Slice units may be used to partition system resources in an
1972 hierarchial fashion and then assign other units to them. By
1973 default there are now three slices: system.slice (for all
1974 system services), user.slice (for all user sessions),
1975 machine.slice (for VMs and containers).
1977 Slices and scopes have been introduced primarily in
1978 context of the work to move cgroup handling to a
1979 single-writer scheme, where only PID 1
1980 creates/removes/manages cgroups.
1982 * There's a new concept of "transient" units. In contrast to
1983 normal units these units are created via an API at runtime,
1984 not from configuration from disk. More specifically this
1985 means it is now possible to run arbitrary programs as
1986 independent services, with all execution parameters passed
1987 in via bus APIs rather than read from disk. Transient units
1988 make systemd substantially more dynamic then it ever was,
1989 and useful as a general batch manager.
1991 * logind has been updated to make use of scope and slice units
1992 for managing user sessions. As a user logs in he will get
1993 his own private slice unit, to which all sessions are added
1994 as scope units. We also added support for automatically
1995 adding an instance of user@.service for the user into the
1996 slice. Effectively logind will no longer create cgroup
1997 hierarchies on its own now, it will defer entirely to PID 1
1998 for this by means of scope, service and slice units. Since
1999 user sessions this way become entities managed by PID 1
2000 the output of "systemctl" is now a lot more comprehensive.
2002 * A new mini-daemon "systemd-machined" has been added which
2003 may be used by virtualization managers to register local
2004 VMs/containers. nspawn has been updated accordingly, and
2005 libvirt will be updated shortly. machined will collect a bit
2006 of meta information about the VMs/containers, and assign
2007 them their own scope unit (see above). The collected
2008 meta-data is then made available via the "machinectl" tool,
2009 and exposed in "ps" and similar tools. machined/machinectl
2010 is compile-time optional.
2012 * As discussed earlier, the low-level cgroup configuration
2013 options ControlGroup=, ControlGroupModify=,
2014 ControlGroupPersistent=, ControlGroupAttribute= have been
2015 removed. Please use high-level attribute settings instead as
2016 well as slice units.
2018 * A new bus call SetUnitProperties() has been added to alter
2019 various runtime parameters of a unit. This is primarily
2020 useful to alter cgroup parameters dynamically in a nice way,
2021 but will be extended later on to make more properties
2022 modifiable at runtime. systemctl gained a new set-properties
2023 command that wraps this call.
2025 * A new tool "systemd-run" has been added which can be used to
2026 run arbitrary command lines as transient services or scopes,
2027 while configuring a number of settings via the command
2028 line. This tool is currently very basic, however already
2029 very useful. We plan to extend this tool to even allow
2030 queuing of execution jobs with time triggers from the
2031 command line, similar in fashion to "at".
2033 * nspawn will now inform the user explicitly that kernels with
2034 audit enabled break containers, and suggest the user to turn
2037 * Support for detecting the IMA and AppArmor security
2038 frameworks with ConditionSecurity= has been added.
2040 * journalctl gained a new "-k" switch for showing only kernel
2041 messages, mimicking dmesg output; in addition to "--user"
2042 and "--system" switches for showing only user's own logs
2045 * systemd-delta can now show information about drop-in
2046 snippets extending unit files.
2048 * libsystemd-bus has been substantially updated but is still
2049 not available as public API.
2051 * systemd will now look for the "debug" argument on the kernel
2052 command line and enable debug logging, similar to what
2053 "systemd.log_level=debug" already did before.
2055 * "systemctl set-default", "systemctl get-default" has been
2056 added to configure the default.target symlink, which
2057 controls what to boot into by default.
2059 * "systemctl set-log-level" has been added as a convenient
2060 way to raise and lower systemd logging threshold.
2062 * "systemd-analyze plot" will now show the time the various
2063 generators needed for execution, as well as information
2064 about the unit file loading.
2066 * libsystemd-journal gained a new sd_journal_open_files() call
2067 for opening specific journal files. journactl also gained a
2068 new switch to expose this new functionality. Previously we
2069 only supported opening all files from a directory, or all
2070 files from the system, as opening individual files only is
2071 racy due to journal file rotation.
2073 * systemd gained the new DefaultEnvironment= setting in
2074 /etc/systemd/system.conf to set environment variables for
2077 * If a privileged process logs a journal message with the
2078 OBJECT_PID= field set, then journald will automatically
2079 augment this with additional OBJECT_UID=, OBJECT_GID=,
2080 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2081 system services want to log events about specific client
2082 processes. journactl/systemctl has been updated to make use
2083 of this information if all log messages regarding a specific
2086 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2087 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2088 Reisner, David Coppa, David King, David Strauss, Eelco
2089 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2090 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2091 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2092 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2093 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2094 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2095 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2096 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2097 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2098 Łukasz Stelmach, 장동준
2102 * The Python bindings gained some minimal support for the APIs
2103 exposed by libsystemd-logind.
2105 * ConditionSecurity= gained support for detecting SMACK. Since
2106 this condition already supports SELinux and AppArmor we only
2107 miss IMA for this. Patches welcome!
2109 Contributions from: Karol Lewandowski, Lennart Poettering,
2110 Zbigniew Jędrzejewski-Szmek
2114 * systemd-nspawn will now create /etc/resolv.conf if
2115 necessary, before bind-mounting the host's file onto it.
2117 * systemd-nspawn will now store meta information about a
2118 container on the container's cgroup as extended attribute
2119 fields, including the root directory.
2121 * The cgroup hierarchy has been reworked in many ways. All
2122 objects any of the components systemd creates in the cgroup
2123 tree are now suffixed. More specifically, user sessions are
2124 now placed in cgroups suffixed with ".session", users in
2125 cgroups suffixed with ".user", and nspawn containers in
2126 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2127 names are now escaped in a simple scheme to avoid collision
2128 of userspace object names with kernel filenames. This work
2129 is preparation for making these objects relocatable in the
2130 cgroup tree, in order to allow easy resource partitioning of
2131 these objects without causing naming conflicts.
2133 * systemctl list-dependencies gained the new switches
2134 --plain, --reverse, --after and --before.
2136 * systemd-inhibit now shows the process name of processes that
2137 have taken an inhibitor lock.
2139 * nss-myhostname will now also resolve "localhost"
2140 implicitly. This makes /etc/hosts an optional file and
2141 nicely handles that on IPv6 ::1 maps to both "localhost" and
2144 * libsystemd-logind.so gained a new call
2145 sd_get_machine_names() to enumerate running containers and
2146 VMs (currently only supported by very new libvirt and
2147 nspawn). sd_login_monitor can now be used to watch
2148 VMs/containers coming and going.
2150 * .include is not allowed recursively anymore, and only in
2151 unit files. Usually it is better to use drop-in snippets in
2152 .d/*.conf anyway, as introduced with systemd 198.
2154 * systemd-analyze gained a new "critical-chain" command that
2155 determines the slowest chain of units run during system
2156 boot-up. It is very useful for tracking down where
2157 optimizing boot time is the most beneficial.
2159 * systemd will no longer allow manipulating service paths in
2160 the name=systemd:/system cgroup tree using ControlGroup= in
2161 units. (But is still fine with it in all other dirs.)
2163 * There's a new systemd-nspawn@.service service file that may
2164 be used to easily run nspawn containers as system
2165 services. With the container's root directory in
2166 /var/lib/container/foobar it is now sufficient to run
2167 "systemctl start systemd-nspawn@foobar.service" to boot it.
2169 * systemd-cgls gained a new parameter "--machine" to list only
2170 the processes within a certain container.
2172 * ConditionSecurity= now can check for "apparmor". We still
2173 are lacking checks for SMACK and IMA for this condition
2174 check though. Patches welcome!
2176 * A new configuration file /etc/systemd/sleep.conf has been
2177 added that may be used to configure which kernel operation
2178 systemd is supposed to execute when "suspend", "hibernate"
2179 or "hybrid-sleep" is requested. This makes the new kernel
2180 "freeze" state accessible to the user.
2182 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2183 the passed argument if applicable.
2185 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2186 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2187 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2188 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2189 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2190 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2191 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2196 * The output of 'systemctl list-jobs' got some polishing. The
2197 '--type=' argument may now be passed more than once. A new
2198 command 'systemctl list-sockets' has been added which shows
2199 a list of kernel sockets systemd is listening on with the
2200 socket units they belong to, plus the units these socket
2203 * The experimental libsystemd-bus library got substantial
2204 updates to work in conjunction with the (also experimental)
2205 kdbus kernel project. It works well enough to exchange
2206 messages with some sophistication. Note that kdbus is not
2207 ready yet, and the library is mostly an elaborate test case
2208 for now, and not installable.
2210 * systemd gained a new unit 'systemd-static-nodes.service'
2211 that generates static device nodes earlier during boot, and
2212 can run in conjunction with udev.
2214 * libsystemd-login gained a new call sd_pid_get_user_unit()
2215 to retrieve the user systemd unit a process is running
2216 in. This is useful for systems where systemd is used as
2219 * systemd-nspawn now places all containers in the new /machine
2220 top-level cgroup directory in the name=systemd
2221 hierarchy. libvirt will soon do the same, so that we get a
2222 uniform separation of /system, /user and /machine for system
2223 services, user processes and containers/virtual
2224 machines. This new cgroup hierarchy is also useful to stick
2225 stable names to specific container instances, which can be
2226 recognized later this way (this name may be controlled
2227 via systemd-nspawn's new -M switch). libsystemd-login also
2228 gained a new call sd_pid_get_machine_name() to retrieve the
2229 name of the container/VM a specific process belongs to.
2231 * bootchart can now store its data in the journal.
2233 * libsystemd-journal gained a new call
2234 sd_journal_add_conjunction() for AND expressions to the
2235 matching logic. This can be used to express more complex
2236 logical expressions.
2238 * journactl can now take multiple --unit= and --user-unit=
2241 * The cryptsetup logic now understands the "luks.key=" kernel
2242 command line switch for specifying a file to read the
2243 decryption key from. Also, if a configured key file is not
2244 found the tool will now automatically fall back to prompting
2247 * Python systemd.journal module was updated to wrap recently
2248 added functions from libsystemd-journal. The interface was
2249 changed to bring the low level interface in s.j._Reader
2250 closer to the C API, and the high level interface in
2251 s.j.Reader was updated to wrap and convert all data about
2254 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2255 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2256 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2257 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2258 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2259 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2263 * journalctl --update-catalog now understands a new --root=
2264 option to operate on catalogs found in a different root
2267 * During shutdown after systemd has terminated all running
2268 services a final killing loop kills all remaining left-over
2269 processes. We will now print the name of these processes
2270 when we send SIGKILL to them, since this usually indicates a
2273 * If /etc/crypttab refers to password files stored on
2274 configured mount points automatic dependencies will now be
2275 generated to ensure the specific mount is established first
2276 before the key file is attempted to be read.
2278 * 'systemctl status' will now show information about the
2279 network sockets a socket unit is listening on.
2281 * 'systemctl status' will also shown information about any
2282 drop-in configuration file for units. (Drop-In configuration
2283 files in this context are files such as
2284 /etc/systemd/systemd/foobar.service.d/*.conf)
2286 * systemd-cgtop now optionally shows summed up CPU times of
2287 cgroups. Press '%' while running cgtop to switch between
2288 percentage and absolute mode. This is useful to determine
2289 which cgroups use up the most CPU time over the entire
2290 runtime of the system. systemd-cgtop has also been updated
2291 to be 'pipeable' for processing with further shell tools.
2293 * 'hostnamectl set-hostname' will now allow setting of FQDN
2296 * The formatting and parsing of time span values has been
2297 changed. The parser now understands fractional expressions
2298 such as "5.5h". The formatter will now output fractional
2299 expressions for all time spans under 1min, i.e. "5.123456s"
2300 rather than "5s 123ms 456us". For time spans under 1s
2301 millisecond values are shown, for those under 1ms
2302 microsecond values are shown. This should greatly improve
2303 all time-related output of systemd.
2305 * libsystemd-login and libsystemd-journal gained new
2306 functions for querying the poll() events mask and poll()
2307 timeout value for integration into arbitrary event
2310 * localectl gained the ability to list available X11 keymaps
2311 (models, layouts, variants, options).
2313 * 'systemd-analyze dot' gained the ability to filter for
2314 specific units via shell-style globs, to create smaller,
2315 more useful graphs. I.e. it is now possible to create simple
2316 graphs of all the dependencies between only target units, or
2317 of all units that Avahi has dependencies with.
2319 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2320 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2321 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2322 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2323 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2324 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2325 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2329 * The boot-time readahead implementation for rotating media
2330 will now read the read-ahead data in multiple passes which
2331 consist of all read requests made in equidistant time
2332 intervals. This means instead of strictly reading read-ahead
2333 data in its physical order on disk we now try to find a
2334 middle ground between physical and access time order.
2336 * /etc/os-release files gained a new BUILD_ID= field for usage
2337 on operating systems that provide continuous builds of OS
2340 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2341 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2342 William Douglas, Zbigniew Jędrzejewski-Szmek
2346 * systemd-python gained an API exposing libsystemd-daemon.
2348 * The SMACK setup logic gained support for uploading CIPSO
2351 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2352 ReadOnlyDirectories= and InaccessibleDirectories= has
2353 changed. The private /tmp and /var/tmp directories are now
2354 shared by all processes of a service (which means
2355 ExecStartPre= may now leave data in /tmp that ExecStart= of
2356 the same service can still access). When a service is
2357 stopped its temporary directories are immediately deleted
2358 (normal clean-up with tmpfiles is still done in addition to
2361 * By default, systemd will now set a couple of sysctl
2362 variables in the kernel: the safe sysrq options are turned
2363 on, IP route verification is turned on, and source routing
2364 disabled. The recently added hardlink and softlink
2365 protection of the kernel is turned on. These settings should
2366 be reasonably safe, and good defaults for all new systems.
2368 * The predictable network naming logic may now be turned off
2369 with a new kernel command line switch: net.ifnames=0.
2371 * A new libsystemd-bus module has been added that implements a
2372 pretty complete D-Bus client library. For details see:
2374 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2376 * journald will now explicitly flush the journal files to disk
2377 at the latest 5min after each write. The file will then also
2378 be marked offline until the next write. This should increase
2379 reliability in case of a crash. The synchronization delay
2380 can be configured via SyncIntervalSec= in journald.conf.
2382 * There's a new remote-fs-setup.target unit that can be used
2383 to pull in specific services when at least one remote file
2384 system is to be mounted.
2386 * There are new targets timers.target and paths.target as
2387 canonical targets to pull user timer and path units in
2388 from. This complements sockets.target with a similar
2389 purpose for socket units.
2391 * libudev gained a new call udev_device_set_attribute_value()
2392 to set sysfs attributes of a device.
2394 * The udev daemon now sets the default number of worker
2395 processes executed in parallel based on the number of available
2396 CPUs instead of the amount of available RAM. This is supposed
2397 to provide a more reliable default and limit a too aggressive
2398 paralellism for setups with 1000s of devices connected.
2400 Contributions from: Auke Kok, Colin Walters, Cristian
2401 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2402 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2403 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2404 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2405 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2406 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2407 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2408 Zbigniew Jędrzejewski-Szmek
2412 * Configuration of unit files may now be extended via drop-in
2413 files without having to edit/override the unit files
2414 themselves. More specifically, if the administrator wants to
2415 change one value for a service file foobar.service he can
2416 now do so by dropping in a configuration snippet into
2417 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2418 will load all these snippets and apply them on top of the
2419 main unit configuration file, possibly extending or
2420 overriding its settings. Using these drop-in snippets is
2421 generally nicer than the two earlier options for changing
2422 unit files locally: copying the files from
2423 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2424 them there; or creating a new file in /etc/systemd/system/
2425 that incorporates the original one via ".include". Drop-in
2426 snippets into these .d/ directories can be placed in any
2427 directory systemd looks for units in, and the usual
2428 overriding semantics between /usr/lib, /etc and /run apply
2431 * Most unit file settings which take lists of items can now be
2432 reset by assigning the empty string to them. For example,
2433 normally, settings such as Environment=FOO=BAR append a new
2434 environment variable assignment to the environment block,
2435 each time they are used. By assigning Environment= the empty
2436 string the environment block can be reset to empty. This is
2437 particularly useful with the .d/*.conf drop-in snippets
2438 mentioned above, since this adds the ability to reset list
2439 settings from vendor unit files via these drop-ins.
2441 * systemctl gained a new "list-dependencies" command for
2442 listing the dependencies of a unit recursively.
2444 * Inhibitors are now honored and listed by "systemctl
2445 suspend", "systemctl poweroff" (and similar) too, not only
2446 GNOME. These commands will also list active sessions by
2449 * Resource limits (as exposed by the various control group
2450 controllers) can now be controlled dynamically at runtime
2451 for all units. More specifically, you can now use a command
2452 like "systemctl set-cgroup-attr foobar.service cpu.shares
2453 2000" to alter the CPU shares a specific service gets. These
2454 settings are stored persistently on disk, and thus allow the
2455 administrator to easily adjust the resource usage of
2456 services with a few simple commands. This dynamic resource
2457 management logic is also available to other programs via the
2458 bus. Almost any kernel cgroup attribute and controller is
2461 * systemd-vconsole-setup will now copy all font settings to
2462 all allocated VTs, where it previously applied them only to
2465 * libsystemd-login gained the new sd_session_get_tty() API
2468 * This release drops support for a few legacy or
2469 distribution-specific LSB facility names when parsing init
2470 scripts: $x-display-manager, $mail-transfer-agent,
2471 $mail-transport-agent, $mail-transfer-agent, $smtp,
2472 $null. Also, the mail-transfer-agent.target unit backing
2473 this has been removed. Distributions which want to retain
2474 compatibility with this should carry the burden for
2475 supporting this themselves and patch support for these back
2476 in, if they really need to. Also, the facilities $syslog and
2477 $local_fs are now ignored, since systemd does not support
2478 early-boot LSB init scripts anymore, and these facilities
2479 are implied anyway for normal services. syslog.target has
2482 * There are new bus calls on PID1's Manager object for
2483 cancelling jobs, and removing snapshot units. Previously,
2484 both calls were only available on the Job and Snapshot
2487 * systemd-journal-gatewayd gained SSL support.
2489 * The various "environment" files, such as /etc/locale.conf
2490 now support continuation lines with a backslash ("\") as
2491 last character in the line, similarly in style (but different)
2492 to how this is supported in shells.
2494 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2495 now implicitly appended to every log entry logged. systemctl
2496 has been updated to filter by this field when operating on a
2497 user systemd instance.
2499 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2500 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2501 the container. This makes it easier to boot unmodified
2502 Fedora systems in a container, which however still requires
2503 audit=0 to be passed on the kernel command line. Auditing in
2504 kernel and userspace is unfortunately still too broken in
2505 context of containers, hence we recommend compiling it out
2506 of the kernel or using audit=0. Hopefully this will be fixed
2507 one day for good in the kernel.
2509 * nspawn gained the new --bind= and --bind-ro= parameters to
2510 bind mount specific directories from the host into the
2513 * nspawn will now mount its own devpts file system instance
2514 into the container, in order not to leak pty devices from
2515 the host into the container.
2517 * systemd will now read the firmware boot time performance
2518 information from the EFI variables, if the used boot loader
2519 supports this, and takes it into account for boot performance
2520 analysis via "systemd-analyze". This is currently supported
2521 only in conjunction with Gummiboot, but could be supported
2522 by other boot loaders too. For details see:
2524 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2526 * A new generator has been added that automatically mounts the
2527 EFI System Partition (ESP) to /boot, if that directory
2528 exists, is empty, and no other file system has been
2529 configured to be mounted there.
2531 * logind will now send out PrepareForSleep(false) out
2532 unconditionally, after coming back from suspend. This may be
2533 used by applications as asynchronous notification for
2534 system resume events.
2536 * "systemctl unlock-sessions" has been added, that allows
2537 unlocking the screens of all user sessions at once, similar
2538 to how "systemctl lock-sessions" already locked all users
2539 sessions. This is backed by a new D-Bus call UnlockSessions().
2541 * "loginctl seat-status" will now show the master device of a
2542 seat. (i.e. the device of a seat that needs to be around for
2543 the seat to be considered available, usually the graphics
2546 * tmpfiles gained a new "X" line type, that allows
2547 configuration of files and directories (with wildcards) that
2548 shall be excluded from automatic cleanup ("aging").
2550 * udev default rules set the device node permissions now only
2551 at "add" events, and do not change them any longer with a
2552 later "change" event.
2554 * The log messages for lid events and power/sleep keypresses
2555 now carry a message ID.
2557 * We now have a substantially larger unit test suite, but this
2558 continues to be work in progress.
2560 * udevadm hwdb gained a new --root= parameter to change the
2561 root directory to operate relative to.
2563 * logind will now issue a background sync() request to the kernel
2564 early at shutdown, so that dirty buffers are flushed to disk early
2565 instead of at the last moment, in order to optimize shutdown
2568 * A new bootctl tool has been added that is an interface for
2569 certain boot loader operations. This is currently a preview
2570 and is likely to be extended into a small mechanism daemon
2571 like timedated, localed, hostnamed, and can be used by
2572 graphical UIs to enumerate available boot options, and
2573 request boot into firmware operations.
2575 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2576 the rest of the package. It also has been updated to work
2577 correctly in initrds.
2579 * Policykit previously has been runtime optional, and is now
2580 also compile time optional via a configure switch.
2582 * systemd-analyze has been reimplemented in C. Also "systemctl
2583 dot" has moved into systemd-analyze.
2585 * "systemctl status" with no further parameters will now print
2586 the status of all active or failed units.
2588 * Operations such as "systemctl start" can now be executed
2589 with a new mode "--irreversible" which may be used to queue
2590 operations that cannot accidentally be reversed by a later
2591 job queuing. This is by default used to make shutdown
2592 requests more robust.
2594 * The Python API of systemd now gained a new module for
2595 reading journal files.
2597 * A new tool kernel-install has been added that can install
2598 kernel images according to the Boot Loader Specification:
2600 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2602 * Boot time console output has been improved to provide
2603 animated boot time output for hanging jobs.
2605 * A new tool systemd-activate has been added which can be used
2606 to test socket activation with, directly from the command
2607 line. This should make it much easier to test and debug
2608 socket activation in daemons.
2610 * journalctl gained a new "--reverse" (or -r) option to show
2611 journal output in reverse order (i.e. newest line first).
2613 * journalctl gained a new "--pager-end" (or -e) option to jump
2614 to immediately jump to the end of the journal in the
2615 pager. This is only supported in conjunction with "less".
2617 * journalctl gained a new "--user-unit=" option, that works
2618 similarly to "--unit=" but filters for user units rather than
2621 * A number of unit files to ease adoption of systemd in
2622 initrds has been added. This moves some minimal logic from
2623 the various initrd implementations into systemd proper.
2625 * The journal files are now owned by a new group
2626 "systemd-journal", which exists specifically to allow access
2627 to the journal, and nothing else. Previously, we used the
2628 "adm" group for that, which however possibly covers more
2629 than just journal/log file access. This new group is now
2630 already used by systemd-journal-gatewayd to ensure this
2631 daemon gets access to the journal files and as little else
2632 as possible. Note that "make install" will also set FS ACLs
2633 up for /var/log/journal to give "adm" and "wheel" read
2634 access to it, in addition to "systemd-journal" which owns
2635 the journal files. We recommend that packaging scripts also
2636 add read access to "adm" + "wheel" to /var/log/journal, and
2637 all existing/future journal files. To normal users and
2638 administrators little changes, however packagers need to
2639 ensure to create the "systemd-journal" system group at
2640 package installation time.
2642 * The systemd-journal-gatewayd now runs as unprivileged user
2643 systemd-journal-gateway:systemd-journal-gateway. Packaging
2644 scripts need to create these system user/group at
2647 * timedated now exposes a new boolean property CanNTP that
2648 indicates whether a local NTP service is available or not.
2650 * systemd-detect-virt will now also detect xen PVs
2652 * The pstore file system is now mounted by default, if it is
2655 * In addition to the SELinux and IMA policies we will now also
2656 load SMACK policies at early boot.
2658 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2659 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2660 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2661 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2662 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2663 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2664 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2665 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2666 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2667 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2668 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2669 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2670 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2671 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2675 * Timer units now support calendar time events in addition to
2676 monotonic time events. That means you can now trigger a unit
2677 based on a calendar time specification such as "Thu,Fri
2678 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2679 or fifth day of any month of the year 2013, given that it is
2680 a thursday or friday. This brings timer event support
2681 considerably closer to cron's capabilities. For details on
2682 the supported calendar time specification language see
2685 * udev now supports a number of different naming policies for
2686 network interfaces for predictable names, and a combination
2687 of these policies is now the default. Please see this wiki
2688 document for details:
2690 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2692 * Auke Kok's bootchart implementation has been added to the
2693 systemd tree. It is an optional component that can graph the
2694 boot in quite some detail. It is one of the best bootchart
2695 implementations around and minimal in its code and
2698 * nss-myhostname has been integrated into the systemd source
2699 tree. nss-myhostname guarantees that the local hostname
2700 always stays resolvable via NSS. It has been a weak
2701 requirement of systemd-hostnamed since a long time, and
2702 since its code is actually trivial we decided to just
2703 include it in systemd's source tree. It can be turned off
2704 with a configure switch.
2706 * The read-ahead logic is now capable of properly detecting
2707 whether a btrfs file system is on SSD or rotating media, in
2708 order to optimize the read-ahead scheme. Previously, it was
2709 only capable of detecting this on traditional file systems
2712 * In udev, additional device properties are now read from the
2713 IAB in addition to the OUI database. Also, Bluetooth company
2714 identities are attached to the devices as well.
2716 * In service files %U may be used as specifier that is
2717 replaced by the configured user name of the service.
2719 * nspawn may now be invoked without a controlling TTY. This
2720 makes it suitable for invocation as its own service. This
2721 may be used to set up a simple containerized server system
2722 using only core OS tools.
2724 * systemd and nspawn can now accept socket file descriptors
2725 when they are started for socket activation. This enables
2726 implementation of socket activated nspawn
2727 containers. i.e. think about autospawning an entire OS image
2728 when the first SSH or HTTP connection is received. We expect
2729 that similar functionality will also be added to libvirt-lxc
2732 * journalctl will now suppress ANSI color codes when
2733 presenting log data.
2735 * systemctl will no longer show control group information for
2736 a unit if a the control group is empty anyway.
2738 * logind can now automatically suspend/hibernate/shutdown the
2741 * /etc/machine-info and hostnamed now also expose the chassis
2742 type of the system. This can be used to determine whether
2743 the local system is a laptop, desktop, handset or
2744 tablet. This information may either be configured by the
2745 user/vendor or is automatically determined from ACPI and DMI
2746 information if possible.
2748 * A number of PolicyKit actions are now bound together with
2749 "imply" rules. This should simplify creating UIs because
2750 many actions will now authenticate similar ones as well.
2752 * Unit files learnt a new condition ConditionACPower= which
2753 may be used to conditionalize a unit depending on whether an
2754 AC power source is connected or not, of whether the system
2755 is running on battery power.
2757 * systemctl gained a new "is-failed" verb that may be used in
2758 shell scripts and suchlike to check whether a specific unit
2759 is in the "failed" state.
2761 * The EnvironmentFile= setting in unit files now supports file
2762 globbing, and can hence be used to easily read a number of
2763 environment files at once.
2765 * systemd will no longer detect and recognize specific
2766 distributions. All distribution-specific #ifdeffery has been
2767 removed, systemd is now fully generic and
2768 distribution-agnostic. Effectively, not too much is lost as
2769 a lot of the code is still accessible via explicit configure
2770 switches. However, support for some distribution specific
2771 legacy configuration file formats has been dropped. We
2772 recommend distributions to simply adopt the configuration
2773 files everybody else uses now and convert the old
2774 configuration from packaging scripts. Most distributions
2775 already did that. If that's not possible or desirable,
2776 distributions are welcome to forward port the specific
2777 pieces of code locally from the git history.
2779 * When logging a message about a unit systemd will now always
2780 log the unit name in the message meta data.
2782 * localectl will now also discover system locale data that is
2783 not stored in locale archives, but directly unpacked.
2785 * logind will no longer unconditionally use framebuffer
2786 devices as seat masters, i.e. as devices that are required
2787 to be existing before a seat is considered preset. Instead,
2788 it will now look for all devices that are tagged as
2789 "seat-master" in udev. By default framebuffer devices will
2790 be marked as such, but depending on local systems other
2791 devices might be marked as well. This may be used to
2792 integrate graphics cards using closed source drivers (such
2793 as NVidia ones) more nicely into logind. Note however, that
2794 we recommend using the open source NVidia drivers instead,
2795 and no udev rules for the closed-source drivers will be
2796 shipped from us upstream.
2798 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2799 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2800 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2801 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2802 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2803 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2804 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2805 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2806 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2807 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2808 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2809 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2814 * udev gained support for loading additional device properties
2815 from an indexed database that is keyed by vendor/product IDs
2816 and similar device identifiers. For the beginning this
2817 "hwdb" is populated with data from the well-known PCI and
2818 USB database, but also includes PNP, ACPI and OID data. In
2819 the longer run this indexed database shall grow into
2820 becoming the one central database for non-essential
2821 userspace device metadata. Previously, data from the PCI/USB
2822 database was only attached to select devices, since the
2823 lookup was a relatively expensive operation due to O(n) time
2824 complexity (with n being the number of entries in the
2825 database). Since this is now O(1), we decided to add in this
2826 data for all devices where this is available, by
2827 default. Note that the indexed database needs to be rebuilt
2828 when new data files are installed. To achieve this you need
2829 to update your packaging scripts to invoke "udevadm hwdb
2830 --update" after installation of hwdb data files. For
2831 RPM-based distributions we introduced the new
2832 %udev_hwdb_update macro for this purpose.
2834 * The Journal gained support for the "Message Catalog", an
2835 indexed database to link up additional information with
2836 journal entries. For further details please check:
2838 http://www.freedesktop.org/wiki/Software/systemd/catalog
2840 The indexed message catalog database also needs to be
2841 rebuilt after installation of message catalog files. Use
2842 "journalctl --update-catalog" for this. For RPM-based
2843 distributions we introduced the %journal_catalog_update
2844 macro for this purpose.
2846 * The Python Journal bindings gained support for the standard
2847 Python logging framework.
2849 * The Journal API gained new functions for checking whether
2850 the underlying file system of a journal file is capable of
2851 properly reporting file change notifications, or whether
2852 applications that want to reflect journal changes "live"
2853 need to recheck journal files continuously in appropriate
2856 * It is now possible to set the "age" field for tmpfiles
2857 entries to 0, indicating that files matching this entry
2858 shall always be removed when the directories are cleaned up.
2860 * coredumpctl gained a new "gdb" verb which invokes gdb
2861 right-away on the selected coredump.
2863 * There's now support for "hybrid sleep" on kernels that
2864 support this, in addition to "suspend" and "hibernate". Use
2865 "systemctl hybrid-sleep" to make use of this.
2867 * logind's HandleSuspendKey= setting (and related settings)
2868 now gained support for a new "lock" setting to simply
2869 request the screen lock on all local sessions, instead of
2870 actually executing a suspend or hibernation.
2872 * systemd will now mount the EFI variables file system by
2875 * Socket units now gained support for configuration of the
2876 SMACK security label.
2878 * timedatectl will now output the time of the last and next
2879 daylight saving change.
2881 * We dropped support for various legacy and distro-specific
2882 concepts, such as insserv, early-boot SysV services
2883 (i.e. those for non-standard runlevels such as 'b' or 'S')
2884 or ArchLinux /etc/rc.conf support. We recommend the
2885 distributions who still need support this to either continue
2886 to maintain the necessary patches downstream, or find a
2887 different solution. (Talk to us if you have questions!)
2889 * Various systemd components will now bypass PolicyKit checks
2890 for root and otherwise handle properly if PolicyKit is not
2891 found to be around. This should fix most issues for
2892 PolicyKit-less systems. Quite frankly this should have been
2893 this way since day one. It is absolutely our intention to
2894 make systemd work fine on PolicyKit-less systems, and we
2895 consider it a bug if something does not work as it should if
2896 PolicyKit is not around.
2898 * For embedded systems it is now possible to build udev and
2899 systemd without blkid and/or kmod support.
2901 * "systemctl switch-root" is now capable of switching root
2902 more than once. I.e. in addition to transitions from the
2903 initrd to the host OS it is now possible to transition to
2904 further OS images from the host. This is useful to implement
2905 offline updating tools.
2907 * Various other additions have been made to the RPM macros
2908 shipped with systemd. Use %udev_rules_update() after
2909 installing new udev rules files. %_udevhwdbdir,
2910 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2911 %_sysctldir are now available which resolve to the right
2912 directories for packages to place various data files in.
2914 * journalctl gained the new --full switch (in addition to
2915 --all, to disable ellipsation for long messages.
2917 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2918 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2919 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2920 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2921 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2922 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2923 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2924 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2925 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2929 * journalctl gained new --since= and --until= switches to
2930 filter by time. It also now supports nice filtering for
2931 units via --unit=/-u.
2933 * Type=oneshot services may use ExecReload= and do the
2936 * The journal daemon now supports time-based rotation and
2937 vacuuming, in addition to the usual disk-space based
2940 * The journal will now index the available field values for
2941 each field name. This enables clients to show pretty drop
2942 downs of available match values when filtering. The bash
2943 completion of journalctl has been updated
2944 accordingly. journalctl gained a new switch -F to list all
2945 values a certain field takes in the journal database.
2947 * More service events are now written as structured messages
2948 to the journal, and made recognizable via message IDs.
2950 * The timedated, localed and hostnamed mini-services which
2951 previously only provided support for changing time, locale
2952 and hostname settings from graphical DEs such as GNOME now
2953 also have a minimal (but very useful) text-based client
2954 utility each. This is probably the nicest way to changing
2955 these settings from the command line now, especially since
2956 it lists available options and is fully integrated with bash
2959 * There's now a new tool "systemd-coredumpctl" to list and
2960 extract coredumps from the journal.
2962 * We now install a README each in /var/log/ and
2963 /etc/rc.d/init.d explaining where the system logs and init
2964 scripts went. This hopefully should help folks who go to
2965 that dirs and look into the otherwise now empty void and
2966 scratch their heads.
2968 * When user-services are invoked (by systemd --user) the
2969 $MANAGERPID env var is set to the PID of systemd.
2971 * SIGRTMIN+24 when sent to a --user instance will now result
2972 in immediate termination of systemd.
2974 * gatewayd received numerous feature additions such as a
2975 "follow" mode, for live syncing and filtering.
2977 * browse.html now allows filtering and showing detailed
2978 information on specific entries. Keyboard navigation and
2979 mouse screen support has been added.
2981 * gatewayd/journalctl now supports HTML5/JSON
2982 Server-Sent-Events as output.
2984 * The SysV init script compatibility logic will now
2985 heuristically determine whether a script supports the
2986 "reload" verb, and only then make this available as
2989 * "systemctl status --follow" has been removed, use "journalctl
2992 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
2993 have been removed since they are hardly useful to be
2996 * And I'd like to take the opportunity to specifically mention
2997 Zbigniew for his great contributions. Zbigniew, you rock!
2999 Contributions from: Andrew Eikum, Christian Hesse, Colin
3000 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3001 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3002 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3003 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3004 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3005 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3009 * If /etc/vconsole.conf is non-existent or empty we will no
3010 longer load any console font or key map at boot by
3011 default. Instead the kernel defaults will be left
3012 intact. This is definitely the right thing to do, as no
3013 configuration should mean no configuration, and hard-coding
3014 font names that are different on all archs is probably a bad
3015 idea. Also, the kernel default key map and font should be
3016 good enough for most cases anyway, and mostly identical to
3017 the userspace fonts/key maps we previously overloaded them
3018 with. If distributions want to continue to default to a
3019 non-kernel font or key map they should ship a default
3020 /etc/vconsole.conf with the appropriate contents.
3022 Contributions from: Colin Walters, Daniel J Walsh, Dave
3023 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3024 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3028 * journalctl gained a new --cursor= switch to show entries
3029 starting from the specified location in the journal.
3031 * We now enforce a size limit on journal entry fields exported
3032 with "-o json" in journalctl. Fields larger than 4K will be
3033 assigned null. This can be turned off with --all.
3035 * An (optional) journal gateway daemon is now available as
3036 "systemd-journal-gatewayd.service". This service provides
3037 access to the journal via HTTP and JSON. This functionality
3038 will be used to implement live log synchronization in both
3039 pull and push modes, but has various other users too, such
3040 as easy log access for debugging of embedded devices. Right
3041 now it is already useful to retrieve the journal via HTTP:
3043 # systemctl start systemd-journal-gatewayd.service
3044 # wget http://localhost:19531/entries
3046 This will download the journal contents in a
3047 /var/log/messages compatible format. The same as JSON:
3049 # curl -H"Accept: application/json" http://localhost:19531/entries
3051 This service is also accessible via a web browser where a
3052 single static HTML5 app is served that uses the JSON logic
3053 to enable the user to do some basic browsing of the
3054 journal. This will be extended later on. Here's an example
3055 screenshot of this app in its current state:
3057 http://0pointer.de/public/journal-gatewayd
3059 Contributions from: Kay Sievers, Lennart Poettering, Robert
3060 Milasan, Tom Gundersen
3064 * The bash completion logic is now available for journalctl
3067 * We do not mount the "cpuset" controller anymore together with
3068 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3069 started if no parameters are assigned to it. "cpuset" hence
3070 broke code that assumed it it could create "cpu" groups and
3073 * journalctl -f will now subscribe to terminal size changes,
3074 and line break accordingly.
3076 Contributions from: Dave Reisner, Kay Sievers, Lennart
3077 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3081 * nspawn will now create a symlink /etc/localtime in the
3082 container environment, copying the host's timezone
3083 setting. Previously this has been done via a bind mount, but
3084 since symlinks cannot be bind mounted this has now been
3085 changed to create/update the appropriate symlink.
3087 * journalctl -n's line number argument is now optional, and
3088 will default to 10 if omitted.
3090 * journald will now log the maximum size the journal files may
3091 take up on disk. This is particularly useful if the default
3092 built-in logic of determining this parameter from the file
3093 system size is used. Use "systemctl status
3094 systemd-journald.service" to see this information.
3096 * The multi-seat X wrapper tool has been stripped down. As X
3097 is now capable of enumerating graphics devices via udev in a
3098 seat-aware way the wrapper is not strictly necessary
3099 anymore. A stripped down temporary stop-gap is still shipped
3100 until the upstream display managers have been updated to
3101 fully support the new X logic. Expect this wrapper to be
3102 removed entirely in one of the next releases.
3104 * HandleSleepKey= in logind.conf has been split up into
3105 HandleSuspendKey= and HandleHibernateKey=. The old setting
3106 is not available anymore. X11 and the kernel are
3107 distuingishing between these keys and we should too. This
3108 also means the inhibition lock for these keys has been split
3111 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3112 Poettering, Lukas Nykryn, Václav Pavlín
3116 * Whenever a unit changes state we will now log this to the
3117 journal and show along the unit's own log output in
3120 * ConditionPathIsMountPoint= can now properly detect bind
3121 mount points too. (Previously, a bind mount of one file
3122 system to another place in the same file system could not be
3123 detected as mount, since they shared struct stat's st_dev
3126 * We will now mount the cgroup controllers cpu, cpuacct,
3127 cpuset and the controllers net_cls, net_prio together by
3130 * nspawn containers will now have a virtualized boot
3131 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3132 over with a randomized ID at container initialization). This
3133 has the effect of making "journalctl -b" do the right thing
3136 * The JSON output journal serialization has been updated not
3137 to generate "endless" list objects anymore, but rather one
3138 JSON object per line. This is more in line how most JSON
3139 parsers expect JSON objects. The new output mode
3140 "json-pretty" has been added to provide similar output, but
3141 neatly aligned for readability by humans.
3143 * We dropped all explicit sync() invocations in the shutdown
3144 code. The kernel does this implicitly anyway in the kernel
3145 reboot() syscall. halt(8)'s -n option is now a compatibility
3148 * We now support virtualized reboot() in containers, as
3149 supported by newer kernels. We will fall back to exit() if
3150 CAP_SYS_REBOOT is not available to the container. Also,
3151 nspawn makes use of this now and will actually reboot the
3152 container if the containerized OS asks for that.
3154 * journalctl will only show local log output by default
3155 now. Use --merge (-m) to show remote log output, too.
3157 * libsystemd-journal gained the new sd_journal_get_usage()
3158 call to determine the current disk usage of all journal
3159 files. This is exposed in the new "journalctl --disk-usage"
3162 * journald gained a new configuration setting SplitMode= in
3163 journald.conf which may be used to control how user journals
3164 are split off. See journald.conf(5) for details.
3166 * A new condition type ConditionFileNotEmpty= has been added.
3168 * tmpfiles' "w" lines now support file globbing, to write
3169 multiple files at once.
3171 * We added Python bindings for the journal submission
3172 APIs. More Python APIs for a number of selected APIs will
3173 likely follow. Note that we intend to add native bindings
3174 only for the Python language, as we consider it common
3175 enough to deserve bindings shipped within systemd. There are
3176 various projects outside of systemd that provide bindings
3177 for languages such as PHP or Lua.
3179 * Many conditions will now resolve specifiers such as %i. In
3180 addition, PathChanged= and related directives of .path units
3181 now support specifiers as well.
3183 * There's now a new RPM macro definition for the system preset
3186 * journald will now warn if it ca not forward a message to the
3187 syslog daemon because its socket is full.
3189 * timedated will no longer write or process /etc/timezone,
3190 except on Debian. As we do not support late mounted /usr
3191 anymore /etc/localtime always being a symlink is now safe,
3192 and hence the information in /etc/timezone is not necessary
3195 * logind will now always reserve one VT for a text getty (VT6
3196 by default). Previously if more than 6 X sessions where
3197 started they took up all the VTs with auto-spawned gettys,
3198 so that no text gettys were available anymore.
3200 * udev will now automatically inform the btrfs kernel logic
3201 about btrfs RAID components showing up. This should make
3202 simple hotplug based btrfs RAID assembly work.
3204 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3205 (but not for its children which will stay at the kernel
3206 default). This should allow setups with a lot more listening
3209 * systemd will now always pass the configured timezone to the
3210 kernel at boot. timedated will do the same when the timezone
3213 * logind's inhibition logic has been updated. By default,
3214 logind will now handle the lid switch, the power and sleep
3215 keys all the time, even in graphical sessions. If DEs want
3216 to handle these events on their own they should take the new
3217 handle-power-key, handle-sleep-key and handle-lid-switch
3218 inhibitors during their runtime. A simple way to achiveve
3219 that is to invoke the DE wrapped in an invocation of:
3221 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3223 * Access to unit operations is now checked via SELinux taking
3224 the unit file label and client process label into account.
3226 * systemd will now notify the administrator in the journal
3227 when he over-mounts a non-empty directory.
3229 * There are new specifiers that are resolved in unit files,
3230 for the host name (%H), the machine ID (%m) and the boot ID
3233 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3234 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3235 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3236 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3237 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3238 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3239 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3243 * Support for reading structured kernel messages from
3244 /dev/kmsg has now been added and is enabled by default.
3246 * Support for reading kernel messages from /proc/kmsg has now
3247 been removed. If you want kernel messages in the journal
3248 make sure to run a recent kernel (>= 3.5) that supports
3249 reading structured messages from /dev/kmsg (see
3250 above). /proc/kmsg is now exclusive property of classic
3251 syslog daemons again.
3253 * The libudev API gained the new
3254 udev_device_new_from_device_id() call.
3256 * The logic for file system namespace (ReadOnlyDirectory=,
3257 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3258 require pivot_root() anymore. This means fewer temporary
3259 directories are created below /tmp for this feature.
3261 * nspawn containers will now see and receive all submounts
3262 made on the host OS below the root file system of the
3265 * Forward Secure Sealing is now supported for Journal files,
3266 which provide cryptographical sealing of journal files so
3267 that attackers cannot alter log history anymore without this
3268 being detectable. Lennart will soon post a blog story about
3269 this explaining it in more detail.
3271 * There are two new service settings RestartPreventExitStatus=
3272 and SuccessExitStatus= which allow configuration of exit
3273 status (exit code or signal) which will be excepted from the
3274 restart logic, resp. consider successful.
3276 * journalctl gained the new --verify switch that can be used
3277 to check the integrity of the structure of journal files and
3278 (if Forward Secure Sealing is enabled) the contents of
3281 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3282 and similar symlinks pre-created. This makes running shells
3283 as container init process a lot more fun.
3285 * The fstab support can now handle PARTUUID= and PARTLABEL=
3288 * A new ConditionHost= condition has been added to match
3289 against the hostname (with globs) and machine ID. This is
3290 useful for clusters where a single OS image is used to
3291 provision a large number of hosts which shall run slightly
3292 different sets of services.
3294 * Services which hit the restart limit will now be placed in a
3297 Contributions from: Bertram Poettering, Dave Reisner, Huang
3298 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3299 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3303 * When running in --user mode systemd will now become a
3304 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3305 tree a lot more organized.
3307 * A new PartOf= unit dependency type has been introduced that
3308 may be used to group services in a natural way.
3310 * "systemctl enable" may now be used to enable instances of
3313 * journalctl now prints error log levels in red, and
3314 warning/notice log levels in bright white. It also supports
3315 filtering by log level now.
3317 * cgtop gained a new -n switch (similar to top), to configure
3318 the maximum number of iterations to run for. It also gained
3319 -b, to run in batch mode (accepting no input).
3321 * The suffix ".service" may now be omitted on most systemctl
3322 command lines involving service unit names.
3324 * There's a new bus call in logind to lock all sessions, as
3325 well as a loginctl verb for it "lock-sessions".
3327 * libsystemd-logind.so gained a new call sd_journal_perror()
3328 that works similar to libc perror() but logs to the journal
3329 and encodes structured information about the error number.
3331 * /etc/crypttab entries now understand the new keyfile-size=
3334 * shutdown(8) now can send a (configurable) wall message when
3335 a shutdown is cancelled.
3337 * The mount propagation mode for the root file system will now
3338 default to "shared", which is useful to make containers work
3339 nicely out-of-the-box so that they receive new mounts from
3340 the host. This can be undone locally by running "mount
3341 --make-rprivate /" if needed.
3343 * The prefdm.service file has been removed. Distributions
3344 should maintain this unit downstream if they intend to keep
3345 it around. However, we recommend writing normal unit files
3346 for display managers instead.
3348 * Since systemd is a crucial part of the OS we will now
3349 default to a number of compiler switches that improve
3350 security (hardening) such as read-only relocations, stack
3351 protection, and suchlike.
3353 * The TimeoutSec= setting for services is now split into
3354 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3355 of individual time outs for the start and the stop phase of
3358 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3359 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3360 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3361 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3362 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3363 Gundersen, Zbigniew Jędrzejewski-Szmek
3367 * The journal and id128 C APIs are now fully documented as man
3370 * Extra safety checks have been added when transitioning from
3371 the initial RAM disk to the main system to avoid accidental
3374 * /etc/crypttab entries now understand the new keyfile-offset=
3377 * systemctl -t can now be used to filter by unit load state.
3379 * The journal C API gained the new sd_journal_wait() call to
3380 make writing synchronous journal clients easier.
3382 * journalctl gained the new -D switch to show journals from a
3385 * journalctl now displays a special marker between log
3386 messages of two different boots.
3388 * The journal is now explicitly flushed to /var via a service
3389 systemd-journal-flush.service, rather than implicitly simply
3390 by seeing /var/log/journal to be writable.
3392 * journalctl (and the journal C APIs) can now match for much
3393 more complex expressions, with alternatives and
3396 * When transitioning from the initial RAM disk to the main
3397 system we will now kill all processes in a killing spree to
3398 ensure no processes stay around by accident.
3400 * Three new specifiers may be used in unit files: %u, %h, %s
3401 resolve to the user name, user home directory resp. user
3402 shell. This is useful for running systemd user instances.
3404 * We now automatically rotate journal files if their data
3405 object hash table gets a fill level > 75%. We also size the
3406 hash table based on the configured maximum file size. This
3407 together should lower hash collisions drastically and thus
3408 speed things up a bit.
3410 * journalctl gained the new "--header" switch to introspect
3411 header data of journal files.
3413 * A new setting SystemCallFilters= has been added to services
3414 which may be used to apply blacklists or whitelists to
3415 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3417 * nspawn gained a new --link-journal= switch (and quicker: -j)
3418 to link the container journal with the host. This makes it
3419 very easy to centralize log viewing on the host for all
3420 guests while still keeping the journal files separated.
3422 * Many bugfixes and optimizations
3424 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3425 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3426 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3431 * Several tools now understand kernel command line arguments,
3432 which are only read when run in an initial RAM disk. They
3433 usually follow closely their normal counterparts, but are
3436 * There's a new tool to analyze the readahead files that are
3437 automatically generated at boot. Use:
3439 /usr/lib/systemd/systemd-readahead analyze /.readahead
3441 * We now provide an early debug shell on tty9 if this enabled. Use:
3443 systemctl enable debug-shell.service
3445 * All plymouth related units have been moved into the Plymouth
3446 package. Please make sure to upgrade your Plymouth version
3449 * systemd-tmpfiles now supports getting passed the basename of
3450 a configuration file only, in which case it will look for it
3451 in all appropriate directories automatically.
3453 * udevadm info now takes a /dev or /sys path as argument, and
3454 does the right thing. Example:
3456 udevadm info /dev/sda
3457 udevadm info /sys/class/block/sda
3459 * systemctl now prints a warning if a unit is stopped but a
3460 unit that might trigger it continues to run. Example: a
3461 service is stopped but the socket that activates it is left
3464 * "systemctl status" will now mention if the log output was
3465 shortened due to rotation since a service has been started.
3467 * The journal API now exposes functions to determine the
3468 "cutoff" times due to rotation.
3470 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3471 immediately flushing of runtime logs to /var if possible,
3472 resp. for triggering immediate rotation of the journal
3475 * It is now considered an error if a service is attempted to
3476 be stopped that is not loaded.
3478 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3480 * systemd-analyze now supports Python 3
3482 * tmpfiles now supports cleaning up directories via aging
3483 where the first level dirs are always kept around but
3484 directories beneath it automatically aged. This is enabled
3485 by prefixing the age field with '~'.
3487 * Seat objects now expose CanGraphical, CanTTY properties
3488 which is required to deal with very fast bootups where the
3489 display manager might be running before the graphics drivers
3490 completed initialization.
3492 * Seat objects now expose a State property.
3494 * We now include RPM macros for service enabling/disabling
3495 based on the preset logic. We recommend RPM based
3496 distributions to make use of these macros if possible. This
3497 makes it simpler to reuse RPM spec files across
3500 * We now make sure that the collected systemd unit name is
3501 always valid when services log to the journal via
3504 * There's a new man page kernel-command-line(7) detailing all
3505 command line options we understand.
3507 * The fstab generator may now be disabled at boot by passing
3508 fstab=0 on the kernel command line.
3510 * A new kernel command line option modules-load= is now understood
3511 to load a specific kernel module statically, early at boot.
3513 * Unit names specified on the systemctl command line are now
3514 automatically escaped as needed. Also, if file system or
3515 device paths are specified they are automatically turned
3516 into the appropriate mount or device unit names. Example:
3518 systemctl status /home
3519 systemctl status /dev/sda
3521 * The SysVConsole= configuration option has been removed from
3522 system.conf parsing.
3524 * The SysV search path is no longer exported on the D-Bus
3527 * The Names= option is been removed from unit file parsing.
3529 * There's a new man page bootup(7) detailing the boot process.
3531 * Every unit and every generator we ship with systemd now
3532 comes with full documentation. The self-explanatory boot is
3535 * A couple of services gained "systemd-" prefixes in their
3536 name if they wrap systemd code, rather than only external
3537 code. Among them fsck@.service which is now
3538 systemd-fsck@.service.
3540 * The HaveWatchdog property has been removed from the D-Bus
3543 * systemd.confirm_spawn= on the kernel command line should now
3546 * There's a new man page crypttab(5) which details all options
3547 we actually understand.
3549 * systemd-nspawn gained a new --capability= switch to pass
3550 additional capabilities to the container.
3552 * timedated will now read known NTP implementation unit names
3553 from /usr/lib/systemd/ntp-units.d/*.list,
3554 systemd-timedated-ntp.target has been removed.
3556 * journalctl gained a new switch "-b" that lists log data of
3557 the current boot only.
3559 * The notify socket is in the abstract namespace again, in
3560 order to support daemons which chroot() at start-up.
3562 * There is a new Storage= configuration option for journald
3563 which allows configuration of where log data should go. This
3564 also provides a way to disable journal logging entirely, so
3565 that data collected is only forwarded to the console, the
3566 kernel log buffer or another syslog implementation.
3568 * Many bugfixes and optimizations
3570 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3571 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3572 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3573 Shawn Landden, Tom Gundersen
3577 * "systemctl help <unit>" now shows the man page if one is
3580 * Several new man pages have been added.
3582 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3583 MaxLevelConsole= can now be specified in
3584 journald.conf. These options allow reducing the amount of
3585 data stored on disk or forwarded by the log level.
3587 * TimerSlackNSec= can now be specified in system.conf for
3588 PID1. This allows system-wide power savings.
3590 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3591 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3596 * logind is now capable of (optionally) handling power and
3597 sleep keys as well as the lid switch.
3599 * journalctl now understands the syntax "journalctl
3600 /usr/bin/avahi-daemon" to get all log output of a specific
3603 * CapabilityBoundingSet= in system.conf now also influences
3604 the capability bound set of usermode helpers of the kernel.
3606 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3607 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3608 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3609 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3613 * Note that we skipped 139 releases here in order to set the
3614 new version to something that is greater than both udev's
3615 and systemd's most recent version number.
3617 * udev: all udev sources are merged into the systemd source tree now.
3618 All future udev development will happen in the systemd tree. It
3619 is still fully supported to use the udev daemon and tools without
3620 systemd running, like in initramfs or other init systems. Building
3621 udev though, will require the *build* of the systemd tree, but
3622 udev can be properly *run* without systemd.
3624 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3625 should be used to create dead device nodes as workarounds for broken
3628 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3629 no longer supported. udev_monitor_new_from_netlink() needs to be
3630 used to subscribe to events.
3632 * udev: when udevd is started by systemd, processes which are left
3633 behind by forking them off of udev rules, are unconditionally cleaned
3634 up and killed now after the event handling has finished. Services or
3635 daemons must be started as systemd services. Services can be
3636 pulled-in by udev to get started, but they can no longer be directly
3637 forked by udev rules.
3639 * udev: the daemon binary is called systemd-udevd now and installed
3640 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3641 to adapt to that, create symlink, or rename the binary after building
3644 * libudev no longer provides these symbols:
3645 udev_monitor_from_socket()
3646 udev_queue_get_failed_list_entry()
3647 udev_get_{dev,sys,run}_path()
3648 The versions number was bumped and symbol versioning introduced.
3650 * systemd-loginctl and systemd-journalctl have been renamed
3651 to loginctl and journalctl to match systemctl.
3653 * The config files: /etc/systemd/systemd-logind.conf and
3654 /etc/systemd/systemd-journald.conf have been renamed to
3655 logind.conf and journald.conf. Package updates should rename
3656 the files to the new names on upgrade.
3658 * For almost all files the license is now LGPL2.1+, changed
3659 from the previous GPL2.0+. Exceptions are some minor stuff
3660 of udev (which will be changed to LGPL2.1 eventually, too),
3661 and the MIT licensed sd-daemon.[ch] library that is suitable
3662 to be used as drop-in files.
3664 * systemd and logind now handle system sleep states, in
3665 particular suspending and hibernating.
3667 * logind now implements a sleep/shutdown/idle inhibiting logic
3668 suitable for a variety of uses. Soonishly Lennart will blog
3669 about this in more detail.
3671 * var-run.mount and var-lock.mount are no longer provided
3672 (which prevously bind mounted these directories to their new
3673 places). Distributions which have not converted these
3674 directories to symlinks should consider stealing these files
3675 from git history and add them downstream.
3677 * We introduced the Documentation= field for units and added
3678 this to all our shipped units. This is useful to make it
3679 easier to explore the boot and the purpose of the various
3682 * All smaller setup units (such as
3683 systemd-vconsole-setup.service) now detect properly if they
3684 are run in a container and are skipped when
3685 appropriate. This guarantees an entirely noise-free boot in
3686 Linux container environments such as systemd-nspawn.
3688 * A framework for implementing offline system updates is now
3689 integrated, for details see:
3690 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3692 * A new service type Type=idle is available now which helps us
3693 avoiding ugly interleaving of getty output and boot status
3696 * There's now a system-wide CapabilityBoundingSet= option to
3697 globally reduce the set of capabilities for the
3698 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3699 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3700 even CAP_NET_ADMIN system-wide for secure systems.
3702 * There are now system-wide DefaultLimitXXX= options to
3703 globally change the defaults of the various resource limits
3704 for all units started by PID 1.
3706 * Harald Hoyer's systemd test suite has been integrated into
3707 systemd which allows easy testing of systemd builds in qemu
3708 and nspawn. (This is really awesome! Ask us for details!)
3710 * The fstab parser is now implemented as generator, not inside
3713 * systemctl will now warn you if .mount units generated from
3714 /etc/fstab are out of date due to changes in fstab that
3715 have not been read by systemd yet.
3717 * systemd is now suitable for usage in initrds. Dracut has
3718 already been updated to make use of this. With this in place
3719 initrds get a slight bit faster but primarily are much
3720 easier to introspect and debug since "systemctl status" in
3721 the host system can be used to introspect initrd services,
3722 and the journal from the initrd is kept around too.
3724 * systemd-delta has been added, a tool to explore differences
3725 between user/admin configuration and vendor defaults.
3727 * PrivateTmp= now affects both /tmp and /var/tmp.
3729 * Boot time status messages are now much prettier and feature
3730 proper english language. Booting up systemd has never been
3733 * Read-ahead pack files now include the inode number of all
3734 files to pre-cache. When the inode changes the pre-caching
3735 is not attempted. This should be nicer to deal with updated
3736 packages which might result in changes of read-ahead
3739 * We now temporaritly lower the kernel's read_ahead_kb variable
3740 when collecting read-ahead data to ensure the kernel's
3741 built-in read-ahead does not add noise to our measurements
3742 of necessary blocks to pre-cache.
3744 * There's now RequiresMountsFor= to add automatic dependencies
3745 for all mounts necessary for a specific file system path.
3747 * MountAuto= and SwapAuto= have been removed from
3748 system.conf. Mounting file systems at boot has to take place
3751 * nspawn now learned a new switch --uuid= to set the machine
3752 ID on the command line.
3754 * nspawn now learned the -b switch to automatically search
3757 * vt102 is now the default TERM for serial TTYs, upgraded from
3760 * systemd-logind now works on VT-less systems.
3762 * The build tree has been reorganized. The individual
3763 components now have directories of their own.
3765 * A new condition type ConditionPathIsReadWrite= is now available.
3767 * nspawn learned the new -C switch to create cgroups for the
3768 container in other hierarchies.
3770 * We now have support for hardware watchdogs, configurable in
3773 * The scheduled shutdown logic now has a public API.
3775 * We now mount /tmp as tmpfs by default, but this can be
3776 masked and /etc/fstab can override it.
3778 * Since udisks does not make use of /media anymore we are not
3779 mounting a tmpfs on it anymore.
3781 * journalctl gained a new --local switch to only interleave
3782 locally generated journal files.
3784 * We can now load the IMA policy at boot automatically.
3786 * The GTK tools have been split off into a systemd-ui.
3788 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3789 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3790 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3791 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3792 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3793 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3794 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3795 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3796 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3801 * This is mostly a bugfix release
3803 * Support optional initialization of the machine ID from the
3804 KVM or container configured UUID.
3806 * Support immediate reboots with "systemctl reboot -ff"
3808 * Show /etc/os-release data in systemd-analyze output
3810 * Many bugfixes for the journal, including endianness fixes and
3811 ensuring that disk space enforcement works
3813 * sd-login.h is C++ comptaible again
3815 * Extend the /etc/os-release format on request of the Debian
3818 * We now refuse non-UTF8 strings used in various configuration
3819 and unit files. This is done to ensure we do not pass invalid
3820 data over D-Bus or expose it elsewhere.
3822 * Register Mimo USB Screens as suitable for automatic seat
3825 * Read SELinux client context from journal clients in a race
3828 * Reorder configuration file lookup order. /etc now always
3829 overrides /run in order to allow the administrator to always
3830 and unconditionally override vendor supplied or
3831 automatically generated data.
3833 * The various user visible bits of the journal now have man
3834 pages. We still lack man pages for the journal API calls
3837 * We now ship all man pages in HTML format again in the
3840 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3841 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3842 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3847 * This is mostly a bugfix release
3849 * systems lacking /etc/os-release are no longer supported.
3851 * Various functionality updates to libsystemd-login.so
3853 * Track class of PAM logins to distuingish greeters from
3856 Contributions from: Kay Sievers, Lennart Poettering, Michael
3861 * This is an important bugfix release for v41.
3863 * Building man pages is now optional which should be useful
3864 for those building systemd from git but unwilling to install
3867 * Watchdog support for supervising services is now usable. In
3868 a future release support for hardware watchdogs
3869 (i.e. /dev/watchdog) will be added building on this.
3871 * Service start rate limiting is now configurable and can be
3872 turned off per service. When a start rate limit is hit a
3873 reboot can automatically be triggered.
3875 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3877 Contributions from: Benjamin Franzke, Bill Nottingham,
3878 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3879 Schmidt, Michał Górny, Piotr Drąg
3883 * The systemd binary is installed /usr/lib/systemd/systemd now;
3884 An existing /sbin/init symlink needs to be adapted with the
3887 * The code that loads kernel modules has been ported to invoke
3888 libkmod directly, instead of modprobe. This means we do not
3889 support systems with module-init-tools anymore.
3891 * Watchdog support is now already useful, but still not
3894 * A new kernel command line option systemd.setenv= is
3895 understood to set system wide environment variables
3896 dynamically at boot.
3898 * We now limit the set of capabilities of systemd-journald.
3900 * We now set SIGPIPE to ignore by default, since it only is
3901 useful in shell pipelines, and has little use in general
3902 code. This can be disabled with IgnoreSIPIPE=no in unit
3905 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3906 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3911 * This is mostly a bugfix release
3913 * We now expose the reason why a service failed in the
3914 "Result" D-Bus property.
3916 * Rudimentary service watchdog support (will be completed over
3917 the next few releases.)
3919 * When systemd forks off in order execute some service we will
3920 now immediately changes its argv[0] to reflect which process
3921 it will execute. This is useful to minimize the time window
3922 with a generic argv[0], which makes bootcharts more useful
3924 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3925 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3926 Mike Kazantsev, Ray Strode
3930 * This is mostly a test release, but incorporates many
3933 * New systemd-cgtop tool to show control groups by their
3936 * Linking against libacl for ACLs is optional again. If
3937 disabled, support tracking device access for active logins
3938 goes becomes unavailable, and so does access to the user
3939 journals by the respective users.
3941 * If a group "adm" exists, journal files are automatically
3942 owned by them, thus allow members of this group full access
3943 to the system journal as well as all user journals.
3945 * The journal now stores the SELinux context of the logging
3946 client for all entries.
3948 * Add C++ inclusion guards to all public headers
3950 * New output mode "cat" in the journal to print only text
3951 messages, without any meta data like date or time.
3953 * Include tiny X server wrapper as a temporary stop-gap to
3954 teach XOrg udev display enumeration. This is used by display
3955 managers such as gdm, and will go away as soon as XOrg
3956 learned native udev hotplugging for display devices.
3958 * Add new systemd-cat tool for executing arbitrary programs
3959 with STDERR/STDOUT connected to the journal. Can also act as
3960 BSD logger replacement, and does so by default.
3962 * Optionally store all locally generated coredumps in the
3963 journal along with meta data.
3965 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3966 writing short strings to files (for usage for /sys), and for
3967 creating symlinks, character and block device nodes.
3969 * New unit file option ControlGroupPersistent= to make cgroups
3970 persistent, following the mechanisms outlined in
3971 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3973 * Support multiple local RTCs in a sane way
3975 * No longer monopolize IO when replaying readahead data on
3976 rotating disks, since we might starve non-file-system IO to
3977 death, since fanotify() will not see accesses done by blkid,
3980 * Do not show kernel threads in systemd-cgls anymore, unless
3981 requested with new -k switch.
3983 Contributions from: Dan Horák, Kay Sievers, Lennart
3984 Poettering, Michal Schmidt
3988 * This is mostly a test release, but incorporates many
3991 * The git repository moved to:
3992 git://anongit.freedesktop.org/systemd/systemd
3993 ssh://git.freedesktop.org/git/systemd/systemd
3995 * First release with the journal
3996 http://0pointer.de/blog/projects/the-journal.html
3998 * The journal replaces both systemd-kmsg-syslogd and
3999 systemd-stdout-bridge.
4001 * New sd_pid_get_unit() API call in libsystemd-logind
4003 * Many systemadm clean-ups
4005 * Introduce remote-fs-pre.target which is ordered before all
4006 remote mounts and may be used to start services before all
4009 * Added Mageia support
4011 * Add bash completion for systemd-loginctl
4013 * Actively monitor PID file creation for daemons which exit in
4014 the parent process before having finished writing the PID
4015 file in the daemon process. Daemons which do this need to be
4016 fixed (i.e. PID file creation must have finished before the
4017 parent exits), but we now react a bit more gracefully to them.
4019 * Add colourful boot output, mimicking the well-known output
4020 of existing distributions.
4022 * New option PassCredentials= for socket units, for
4023 compatibility with a recent kernel ABI breakage.
4025 * /etc/rc.local is now hooked in via a generator binary, and
4026 thus will no longer act as synchronization point during
4029 * systemctl list-unit-files now supports --root=.
4031 * systemd-tmpfiles now understands two new commands: z, Z for
4032 relabelling files according to the SELinux database. This is
4033 useful to apply SELinux labels to specific files in /sys,
4036 * Output of SysV services is now forwarded to both the console
4037 and the journal by default, not only just the console.
4039 * New man pages for all APIs from libsystemd-login.
4041 * The build tree got reorganized and a the build system is a
4042 lot more modular allowing embedded setups to specifically
4043 select the components of systemd they are interested in.
4045 * Support for Linux systems lacking the kernel VT subsystem is
4048 * configure's --with-rootdir= got renamed to
4049 --with-rootprefix= to follow the naming used by udev and
4052 * Unless specified otherwise we will now install to /usr instead
4053 of /usr/local by default.
4055 * Processes with '@' in argv[0][0] are now excluded from the
4056 final shut-down killing spree, following the logic explained
4058 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4060 * All processes remaining in a service cgroup when we enter
4061 the START or START_PRE states are now killed with
4062 SIGKILL. That means it is no longer possible to spawn
4063 background processes from ExecStart= lines (which was never
4064 supported anyway, and bad style).
4066 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4067 reloading of units together.
4069 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4070 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4071 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4072 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4073 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek