1 systemd System and Service Manager
5 * timedated no longer reads NTP implementation unit names from
6 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
7 implementations should add a
9 Conflicts=systemd-timesyncd.service
11 to their unit files to take over and replace systemd's NTP
12 default functionality.
14 * systemd-sysusers gained a new line type "r" for configuring
15 which UID/GID ranges to allocate system users/groups
16 from. Lines of type "u" may now add an additional column
17 that specifies the home directory for the system user to be
18 created. Also, systemd-sysusers may now optionally read user
19 information from STDIN instead of a file. This is useful for
20 invoking it from RPM preinst scriptlets that need to create
21 users before the first RPM file is installed since these
22 files might need to be owned by them. A new
23 %sysusers_create_inline RPM macro has been introduced to do
24 just that. systemd-sysusers now updates the shadow files as
25 well as the user/group databases, which should enhance
26 compatibility with certain tools like grpck.
28 * A number of bus APIs of PID 1 now optionally consult
29 PolicyKit to permit access for otherwise unprivileged
30 clients under certain conditions. Note that this currently
31 doesn't support interactive authentication yet, but this is
32 expected to be added eventually, too.
34 * /etc/machine-info now has new fields for configuring the
35 deployment environment of the machine, as well as the
36 location of the machine. hostnamectl has been updated with
37 new command to update these fields.
39 * systemd-timesyncd has been updated to automatically acquire
40 NTP server information from systemd-networkd, which might
41 have been discovered via DHCP.
43 * systemd-resolved now includes a caching DNS stub resolver
44 and a complete LLMNR name resolution implementation. A new
45 NSS module "nss-resolve" has been added which make be used
46 of glibc's own "nss-dns" to resolve hostnames via
47 systemd-resolved. Hostnames, addresses and arbitrary RRs may
48 be resolved via systemd-resolved D-Bus APIs. In contrast to
49 the glibc internal resolver systemd-resolved is aware of
50 multi-homed system, and keeps DNS server and caches separate
51 and per-interface. Queries are sent simultaneously on all
52 interfaces that have DNS servers configured, in order to
53 properly handle VPNs and local LANs which might resolve
54 separate sets of domain names. systemd-resolved may acquire
55 DNS server information from systemd-networkd automatically,
56 which in turn might have discovered them via DHCP. A tool
57 "systemd-resolve-host" has been added that may be used to
58 query the DNS logic in resolved. systemd-resolved implements
59 IDNA and automatically uses IDNA or UTF-8 encoding depending
60 on whether classic DNS or LLMNR is used as transport. In the
61 next releases we intend to add a DNSSEC and mDNS/DNS-SD
62 implementation to systemd-resolved.
64 * A new NSS module nss-mymachines has been added, that
65 automatically resolves the names of all local registered
66 containers to their respective IP addresses.
68 * A new client tool "networkctl" for systemd-networkd has been
69 added. It currently is entirely passive and will query
70 networking configuration from udev, rtnetlink and networkd,
71 and present it to the user in a very friendly
72 way. Eventually, we hope to extend it to become a full
73 control utility for networkd.
75 * .socket units gained a new DeferAcceptSec= setting that
76 controls the kernels' TCP_DEFER_ACCEPT sockopt for
77 TCP. Similar, support for controlling TCP keep-alive
78 settings has been added (KeepAliveTimeSec=,
79 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
80 turning off Nagle's algorithm on TCP has been added
83 * logind learned a new session type "web", for use in projects
84 like Cockpit which register web clients as PAM sessions.
86 * timer units with at least one OnCalendar= setting will now
87 be started only after timer-sync.target has been
88 reached. This way they will not elapse before the system
89 clock has been corrected by a local NTP client or
90 similar. This is particular useful on RTC-less embedded
91 machines, that come up with an invalid system clock.
93 * systemd-nspawn's --network-veth= switch should now result in
94 stable MAC addresses for both the outer and the inner side
97 * systemd-nspawn gained a new --volatile= switch for running
98 container instances with /etc or /var unpopulated.
100 * The kdbus client code has been updated to use the new Linux
101 3.17 memfd subsystem instead of the old kdbus-specific one.
103 * systemd-networkd's DHCP client and server now support
106 * systemd will no longer inform the kernel about the current
107 timezone, as this is necessarily incorrect and racy as the
108 kernel has no understanding of DST and similar
109 concepts. This hence means FAT timestamps will be always
110 considered UTC, similar to what Android is already
111 doing. Also, when the RTC is configured to the local time
112 (rather than UTC) systemd will never synchronize back to it,
113 as this might confuse Windows at a later boot.
115 * systemd-analyze gained a new command "verify" for offline
116 validation of unit files.
118 * systemd-networkd gained support for a couple of additional
119 settings for bonding networking setups. Also, the metric for
120 statically configured routes may now be configured. For
121 network interfaces where this is appropriate the peer IP
122 address may now be configured.
124 * systemd-networkd's DHCP client will no longer request
125 broadcasting by default, as this tripped up some networks.
126 For hardware where broadcast is required the feature should
127 be switched back on using RequestBroadcast=yes.
129 * systemd-networkd will now set up IPv4LL addresses (when
130 enabled) even if DHCP is configured successfully.
132 * udev will now default to respect network device names given
133 by the kernel when the kernel indicates that these are
134 predictable. This behavior can be tweaked by changing
135 NamePolicy= in the relevant .link file.
137 * A new library systemd-terminal has been added that
138 implements full TTY stream parsing and rendering. This
139 library is supposed to be used later on for implementing a
140 full userspace VT subsystem, replacing the current kernel
143 * A new tool systemd-journal-upload has been added to push
144 journal data to a remote system running
145 systemd-journal-remote.
147 * journald will no longer forward all local data to another
148 running syslog daemon. This change has been made because
149 rsyslog (which appears to be the most commonly used syslog
150 implementation these days) no longer makes use of this, and
151 instead pulls the data out of the journal on its own. Since
152 forwarding the messages to a non-existent syslog server is
153 more expensive than we assumed we have now turned this
154 off. If you run a syslog server that is not a recent rsyslog
155 version, you have to turn this option on again
156 (ForwardToSyslog= in journald.conf).
158 * journald now optionally supports the LZ4 compressor for
159 larger journal fields. This compressor should perform much
160 better than XZ which was the previous default.
162 * machinectl now shows the IP addresses of local containers,
163 if it knows them, plus the interface name of the container.
165 * A new tool "systemd-escape" has been added that makes it
166 easy to escape strings to build unit names and similar.
168 * sd_notify() messages may now include a new ERRNO= field
169 which is parsed and collected by systemd and shown among the
170 "systemctl status" output for a service.
172 * A new component "systemd-firstboot" has been added that
173 queries the most basic systemd information (timezone,
174 hostname, root password) interactively on first
175 boot. Alternatively it may also be used to provision these
176 things offline on OS images installed into directories.
178 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
179 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
180 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
181 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
182 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
183 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
184 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
185 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
186 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
187 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
188 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
189 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
190 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
191 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
192 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
194 -- Berlin, 2014-08-19
198 * A new tool systemd-sysusers has been added. This tool
199 creates system users and groups in /etc/passwd and
200 /etc/group, based on static declarative system user/group
201 definitions in /usr/lib/sysusers.d/. This is useful to
202 enable factory resets and volatile systems that boot up with
203 an empty /etc directory, and thus need system users and
204 groups created during early boot. systemd now also ships
205 with two default sysusers.d/ files for the most basic
206 users and groups systemd and the core operating system
209 * A new tmpfiles snippet has been added that rebuilds the
210 essential files in /etc on boot, should they be missing.
212 * A directive for ensuring automatic clean-up of
213 /var/cache/man/ has been removed from the default
214 configuration. This line should now be shipped by the man
215 implementation. The necessary change has been made to the
216 man-db implementation. Note that you need to update your man
217 implementation to one that ships this line, otherwise no
218 automatic clean-up of /var/cache/man will take place.
220 * A new condition ConditionNeedsUpdate= has been added that
221 may conditionalize services to only run when /etc or /var
222 are "older" than the vendor operating system resources in
223 /usr. This is useful for reconstructing or updating /etc
224 after an offline update of /usr or a factory reset, on the
225 next reboot. Services that want to run once after such an
226 update or reset should use this condition and order
227 themselves before the new systemd-update-done.service, which
228 will mark the two directories as fully updated. A number of
229 service files have been added making use of this, to rebuild
230 the udev hardware database, the journald message catalog and
231 dynamic loader cache (ldconfig). The systemd-sysusers tool
232 described above also makes use of this now. With this in
233 place it is now possible to start up a minimal operating
234 system with /etc empty cleanly. For more information on the
235 concepts involved see this recent blog story:
237 http://0pointer.de/blog/projects/stateless.html
239 * A new system group "input" has been introduced, and all
240 input device nodes get this group assigned. This is useful
241 for system-level software to get access to input devices. It
242 complements what is already done for "audio" and "video".
244 * systemd-networkd learnt minimal DHCPv4 server support in
245 addition to the existing DHCPv4 client support. It also
246 learnt DHCPv6 client and IPv6 Router Solicitation client
247 support. The DHCPv4 client gained support for static routes
248 passed in from the server. Note that the [DHCPv4] section
249 known in older systemd-networkd versions has been renamed to
250 [DHCP] and is now also used by the DHCPv6 client. Existing
251 .network files using settings of this section should be
252 updated, though compatibility is maintained. Optionally, the
253 client hostname may now be sent to the DHCP server.
255 * networkd gained support for vxlan virtual networks as well
256 as tun/tap and dummy devices.
258 * networkd gained support for automatic allocation of address
259 ranges for interfaces from a system-wide pool of
260 addresses. This is useful for dynamically managing a large
261 number of interfaces with a single network configuration
262 file. In particular this is useful to easily assign
263 appropriate IP addresses to the veth links of a large number
266 * RPM macros for processing sysusers, sysctl and binfmt
267 drop-in snippets at package installation time have been
270 * The /etc/os-release file should now be placed in
271 /usr/lib/os-release. The old location is automatically
272 created as symlink. /usr/lib is the more appropriate
273 location of this file, since it shall actually describe the
274 vendor operating system shipped in /usr, and not the
275 configuration stored in /etc.
277 * .mount units gained a new boolean SloppyOptions= setting
278 that maps to mount(8)'s -s option which enables permissive
279 parsing of unknown mount options.
281 * tmpfiles learnt a new "L+" directive which creates a symlink
282 but (unlike "L") deletes a pre-existing file first, should
283 it already exist and not already be the correct
284 symlink. Similar, "b+", "c+" and "p+" directives have been
285 added as well, which create block and character devices, as
286 well as fifos in the filesystem, possibly removing any
287 pre-existing files of different types.
289 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
290 'argument' field (which so far specified the source to
291 symlink/copy the files from) is now optional. If omitted the
292 same file os copied from /usr/share/factory/ suffixed by the
293 full destination path. This is useful for populating /etc
294 with essential files, by copying them from vendor defaults
295 shipped in /usr/share/factory/etc.
297 * A new command "systemctl preset-all" has been added that
298 applies the service preset settings to all installed unit
299 files. A new switch --preset-mode= has been added that
300 controls whether only enable or only disable operations
303 * A new command "systemctl is-system-running" has been added
304 that allows checking the overall state of the system, for
305 example whether it is fully up and running.
307 * When the system boots up with an empty /etc, the equivalent
308 to "systemctl preset-all" is executed during early boot, to
309 make sure all default services are enabled after a factory
312 * systemd now contains a minimal preset file that enables the
313 most basic services systemd ships by default.
315 * Unit files' [Install] section gained a new DefaultInstance=
316 field for defining the default instance to create if a
317 template unit is enabled with no instance specified.
319 * A new passive target cryptsetup-pre.target has been added
320 that may be used by services that need to make they run and
321 finish before the first LUKS cryptographic device is set up.
323 * The /dev/loop-control and /dev/btrfs-control device nodes
324 are now owned by the "disk" group by default, opening up
325 access to this group.
327 * systemd-coredump will now automatically generate a
328 stack trace of all core dumps taking place on the system,
329 based on elfutils' libdw library. This stack trace is logged
332 * systemd-coredump may now optionally store coredumps directly
333 on disk (in /var/lib/systemd/coredump, possibly compressed),
334 instead of storing them unconditionally in the journal. This
335 mode is the new default. A new configuration file
336 /etc/systemd/coredump.conf has been added to configure this
337 and other parameters of systemd-coredump.
339 * coredumpctl gained a new "info" verb to show details about a
340 specific coredump. A new switch "-1" has also been added
341 that makes sure to only show information about the most
342 recent entry instead of all entries. Also, as the tool is
343 generally useful now the "systemd-" prefix of the binary
344 name has been removed. Distributions that want to maintain
345 compatibility with the old name should add a symlink from
346 the old name to the new name.
348 * journald's SplitMode= now defaults to "uid". This makes sure
349 that unprivileged users can access their own coredumps with
350 coredumpctl without restrictions.
352 * New kernel command line options "systemd.wants=" (for
353 pulling an additional unit during boot), "systemd.mask="
354 (for masking a specific unit for the boot), and
355 "systemd.debug-shell" (for enabling the debug shell on tty9)
356 have been added. This is implemented in the new generator
357 "systemd-debug-generator".
359 * systemd-nspawn will now by default filter a couple of
360 syscalls for containers, among them those required for
361 kernel module loading, direct x86 IO port access, swap
362 management, and kexec. Most importantly though
363 open_by_handle_at() is now prohibited for containers,
364 closing a hole similar to a recently discussed vulnerability
365 in docker regarding access to files on file hierarchies the
366 container should normally not have access to. Note that for
367 nspawn we generally make no security claims anyway (and
368 this is explicitly documented in the man page), so this is
369 just a fix for one of the most obvious problems.
371 * A new man page file-hierarchy(7) has been added that
372 contains a minimized, modernized version of the file system
373 layout systemd expects, similar in style to the FHS
374 specification or hier(5). A new tool systemd-path(1) has
375 been added to query many of these paths for the local
378 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
379 longer done. Since the directory now has a per-user size
380 limit, and is cleaned on logout this appears unnecessary,
381 in particular since this now brings the lifecycle of this
382 directory closer in line with how IPC objects are handled.
384 * systemd.pc now exports a number of additional directories,
385 including $libdir (which is useful to identify the library
386 path for the primary architecture of the system), and a
387 couple of drop-in directories.
389 * udev's predictable network interface names now use the dev_port
390 sysfs attribute, introduced in linux 3.15 instead of dev_id to
391 distinguish between ports of the same PCI function. dev_id should
392 only be used for ports using the same HW address, hence the need
395 * machined has been updated to export the OS version of a
396 container (read from /etc/os-release and
397 /usr/lib/os-release) on the bus. This is now shown in
398 "machinectl status" for a machine.
400 * A new service setting RestartForceExitStatus= has been
401 added. If configured to a set of exit signals or process
402 return values, the service will be restarted when the main
403 daemon process exits with any of them, regardless of the
406 * systemctl's -H switch for connecting to remote systemd
407 machines has been extended so that it may be used to
408 directly connect to a specific container on the
409 host. "systemctl -H root@foobar:waldi" will now connect as
410 user "root" to host "foobar", and then proceed directly to
411 the container named "waldi". Note that currently you have to
412 authenticate as user "root" for this to work, as entering
413 containers is a privileged operation.
415 Contributions from: Andreas Henriksson, Benjamin Steinwender,
416 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
417 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
418 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
419 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
420 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
421 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
422 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
423 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
424 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
425 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
426 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
428 -- Berlin, 2014-07-03
432 * As an experimental feature, udev now tries to lock the
433 disk device node (flock(LOCK_SH|LOCK_NB)) while it
434 executes events for the disk or any of its partitions.
435 Applications like partitioning programs can lock the
436 disk device node (flock(LOCK_EX)) and claim temporary
437 device ownership that way; udev will entirely skip all event
438 handling for this disk and its partitions. If the disk
439 was opened for writing, the close will trigger a partition
440 table rescan in udev's "watch" facility, and if needed
441 synthesize "change" events for the disk and all its partitions.
442 This is now unconditionally enabled, and if it turns out to
443 cause major problems, we might turn it on only for specific
444 devices, or might need to disable it entirely. Device Mapper
445 devices are excluded from this logic.
447 * We temporarily dropped the "-l" switch for fsck invocations,
448 since they collide with the flock() logic above. util-linux
449 upstream has been changed already to avoid this conflict,
450 and we will readd "-l" as soon as util-linux with this
451 change has been released.
453 * The dependency on libattr has been removed. Since a long
454 time, the extended attribute calls have moved to glibc, and
455 libattr is thus unnecessary.
457 * Virtualization detection works without priviliges now. This
458 means the systemd-detect-virt binary no longer requires
459 CAP_SYS_PTRACE file capabilities, and our daemons can run
460 with fewer privileges.
462 * systemd-networkd now runs under its own "systemd-network"
463 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
464 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
465 loses the ability to write to files owned by root this way.
467 * Similar, systemd-resolved now runs under its own
468 "systemd-resolve" user with no capabilities remaining.
470 * Similar, systemd-bus-proxyd now runs under its own
471 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
473 * systemd-networkd gained support for setting up "veth"
474 virtual ethernet devices for container connectivity, as well
475 as GRE and VTI tunnels.
477 * systemd-networkd will no longer automatically attempt to
478 manually load kernel modules necessary for certain tunnel
479 transports. Instead, it is assumed the kernel loads them
480 automatically when required. This only works correctly on
481 very new kernels. On older kernels, please consider adding
482 the kernel modules to /etc/modules-load.d/ as a work-around.
484 * The resolv.conf file systemd-resolved generates has been
485 moved to /run/systemd/resolve/. If you have a symlink from
486 /etc/resolv.conf, it might be necessary to correct it.
488 * Two new service settings, ProtectHome= and ProtectSystem=,
489 have been added. When enabled, they will make the user data
490 (such as /home) inaccessible or read-only and the system
491 (such as /usr) read-only, for specific services. This allows
492 very light-weight per-service sandboxing to avoid
493 modifications of user data or system files from
494 services. These two new switches have been enabled for all
495 of systemd's long-running services, where appropriate.
497 * Socket units gained new SocketUser= and SocketGroup=
498 settings to set the owner user and group of AF_UNIX sockets
499 and FIFOs in the file system.
501 * Socket units gained a new RemoveOnStop= setting. If enabled,
502 all FIFOS and sockets in the file system will be removed
503 when the specific socket unit is stopped.
505 * Socket units gained a new Symlinks= setting. It takes a list
506 of symlinks to create to file system sockets or FIFOs
507 created by the specific Unix sockets. This is useful to
508 manage symlinks to socket nodes with the same life-cycle as
511 * The /dev/log socket and /dev/initctl FIFO have been moved to
512 /run, and have been replaced by symlinks. This allows
513 connecting to these facilities even if PrivateDevices=yes is
514 used for a service (which makes /dev/log itself unavailable,
515 but /run is left). This also has the benefit of ensuring
516 that /dev only contains device nodes, directories and
517 symlinks, and nothing else.
519 * sd-daemon gained two new calls sd_pid_notify() and
520 sd_pid_notifyf(). They are similar to sd_notify() and
521 sd_notifyf(), but allow overriding of the source PID of
522 notification messages if permissions permit this. This is
523 useful to send notify messages on behalf of a different
524 process (for example, the parent process). The
525 systemd-notify tool has been updated to make use of this
526 when sending messages (so that notification messages now
527 originate from the shell script invoking systemd-notify and
528 not the systemd-notify process itself. This should minimize
529 a race where systemd fails to associate notification
530 messages to services when the originating process already
533 * A new "on-abnormal" setting for Restart= has been added. If
534 set, it will result in automatic restarts on all "abnormal"
535 reasons for a process to exit, which includes unclean
536 signals, core dumps, timeouts and watchdog timeouts, but
537 does not include clean and unclean exit codes or clean
538 signals. Restart=on-abnormal is an alternative for
539 Restart=on-failure for services that shall be able to
540 terminate and avoid restarts on certain errors, by
541 indicating so with an unclean exit code. Restart=on-failure
542 or Restart=on-abnormal is now the recommended setting for
543 all long-running services.
545 * If the InaccessibleDirectories= service setting points to a
546 mount point (or if there are any submounts contained within
547 it), it is now attempted to completely unmount it, to make
548 the file systems truly unavailable for the respective
551 * The ReadOnlyDirectories= service setting and
552 systemd-nspawn's --read-only parameter are now recursively
553 applied to all submounts, too.
555 * Mount units may now be created transiently via the bus APIs.
557 * The support for SysV and LSB init scripts has been removed
558 from the systemd daemon itself. Instead, it is now
559 implemented as a generator that creates native systemd units
560 from these scripts when needed. This enables us to remove a
561 substantial amount of legacy code from PID 1, following the
562 fact that many distributions only ship a very small number
563 of LSB/SysV init scripts nowadays.
565 * Priviliged Xen (dom0) domains are not considered
566 virtualization anymore by the virtualization detection
567 logic. After all, they generally have unrestricted access to
568 the hardware and usually are used to manage the unprivileged
571 * systemd-tmpfiles gained a new "C" line type, for copying
572 files or entire directories.
574 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
575 lines. So far, they have been non-globbing versions of the
576 latter, and have thus been redundant. In future, it is
577 recommended to only use "z". "m" has hence been removed
578 from the documentation, even though it stays supported.
580 * A tmpfiles snippet to recreate the most basic structure in
581 /var has been added. This is enough to create the /var/run →
582 /run symlink and create a couple of structural
583 directories. This allows systems to boot up with an empty or
584 volatile /var. Of course, while with this change, the core OS
585 now is capable with dealing with a volatile /var, not all
586 user services are ready for it. However, we hope that sooner
587 or later, many service daemons will be changed upstream so
588 that they are able to automatically create their necessary
589 directories in /var at boot, should they be missing. This is
590 the first step to allow state-less systems that only require
591 the vendor image for /usr to boot.
593 * systemd-nspawn has gained a new --tmpfs= switch to mount an
594 empty tmpfs instance to a specific directory. This is
595 particularly useful for making use of the automatic
596 reconstruction of /var (see above), by passing --tmpfs=/var.
598 * Access modes specified in tmpfiles snippets may now be
599 prefixed with "~", which indicates that they shall be masked
600 by whether the existing file or directly is currently
601 writable, readable or executable at all. Also, if specified,
602 the sgid/suid/sticky bits will be masked for all
605 * A new passive target unit "network-pre.target" has been
606 added which is useful for services that shall run before any
607 network is configured, for example firewall scripts.
609 * The "floppy" group that previously owned the /dev/fd*
610 devices is no longer used. The "disk" group is now used
611 instead. Distributions should probably deprecate usage of
614 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
615 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
616 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
617 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
618 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
619 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
622 -- Berlin, 2014-06-11
626 * A new "systemd-timesyncd" daemon has been added for
627 synchronizing the system clock across the network. It
628 implements an SNTP client. In contrast to NTP
629 implementations such as chrony or the NTP reference server,
630 this only implements a client side, and does not bother with
631 the full NTP complexity, focusing only on querying time from
632 one remote server and synchronizing the local clock to
633 it. Unless you intend to serve NTP to networked clients or
634 want to connect to local hardware clocks, this simple NTP
635 client should be more than appropriate for most
636 installations. The daemon runs with minimal privileges, and
637 has been hooked up with networkd to only operate when
638 network connectivity is available. The daemon saves the
639 current clock to disk every time a new NTP sync has been
640 acquired, and uses this to possibly correct the system clock
641 early at bootup, in order to accommodate for systems that
642 lack an RTC such as the Raspberry Pi and embedded devices,
643 and to make sure that time monotonically progresses on these
644 systems, even if it is not always correct. To make use of
645 this daemon, a new system user and group "systemd-timesync"
646 needs to be created on installation of systemd.
648 * The queue "seqnum" interface of libudev has been disabled, as
649 it was generally incompatible with device namespacing as
650 sequence numbers of devices go "missing" if the devices are
651 part of a different namespace.
653 * "systemctl list-timers" and "systemctl list-sockets" gained
654 a --recursive switch for showing units of these types also
655 for all local containers, similar in style to the already
656 supported --recursive switch for "systemctl list-units".
658 * A new RebootArgument= setting has been added for service
659 units, which may be used to specify a kernel reboot argument
660 to use when triggering reboots with StartLimitAction=.
662 * A new FailureAction= setting has been added for service
663 units which may be used to specify an operation to trigger
664 when a service fails. This works similarly to
665 StartLimitAction=, but unlike it, controls what is done
666 immediately rather than only after several attempts to
667 restart the service in question.
669 * hostnamed got updated to also expose the kernel name,
670 release, and version on the bus. This is useful for
671 executing commands like hostnamectl with the -H switch.
672 systemd-analyze makes use of this to properly display
673 details when running non-locally.
675 * The bootchart tool can now show cgroup information in the
678 * The CFS CPU quota cgroup attribute is now exposed for
679 services. The new CPUQuota= switch has been added for this
680 which takes a percentage value. Setting this will have the
681 result that a service may never get more CPU time than the
682 specified percentage, even if the machine is otherwise idle.
684 * systemd-networkd learned IPIP and SIT tunnel support.
686 * LSB init scripts exposing a dependency on $network will now
687 get a dependency on network-online.target rather than simply
688 network.target. This should bring LSB handling closer to
689 what it was on SysV systems.
691 * A new fsck.repair= kernel option has been added to control
692 how fsck shall deal with unclean file systems at boot.
694 * The (.ini) configuration file parser will now silently
695 ignore sections whose name begins with "X-". This may be
696 used to maintain application-specific extension sections in unit
699 * machined gained a new API to query the IP addresses of
700 registered containers. "machinectl status" has been updated
701 to show these addresses in its output.
703 * A new call sd_uid_get_display() has been added to the
704 sd-login APIs for querying the "primary" session of a
705 user. The "primary" session of the user is elected from the
706 user's sessions and generally a graphical session is
707 preferred over a text one.
709 * A minimal systemd-resolved daemon has been added. It
710 currently simply acts as a companion to systemd-networkd and
711 manages resolv.conf based on per-interface DNS
712 configuration, possibly supplied via DHCP. In the long run
713 we hope to extend this into a local DNSSEC enabled DNS and
716 * The systemd-networkd-wait-online tool is now enabled by
717 default. It will delay network-online.target until a network
718 connection has been configured. The tool primarily integrates
719 with networkd, but will also make a best effort to make sense
720 of network configuration performed in some other way.
722 * Two new service options StartupCPUShares= and
723 StartupBlockIOWeight= have been added that work similarly to
724 CPUShares= and BlockIOWeight= however only apply during
725 system startup. This is useful to prioritize certain services
726 differently during bootup than during normal runtime.
728 * hostnamed has been changed to prefer the statically
729 configured hostname in /etc/hostname (unless set to
730 'localhost' or empty) over any dynamic one supplied by
731 dhcp. With this change, the rules for picking the hostname
732 match more closely the rules of other configuration settings
733 where the local administrator's configuration in /etc always
734 overrides any other settings.
736 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
737 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
738 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
739 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
740 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
741 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
742 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
743 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
744 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
745 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
746 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
747 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
748 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
749 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
750 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
751 Lindskog, WaLyong Cho, Will Woods, Zbigniew
754 -- Beijing, 2014-05-28
758 * When restoring the screen brightness at boot, stay away from
759 the darkest setting or from the lowest 5% of the available
760 range, depending on which is the larger value of both. This
761 should effectively protect the user from rebooting into a
762 black screen, should the brightness have been set to minimum
765 * sd-login gained a new sd_machine_get_class() call to
766 determine the class ("vm" or "container") of a machine
767 registered with machined.
769 * sd-login gained new calls
770 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
771 to query the identity of the peer of a local AF_UNIX
772 connection. They operate similarly to their sd_pid_get_xyz()
775 * PID 1 will now maintain a system-wide system state engine
776 with the states "starting", "running", "degraded",
777 "maintenance", "stopping". These states are bound to system
778 startup, normal runtime, runtime with at least one failed
779 service, rescue/emergency mode and system shutdown. This
780 state is shown in the "systemctl status" output when no unit
781 name is passed. It is useful to determine system state, in
782 particularly when doing so for many systems or containers at
785 * A new command "list-machines" has been added to "systemctl"
786 that lists all local OS containers and shows their system
787 state (see above), if systemd runs inside of them.
789 * systemctl gained a new "-r" switch to recursively enumerate
790 units on all local containers, when used with the
791 "list-unit" command (which is the default one that is
792 executed when no parameters are specified).
794 * The GPT automatic partition discovery logic will now honour
795 two GPT partition flags: one may be set on a partition to
796 cause it to be mounted read-only, and the other may be set
797 on a partition to ignore it during automatic discovery.
799 * Two new GPT type UUIDs have been added for automatic root
800 partition discovery, for 32-bit and 64-bit ARM. This is not
801 particularly useful for discovering the root directory on
802 these architectures during bare-metal boots (since UEFI is
803 not common there), but still very useful to allow booting of
804 ARM disk images in nspawn with the -i option.
806 * MAC addresses of interfaces created with nspawn's
807 --network-interface= switch will now be generated from the
808 machine name, and thus be stable between multiple invocations
811 * logind will now automatically remove all IPC objects owned
812 by a user if she or he fully logs out. This makes sure that
813 users who are logged out cannot continue to consume IPC
814 resources. This covers SysV memory, semaphores and message
815 queues as well as POSIX shared memory and message
816 queues. Traditionally, SysV and POSIX IPC had no life-cycle
817 limits. With this functionality, that is corrected. This may
818 be turned off by using the RemoveIPC= switch of logind.conf.
820 * The systemd-machine-id-setup and tmpfiles tools gained a
821 --root= switch to operate on a specific root directory,
824 * journald can now forward logged messages to the TTYs of all
825 logged in users ("wall"). This is the default for all
826 emergency messages now.
828 * A new tool systemd-journal-remote has been added to stream
829 journal log messages across the network.
831 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
832 controller trees are mounted into it. Note that the
833 directories mounted beneath it are not read-only. This is a
834 security measure and is particularly useful because glibc
835 actually includes a search logic to pick any tmpfs it can
836 find to implement shm_open() if /dev/shm is not available
837 (which it might very well be in namespaced setups).
839 * machinectl gained a new "poweroff" command to cleanly power
840 down a local OS container.
842 * The PrivateDevices= unit file setting will now also drop the
843 CAP_MKNOD capability from the capability bound set, and
844 imply DevicePolicy=closed.
846 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
847 comprehensively on all long-running systemd services where
850 * systemd-udevd will now run in a disassociated mount
851 namespace. To mount directories from udev rules, make sure to
852 pull in mount units via SYSTEMD_WANTS properties.
854 * The kdbus support gained support for uploading policy into
855 the kernel. sd-bus gained support for creating "monitoring"
856 connections that can eavesdrop into all bus communication
857 for debugging purposes.
859 * Timestamps may now be specified in seconds since the UNIX
860 epoch Jan 1st, 1970 by specifying "@" followed by the value
863 * Native tcpwrap support in systemd has been removed. tcpwrap
864 is old code, not really maintained anymore and has serious
865 shortcomings, and better options such as firewalls
866 exist. For setups that require tcpwrap usage, please
867 consider invoking your socket-activated service via tcpd,
868 like on traditional inetd.
870 * A new system.conf configuration option
871 DefaultTimerAccuracySec= has been added that controls the
872 default AccuracySec= setting of .timer units.
874 * Timer units gained a new WakeSystem= switch. If enabled,
875 timers configured this way will cause the system to resume
876 from system suspend (if the system supports that, which most
879 * Timer units gained a new Persistent= switch. If enabled,
880 timers configured this way will save to disk when they have
881 been last triggered. This information is then used on next
882 reboot to possible execute overdue timer events, that
883 could not take place because the system was powered off.
884 This enables simple anacron-like behaviour for timer units.
886 * systemctl's "list-timers" will now also list the time a
887 timer unit was last triggered in addition to the next time
888 it will be triggered.
890 * systemd-networkd will now assign predictable IPv4LL
891 addresses to its local interfaces.
893 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
894 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
895 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
896 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
897 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
898 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
899 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
900 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
903 -- Berlin, 2014-03-25
907 * A new unit file setting RestrictAddressFamilies= has been
908 added to restrict which socket address families unit
909 processes gain access to. This takes address family names
910 like "AF_INET" or "AF_UNIX", and is useful to minimize the
911 attack surface of services via exotic protocol stacks. This
912 is built on seccomp system call filters.
914 * Two new unit file settings RuntimeDirectory= and
915 RuntimeDirectoryMode= have been added that may be used to
916 manage a per-daemon runtime directories below /run. This is
917 an alternative for setting up directory permissions with
918 tmpfiles snippets, and has the advantage that the runtime
919 directory's lifetime is bound to the daemon runtime and that
920 the daemon starts up with an empty directory each time. This
921 is particularly useful when writing services that drop
922 privileges using the User= or Group= setting.
924 * The DeviceAllow= unit setting now supports globbing for
925 matching against device group names.
927 * The systemd configuration file system.conf gained new
928 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
929 DefaultMemoryAccounting= to globally turn on/off accounting
930 for specific resources (cgroups) for all units. These
931 settings may still be overridden individually in each unit
934 * systemd-gpt-auto-generator is now able to discover /srv and
935 root partitions in addition to /home and swap partitions. It
936 also supports LUKS-encrypted partitions now. With this in
937 place, automatic discovery of partitions to mount following
938 the Discoverable Partitions Specification
939 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
940 is now a lot more complete. This allows booting without
941 /etc/fstab and without root= on the kernel command line on
942 systems prepared appropriately.
944 * systemd-nspawn gained a new --image= switch which allows
945 booting up disk images and Linux installations on any block
946 device that follow the Discoverable Partitions Specification
947 (see above). This means that installations made with
948 appropriately updated installers may now be started and
949 deployed using container managers, completely
950 unmodified. (We hope that libvirt-lxc will add support for
951 this feature soon, too.)
953 * systemd-nspawn gained a new --network-macvlan= setting to
954 set up a private macvlan interface for the
955 container. Similarly, systemd-networkd gained a new
956 Kind=macvlan setting in .netdev files.
958 * systemd-networkd now supports configuring local addresses
961 * A new tool systemd-network-wait-online has been added to
962 synchronously wait for network connectivity using
965 * The sd-bus.h bus API gained a new sd_bus_track object for
966 tracking the life-cycle of bus peers. Note that sd-bus.h is
967 still not a public API though (unless you specify
968 --enable-kdbus on the configure command line, which however
969 voids your warranty and you get no API stability guarantee).
971 * The $XDG_RUNTIME_DIR runtime directories for each user are
972 now individual tmpfs instances, which has the benefit of
973 introducing separate pools for each user, with individual
974 size limits, and thus making sure that unprivileged clients
975 can no longer negatively impact the system or other users by
976 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
977 RuntimeDirectorySize= has been introduced that allows
978 controlling the default size limit for all users. It
979 defaults to 10% of the available physical memory. This is no
980 replacement for quotas on tmpfs though (which the kernel
981 still does not support), as /dev/shm and /tmp are still
982 shared resources used by both the system and unprivileged
985 * logind will now automatically turn off automatic suspending
986 on laptop lid close when more than one display is
987 connected. This was previously expected to be implemented
988 individually in desktop environments (such as GNOME),
989 however has been added to logind now, in order to fix a
990 boot-time race where a desktop environment might not have
991 been started yet and thus not been able to take an inhibitor
992 lock at the time where logind already suspends the system
995 * logind will now wait at least 30s after each system
996 suspend/resume cycle, and 3min after system boot before
997 suspending the system due to a closed laptop lid. This
998 should give USB docking stations and similar enough time to
999 be probed and configured after system resume and boot in
1000 order to then act as suspend blocker.
1002 * systemd-run gained a new --property= setting which allows
1003 initialization of resource control properties (and others)
1004 for the created scope or service unit. Example: "systemd-run
1005 --property=BlockIOWeight=10 updatedb" may be used to run
1006 updatedb at a low block IO scheduling weight.
1008 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1009 now also work in --scope mode.
1011 * When systemd is compiled with kdbus support, basic support
1012 for enforced policies is now in place. (Note that enabling
1013 kdbus still voids your warranty and no API compatibility
1016 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1017 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1018 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1019 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1020 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1021 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1022 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1023 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1024 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1025 Zbigniew Jędrzejewski-Szmek
1027 -- Berlin, 2014-03-12
1031 * systemd will now relabel /dev after loading the SMACK policy
1032 according to SMACK rules.
1034 * A new unit file option AppArmorProfile= has been added to
1035 set the AppArmor profile for the processes of a unit.
1037 * A new condition check ConditionArchitecture= has been added
1038 to conditionalize units based on the system architecture, as
1039 reported by uname()'s "machine" field.
1041 * systemd-networkd now supports matching on the system
1042 virtualization, architecture, kernel command line, host name
1045 * logind is now a lot more aggressive when suspending the
1046 machine due to a closed laptop lid. Instead of acting only
1047 on the lid close action, it will continuously watch the lid
1048 status and act on it. This is useful for laptops where the
1049 power button is on the outside of the chassis so that it can
1050 be reached without opening the lid (such as the Lenovo
1051 Yoga). On those machines, logind will now immediately
1052 re-suspend the machine if the power button has been
1053 accidentally pressed while the laptop was suspended and in a
1054 backpack or similar.
1056 * logind will now watch SW_DOCK switches and inhibit reaction
1057 to the lid switch if it is pressed. This means that logind
1058 will not suspend the machine anymore if the lid is closed
1059 and the system is docked, if the laptop supports SW_DOCK
1060 notifications via the input layer. Note that ACPI docking
1061 stations do not generate this currently. Also note that this
1062 logic is usually not fully sufficient and Desktop
1063 Environments should take a lid switch inhibitor lock when an
1064 external display is connected, as systemd will not watch
1067 * nspawn will now make use of the devices cgroup controller by
1068 default, and only permit creation of and access to the usual
1069 API device nodes like /dev/null or /dev/random, as well as
1070 access to (but not creation of) the pty devices.
1072 * We will now ship a default .network file for
1073 systemd-networkd that automatically configures DHCP for
1074 network interfaces created by nspawn's --network-veth or
1075 --network-bridge= switches.
1077 * systemd will now understand the usual M, K, G, T suffixes
1078 according to SI conventions (i.e. to the base 1000) when
1079 referring to throughput and hardware metrics. It will stay
1080 with IEC conventions (i.e. to the base 1024) for software
1081 metrics, according to what is customary according to
1082 Wikipedia. We explicitly document which base applies for
1083 each configuration option.
1085 * The DeviceAllow= setting in unit files now supports a syntax
1086 to whitelist an entire group of devices node majors at once,
1087 based on the /proc/devices listing. For example, with the
1088 string "char-pts", it is now possible to whitelist all
1089 current and future pseudo-TTYs at once.
1091 * sd-event learned a new "post" event source. Event sources of
1092 this type are triggered by the dispatching of any event
1093 source of a type that is not "post". This is useful for
1094 implementing clean-up and check event sources that are
1095 triggered by other work being done in the program.
1097 * systemd-networkd is no longer statically enabled, but uses
1098 the usual [Install] sections so that it can be
1099 enabled/disabled using systemctl. It still is enabled by
1102 * When creating a veth interface pair with systemd-nspawn, the
1103 host side will now be prefixed with "vb-" if
1104 --network-bridge= is used, and with "ve-" if --network-veth
1105 is used. This way, it is easy to distinguish these cases on
1106 the host, for example to apply different configuration to
1107 them with systemd-networkd.
1109 * The compatibility libraries for libsystemd-journal.so,
1110 libsystem-id128.so, libsystemd-login.so and
1111 libsystemd-daemon.so do not make use of IFUNC
1112 anymore. Instead, we now build libsystemd.so multiple times
1113 under these alternative names. This means that the footprint
1114 is drastically increased, but given that these are
1115 transitional compatibility libraries, this should not matter
1116 much. This change has been made necessary to support the ARM
1117 platform for these compatibility libraries, as the ARM
1118 toolchain is not really at the same level as the toolchain
1119 for other architectures like x86 and does not support
1120 IFUNC. Please make sure to use --enable-compat-libs only
1121 during a transitional period!
1123 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1124 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1125 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1126 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1127 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1128 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1129 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1130 Zbigniew Jędrzejewski-Szmek
1132 -- Berlin, 2014-02-24
1136 * A new component "systemd-networkd" has been added that can
1137 be used to configure local network interfaces statically or
1138 via DHCP. It is capable of bringing up bridges, VLANs, and
1139 bonding. Currently, no hook-ups for interactive network
1140 configuration are provided. Use this for your initrd,
1141 container, embedded, or server setup if you need a simple,
1142 yet powerful, network configuration solution. This
1143 configuration subsystem is quite nifty, as it allows wildcard
1144 hotplug matching in interfaces. For example, with a single
1145 configuration snippet, you can configure that all Ethernet
1146 interfaces showing up are automatically added to a bridge,
1147 or similar. It supports link-sensing and more.
1149 * A new tool "systemd-socket-proxyd" has been added which can
1150 act as a bidirectional proxy for TCP sockets. This is
1151 useful for adding socket activation support to services that
1152 do not actually support socket activation, including virtual
1153 machines and the like.
1155 * Add a new tool to save/restore rfkill state on
1158 * Save/restore state of keyboard backlights in addition to
1159 display backlights on shutdown/boot.
1161 * udev learned a new SECLABEL{} construct to label device
1162 nodes with a specific security label when they appear. For
1163 now, only SECLABEL{selinux} is supported, but the syntax is
1164 prepared for additional security frameworks.
1166 * udev gained a new scheme to configure link-level attributes
1167 from files in /etc/systemd/network/*.link. These files can
1168 match against MAC address, device path, driver name and type,
1169 and will apply attributes like the naming policy, link speed,
1170 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1171 address assignment policy (randomized, ...).
1173 * The configuration of network interface naming rules for
1174 "permanent interface names" has changed: a new NamePolicy=
1175 setting in the [Link] section of .link files determines the
1176 priority of possible naming schemes (onboard, slot, mac,
1177 path). The default value of this setting is determined by
1178 /usr/lib/net/links/99-default.link. Old
1179 80-net-name-slot.rules udev configuration file has been
1180 removed, so local configuration overriding this file should
1181 be adapated to override 99-default.link instead.
1183 * When the User= switch is used in a unit file, also
1184 initialize $SHELL= based on the user database entry.
1186 * systemd no longer depends on libdbus. All communication is
1187 now done with sd-bus, systemd's low-level bus library
1190 * kdbus support has been added to PID 1 itself. When kdbus is
1191 enabled, this causes PID 1 to set up the system bus and
1192 enable support for a new ".busname" unit type that
1193 encapsulates bus name activation on kdbus. It works a little
1194 bit like ".socket" units, except for bus names. A new
1195 generator has been added that converts classic dbus1 service
1196 activation files automatically into native systemd .busname
1199 * sd-bus: add a light-weight vtable implementation that allows
1200 defining objects on the bus with a simple static const
1201 vtable array of its methods, signals and properties.
1203 * systemd will not generate or install static dbus
1204 introspection data anymore to /usr/share/dbus-1/interfaces,
1205 as the precise format of these files is unclear, and
1206 nothing makes use of it.
1208 * A proxy daemon is now provided to proxy clients connecting
1209 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1210 compatibility with classic D-Bus.
1212 * A bus driver implementation has been added that supports the
1213 classic D-Bus bus driver calls on kdbus, also for
1214 compatibility purposes.
1216 * A new API "sd-event.h" has been added that implements a
1217 minimal event loop API built around epoll. It provides a
1218 couple of features that direct epoll usage is lacking:
1219 prioritization of events, scales to large numbers of timer
1220 events, per-event timer slack (accuracy), system-wide
1221 coalescing of timer events, exit handlers, watchdog
1222 supervision support using systemd's sd_notify() API, child
1225 * A new API "sd-rntl.h" has been added that provides an API
1226 around the route netlink interface of the kernel, similar in
1227 style to "sd-bus.h".
1229 * A new API "sd-dhcp-client.h" has been added that provides a
1230 small DHCPv4 client-side implementation. This is used by
1233 * There is a new kernel command line option
1234 "systemd.restore_state=0|1". When set to "0", none of the
1235 systemd tools will restore saved runtime state to hardware
1236 devices. More specifically, the rfkill and backlight states
1239 * The FsckPassNo= compatibility option in mount/service units
1240 has been removed. The fstab generator will now add the
1241 necessary dependencies automatically, and does not require
1242 PID1's support for that anymore.
1244 * journalctl gained a new switch, --list-boots, that lists
1245 recent boots with their times and boot IDs.
1247 * The various tools like systemctl, loginctl, timedatectl,
1248 busctl, systemd-run, ... have gained a new switch "-M" to
1249 connect to a specific, local OS container (as direct
1250 connection, without requiring SSH). This works on any
1251 container that is registered with machined, such as those
1252 created by libvirt-lxc or nspawn.
1254 * systemd-run and systemd-analyze also gained support for "-H"
1255 to connect to remote hosts via SSH. This is particularly
1256 useful for systemd-run because it enables queuing of jobs
1257 onto remote systems.
1259 * machinectl gained a new command "login" to open a getty
1260 login in any local container. This works with any container
1261 that is registered with machined (such as those created by
1262 libvirt-lxc or nspawn), and which runs systemd inside.
1264 * machinectl gained a new "reboot" command that may be used to
1265 trigger a reboot on a specific container that is registered
1266 with machined. This works on any container that runs an init
1267 system of some kind.
1269 * systemctl gained a new "list-timers" command to print a nice
1270 listing of installed timer units with the times they elapse
1273 * Alternative reboot() parameters may now be specified on the
1274 "systemctl reboot" command line and are passed to the
1275 reboot() system call.
1277 * systemctl gained a new --job-mode= switch to configure the
1278 mode to queue a job with. This is a more generic version of
1279 --fail, --irreversible, and --ignore-dependencies, which are
1280 still available but not advertised anymore.
1282 * /etc/systemd/system.conf gained new settings to configure
1283 various default timeouts of units, as well as the default
1284 start limit interval and burst. These may still be overridden
1287 * PID1 will now export on the bus profile data of the security
1288 policy upload process (such as the SELinux policy upload to
1291 * journald: when forwarding logs to the console, include
1292 timestamps (following the setting in
1293 /sys/module/printk/parameters/time).
1295 * OnCalendar= in timer units now understands the special
1296 strings "yearly" and "annually". (Both are equivalent)
1298 * The accuracy of timer units is now configurable with the new
1299 AccuracySec= setting. It defaults to 1min.
1301 * A new dependency type JoinsNamespaceOf= has been added that
1302 allows running two services within the same /tmp and network
1303 namespace, if PrivateNetwork= or PrivateTmp= are used.
1305 * A new command "cat" has been added to systemctl. It outputs
1306 the original unit file of a unit, and concatenates the
1307 contents of additional "drop-in" unit file snippets, so that
1308 the full configuration is shown.
1310 * systemctl now supports globbing on the various "list-xyz"
1311 commands, like "list-units" or "list-sockets", as well as on
1312 those commands which take multiple unit names.
1314 * journalctl's --unit= switch gained support for globbing.
1316 * All systemd daemons now make use of the watchdog logic so
1317 that systemd automatically notices when they hang.
1319 * If the $container_ttys environment variable is set,
1320 getty-generator will automatically spawn a getty for each
1321 listed tty. This is useful for container managers to request
1322 login gettys to be spawned on as many ttys as needed.
1324 * %h, %s, %U specifier support is not available anymore when
1325 used in unit files for PID 1. This is because NSS calls are
1326 not safe from PID 1. They stay available for --user
1327 instances of systemd, and as special case for the root user.
1329 * loginctl gained a new "--no-legend" switch to turn off output
1332 * The "sd-login.h" API gained three new calls:
1333 sd_session_is_remote(), sd_session_get_remote_user(),
1334 sd_session_get_remote_host() to query information about
1337 * The udev hardware database now also carries vendor/product
1338 information of SDIO devices.
1340 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1341 determine whether watchdog notifications are requested by
1344 * Socket-activated per-connection services now include a
1345 short description of the connection parameters in the
1348 * tmpfiles gained a new "--boot" option. When this is not used,
1349 only lines where the command character is not suffixed with
1350 "!" are executed. When this option is specified, those
1351 options are executed too. This partitions tmpfiles
1352 directives into those that can be safely executed at any
1353 time, and those which should be run only at boot (for
1354 example, a line that creates /run/nologin).
1356 * A new API "sd-resolve.h" has been added which provides a simple
1357 asynchronous wrapper around glibc NSS host name resolution
1358 calls, such as getaddrinfo(). In contrast to glibc's
1359 getaddrinfo_a(), it does not use signals. In contrast to most
1360 other asynchronous name resolution libraries, this one does
1361 not reimplement DNS, but reuses NSS, so that alternate
1362 host name resolution systems continue to work, such as mDNS,
1363 LDAP, etc. This API is based on libasyncns, but it has been
1364 cleaned up for inclusion in systemd.
1366 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1367 "sd-daemon.h" are no longer found in individual libraries
1368 libsystemd-journal.so, libsystemd-login.so,
1369 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1370 merged them into a single library, libsystemd.so, which
1371 provides all symbols. The reason for this is cyclic
1372 dependencies, as these libraries tend to use each other's
1373 symbols. So far, we have managed to workaround that by linking
1374 a copy of a good part of our code into each of these
1375 libraries again and again, which, however, makes certain
1376 things hard to do, like sharing static variables. Also, it
1377 substantially increases footprint. With this change, there
1378 is only one library for the basic APIs systemd
1379 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1380 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1381 library as well, however are subject to the --enable-kdbus
1382 switch (see below). Note that "sd-dhcp-client.h" is not part
1383 of this library (this is because it only consumes, never
1384 provides, services of/to other APIs). To make the transition
1385 easy from the separate libraries to the unified one, we
1386 provide the --enable-compat-libs compile-time switch which
1387 will generate stub libraries that are compatible with the
1388 old ones but redirect all calls to the new one.
1390 * All of the kdbus logic and the new APIs "sd-bus.h",
1391 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1392 and "sd-utf8.h" are compile-time optional via the
1393 "--enable-kdbus" switch, and they are not compiled in by
1394 default. To make use of kdbus, you have to explicitly enable
1395 the switch. Note however, that neither the kernel nor the
1396 userspace API for all of this is considered stable yet. We
1397 want to maintain the freedom to still change the APIs for
1398 now. By specifying this build-time switch, you acknowledge
1399 that you are aware of the instability of the current
1402 * Also, note that while kdbus is pretty much complete,
1403 it lacks one thing: proper policy support. This means you
1404 can build a fully working system with all features; however,
1405 it will be highly insecure. Policy support will be added in
1406 one of the next releases, at the same time that we will
1407 declare the APIs stable.
1409 * When the kernel command-line argument "kdbus" is specified,
1410 systemd will automatically load the kdbus.ko kernel module. At
1411 this stage of development, it is only useful for testing kdbus
1412 and should not be used in production. Note: if "--enable-kdbus"
1413 is specified, and the kdbus.ko kernel module is available, and
1414 "kdbus" is added to the kernel command line, the entire system
1415 runs with kdbus instead of dbus-daemon, with the above mentioned
1416 problem of missing the system policy enforcement. Also a future
1417 version of kdbus.ko or a newer systemd will not be compatible with
1418 each other, and will unlikely be able to boot the machine if only
1419 one of them is updated.
1421 * systemctl gained a new "import-environment" command which
1422 uploads the caller's environment (or parts thereof) into the
1423 service manager so that it is inherited by services started
1424 by the manager. This is useful to upload variables like
1425 $DISPLAY into the user service manager.
1427 * A new PrivateDevices= switch has been added to service units
1428 which allows running a service with a namespaced /dev
1429 directory that does not contain any device nodes for
1430 physical devices. More specifically, it only includes devices
1431 such as /dev/null, /dev/urandom, and /dev/zero which are API
1434 * logind has been extended to support behaviour like VT
1435 switching on seats that do not support a VT. This makes
1436 multi-session available on seats that are not the first seat
1437 (seat0), and on systems where kernel support for VTs has
1438 been disabled at compile-time.
1440 * If a process holds a delay lock for system sleep or shutdown
1441 and fails to release it in time, we will now log its
1442 identity. This makes it easier to identify processes that
1443 cause slow suspends or power-offs.
1445 * When parsing /etc/crypttab, support for a new key-slot=
1446 option as supported by Debian is added. It allows indicating
1447 which LUKS slot to use on disk, speeding up key loading.
1449 * The sd_journald_sendv() API call has been checked and
1450 officially declared to be async-signal-safe so that it may
1451 be invoked from signal handlers for logging purposes.
1453 * Boot-time status output is now enabled automatically after a
1454 short timeout if boot does not progress, in order to give
1455 the user an indication what she or he is waiting for.
1457 * The boot-time output has been improved to show how much time
1458 remains until jobs expire.
1460 * The KillMode= switch in service units gained a new possible
1461 value "mixed". If set, and the unit is shut down, then the
1462 initial SIGTERM signal is sent only to the main daemon
1463 process, while the following SIGKILL signal is sent to
1464 all remaining processes of the service.
1466 * When a scope unit is registered, a new property "Controller"
1467 may be set. If set to a valid bus name, systemd will send a
1468 RequestStop() signal to this name when it would like to shut
1469 down the scope. This may be used to hook manager logic into
1470 the shutdown logic of scope units. Also, scope units may now
1471 be put in a special "abandoned" state, in which case the
1472 manager process which created them takes no further
1473 responsibilities for it.
1475 * When reading unit files, systemd will now verify
1476 the access mode of these files, and warn about certain
1477 suspicious combinations. This has been added to make it
1478 easier to track down packaging bugs where unit files are
1479 marked executable or world-writable.
1481 * systemd-nspawn gained a new "--setenv=" switch to set
1482 container-wide environment variables. The similar option in
1483 systemd-activate was renamed from "--environment=" to
1484 "--setenv=" for consistency.
1486 * systemd-nspawn has been updated to create a new kdbus domain
1487 for each container that is invoked, thus allowing each
1488 container to have its own set of system and user buses,
1489 independent of the host.
1491 * systemd-nspawn gained a new --drop-capability= switch to run
1492 the container with less capabilities than the default. Both
1493 --drop-capability= and --capability= now take the special
1494 string "all" for dropping or keeping all capabilities.
1496 * systemd-nspawn gained new switches for executing containers
1497 with specific SELinux labels set.
1499 * systemd-nspawn gained a new --quiet switch to not generate
1500 any additional output but the container's own console
1503 * systemd-nspawn gained a new --share-system switch to run a
1504 container without PID namespacing enabled.
1506 * systemd-nspawn gained a new --register= switch to control
1507 whether the container is registered with systemd-machined or
1508 not. This is useful for containers that do not run full
1509 OS images, but only specific apps.
1511 * systemd-nspawn gained a new --keep-unit which may be used
1512 when invoked as the only program from a service unit, and
1513 results in registration of the unit service itself in
1514 systemd-machined, instead of a newly opened scope unit.
1516 * systemd-nspawn gained a new --network-interface= switch for
1517 moving arbitrary interfaces to the container. The new
1518 --network-veth switch creates a virtual Ethernet connection
1519 between host and container. The new --network-bridge=
1520 switch then allows assigning the host side of this virtual
1521 Ethernet connection to a bridge device.
1523 * systemd-nspawn gained a new --personality= switch for
1524 setting the kernel personality for the container. This is
1525 useful when running a 32-bit container on a 64-bit host. A
1526 similar option Personality= is now also available for service
1529 * logind will now also track a "Desktop" identifier for each
1530 session which encodes the desktop environment of it. This is
1531 useful for desktop environments that want to identify
1532 multiple running sessions of itself easily.
1534 * A new SELinuxContext= setting for service units has been
1535 added that allows setting a specific SELinux execution
1536 context for a service.
1538 * Most systemd client tools will now honour $SYSTEMD_LESS for
1539 settings of the "less" pager. By default, these tools will
1540 override $LESS to allow certain operations to work, such as
1541 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1542 influence this logic.
1544 * systemd's "seccomp" hook-up has been changed to make use of
1545 the libseccomp library instead of using its own
1546 implementation. This has benefits for portability among
1549 * For usage together with SystemCallFilter=, a new
1550 SystemCallErrorNumber= setting has been introduced that
1551 allows configuration of a system error number to be returned
1552 on filtered system calls, instead of immediately killing the
1553 process. Also, SystemCallArchitectures= has been added to
1554 limit access to system calls of a particular architecture
1555 (in order to turn off support for unused secondary
1556 architectures). There is also a global
1557 SystemCallArchitectures= setting in system.conf now to turn
1558 off support for non-native system calls system-wide.
1560 * systemd requires a kernel with a working name_to_handle_at(),
1561 please see the kernel config requirements in the README file.
1563 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1564 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1565 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1566 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1567 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1568 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1569 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1570 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1571 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1572 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1573 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1574 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1575 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1576 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1577 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1578 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1579 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1580 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1581 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1582 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1583 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1584 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1585 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1586 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1588 -- Berlin, 2014-02-20
1592 * logind has gained support for facilitating privileged input
1593 and drm device access for unprivileged clients. This work is
1594 useful to allow Wayland display servers (and similar
1595 programs, such as kmscon) to run under the user's ID and
1596 access input and drm devices which are normally
1597 protected. When this is used (and the kernel is new enough)
1598 logind will "mute" IO on the file descriptors passed to
1599 Wayland as long as it is in the background and "unmute" it
1600 if it returns into the foreground. This allows secure
1601 session switching without allowing background sessions to
1602 eavesdrop on input and display data. This also introduces
1603 session switching support if VT support is turned off in the
1604 kernel, and on seats that are not seat0.
1606 * A new kernel command line option luks.options= is understood
1607 now which allows specifiying LUKS options for usage for LUKS
1608 encrypted partitions specified with luks.uuid=.
1610 * tmpfiles.d(5) snippets may now use specifier expansion in
1611 path names. More specifically %m, %b, %H, %v, are now
1612 replaced by the local machine id, boot id, hostname, and
1613 kernel version number.
1615 * A new tmpfiles.d(5) command "m" has been introduced which
1616 may be used to change the owner/group/access mode of a file
1617 or directory if it exists, but do nothing if it does not.
1619 * This release removes high-level support for the
1620 MemorySoftLimit= cgroup setting. The underlying kernel
1621 cgroup attribute memory.soft_limit= is currently badly
1622 designed and likely to be removed from the kernel API in its
1623 current form, hence we should not expose it for now.
1625 * The memory.use_hierarchy cgroup attribute is now enabled for
1626 all cgroups systemd creates in the memory cgroup
1627 hierarchy. This option is likely to be come the built-in
1628 default in the kernel anyway, and the non-hierarchial mode
1629 never made much sense in the intrinsically hierarchial
1632 * A new field _SYSTEMD_SLICE= is logged along with all journal
1633 messages containing the slice a message was generated
1634 from. This is useful to allow easy per-customer filtering of
1635 logs among other things.
1637 * systemd-journald will no longer adjust the group of journal
1638 files it creates to the "systemd-journal" group. Instead we
1639 rely on the journal directory to be owned by the
1640 "systemd-journal" group, and its setgid bit set, so that the
1641 kernel file system layer will automatically enforce that
1642 journal files inherit this group assignment. The reason for
1643 this change is that we cannot allow NSS look-ups from
1644 journald which would be necessary to resolve
1645 "systemd-journal" to a numeric GID, because this might
1646 create deadlocks if NSS involves synchronous queries to
1647 other daemons (such as nscd, or sssd) which in turn are
1648 logging clients of journald and might block on it, which
1649 would then dead lock. A tmpfiles.d(5) snippet included in
1650 systemd will make sure the setgid bit and group are
1651 properly set on the journal directory if it exists on every
1652 boot. However, we recommend adjusting it manually after
1653 upgrades too (or from RPM scriptlets), so that the change is
1654 not delayed until next reboot.
1656 * Backlight and random seed files in /var/lib/ have moved into
1657 the /var/lib/systemd/ directory, in order to centralize all
1658 systemd generated files in one directory.
1660 * Boot time performance measurements (as displayed by
1661 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1662 performance information if that's available to determine how
1663 much time BIOS and boot loader initialization required. With
1664 a sufficiently new BIOS you hence no longer need to boot
1665 with Gummiboot to get access to such information.
1667 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1668 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1669 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1670 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1671 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1672 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1673 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1675 -- Berlin, 2013-10-02
1679 * The Restart= option for services now understands a new
1680 on-watchdog setting, which will restart the service
1681 automatically if the service stops sending out watchdog keep
1682 alive messages (as configured with WatchdogSec=).
1684 * The getty generator (which is responsible for bringing up a
1685 getty on configured serial consoles) will no longer only
1686 start a getty on the primary kernel console but on all
1687 others, too. This makes the order in which console= is
1688 specified on the kernel command line less important.
1690 * libsystemd-logind gained a new sd_session_get_vt() call to
1691 retrieve the VT number of a session.
1693 * If the option "tries=0" is set for an entry of /etc/crypttab
1694 its passphrase is queried indefinitely instead of any
1695 maximum number of tries.
1697 * If a service with a configure PID file terminates its PID
1698 file will now be removed automatically if it still exists
1699 afterwards. This should put an end to stale PID files.
1701 * systemd-run will now also take relative binary path names
1702 for execution and no longer insists on absolute paths.
1704 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1705 paths that are optionally prefixed with "-" to indicate that
1706 it should not be considered a failure if they do not exist.
1708 * journalctl -o (and similar commands) now understands a new
1709 output mode "short-precise", it is similar to "short" but
1710 shows timestamps with usec accuracy.
1712 * The option "discard" (as known from Debian) is now
1713 synonymous to "allow-discards" in /etc/crypttab. In fact,
1714 "discard" is preferred now (since it is easier to remember
1717 * Some licensing clean-ups were made, so that more code is now
1718 LGPL-2.1 licensed than before.
1720 * A minimal tool to save/restore the display backlight
1721 brightness across reboots has been added. It will store the
1722 backlight setting as late as possible at shutdown, and
1723 restore it as early as possible during reboot.
1725 * A logic to automatically discover and enable home and swap
1726 partitions on GPT disks has been added. With this in place
1727 /etc/fstab becomes optional for many setups as systemd can
1728 discover certain partitions located on the root disk
1729 automatically. Home partitions are recognized under their
1730 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1731 partitions are recognized under their GPT type ID
1732 0657fd6da4ab43c484e50933c84b4f4f.
1734 * systemd will no longer pass any environment from the kernel
1735 or initrd to system services. If you want to set an
1736 environment for all services, do so via the kernel command
1737 line systemd.setenv= assignment.
1739 * The systemd-sysctl tool no longer natively reads the file
1740 /etc/sysctl.conf. If desired, the file should be symlinked
1741 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1742 legacy support by a symlink rather than built-in code, it
1743 also makes the otherwise hidden order of application of the
1744 different files visible. (Note that this partly reverts to a
1745 pre-198 application order of sysctl knobs!)
1747 * The "systemctl set-log-level" and "systemctl dump" commands
1748 have been moved to systemd-analyze.
1750 * systemd-run learned the new --remain-after-exit switch,
1751 which causes the scope unit not to be cleaned up
1752 automatically after the process terminated.
1754 * tmpfiles learned a new --exclude-prefix= switch to exclude
1755 certain paths from operation.
1757 * journald will now automatically flush all messages to disk
1758 as soon as a message of the log priorities CRIT, ALERT or
1761 Contributions from: Andrew Cook, Brandon Philips, Christian
1762 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1763 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1764 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1765 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1766 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1767 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1768 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1769 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1770 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1771 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1772 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1773 William Giokas, Zbigniew Jędrzejewski-Szmek
1775 -- Berlin, 2013-09-13
1779 * The documentation has been updated to cover the various new
1780 concepts introduced with 205.
1782 * Unit files now understand the new %v specifier which
1783 resolves to the kernel version string as returned by "uname
1786 * systemctl now supports filtering the unit list output by
1787 load state, active state and sub state, using the new
1790 * "systemctl status" will now show the results of the
1791 condition checks (like ConditionPathExists= and similar) of
1792 the last start attempts of the unit. They are also logged to
1795 * "journalctl -b" may now be used to look for boot output of a
1796 specific boot. Try "journalctl -b -1" for the previous boot,
1797 but the syntax is substantially more powerful.
1799 * "journalctl --show-cursor" has been added which prints the
1800 cursor string the last shown log line. This may then be used
1801 with the new "journalctl --after-cursor=" switch to continue
1802 browsing logs from that point on.
1804 * "journalctl --force" may now be used to force regeneration
1807 * Creation of "dead" device nodes has been moved from udev
1808 into kmod and tmpfiles. Previously, udev would read the kmod
1809 databases to pre-generate dead device nodes based on meta
1810 information contained in kernel modules, so that these would
1811 be auto-loaded on access rather then at boot. As this
1812 does not really have much to do with the exposing actual
1813 kernel devices to userspace this has always been slightly
1814 alien in the udev codebase. Following the new scheme kmod
1815 will now generate a runtime snippet for tmpfiles from the
1816 module meta information and it now is tmpfiles' job to the
1817 create the nodes. This also allows overriding access and
1818 other parameters for the nodes using the usual tmpfiles
1819 facilities. As side effect this allows us to remove the
1820 CAP_SYS_MKNOD capability bit from udevd entirely.
1822 * logind's device ACLs may now be applied to these "dead"
1823 devices nodes too, thus finally allowing managed access to
1824 devices such as /dev/snd/sequencer whithout loading the
1825 backing module right-away.
1827 * A new RPM macro has been added that may be used to apply
1828 tmpfiles configuration during package installation.
1830 * systemd-detect-virt and ConditionVirtualization= now can
1831 detect User-Mode-Linux machines (UML).
1833 * journald will now implicitly log the effective capabilities
1834 set of processes in the message metadata.
1836 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1838 * The initrd interface has been simplified (more specifically,
1839 support for passing performance data via environment
1840 variables and fsck results via files in /run has been
1841 removed). These features were non-essential, and are
1842 nowadays available in a much nicer way by having systemd in
1843 the initrd serialize its state and have the hosts systemd
1844 deserialize it again.
1846 * The udev "keymap" data files and tools to apply keyboard
1847 specific mappings of scan to key codes, and force-release
1848 scan code lists have been entirely replaced by a udev
1849 "keyboard" builtin and a hwdb data file.
1851 * systemd will now honour the kernel's "quiet" command line
1852 argument also during late shutdown, resulting in a
1853 completely silent shutdown when used.
1855 * There's now an option to control the SO_REUSEPORT socket
1856 option in .socket units.
1858 * Instance units will now automatically get a per-template
1859 subslice of system.slice unless something else is explicitly
1860 configured. For example, instances of sshd@.service will now
1861 implicitly be placed in system-sshd.slice rather than
1862 system.slice as before.
1864 * Test coverage support may now be enabled at build time.
1866 Contributions from: Dave Reisner, Frederic Crozat, Harald
1867 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1868 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1869 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1870 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1871 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1872 Giokas, Zbigniew Jędrzejewski-Szmek
1874 -- Berlin, 2013-07-23
1878 * Two new unit types have been introduced:
1880 Scope units are very similar to service units, however, are
1881 created out of pre-existing processes -- instead of PID 1
1882 forking off the processes. By using scope units it is
1883 possible for system services and applications to group their
1884 own child processes (worker processes) in a powerful way
1885 which then maybe used to organize them, or kill them
1886 together, or apply resource limits on them.
1888 Slice units may be used to partition system resources in an
1889 hierarchial fashion and then assign other units to them. By
1890 default there are now three slices: system.slice (for all
1891 system services), user.slice (for all user sessions),
1892 machine.slice (for VMs and containers).
1894 Slices and scopes have been introduced primarily in
1895 context of the work to move cgroup handling to a
1896 single-writer scheme, where only PID 1
1897 creates/removes/manages cgroups.
1899 * There's a new concept of "transient" units. In contrast to
1900 normal units these units are created via an API at runtime,
1901 not from configuration from disk. More specifically this
1902 means it is now possible to run arbitrary programs as
1903 independent services, with all execution parameters passed
1904 in via bus APIs rather than read from disk. Transient units
1905 make systemd substantially more dynamic then it ever was,
1906 and useful as a general batch manager.
1908 * logind has been updated to make use of scope and slice units
1909 for managing user sessions. As a user logs in he will get
1910 his own private slice unit, to which all sessions are added
1911 as scope units. We also added support for automatically
1912 adding an instance of user@.service for the user into the
1913 slice. Effectively logind will no longer create cgroup
1914 hierarchies on its own now, it will defer entirely to PID 1
1915 for this by means of scope, service and slice units. Since
1916 user sessions this way become entities managed by PID 1
1917 the output of "systemctl" is now a lot more comprehensive.
1919 * A new mini-daemon "systemd-machined" has been added which
1920 may be used by virtualization managers to register local
1921 VMs/containers. nspawn has been updated accordingly, and
1922 libvirt will be updated shortly. machined will collect a bit
1923 of meta information about the VMs/containers, and assign
1924 them their own scope unit (see above). The collected
1925 meta-data is then made available via the "machinectl" tool,
1926 and exposed in "ps" and similar tools. machined/machinectl
1927 is compile-time optional.
1929 * As discussed earlier, the low-level cgroup configuration
1930 options ControlGroup=, ControlGroupModify=,
1931 ControlGroupPersistent=, ControlGroupAttribute= have been
1932 removed. Please use high-level attribute settings instead as
1933 well as slice units.
1935 * A new bus call SetUnitProperties() has been added to alter
1936 various runtime parameters of a unit. This is primarily
1937 useful to alter cgroup parameters dynamically in a nice way,
1938 but will be extended later on to make more properties
1939 modifiable at runtime. systemctl gained a new set-properties
1940 command that wraps this call.
1942 * A new tool "systemd-run" has been added which can be used to
1943 run arbitrary command lines as transient services or scopes,
1944 while configuring a number of settings via the command
1945 line. This tool is currently very basic, however already
1946 very useful. We plan to extend this tool to even allow
1947 queuing of execution jobs with time triggers from the
1948 command line, similar in fashion to "at".
1950 * nspawn will now inform the user explicitly that kernels with
1951 audit enabled break containers, and suggest the user to turn
1954 * Support for detecting the IMA and AppArmor security
1955 frameworks with ConditionSecurity= has been added.
1957 * journalctl gained a new "-k" switch for showing only kernel
1958 messages, mimicking dmesg output; in addition to "--user"
1959 and "--system" switches for showing only user's own logs
1962 * systemd-delta can now show information about drop-in
1963 snippets extending unit files.
1965 * libsystemd-bus has been substantially updated but is still
1966 not available as public API.
1968 * systemd will now look for the "debug" argument on the kernel
1969 command line and enable debug logging, similar to what
1970 "systemd.log_level=debug" already did before.
1972 * "systemctl set-default", "systemctl get-default" has been
1973 added to configure the default.target symlink, which
1974 controls what to boot into by default.
1976 * "systemctl set-log-level" has been added as a convenient
1977 way to raise and lower systemd logging threshold.
1979 * "systemd-analyze plot" will now show the time the various
1980 generators needed for execution, as well as information
1981 about the unit file loading.
1983 * libsystemd-journal gained a new sd_journal_open_files() call
1984 for opening specific journal files. journactl also gained a
1985 new switch to expose this new functionality. Previously we
1986 only supported opening all files from a directory, or all
1987 files from the system, as opening individual files only is
1988 racy due to journal file rotation.
1990 * systemd gained the new DefaultEnvironment= setting in
1991 /etc/systemd/system.conf to set environment variables for
1994 * If a privileged process logs a journal message with the
1995 OBJECT_PID= field set, then journald will automatically
1996 augment this with additional OBJECT_UID=, OBJECT_GID=,
1997 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
1998 system services want to log events about specific client
1999 processes. journactl/systemctl has been updated to make use
2000 of this information if all log messages regarding a specific
2003 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2004 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2005 Reisner, David Coppa, David King, David Strauss, Eelco
2006 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2007 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2008 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2009 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2010 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2011 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2012 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2013 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2014 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2015 Łukasz Stelmach, 장동준
2019 * The Python bindings gained some minimal support for the APIs
2020 exposed by libsystemd-logind.
2022 * ConditionSecurity= gained support for detecting SMACK. Since
2023 this condition already supports SELinux and AppArmor we only
2024 miss IMA for this. Patches welcome!
2026 Contributions from: Karol Lewandowski, Lennart Poettering,
2027 Zbigniew Jędrzejewski-Szmek
2031 * systemd-nspawn will now create /etc/resolv.conf if
2032 necessary, before bind-mounting the host's file onto it.
2034 * systemd-nspawn will now store meta information about a
2035 container on the container's cgroup as extended attribute
2036 fields, including the root directory.
2038 * The cgroup hierarchy has been reworked in many ways. All
2039 objects any of the components systemd creates in the cgroup
2040 tree are now suffixed. More specifically, user sessions are
2041 now placed in cgroups suffixed with ".session", users in
2042 cgroups suffixed with ".user", and nspawn containers in
2043 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2044 names are now escaped in a simple scheme to avoid collision
2045 of userspace object names with kernel filenames. This work
2046 is preparation for making these objects relocatable in the
2047 cgroup tree, in order to allow easy resource partitioning of
2048 these objects without causing naming conflicts.
2050 * systemctl list-dependencies gained the new switches
2051 --plain, --reverse, --after and --before.
2053 * systemd-inhibit now shows the process name of processes that
2054 have taken an inhibitor lock.
2056 * nss-myhostname will now also resolve "localhost"
2057 implicitly. This makes /etc/hosts an optional file and
2058 nicely handles that on IPv6 ::1 maps to both "localhost" and
2061 * libsystemd-logind.so gained a new call
2062 sd_get_machine_names() to enumerate running containers and
2063 VMs (currently only supported by very new libvirt and
2064 nspawn). sd_login_monitor can now be used to watch
2065 VMs/containers coming and going.
2067 * .include is not allowed recursively anymore, and only in
2068 unit files. Usually it is better to use drop-in snippets in
2069 .d/*.conf anyway, as introduced with systemd 198.
2071 * systemd-analyze gained a new "critical-chain" command that
2072 determines the slowest chain of units run during system
2073 boot-up. It is very useful for tracking down where
2074 optimizing boot time is the most beneficial.
2076 * systemd will no longer allow manipulating service paths in
2077 the name=systemd:/system cgroup tree using ControlGroup= in
2078 units. (But is still fine with it in all other dirs.)
2080 * There's a new systemd-nspawn@.service service file that may
2081 be used to easily run nspawn containers as system
2082 services. With the container's root directory in
2083 /var/lib/container/foobar it is now sufficient to run
2084 "systemctl start systemd-nspawn@foobar.service" to boot it.
2086 * systemd-cgls gained a new parameter "--machine" to list only
2087 the processes within a certain container.
2089 * ConditionSecurity= now can check for "apparmor". We still
2090 are lacking checks for SMACK and IMA for this condition
2091 check though. Patches welcome!
2093 * A new configuration file /etc/systemd/sleep.conf has been
2094 added that may be used to configure which kernel operation
2095 systemd is supposed to execute when "suspend", "hibernate"
2096 or "hybrid-sleep" is requested. This makes the new kernel
2097 "freeze" state accessible to the user.
2099 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2100 the passed argument if applicable.
2102 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2103 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2104 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2105 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2106 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2107 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2108 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2113 * The output of 'systemctl list-jobs' got some polishing. The
2114 '--type=' argument may now be passed more than once. A new
2115 command 'systemctl list-sockets' has been added which shows
2116 a list of kernel sockets systemd is listening on with the
2117 socket units they belong to, plus the units these socket
2120 * The experimental libsystemd-bus library got substantial
2121 updates to work in conjunction with the (also experimental)
2122 kdbus kernel project. It works well enough to exchange
2123 messages with some sophistication. Note that kdbus is not
2124 ready yet, and the library is mostly an elaborate test case
2125 for now, and not installable.
2127 * systemd gained a new unit 'systemd-static-nodes.service'
2128 that generates static device nodes earlier during boot, and
2129 can run in conjunction with udev.
2131 * libsystemd-login gained a new call sd_pid_get_user_unit()
2132 to retrieve the user systemd unit a process is running
2133 in. This is useful for systems where systemd is used as
2136 * systemd-nspawn now places all containers in the new /machine
2137 top-level cgroup directory in the name=systemd
2138 hierarchy. libvirt will soon do the same, so that we get a
2139 uniform separation of /system, /user and /machine for system
2140 services, user processes and containers/virtual
2141 machines. This new cgroup hierarchy is also useful to stick
2142 stable names to specific container instances, which can be
2143 recognized later this way (this name may be controlled
2144 via systemd-nspawn's new -M switch). libsystemd-login also
2145 gained a new call sd_pid_get_machine_name() to retrieve the
2146 name of the container/VM a specific process belongs to.
2148 * bootchart can now store its data in the journal.
2150 * libsystemd-journal gained a new call
2151 sd_journal_add_conjunction() for AND expressions to the
2152 matching logic. This can be used to express more complex
2153 logical expressions.
2155 * journactl can now take multiple --unit= and --user-unit=
2158 * The cryptsetup logic now understands the "luks.key=" kernel
2159 command line switch for specifying a file to read the
2160 decryption key from. Also, if a configured key file is not
2161 found the tool will now automatically fall back to prompting
2164 * Python systemd.journal module was updated to wrap recently
2165 added functions from libsystemd-journal. The interface was
2166 changed to bring the low level interface in s.j._Reader
2167 closer to the C API, and the high level interface in
2168 s.j.Reader was updated to wrap and convert all data about
2171 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2172 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2173 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2174 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2175 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2176 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2180 * journalctl --update-catalog now understands a new --root=
2181 option to operate on catalogs found in a different root
2184 * During shutdown after systemd has terminated all running
2185 services a final killing loop kills all remaining left-over
2186 processes. We will now print the name of these processes
2187 when we send SIGKILL to them, since this usually indicates a
2190 * If /etc/crypttab refers to password files stored on
2191 configured mount points automatic dependencies will now be
2192 generated to ensure the specific mount is established first
2193 before the key file is attempted to be read.
2195 * 'systemctl status' will now show information about the
2196 network sockets a socket unit is listening on.
2198 * 'systemctl status' will also shown information about any
2199 drop-in configuration file for units. (Drop-In configuration
2200 files in this context are files such as
2201 /etc/systemd/systemd/foobar.service.d/*.conf)
2203 * systemd-cgtop now optionally shows summed up CPU times of
2204 cgroups. Press '%' while running cgtop to switch between
2205 percentage and absolute mode. This is useful to determine
2206 which cgroups use up the most CPU time over the entire
2207 runtime of the system. systemd-cgtop has also been updated
2208 to be 'pipeable' for processing with further shell tools.
2210 * 'hostnamectl set-hostname' will now allow setting of FQDN
2213 * The formatting and parsing of time span values has been
2214 changed. The parser now understands fractional expressions
2215 such as "5.5h". The formatter will now output fractional
2216 expressions for all time spans under 1min, i.e. "5.123456s"
2217 rather than "5s 123ms 456us". For time spans under 1s
2218 millisecond values are shown, for those under 1ms
2219 microsecond values are shown. This should greatly improve
2220 all time-related output of systemd.
2222 * libsystemd-login and libsystemd-journal gained new
2223 functions for querying the poll() events mask and poll()
2224 timeout value for integration into arbitrary event
2227 * localectl gained the ability to list available X11 keymaps
2228 (models, layouts, variants, options).
2230 * 'systemd-analyze dot' gained the ability to filter for
2231 specific units via shell-style globs, to create smaller,
2232 more useful graphs. I.e. it is now possible to create simple
2233 graphs of all the dependencies between only target units, or
2234 of all units that Avahi has dependencies with.
2236 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2237 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2238 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2239 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2240 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2241 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2242 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2246 * The boot-time readahead implementation for rotating media
2247 will now read the read-ahead data in multiple passes which
2248 consist of all read requests made in equidistant time
2249 intervals. This means instead of strictly reading read-ahead
2250 data in its physical order on disk we now try to find a
2251 middle ground between physical and access time order.
2253 * /etc/os-release files gained a new BUILD_ID= field for usage
2254 on operating systems that provide continuous builds of OS
2257 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2258 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2259 William Douglas, Zbigniew Jędrzejewski-Szmek
2263 * systemd-python gained an API exposing libsystemd-daemon.
2265 * The SMACK setup logic gained support for uploading CIPSO
2268 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2269 ReadOnlyDirectories= and InaccessibleDirectories= has
2270 changed. The private /tmp and /var/tmp directories are now
2271 shared by all processes of a service (which means
2272 ExecStartPre= may now leave data in /tmp that ExecStart= of
2273 the same service can still access). When a service is
2274 stopped its temporary directories are immediately deleted
2275 (normal clean-up with tmpfiles is still done in addition to
2278 * By default, systemd will now set a couple of sysctl
2279 variables in the kernel: the safe sysrq options are turned
2280 on, IP route verification is turned on, and source routing
2281 disabled. The recently added hardlink and softlink
2282 protection of the kernel is turned on. These settings should
2283 be reasonably safe, and good defaults for all new systems.
2285 * The predictable network naming logic may now be turned off
2286 with a new kernel command line switch: net.ifnames=0.
2288 * A new libsystemd-bus module has been added that implements a
2289 pretty complete D-Bus client library. For details see:
2291 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2293 * journald will now explicitly flush the journal files to disk
2294 at the latest 5min after each write. The file will then also
2295 be marked offline until the next write. This should increase
2296 reliability in case of a crash. The synchronization delay
2297 can be configured via SyncIntervalSec= in journald.conf.
2299 * There's a new remote-fs-setup.target unit that can be used
2300 to pull in specific services when at least one remote file
2301 system is to be mounted.
2303 * There are new targets timers.target and paths.target as
2304 canonical targets to pull user timer and path units in
2305 from. This complements sockets.target with a similar
2306 purpose for socket units.
2308 * libudev gained a new call udev_device_set_attribute_value()
2309 to set sysfs attributes of a device.
2311 * The udev daemon now sets the default number of worker
2312 processes executed in parallel based on the number of available
2313 CPUs instead of the amount of available RAM. This is supposed
2314 to provide a more reliable default and limit a too aggressive
2315 paralellism for setups with 1000s of devices connected.
2317 Contributions from: Auke Kok, Colin Walters, Cristian
2318 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2319 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2320 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2321 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2322 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2323 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2324 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2325 Zbigniew Jędrzejewski-Szmek
2329 * Configuration of unit files may now be extended via drop-in
2330 files without having to edit/override the unit files
2331 themselves. More specifically, if the administrator wants to
2332 change one value for a service file foobar.service he can
2333 now do so by dropping in a configuration snippet into
2334 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2335 will load all these snippets and apply them on top of the
2336 main unit configuration file, possibly extending or
2337 overriding its settings. Using these drop-in snippets is
2338 generally nicer than the two earlier options for changing
2339 unit files locally: copying the files from
2340 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2341 them there; or creating a new file in /etc/systemd/system/
2342 that incorporates the original one via ".include". Drop-in
2343 snippets into these .d/ directories can be placed in any
2344 directory systemd looks for units in, and the usual
2345 overriding semantics between /usr/lib, /etc and /run apply
2348 * Most unit file settings which take lists of items can now be
2349 reset by assigning the empty string to them. For example,
2350 normally, settings such as Environment=FOO=BAR append a new
2351 environment variable assignment to the environment block,
2352 each time they are used. By assigning Environment= the empty
2353 string the environment block can be reset to empty. This is
2354 particularly useful with the .d/*.conf drop-in snippets
2355 mentioned above, since this adds the ability to reset list
2356 settings from vendor unit files via these drop-ins.
2358 * systemctl gained a new "list-dependencies" command for
2359 listing the dependencies of a unit recursively.
2361 * Inhibitors are now honored and listed by "systemctl
2362 suspend", "systemctl poweroff" (and similar) too, not only
2363 GNOME. These commands will also list active sessions by
2366 * Resource limits (as exposed by the various control group
2367 controllers) can now be controlled dynamically at runtime
2368 for all units. More specifically, you can now use a command
2369 like "systemctl set-cgroup-attr foobar.service cpu.shares
2370 2000" to alter the CPU shares a specific service gets. These
2371 settings are stored persistently on disk, and thus allow the
2372 administrator to easily adjust the resource usage of
2373 services with a few simple commands. This dynamic resource
2374 management logic is also available to other programs via the
2375 bus. Almost any kernel cgroup attribute and controller is
2378 * systemd-vconsole-setup will now copy all font settings to
2379 all allocated VTs, where it previously applied them only to
2382 * libsystemd-login gained the new sd_session_get_tty() API
2385 * This release drops support for a few legacy or
2386 distribution-specific LSB facility names when parsing init
2387 scripts: $x-display-manager, $mail-transfer-agent,
2388 $mail-transport-agent, $mail-transfer-agent, $smtp,
2389 $null. Also, the mail-transfer-agent.target unit backing
2390 this has been removed. Distributions which want to retain
2391 compatibility with this should carry the burden for
2392 supporting this themselves and patch support for these back
2393 in, if they really need to. Also, the facilities $syslog and
2394 $local_fs are now ignored, since systemd does not support
2395 early-boot LSB init scripts anymore, and these facilities
2396 are implied anyway for normal services. syslog.target has
2399 * There are new bus calls on PID1's Manager object for
2400 cancelling jobs, and removing snapshot units. Previously,
2401 both calls were only available on the Job and Snapshot
2404 * systemd-journal-gatewayd gained SSL support.
2406 * The various "environment" files, such as /etc/locale.conf
2407 now support continuation lines with a backslash ("\") as
2408 last character in the line, similarly in style (but different)
2409 to how this is supported in shells.
2411 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2412 now implicitly appended to every log entry logged. systemctl
2413 has been updated to filter by this field when operating on a
2414 user systemd instance.
2416 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2417 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2418 the container. This makes it easier to boot unmodified
2419 Fedora systems in a container, which however still requires
2420 audit=0 to be passed on the kernel command line. Auditing in
2421 kernel and userspace is unfortunately still too broken in
2422 context of containers, hence we recommend compiling it out
2423 of the kernel or using audit=0. Hopefully this will be fixed
2424 one day for good in the kernel.
2426 * nspawn gained the new --bind= and --bind-ro= parameters to
2427 bind mount specific directories from the host into the
2430 * nspawn will now mount its own devpts file system instance
2431 into the container, in order not to leak pty devices from
2432 the host into the container.
2434 * systemd will now read the firmware boot time performance
2435 information from the EFI variables, if the used boot loader
2436 supports this, and takes it into account for boot performance
2437 analysis via "systemd-analyze". This is currently supported
2438 only in conjunction with Gummiboot, but could be supported
2439 by other boot loaders too. For details see:
2441 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2443 * A new generator has been added that automatically mounts the
2444 EFI System Partition (ESP) to /boot, if that directory
2445 exists, is empty, and no other file system has been
2446 configured to be mounted there.
2448 * logind will now send out PrepareForSleep(false) out
2449 unconditionally, after coming back from suspend. This may be
2450 used by applications as asynchronous notification for
2451 system resume events.
2453 * "systemctl unlock-sessions" has been added, that allows
2454 unlocking the screens of all user sessions at once, similar
2455 to how "systemctl lock-sessions" already locked all users
2456 sessions. This is backed by a new D-Bus call UnlockSessions().
2458 * "loginctl seat-status" will now show the master device of a
2459 seat. (i.e. the device of a seat that needs to be around for
2460 the seat to be considered available, usually the graphics
2463 * tmpfiles gained a new "X" line type, that allows
2464 configuration of files and directories (with wildcards) that
2465 shall be excluded from automatic cleanup ("aging").
2467 * udev default rules set the device node permissions now only
2468 at "add" events, and do not change them any longer with a
2469 later "change" event.
2471 * The log messages for lid events and power/sleep keypresses
2472 now carry a message ID.
2474 * We now have a substantially larger unit test suite, but this
2475 continues to be work in progress.
2477 * udevadm hwdb gained a new --root= parameter to change the
2478 root directory to operate relative to.
2480 * logind will now issue a background sync() request to the kernel
2481 early at shutdown, so that dirty buffers are flushed to disk early
2482 instead of at the last moment, in order to optimize shutdown
2485 * A new bootctl tool has been added that is an interface for
2486 certain boot loader operations. This is currently a preview
2487 and is likely to be extended into a small mechanism daemon
2488 like timedated, localed, hostnamed, and can be used by
2489 graphical UIs to enumerate available boot options, and
2490 request boot into firmware operations.
2492 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2493 the rest of the package. It also has been updated to work
2494 correctly in initrds.
2496 * Policykit previously has been runtime optional, and is now
2497 also compile time optional via a configure switch.
2499 * systemd-analyze has been reimplemented in C. Also "systemctl
2500 dot" has moved into systemd-analyze.
2502 * "systemctl status" with no further parameters will now print
2503 the status of all active or failed units.
2505 * Operations such as "systemctl start" can now be executed
2506 with a new mode "--irreversible" which may be used to queue
2507 operations that cannot accidentally be reversed by a later
2508 job queuing. This is by default used to make shutdown
2509 requests more robust.
2511 * The Python API of systemd now gained a new module for
2512 reading journal files.
2514 * A new tool kernel-install has been added that can install
2515 kernel images according to the Boot Loader Specification:
2517 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2519 * Boot time console output has been improved to provide
2520 animated boot time output for hanging jobs.
2522 * A new tool systemd-activate has been added which can be used
2523 to test socket activation with, directly from the command
2524 line. This should make it much easier to test and debug
2525 socket activation in daemons.
2527 * journalctl gained a new "--reverse" (or -r) option to show
2528 journal output in reverse order (i.e. newest line first).
2530 * journalctl gained a new "--pager-end" (or -e) option to jump
2531 to immediately jump to the end of the journal in the
2532 pager. This is only supported in conjunction with "less".
2534 * journalctl gained a new "--user-unit=" option, that works
2535 similarly to "--unit=" but filters for user units rather than
2538 * A number of unit files to ease adoption of systemd in
2539 initrds has been added. This moves some minimal logic from
2540 the various initrd implementations into systemd proper.
2542 * The journal files are now owned by a new group
2543 "systemd-journal", which exists specifically to allow access
2544 to the journal, and nothing else. Previously, we used the
2545 "adm" group for that, which however possibly covers more
2546 than just journal/log file access. This new group is now
2547 already used by systemd-journal-gatewayd to ensure this
2548 daemon gets access to the journal files and as little else
2549 as possible. Note that "make install" will also set FS ACLs
2550 up for /var/log/journal to give "adm" and "wheel" read
2551 access to it, in addition to "systemd-journal" which owns
2552 the journal files. We recommend that packaging scripts also
2553 add read access to "adm" + "wheel" to /var/log/journal, and
2554 all existing/future journal files. To normal users and
2555 administrators little changes, however packagers need to
2556 ensure to create the "systemd-journal" system group at
2557 package installation time.
2559 * The systemd-journal-gatewayd now runs as unprivileged user
2560 systemd-journal-gateway:systemd-journal-gateway. Packaging
2561 scripts need to create these system user/group at
2564 * timedated now exposes a new boolean property CanNTP that
2565 indicates whether a local NTP service is available or not.
2567 * systemd-detect-virt will now also detect xen PVs
2569 * The pstore file system is now mounted by default, if it is
2572 * In addition to the SELinux and IMA policies we will now also
2573 load SMACK policies at early boot.
2575 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2576 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2577 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2578 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2579 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2580 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2581 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2582 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2583 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2584 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2585 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2586 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2587 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2588 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2592 * Timer units now support calendar time events in addition to
2593 monotonic time events. That means you can now trigger a unit
2594 based on a calendar time specification such as "Thu,Fri
2595 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2596 or fifth day of any month of the year 2013, given that it is
2597 a thursday or friday. This brings timer event support
2598 considerably closer to cron's capabilities. For details on
2599 the supported calendar time specification language see
2602 * udev now supports a number of different naming policies for
2603 network interfaces for predictable names, and a combination
2604 of these policies is now the default. Please see this wiki
2605 document for details:
2607 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2609 * Auke Kok's bootchart implementation has been added to the
2610 systemd tree. It is an optional component that can graph the
2611 boot in quite some detail. It is one of the best bootchart
2612 implementations around and minimal in its code and
2615 * nss-myhostname has been integrated into the systemd source
2616 tree. nss-myhostname guarantees that the local hostname
2617 always stays resolvable via NSS. It has been a weak
2618 requirement of systemd-hostnamed since a long time, and
2619 since its code is actually trivial we decided to just
2620 include it in systemd's source tree. It can be turned off
2621 with a configure switch.
2623 * The read-ahead logic is now capable of properly detecting
2624 whether a btrfs file system is on SSD or rotating media, in
2625 order to optimize the read-ahead scheme. Previously, it was
2626 only capable of detecting this on traditional file systems
2629 * In udev, additional device properties are now read from the
2630 IAB in addition to the OUI database. Also, Bluetooth company
2631 identities are attached to the devices as well.
2633 * In service files %U may be used as specifier that is
2634 replaced by the configured user name of the service.
2636 * nspawn may now be invoked without a controlling TTY. This
2637 makes it suitable for invocation as its own service. This
2638 may be used to set up a simple containerized server system
2639 using only core OS tools.
2641 * systemd and nspawn can now accept socket file descriptors
2642 when they are started for socket activation. This enables
2643 implementation of socket activated nspawn
2644 containers. i.e. think about autospawning an entire OS image
2645 when the first SSH or HTTP connection is received. We expect
2646 that similar functionality will also be added to libvirt-lxc
2649 * journalctl will now suppress ANSI color codes when
2650 presenting log data.
2652 * systemctl will no longer show control group information for
2653 a unit if a the control group is empty anyway.
2655 * logind can now automatically suspend/hibernate/shutdown the
2658 * /etc/machine-info and hostnamed now also expose the chassis
2659 type of the system. This can be used to determine whether
2660 the local system is a laptop, desktop, handset or
2661 tablet. This information may either be configured by the
2662 user/vendor or is automatically determined from ACPI and DMI
2663 information if possible.
2665 * A number of PolicyKit actions are now bound together with
2666 "imply" rules. This should simplify creating UIs because
2667 many actions will now authenticate similar ones as well.
2669 * Unit files learnt a new condition ConditionACPower= which
2670 may be used to conditionalize a unit depending on whether an
2671 AC power source is connected or not, of whether the system
2672 is running on battery power.
2674 * systemctl gained a new "is-failed" verb that may be used in
2675 shell scripts and suchlike to check whether a specific unit
2676 is in the "failed" state.
2678 * The EnvironmentFile= setting in unit files now supports file
2679 globbing, and can hence be used to easily read a number of
2680 environment files at once.
2682 * systemd will no longer detect and recognize specific
2683 distributions. All distribution-specific #ifdeffery has been
2684 removed, systemd is now fully generic and
2685 distribution-agnostic. Effectively, not too much is lost as
2686 a lot of the code is still accessible via explicit configure
2687 switches. However, support for some distribution specific
2688 legacy configuration file formats has been dropped. We
2689 recommend distributions to simply adopt the configuration
2690 files everybody else uses now and convert the old
2691 configuration from packaging scripts. Most distributions
2692 already did that. If that's not possible or desirable,
2693 distributions are welcome to forward port the specific
2694 pieces of code locally from the git history.
2696 * When logging a message about a unit systemd will now always
2697 log the unit name in the message meta data.
2699 * localectl will now also discover system locale data that is
2700 not stored in locale archives, but directly unpacked.
2702 * logind will no longer unconditionally use framebuffer
2703 devices as seat masters, i.e. as devices that are required
2704 to be existing before a seat is considered preset. Instead,
2705 it will now look for all devices that are tagged as
2706 "seat-master" in udev. By default framebuffer devices will
2707 be marked as such, but depending on local systems other
2708 devices might be marked as well. This may be used to
2709 integrate graphics cards using closed source drivers (such
2710 as NVidia ones) more nicely into logind. Note however, that
2711 we recommend using the open source NVidia drivers instead,
2712 and no udev rules for the closed-source drivers will be
2713 shipped from us upstream.
2715 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2716 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2717 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2718 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2719 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2720 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2721 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2722 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2723 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2724 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2725 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2726 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2731 * udev gained support for loading additional device properties
2732 from an indexed database that is keyed by vendor/product IDs
2733 and similar device identifiers. For the beginning this
2734 "hwdb" is populated with data from the well-known PCI and
2735 USB database, but also includes PNP, ACPI and OID data. In
2736 the longer run this indexed database shall grow into
2737 becoming the one central database for non-essential
2738 userspace device metadata. Previously, data from the PCI/USB
2739 database was only attached to select devices, since the
2740 lookup was a relatively expensive operation due to O(n) time
2741 complexity (with n being the number of entries in the
2742 database). Since this is now O(1), we decided to add in this
2743 data for all devices where this is available, by
2744 default. Note that the indexed database needs to be rebuilt
2745 when new data files are installed. To achieve this you need
2746 to update your packaging scripts to invoke "udevadm hwdb
2747 --update" after installation of hwdb data files. For
2748 RPM-based distributions we introduced the new
2749 %udev_hwdb_update macro for this purpose.
2751 * The Journal gained support for the "Message Catalog", an
2752 indexed database to link up additional information with
2753 journal entries. For further details please check:
2755 http://www.freedesktop.org/wiki/Software/systemd/catalog
2757 The indexed message catalog database also needs to be
2758 rebuilt after installation of message catalog files. Use
2759 "journalctl --update-catalog" for this. For RPM-based
2760 distributions we introduced the %journal_catalog_update
2761 macro for this purpose.
2763 * The Python Journal bindings gained support for the standard
2764 Python logging framework.
2766 * The Journal API gained new functions for checking whether
2767 the underlying file system of a journal file is capable of
2768 properly reporting file change notifications, or whether
2769 applications that want to reflect journal changes "live"
2770 need to recheck journal files continuously in appropriate
2773 * It is now possible to set the "age" field for tmpfiles
2774 entries to 0, indicating that files matching this entry
2775 shall always be removed when the directories are cleaned up.
2777 * coredumpctl gained a new "gdb" verb which invokes gdb
2778 right-away on the selected coredump.
2780 * There's now support for "hybrid sleep" on kernels that
2781 support this, in addition to "suspend" and "hibernate". Use
2782 "systemctl hybrid-sleep" to make use of this.
2784 * logind's HandleSuspendKey= setting (and related settings)
2785 now gained support for a new "lock" setting to simply
2786 request the screen lock on all local sessions, instead of
2787 actually executing a suspend or hibernation.
2789 * systemd will now mount the EFI variables file system by
2792 * Socket units now gained support for configuration of the
2793 SMACK security label.
2795 * timedatectl will now output the time of the last and next
2796 daylight saving change.
2798 * We dropped support for various legacy and distro-specific
2799 concepts, such as insserv, early-boot SysV services
2800 (i.e. those for non-standard runlevels such as 'b' or 'S')
2801 or ArchLinux /etc/rc.conf support. We recommend the
2802 distributions who still need support this to either continue
2803 to maintain the necessary patches downstream, or find a
2804 different solution. (Talk to us if you have questions!)
2806 * Various systemd components will now bypass PolicyKit checks
2807 for root and otherwise handle properly if PolicyKit is not
2808 found to be around. This should fix most issues for
2809 PolicyKit-less systems. Quite frankly this should have been
2810 this way since day one. It is absolutely our intention to
2811 make systemd work fine on PolicyKit-less systems, and we
2812 consider it a bug if something does not work as it should if
2813 PolicyKit is not around.
2815 * For embedded systems it is now possible to build udev and
2816 systemd without blkid and/or kmod support.
2818 * "systemctl switch-root" is now capable of switching root
2819 more than once. I.e. in addition to transitions from the
2820 initrd to the host OS it is now possible to transition to
2821 further OS images from the host. This is useful to implement
2822 offline updating tools.
2824 * Various other additions have been made to the RPM macros
2825 shipped with systemd. Use %udev_rules_update() after
2826 installing new udev rules files. %_udevhwdbdir,
2827 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2828 %_sysctldir are now available which resolve to the right
2829 directories for packages to place various data files in.
2831 * journalctl gained the new --full switch (in addition to
2832 --all, to disable ellipsation for long messages.
2834 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2835 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2836 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2837 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2838 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2839 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2840 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2841 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2842 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2846 * journalctl gained new --since= and --until= switches to
2847 filter by time. It also now supports nice filtering for
2848 units via --unit=/-u.
2850 * Type=oneshot services may use ExecReload= and do the
2853 * The journal daemon now supports time-based rotation and
2854 vacuuming, in addition to the usual disk-space based
2857 * The journal will now index the available field values for
2858 each field name. This enables clients to show pretty drop
2859 downs of available match values when filtering. The bash
2860 completion of journalctl has been updated
2861 accordingly. journalctl gained a new switch -F to list all
2862 values a certain field takes in the journal database.
2864 * More service events are now written as structured messages
2865 to the journal, and made recognizable via message IDs.
2867 * The timedated, localed and hostnamed mini-services which
2868 previously only provided support for changing time, locale
2869 and hostname settings from graphical DEs such as GNOME now
2870 also have a minimal (but very useful) text-based client
2871 utility each. This is probably the nicest way to changing
2872 these settings from the command line now, especially since
2873 it lists available options and is fully integrated with bash
2876 * There's now a new tool "systemd-coredumpctl" to list and
2877 extract coredumps from the journal.
2879 * We now install a README each in /var/log/ and
2880 /etc/rc.d/init.d explaining where the system logs and init
2881 scripts went. This hopefully should help folks who go to
2882 that dirs and look into the otherwise now empty void and
2883 scratch their heads.
2885 * When user-services are invoked (by systemd --user) the
2886 $MANAGERPID env var is set to the PID of systemd.
2888 * SIGRTMIN+24 when sent to a --user instance will now result
2889 in immediate termination of systemd.
2891 * gatewayd received numerous feature additions such as a
2892 "follow" mode, for live syncing and filtering.
2894 * browse.html now allows filtering and showing detailed
2895 information on specific entries. Keyboard navigation and
2896 mouse screen support has been added.
2898 * gatewayd/journalctl now supports HTML5/JSON
2899 Server-Sent-Events as output.
2901 * The SysV init script compatibility logic will now
2902 heuristically determine whether a script supports the
2903 "reload" verb, and only then make this available as
2906 * "systemctl status --follow" has been removed, use "journalctl
2909 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
2910 have been removed since they are hardly useful to be
2913 * And I'd like to take the opportunity to specifically mention
2914 Zbigniew for his great contributions. Zbigniew, you rock!
2916 Contributions from: Andrew Eikum, Christian Hesse, Colin
2917 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
2918 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
2919 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
2920 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
2921 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
2922 Jędrzejewski-Szmek, Сковорода Никита Андреевич
2926 * If /etc/vconsole.conf is non-existent or empty we will no
2927 longer load any console font or key map at boot by
2928 default. Instead the kernel defaults will be left
2929 intact. This is definitely the right thing to do, as no
2930 configuration should mean no configuration, and hard-coding
2931 font names that are different on all archs is probably a bad
2932 idea. Also, the kernel default key map and font should be
2933 good enough for most cases anyway, and mostly identical to
2934 the userspace fonts/key maps we previously overloaded them
2935 with. If distributions want to continue to default to a
2936 non-kernel font or key map they should ship a default
2937 /etc/vconsole.conf with the appropriate contents.
2939 Contributions from: Colin Walters, Daniel J Walsh, Dave
2940 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
2941 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2945 * journalctl gained a new --cursor= switch to show entries
2946 starting from the specified location in the journal.
2948 * We now enforce a size limit on journal entry fields exported
2949 with "-o json" in journalctl. Fields larger than 4K will be
2950 assigned null. This can be turned off with --all.
2952 * An (optional) journal gateway daemon is now available as
2953 "systemd-journal-gatewayd.service". This service provides
2954 access to the journal via HTTP and JSON. This functionality
2955 will be used to implement live log synchronization in both
2956 pull and push modes, but has various other users too, such
2957 as easy log access for debugging of embedded devices. Right
2958 now it is already useful to retrieve the journal via HTTP:
2960 # systemctl start systemd-journal-gatewayd.service
2961 # wget http://localhost:19531/entries
2963 This will download the journal contents in a
2964 /var/log/messages compatible format. The same as JSON:
2966 # curl -H"Accept: application/json" http://localhost:19531/entries
2968 This service is also accessible via a web browser where a
2969 single static HTML5 app is served that uses the JSON logic
2970 to enable the user to do some basic browsing of the
2971 journal. This will be extended later on. Here's an example
2972 screenshot of this app in its current state:
2974 http://0pointer.de/public/journal-gatewayd
2976 Contributions from: Kay Sievers, Lennart Poettering, Robert
2977 Milasan, Tom Gundersen
2981 * The bash completion logic is now available for journalctl
2984 * We do not mount the "cpuset" controller anymore together with
2985 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
2986 started if no parameters are assigned to it. "cpuset" hence
2987 broke code that assumed it it could create "cpu" groups and
2990 * journalctl -f will now subscribe to terminal size changes,
2991 and line break accordingly.
2993 Contributions from: Dave Reisner, Kay Sievers, Lennart
2994 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
2998 * nspawn will now create a symlink /etc/localtime in the
2999 container environment, copying the host's timezone
3000 setting. Previously this has been done via a bind mount, but
3001 since symlinks cannot be bind mounted this has now been
3002 changed to create/update the appropriate symlink.
3004 * journalctl -n's line number argument is now optional, and
3005 will default to 10 if omitted.
3007 * journald will now log the maximum size the journal files may
3008 take up on disk. This is particularly useful if the default
3009 built-in logic of determining this parameter from the file
3010 system size is used. Use "systemctl status
3011 systemd-journald.service" to see this information.
3013 * The multi-seat X wrapper tool has been stripped down. As X
3014 is now capable of enumerating graphics devices via udev in a
3015 seat-aware way the wrapper is not strictly necessary
3016 anymore. A stripped down temporary stop-gap is still shipped
3017 until the upstream display managers have been updated to
3018 fully support the new X logic. Expect this wrapper to be
3019 removed entirely in one of the next releases.
3021 * HandleSleepKey= in logind.conf has been split up into
3022 HandleSuspendKey= and HandleHibernateKey=. The old setting
3023 is not available anymore. X11 and the kernel are
3024 distuingishing between these keys and we should too. This
3025 also means the inhibition lock for these keys has been split
3028 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3029 Poettering, Lukas Nykryn, Václav Pavlín
3033 * Whenever a unit changes state we will now log this to the
3034 journal and show along the unit's own log output in
3037 * ConditionPathIsMountPoint= can now properly detect bind
3038 mount points too. (Previously, a bind mount of one file
3039 system to another place in the same file system could not be
3040 detected as mount, since they shared struct stat's st_dev
3043 * We will now mount the cgroup controllers cpu, cpuacct,
3044 cpuset and the controllers net_cls, net_prio together by
3047 * nspawn containers will now have a virtualized boot
3048 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3049 over with a randomized ID at container initialization). This
3050 has the effect of making "journalctl -b" do the right thing
3053 * The JSON output journal serialization has been updated not
3054 to generate "endless" list objects anymore, but rather one
3055 JSON object per line. This is more in line how most JSON
3056 parsers expect JSON objects. The new output mode
3057 "json-pretty" has been added to provide similar output, but
3058 neatly aligned for readability by humans.
3060 * We dropped all explicit sync() invocations in the shutdown
3061 code. The kernel does this implicitly anyway in the kernel
3062 reboot() syscall. halt(8)'s -n option is now a compatibility
3065 * We now support virtualized reboot() in containers, as
3066 supported by newer kernels. We will fall back to exit() if
3067 CAP_SYS_REBOOT is not available to the container. Also,
3068 nspawn makes use of this now and will actually reboot the
3069 container if the containerized OS asks for that.
3071 * journalctl will only show local log output by default
3072 now. Use --merge (-m) to show remote log output, too.
3074 * libsystemd-journal gained the new sd_journal_get_usage()
3075 call to determine the current disk usage of all journal
3076 files. This is exposed in the new "journalctl --disk-usage"
3079 * journald gained a new configuration setting SplitMode= in
3080 journald.conf which may be used to control how user journals
3081 are split off. See journald.conf(5) for details.
3083 * A new condition type ConditionFileNotEmpty= has been added.
3085 * tmpfiles' "w" lines now support file globbing, to write
3086 multiple files at once.
3088 * We added Python bindings for the journal submission
3089 APIs. More Python APIs for a number of selected APIs will
3090 likely follow. Note that we intend to add native bindings
3091 only for the Python language, as we consider it common
3092 enough to deserve bindings shipped within systemd. There are
3093 various projects outside of systemd that provide bindings
3094 for languages such as PHP or Lua.
3096 * Many conditions will now resolve specifiers such as %i. In
3097 addition, PathChanged= and related directives of .path units
3098 now support specifiers as well.
3100 * There's now a new RPM macro definition for the system preset
3103 * journald will now warn if it ca not forward a message to the
3104 syslog daemon because its socket is full.
3106 * timedated will no longer write or process /etc/timezone,
3107 except on Debian. As we do not support late mounted /usr
3108 anymore /etc/localtime always being a symlink is now safe,
3109 and hence the information in /etc/timezone is not necessary
3112 * logind will now always reserve one VT for a text getty (VT6
3113 by default). Previously if more than 6 X sessions where
3114 started they took up all the VTs with auto-spawned gettys,
3115 so that no text gettys were available anymore.
3117 * udev will now automatically inform the btrfs kernel logic
3118 about btrfs RAID components showing up. This should make
3119 simple hotplug based btrfs RAID assembly work.
3121 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3122 (but not for its children which will stay at the kernel
3123 default). This should allow setups with a lot more listening
3126 * systemd will now always pass the configured timezone to the
3127 kernel at boot. timedated will do the same when the timezone
3130 * logind's inhibition logic has been updated. By default,
3131 logind will now handle the lid switch, the power and sleep
3132 keys all the time, even in graphical sessions. If DEs want
3133 to handle these events on their own they should take the new
3134 handle-power-key, handle-sleep-key and handle-lid-switch
3135 inhibitors during their runtime. A simple way to achiveve
3136 that is to invoke the DE wrapped in an invocation of:
3138 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3140 * Access to unit operations is now checked via SELinux taking
3141 the unit file label and client process label into account.
3143 * systemd will now notify the administrator in the journal
3144 when he over-mounts a non-empty directory.
3146 * There are new specifiers that are resolved in unit files,
3147 for the host name (%H), the machine ID (%m) and the boot ID
3150 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3151 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3152 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3153 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3154 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3155 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3156 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3160 * Support for reading structured kernel messages from
3161 /dev/kmsg has now been added and is enabled by default.
3163 * Support for reading kernel messages from /proc/kmsg has now
3164 been removed. If you want kernel messages in the journal
3165 make sure to run a recent kernel (>= 3.5) that supports
3166 reading structured messages from /dev/kmsg (see
3167 above). /proc/kmsg is now exclusive property of classic
3168 syslog daemons again.
3170 * The libudev API gained the new
3171 udev_device_new_from_device_id() call.
3173 * The logic for file system namespace (ReadOnlyDirectory=,
3174 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3175 require pivot_root() anymore. This means fewer temporary
3176 directories are created below /tmp for this feature.
3178 * nspawn containers will now see and receive all submounts
3179 made on the host OS below the root file system of the
3182 * Forward Secure Sealing is now supported for Journal files,
3183 which provide cryptographical sealing of journal files so
3184 that attackers cannot alter log history anymore without this
3185 being detectable. Lennart will soon post a blog story about
3186 this explaining it in more detail.
3188 * There are two new service settings RestartPreventExitStatus=
3189 and SuccessExitStatus= which allow configuration of exit
3190 status (exit code or signal) which will be excepted from the
3191 restart logic, resp. consider successful.
3193 * journalctl gained the new --verify switch that can be used
3194 to check the integrity of the structure of journal files and
3195 (if Forward Secure Sealing is enabled) the contents of
3198 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3199 and similar symlinks pre-created. This makes running shells
3200 as container init process a lot more fun.
3202 * The fstab support can now handle PARTUUID= and PARTLABEL=
3205 * A new ConditionHost= condition has been added to match
3206 against the hostname (with globs) and machine ID. This is
3207 useful for clusters where a single OS image is used to
3208 provision a large number of hosts which shall run slightly
3209 different sets of services.
3211 * Services which hit the restart limit will now be placed in a
3214 Contributions from: Bertram Poettering, Dave Reisner, Huang
3215 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3216 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3220 * When running in --user mode systemd will now become a
3221 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3222 tree a lot more organized.
3224 * A new PartOf= unit dependency type has been introduced that
3225 may be used to group services in a natural way.
3227 * "systemctl enable" may now be used to enable instances of
3230 * journalctl now prints error log levels in red, and
3231 warning/notice log levels in bright white. It also supports
3232 filtering by log level now.
3234 * cgtop gained a new -n switch (similar to top), to configure
3235 the maximum number of iterations to run for. It also gained
3236 -b, to run in batch mode (accepting no input).
3238 * The suffix ".service" may now be omitted on most systemctl
3239 command lines involving service unit names.
3241 * There's a new bus call in logind to lock all sessions, as
3242 well as a loginctl verb for it "lock-sessions".
3244 * libsystemd-logind.so gained a new call sd_journal_perror()
3245 that works similar to libc perror() but logs to the journal
3246 and encodes structured information about the error number.
3248 * /etc/crypttab entries now understand the new keyfile-size=
3251 * shutdown(8) now can send a (configurable) wall message when
3252 a shutdown is cancelled.
3254 * The mount propagation mode for the root file system will now
3255 default to "shared", which is useful to make containers work
3256 nicely out-of-the-box so that they receive new mounts from
3257 the host. This can be undone locally by running "mount
3258 --make-rprivate /" if needed.
3260 * The prefdm.service file has been removed. Distributions
3261 should maintain this unit downstream if they intend to keep
3262 it around. However, we recommend writing normal unit files
3263 for display managers instead.
3265 * Since systemd is a crucial part of the OS we will now
3266 default to a number of compiler switches that improve
3267 security (hardening) such as read-only relocations, stack
3268 protection, and suchlike.
3270 * The TimeoutSec= setting for services is now split into
3271 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3272 of individual time outs for the start and the stop phase of
3275 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3276 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3277 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3278 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3279 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3280 Gundersen, Zbigniew Jędrzejewski-Szmek
3284 * The journal and id128 C APIs are now fully documented as man
3287 * Extra safety checks have been added when transitioning from
3288 the initial RAM disk to the main system to avoid accidental
3291 * /etc/crypttab entries now understand the new keyfile-offset=
3294 * systemctl -t can now be used to filter by unit load state.
3296 * The journal C API gained the new sd_journal_wait() call to
3297 make writing synchronous journal clients easier.
3299 * journalctl gained the new -D switch to show journals from a
3302 * journalctl now displays a special marker between log
3303 messages of two different boots.
3305 * The journal is now explicitly flushed to /var via a service
3306 systemd-journal-flush.service, rather than implicitly simply
3307 by seeing /var/log/journal to be writable.
3309 * journalctl (and the journal C APIs) can now match for much
3310 more complex expressions, with alternatives and
3313 * When transitioning from the initial RAM disk to the main
3314 system we will now kill all processes in a killing spree to
3315 ensure no processes stay around by accident.
3317 * Three new specifiers may be used in unit files: %u, %h, %s
3318 resolve to the user name, user home directory resp. user
3319 shell. This is useful for running systemd user instances.
3321 * We now automatically rotate journal files if their data
3322 object hash table gets a fill level > 75%. We also size the
3323 hash table based on the configured maximum file size. This
3324 together should lower hash collisions drastically and thus
3325 speed things up a bit.
3327 * journalctl gained the new "--header" switch to introspect
3328 header data of journal files.
3330 * A new setting SystemCallFilters= has been added to services
3331 which may be used to apply blacklists or whitelists to
3332 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3334 * nspawn gained a new --link-journal= switch (and quicker: -j)
3335 to link the container journal with the host. This makes it
3336 very easy to centralize log viewing on the host for all
3337 guests while still keeping the journal files separated.
3339 * Many bugfixes and optimizations
3341 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3342 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3343 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3348 * Several tools now understand kernel command line arguments,
3349 which are only read when run in an initial RAM disk. They
3350 usually follow closely their normal counterparts, but are
3353 * There's a new tool to analyze the readahead files that are
3354 automatically generated at boot. Use:
3356 /usr/lib/systemd/systemd-readahead analyze /.readahead
3358 * We now provide an early debug shell on tty9 if this enabled. Use:
3360 systemctl enable debug-shell.service
3362 * All plymouth related units have been moved into the Plymouth
3363 package. Please make sure to upgrade your Plymouth version
3366 * systemd-tmpfiles now supports getting passed the basename of
3367 a configuration file only, in which case it will look for it
3368 in all appropriate directories automatically.
3370 * udevadm info now takes a /dev or /sys path as argument, and
3371 does the right thing. Example:
3373 udevadm info /dev/sda
3374 udevadm info /sys/class/block/sda
3376 * systemctl now prints a warning if a unit is stopped but a
3377 unit that might trigger it continues to run. Example: a
3378 service is stopped but the socket that activates it is left
3381 * "systemctl status" will now mention if the log output was
3382 shortened due to rotation since a service has been started.
3384 * The journal API now exposes functions to determine the
3385 "cutoff" times due to rotation.
3387 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3388 immediately flushing of runtime logs to /var if possible,
3389 resp. for triggering immediate rotation of the journal
3392 * It is now considered an error if a service is attempted to
3393 be stopped that is not loaded.
3395 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3397 * systemd-analyze now supports Python 3
3399 * tmpfiles now supports cleaning up directories via aging
3400 where the first level dirs are always kept around but
3401 directories beneath it automatically aged. This is enabled
3402 by prefixing the age field with '~'.
3404 * Seat objects now expose CanGraphical, CanTTY properties
3405 which is required to deal with very fast bootups where the
3406 display manager might be running before the graphics drivers
3407 completed initialization.
3409 * Seat objects now expose a State property.
3411 * We now include RPM macros for service enabling/disabling
3412 based on the preset logic. We recommend RPM based
3413 distributions to make use of these macros if possible. This
3414 makes it simpler to reuse RPM spec files across
3417 * We now make sure that the collected systemd unit name is
3418 always valid when services log to the journal via
3421 * There's a new man page kernel-command-line(7) detailing all
3422 command line options we understand.
3424 * The fstab generator may now be disabled at boot by passing
3425 fstab=0 on the kernel command line.
3427 * A new kernel command line option modules-load= is now understood
3428 to load a specific kernel module statically, early at boot.
3430 * Unit names specified on the systemctl command line are now
3431 automatically escaped as needed. Also, if file system or
3432 device paths are specified they are automatically turned
3433 into the appropriate mount or device unit names. Example:
3435 systemctl status /home
3436 systemctl status /dev/sda
3438 * The SysVConsole= configuration option has been removed from
3439 system.conf parsing.
3441 * The SysV search path is no longer exported on the D-Bus
3444 * The Names= option is been removed from unit file parsing.
3446 * There's a new man page bootup(7) detailing the boot process.
3448 * Every unit and every generator we ship with systemd now
3449 comes with full documentation. The self-explanatory boot is
3452 * A couple of services gained "systemd-" prefixes in their
3453 name if they wrap systemd code, rather than only external
3454 code. Among them fsck@.service which is now
3455 systemd-fsck@.service.
3457 * The HaveWatchdog property has been removed from the D-Bus
3460 * systemd.confirm_spawn= on the kernel command line should now
3463 * There's a new man page crypttab(5) which details all options
3464 we actually understand.
3466 * systemd-nspawn gained a new --capability= switch to pass
3467 additional capabilities to the container.
3469 * timedated will now read known NTP implementation unit names
3470 from /usr/lib/systemd/ntp-units.d/*.list,
3471 systemd-timedated-ntp.target has been removed.
3473 * journalctl gained a new switch "-b" that lists log data of
3474 the current boot only.
3476 * The notify socket is in the abstract namespace again, in
3477 order to support daemons which chroot() at start-up.
3479 * There is a new Storage= configuration option for journald
3480 which allows configuration of where log data should go. This
3481 also provides a way to disable journal logging entirely, so
3482 that data collected is only forwarded to the console, the
3483 kernel log buffer or another syslog implementation.
3485 * Many bugfixes and optimizations
3487 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3488 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3489 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3490 Shawn Landden, Tom Gundersen
3494 * "systemctl help <unit>" now shows the man page if one is
3497 * Several new man pages have been added.
3499 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3500 MaxLevelConsole= can now be specified in
3501 journald.conf. These options allow reducing the amount of
3502 data stored on disk or forwarded by the log level.
3504 * TimerSlackNSec= can now be specified in system.conf for
3505 PID1. This allows system-wide power savings.
3507 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3508 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3513 * logind is now capable of (optionally) handling power and
3514 sleep keys as well as the lid switch.
3516 * journalctl now understands the syntax "journalctl
3517 /usr/bin/avahi-daemon" to get all log output of a specific
3520 * CapabilityBoundingSet= in system.conf now also influences
3521 the capability bound set of usermode helpers of the kernel.
3523 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3524 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3525 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3526 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3530 * Note that we skipped 139 releases here in order to set the
3531 new version to something that is greater than both udev's
3532 and systemd's most recent version number.
3534 * udev: all udev sources are merged into the systemd source tree now.
3535 All future udev development will happen in the systemd tree. It
3536 is still fully supported to use the udev daemon and tools without
3537 systemd running, like in initramfs or other init systems. Building
3538 udev though, will require the *build* of the systemd tree, but
3539 udev can be properly *run* without systemd.
3541 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3542 should be used to create dead device nodes as workarounds for broken
3545 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3546 no longer supported. udev_monitor_new_from_netlink() needs to be
3547 used to subscribe to events.
3549 * udev: when udevd is started by systemd, processes which are left
3550 behind by forking them off of udev rules, are unconditionally cleaned
3551 up and killed now after the event handling has finished. Services or
3552 daemons must be started as systemd services. Services can be
3553 pulled-in by udev to get started, but they can no longer be directly
3554 forked by udev rules.
3556 * udev: the daemon binary is called systemd-udevd now and installed
3557 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3558 to adapt to that, create symlink, or rename the binary after building
3561 * libudev no longer provides these symbols:
3562 udev_monitor_from_socket()
3563 udev_queue_get_failed_list_entry()
3564 udev_get_{dev,sys,run}_path()
3565 The versions number was bumped and symbol versioning introduced.
3567 * systemd-loginctl and systemd-journalctl have been renamed
3568 to loginctl and journalctl to match systemctl.
3570 * The config files: /etc/systemd/systemd-logind.conf and
3571 /etc/systemd/systemd-journald.conf have been renamed to
3572 logind.conf and journald.conf. Package updates should rename
3573 the files to the new names on upgrade.
3575 * For almost all files the license is now LGPL2.1+, changed
3576 from the previous GPL2.0+. Exceptions are some minor stuff
3577 of udev (which will be changed to LGPL2.1 eventually, too),
3578 and the MIT licensed sd-daemon.[ch] library that is suitable
3579 to be used as drop-in files.
3581 * systemd and logind now handle system sleep states, in
3582 particular suspending and hibernating.
3584 * logind now implements a sleep/shutdown/idle inhibiting logic
3585 suitable for a variety of uses. Soonishly Lennart will blog
3586 about this in more detail.
3588 * var-run.mount and var-lock.mount are no longer provided
3589 (which prevously bind mounted these directories to their new
3590 places). Distributions which have not converted these
3591 directories to symlinks should consider stealing these files
3592 from git history and add them downstream.
3594 * We introduced the Documentation= field for units and added
3595 this to all our shipped units. This is useful to make it
3596 easier to explore the boot and the purpose of the various
3599 * All smaller setup units (such as
3600 systemd-vconsole-setup.service) now detect properly if they
3601 are run in a container and are skipped when
3602 appropriate. This guarantees an entirely noise-free boot in
3603 Linux container environments such as systemd-nspawn.
3605 * A framework for implementing offline system updates is now
3606 integrated, for details see:
3607 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3609 * A new service type Type=idle is available now which helps us
3610 avoiding ugly interleaving of getty output and boot status
3613 * There's now a system-wide CapabilityBoundingSet= option to
3614 globally reduce the set of capabilities for the
3615 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3616 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3617 even CAP_NET_ADMIN system-wide for secure systems.
3619 * There are now system-wide DefaultLimitXXX= options to
3620 globally change the defaults of the various resource limits
3621 for all units started by PID 1.
3623 * Harald Hoyer's systemd test suite has been integrated into
3624 systemd which allows easy testing of systemd builds in qemu
3625 and nspawn. (This is really awesome! Ask us for details!)
3627 * The fstab parser is now implemented as generator, not inside
3630 * systemctl will now warn you if .mount units generated from
3631 /etc/fstab are out of date due to changes in fstab that
3632 have not been read by systemd yet.
3634 * systemd is now suitable for usage in initrds. Dracut has
3635 already been updated to make use of this. With this in place
3636 initrds get a slight bit faster but primarily are much
3637 easier to introspect and debug since "systemctl status" in
3638 the host system can be used to introspect initrd services,
3639 and the journal from the initrd is kept around too.
3641 * systemd-delta has been added, a tool to explore differences
3642 between user/admin configuration and vendor defaults.
3644 * PrivateTmp= now affects both /tmp and /var/tmp.
3646 * Boot time status messages are now much prettier and feature
3647 proper english language. Booting up systemd has never been
3650 * Read-ahead pack files now include the inode number of all
3651 files to pre-cache. When the inode changes the pre-caching
3652 is not attempted. This should be nicer to deal with updated
3653 packages which might result in changes of read-ahead
3656 * We now temporaritly lower the kernel's read_ahead_kb variable
3657 when collecting read-ahead data to ensure the kernel's
3658 built-in read-ahead does not add noise to our measurements
3659 of necessary blocks to pre-cache.
3661 * There's now RequiresMountsFor= to add automatic dependencies
3662 for all mounts necessary for a specific file system path.
3664 * MountAuto= and SwapAuto= have been removed from
3665 system.conf. Mounting file systems at boot has to take place
3668 * nspawn now learned a new switch --uuid= to set the machine
3669 ID on the command line.
3671 * nspawn now learned the -b switch to automatically search
3674 * vt102 is now the default TERM for serial TTYs, upgraded from
3677 * systemd-logind now works on VT-less systems.
3679 * The build tree has been reorganized. The individual
3680 components now have directories of their own.
3682 * A new condition type ConditionPathIsReadWrite= is now available.
3684 * nspawn learned the new -C switch to create cgroups for the
3685 container in other hierarchies.
3687 * We now have support for hardware watchdogs, configurable in
3690 * The scheduled shutdown logic now has a public API.
3692 * We now mount /tmp as tmpfs by default, but this can be
3693 masked and /etc/fstab can override it.
3695 * Since udisks does not make use of /media anymore we are not
3696 mounting a tmpfs on it anymore.
3698 * journalctl gained a new --local switch to only interleave
3699 locally generated journal files.
3701 * We can now load the IMA policy at boot automatically.
3703 * The GTK tools have been split off into a systemd-ui.
3705 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3706 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3707 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3708 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3709 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3710 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3711 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3712 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3713 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3718 * This is mostly a bugfix release
3720 * Support optional initialization of the machine ID from the
3721 KVM or container configured UUID.
3723 * Support immediate reboots with "systemctl reboot -ff"
3725 * Show /etc/os-release data in systemd-analyze output
3727 * Many bugfixes for the journal, including endianness fixes and
3728 ensuring that disk space enforcement works
3730 * sd-login.h is C++ comptaible again
3732 * Extend the /etc/os-release format on request of the Debian
3735 * We now refuse non-UTF8 strings used in various configuration
3736 and unit files. This is done to ensure we do not pass invalid
3737 data over D-Bus or expose it elsewhere.
3739 * Register Mimo USB Screens as suitable for automatic seat
3742 * Read SELinux client context from journal clients in a race
3745 * Reorder configuration file lookup order. /etc now always
3746 overrides /run in order to allow the administrator to always
3747 and unconditionally override vendor supplied or
3748 automatically generated data.
3750 * The various user visible bits of the journal now have man
3751 pages. We still lack man pages for the journal API calls
3754 * We now ship all man pages in HTML format again in the
3757 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3758 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3759 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3764 * This is mostly a bugfix release
3766 * systems lacking /etc/os-release are no longer supported.
3768 * Various functionality updates to libsystemd-login.so
3770 * Track class of PAM logins to distuingish greeters from
3773 Contributions from: Kay Sievers, Lennart Poettering, Michael
3778 * This is an important bugfix release for v41.
3780 * Building man pages is now optional which should be useful
3781 for those building systemd from git but unwilling to install
3784 * Watchdog support for supervising services is now usable. In
3785 a future release support for hardware watchdogs
3786 (i.e. /dev/watchdog) will be added building on this.
3788 * Service start rate limiting is now configurable and can be
3789 turned off per service. When a start rate limit is hit a
3790 reboot can automatically be triggered.
3792 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3794 Contributions from: Benjamin Franzke, Bill Nottingham,
3795 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3796 Schmidt, Michał Górny, Piotr Drąg
3800 * The systemd binary is installed /usr/lib/systemd/systemd now;
3801 An existing /sbin/init symlink needs to be adapted with the
3804 * The code that loads kernel modules has been ported to invoke
3805 libkmod directly, instead of modprobe. This means we do not
3806 support systems with module-init-tools anymore.
3808 * Watchdog support is now already useful, but still not
3811 * A new kernel command line option systemd.setenv= is
3812 understood to set system wide environment variables
3813 dynamically at boot.
3815 * We now limit the set of capabilities of systemd-journald.
3817 * We now set SIGPIPE to ignore by default, since it only is
3818 useful in shell pipelines, and has little use in general
3819 code. This can be disabled with IgnoreSIPIPE=no in unit
3822 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3823 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3828 * This is mostly a bugfix release
3830 * We now expose the reason why a service failed in the
3831 "Result" D-Bus property.
3833 * Rudimentary service watchdog support (will be completed over
3834 the next few releases.)
3836 * When systemd forks off in order execute some service we will
3837 now immediately changes its argv[0] to reflect which process
3838 it will execute. This is useful to minimize the time window
3839 with a generic argv[0], which makes bootcharts more useful
3841 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3842 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3843 Mike Kazantsev, Ray Strode
3847 * This is mostly a test release, but incorporates many
3850 * New systemd-cgtop tool to show control groups by their
3853 * Linking against libacl for ACLs is optional again. If
3854 disabled, support tracking device access for active logins
3855 goes becomes unavailable, and so does access to the user
3856 journals by the respective users.
3858 * If a group "adm" exists, journal files are automatically
3859 owned by them, thus allow members of this group full access
3860 to the system journal as well as all user journals.
3862 * The journal now stores the SELinux context of the logging
3863 client for all entries.
3865 * Add C++ inclusion guards to all public headers
3867 * New output mode "cat" in the journal to print only text
3868 messages, without any meta data like date or time.
3870 * Include tiny X server wrapper as a temporary stop-gap to
3871 teach XOrg udev display enumeration. This is used by display
3872 managers such as gdm, and will go away as soon as XOrg
3873 learned native udev hotplugging for display devices.
3875 * Add new systemd-cat tool for executing arbitrary programs
3876 with STDERR/STDOUT connected to the journal. Can also act as
3877 BSD logger replacement, and does so by default.
3879 * Optionally store all locally generated coredumps in the
3880 journal along with meta data.
3882 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3883 writing short strings to files (for usage for /sys), and for
3884 creating symlinks, character and block device nodes.
3886 * New unit file option ControlGroupPersistent= to make cgroups
3887 persistent, following the mechanisms outlined in
3888 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3890 * Support multiple local RTCs in a sane way
3892 * No longer monopolize IO when replaying readahead data on
3893 rotating disks, since we might starve non-file-system IO to
3894 death, since fanotify() will not see accesses done by blkid,
3897 * Do not show kernel threads in systemd-cgls anymore, unless
3898 requested with new -k switch.
3900 Contributions from: Dan Horák, Kay Sievers, Lennart
3901 Poettering, Michal Schmidt
3905 * This is mostly a test release, but incorporates many
3908 * The git repository moved to:
3909 git://anongit.freedesktop.org/systemd/systemd
3910 ssh://git.freedesktop.org/git/systemd/systemd
3912 * First release with the journal
3913 http://0pointer.de/blog/projects/the-journal.html
3915 * The journal replaces both systemd-kmsg-syslogd and
3916 systemd-stdout-bridge.
3918 * New sd_pid_get_unit() API call in libsystemd-logind
3920 * Many systemadm clean-ups
3922 * Introduce remote-fs-pre.target which is ordered before all
3923 remote mounts and may be used to start services before all
3926 * Added Mageia support
3928 * Add bash completion for systemd-loginctl
3930 * Actively monitor PID file creation for daemons which exit in
3931 the parent process before having finished writing the PID
3932 file in the daemon process. Daemons which do this need to be
3933 fixed (i.e. PID file creation must have finished before the
3934 parent exits), but we now react a bit more gracefully to them.
3936 * Add colourful boot output, mimicking the well-known output
3937 of existing distributions.
3939 * New option PassCredentials= for socket units, for
3940 compatibility with a recent kernel ABI breakage.
3942 * /etc/rc.local is now hooked in via a generator binary, and
3943 thus will no longer act as synchronization point during
3946 * systemctl list-unit-files now supports --root=.
3948 * systemd-tmpfiles now understands two new commands: z, Z for
3949 relabelling files according to the SELinux database. This is
3950 useful to apply SELinux labels to specific files in /sys,
3953 * Output of SysV services is now forwarded to both the console
3954 and the journal by default, not only just the console.
3956 * New man pages for all APIs from libsystemd-login.
3958 * The build tree got reorganized and a the build system is a
3959 lot more modular allowing embedded setups to specifically
3960 select the components of systemd they are interested in.
3962 * Support for Linux systems lacking the kernel VT subsystem is
3965 * configure's --with-rootdir= got renamed to
3966 --with-rootprefix= to follow the naming used by udev and
3969 * Unless specified otherwise we will now install to /usr instead
3970 of /usr/local by default.
3972 * Processes with '@' in argv[0][0] are now excluded from the
3973 final shut-down killing spree, following the logic explained
3975 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
3977 * All processes remaining in a service cgroup when we enter
3978 the START or START_PRE states are now killed with
3979 SIGKILL. That means it is no longer possible to spawn
3980 background processes from ExecStart= lines (which was never
3981 supported anyway, and bad style).
3983 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
3984 reloading of units together.
3986 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
3987 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
3988 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
3989 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
3990 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek