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
77 * The default sysctl.d/ snippets will now set:
79 net.core.default_qdisc = fq_codel
81 This selects Fair Queueing Controlled Delay as the default
82 queueing discipline for network interfaces. fq_codel helps
83 fight the network bufferbloat problem. It is believed to be
84 a good default with no tuning required for most workloads.
85 Downstream distributions may override this choice. On 10Gbit
86 servers that do not do forwarding, "fq" may perform better.
87 Systems without a good clocksource should use "pfifo_fast".
91 * timedated no longer reads NTP implementation unit names from
92 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
93 implementations should add a
95 Conflicts=systemd-timesyncd.service
97 to their unit files to take over and replace systemd's NTP
98 default functionality.
100 * systemd-sysusers gained a new line type "r" for configuring
101 which UID/GID ranges to allocate system users/groups
102 from. Lines of type "u" may now add an additional column
103 that specifies the home directory for the system user to be
104 created. Also, systemd-sysusers may now optionally read user
105 information from STDIN instead of a file. This is useful for
106 invoking it from RPM preinst scriptlets that need to create
107 users before the first RPM file is installed since these
108 files might need to be owned by them. A new
109 %sysusers_create_inline RPM macro has been introduced to do
110 just that. systemd-sysusers now updates the shadow files as
111 well as the user/group databases, which should enhance
112 compatibility with certain tools like grpck.
114 * A number of bus APIs of PID 1 now optionally consult
115 PolicyKit to permit access for otherwise unprivileged
116 clients under certain conditions. Note that this currently
117 doesn't support interactive authentication yet, but this is
118 expected to be added eventually, too.
120 * /etc/machine-info now has new fields for configuring the
121 deployment environment of the machine, as well as the
122 location of the machine. hostnamectl has been updated with
123 new command to update these fields.
125 * systemd-timesyncd has been updated to automatically acquire
126 NTP server information from systemd-networkd, which might
127 have been discovered via DHCP.
129 * systemd-resolved now includes a caching DNS stub resolver
130 and a complete LLMNR name resolution implementation. A new
131 NSS module "nss-resolve" has been added which can be used
132 instead of glibc's own "nss-dns" to resolve hostnames via
133 systemd-resolved. Hostnames, addresses and arbitrary RRs may
134 be resolved via systemd-resolved D-Bus APIs. In contrast to
135 the glibc internal resolver systemd-resolved is aware of
136 multi-homed system, and keeps DNS server and caches separate
137 and per-interface. Queries are sent simultaneously on all
138 interfaces that have DNS servers configured, in order to
139 properly handle VPNs and local LANs which might resolve
140 separate sets of domain names. systemd-resolved may acquire
141 DNS server information from systemd-networkd automatically,
142 which in turn might have discovered them via DHCP. A tool
143 "systemd-resolve-host" has been added that may be used to
144 query the DNS logic in resolved. systemd-resolved implements
145 IDNA and automatically uses IDNA or UTF-8 encoding depending
146 on whether classic DNS or LLMNR is used as transport. In the
147 next releases we intend to add a DNSSEC and mDNS/DNS-SD
148 implementation to systemd-resolved.
150 * A new NSS module nss-mymachines has been added, that
151 automatically resolves the names of all local registered
152 containers to their respective IP addresses.
154 * A new client tool "networkctl" for systemd-networkd has been
155 added. It currently is entirely passive and will query
156 networking configuration from udev, rtnetlink and networkd,
157 and present it to the user in a very friendly
158 way. Eventually, we hope to extend it to become a full
159 control utility for networkd.
161 * .socket units gained a new DeferAcceptSec= setting that
162 controls the kernels' TCP_DEFER_ACCEPT sockopt for
163 TCP. Similar, support for controlling TCP keep-alive
164 settings has been added (KeepAliveTimeSec=,
165 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
166 turning off Nagle's algorithm on TCP has been added
169 * logind learned a new session type "web", for use in projects
170 like Cockpit which register web clients as PAM sessions.
172 * timer units with at least one OnCalendar= setting will now
173 be started only after timer-sync.target has been
174 reached. This way they will not elapse before the system
175 clock has been corrected by a local NTP client or
176 similar. This is particular useful on RTC-less embedded
177 machines, that come up with an invalid system clock.
179 * systemd-nspawn's --network-veth= switch should now result in
180 stable MAC addresses for both the outer and the inner side
183 * systemd-nspawn gained a new --volatile= switch for running
184 container instances with /etc or /var unpopulated.
186 * The kdbus client code has been updated to use the new Linux
187 3.17 memfd subsystem instead of the old kdbus-specific one.
189 * systemd-networkd's DHCP client and server now support
190 FORCERENEW. There are also new configuration options to
191 configure the vendor client identifier and broadcast mode
194 * systemd will no longer inform the kernel about the current
195 timezone, as this is necessarily incorrect and racy as the
196 kernel has no understanding of DST and similar
197 concepts. This hence means FAT timestamps will be always
198 considered UTC, similar to what Android is already
199 doing. Also, when the RTC is configured to the local time
200 (rather than UTC) systemd will never synchronize back to it,
201 as this might confuse Windows at a later boot.
203 * systemd-analyze gained a new command "verify" for offline
204 validation of unit files.
206 * systemd-networkd gained support for a couple of additional
207 settings for bonding networking setups. Also, the metric for
208 statically configured routes may now be configured. For
209 network interfaces where this is appropriate the peer IP
210 address may now be configured.
212 * systemd-networkd's DHCP client will no longer request
213 broadcasting by default, as this tripped up some networks.
214 For hardware where broadcast is required the feature should
215 be switched back on using RequestBroadcast=yes.
217 * systemd-networkd will now set up IPv4LL addresses (when
218 enabled) even if DHCP is configured successfully.
220 * udev will now default to respect network device names given
221 by the kernel when the kernel indicates that these are
222 predictable. This behavior can be tweaked by changing
223 NamePolicy= in the relevant .link file.
225 * A new library systemd-terminal has been added that
226 implements full TTY stream parsing and rendering. This
227 library is supposed to be used later on for implementing a
228 full userspace VT subsystem, replacing the current kernel
231 * A new tool systemd-journal-upload has been added to push
232 journal data to a remote system running
233 systemd-journal-remote.
235 * journald will no longer forward all local data to another
236 running syslog daemon. This change has been made because
237 rsyslog (which appears to be the most commonly used syslog
238 implementation these days) no longer makes use of this, and
239 instead pulls the data out of the journal on its own. Since
240 forwarding the messages to a non-existent syslog server is
241 more expensive than we assumed we have now turned this
242 off. If you run a syslog server that is not a recent rsyslog
243 version, you have to turn this option on again
244 (ForwardToSyslog= in journald.conf).
246 * journald now optionally supports the LZ4 compressor for
247 larger journal fields. This compressor should perform much
248 better than XZ which was the previous default.
250 * machinectl now shows the IP addresses of local containers,
251 if it knows them, plus the interface name of the container.
253 * A new tool "systemd-escape" has been added that makes it
254 easy to escape strings to build unit names and similar.
256 * sd_notify() messages may now include a new ERRNO= field
257 which is parsed and collected by systemd and shown among the
258 "systemctl status" output for a service.
260 * A new component "systemd-firstboot" has been added that
261 queries the most basic systemd information (timezone,
262 hostname, root password) interactively on first
263 boot. Alternatively it may also be used to provision these
264 things offline on OS images installed into directories.
266 * The default sysctl.d/ snippets will now set
268 net.ipv4.conf.default.promote_secondaries=1
270 This has the benefit of no flushing secondary IP addresses
271 when primary addresses are removed.
273 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
274 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
275 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
276 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
277 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
278 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
279 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
280 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
281 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
282 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
283 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
284 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
285 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
286 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
287 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
289 -- Berlin, 2014-08-19
293 * A new tool systemd-sysusers has been added. This tool
294 creates system users and groups in /etc/passwd and
295 /etc/group, based on static declarative system user/group
296 definitions in /usr/lib/sysusers.d/. This is useful to
297 enable factory resets and volatile systems that boot up with
298 an empty /etc directory, and thus need system users and
299 groups created during early boot. systemd now also ships
300 with two default sysusers.d/ files for the most basic
301 users and groups systemd and the core operating system
304 * A new tmpfiles snippet has been added that rebuilds the
305 essential files in /etc on boot, should they be missing.
307 * A directive for ensuring automatic clean-up of
308 /var/cache/man/ has been removed from the default
309 configuration. This line should now be shipped by the man
310 implementation. The necessary change has been made to the
311 man-db implementation. Note that you need to update your man
312 implementation to one that ships this line, otherwise no
313 automatic clean-up of /var/cache/man will take place.
315 * A new condition ConditionNeedsUpdate= has been added that
316 may conditionalize services to only run when /etc or /var
317 are "older" than the vendor operating system resources in
318 /usr. This is useful for reconstructing or updating /etc
319 after an offline update of /usr or a factory reset, on the
320 next reboot. Services that want to run once after such an
321 update or reset should use this condition and order
322 themselves before the new systemd-update-done.service, which
323 will mark the two directories as fully updated. A number of
324 service files have been added making use of this, to rebuild
325 the udev hardware database, the journald message catalog and
326 dynamic loader cache (ldconfig). The systemd-sysusers tool
327 described above also makes use of this now. With this in
328 place it is now possible to start up a minimal operating
329 system with /etc empty cleanly. For more information on the
330 concepts involved see this recent blog story:
332 http://0pointer.de/blog/projects/stateless.html
334 * A new system group "input" has been introduced, and all
335 input device nodes get this group assigned. This is useful
336 for system-level software to get access to input devices. It
337 complements what is already done for "audio" and "video".
339 * systemd-networkd learnt minimal DHCPv4 server support in
340 addition to the existing DHCPv4 client support. It also
341 learnt DHCPv6 client and IPv6 Router Solicitation client
342 support. The DHCPv4 client gained support for static routes
343 passed in from the server. Note that the [DHCPv4] section
344 known in older systemd-networkd versions has been renamed to
345 [DHCP] and is now also used by the DHCPv6 client. Existing
346 .network files using settings of this section should be
347 updated, though compatibility is maintained. Optionally, the
348 client hostname may now be sent to the DHCP server.
350 * networkd gained support for vxlan virtual networks as well
351 as tun/tap and dummy devices.
353 * networkd gained support for automatic allocation of address
354 ranges for interfaces from a system-wide pool of
355 addresses. This is useful for dynamically managing a large
356 number of interfaces with a single network configuration
357 file. In particular this is useful to easily assign
358 appropriate IP addresses to the veth links of a large number
361 * RPM macros for processing sysusers, sysctl and binfmt
362 drop-in snippets at package installation time have been
365 * The /etc/os-release file should now be placed in
366 /usr/lib/os-release. The old location is automatically
367 created as symlink. /usr/lib is the more appropriate
368 location of this file, since it shall actually describe the
369 vendor operating system shipped in /usr, and not the
370 configuration stored in /etc.
372 * .mount units gained a new boolean SloppyOptions= setting
373 that maps to mount(8)'s -s option which enables permissive
374 parsing of unknown mount options.
376 * tmpfiles learnt a new "L+" directive which creates a symlink
377 but (unlike "L") deletes a pre-existing file first, should
378 it already exist and not already be the correct
379 symlink. Similar, "b+", "c+" and "p+" directives have been
380 added as well, which create block and character devices, as
381 well as fifos in the filesystem, possibly removing any
382 pre-existing files of different types.
384 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
385 'argument' field (which so far specified the source to
386 symlink/copy the files from) is now optional. If omitted the
387 same file os copied from /usr/share/factory/ suffixed by the
388 full destination path. This is useful for populating /etc
389 with essential files, by copying them from vendor defaults
390 shipped in /usr/share/factory/etc.
392 * A new command "systemctl preset-all" has been added that
393 applies the service preset settings to all installed unit
394 files. A new switch --preset-mode= has been added that
395 controls whether only enable or only disable operations
398 * A new command "systemctl is-system-running" has been added
399 that allows checking the overall state of the system, for
400 example whether it is fully up and running.
402 * When the system boots up with an empty /etc, the equivalent
403 to "systemctl preset-all" is executed during early boot, to
404 make sure all default services are enabled after a factory
407 * systemd now contains a minimal preset file that enables the
408 most basic services systemd ships by default.
410 * Unit files' [Install] section gained a new DefaultInstance=
411 field for defining the default instance to create if a
412 template unit is enabled with no instance specified.
414 * A new passive target cryptsetup-pre.target has been added
415 that may be used by services that need to make they run and
416 finish before the first LUKS cryptographic device is set up.
418 * The /dev/loop-control and /dev/btrfs-control device nodes
419 are now owned by the "disk" group by default, opening up
420 access to this group.
422 * systemd-coredump will now automatically generate a
423 stack trace of all core dumps taking place on the system,
424 based on elfutils' libdw library. This stack trace is logged
427 * systemd-coredump may now optionally store coredumps directly
428 on disk (in /var/lib/systemd/coredump, possibly compressed),
429 instead of storing them unconditionally in the journal. This
430 mode is the new default. A new configuration file
431 /etc/systemd/coredump.conf has been added to configure this
432 and other parameters of systemd-coredump.
434 * coredumpctl gained a new "info" verb to show details about a
435 specific coredump. A new switch "-1" has also been added
436 that makes sure to only show information about the most
437 recent entry instead of all entries. Also, as the tool is
438 generally useful now the "systemd-" prefix of the binary
439 name has been removed. Distributions that want to maintain
440 compatibility with the old name should add a symlink from
441 the old name to the new name.
443 * journald's SplitMode= now defaults to "uid". This makes sure
444 that unprivileged users can access their own coredumps with
445 coredumpctl without restrictions.
447 * New kernel command line options "systemd.wants=" (for
448 pulling an additional unit during boot), "systemd.mask="
449 (for masking a specific unit for the boot), and
450 "systemd.debug-shell" (for enabling the debug shell on tty9)
451 have been added. This is implemented in the new generator
452 "systemd-debug-generator".
454 * systemd-nspawn will now by default filter a couple of
455 syscalls for containers, among them those required for
456 kernel module loading, direct x86 IO port access, swap
457 management, and kexec. Most importantly though
458 open_by_handle_at() is now prohibited for containers,
459 closing a hole similar to a recently discussed vulnerability
460 in docker regarding access to files on file hierarchies the
461 container should normally not have access to. Note that for
462 nspawn we generally make no security claims anyway (and
463 this is explicitly documented in the man page), so this is
464 just a fix for one of the most obvious problems.
466 * A new man page file-hierarchy(7) has been added that
467 contains a minimized, modernized version of the file system
468 layout systemd expects, similar in style to the FHS
469 specification or hier(5). A new tool systemd-path(1) has
470 been added to query many of these paths for the local
473 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
474 longer done. Since the directory now has a per-user size
475 limit, and is cleaned on logout this appears unnecessary,
476 in particular since this now brings the lifecycle of this
477 directory closer in line with how IPC objects are handled.
479 * systemd.pc now exports a number of additional directories,
480 including $libdir (which is useful to identify the library
481 path for the primary architecture of the system), and a
482 couple of drop-in directories.
484 * udev's predictable network interface names now use the dev_port
485 sysfs attribute, introduced in linux 3.15 instead of dev_id to
486 distinguish between ports of the same PCI function. dev_id should
487 only be used for ports using the same HW address, hence the need
490 * machined has been updated to export the OS version of a
491 container (read from /etc/os-release and
492 /usr/lib/os-release) on the bus. This is now shown in
493 "machinectl status" for a machine.
495 * A new service setting RestartForceExitStatus= has been
496 added. If configured to a set of exit signals or process
497 return values, the service will be restarted when the main
498 daemon process exits with any of them, regardless of the
501 * systemctl's -H switch for connecting to remote systemd
502 machines has been extended so that it may be used to
503 directly connect to a specific container on the
504 host. "systemctl -H root@foobar:waldi" will now connect as
505 user "root" to host "foobar", and then proceed directly to
506 the container named "waldi". Note that currently you have to
507 authenticate as user "root" for this to work, as entering
508 containers is a privileged operation.
510 Contributions from: Andreas Henriksson, Benjamin Steinwender,
511 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
512 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
513 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
514 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
515 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
516 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
517 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
518 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
519 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
520 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
521 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
523 -- Berlin, 2014-07-03
527 * As an experimental feature, udev now tries to lock the
528 disk device node (flock(LOCK_SH|LOCK_NB)) while it
529 executes events for the disk or any of its partitions.
530 Applications like partitioning programs can lock the
531 disk device node (flock(LOCK_EX)) and claim temporary
532 device ownership that way; udev will entirely skip all event
533 handling for this disk and its partitions. If the disk
534 was opened for writing, the close will trigger a partition
535 table rescan in udev's "watch" facility, and if needed
536 synthesize "change" events for the disk and all its partitions.
537 This is now unconditionally enabled, and if it turns out to
538 cause major problems, we might turn it on only for specific
539 devices, or might need to disable it entirely. Device Mapper
540 devices are excluded from this logic.
542 * We temporarily dropped the "-l" switch for fsck invocations,
543 since they collide with the flock() logic above. util-linux
544 upstream has been changed already to avoid this conflict,
545 and we will readd "-l" as soon as util-linux with this
546 change has been released.
548 * The dependency on libattr has been removed. Since a long
549 time, the extended attribute calls have moved to glibc, and
550 libattr is thus unnecessary.
552 * Virtualization detection works without priviliges now. This
553 means the systemd-detect-virt binary no longer requires
554 CAP_SYS_PTRACE file capabilities, and our daemons can run
555 with fewer privileges.
557 * systemd-networkd now runs under its own "systemd-network"
558 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
559 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
560 loses the ability to write to files owned by root this way.
562 * Similar, systemd-resolved now runs under its own
563 "systemd-resolve" user with no capabilities remaining.
565 * Similar, systemd-bus-proxyd now runs under its own
566 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
568 * systemd-networkd gained support for setting up "veth"
569 virtual ethernet devices for container connectivity, as well
570 as GRE and VTI tunnels.
572 * systemd-networkd will no longer automatically attempt to
573 manually load kernel modules necessary for certain tunnel
574 transports. Instead, it is assumed the kernel loads them
575 automatically when required. This only works correctly on
576 very new kernels. On older kernels, please consider adding
577 the kernel modules to /etc/modules-load.d/ as a work-around.
579 * The resolv.conf file systemd-resolved generates has been
580 moved to /run/systemd/resolve/. If you have a symlink from
581 /etc/resolv.conf, it might be necessary to correct it.
583 * Two new service settings, ProtectHome= and ProtectSystem=,
584 have been added. When enabled, they will make the user data
585 (such as /home) inaccessible or read-only and the system
586 (such as /usr) read-only, for specific services. This allows
587 very light-weight per-service sandboxing to avoid
588 modifications of user data or system files from
589 services. These two new switches have been enabled for all
590 of systemd's long-running services, where appropriate.
592 * Socket units gained new SocketUser= and SocketGroup=
593 settings to set the owner user and group of AF_UNIX sockets
594 and FIFOs in the file system.
596 * Socket units gained a new RemoveOnStop= setting. If enabled,
597 all FIFOS and sockets in the file system will be removed
598 when the specific socket unit is stopped.
600 * Socket units gained a new Symlinks= setting. It takes a list
601 of symlinks to create to file system sockets or FIFOs
602 created by the specific Unix sockets. This is useful to
603 manage symlinks to socket nodes with the same life-cycle as
606 * The /dev/log socket and /dev/initctl FIFO have been moved to
607 /run, and have been replaced by symlinks. This allows
608 connecting to these facilities even if PrivateDevices=yes is
609 used for a service (which makes /dev/log itself unavailable,
610 but /run is left). This also has the benefit of ensuring
611 that /dev only contains device nodes, directories and
612 symlinks, and nothing else.
614 * sd-daemon gained two new calls sd_pid_notify() and
615 sd_pid_notifyf(). They are similar to sd_notify() and
616 sd_notifyf(), but allow overriding of the source PID of
617 notification messages if permissions permit this. This is
618 useful to send notify messages on behalf of a different
619 process (for example, the parent process). The
620 systemd-notify tool has been updated to make use of this
621 when sending messages (so that notification messages now
622 originate from the shell script invoking systemd-notify and
623 not the systemd-notify process itself. This should minimize
624 a race where systemd fails to associate notification
625 messages to services when the originating process already
628 * A new "on-abnormal" setting for Restart= has been added. If
629 set, it will result in automatic restarts on all "abnormal"
630 reasons for a process to exit, which includes unclean
631 signals, core dumps, timeouts and watchdog timeouts, but
632 does not include clean and unclean exit codes or clean
633 signals. Restart=on-abnormal is an alternative for
634 Restart=on-failure for services that shall be able to
635 terminate and avoid restarts on certain errors, by
636 indicating so with an unclean exit code. Restart=on-failure
637 or Restart=on-abnormal is now the recommended setting for
638 all long-running services.
640 * If the InaccessibleDirectories= service setting points to a
641 mount point (or if there are any submounts contained within
642 it), it is now attempted to completely unmount it, to make
643 the file systems truly unavailable for the respective
646 * The ReadOnlyDirectories= service setting and
647 systemd-nspawn's --read-only parameter are now recursively
648 applied to all submounts, too.
650 * Mount units may now be created transiently via the bus APIs.
652 * The support for SysV and LSB init scripts has been removed
653 from the systemd daemon itself. Instead, it is now
654 implemented as a generator that creates native systemd units
655 from these scripts when needed. This enables us to remove a
656 substantial amount of legacy code from PID 1, following the
657 fact that many distributions only ship a very small number
658 of LSB/SysV init scripts nowadays.
660 * Priviliged Xen (dom0) domains are not considered
661 virtualization anymore by the virtualization detection
662 logic. After all, they generally have unrestricted access to
663 the hardware and usually are used to manage the unprivileged
666 * systemd-tmpfiles gained a new "C" line type, for copying
667 files or entire directories.
669 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
670 lines. So far, they have been non-globbing versions of the
671 latter, and have thus been redundant. In future, it is
672 recommended to only use "z". "m" has hence been removed
673 from the documentation, even though it stays supported.
675 * A tmpfiles snippet to recreate the most basic structure in
676 /var has been added. This is enough to create the /var/run →
677 /run symlink and create a couple of structural
678 directories. This allows systems to boot up with an empty or
679 volatile /var. Of course, while with this change, the core OS
680 now is capable with dealing with a volatile /var, not all
681 user services are ready for it. However, we hope that sooner
682 or later, many service daemons will be changed upstream so
683 that they are able to automatically create their necessary
684 directories in /var at boot, should they be missing. This is
685 the first step to allow state-less systems that only require
686 the vendor image for /usr to boot.
688 * systemd-nspawn has gained a new --tmpfs= switch to mount an
689 empty tmpfs instance to a specific directory. This is
690 particularly useful for making use of the automatic
691 reconstruction of /var (see above), by passing --tmpfs=/var.
693 * Access modes specified in tmpfiles snippets may now be
694 prefixed with "~", which indicates that they shall be masked
695 by whether the existing file or directory is currently
696 writable, readable or executable at all. Also, if specified,
697 the sgid/suid/sticky bits will be masked for all
700 * A new passive target unit "network-pre.target" has been
701 added which is useful for services that shall run before any
702 network is configured, for example firewall scripts.
704 * The "floppy" group that previously owned the /dev/fd*
705 devices is no longer used. The "disk" group is now used
706 instead. Distributions should probably deprecate usage of
709 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
710 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
711 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
712 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
713 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
714 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
717 -- Berlin, 2014-06-11
721 * A new "systemd-timesyncd" daemon has been added for
722 synchronizing the system clock across the network. It
723 implements an SNTP client. In contrast to NTP
724 implementations such as chrony or the NTP reference server,
725 this only implements a client side, and does not bother with
726 the full NTP complexity, focusing only on querying time from
727 one remote server and synchronizing the local clock to
728 it. Unless you intend to serve NTP to networked clients or
729 want to connect to local hardware clocks, this simple NTP
730 client should be more than appropriate for most
731 installations. The daemon runs with minimal privileges, and
732 has been hooked up with networkd to only operate when
733 network connectivity is available. The daemon saves the
734 current clock to disk every time a new NTP sync has been
735 acquired, and uses this to possibly correct the system clock
736 early at bootup, in order to accommodate for systems that
737 lack an RTC such as the Raspberry Pi and embedded devices,
738 and to make sure that time monotonically progresses on these
739 systems, even if it is not always correct. To make use of
740 this daemon, a new system user and group "systemd-timesync"
741 needs to be created on installation of systemd.
743 * The queue "seqnum" interface of libudev has been disabled, as
744 it was generally incompatible with device namespacing as
745 sequence numbers of devices go "missing" if the devices are
746 part of a different namespace.
748 * "systemctl list-timers" and "systemctl list-sockets" gained
749 a --recursive switch for showing units of these types also
750 for all local containers, similar in style to the already
751 supported --recursive switch for "systemctl list-units".
753 * A new RebootArgument= setting has been added for service
754 units, which may be used to specify a kernel reboot argument
755 to use when triggering reboots with StartLimitAction=.
757 * A new FailureAction= setting has been added for service
758 units which may be used to specify an operation to trigger
759 when a service fails. This works similarly to
760 StartLimitAction=, but unlike it, controls what is done
761 immediately rather than only after several attempts to
762 restart the service in question.
764 * hostnamed got updated to also expose the kernel name,
765 release, and version on the bus. This is useful for
766 executing commands like hostnamectl with the -H switch.
767 systemd-analyze makes use of this to properly display
768 details when running non-locally.
770 * The bootchart tool can now show cgroup information in the
773 * The CFS CPU quota cgroup attribute is now exposed for
774 services. The new CPUQuota= switch has been added for this
775 which takes a percentage value. Setting this will have the
776 result that a service may never get more CPU time than the
777 specified percentage, even if the machine is otherwise idle.
779 * systemd-networkd learned IPIP and SIT tunnel support.
781 * LSB init scripts exposing a dependency on $network will now
782 get a dependency on network-online.target rather than simply
783 network.target. This should bring LSB handling closer to
784 what it was on SysV systems.
786 * A new fsck.repair= kernel option has been added to control
787 how fsck shall deal with unclean file systems at boot.
789 * The (.ini) configuration file parser will now silently
790 ignore sections whose name begins with "X-". This may be
791 used to maintain application-specific extension sections in unit
794 * machined gained a new API to query the IP addresses of
795 registered containers. "machinectl status" has been updated
796 to show these addresses in its output.
798 * A new call sd_uid_get_display() has been added to the
799 sd-login APIs for querying the "primary" session of a
800 user. The "primary" session of the user is elected from the
801 user's sessions and generally a graphical session is
802 preferred over a text one.
804 * A minimal systemd-resolved daemon has been added. It
805 currently simply acts as a companion to systemd-networkd and
806 manages resolv.conf based on per-interface DNS
807 configuration, possibly supplied via DHCP. In the long run
808 we hope to extend this into a local DNSSEC enabled DNS and
811 * The systemd-networkd-wait-online tool is now enabled by
812 default. It will delay network-online.target until a network
813 connection has been configured. The tool primarily integrates
814 with networkd, but will also make a best effort to make sense
815 of network configuration performed in some other way.
817 * Two new service options StartupCPUShares= and
818 StartupBlockIOWeight= have been added that work similarly to
819 CPUShares= and BlockIOWeight= however only apply during
820 system startup. This is useful to prioritize certain services
821 differently during bootup than during normal runtime.
823 * hostnamed has been changed to prefer the statically
824 configured hostname in /etc/hostname (unless set to
825 'localhost' or empty) over any dynamic one supplied by
826 dhcp. With this change, the rules for picking the hostname
827 match more closely the rules of other configuration settings
828 where the local administrator's configuration in /etc always
829 overrides any other settings.
831 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
832 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
833 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
834 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
835 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
836 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
837 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
838 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
839 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
840 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
841 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
842 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
843 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
844 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
845 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
846 Lindskog, WaLyong Cho, Will Woods, Zbigniew
849 -- Beijing, 2014-05-28
853 * When restoring the screen brightness at boot, stay away from
854 the darkest setting or from the lowest 5% of the available
855 range, depending on which is the larger value of both. This
856 should effectively protect the user from rebooting into a
857 black screen, should the brightness have been set to minimum
860 * sd-login gained a new sd_machine_get_class() call to
861 determine the class ("vm" or "container") of a machine
862 registered with machined.
864 * sd-login gained new calls
865 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
866 to query the identity of the peer of a local AF_UNIX
867 connection. They operate similarly to their sd_pid_get_xyz()
870 * PID 1 will now maintain a system-wide system state engine
871 with the states "starting", "running", "degraded",
872 "maintenance", "stopping". These states are bound to system
873 startup, normal runtime, runtime with at least one failed
874 service, rescue/emergency mode and system shutdown. This
875 state is shown in the "systemctl status" output when no unit
876 name is passed. It is useful to determine system state, in
877 particularly when doing so for many systems or containers at
880 * A new command "list-machines" has been added to "systemctl"
881 that lists all local OS containers and shows their system
882 state (see above), if systemd runs inside of them.
884 * systemctl gained a new "-r" switch to recursively enumerate
885 units on all local containers, when used with the
886 "list-unit" command (which is the default one that is
887 executed when no parameters are specified).
889 * The GPT automatic partition discovery logic will now honour
890 two GPT partition flags: one may be set on a partition to
891 cause it to be mounted read-only, and the other may be set
892 on a partition to ignore it during automatic discovery.
894 * Two new GPT type UUIDs have been added for automatic root
895 partition discovery, for 32-bit and 64-bit ARM. This is not
896 particularly useful for discovering the root directory on
897 these architectures during bare-metal boots (since UEFI is
898 not common there), but still very useful to allow booting of
899 ARM disk images in nspawn with the -i option.
901 * MAC addresses of interfaces created with nspawn's
902 --network-interface= switch will now be generated from the
903 machine name, and thus be stable between multiple invocations
906 * logind will now automatically remove all IPC objects owned
907 by a user if she or he fully logs out. This makes sure that
908 users who are logged out cannot continue to consume IPC
909 resources. This covers SysV memory, semaphores and message
910 queues as well as POSIX shared memory and message
911 queues. Traditionally, SysV and POSIX IPC had no life-cycle
912 limits. With this functionality, that is corrected. This may
913 be turned off by using the RemoveIPC= switch of logind.conf.
915 * The systemd-machine-id-setup and tmpfiles tools gained a
916 --root= switch to operate on a specific root directory,
919 * journald can now forward logged messages to the TTYs of all
920 logged in users ("wall"). This is the default for all
921 emergency messages now.
923 * A new tool systemd-journal-remote has been added to stream
924 journal log messages across the network.
926 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
927 controller trees are mounted into it. Note that the
928 directories mounted beneath it are not read-only. This is a
929 security measure and is particularly useful because glibc
930 actually includes a search logic to pick any tmpfs it can
931 find to implement shm_open() if /dev/shm is not available
932 (which it might very well be in namespaced setups).
934 * machinectl gained a new "poweroff" command to cleanly power
935 down a local OS container.
937 * The PrivateDevices= unit file setting will now also drop the
938 CAP_MKNOD capability from the capability bound set, and
939 imply DevicePolicy=closed.
941 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
942 comprehensively on all long-running systemd services where
945 * systemd-udevd will now run in a disassociated mount
946 namespace. To mount directories from udev rules, make sure to
947 pull in mount units via SYSTEMD_WANTS properties.
949 * The kdbus support gained support for uploading policy into
950 the kernel. sd-bus gained support for creating "monitoring"
951 connections that can eavesdrop into all bus communication
952 for debugging purposes.
954 * Timestamps may now be specified in seconds since the UNIX
955 epoch Jan 1st, 1970 by specifying "@" followed by the value
958 * Native tcpwrap support in systemd has been removed. tcpwrap
959 is old code, not really maintained anymore and has serious
960 shortcomings, and better options such as firewalls
961 exist. For setups that require tcpwrap usage, please
962 consider invoking your socket-activated service via tcpd,
963 like on traditional inetd.
965 * A new system.conf configuration option
966 DefaultTimerAccuracySec= has been added that controls the
967 default AccuracySec= setting of .timer units.
969 * Timer units gained a new WakeSystem= switch. If enabled,
970 timers configured this way will cause the system to resume
971 from system suspend (if the system supports that, which most
974 * Timer units gained a new Persistent= switch. If enabled,
975 timers configured this way will save to disk when they have
976 been last triggered. This information is then used on next
977 reboot to possible execute overdue timer events, that
978 could not take place because the system was powered off.
979 This enables simple anacron-like behaviour for timer units.
981 * systemctl's "list-timers" will now also list the time a
982 timer unit was last triggered in addition to the next time
983 it will be triggered.
985 * systemd-networkd will now assign predictable IPv4LL
986 addresses to its local interfaces.
988 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
989 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
990 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
991 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
992 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
993 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
994 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
995 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
998 -- Berlin, 2014-03-25
1002 * A new unit file setting RestrictAddressFamilies= has been
1003 added to restrict which socket address families unit
1004 processes gain access to. This takes address family names
1005 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1006 attack surface of services via exotic protocol stacks. This
1007 is built on seccomp system call filters.
1009 * Two new unit file settings RuntimeDirectory= and
1010 RuntimeDirectoryMode= have been added that may be used to
1011 manage a per-daemon runtime directories below /run. This is
1012 an alternative for setting up directory permissions with
1013 tmpfiles snippets, and has the advantage that the runtime
1014 directory's lifetime is bound to the daemon runtime and that
1015 the daemon starts up with an empty directory each time. This
1016 is particularly useful when writing services that drop
1017 privileges using the User= or Group= setting.
1019 * The DeviceAllow= unit setting now supports globbing for
1020 matching against device group names.
1022 * The systemd configuration file system.conf gained new
1023 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1024 DefaultMemoryAccounting= to globally turn on/off accounting
1025 for specific resources (cgroups) for all units. These
1026 settings may still be overridden individually in each unit
1029 * systemd-gpt-auto-generator is now able to discover /srv and
1030 root partitions in addition to /home and swap partitions. It
1031 also supports LUKS-encrypted partitions now. With this in
1032 place, automatic discovery of partitions to mount following
1033 the Discoverable Partitions Specification
1034 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1035 is now a lot more complete. This allows booting without
1036 /etc/fstab and without root= on the kernel command line on
1037 systems prepared appropriately.
1039 * systemd-nspawn gained a new --image= switch which allows
1040 booting up disk images and Linux installations on any block
1041 device that follow the Discoverable Partitions Specification
1042 (see above). This means that installations made with
1043 appropriately updated installers may now be started and
1044 deployed using container managers, completely
1045 unmodified. (We hope that libvirt-lxc will add support for
1046 this feature soon, too.)
1048 * systemd-nspawn gained a new --network-macvlan= setting to
1049 set up a private macvlan interface for the
1050 container. Similarly, systemd-networkd gained a new
1051 Kind=macvlan setting in .netdev files.
1053 * systemd-networkd now supports configuring local addresses
1056 * A new tool systemd-network-wait-online has been added to
1057 synchronously wait for network connectivity using
1060 * The sd-bus.h bus API gained a new sd_bus_track object for
1061 tracking the life-cycle of bus peers. Note that sd-bus.h is
1062 still not a public API though (unless you specify
1063 --enable-kdbus on the configure command line, which however
1064 voids your warranty and you get no API stability guarantee).
1066 * The $XDG_RUNTIME_DIR runtime directories for each user are
1067 now individual tmpfs instances, which has the benefit of
1068 introducing separate pools for each user, with individual
1069 size limits, and thus making sure that unprivileged clients
1070 can no longer negatively impact the system or other users by
1071 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1072 RuntimeDirectorySize= has been introduced that allows
1073 controlling the default size limit for all users. It
1074 defaults to 10% of the available physical memory. This is no
1075 replacement for quotas on tmpfs though (which the kernel
1076 still does not support), as /dev/shm and /tmp are still
1077 shared resources used by both the system and unprivileged
1080 * logind will now automatically turn off automatic suspending
1081 on laptop lid close when more than one display is
1082 connected. This was previously expected to be implemented
1083 individually in desktop environments (such as GNOME),
1084 however has been added to logind now, in order to fix a
1085 boot-time race where a desktop environment might not have
1086 been started yet and thus not been able to take an inhibitor
1087 lock at the time where logind already suspends the system
1088 due to a closed lid.
1090 * logind will now wait at least 30s after each system
1091 suspend/resume cycle, and 3min after system boot before
1092 suspending the system due to a closed laptop lid. This
1093 should give USB docking stations and similar enough time to
1094 be probed and configured after system resume and boot in
1095 order to then act as suspend blocker.
1097 * systemd-run gained a new --property= setting which allows
1098 initialization of resource control properties (and others)
1099 for the created scope or service unit. Example: "systemd-run
1100 --property=BlockIOWeight=10 updatedb" may be used to run
1101 updatedb at a low block IO scheduling weight.
1103 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1104 now also work in --scope mode.
1106 * When systemd is compiled with kdbus support, basic support
1107 for enforced policies is now in place. (Note that enabling
1108 kdbus still voids your warranty and no API compatibility
1111 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1112 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1113 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1114 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1115 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1116 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1117 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1118 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1119 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1120 Zbigniew Jędrzejewski-Szmek
1122 -- Berlin, 2014-03-12
1126 * systemd will now relabel /dev after loading the SMACK policy
1127 according to SMACK rules.
1129 * A new unit file option AppArmorProfile= has been added to
1130 set the AppArmor profile for the processes of a unit.
1132 * A new condition check ConditionArchitecture= has been added
1133 to conditionalize units based on the system architecture, as
1134 reported by uname()'s "machine" field.
1136 * systemd-networkd now supports matching on the system
1137 virtualization, architecture, kernel command line, host name
1140 * logind is now a lot more aggressive when suspending the
1141 machine due to a closed laptop lid. Instead of acting only
1142 on the lid close action, it will continuously watch the lid
1143 status and act on it. This is useful for laptops where the
1144 power button is on the outside of the chassis so that it can
1145 be reached without opening the lid (such as the Lenovo
1146 Yoga). On those machines, logind will now immediately
1147 re-suspend the machine if the power button has been
1148 accidentally pressed while the laptop was suspended and in a
1149 backpack or similar.
1151 * logind will now watch SW_DOCK switches and inhibit reaction
1152 to the lid switch if it is pressed. This means that logind
1153 will not suspend the machine anymore if the lid is closed
1154 and the system is docked, if the laptop supports SW_DOCK
1155 notifications via the input layer. Note that ACPI docking
1156 stations do not generate this currently. Also note that this
1157 logic is usually not fully sufficient and Desktop
1158 Environments should take a lid switch inhibitor lock when an
1159 external display is connected, as systemd will not watch
1162 * nspawn will now make use of the devices cgroup controller by
1163 default, and only permit creation of and access to the usual
1164 API device nodes like /dev/null or /dev/random, as well as
1165 access to (but not creation of) the pty devices.
1167 * We will now ship a default .network file for
1168 systemd-networkd that automatically configures DHCP for
1169 network interfaces created by nspawn's --network-veth or
1170 --network-bridge= switches.
1172 * systemd will now understand the usual M, K, G, T suffixes
1173 according to SI conventions (i.e. to the base 1000) when
1174 referring to throughput and hardware metrics. It will stay
1175 with IEC conventions (i.e. to the base 1024) for software
1176 metrics, according to what is customary according to
1177 Wikipedia. We explicitly document which base applies for
1178 each configuration option.
1180 * The DeviceAllow= setting in unit files now supports a syntax
1181 to whitelist an entire group of devices node majors at once,
1182 based on the /proc/devices listing. For example, with the
1183 string "char-pts", it is now possible to whitelist all
1184 current and future pseudo-TTYs at once.
1186 * sd-event learned a new "post" event source. Event sources of
1187 this type are triggered by the dispatching of any event
1188 source of a type that is not "post". This is useful for
1189 implementing clean-up and check event sources that are
1190 triggered by other work being done in the program.
1192 * systemd-networkd is no longer statically enabled, but uses
1193 the usual [Install] sections so that it can be
1194 enabled/disabled using systemctl. It still is enabled by
1197 * When creating a veth interface pair with systemd-nspawn, the
1198 host side will now be prefixed with "vb-" if
1199 --network-bridge= is used, and with "ve-" if --network-veth
1200 is used. This way, it is easy to distinguish these cases on
1201 the host, for example to apply different configuration to
1202 them with systemd-networkd.
1204 * The compatibility libraries for libsystemd-journal.so,
1205 libsystem-id128.so, libsystemd-login.so and
1206 libsystemd-daemon.so do not make use of IFUNC
1207 anymore. Instead, we now build libsystemd.so multiple times
1208 under these alternative names. This means that the footprint
1209 is drastically increased, but given that these are
1210 transitional compatibility libraries, this should not matter
1211 much. This change has been made necessary to support the ARM
1212 platform for these compatibility libraries, as the ARM
1213 toolchain is not really at the same level as the toolchain
1214 for other architectures like x86 and does not support
1215 IFUNC. Please make sure to use --enable-compat-libs only
1216 during a transitional period!
1218 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1219 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1220 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1221 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1222 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1223 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1224 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1225 Zbigniew Jędrzejewski-Szmek
1227 -- Berlin, 2014-02-24
1231 * A new component "systemd-networkd" has been added that can
1232 be used to configure local network interfaces statically or
1233 via DHCP. It is capable of bringing up bridges, VLANs, and
1234 bonding. Currently, no hook-ups for interactive network
1235 configuration are provided. Use this for your initrd,
1236 container, embedded, or server setup if you need a simple,
1237 yet powerful, network configuration solution. This
1238 configuration subsystem is quite nifty, as it allows wildcard
1239 hotplug matching in interfaces. For example, with a single
1240 configuration snippet, you can configure that all Ethernet
1241 interfaces showing up are automatically added to a bridge,
1242 or similar. It supports link-sensing and more.
1244 * A new tool "systemd-socket-proxyd" has been added which can
1245 act as a bidirectional proxy for TCP sockets. This is
1246 useful for adding socket activation support to services that
1247 do not actually support socket activation, including virtual
1248 machines and the like.
1250 * Add a new tool to save/restore rfkill state on
1253 * Save/restore state of keyboard backlights in addition to
1254 display backlights on shutdown/boot.
1256 * udev learned a new SECLABEL{} construct to label device
1257 nodes with a specific security label when they appear. For
1258 now, only SECLABEL{selinux} is supported, but the syntax is
1259 prepared for additional security frameworks.
1261 * udev gained a new scheme to configure link-level attributes
1262 from files in /etc/systemd/network/*.link. These files can
1263 match against MAC address, device path, driver name and type,
1264 and will apply attributes like the naming policy, link speed,
1265 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1266 address assignment policy (randomized, ...).
1268 * The configuration of network interface naming rules for
1269 "permanent interface names" has changed: a new NamePolicy=
1270 setting in the [Link] section of .link files determines the
1271 priority of possible naming schemes (onboard, slot, mac,
1272 path). The default value of this setting is determined by
1273 /usr/lib/net/links/99-default.link. Old
1274 80-net-name-slot.rules udev configuration file has been
1275 removed, so local configuration overriding this file should
1276 be adapated to override 99-default.link instead.
1278 * When the User= switch is used in a unit file, also
1279 initialize $SHELL= based on the user database entry.
1281 * systemd no longer depends on libdbus. All communication is
1282 now done with sd-bus, systemd's low-level bus library
1285 * kdbus support has been added to PID 1 itself. When kdbus is
1286 enabled, this causes PID 1 to set up the system bus and
1287 enable support for a new ".busname" unit type that
1288 encapsulates bus name activation on kdbus. It works a little
1289 bit like ".socket" units, except for bus names. A new
1290 generator has been added that converts classic dbus1 service
1291 activation files automatically into native systemd .busname
1294 * sd-bus: add a light-weight vtable implementation that allows
1295 defining objects on the bus with a simple static const
1296 vtable array of its methods, signals and properties.
1298 * systemd will not generate or install static dbus
1299 introspection data anymore to /usr/share/dbus-1/interfaces,
1300 as the precise format of these files is unclear, and
1301 nothing makes use of it.
1303 * A proxy daemon is now provided to proxy clients connecting
1304 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1305 compatibility with classic D-Bus.
1307 * A bus driver implementation has been added that supports the
1308 classic D-Bus bus driver calls on kdbus, also for
1309 compatibility purposes.
1311 * A new API "sd-event.h" has been added that implements a
1312 minimal event loop API built around epoll. It provides a
1313 couple of features that direct epoll usage is lacking:
1314 prioritization of events, scales to large numbers of timer
1315 events, per-event timer slack (accuracy), system-wide
1316 coalescing of timer events, exit handlers, watchdog
1317 supervision support using systemd's sd_notify() API, child
1320 * A new API "sd-rntl.h" has been added that provides an API
1321 around the route netlink interface of the kernel, similar in
1322 style to "sd-bus.h".
1324 * A new API "sd-dhcp-client.h" has been added that provides a
1325 small DHCPv4 client-side implementation. This is used by
1328 * There is a new kernel command line option
1329 "systemd.restore_state=0|1". When set to "0", none of the
1330 systemd tools will restore saved runtime state to hardware
1331 devices. More specifically, the rfkill and backlight states
1334 * The FsckPassNo= compatibility option in mount/service units
1335 has been removed. The fstab generator will now add the
1336 necessary dependencies automatically, and does not require
1337 PID1's support for that anymore.
1339 * journalctl gained a new switch, --list-boots, that lists
1340 recent boots with their times and boot IDs.
1342 * The various tools like systemctl, loginctl, timedatectl,
1343 busctl, systemd-run, ... have gained a new switch "-M" to
1344 connect to a specific, local OS container (as direct
1345 connection, without requiring SSH). This works on any
1346 container that is registered with machined, such as those
1347 created by libvirt-lxc or nspawn.
1349 * systemd-run and systemd-analyze also gained support for "-H"
1350 to connect to remote hosts via SSH. This is particularly
1351 useful for systemd-run because it enables queuing of jobs
1352 onto remote systems.
1354 * machinectl gained a new command "login" to open a getty
1355 login in any local container. This works with any container
1356 that is registered with machined (such as those created by
1357 libvirt-lxc or nspawn), and which runs systemd inside.
1359 * machinectl gained a new "reboot" command that may be used to
1360 trigger a reboot on a specific container that is registered
1361 with machined. This works on any container that runs an init
1362 system of some kind.
1364 * systemctl gained a new "list-timers" command to print a nice
1365 listing of installed timer units with the times they elapse
1368 * Alternative reboot() parameters may now be specified on the
1369 "systemctl reboot" command line and are passed to the
1370 reboot() system call.
1372 * systemctl gained a new --job-mode= switch to configure the
1373 mode to queue a job with. This is a more generic version of
1374 --fail, --irreversible, and --ignore-dependencies, which are
1375 still available but not advertised anymore.
1377 * /etc/systemd/system.conf gained new settings to configure
1378 various default timeouts of units, as well as the default
1379 start limit interval and burst. These may still be overridden
1382 * PID1 will now export on the bus profile data of the security
1383 policy upload process (such as the SELinux policy upload to
1386 * journald: when forwarding logs to the console, include
1387 timestamps (following the setting in
1388 /sys/module/printk/parameters/time).
1390 * OnCalendar= in timer units now understands the special
1391 strings "yearly" and "annually". (Both are equivalent)
1393 * The accuracy of timer units is now configurable with the new
1394 AccuracySec= setting. It defaults to 1min.
1396 * A new dependency type JoinsNamespaceOf= has been added that
1397 allows running two services within the same /tmp and network
1398 namespace, if PrivateNetwork= or PrivateTmp= are used.
1400 * A new command "cat" has been added to systemctl. It outputs
1401 the original unit file of a unit, and concatenates the
1402 contents of additional "drop-in" unit file snippets, so that
1403 the full configuration is shown.
1405 * systemctl now supports globbing on the various "list-xyz"
1406 commands, like "list-units" or "list-sockets", as well as on
1407 those commands which take multiple unit names.
1409 * journalctl's --unit= switch gained support for globbing.
1411 * All systemd daemons now make use of the watchdog logic so
1412 that systemd automatically notices when they hang.
1414 * If the $container_ttys environment variable is set,
1415 getty-generator will automatically spawn a getty for each
1416 listed tty. This is useful for container managers to request
1417 login gettys to be spawned on as many ttys as needed.
1419 * %h, %s, %U specifier support is not available anymore when
1420 used in unit files for PID 1. This is because NSS calls are
1421 not safe from PID 1. They stay available for --user
1422 instances of systemd, and as special case for the root user.
1424 * loginctl gained a new "--no-legend" switch to turn off output
1427 * The "sd-login.h" API gained three new calls:
1428 sd_session_is_remote(), sd_session_get_remote_user(),
1429 sd_session_get_remote_host() to query information about
1432 * The udev hardware database now also carries vendor/product
1433 information of SDIO devices.
1435 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1436 determine whether watchdog notifications are requested by
1439 * Socket-activated per-connection services now include a
1440 short description of the connection parameters in the
1443 * tmpfiles gained a new "--boot" option. When this is not used,
1444 only lines where the command character is not suffixed with
1445 "!" are executed. When this option is specified, those
1446 options are executed too. This partitions tmpfiles
1447 directives into those that can be safely executed at any
1448 time, and those which should be run only at boot (for
1449 example, a line that creates /run/nologin).
1451 * A new API "sd-resolve.h" has been added which provides a simple
1452 asynchronous wrapper around glibc NSS host name resolution
1453 calls, such as getaddrinfo(). In contrast to glibc's
1454 getaddrinfo_a(), it does not use signals. In contrast to most
1455 other asynchronous name resolution libraries, this one does
1456 not reimplement DNS, but reuses NSS, so that alternate
1457 host name resolution systems continue to work, such as mDNS,
1458 LDAP, etc. This API is based on libasyncns, but it has been
1459 cleaned up for inclusion in systemd.
1461 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1462 "sd-daemon.h" are no longer found in individual libraries
1463 libsystemd-journal.so, libsystemd-login.so,
1464 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1465 merged them into a single library, libsystemd.so, which
1466 provides all symbols. The reason for this is cyclic
1467 dependencies, as these libraries tend to use each other's
1468 symbols. So far, we have managed to workaround that by linking
1469 a copy of a good part of our code into each of these
1470 libraries again and again, which, however, makes certain
1471 things hard to do, like sharing static variables. Also, it
1472 substantially increases footprint. With this change, there
1473 is only one library for the basic APIs systemd
1474 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1475 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1476 library as well, however are subject to the --enable-kdbus
1477 switch (see below). Note that "sd-dhcp-client.h" is not part
1478 of this library (this is because it only consumes, never
1479 provides, services of/to other APIs). To make the transition
1480 easy from the separate libraries to the unified one, we
1481 provide the --enable-compat-libs compile-time switch which
1482 will generate stub libraries that are compatible with the
1483 old ones but redirect all calls to the new one.
1485 * All of the kdbus logic and the new APIs "sd-bus.h",
1486 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1487 and "sd-utf8.h" are compile-time optional via the
1488 "--enable-kdbus" switch, and they are not compiled in by
1489 default. To make use of kdbus, you have to explicitly enable
1490 the switch. Note however, that neither the kernel nor the
1491 userspace API for all of this is considered stable yet. We
1492 want to maintain the freedom to still change the APIs for
1493 now. By specifying this build-time switch, you acknowledge
1494 that you are aware of the instability of the current
1497 * Also, note that while kdbus is pretty much complete,
1498 it lacks one thing: proper policy support. This means you
1499 can build a fully working system with all features; however,
1500 it will be highly insecure. Policy support will be added in
1501 one of the next releases, at the same time that we will
1502 declare the APIs stable.
1504 * When the kernel command-line argument "kdbus" is specified,
1505 systemd will automatically load the kdbus.ko kernel module. At
1506 this stage of development, it is only useful for testing kdbus
1507 and should not be used in production. Note: if "--enable-kdbus"
1508 is specified, and the kdbus.ko kernel module is available, and
1509 "kdbus" is added to the kernel command line, the entire system
1510 runs with kdbus instead of dbus-daemon, with the above mentioned
1511 problem of missing the system policy enforcement. Also a future
1512 version of kdbus.ko or a newer systemd will not be compatible with
1513 each other, and will unlikely be able to boot the machine if only
1514 one of them is updated.
1516 * systemctl gained a new "import-environment" command which
1517 uploads the caller's environment (or parts thereof) into the
1518 service manager so that it is inherited by services started
1519 by the manager. This is useful to upload variables like
1520 $DISPLAY into the user service manager.
1522 * A new PrivateDevices= switch has been added to service units
1523 which allows running a service with a namespaced /dev
1524 directory that does not contain any device nodes for
1525 physical devices. More specifically, it only includes devices
1526 such as /dev/null, /dev/urandom, and /dev/zero which are API
1529 * logind has been extended to support behaviour like VT
1530 switching on seats that do not support a VT. This makes
1531 multi-session available on seats that are not the first seat
1532 (seat0), and on systems where kernel support for VTs has
1533 been disabled at compile-time.
1535 * If a process holds a delay lock for system sleep or shutdown
1536 and fails to release it in time, we will now log its
1537 identity. This makes it easier to identify processes that
1538 cause slow suspends or power-offs.
1540 * When parsing /etc/crypttab, support for a new key-slot=
1541 option as supported by Debian is added. It allows indicating
1542 which LUKS slot to use on disk, speeding up key loading.
1544 * The sd_journald_sendv() API call has been checked and
1545 officially declared to be async-signal-safe so that it may
1546 be invoked from signal handlers for logging purposes.
1548 * Boot-time status output is now enabled automatically after a
1549 short timeout if boot does not progress, in order to give
1550 the user an indication what she or he is waiting for.
1552 * The boot-time output has been improved to show how much time
1553 remains until jobs expire.
1555 * The KillMode= switch in service units gained a new possible
1556 value "mixed". If set, and the unit is shut down, then the
1557 initial SIGTERM signal is sent only to the main daemon
1558 process, while the following SIGKILL signal is sent to
1559 all remaining processes of the service.
1561 * When a scope unit is registered, a new property "Controller"
1562 may be set. If set to a valid bus name, systemd will send a
1563 RequestStop() signal to this name when it would like to shut
1564 down the scope. This may be used to hook manager logic into
1565 the shutdown logic of scope units. Also, scope units may now
1566 be put in a special "abandoned" state, in which case the
1567 manager process which created them takes no further
1568 responsibilities for it.
1570 * When reading unit files, systemd will now verify
1571 the access mode of these files, and warn about certain
1572 suspicious combinations. This has been added to make it
1573 easier to track down packaging bugs where unit files are
1574 marked executable or world-writable.
1576 * systemd-nspawn gained a new "--setenv=" switch to set
1577 container-wide environment variables. The similar option in
1578 systemd-activate was renamed from "--environment=" to
1579 "--setenv=" for consistency.
1581 * systemd-nspawn has been updated to create a new kdbus domain
1582 for each container that is invoked, thus allowing each
1583 container to have its own set of system and user buses,
1584 independent of the host.
1586 * systemd-nspawn gained a new --drop-capability= switch to run
1587 the container with less capabilities than the default. Both
1588 --drop-capability= and --capability= now take the special
1589 string "all" for dropping or keeping all capabilities.
1591 * systemd-nspawn gained new switches for executing containers
1592 with specific SELinux labels set.
1594 * systemd-nspawn gained a new --quiet switch to not generate
1595 any additional output but the container's own console
1598 * systemd-nspawn gained a new --share-system switch to run a
1599 container without PID namespacing enabled.
1601 * systemd-nspawn gained a new --register= switch to control
1602 whether the container is registered with systemd-machined or
1603 not. This is useful for containers that do not run full
1604 OS images, but only specific apps.
1606 * systemd-nspawn gained a new --keep-unit which may be used
1607 when invoked as the only program from a service unit, and
1608 results in registration of the unit service itself in
1609 systemd-machined, instead of a newly opened scope unit.
1611 * systemd-nspawn gained a new --network-interface= switch for
1612 moving arbitrary interfaces to the container. The new
1613 --network-veth switch creates a virtual Ethernet connection
1614 between host and container. The new --network-bridge=
1615 switch then allows assigning the host side of this virtual
1616 Ethernet connection to a bridge device.
1618 * systemd-nspawn gained a new --personality= switch for
1619 setting the kernel personality for the container. This is
1620 useful when running a 32-bit container on a 64-bit host. A
1621 similar option Personality= is now also available for service
1624 * logind will now also track a "Desktop" identifier for each
1625 session which encodes the desktop environment of it. This is
1626 useful for desktop environments that want to identify
1627 multiple running sessions of itself easily.
1629 * A new SELinuxContext= setting for service units has been
1630 added that allows setting a specific SELinux execution
1631 context for a service.
1633 * Most systemd client tools will now honour $SYSTEMD_LESS for
1634 settings of the "less" pager. By default, these tools will
1635 override $LESS to allow certain operations to work, such as
1636 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1637 influence this logic.
1639 * systemd's "seccomp" hook-up has been changed to make use of
1640 the libseccomp library instead of using its own
1641 implementation. This has benefits for portability among
1644 * For usage together with SystemCallFilter=, a new
1645 SystemCallErrorNumber= setting has been introduced that
1646 allows configuration of a system error number to be returned
1647 on filtered system calls, instead of immediately killing the
1648 process. Also, SystemCallArchitectures= has been added to
1649 limit access to system calls of a particular architecture
1650 (in order to turn off support for unused secondary
1651 architectures). There is also a global
1652 SystemCallArchitectures= setting in system.conf now to turn
1653 off support for non-native system calls system-wide.
1655 * systemd requires a kernel with a working name_to_handle_at(),
1656 please see the kernel config requirements in the README file.
1658 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1659 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1660 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1661 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1662 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1663 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1664 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1665 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1666 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1667 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1668 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1669 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1670 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1671 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1672 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1673 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1674 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1675 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1676 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1677 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1678 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1679 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1680 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1681 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1683 -- Berlin, 2014-02-20
1687 * logind has gained support for facilitating privileged input
1688 and drm device access for unprivileged clients. This work is
1689 useful to allow Wayland display servers (and similar
1690 programs, such as kmscon) to run under the user's ID and
1691 access input and drm devices which are normally
1692 protected. When this is used (and the kernel is new enough)
1693 logind will "mute" IO on the file descriptors passed to
1694 Wayland as long as it is in the background and "unmute" it
1695 if it returns into the foreground. This allows secure
1696 session switching without allowing background sessions to
1697 eavesdrop on input and display data. This also introduces
1698 session switching support if VT support is turned off in the
1699 kernel, and on seats that are not seat0.
1701 * A new kernel command line option luks.options= is understood
1702 now which allows specifying LUKS options for usage for LUKS
1703 encrypted partitions specified with luks.uuid=.
1705 * tmpfiles.d(5) snippets may now use specifier expansion in
1706 path names. More specifically %m, %b, %H, %v, are now
1707 replaced by the local machine id, boot id, hostname, and
1708 kernel version number.
1710 * A new tmpfiles.d(5) command "m" has been introduced which
1711 may be used to change the owner/group/access mode of a file
1712 or directory if it exists, but do nothing if it does not.
1714 * This release removes high-level support for the
1715 MemorySoftLimit= cgroup setting. The underlying kernel
1716 cgroup attribute memory.soft_limit= is currently badly
1717 designed and likely to be removed from the kernel API in its
1718 current form, hence we should not expose it for now.
1720 * The memory.use_hierarchy cgroup attribute is now enabled for
1721 all cgroups systemd creates in the memory cgroup
1722 hierarchy. This option is likely to be come the built-in
1723 default in the kernel anyway, and the non-hierarchial mode
1724 never made much sense in the intrinsically hierarchial
1727 * A new field _SYSTEMD_SLICE= is logged along with all journal
1728 messages containing the slice a message was generated
1729 from. This is useful to allow easy per-customer filtering of
1730 logs among other things.
1732 * systemd-journald will no longer adjust the group of journal
1733 files it creates to the "systemd-journal" group. Instead we
1734 rely on the journal directory to be owned by the
1735 "systemd-journal" group, and its setgid bit set, so that the
1736 kernel file system layer will automatically enforce that
1737 journal files inherit this group assignment. The reason for
1738 this change is that we cannot allow NSS look-ups from
1739 journald which would be necessary to resolve
1740 "systemd-journal" to a numeric GID, because this might
1741 create deadlocks if NSS involves synchronous queries to
1742 other daemons (such as nscd, or sssd) which in turn are
1743 logging clients of journald and might block on it, which
1744 would then dead lock. A tmpfiles.d(5) snippet included in
1745 systemd will make sure the setgid bit and group are
1746 properly set on the journal directory if it exists on every
1747 boot. However, we recommend adjusting it manually after
1748 upgrades too (or from RPM scriptlets), so that the change is
1749 not delayed until next reboot.
1751 * Backlight and random seed files in /var/lib/ have moved into
1752 the /var/lib/systemd/ directory, in order to centralize all
1753 systemd generated files in one directory.
1755 * Boot time performance measurements (as displayed by
1756 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1757 performance information if that's available to determine how
1758 much time BIOS and boot loader initialization required. With
1759 a sufficiently new BIOS you hence no longer need to boot
1760 with Gummiboot to get access to such information.
1762 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1763 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1764 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1765 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1766 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1767 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1768 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1770 -- Berlin, 2013-10-02
1774 * The Restart= option for services now understands a new
1775 on-watchdog setting, which will restart the service
1776 automatically if the service stops sending out watchdog keep
1777 alive messages (as configured with WatchdogSec=).
1779 * The getty generator (which is responsible for bringing up a
1780 getty on configured serial consoles) will no longer only
1781 start a getty on the primary kernel console but on all
1782 others, too. This makes the order in which console= is
1783 specified on the kernel command line less important.
1785 * libsystemd-logind gained a new sd_session_get_vt() call to
1786 retrieve the VT number of a session.
1788 * If the option "tries=0" is set for an entry of /etc/crypttab
1789 its passphrase is queried indefinitely instead of any
1790 maximum number of tries.
1792 * If a service with a configure PID file terminates its PID
1793 file will now be removed automatically if it still exists
1794 afterwards. This should put an end to stale PID files.
1796 * systemd-run will now also take relative binary path names
1797 for execution and no longer insists on absolute paths.
1799 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1800 paths that are optionally prefixed with "-" to indicate that
1801 it should not be considered a failure if they do not exist.
1803 * journalctl -o (and similar commands) now understands a new
1804 output mode "short-precise", it is similar to "short" but
1805 shows timestamps with usec accuracy.
1807 * The option "discard" (as known from Debian) is now
1808 synonymous to "allow-discards" in /etc/crypttab. In fact,
1809 "discard" is preferred now (since it is easier to remember
1812 * Some licensing clean-ups were made, so that more code is now
1813 LGPL-2.1 licensed than before.
1815 * A minimal tool to save/restore the display backlight
1816 brightness across reboots has been added. It will store the
1817 backlight setting as late as possible at shutdown, and
1818 restore it as early as possible during reboot.
1820 * A logic to automatically discover and enable home and swap
1821 partitions on GPT disks has been added. With this in place
1822 /etc/fstab becomes optional for many setups as systemd can
1823 discover certain partitions located on the root disk
1824 automatically. Home partitions are recognized under their
1825 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1826 partitions are recognized under their GPT type ID
1827 0657fd6da4ab43c484e50933c84b4f4f.
1829 * systemd will no longer pass any environment from the kernel
1830 or initrd to system services. If you want to set an
1831 environment for all services, do so via the kernel command
1832 line systemd.setenv= assignment.
1834 * The systemd-sysctl tool no longer natively reads the file
1835 /etc/sysctl.conf. If desired, the file should be symlinked
1836 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1837 legacy support by a symlink rather than built-in code, it
1838 also makes the otherwise hidden order of application of the
1839 different files visible. (Note that this partly reverts to a
1840 pre-198 application order of sysctl knobs!)
1842 * The "systemctl set-log-level" and "systemctl dump" commands
1843 have been moved to systemd-analyze.
1845 * systemd-run learned the new --remain-after-exit switch,
1846 which causes the scope unit not to be cleaned up
1847 automatically after the process terminated.
1849 * tmpfiles learned a new --exclude-prefix= switch to exclude
1850 certain paths from operation.
1852 * journald will now automatically flush all messages to disk
1853 as soon as a message of the log priorities CRIT, ALERT or
1856 Contributions from: Andrew Cook, Brandon Philips, Christian
1857 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1858 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1859 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1860 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1861 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1862 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1863 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1864 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1865 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1866 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1867 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1868 William Giokas, Zbigniew Jędrzejewski-Szmek
1870 -- Berlin, 2013-09-13
1874 * The documentation has been updated to cover the various new
1875 concepts introduced with 205.
1877 * Unit files now understand the new %v specifier which
1878 resolves to the kernel version string as returned by "uname
1881 * systemctl now supports filtering the unit list output by
1882 load state, active state and sub state, using the new
1885 * "systemctl status" will now show the results of the
1886 condition checks (like ConditionPathExists= and similar) of
1887 the last start attempts of the unit. They are also logged to
1890 * "journalctl -b" may now be used to look for boot output of a
1891 specific boot. Try "journalctl -b -1" for the previous boot,
1892 but the syntax is substantially more powerful.
1894 * "journalctl --show-cursor" has been added which prints the
1895 cursor string the last shown log line. This may then be used
1896 with the new "journalctl --after-cursor=" switch to continue
1897 browsing logs from that point on.
1899 * "journalctl --force" may now be used to force regeneration
1902 * Creation of "dead" device nodes has been moved from udev
1903 into kmod and tmpfiles. Previously, udev would read the kmod
1904 databases to pre-generate dead device nodes based on meta
1905 information contained in kernel modules, so that these would
1906 be auto-loaded on access rather then at boot. As this
1907 does not really have much to do with the exposing actual
1908 kernel devices to userspace this has always been slightly
1909 alien in the udev codebase. Following the new scheme kmod
1910 will now generate a runtime snippet for tmpfiles from the
1911 module meta information and it now is tmpfiles' job to the
1912 create the nodes. This also allows overriding access and
1913 other parameters for the nodes using the usual tmpfiles
1914 facilities. As side effect this allows us to remove the
1915 CAP_SYS_MKNOD capability bit from udevd entirely.
1917 * logind's device ACLs may now be applied to these "dead"
1918 devices nodes too, thus finally allowing managed access to
1919 devices such as /dev/snd/sequencer whithout loading the
1920 backing module right-away.
1922 * A new RPM macro has been added that may be used to apply
1923 tmpfiles configuration during package installation.
1925 * systemd-detect-virt and ConditionVirtualization= now can
1926 detect User-Mode-Linux machines (UML).
1928 * journald will now implicitly log the effective capabilities
1929 set of processes in the message metadata.
1931 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1933 * The initrd interface has been simplified (more specifically,
1934 support for passing performance data via environment
1935 variables and fsck results via files in /run has been
1936 removed). These features were non-essential, and are
1937 nowadays available in a much nicer way by having systemd in
1938 the initrd serialize its state and have the hosts systemd
1939 deserialize it again.
1941 * The udev "keymap" data files and tools to apply keyboard
1942 specific mappings of scan to key codes, and force-release
1943 scan code lists have been entirely replaced by a udev
1944 "keyboard" builtin and a hwdb data file.
1946 * systemd will now honour the kernel's "quiet" command line
1947 argument also during late shutdown, resulting in a
1948 completely silent shutdown when used.
1950 * There's now an option to control the SO_REUSEPORT socket
1951 option in .socket units.
1953 * Instance units will now automatically get a per-template
1954 subslice of system.slice unless something else is explicitly
1955 configured. For example, instances of sshd@.service will now
1956 implicitly be placed in system-sshd.slice rather than
1957 system.slice as before.
1959 * Test coverage support may now be enabled at build time.
1961 Contributions from: Dave Reisner, Frederic Crozat, Harald
1962 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1963 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1964 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1965 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1966 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1967 Giokas, Zbigniew Jędrzejewski-Szmek
1969 -- Berlin, 2013-07-23
1973 * Two new unit types have been introduced:
1975 Scope units are very similar to service units, however, are
1976 created out of pre-existing processes -- instead of PID 1
1977 forking off the processes. By using scope units it is
1978 possible for system services and applications to group their
1979 own child processes (worker processes) in a powerful way
1980 which then maybe used to organize them, or kill them
1981 together, or apply resource limits on them.
1983 Slice units may be used to partition system resources in an
1984 hierarchial fashion and then assign other units to them. By
1985 default there are now three slices: system.slice (for all
1986 system services), user.slice (for all user sessions),
1987 machine.slice (for VMs and containers).
1989 Slices and scopes have been introduced primarily in
1990 context of the work to move cgroup handling to a
1991 single-writer scheme, where only PID 1
1992 creates/removes/manages cgroups.
1994 * There's a new concept of "transient" units. In contrast to
1995 normal units these units are created via an API at runtime,
1996 not from configuration from disk. More specifically this
1997 means it is now possible to run arbitrary programs as
1998 independent services, with all execution parameters passed
1999 in via bus APIs rather than read from disk. Transient units
2000 make systemd substantially more dynamic then it ever was,
2001 and useful as a general batch manager.
2003 * logind has been updated to make use of scope and slice units
2004 for managing user sessions. As a user logs in he will get
2005 his own private slice unit, to which all sessions are added
2006 as scope units. We also added support for automatically
2007 adding an instance of user@.service for the user into the
2008 slice. Effectively logind will no longer create cgroup
2009 hierarchies on its own now, it will defer entirely to PID 1
2010 for this by means of scope, service and slice units. Since
2011 user sessions this way become entities managed by PID 1
2012 the output of "systemctl" is now a lot more comprehensive.
2014 * A new mini-daemon "systemd-machined" has been added which
2015 may be used by virtualization managers to register local
2016 VMs/containers. nspawn has been updated accordingly, and
2017 libvirt will be updated shortly. machined will collect a bit
2018 of meta information about the VMs/containers, and assign
2019 them their own scope unit (see above). The collected
2020 meta-data is then made available via the "machinectl" tool,
2021 and exposed in "ps" and similar tools. machined/machinectl
2022 is compile-time optional.
2024 * As discussed earlier, the low-level cgroup configuration
2025 options ControlGroup=, ControlGroupModify=,
2026 ControlGroupPersistent=, ControlGroupAttribute= have been
2027 removed. Please use high-level attribute settings instead as
2028 well as slice units.
2030 * A new bus call SetUnitProperties() has been added to alter
2031 various runtime parameters of a unit. This is primarily
2032 useful to alter cgroup parameters dynamically in a nice way,
2033 but will be extended later on to make more properties
2034 modifiable at runtime. systemctl gained a new set-properties
2035 command that wraps this call.
2037 * A new tool "systemd-run" has been added which can be used to
2038 run arbitrary command lines as transient services or scopes,
2039 while configuring a number of settings via the command
2040 line. This tool is currently very basic, however already
2041 very useful. We plan to extend this tool to even allow
2042 queuing of execution jobs with time triggers from the
2043 command line, similar in fashion to "at".
2045 * nspawn will now inform the user explicitly that kernels with
2046 audit enabled break containers, and suggest the user to turn
2049 * Support for detecting the IMA and AppArmor security
2050 frameworks with ConditionSecurity= has been added.
2052 * journalctl gained a new "-k" switch for showing only kernel
2053 messages, mimicking dmesg output; in addition to "--user"
2054 and "--system" switches for showing only user's own logs
2057 * systemd-delta can now show information about drop-in
2058 snippets extending unit files.
2060 * libsystemd-bus has been substantially updated but is still
2061 not available as public API.
2063 * systemd will now look for the "debug" argument on the kernel
2064 command line and enable debug logging, similar to what
2065 "systemd.log_level=debug" already did before.
2067 * "systemctl set-default", "systemctl get-default" has been
2068 added to configure the default.target symlink, which
2069 controls what to boot into by default.
2071 * "systemctl set-log-level" has been added as a convenient
2072 way to raise and lower systemd logging threshold.
2074 * "systemd-analyze plot" will now show the time the various
2075 generators needed for execution, as well as information
2076 about the unit file loading.
2078 * libsystemd-journal gained a new sd_journal_open_files() call
2079 for opening specific journal files. journactl also gained a
2080 new switch to expose this new functionality. Previously we
2081 only supported opening all files from a directory, or all
2082 files from the system, as opening individual files only is
2083 racy due to journal file rotation.
2085 * systemd gained the new DefaultEnvironment= setting in
2086 /etc/systemd/system.conf to set environment variables for
2089 * If a privileged process logs a journal message with the
2090 OBJECT_PID= field set, then journald will automatically
2091 augment this with additional OBJECT_UID=, OBJECT_GID=,
2092 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2093 system services want to log events about specific client
2094 processes. journactl/systemctl has been updated to make use
2095 of this information if all log messages regarding a specific
2098 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2099 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2100 Reisner, David Coppa, David King, David Strauss, Eelco
2101 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2102 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2103 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2104 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2105 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2106 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2107 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2108 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2109 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2110 Łukasz Stelmach, 장동준
2114 * The Python bindings gained some minimal support for the APIs
2115 exposed by libsystemd-logind.
2117 * ConditionSecurity= gained support for detecting SMACK. Since
2118 this condition already supports SELinux and AppArmor we only
2119 miss IMA for this. Patches welcome!
2121 Contributions from: Karol Lewandowski, Lennart Poettering,
2122 Zbigniew Jędrzejewski-Szmek
2126 * systemd-nspawn will now create /etc/resolv.conf if
2127 necessary, before bind-mounting the host's file onto it.
2129 * systemd-nspawn will now store meta information about a
2130 container on the container's cgroup as extended attribute
2131 fields, including the root directory.
2133 * The cgroup hierarchy has been reworked in many ways. All
2134 objects any of the components systemd creates in the cgroup
2135 tree are now suffixed. More specifically, user sessions are
2136 now placed in cgroups suffixed with ".session", users in
2137 cgroups suffixed with ".user", and nspawn containers in
2138 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2139 names are now escaped in a simple scheme to avoid collision
2140 of userspace object names with kernel filenames. This work
2141 is preparation for making these objects relocatable in the
2142 cgroup tree, in order to allow easy resource partitioning of
2143 these objects without causing naming conflicts.
2145 * systemctl list-dependencies gained the new switches
2146 --plain, --reverse, --after and --before.
2148 * systemd-inhibit now shows the process name of processes that
2149 have taken an inhibitor lock.
2151 * nss-myhostname will now also resolve "localhost"
2152 implicitly. This makes /etc/hosts an optional file and
2153 nicely handles that on IPv6 ::1 maps to both "localhost" and
2156 * libsystemd-logind.so gained a new call
2157 sd_get_machine_names() to enumerate running containers and
2158 VMs (currently only supported by very new libvirt and
2159 nspawn). sd_login_monitor can now be used to watch
2160 VMs/containers coming and going.
2162 * .include is not allowed recursively anymore, and only in
2163 unit files. Usually it is better to use drop-in snippets in
2164 .d/*.conf anyway, as introduced with systemd 198.
2166 * systemd-analyze gained a new "critical-chain" command that
2167 determines the slowest chain of units run during system
2168 boot-up. It is very useful for tracking down where
2169 optimizing boot time is the most beneficial.
2171 * systemd will no longer allow manipulating service paths in
2172 the name=systemd:/system cgroup tree using ControlGroup= in
2173 units. (But is still fine with it in all other dirs.)
2175 * There's a new systemd-nspawn@.service service file that may
2176 be used to easily run nspawn containers as system
2177 services. With the container's root directory in
2178 /var/lib/container/foobar it is now sufficient to run
2179 "systemctl start systemd-nspawn@foobar.service" to boot it.
2181 * systemd-cgls gained a new parameter "--machine" to list only
2182 the processes within a certain container.
2184 * ConditionSecurity= now can check for "apparmor". We still
2185 are lacking checks for SMACK and IMA for this condition
2186 check though. Patches welcome!
2188 * A new configuration file /etc/systemd/sleep.conf has been
2189 added that may be used to configure which kernel operation
2190 systemd is supposed to execute when "suspend", "hibernate"
2191 or "hybrid-sleep" is requested. This makes the new kernel
2192 "freeze" state accessible to the user.
2194 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2195 the passed argument if applicable.
2197 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2198 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2199 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2200 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2201 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2202 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2203 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2208 * The output of 'systemctl list-jobs' got some polishing. The
2209 '--type=' argument may now be passed more than once. A new
2210 command 'systemctl list-sockets' has been added which shows
2211 a list of kernel sockets systemd is listening on with the
2212 socket units they belong to, plus the units these socket
2215 * The experimental libsystemd-bus library got substantial
2216 updates to work in conjunction with the (also experimental)
2217 kdbus kernel project. It works well enough to exchange
2218 messages with some sophistication. Note that kdbus is not
2219 ready yet, and the library is mostly an elaborate test case
2220 for now, and not installable.
2222 * systemd gained a new unit 'systemd-static-nodes.service'
2223 that generates static device nodes earlier during boot, and
2224 can run in conjunction with udev.
2226 * libsystemd-login gained a new call sd_pid_get_user_unit()
2227 to retrieve the user systemd unit a process is running
2228 in. This is useful for systems where systemd is used as
2231 * systemd-nspawn now places all containers in the new /machine
2232 top-level cgroup directory in the name=systemd
2233 hierarchy. libvirt will soon do the same, so that we get a
2234 uniform separation of /system, /user and /machine for system
2235 services, user processes and containers/virtual
2236 machines. This new cgroup hierarchy is also useful to stick
2237 stable names to specific container instances, which can be
2238 recognized later this way (this name may be controlled
2239 via systemd-nspawn's new -M switch). libsystemd-login also
2240 gained a new call sd_pid_get_machine_name() to retrieve the
2241 name of the container/VM a specific process belongs to.
2243 * bootchart can now store its data in the journal.
2245 * libsystemd-journal gained a new call
2246 sd_journal_add_conjunction() for AND expressions to the
2247 matching logic. This can be used to express more complex
2248 logical expressions.
2250 * journactl can now take multiple --unit= and --user-unit=
2253 * The cryptsetup logic now understands the "luks.key=" kernel
2254 command line switch for specifying a file to read the
2255 decryption key from. Also, if a configured key file is not
2256 found the tool will now automatically fall back to prompting
2259 * Python systemd.journal module was updated to wrap recently
2260 added functions from libsystemd-journal. The interface was
2261 changed to bring the low level interface in s.j._Reader
2262 closer to the C API, and the high level interface in
2263 s.j.Reader was updated to wrap and convert all data about
2266 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2267 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2268 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2269 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2270 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2271 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2275 * journalctl --update-catalog now understands a new --root=
2276 option to operate on catalogs found in a different root
2279 * During shutdown after systemd has terminated all running
2280 services a final killing loop kills all remaining left-over
2281 processes. We will now print the name of these processes
2282 when we send SIGKILL to them, since this usually indicates a
2285 * If /etc/crypttab refers to password files stored on
2286 configured mount points automatic dependencies will now be
2287 generated to ensure the specific mount is established first
2288 before the key file is attempted to be read.
2290 * 'systemctl status' will now show information about the
2291 network sockets a socket unit is listening on.
2293 * 'systemctl status' will also shown information about any
2294 drop-in configuration file for units. (Drop-In configuration
2295 files in this context are files such as
2296 /etc/systemd/systemd/foobar.service.d/*.conf)
2298 * systemd-cgtop now optionally shows summed up CPU times of
2299 cgroups. Press '%' while running cgtop to switch between
2300 percentage and absolute mode. This is useful to determine
2301 which cgroups use up the most CPU time over the entire
2302 runtime of the system. systemd-cgtop has also been updated
2303 to be 'pipeable' for processing with further shell tools.
2305 * 'hostnamectl set-hostname' will now allow setting of FQDN
2308 * The formatting and parsing of time span values has been
2309 changed. The parser now understands fractional expressions
2310 such as "5.5h". The formatter will now output fractional
2311 expressions for all time spans under 1min, i.e. "5.123456s"
2312 rather than "5s 123ms 456us". For time spans under 1s
2313 millisecond values are shown, for those under 1ms
2314 microsecond values are shown. This should greatly improve
2315 all time-related output of systemd.
2317 * libsystemd-login and libsystemd-journal gained new
2318 functions for querying the poll() events mask and poll()
2319 timeout value for integration into arbitrary event
2322 * localectl gained the ability to list available X11 keymaps
2323 (models, layouts, variants, options).
2325 * 'systemd-analyze dot' gained the ability to filter for
2326 specific units via shell-style globs, to create smaller,
2327 more useful graphs. I.e. it is now possible to create simple
2328 graphs of all the dependencies between only target units, or
2329 of all units that Avahi has dependencies with.
2331 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2332 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2333 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2334 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2335 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2336 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2337 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2341 * The boot-time readahead implementation for rotating media
2342 will now read the read-ahead data in multiple passes which
2343 consist of all read requests made in equidistant time
2344 intervals. This means instead of strictly reading read-ahead
2345 data in its physical order on disk we now try to find a
2346 middle ground between physical and access time order.
2348 * /etc/os-release files gained a new BUILD_ID= field for usage
2349 on operating systems that provide continuous builds of OS
2352 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2353 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2354 William Douglas, Zbigniew Jędrzejewski-Szmek
2358 * systemd-python gained an API exposing libsystemd-daemon.
2360 * The SMACK setup logic gained support for uploading CIPSO
2363 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2364 ReadOnlyDirectories= and InaccessibleDirectories= has
2365 changed. The private /tmp and /var/tmp directories are now
2366 shared by all processes of a service (which means
2367 ExecStartPre= may now leave data in /tmp that ExecStart= of
2368 the same service can still access). When a service is
2369 stopped its temporary directories are immediately deleted
2370 (normal clean-up with tmpfiles is still done in addition to
2373 * By default, systemd will now set a couple of sysctl
2374 variables in the kernel: the safe sysrq options are turned
2375 on, IP route verification is turned on, and source routing
2376 disabled. The recently added hardlink and softlink
2377 protection of the kernel is turned on. These settings should
2378 be reasonably safe, and good defaults for all new systems.
2380 * The predictable network naming logic may now be turned off
2381 with a new kernel command line switch: net.ifnames=0.
2383 * A new libsystemd-bus module has been added that implements a
2384 pretty complete D-Bus client library. For details see:
2386 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2388 * journald will now explicitly flush the journal files to disk
2389 at the latest 5min after each write. The file will then also
2390 be marked offline until the next write. This should increase
2391 reliability in case of a crash. The synchronization delay
2392 can be configured via SyncIntervalSec= in journald.conf.
2394 * There's a new remote-fs-setup.target unit that can be used
2395 to pull in specific services when at least one remote file
2396 system is to be mounted.
2398 * There are new targets timers.target and paths.target as
2399 canonical targets to pull user timer and path units in
2400 from. This complements sockets.target with a similar
2401 purpose for socket units.
2403 * libudev gained a new call udev_device_set_attribute_value()
2404 to set sysfs attributes of a device.
2406 * The udev daemon now sets the default number of worker
2407 processes executed in parallel based on the number of available
2408 CPUs instead of the amount of available RAM. This is supposed
2409 to provide a more reliable default and limit a too aggressive
2410 paralellism for setups with 1000s of devices connected.
2412 Contributions from: Auke Kok, Colin Walters, Cristian
2413 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2414 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2415 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2416 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2417 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2418 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2419 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2420 Zbigniew Jędrzejewski-Szmek
2424 * Configuration of unit files may now be extended via drop-in
2425 files without having to edit/override the unit files
2426 themselves. More specifically, if the administrator wants to
2427 change one value for a service file foobar.service he can
2428 now do so by dropping in a configuration snippet into
2429 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2430 will load all these snippets and apply them on top of the
2431 main unit configuration file, possibly extending or
2432 overriding its settings. Using these drop-in snippets is
2433 generally nicer than the two earlier options for changing
2434 unit files locally: copying the files from
2435 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2436 them there; or creating a new file in /etc/systemd/system/
2437 that incorporates the original one via ".include". Drop-in
2438 snippets into these .d/ directories can be placed in any
2439 directory systemd looks for units in, and the usual
2440 overriding semantics between /usr/lib, /etc and /run apply
2443 * Most unit file settings which take lists of items can now be
2444 reset by assigning the empty string to them. For example,
2445 normally, settings such as Environment=FOO=BAR append a new
2446 environment variable assignment to the environment block,
2447 each time they are used. By assigning Environment= the empty
2448 string the environment block can be reset to empty. This is
2449 particularly useful with the .d/*.conf drop-in snippets
2450 mentioned above, since this adds the ability to reset list
2451 settings from vendor unit files via these drop-ins.
2453 * systemctl gained a new "list-dependencies" command for
2454 listing the dependencies of a unit recursively.
2456 * Inhibitors are now honored and listed by "systemctl
2457 suspend", "systemctl poweroff" (and similar) too, not only
2458 GNOME. These commands will also list active sessions by
2461 * Resource limits (as exposed by the various control group
2462 controllers) can now be controlled dynamically at runtime
2463 for all units. More specifically, you can now use a command
2464 like "systemctl set-cgroup-attr foobar.service cpu.shares
2465 2000" to alter the CPU shares a specific service gets. These
2466 settings are stored persistently on disk, and thus allow the
2467 administrator to easily adjust the resource usage of
2468 services with a few simple commands. This dynamic resource
2469 management logic is also available to other programs via the
2470 bus. Almost any kernel cgroup attribute and controller is
2473 * systemd-vconsole-setup will now copy all font settings to
2474 all allocated VTs, where it previously applied them only to
2477 * libsystemd-login gained the new sd_session_get_tty() API
2480 * This release drops support for a few legacy or
2481 distribution-specific LSB facility names when parsing init
2482 scripts: $x-display-manager, $mail-transfer-agent,
2483 $mail-transport-agent, $mail-transfer-agent, $smtp,
2484 $null. Also, the mail-transfer-agent.target unit backing
2485 this has been removed. Distributions which want to retain
2486 compatibility with this should carry the burden for
2487 supporting this themselves and patch support for these back
2488 in, if they really need to. Also, the facilities $syslog and
2489 $local_fs are now ignored, since systemd does not support
2490 early-boot LSB init scripts anymore, and these facilities
2491 are implied anyway for normal services. syslog.target has
2494 * There are new bus calls on PID1's Manager object for
2495 cancelling jobs, and removing snapshot units. Previously,
2496 both calls were only available on the Job and Snapshot
2499 * systemd-journal-gatewayd gained SSL support.
2501 * The various "environment" files, such as /etc/locale.conf
2502 now support continuation lines with a backslash ("\") as
2503 last character in the line, similarly in style (but different)
2504 to how this is supported in shells.
2506 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2507 now implicitly appended to every log entry logged. systemctl
2508 has been updated to filter by this field when operating on a
2509 user systemd instance.
2511 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2512 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2513 the container. This makes it easier to boot unmodified
2514 Fedora systems in a container, which however still requires
2515 audit=0 to be passed on the kernel command line. Auditing in
2516 kernel and userspace is unfortunately still too broken in
2517 context of containers, hence we recommend compiling it out
2518 of the kernel or using audit=0. Hopefully this will be fixed
2519 one day for good in the kernel.
2521 * nspawn gained the new --bind= and --bind-ro= parameters to
2522 bind mount specific directories from the host into the
2525 * nspawn will now mount its own devpts file system instance
2526 into the container, in order not to leak pty devices from
2527 the host into the container.
2529 * systemd will now read the firmware boot time performance
2530 information from the EFI variables, if the used boot loader
2531 supports this, and takes it into account for boot performance
2532 analysis via "systemd-analyze". This is currently supported
2533 only in conjunction with Gummiboot, but could be supported
2534 by other boot loaders too. For details see:
2536 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2538 * A new generator has been added that automatically mounts the
2539 EFI System Partition (ESP) to /boot, if that directory
2540 exists, is empty, and no other file system has been
2541 configured to be mounted there.
2543 * logind will now send out PrepareForSleep(false) out
2544 unconditionally, after coming back from suspend. This may be
2545 used by applications as asynchronous notification for
2546 system resume events.
2548 * "systemctl unlock-sessions" has been added, that allows
2549 unlocking the screens of all user sessions at once, similar
2550 to how "systemctl lock-sessions" already locked all users
2551 sessions. This is backed by a new D-Bus call UnlockSessions().
2553 * "loginctl seat-status" will now show the master device of a
2554 seat. (i.e. the device of a seat that needs to be around for
2555 the seat to be considered available, usually the graphics
2558 * tmpfiles gained a new "X" line type, that allows
2559 configuration of files and directories (with wildcards) that
2560 shall be excluded from automatic cleanup ("aging").
2562 * udev default rules set the device node permissions now only
2563 at "add" events, and do not change them any longer with a
2564 later "change" event.
2566 * The log messages for lid events and power/sleep keypresses
2567 now carry a message ID.
2569 * We now have a substantially larger unit test suite, but this
2570 continues to be work in progress.
2572 * udevadm hwdb gained a new --root= parameter to change the
2573 root directory to operate relative to.
2575 * logind will now issue a background sync() request to the kernel
2576 early at shutdown, so that dirty buffers are flushed to disk early
2577 instead of at the last moment, in order to optimize shutdown
2580 * A new bootctl tool has been added that is an interface for
2581 certain boot loader operations. This is currently a preview
2582 and is likely to be extended into a small mechanism daemon
2583 like timedated, localed, hostnamed, and can be used by
2584 graphical UIs to enumerate available boot options, and
2585 request boot into firmware operations.
2587 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2588 the rest of the package. It also has been updated to work
2589 correctly in initrds.
2591 * Policykit previously has been runtime optional, and is now
2592 also compile time optional via a configure switch.
2594 * systemd-analyze has been reimplemented in C. Also "systemctl
2595 dot" has moved into systemd-analyze.
2597 * "systemctl status" with no further parameters will now print
2598 the status of all active or failed units.
2600 * Operations such as "systemctl start" can now be executed
2601 with a new mode "--irreversible" which may be used to queue
2602 operations that cannot accidentally be reversed by a later
2603 job queuing. This is by default used to make shutdown
2604 requests more robust.
2606 * The Python API of systemd now gained a new module for
2607 reading journal files.
2609 * A new tool kernel-install has been added that can install
2610 kernel images according to the Boot Loader Specification:
2612 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2614 * Boot time console output has been improved to provide
2615 animated boot time output for hanging jobs.
2617 * A new tool systemd-activate has been added which can be used
2618 to test socket activation with, directly from the command
2619 line. This should make it much easier to test and debug
2620 socket activation in daemons.
2622 * journalctl gained a new "--reverse" (or -r) option to show
2623 journal output in reverse order (i.e. newest line first).
2625 * journalctl gained a new "--pager-end" (or -e) option to jump
2626 to immediately jump to the end of the journal in the
2627 pager. This is only supported in conjunction with "less".
2629 * journalctl gained a new "--user-unit=" option, that works
2630 similarly to "--unit=" but filters for user units rather than
2633 * A number of unit files to ease adoption of systemd in
2634 initrds has been added. This moves some minimal logic from
2635 the various initrd implementations into systemd proper.
2637 * The journal files are now owned by a new group
2638 "systemd-journal", which exists specifically to allow access
2639 to the journal, and nothing else. Previously, we used the
2640 "adm" group for that, which however possibly covers more
2641 than just journal/log file access. This new group is now
2642 already used by systemd-journal-gatewayd to ensure this
2643 daemon gets access to the journal files and as little else
2644 as possible. Note that "make install" will also set FS ACLs
2645 up for /var/log/journal to give "adm" and "wheel" read
2646 access to it, in addition to "systemd-journal" which owns
2647 the journal files. We recommend that packaging scripts also
2648 add read access to "adm" + "wheel" to /var/log/journal, and
2649 all existing/future journal files. To normal users and
2650 administrators little changes, however packagers need to
2651 ensure to create the "systemd-journal" system group at
2652 package installation time.
2654 * The systemd-journal-gatewayd now runs as unprivileged user
2655 systemd-journal-gateway:systemd-journal-gateway. Packaging
2656 scripts need to create these system user/group at
2659 * timedated now exposes a new boolean property CanNTP that
2660 indicates whether a local NTP service is available or not.
2662 * systemd-detect-virt will now also detect xen PVs
2664 * The pstore file system is now mounted by default, if it is
2667 * In addition to the SELinux and IMA policies we will now also
2668 load SMACK policies at early boot.
2670 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2671 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2672 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2673 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2674 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2675 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2676 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2677 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2678 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2679 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2680 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2681 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2682 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2683 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2687 * Timer units now support calendar time events in addition to
2688 monotonic time events. That means you can now trigger a unit
2689 based on a calendar time specification such as "Thu,Fri
2690 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2691 or fifth day of any month of the year 2013, given that it is
2692 a thursday or friday. This brings timer event support
2693 considerably closer to cron's capabilities. For details on
2694 the supported calendar time specification language see
2697 * udev now supports a number of different naming policies for
2698 network interfaces for predictable names, and a combination
2699 of these policies is now the default. Please see this wiki
2700 document for details:
2702 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2704 * Auke Kok's bootchart implementation has been added to the
2705 systemd tree. It is an optional component that can graph the
2706 boot in quite some detail. It is one of the best bootchart
2707 implementations around and minimal in its code and
2710 * nss-myhostname has been integrated into the systemd source
2711 tree. nss-myhostname guarantees that the local hostname
2712 always stays resolvable via NSS. It has been a weak
2713 requirement of systemd-hostnamed since a long time, and
2714 since its code is actually trivial we decided to just
2715 include it in systemd's source tree. It can be turned off
2716 with a configure switch.
2718 * The read-ahead logic is now capable of properly detecting
2719 whether a btrfs file system is on SSD or rotating media, in
2720 order to optimize the read-ahead scheme. Previously, it was
2721 only capable of detecting this on traditional file systems
2724 * In udev, additional device properties are now read from the
2725 IAB in addition to the OUI database. Also, Bluetooth company
2726 identities are attached to the devices as well.
2728 * In service files %U may be used as specifier that is
2729 replaced by the configured user name of the service.
2731 * nspawn may now be invoked without a controlling TTY. This
2732 makes it suitable for invocation as its own service. This
2733 may be used to set up a simple containerized server system
2734 using only core OS tools.
2736 * systemd and nspawn can now accept socket file descriptors
2737 when they are started for socket activation. This enables
2738 implementation of socket activated nspawn
2739 containers. i.e. think about autospawning an entire OS image
2740 when the first SSH or HTTP connection is received. We expect
2741 that similar functionality will also be added to libvirt-lxc
2744 * journalctl will now suppress ANSI color codes when
2745 presenting log data.
2747 * systemctl will no longer show control group information for
2748 a unit if a the control group is empty anyway.
2750 * logind can now automatically suspend/hibernate/shutdown the
2753 * /etc/machine-info and hostnamed now also expose the chassis
2754 type of the system. This can be used to determine whether
2755 the local system is a laptop, desktop, handset or
2756 tablet. This information may either be configured by the
2757 user/vendor or is automatically determined from ACPI and DMI
2758 information if possible.
2760 * A number of PolicyKit actions are now bound together with
2761 "imply" rules. This should simplify creating UIs because
2762 many actions will now authenticate similar ones as well.
2764 * Unit files learnt a new condition ConditionACPower= which
2765 may be used to conditionalize a unit depending on whether an
2766 AC power source is connected or not, of whether the system
2767 is running on battery power.
2769 * systemctl gained a new "is-failed" verb that may be used in
2770 shell scripts and suchlike to check whether a specific unit
2771 is in the "failed" state.
2773 * The EnvironmentFile= setting in unit files now supports file
2774 globbing, and can hence be used to easily read a number of
2775 environment files at once.
2777 * systemd will no longer detect and recognize specific
2778 distributions. All distribution-specific #ifdeffery has been
2779 removed, systemd is now fully generic and
2780 distribution-agnostic. Effectively, not too much is lost as
2781 a lot of the code is still accessible via explicit configure
2782 switches. However, support for some distribution specific
2783 legacy configuration file formats has been dropped. We
2784 recommend distributions to simply adopt the configuration
2785 files everybody else uses now and convert the old
2786 configuration from packaging scripts. Most distributions
2787 already did that. If that's not possible or desirable,
2788 distributions are welcome to forward port the specific
2789 pieces of code locally from the git history.
2791 * When logging a message about a unit systemd will now always
2792 log the unit name in the message meta data.
2794 * localectl will now also discover system locale data that is
2795 not stored in locale archives, but directly unpacked.
2797 * logind will no longer unconditionally use framebuffer
2798 devices as seat masters, i.e. as devices that are required
2799 to be existing before a seat is considered preset. Instead,
2800 it will now look for all devices that are tagged as
2801 "seat-master" in udev. By default framebuffer devices will
2802 be marked as such, but depending on local systems other
2803 devices might be marked as well. This may be used to
2804 integrate graphics cards using closed source drivers (such
2805 as NVidia ones) more nicely into logind. Note however, that
2806 we recommend using the open source NVidia drivers instead,
2807 and no udev rules for the closed-source drivers will be
2808 shipped from us upstream.
2810 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2811 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2812 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2813 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2814 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2815 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2816 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2817 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2818 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2819 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2820 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2821 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2826 * udev gained support for loading additional device properties
2827 from an indexed database that is keyed by vendor/product IDs
2828 and similar device identifiers. For the beginning this
2829 "hwdb" is populated with data from the well-known PCI and
2830 USB database, but also includes PNP, ACPI and OID data. In
2831 the longer run this indexed database shall grow into
2832 becoming the one central database for non-essential
2833 userspace device metadata. Previously, data from the PCI/USB
2834 database was only attached to select devices, since the
2835 lookup was a relatively expensive operation due to O(n) time
2836 complexity (with n being the number of entries in the
2837 database). Since this is now O(1), we decided to add in this
2838 data for all devices where this is available, by
2839 default. Note that the indexed database needs to be rebuilt
2840 when new data files are installed. To achieve this you need
2841 to update your packaging scripts to invoke "udevadm hwdb
2842 --update" after installation of hwdb data files. For
2843 RPM-based distributions we introduced the new
2844 %udev_hwdb_update macro for this purpose.
2846 * The Journal gained support for the "Message Catalog", an
2847 indexed database to link up additional information with
2848 journal entries. For further details please check:
2850 http://www.freedesktop.org/wiki/Software/systemd/catalog
2852 The indexed message catalog database also needs to be
2853 rebuilt after installation of message catalog files. Use
2854 "journalctl --update-catalog" for this. For RPM-based
2855 distributions we introduced the %journal_catalog_update
2856 macro for this purpose.
2858 * The Python Journal bindings gained support for the standard
2859 Python logging framework.
2861 * The Journal API gained new functions for checking whether
2862 the underlying file system of a journal file is capable of
2863 properly reporting file change notifications, or whether
2864 applications that want to reflect journal changes "live"
2865 need to recheck journal files continuously in appropriate
2868 * It is now possible to set the "age" field for tmpfiles
2869 entries to 0, indicating that files matching this entry
2870 shall always be removed when the directories are cleaned up.
2872 * coredumpctl gained a new "gdb" verb which invokes gdb
2873 right-away on the selected coredump.
2875 * There's now support for "hybrid sleep" on kernels that
2876 support this, in addition to "suspend" and "hibernate". Use
2877 "systemctl hybrid-sleep" to make use of this.
2879 * logind's HandleSuspendKey= setting (and related settings)
2880 now gained support for a new "lock" setting to simply
2881 request the screen lock on all local sessions, instead of
2882 actually executing a suspend or hibernation.
2884 * systemd will now mount the EFI variables file system by
2887 * Socket units now gained support for configuration of the
2888 SMACK security label.
2890 * timedatectl will now output the time of the last and next
2891 daylight saving change.
2893 * We dropped support for various legacy and distro-specific
2894 concepts, such as insserv, early-boot SysV services
2895 (i.e. those for non-standard runlevels such as 'b' or 'S')
2896 or ArchLinux /etc/rc.conf support. We recommend the
2897 distributions who still need support this to either continue
2898 to maintain the necessary patches downstream, or find a
2899 different solution. (Talk to us if you have questions!)
2901 * Various systemd components will now bypass PolicyKit checks
2902 for root and otherwise handle properly if PolicyKit is not
2903 found to be around. This should fix most issues for
2904 PolicyKit-less systems. Quite frankly this should have been
2905 this way since day one. It is absolutely our intention to
2906 make systemd work fine on PolicyKit-less systems, and we
2907 consider it a bug if something does not work as it should if
2908 PolicyKit is not around.
2910 * For embedded systems it is now possible to build udev and
2911 systemd without blkid and/or kmod support.
2913 * "systemctl switch-root" is now capable of switching root
2914 more than once. I.e. in addition to transitions from the
2915 initrd to the host OS it is now possible to transition to
2916 further OS images from the host. This is useful to implement
2917 offline updating tools.
2919 * Various other additions have been made to the RPM macros
2920 shipped with systemd. Use %udev_rules_update() after
2921 installing new udev rules files. %_udevhwdbdir,
2922 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2923 %_sysctldir are now available which resolve to the right
2924 directories for packages to place various data files in.
2926 * journalctl gained the new --full switch (in addition to
2927 --all, to disable ellipsation for long messages.
2929 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2930 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2931 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2932 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2933 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2934 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2935 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2936 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2937 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2941 * journalctl gained new --since= and --until= switches to
2942 filter by time. It also now supports nice filtering for
2943 units via --unit=/-u.
2945 * Type=oneshot services may use ExecReload= and do the
2948 * The journal daemon now supports time-based rotation and
2949 vacuuming, in addition to the usual disk-space based
2952 * The journal will now index the available field values for
2953 each field name. This enables clients to show pretty drop
2954 downs of available match values when filtering. The bash
2955 completion of journalctl has been updated
2956 accordingly. journalctl gained a new switch -F to list all
2957 values a certain field takes in the journal database.
2959 * More service events are now written as structured messages
2960 to the journal, and made recognizable via message IDs.
2962 * The timedated, localed and hostnamed mini-services which
2963 previously only provided support for changing time, locale
2964 and hostname settings from graphical DEs such as GNOME now
2965 also have a minimal (but very useful) text-based client
2966 utility each. This is probably the nicest way to changing
2967 these settings from the command line now, especially since
2968 it lists available options and is fully integrated with bash
2971 * There's now a new tool "systemd-coredumpctl" to list and
2972 extract coredumps from the journal.
2974 * We now install a README each in /var/log/ and
2975 /etc/rc.d/init.d explaining where the system logs and init
2976 scripts went. This hopefully should help folks who go to
2977 that dirs and look into the otherwise now empty void and
2978 scratch their heads.
2980 * When user-services are invoked (by systemd --user) the
2981 $MANAGERPID env var is set to the PID of systemd.
2983 * SIGRTMIN+24 when sent to a --user instance will now result
2984 in immediate termination of systemd.
2986 * gatewayd received numerous feature additions such as a
2987 "follow" mode, for live syncing and filtering.
2989 * browse.html now allows filtering and showing detailed
2990 information on specific entries. Keyboard navigation and
2991 mouse screen support has been added.
2993 * gatewayd/journalctl now supports HTML5/JSON
2994 Server-Sent-Events as output.
2996 * The SysV init script compatibility logic will now
2997 heuristically determine whether a script supports the
2998 "reload" verb, and only then make this available as
3001 * "systemctl status --follow" has been removed, use "journalctl
3004 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3005 have been removed since they are hardly useful to be
3008 * And I'd like to take the opportunity to specifically mention
3009 Zbigniew for his great contributions. Zbigniew, you rock!
3011 Contributions from: Andrew Eikum, Christian Hesse, Colin
3012 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3013 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3014 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3015 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3016 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3017 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3021 * If /etc/vconsole.conf is non-existent or empty we will no
3022 longer load any console font or key map at boot by
3023 default. Instead the kernel defaults will be left
3024 intact. This is definitely the right thing to do, as no
3025 configuration should mean no configuration, and hard-coding
3026 font names that are different on all archs is probably a bad
3027 idea. Also, the kernel default key map and font should be
3028 good enough for most cases anyway, and mostly identical to
3029 the userspace fonts/key maps we previously overloaded them
3030 with. If distributions want to continue to default to a
3031 non-kernel font or key map they should ship a default
3032 /etc/vconsole.conf with the appropriate contents.
3034 Contributions from: Colin Walters, Daniel J Walsh, Dave
3035 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3036 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3040 * journalctl gained a new --cursor= switch to show entries
3041 starting from the specified location in the journal.
3043 * We now enforce a size limit on journal entry fields exported
3044 with "-o json" in journalctl. Fields larger than 4K will be
3045 assigned null. This can be turned off with --all.
3047 * An (optional) journal gateway daemon is now available as
3048 "systemd-journal-gatewayd.service". This service provides
3049 access to the journal via HTTP and JSON. This functionality
3050 will be used to implement live log synchronization in both
3051 pull and push modes, but has various other users too, such
3052 as easy log access for debugging of embedded devices. Right
3053 now it is already useful to retrieve the journal via HTTP:
3055 # systemctl start systemd-journal-gatewayd.service
3056 # wget http://localhost:19531/entries
3058 This will download the journal contents in a
3059 /var/log/messages compatible format. The same as JSON:
3061 # curl -H"Accept: application/json" http://localhost:19531/entries
3063 This service is also accessible via a web browser where a
3064 single static HTML5 app is served that uses the JSON logic
3065 to enable the user to do some basic browsing of the
3066 journal. This will be extended later on. Here's an example
3067 screenshot of this app in its current state:
3069 http://0pointer.de/public/journal-gatewayd
3071 Contributions from: Kay Sievers, Lennart Poettering, Robert
3072 Milasan, Tom Gundersen
3076 * The bash completion logic is now available for journalctl
3079 * We do not mount the "cpuset" controller anymore together with
3080 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3081 started if no parameters are assigned to it. "cpuset" hence
3082 broke code that assumed it it could create "cpu" groups and
3085 * journalctl -f will now subscribe to terminal size changes,
3086 and line break accordingly.
3088 Contributions from: Dave Reisner, Kay Sievers, Lennart
3089 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3093 * nspawn will now create a symlink /etc/localtime in the
3094 container environment, copying the host's timezone
3095 setting. Previously this has been done via a bind mount, but
3096 since symlinks cannot be bind mounted this has now been
3097 changed to create/update the appropriate symlink.
3099 * journalctl -n's line number argument is now optional, and
3100 will default to 10 if omitted.
3102 * journald will now log the maximum size the journal files may
3103 take up on disk. This is particularly useful if the default
3104 built-in logic of determining this parameter from the file
3105 system size is used. Use "systemctl status
3106 systemd-journald.service" to see this information.
3108 * The multi-seat X wrapper tool has been stripped down. As X
3109 is now capable of enumerating graphics devices via udev in a
3110 seat-aware way the wrapper is not strictly necessary
3111 anymore. A stripped down temporary stop-gap is still shipped
3112 until the upstream display managers have been updated to
3113 fully support the new X logic. Expect this wrapper to be
3114 removed entirely in one of the next releases.
3116 * HandleSleepKey= in logind.conf has been split up into
3117 HandleSuspendKey= and HandleHibernateKey=. The old setting
3118 is not available anymore. X11 and the kernel are
3119 distuingishing between these keys and we should too. This
3120 also means the inhibition lock for these keys has been split
3123 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3124 Poettering, Lukas Nykryn, Václav Pavlín
3128 * Whenever a unit changes state we will now log this to the
3129 journal and show along the unit's own log output in
3132 * ConditionPathIsMountPoint= can now properly detect bind
3133 mount points too. (Previously, a bind mount of one file
3134 system to another place in the same file system could not be
3135 detected as mount, since they shared struct stat's st_dev
3138 * We will now mount the cgroup controllers cpu, cpuacct,
3139 cpuset and the controllers net_cls, net_prio together by
3142 * nspawn containers will now have a virtualized boot
3143 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3144 over with a randomized ID at container initialization). This
3145 has the effect of making "journalctl -b" do the right thing
3148 * The JSON output journal serialization has been updated not
3149 to generate "endless" list objects anymore, but rather one
3150 JSON object per line. This is more in line how most JSON
3151 parsers expect JSON objects. The new output mode
3152 "json-pretty" has been added to provide similar output, but
3153 neatly aligned for readability by humans.
3155 * We dropped all explicit sync() invocations in the shutdown
3156 code. The kernel does this implicitly anyway in the kernel
3157 reboot() syscall. halt(8)'s -n option is now a compatibility
3160 * We now support virtualized reboot() in containers, as
3161 supported by newer kernels. We will fall back to exit() if
3162 CAP_SYS_REBOOT is not available to the container. Also,
3163 nspawn makes use of this now and will actually reboot the
3164 container if the containerized OS asks for that.
3166 * journalctl will only show local log output by default
3167 now. Use --merge (-m) to show remote log output, too.
3169 * libsystemd-journal gained the new sd_journal_get_usage()
3170 call to determine the current disk usage of all journal
3171 files. This is exposed in the new "journalctl --disk-usage"
3174 * journald gained a new configuration setting SplitMode= in
3175 journald.conf which may be used to control how user journals
3176 are split off. See journald.conf(5) for details.
3178 * A new condition type ConditionFileNotEmpty= has been added.
3180 * tmpfiles' "w" lines now support file globbing, to write
3181 multiple files at once.
3183 * We added Python bindings for the journal submission
3184 APIs. More Python APIs for a number of selected APIs will
3185 likely follow. Note that we intend to add native bindings
3186 only for the Python language, as we consider it common
3187 enough to deserve bindings shipped within systemd. There are
3188 various projects outside of systemd that provide bindings
3189 for languages such as PHP or Lua.
3191 * Many conditions will now resolve specifiers such as %i. In
3192 addition, PathChanged= and related directives of .path units
3193 now support specifiers as well.
3195 * There's now a new RPM macro definition for the system preset
3198 * journald will now warn if it ca not forward a message to the
3199 syslog daemon because its socket is full.
3201 * timedated will no longer write or process /etc/timezone,
3202 except on Debian. As we do not support late mounted /usr
3203 anymore /etc/localtime always being a symlink is now safe,
3204 and hence the information in /etc/timezone is not necessary
3207 * logind will now always reserve one VT for a text getty (VT6
3208 by default). Previously if more than 6 X sessions where
3209 started they took up all the VTs with auto-spawned gettys,
3210 so that no text gettys were available anymore.
3212 * udev will now automatically inform the btrfs kernel logic
3213 about btrfs RAID components showing up. This should make
3214 simple hotplug based btrfs RAID assembly work.
3216 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3217 (but not for its children which will stay at the kernel
3218 default). This should allow setups with a lot more listening
3221 * systemd will now always pass the configured timezone to the
3222 kernel at boot. timedated will do the same when the timezone
3225 * logind's inhibition logic has been updated. By default,
3226 logind will now handle the lid switch, the power and sleep
3227 keys all the time, even in graphical sessions. If DEs want
3228 to handle these events on their own they should take the new
3229 handle-power-key, handle-sleep-key and handle-lid-switch
3230 inhibitors during their runtime. A simple way to achiveve
3231 that is to invoke the DE wrapped in an invocation of:
3233 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3235 * Access to unit operations is now checked via SELinux taking
3236 the unit file label and client process label into account.
3238 * systemd will now notify the administrator in the journal
3239 when he over-mounts a non-empty directory.
3241 * There are new specifiers that are resolved in unit files,
3242 for the host name (%H), the machine ID (%m) and the boot ID
3245 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3246 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3247 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3248 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3249 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3250 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3251 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3255 * Support for reading structured kernel messages from
3256 /dev/kmsg has now been added and is enabled by default.
3258 * Support for reading kernel messages from /proc/kmsg has now
3259 been removed. If you want kernel messages in the journal
3260 make sure to run a recent kernel (>= 3.5) that supports
3261 reading structured messages from /dev/kmsg (see
3262 above). /proc/kmsg is now exclusive property of classic
3263 syslog daemons again.
3265 * The libudev API gained the new
3266 udev_device_new_from_device_id() call.
3268 * The logic for file system namespace (ReadOnlyDirectory=,
3269 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3270 require pivot_root() anymore. This means fewer temporary
3271 directories are created below /tmp for this feature.
3273 * nspawn containers will now see and receive all submounts
3274 made on the host OS below the root file system of the
3277 * Forward Secure Sealing is now supported for Journal files,
3278 which provide cryptographical sealing of journal files so
3279 that attackers cannot alter log history anymore without this
3280 being detectable. Lennart will soon post a blog story about
3281 this explaining it in more detail.
3283 * There are two new service settings RestartPreventExitStatus=
3284 and SuccessExitStatus= which allow configuration of exit
3285 status (exit code or signal) which will be excepted from the
3286 restart logic, resp. consider successful.
3288 * journalctl gained the new --verify switch that can be used
3289 to check the integrity of the structure of journal files and
3290 (if Forward Secure Sealing is enabled) the contents of
3293 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3294 and similar symlinks pre-created. This makes running shells
3295 as container init process a lot more fun.
3297 * The fstab support can now handle PARTUUID= and PARTLABEL=
3300 * A new ConditionHost= condition has been added to match
3301 against the hostname (with globs) and machine ID. This is
3302 useful for clusters where a single OS image is used to
3303 provision a large number of hosts which shall run slightly
3304 different sets of services.
3306 * Services which hit the restart limit will now be placed in a
3309 Contributions from: Bertram Poettering, Dave Reisner, Huang
3310 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3311 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3315 * When running in --user mode systemd will now become a
3316 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3317 tree a lot more organized.
3319 * A new PartOf= unit dependency type has been introduced that
3320 may be used to group services in a natural way.
3322 * "systemctl enable" may now be used to enable instances of
3325 * journalctl now prints error log levels in red, and
3326 warning/notice log levels in bright white. It also supports
3327 filtering by log level now.
3329 * cgtop gained a new -n switch (similar to top), to configure
3330 the maximum number of iterations to run for. It also gained
3331 -b, to run in batch mode (accepting no input).
3333 * The suffix ".service" may now be omitted on most systemctl
3334 command lines involving service unit names.
3336 * There's a new bus call in logind to lock all sessions, as
3337 well as a loginctl verb for it "lock-sessions".
3339 * libsystemd-logind.so gained a new call sd_journal_perror()
3340 that works similar to libc perror() but logs to the journal
3341 and encodes structured information about the error number.
3343 * /etc/crypttab entries now understand the new keyfile-size=
3346 * shutdown(8) now can send a (configurable) wall message when
3347 a shutdown is cancelled.
3349 * The mount propagation mode for the root file system will now
3350 default to "shared", which is useful to make containers work
3351 nicely out-of-the-box so that they receive new mounts from
3352 the host. This can be undone locally by running "mount
3353 --make-rprivate /" if needed.
3355 * The prefdm.service file has been removed. Distributions
3356 should maintain this unit downstream if they intend to keep
3357 it around. However, we recommend writing normal unit files
3358 for display managers instead.
3360 * Since systemd is a crucial part of the OS we will now
3361 default to a number of compiler switches that improve
3362 security (hardening) such as read-only relocations, stack
3363 protection, and suchlike.
3365 * The TimeoutSec= setting for services is now split into
3366 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3367 of individual time outs for the start and the stop phase of
3370 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3371 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3372 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3373 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3374 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3375 Gundersen, Zbigniew Jędrzejewski-Szmek
3379 * The journal and id128 C APIs are now fully documented as man
3382 * Extra safety checks have been added when transitioning from
3383 the initial RAM disk to the main system to avoid accidental
3386 * /etc/crypttab entries now understand the new keyfile-offset=
3389 * systemctl -t can now be used to filter by unit load state.
3391 * The journal C API gained the new sd_journal_wait() call to
3392 make writing synchronous journal clients easier.
3394 * journalctl gained the new -D switch to show journals from a
3397 * journalctl now displays a special marker between log
3398 messages of two different boots.
3400 * The journal is now explicitly flushed to /var via a service
3401 systemd-journal-flush.service, rather than implicitly simply
3402 by seeing /var/log/journal to be writable.
3404 * journalctl (and the journal C APIs) can now match for much
3405 more complex expressions, with alternatives and
3408 * When transitioning from the initial RAM disk to the main
3409 system we will now kill all processes in a killing spree to
3410 ensure no processes stay around by accident.
3412 * Three new specifiers may be used in unit files: %u, %h, %s
3413 resolve to the user name, user home directory resp. user
3414 shell. This is useful for running systemd user instances.
3416 * We now automatically rotate journal files if their data
3417 object hash table gets a fill level > 75%. We also size the
3418 hash table based on the configured maximum file size. This
3419 together should lower hash collisions drastically and thus
3420 speed things up a bit.
3422 * journalctl gained the new "--header" switch to introspect
3423 header data of journal files.
3425 * A new setting SystemCallFilters= has been added to services
3426 which may be used to apply blacklists or whitelists to
3427 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3429 * nspawn gained a new --link-journal= switch (and quicker: -j)
3430 to link the container journal with the host. This makes it
3431 very easy to centralize log viewing on the host for all
3432 guests while still keeping the journal files separated.
3434 * Many bugfixes and optimizations
3436 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3437 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3438 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3443 * Several tools now understand kernel command line arguments,
3444 which are only read when run in an initial RAM disk. They
3445 usually follow closely their normal counterparts, but are
3448 * There's a new tool to analyze the readahead files that are
3449 automatically generated at boot. Use:
3451 /usr/lib/systemd/systemd-readahead analyze /.readahead
3453 * We now provide an early debug shell on tty9 if this enabled. Use:
3455 systemctl enable debug-shell.service
3457 * All plymouth related units have been moved into the Plymouth
3458 package. Please make sure to upgrade your Plymouth version
3461 * systemd-tmpfiles now supports getting passed the basename of
3462 a configuration file only, in which case it will look for it
3463 in all appropriate directories automatically.
3465 * udevadm info now takes a /dev or /sys path as argument, and
3466 does the right thing. Example:
3468 udevadm info /dev/sda
3469 udevadm info /sys/class/block/sda
3471 * systemctl now prints a warning if a unit is stopped but a
3472 unit that might trigger it continues to run. Example: a
3473 service is stopped but the socket that activates it is left
3476 * "systemctl status" will now mention if the log output was
3477 shortened due to rotation since a service has been started.
3479 * The journal API now exposes functions to determine the
3480 "cutoff" times due to rotation.
3482 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3483 immediately flushing of runtime logs to /var if possible,
3484 resp. for triggering immediate rotation of the journal
3487 * It is now considered an error if a service is attempted to
3488 be stopped that is not loaded.
3490 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3492 * systemd-analyze now supports Python 3
3494 * tmpfiles now supports cleaning up directories via aging
3495 where the first level dirs are always kept around but
3496 directories beneath it automatically aged. This is enabled
3497 by prefixing the age field with '~'.
3499 * Seat objects now expose CanGraphical, CanTTY properties
3500 which is required to deal with very fast bootups where the
3501 display manager might be running before the graphics drivers
3502 completed initialization.
3504 * Seat objects now expose a State property.
3506 * We now include RPM macros for service enabling/disabling
3507 based on the preset logic. We recommend RPM based
3508 distributions to make use of these macros if possible. This
3509 makes it simpler to reuse RPM spec files across
3512 * We now make sure that the collected systemd unit name is
3513 always valid when services log to the journal via
3516 * There's a new man page kernel-command-line(7) detailing all
3517 command line options we understand.
3519 * The fstab generator may now be disabled at boot by passing
3520 fstab=0 on the kernel command line.
3522 * A new kernel command line option modules-load= is now understood
3523 to load a specific kernel module statically, early at boot.
3525 * Unit names specified on the systemctl command line are now
3526 automatically escaped as needed. Also, if file system or
3527 device paths are specified they are automatically turned
3528 into the appropriate mount or device unit names. Example:
3530 systemctl status /home
3531 systemctl status /dev/sda
3533 * The SysVConsole= configuration option has been removed from
3534 system.conf parsing.
3536 * The SysV search path is no longer exported on the D-Bus
3539 * The Names= option is been removed from unit file parsing.
3541 * There's a new man page bootup(7) detailing the boot process.
3543 * Every unit and every generator we ship with systemd now
3544 comes with full documentation. The self-explanatory boot is
3547 * A couple of services gained "systemd-" prefixes in their
3548 name if they wrap systemd code, rather than only external
3549 code. Among them fsck@.service which is now
3550 systemd-fsck@.service.
3552 * The HaveWatchdog property has been removed from the D-Bus
3555 * systemd.confirm_spawn= on the kernel command line should now
3558 * There's a new man page crypttab(5) which details all options
3559 we actually understand.
3561 * systemd-nspawn gained a new --capability= switch to pass
3562 additional capabilities to the container.
3564 * timedated will now read known NTP implementation unit names
3565 from /usr/lib/systemd/ntp-units.d/*.list,
3566 systemd-timedated-ntp.target has been removed.
3568 * journalctl gained a new switch "-b" that lists log data of
3569 the current boot only.
3571 * The notify socket is in the abstract namespace again, in
3572 order to support daemons which chroot() at start-up.
3574 * There is a new Storage= configuration option for journald
3575 which allows configuration of where log data should go. This
3576 also provides a way to disable journal logging entirely, so
3577 that data collected is only forwarded to the console, the
3578 kernel log buffer or another syslog implementation.
3580 * Many bugfixes and optimizations
3582 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3583 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3584 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3585 Shawn Landden, Tom Gundersen
3589 * "systemctl help <unit>" now shows the man page if one is
3592 * Several new man pages have been added.
3594 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3595 MaxLevelConsole= can now be specified in
3596 journald.conf. These options allow reducing the amount of
3597 data stored on disk or forwarded by the log level.
3599 * TimerSlackNSec= can now be specified in system.conf for
3600 PID1. This allows system-wide power savings.
3602 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3603 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3608 * logind is now capable of (optionally) handling power and
3609 sleep keys as well as the lid switch.
3611 * journalctl now understands the syntax "journalctl
3612 /usr/bin/avahi-daemon" to get all log output of a specific
3615 * CapabilityBoundingSet= in system.conf now also influences
3616 the capability bound set of usermode helpers of the kernel.
3618 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3619 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3620 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3621 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3625 * Note that we skipped 139 releases here in order to set the
3626 new version to something that is greater than both udev's
3627 and systemd's most recent version number.
3629 * udev: all udev sources are merged into the systemd source tree now.
3630 All future udev development will happen in the systemd tree. It
3631 is still fully supported to use the udev daemon and tools without
3632 systemd running, like in initramfs or other init systems. Building
3633 udev though, will require the *build* of the systemd tree, but
3634 udev can be properly *run* without systemd.
3636 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3637 should be used to create dead device nodes as workarounds for broken
3640 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3641 no longer supported. udev_monitor_new_from_netlink() needs to be
3642 used to subscribe to events.
3644 * udev: when udevd is started by systemd, processes which are left
3645 behind by forking them off of udev rules, are unconditionally cleaned
3646 up and killed now after the event handling has finished. Services or
3647 daemons must be started as systemd services. Services can be
3648 pulled-in by udev to get started, but they can no longer be directly
3649 forked by udev rules.
3651 * udev: the daemon binary is called systemd-udevd now and installed
3652 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3653 to adapt to that, create symlink, or rename the binary after building
3656 * libudev no longer provides these symbols:
3657 udev_monitor_from_socket()
3658 udev_queue_get_failed_list_entry()
3659 udev_get_{dev,sys,run}_path()
3660 The versions number was bumped and symbol versioning introduced.
3662 * systemd-loginctl and systemd-journalctl have been renamed
3663 to loginctl and journalctl to match systemctl.
3665 * The config files: /etc/systemd/systemd-logind.conf and
3666 /etc/systemd/systemd-journald.conf have been renamed to
3667 logind.conf and journald.conf. Package updates should rename
3668 the files to the new names on upgrade.
3670 * For almost all files the license is now LGPL2.1+, changed
3671 from the previous GPL2.0+. Exceptions are some minor stuff
3672 of udev (which will be changed to LGPL2.1 eventually, too),
3673 and the MIT licensed sd-daemon.[ch] library that is suitable
3674 to be used as drop-in files.
3676 * systemd and logind now handle system sleep states, in
3677 particular suspending and hibernating.
3679 * logind now implements a sleep/shutdown/idle inhibiting logic
3680 suitable for a variety of uses. Soonishly Lennart will blog
3681 about this in more detail.
3683 * var-run.mount and var-lock.mount are no longer provided
3684 (which prevously bind mounted these directories to their new
3685 places). Distributions which have not converted these
3686 directories to symlinks should consider stealing these files
3687 from git history and add them downstream.
3689 * We introduced the Documentation= field for units and added
3690 this to all our shipped units. This is useful to make it
3691 easier to explore the boot and the purpose of the various
3694 * All smaller setup units (such as
3695 systemd-vconsole-setup.service) now detect properly if they
3696 are run in a container and are skipped when
3697 appropriate. This guarantees an entirely noise-free boot in
3698 Linux container environments such as systemd-nspawn.
3700 * A framework for implementing offline system updates is now
3701 integrated, for details see:
3702 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3704 * A new service type Type=idle is available now which helps us
3705 avoiding ugly interleaving of getty output and boot status
3708 * There's now a system-wide CapabilityBoundingSet= option to
3709 globally reduce the set of capabilities for the
3710 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3711 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3712 even CAP_NET_ADMIN system-wide for secure systems.
3714 * There are now system-wide DefaultLimitXXX= options to
3715 globally change the defaults of the various resource limits
3716 for all units started by PID 1.
3718 * Harald Hoyer's systemd test suite has been integrated into
3719 systemd which allows easy testing of systemd builds in qemu
3720 and nspawn. (This is really awesome! Ask us for details!)
3722 * The fstab parser is now implemented as generator, not inside
3725 * systemctl will now warn you if .mount units generated from
3726 /etc/fstab are out of date due to changes in fstab that
3727 have not been read by systemd yet.
3729 * systemd is now suitable for usage in initrds. Dracut has
3730 already been updated to make use of this. With this in place
3731 initrds get a slight bit faster but primarily are much
3732 easier to introspect and debug since "systemctl status" in
3733 the host system can be used to introspect initrd services,
3734 and the journal from the initrd is kept around too.
3736 * systemd-delta has been added, a tool to explore differences
3737 between user/admin configuration and vendor defaults.
3739 * PrivateTmp= now affects both /tmp and /var/tmp.
3741 * Boot time status messages are now much prettier and feature
3742 proper english language. Booting up systemd has never been
3745 * Read-ahead pack files now include the inode number of all
3746 files to pre-cache. When the inode changes the pre-caching
3747 is not attempted. This should be nicer to deal with updated
3748 packages which might result in changes of read-ahead
3751 * We now temporaritly lower the kernel's read_ahead_kb variable
3752 when collecting read-ahead data to ensure the kernel's
3753 built-in read-ahead does not add noise to our measurements
3754 of necessary blocks to pre-cache.
3756 * There's now RequiresMountsFor= to add automatic dependencies
3757 for all mounts necessary for a specific file system path.
3759 * MountAuto= and SwapAuto= have been removed from
3760 system.conf. Mounting file systems at boot has to take place
3763 * nspawn now learned a new switch --uuid= to set the machine
3764 ID on the command line.
3766 * nspawn now learned the -b switch to automatically search
3769 * vt102 is now the default TERM for serial TTYs, upgraded from
3772 * systemd-logind now works on VT-less systems.
3774 * The build tree has been reorganized. The individual
3775 components now have directories of their own.
3777 * A new condition type ConditionPathIsReadWrite= is now available.
3779 * nspawn learned the new -C switch to create cgroups for the
3780 container in other hierarchies.
3782 * We now have support for hardware watchdogs, configurable in
3785 * The scheduled shutdown logic now has a public API.
3787 * We now mount /tmp as tmpfs by default, but this can be
3788 masked and /etc/fstab can override it.
3790 * Since udisks does not make use of /media anymore we are not
3791 mounting a tmpfs on it anymore.
3793 * journalctl gained a new --local switch to only interleave
3794 locally generated journal files.
3796 * We can now load the IMA policy at boot automatically.
3798 * The GTK tools have been split off into a systemd-ui.
3800 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3801 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3802 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3803 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3804 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3805 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3806 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3807 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3808 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3813 * This is mostly a bugfix release
3815 * Support optional initialization of the machine ID from the
3816 KVM or container configured UUID.
3818 * Support immediate reboots with "systemctl reboot -ff"
3820 * Show /etc/os-release data in systemd-analyze output
3822 * Many bugfixes for the journal, including endianness fixes and
3823 ensuring that disk space enforcement works
3825 * sd-login.h is C++ comptaible again
3827 * Extend the /etc/os-release format on request of the Debian
3830 * We now refuse non-UTF8 strings used in various configuration
3831 and unit files. This is done to ensure we do not pass invalid
3832 data over D-Bus or expose it elsewhere.
3834 * Register Mimo USB Screens as suitable for automatic seat
3837 * Read SELinux client context from journal clients in a race
3840 * Reorder configuration file lookup order. /etc now always
3841 overrides /run in order to allow the administrator to always
3842 and unconditionally override vendor supplied or
3843 automatically generated data.
3845 * The various user visible bits of the journal now have man
3846 pages. We still lack man pages for the journal API calls
3849 * We now ship all man pages in HTML format again in the
3852 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3853 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3854 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3859 * This is mostly a bugfix release
3861 * systems lacking /etc/os-release are no longer supported.
3863 * Various functionality updates to libsystemd-login.so
3865 * Track class of PAM logins to distuingish greeters from
3868 Contributions from: Kay Sievers, Lennart Poettering, Michael
3873 * This is an important bugfix release for v41.
3875 * Building man pages is now optional which should be useful
3876 for those building systemd from git but unwilling to install
3879 * Watchdog support for supervising services is now usable. In
3880 a future release support for hardware watchdogs
3881 (i.e. /dev/watchdog) will be added building on this.
3883 * Service start rate limiting is now configurable and can be
3884 turned off per service. When a start rate limit is hit a
3885 reboot can automatically be triggered.
3887 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3889 Contributions from: Benjamin Franzke, Bill Nottingham,
3890 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3891 Schmidt, Michał Górny, Piotr Drąg
3895 * The systemd binary is installed /usr/lib/systemd/systemd now;
3896 An existing /sbin/init symlink needs to be adapted with the
3899 * The code that loads kernel modules has been ported to invoke
3900 libkmod directly, instead of modprobe. This means we do not
3901 support systems with module-init-tools anymore.
3903 * Watchdog support is now already useful, but still not
3906 * A new kernel command line option systemd.setenv= is
3907 understood to set system wide environment variables
3908 dynamically at boot.
3910 * We now limit the set of capabilities of systemd-journald.
3912 * We now set SIGPIPE to ignore by default, since it only is
3913 useful in shell pipelines, and has little use in general
3914 code. This can be disabled with IgnoreSIPIPE=no in unit
3917 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3918 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3923 * This is mostly a bugfix release
3925 * We now expose the reason why a service failed in the
3926 "Result" D-Bus property.
3928 * Rudimentary service watchdog support (will be completed over
3929 the next few releases.)
3931 * When systemd forks off in order execute some service we will
3932 now immediately changes its argv[0] to reflect which process
3933 it will execute. This is useful to minimize the time window
3934 with a generic argv[0], which makes bootcharts more useful
3936 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3937 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3938 Mike Kazantsev, Ray Strode
3942 * This is mostly a test release, but incorporates many
3945 * New systemd-cgtop tool to show control groups by their
3948 * Linking against libacl for ACLs is optional again. If
3949 disabled, support tracking device access for active logins
3950 goes becomes unavailable, and so does access to the user
3951 journals by the respective users.
3953 * If a group "adm" exists, journal files are automatically
3954 owned by them, thus allow members of this group full access
3955 to the system journal as well as all user journals.
3957 * The journal now stores the SELinux context of the logging
3958 client for all entries.
3960 * Add C++ inclusion guards to all public headers
3962 * New output mode "cat" in the journal to print only text
3963 messages, without any meta data like date or time.
3965 * Include tiny X server wrapper as a temporary stop-gap to
3966 teach XOrg udev display enumeration. This is used by display
3967 managers such as gdm, and will go away as soon as XOrg
3968 learned native udev hotplugging for display devices.
3970 * Add new systemd-cat tool for executing arbitrary programs
3971 with STDERR/STDOUT connected to the journal. Can also act as
3972 BSD logger replacement, and does so by default.
3974 * Optionally store all locally generated coredumps in the
3975 journal along with meta data.
3977 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3978 writing short strings to files (for usage for /sys), and for
3979 creating symlinks, character and block device nodes.
3981 * New unit file option ControlGroupPersistent= to make cgroups
3982 persistent, following the mechanisms outlined in
3983 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3985 * Support multiple local RTCs in a sane way
3987 * No longer monopolize IO when replaying readahead data on
3988 rotating disks, since we might starve non-file-system IO to
3989 death, since fanotify() will not see accesses done by blkid,
3992 * Do not show kernel threads in systemd-cgls anymore, unless
3993 requested with new -k switch.
3995 Contributions from: Dan Horák, Kay Sievers, Lennart
3996 Poettering, Michal Schmidt
4000 * This is mostly a test release, but incorporates many
4003 * The git repository moved to:
4004 git://anongit.freedesktop.org/systemd/systemd
4005 ssh://git.freedesktop.org/git/systemd/systemd
4007 * First release with the journal
4008 http://0pointer.de/blog/projects/the-journal.html
4010 * The journal replaces both systemd-kmsg-syslogd and
4011 systemd-stdout-bridge.
4013 * New sd_pid_get_unit() API call in libsystemd-logind
4015 * Many systemadm clean-ups
4017 * Introduce remote-fs-pre.target which is ordered before all
4018 remote mounts and may be used to start services before all
4021 * Added Mageia support
4023 * Add bash completion for systemd-loginctl
4025 * Actively monitor PID file creation for daemons which exit in
4026 the parent process before having finished writing the PID
4027 file in the daemon process. Daemons which do this need to be
4028 fixed (i.e. PID file creation must have finished before the
4029 parent exits), but we now react a bit more gracefully to them.
4031 * Add colourful boot output, mimicking the well-known output
4032 of existing distributions.
4034 * New option PassCredentials= for socket units, for
4035 compatibility with a recent kernel ABI breakage.
4037 * /etc/rc.local is now hooked in via a generator binary, and
4038 thus will no longer act as synchronization point during
4041 * systemctl list-unit-files now supports --root=.
4043 * systemd-tmpfiles now understands two new commands: z, Z for
4044 relabelling files according to the SELinux database. This is
4045 useful to apply SELinux labels to specific files in /sys,
4048 * Output of SysV services is now forwarded to both the console
4049 and the journal by default, not only just the console.
4051 * New man pages for all APIs from libsystemd-login.
4053 * The build tree got reorganized and a the build system is a
4054 lot more modular allowing embedded setups to specifically
4055 select the components of systemd they are interested in.
4057 * Support for Linux systems lacking the kernel VT subsystem is
4060 * configure's --with-rootdir= got renamed to
4061 --with-rootprefix= to follow the naming used by udev and
4064 * Unless specified otherwise we will now install to /usr instead
4065 of /usr/local by default.
4067 * Processes with '@' in argv[0][0] are now excluded from the
4068 final shut-down killing spree, following the logic explained
4070 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4072 * All processes remaining in a service cgroup when we enter
4073 the START or START_PRE states are now killed with
4074 SIGKILL. That means it is no longer possible to spawn
4075 background processes from ExecStart= lines (which was never
4076 supported anyway, and bad style).
4078 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4079 reloading of units together.
4081 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4082 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4083 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4084 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4085 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek