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 can be used
46 instead 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
104 FORCERENEW. There are also new configuration options to
105 configure the vendor client identifier and broadcast mode
108 * systemd will no longer inform the kernel about the current
109 timezone, as this is necessarily incorrect and racy as the
110 kernel has no understanding of DST and similar
111 concepts. This hence means FAT timestamps will be always
112 considered UTC, similar to what Android is already
113 doing. Also, when the RTC is configured to the local time
114 (rather than UTC) systemd will never synchronize back to it,
115 as this might confuse Windows at a later boot.
117 * systemd-analyze gained a new command "verify" for offline
118 validation of unit files.
120 * systemd-networkd gained support for a couple of additional
121 settings for bonding networking setups. Also, the metric for
122 statically configured routes may now be configured. For
123 network interfaces where this is appropriate the peer IP
124 address may now be configured.
126 * systemd-networkd's DHCP client will no longer request
127 broadcasting by default, as this tripped up some networks.
128 For hardware where broadcast is required the feature should
129 be switched back on using RequestBroadcast=yes.
131 * systemd-networkd will now set up IPv4LL addresses (when
132 enabled) even if DHCP is configured successfully.
134 * udev will now default to respect network device names given
135 by the kernel when the kernel indicates that these are
136 predictable. This behavior can be tweaked by changing
137 NamePolicy= in the relevant .link file.
139 * A new library systemd-terminal has been added that
140 implements full TTY stream parsing and rendering. This
141 library is supposed to be used later on for implementing a
142 full userspace VT subsystem, replacing the current kernel
145 * A new tool systemd-journal-upload has been added to push
146 journal data to a remote system running
147 systemd-journal-remote.
149 * journald will no longer forward all local data to another
150 running syslog daemon. This change has been made because
151 rsyslog (which appears to be the most commonly used syslog
152 implementation these days) no longer makes use of this, and
153 instead pulls the data out of the journal on its own. Since
154 forwarding the messages to a non-existent syslog server is
155 more expensive than we assumed we have now turned this
156 off. If you run a syslog server that is not a recent rsyslog
157 version, you have to turn this option on again
158 (ForwardToSyslog= in journald.conf).
160 * journald now optionally supports the LZ4 compressor for
161 larger journal fields. This compressor should perform much
162 better than XZ which was the previous default.
164 * machinectl now shows the IP addresses of local containers,
165 if it knows them, plus the interface name of the container.
167 * A new tool "systemd-escape" has been added that makes it
168 easy to escape strings to build unit names and similar.
170 * sd_notify() messages may now include a new ERRNO= field
171 which is parsed and collected by systemd and shown among the
172 "systemctl status" output for a service.
174 * A new component "systemd-firstboot" has been added that
175 queries the most basic systemd information (timezone,
176 hostname, root password) interactively on first
177 boot. Alternatively it may also be used to provision these
178 things offline on OS images installed into directories.
180 * The default sysctl.d/ snippets will now set
182 net.ipv4.conf.default.promote_secondaries=1
184 This has the benefit of no flushing secondary IP addresses
185 when primary addresses are removed.
187 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
188 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
189 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
190 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
191 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
192 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
193 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
194 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
195 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
196 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
197 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
198 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
199 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
200 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
201 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
203 -- Berlin, 2014-08-19
207 * A new tool systemd-sysusers has been added. This tool
208 creates system users and groups in /etc/passwd and
209 /etc/group, based on static declarative system user/group
210 definitions in /usr/lib/sysusers.d/. This is useful to
211 enable factory resets and volatile systems that boot up with
212 an empty /etc directory, and thus need system users and
213 groups created during early boot. systemd now also ships
214 with two default sysusers.d/ files for the most basic
215 users and groups systemd and the core operating system
218 * A new tmpfiles snippet has been added that rebuilds the
219 essential files in /etc on boot, should they be missing.
221 * A directive for ensuring automatic clean-up of
222 /var/cache/man/ has been removed from the default
223 configuration. This line should now be shipped by the man
224 implementation. The necessary change has been made to the
225 man-db implementation. Note that you need to update your man
226 implementation to one that ships this line, otherwise no
227 automatic clean-up of /var/cache/man will take place.
229 * A new condition ConditionNeedsUpdate= has been added that
230 may conditionalize services to only run when /etc or /var
231 are "older" than the vendor operating system resources in
232 /usr. This is useful for reconstructing or updating /etc
233 after an offline update of /usr or a factory reset, on the
234 next reboot. Services that want to run once after such an
235 update or reset should use this condition and order
236 themselves before the new systemd-update-done.service, which
237 will mark the two directories as fully updated. A number of
238 service files have been added making use of this, to rebuild
239 the udev hardware database, the journald message catalog and
240 dynamic loader cache (ldconfig). The systemd-sysusers tool
241 described above also makes use of this now. With this in
242 place it is now possible to start up a minimal operating
243 system with /etc empty cleanly. For more information on the
244 concepts involved see this recent blog story:
246 http://0pointer.de/blog/projects/stateless.html
248 * A new system group "input" has been introduced, and all
249 input device nodes get this group assigned. This is useful
250 for system-level software to get access to input devices. It
251 complements what is already done for "audio" and "video".
253 * systemd-networkd learnt minimal DHCPv4 server support in
254 addition to the existing DHCPv4 client support. It also
255 learnt DHCPv6 client and IPv6 Router Solicitation client
256 support. The DHCPv4 client gained support for static routes
257 passed in from the server. Note that the [DHCPv4] section
258 known in older systemd-networkd versions has been renamed to
259 [DHCP] and is now also used by the DHCPv6 client. Existing
260 .network files using settings of this section should be
261 updated, though compatibility is maintained. Optionally, the
262 client hostname may now be sent to the DHCP server.
264 * networkd gained support for vxlan virtual networks as well
265 as tun/tap and dummy devices.
267 * networkd gained support for automatic allocation of address
268 ranges for interfaces from a system-wide pool of
269 addresses. This is useful for dynamically managing a large
270 number of interfaces with a single network configuration
271 file. In particular this is useful to easily assign
272 appropriate IP addresses to the veth links of a large number
275 * RPM macros for processing sysusers, sysctl and binfmt
276 drop-in snippets at package installation time have been
279 * The /etc/os-release file should now be placed in
280 /usr/lib/os-release. The old location is automatically
281 created as symlink. /usr/lib is the more appropriate
282 location of this file, since it shall actually describe the
283 vendor operating system shipped in /usr, and not the
284 configuration stored in /etc.
286 * .mount units gained a new boolean SloppyOptions= setting
287 that maps to mount(8)'s -s option which enables permissive
288 parsing of unknown mount options.
290 * tmpfiles learnt a new "L+" directive which creates a symlink
291 but (unlike "L") deletes a pre-existing file first, should
292 it already exist and not already be the correct
293 symlink. Similar, "b+", "c+" and "p+" directives have been
294 added as well, which create block and character devices, as
295 well as fifos in the filesystem, possibly removing any
296 pre-existing files of different types.
298 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
299 'argument' field (which so far specified the source to
300 symlink/copy the files from) is now optional. If omitted the
301 same file os copied from /usr/share/factory/ suffixed by the
302 full destination path. This is useful for populating /etc
303 with essential files, by copying them from vendor defaults
304 shipped in /usr/share/factory/etc.
306 * A new command "systemctl preset-all" has been added that
307 applies the service preset settings to all installed unit
308 files. A new switch --preset-mode= has been added that
309 controls whether only enable or only disable operations
312 * A new command "systemctl is-system-running" has been added
313 that allows checking the overall state of the system, for
314 example whether it is fully up and running.
316 * When the system boots up with an empty /etc, the equivalent
317 to "systemctl preset-all" is executed during early boot, to
318 make sure all default services are enabled after a factory
321 * systemd now contains a minimal preset file that enables the
322 most basic services systemd ships by default.
324 * Unit files' [Install] section gained a new DefaultInstance=
325 field for defining the default instance to create if a
326 template unit is enabled with no instance specified.
328 * A new passive target cryptsetup-pre.target has been added
329 that may be used by services that need to make they run and
330 finish before the first LUKS cryptographic device is set up.
332 * The /dev/loop-control and /dev/btrfs-control device nodes
333 are now owned by the "disk" group by default, opening up
334 access to this group.
336 * systemd-coredump will now automatically generate a
337 stack trace of all core dumps taking place on the system,
338 based on elfutils' libdw library. This stack trace is logged
341 * systemd-coredump may now optionally store coredumps directly
342 on disk (in /var/lib/systemd/coredump, possibly compressed),
343 instead of storing them unconditionally in the journal. This
344 mode is the new default. A new configuration file
345 /etc/systemd/coredump.conf has been added to configure this
346 and other parameters of systemd-coredump.
348 * coredumpctl gained a new "info" verb to show details about a
349 specific coredump. A new switch "-1" has also been added
350 that makes sure to only show information about the most
351 recent entry instead of all entries. Also, as the tool is
352 generally useful now the "systemd-" prefix of the binary
353 name has been removed. Distributions that want to maintain
354 compatibility with the old name should add a symlink from
355 the old name to the new name.
357 * journald's SplitMode= now defaults to "uid". This makes sure
358 that unprivileged users can access their own coredumps with
359 coredumpctl without restrictions.
361 * New kernel command line options "systemd.wants=" (for
362 pulling an additional unit during boot), "systemd.mask="
363 (for masking a specific unit for the boot), and
364 "systemd.debug-shell" (for enabling the debug shell on tty9)
365 have been added. This is implemented in the new generator
366 "systemd-debug-generator".
368 * systemd-nspawn will now by default filter a couple of
369 syscalls for containers, among them those required for
370 kernel module loading, direct x86 IO port access, swap
371 management, and kexec. Most importantly though
372 open_by_handle_at() is now prohibited for containers,
373 closing a hole similar to a recently discussed vulnerability
374 in docker regarding access to files on file hierarchies the
375 container should normally not have access to. Note that for
376 nspawn we generally make no security claims anyway (and
377 this is explicitly documented in the man page), so this is
378 just a fix for one of the most obvious problems.
380 * A new man page file-hierarchy(7) has been added that
381 contains a minimized, modernized version of the file system
382 layout systemd expects, similar in style to the FHS
383 specification or hier(5). A new tool systemd-path(1) has
384 been added to query many of these paths for the local
387 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
388 longer done. Since the directory now has a per-user size
389 limit, and is cleaned on logout this appears unnecessary,
390 in particular since this now brings the lifecycle of this
391 directory closer in line with how IPC objects are handled.
393 * systemd.pc now exports a number of additional directories,
394 including $libdir (which is useful to identify the library
395 path for the primary architecture of the system), and a
396 couple of drop-in directories.
398 * udev's predictable network interface names now use the dev_port
399 sysfs attribute, introduced in linux 3.15 instead of dev_id to
400 distinguish between ports of the same PCI function. dev_id should
401 only be used for ports using the same HW address, hence the need
404 * machined has been updated to export the OS version of a
405 container (read from /etc/os-release and
406 /usr/lib/os-release) on the bus. This is now shown in
407 "machinectl status" for a machine.
409 * A new service setting RestartForceExitStatus= has been
410 added. If configured to a set of exit signals or process
411 return values, the service will be restarted when the main
412 daemon process exits with any of them, regardless of the
415 * systemctl's -H switch for connecting to remote systemd
416 machines has been extended so that it may be used to
417 directly connect to a specific container on the
418 host. "systemctl -H root@foobar:waldi" will now connect as
419 user "root" to host "foobar", and then proceed directly to
420 the container named "waldi". Note that currently you have to
421 authenticate as user "root" for this to work, as entering
422 containers is a privileged operation.
424 Contributions from: Andreas Henriksson, Benjamin Steinwender,
425 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
426 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
427 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
428 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
429 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
430 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
431 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
432 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
433 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
434 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
435 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
437 -- Berlin, 2014-07-03
441 * As an experimental feature, udev now tries to lock the
442 disk device node (flock(LOCK_SH|LOCK_NB)) while it
443 executes events for the disk or any of its partitions.
444 Applications like partitioning programs can lock the
445 disk device node (flock(LOCK_EX)) and claim temporary
446 device ownership that way; udev will entirely skip all event
447 handling for this disk and its partitions. If the disk
448 was opened for writing, the close will trigger a partition
449 table rescan in udev's "watch" facility, and if needed
450 synthesize "change" events for the disk and all its partitions.
451 This is now unconditionally enabled, and if it turns out to
452 cause major problems, we might turn it on only for specific
453 devices, or might need to disable it entirely. Device Mapper
454 devices are excluded from this logic.
456 * We temporarily dropped the "-l" switch for fsck invocations,
457 since they collide with the flock() logic above. util-linux
458 upstream has been changed already to avoid this conflict,
459 and we will readd "-l" as soon as util-linux with this
460 change has been released.
462 * The dependency on libattr has been removed. Since a long
463 time, the extended attribute calls have moved to glibc, and
464 libattr is thus unnecessary.
466 * Virtualization detection works without priviliges now. This
467 means the systemd-detect-virt binary no longer requires
468 CAP_SYS_PTRACE file capabilities, and our daemons can run
469 with fewer privileges.
471 * systemd-networkd now runs under its own "systemd-network"
472 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
473 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
474 loses the ability to write to files owned by root this way.
476 * Similar, systemd-resolved now runs under its own
477 "systemd-resolve" user with no capabilities remaining.
479 * Similar, systemd-bus-proxyd now runs under its own
480 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
482 * systemd-networkd gained support for setting up "veth"
483 virtual ethernet devices for container connectivity, as well
484 as GRE and VTI tunnels.
486 * systemd-networkd will no longer automatically attempt to
487 manually load kernel modules necessary for certain tunnel
488 transports. Instead, it is assumed the kernel loads them
489 automatically when required. This only works correctly on
490 very new kernels. On older kernels, please consider adding
491 the kernel modules to /etc/modules-load.d/ as a work-around.
493 * The resolv.conf file systemd-resolved generates has been
494 moved to /run/systemd/resolve/. If you have a symlink from
495 /etc/resolv.conf, it might be necessary to correct it.
497 * Two new service settings, ProtectHome= and ProtectSystem=,
498 have been added. When enabled, they will make the user data
499 (such as /home) inaccessible or read-only and the system
500 (such as /usr) read-only, for specific services. This allows
501 very light-weight per-service sandboxing to avoid
502 modifications of user data or system files from
503 services. These two new switches have been enabled for all
504 of systemd's long-running services, where appropriate.
506 * Socket units gained new SocketUser= and SocketGroup=
507 settings to set the owner user and group of AF_UNIX sockets
508 and FIFOs in the file system.
510 * Socket units gained a new RemoveOnStop= setting. If enabled,
511 all FIFOS and sockets in the file system will be removed
512 when the specific socket unit is stopped.
514 * Socket units gained a new Symlinks= setting. It takes a list
515 of symlinks to create to file system sockets or FIFOs
516 created by the specific Unix sockets. This is useful to
517 manage symlinks to socket nodes with the same life-cycle as
520 * The /dev/log socket and /dev/initctl FIFO have been moved to
521 /run, and have been replaced by symlinks. This allows
522 connecting to these facilities even if PrivateDevices=yes is
523 used for a service (which makes /dev/log itself unavailable,
524 but /run is left). This also has the benefit of ensuring
525 that /dev only contains device nodes, directories and
526 symlinks, and nothing else.
528 * sd-daemon gained two new calls sd_pid_notify() and
529 sd_pid_notifyf(). They are similar to sd_notify() and
530 sd_notifyf(), but allow overriding of the source PID of
531 notification messages if permissions permit this. This is
532 useful to send notify messages on behalf of a different
533 process (for example, the parent process). The
534 systemd-notify tool has been updated to make use of this
535 when sending messages (so that notification messages now
536 originate from the shell script invoking systemd-notify and
537 not the systemd-notify process itself. This should minimize
538 a race where systemd fails to associate notification
539 messages to services when the originating process already
542 * A new "on-abnormal" setting for Restart= has been added. If
543 set, it will result in automatic restarts on all "abnormal"
544 reasons for a process to exit, which includes unclean
545 signals, core dumps, timeouts and watchdog timeouts, but
546 does not include clean and unclean exit codes or clean
547 signals. Restart=on-abnormal is an alternative for
548 Restart=on-failure for services that shall be able to
549 terminate and avoid restarts on certain errors, by
550 indicating so with an unclean exit code. Restart=on-failure
551 or Restart=on-abnormal is now the recommended setting for
552 all long-running services.
554 * If the InaccessibleDirectories= service setting points to a
555 mount point (or if there are any submounts contained within
556 it), it is now attempted to completely unmount it, to make
557 the file systems truly unavailable for the respective
560 * The ReadOnlyDirectories= service setting and
561 systemd-nspawn's --read-only parameter are now recursively
562 applied to all submounts, too.
564 * Mount units may now be created transiently via the bus APIs.
566 * The support for SysV and LSB init scripts has been removed
567 from the systemd daemon itself. Instead, it is now
568 implemented as a generator that creates native systemd units
569 from these scripts when needed. This enables us to remove a
570 substantial amount of legacy code from PID 1, following the
571 fact that many distributions only ship a very small number
572 of LSB/SysV init scripts nowadays.
574 * Priviliged Xen (dom0) domains are not considered
575 virtualization anymore by the virtualization detection
576 logic. After all, they generally have unrestricted access to
577 the hardware and usually are used to manage the unprivileged
580 * systemd-tmpfiles gained a new "C" line type, for copying
581 files or entire directories.
583 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
584 lines. So far, they have been non-globbing versions of the
585 latter, and have thus been redundant. In future, it is
586 recommended to only use "z". "m" has hence been removed
587 from the documentation, even though it stays supported.
589 * A tmpfiles snippet to recreate the most basic structure in
590 /var has been added. This is enough to create the /var/run →
591 /run symlink and create a couple of structural
592 directories. This allows systems to boot up with an empty or
593 volatile /var. Of course, while with this change, the core OS
594 now is capable with dealing with a volatile /var, not all
595 user services are ready for it. However, we hope that sooner
596 or later, many service daemons will be changed upstream so
597 that they are able to automatically create their necessary
598 directories in /var at boot, should they be missing. This is
599 the first step to allow state-less systems that only require
600 the vendor image for /usr to boot.
602 * systemd-nspawn has gained a new --tmpfs= switch to mount an
603 empty tmpfs instance to a specific directory. This is
604 particularly useful for making use of the automatic
605 reconstruction of /var (see above), by passing --tmpfs=/var.
607 * Access modes specified in tmpfiles snippets may now be
608 prefixed with "~", which indicates that they shall be masked
609 by whether the existing file or directory is currently
610 writable, readable or executable at all. Also, if specified,
611 the sgid/suid/sticky bits will be masked for all
614 * A new passive target unit "network-pre.target" has been
615 added which is useful for services that shall run before any
616 network is configured, for example firewall scripts.
618 * The "floppy" group that previously owned the /dev/fd*
619 devices is no longer used. The "disk" group is now used
620 instead. Distributions should probably deprecate usage of
623 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
624 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
625 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
626 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
627 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
628 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
631 -- Berlin, 2014-06-11
635 * A new "systemd-timesyncd" daemon has been added for
636 synchronizing the system clock across the network. It
637 implements an SNTP client. In contrast to NTP
638 implementations such as chrony or the NTP reference server,
639 this only implements a client side, and does not bother with
640 the full NTP complexity, focusing only on querying time from
641 one remote server and synchronizing the local clock to
642 it. Unless you intend to serve NTP to networked clients or
643 want to connect to local hardware clocks, this simple NTP
644 client should be more than appropriate for most
645 installations. The daemon runs with minimal privileges, and
646 has been hooked up with networkd to only operate when
647 network connectivity is available. The daemon saves the
648 current clock to disk every time a new NTP sync has been
649 acquired, and uses this to possibly correct the system clock
650 early at bootup, in order to accommodate for systems that
651 lack an RTC such as the Raspberry Pi and embedded devices,
652 and to make sure that time monotonically progresses on these
653 systems, even if it is not always correct. To make use of
654 this daemon, a new system user and group "systemd-timesync"
655 needs to be created on installation of systemd.
657 * The queue "seqnum" interface of libudev has been disabled, as
658 it was generally incompatible with device namespacing as
659 sequence numbers of devices go "missing" if the devices are
660 part of a different namespace.
662 * "systemctl list-timers" and "systemctl list-sockets" gained
663 a --recursive switch for showing units of these types also
664 for all local containers, similar in style to the already
665 supported --recursive switch for "systemctl list-units".
667 * A new RebootArgument= setting has been added for service
668 units, which may be used to specify a kernel reboot argument
669 to use when triggering reboots with StartLimitAction=.
671 * A new FailureAction= setting has been added for service
672 units which may be used to specify an operation to trigger
673 when a service fails. This works similarly to
674 StartLimitAction=, but unlike it, controls what is done
675 immediately rather than only after several attempts to
676 restart the service in question.
678 * hostnamed got updated to also expose the kernel name,
679 release, and version on the bus. This is useful for
680 executing commands like hostnamectl with the -H switch.
681 systemd-analyze makes use of this to properly display
682 details when running non-locally.
684 * The bootchart tool can now show cgroup information in the
687 * The CFS CPU quota cgroup attribute is now exposed for
688 services. The new CPUQuota= switch has been added for this
689 which takes a percentage value. Setting this will have the
690 result that a service may never get more CPU time than the
691 specified percentage, even if the machine is otherwise idle.
693 * systemd-networkd learned IPIP and SIT tunnel support.
695 * LSB init scripts exposing a dependency on $network will now
696 get a dependency on network-online.target rather than simply
697 network.target. This should bring LSB handling closer to
698 what it was on SysV systems.
700 * A new fsck.repair= kernel option has been added to control
701 how fsck shall deal with unclean file systems at boot.
703 * The (.ini) configuration file parser will now silently
704 ignore sections whose name begins with "X-". This may be
705 used to maintain application-specific extension sections in unit
708 * machined gained a new API to query the IP addresses of
709 registered containers. "machinectl status" has been updated
710 to show these addresses in its output.
712 * A new call sd_uid_get_display() has been added to the
713 sd-login APIs for querying the "primary" session of a
714 user. The "primary" session of the user is elected from the
715 user's sessions and generally a graphical session is
716 preferred over a text one.
718 * A minimal systemd-resolved daemon has been added. It
719 currently simply acts as a companion to systemd-networkd and
720 manages resolv.conf based on per-interface DNS
721 configuration, possibly supplied via DHCP. In the long run
722 we hope to extend this into a local DNSSEC enabled DNS and
725 * The systemd-networkd-wait-online tool is now enabled by
726 default. It will delay network-online.target until a network
727 connection has been configured. The tool primarily integrates
728 with networkd, but will also make a best effort to make sense
729 of network configuration performed in some other way.
731 * Two new service options StartupCPUShares= and
732 StartupBlockIOWeight= have been added that work similarly to
733 CPUShares= and BlockIOWeight= however only apply during
734 system startup. This is useful to prioritize certain services
735 differently during bootup than during normal runtime.
737 * hostnamed has been changed to prefer the statically
738 configured hostname in /etc/hostname (unless set to
739 'localhost' or empty) over any dynamic one supplied by
740 dhcp. With this change, the rules for picking the hostname
741 match more closely the rules of other configuration settings
742 where the local administrator's configuration in /etc always
743 overrides any other settings.
745 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
746 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
747 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
748 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
749 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
750 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
751 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
752 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
753 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
754 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
755 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
756 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
757 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
758 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
759 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
760 Lindskog, WaLyong Cho, Will Woods, Zbigniew
763 -- Beijing, 2014-05-28
767 * When restoring the screen brightness at boot, stay away from
768 the darkest setting or from the lowest 5% of the available
769 range, depending on which is the larger value of both. This
770 should effectively protect the user from rebooting into a
771 black screen, should the brightness have been set to minimum
774 * sd-login gained a new sd_machine_get_class() call to
775 determine the class ("vm" or "container") of a machine
776 registered with machined.
778 * sd-login gained new calls
779 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
780 to query the identity of the peer of a local AF_UNIX
781 connection. They operate similarly to their sd_pid_get_xyz()
784 * PID 1 will now maintain a system-wide system state engine
785 with the states "starting", "running", "degraded",
786 "maintenance", "stopping". These states are bound to system
787 startup, normal runtime, runtime with at least one failed
788 service, rescue/emergency mode and system shutdown. This
789 state is shown in the "systemctl status" output when no unit
790 name is passed. It is useful to determine system state, in
791 particularly when doing so for many systems or containers at
794 * A new command "list-machines" has been added to "systemctl"
795 that lists all local OS containers and shows their system
796 state (see above), if systemd runs inside of them.
798 * systemctl gained a new "-r" switch to recursively enumerate
799 units on all local containers, when used with the
800 "list-unit" command (which is the default one that is
801 executed when no parameters are specified).
803 * The GPT automatic partition discovery logic will now honour
804 two GPT partition flags: one may be set on a partition to
805 cause it to be mounted read-only, and the other may be set
806 on a partition to ignore it during automatic discovery.
808 * Two new GPT type UUIDs have been added for automatic root
809 partition discovery, for 32-bit and 64-bit ARM. This is not
810 particularly useful for discovering the root directory on
811 these architectures during bare-metal boots (since UEFI is
812 not common there), but still very useful to allow booting of
813 ARM disk images in nspawn with the -i option.
815 * MAC addresses of interfaces created with nspawn's
816 --network-interface= switch will now be generated from the
817 machine name, and thus be stable between multiple invocations
820 * logind will now automatically remove all IPC objects owned
821 by a user if she or he fully logs out. This makes sure that
822 users who are logged out cannot continue to consume IPC
823 resources. This covers SysV memory, semaphores and message
824 queues as well as POSIX shared memory and message
825 queues. Traditionally, SysV and POSIX IPC had no life-cycle
826 limits. With this functionality, that is corrected. This may
827 be turned off by using the RemoveIPC= switch of logind.conf.
829 * The systemd-machine-id-setup and tmpfiles tools gained a
830 --root= switch to operate on a specific root directory,
833 * journald can now forward logged messages to the TTYs of all
834 logged in users ("wall"). This is the default for all
835 emergency messages now.
837 * A new tool systemd-journal-remote has been added to stream
838 journal log messages across the network.
840 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
841 controller trees are mounted into it. Note that the
842 directories mounted beneath it are not read-only. This is a
843 security measure and is particularly useful because glibc
844 actually includes a search logic to pick any tmpfs it can
845 find to implement shm_open() if /dev/shm is not available
846 (which it might very well be in namespaced setups).
848 * machinectl gained a new "poweroff" command to cleanly power
849 down a local OS container.
851 * The PrivateDevices= unit file setting will now also drop the
852 CAP_MKNOD capability from the capability bound set, and
853 imply DevicePolicy=closed.
855 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
856 comprehensively on all long-running systemd services where
859 * systemd-udevd will now run in a disassociated mount
860 namespace. To mount directories from udev rules, make sure to
861 pull in mount units via SYSTEMD_WANTS properties.
863 * The kdbus support gained support for uploading policy into
864 the kernel. sd-bus gained support for creating "monitoring"
865 connections that can eavesdrop into all bus communication
866 for debugging purposes.
868 * Timestamps may now be specified in seconds since the UNIX
869 epoch Jan 1st, 1970 by specifying "@" followed by the value
872 * Native tcpwrap support in systemd has been removed. tcpwrap
873 is old code, not really maintained anymore and has serious
874 shortcomings, and better options such as firewalls
875 exist. For setups that require tcpwrap usage, please
876 consider invoking your socket-activated service via tcpd,
877 like on traditional inetd.
879 * A new system.conf configuration option
880 DefaultTimerAccuracySec= has been added that controls the
881 default AccuracySec= setting of .timer units.
883 * Timer units gained a new WakeSystem= switch. If enabled,
884 timers configured this way will cause the system to resume
885 from system suspend (if the system supports that, which most
888 * Timer units gained a new Persistent= switch. If enabled,
889 timers configured this way will save to disk when they have
890 been last triggered. This information is then used on next
891 reboot to possible execute overdue timer events, that
892 could not take place because the system was powered off.
893 This enables simple anacron-like behaviour for timer units.
895 * systemctl's "list-timers" will now also list the time a
896 timer unit was last triggered in addition to the next time
897 it will be triggered.
899 * systemd-networkd will now assign predictable IPv4LL
900 addresses to its local interfaces.
902 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
903 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
904 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
905 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
906 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
907 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
908 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
909 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
912 -- Berlin, 2014-03-25
916 * A new unit file setting RestrictAddressFamilies= has been
917 added to restrict which socket address families unit
918 processes gain access to. This takes address family names
919 like "AF_INET" or "AF_UNIX", and is useful to minimize the
920 attack surface of services via exotic protocol stacks. This
921 is built on seccomp system call filters.
923 * Two new unit file settings RuntimeDirectory= and
924 RuntimeDirectoryMode= have been added that may be used to
925 manage a per-daemon runtime directories below /run. This is
926 an alternative for setting up directory permissions with
927 tmpfiles snippets, and has the advantage that the runtime
928 directory's lifetime is bound to the daemon runtime and that
929 the daemon starts up with an empty directory each time. This
930 is particularly useful when writing services that drop
931 privileges using the User= or Group= setting.
933 * The DeviceAllow= unit setting now supports globbing for
934 matching against device group names.
936 * The systemd configuration file system.conf gained new
937 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
938 DefaultMemoryAccounting= to globally turn on/off accounting
939 for specific resources (cgroups) for all units. These
940 settings may still be overridden individually in each unit
943 * systemd-gpt-auto-generator is now able to discover /srv and
944 root partitions in addition to /home and swap partitions. It
945 also supports LUKS-encrypted partitions now. With this in
946 place, automatic discovery of partitions to mount following
947 the Discoverable Partitions Specification
948 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
949 is now a lot more complete. This allows booting without
950 /etc/fstab and without root= on the kernel command line on
951 systems prepared appropriately.
953 * systemd-nspawn gained a new --image= switch which allows
954 booting up disk images and Linux installations on any block
955 device that follow the Discoverable Partitions Specification
956 (see above). This means that installations made with
957 appropriately updated installers may now be started and
958 deployed using container managers, completely
959 unmodified. (We hope that libvirt-lxc will add support for
960 this feature soon, too.)
962 * systemd-nspawn gained a new --network-macvlan= setting to
963 set up a private macvlan interface for the
964 container. Similarly, systemd-networkd gained a new
965 Kind=macvlan setting in .netdev files.
967 * systemd-networkd now supports configuring local addresses
970 * A new tool systemd-network-wait-online has been added to
971 synchronously wait for network connectivity using
974 * The sd-bus.h bus API gained a new sd_bus_track object for
975 tracking the life-cycle of bus peers. Note that sd-bus.h is
976 still not a public API though (unless you specify
977 --enable-kdbus on the configure command line, which however
978 voids your warranty and you get no API stability guarantee).
980 * The $XDG_RUNTIME_DIR runtime directories for each user are
981 now individual tmpfs instances, which has the benefit of
982 introducing separate pools for each user, with individual
983 size limits, and thus making sure that unprivileged clients
984 can no longer negatively impact the system or other users by
985 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
986 RuntimeDirectorySize= has been introduced that allows
987 controlling the default size limit for all users. It
988 defaults to 10% of the available physical memory. This is no
989 replacement for quotas on tmpfs though (which the kernel
990 still does not support), as /dev/shm and /tmp are still
991 shared resources used by both the system and unprivileged
994 * logind will now automatically turn off automatic suspending
995 on laptop lid close when more than one display is
996 connected. This was previously expected to be implemented
997 individually in desktop environments (such as GNOME),
998 however has been added to logind now, in order to fix a
999 boot-time race where a desktop environment might not have
1000 been started yet and thus not been able to take an inhibitor
1001 lock at the time where logind already suspends the system
1002 due to a closed lid.
1004 * logind will now wait at least 30s after each system
1005 suspend/resume cycle, and 3min after system boot before
1006 suspending the system due to a closed laptop lid. This
1007 should give USB docking stations and similar enough time to
1008 be probed and configured after system resume and boot in
1009 order to then act as suspend blocker.
1011 * systemd-run gained a new --property= setting which allows
1012 initialization of resource control properties (and others)
1013 for the created scope or service unit. Example: "systemd-run
1014 --property=BlockIOWeight=10 updatedb" may be used to run
1015 updatedb at a low block IO scheduling weight.
1017 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1018 now also work in --scope mode.
1020 * When systemd is compiled with kdbus support, basic support
1021 for enforced policies is now in place. (Note that enabling
1022 kdbus still voids your warranty and no API compatibility
1025 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1026 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1027 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1028 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1029 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1030 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1031 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1032 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1033 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1034 Zbigniew Jędrzejewski-Szmek
1036 -- Berlin, 2014-03-12
1040 * systemd will now relabel /dev after loading the SMACK policy
1041 according to SMACK rules.
1043 * A new unit file option AppArmorProfile= has been added to
1044 set the AppArmor profile for the processes of a unit.
1046 * A new condition check ConditionArchitecture= has been added
1047 to conditionalize units based on the system architecture, as
1048 reported by uname()'s "machine" field.
1050 * systemd-networkd now supports matching on the system
1051 virtualization, architecture, kernel command line, host name
1054 * logind is now a lot more aggressive when suspending the
1055 machine due to a closed laptop lid. Instead of acting only
1056 on the lid close action, it will continuously watch the lid
1057 status and act on it. This is useful for laptops where the
1058 power button is on the outside of the chassis so that it can
1059 be reached without opening the lid (such as the Lenovo
1060 Yoga). On those machines, logind will now immediately
1061 re-suspend the machine if the power button has been
1062 accidentally pressed while the laptop was suspended and in a
1063 backpack or similar.
1065 * logind will now watch SW_DOCK switches and inhibit reaction
1066 to the lid switch if it is pressed. This means that logind
1067 will not suspend the machine anymore if the lid is closed
1068 and the system is docked, if the laptop supports SW_DOCK
1069 notifications via the input layer. Note that ACPI docking
1070 stations do not generate this currently. Also note that this
1071 logic is usually not fully sufficient and Desktop
1072 Environments should take a lid switch inhibitor lock when an
1073 external display is connected, as systemd will not watch
1076 * nspawn will now make use of the devices cgroup controller by
1077 default, and only permit creation of and access to the usual
1078 API device nodes like /dev/null or /dev/random, as well as
1079 access to (but not creation of) the pty devices.
1081 * We will now ship a default .network file for
1082 systemd-networkd that automatically configures DHCP for
1083 network interfaces created by nspawn's --network-veth or
1084 --network-bridge= switches.
1086 * systemd will now understand the usual M, K, G, T suffixes
1087 according to SI conventions (i.e. to the base 1000) when
1088 referring to throughput and hardware metrics. It will stay
1089 with IEC conventions (i.e. to the base 1024) for software
1090 metrics, according to what is customary according to
1091 Wikipedia. We explicitly document which base applies for
1092 each configuration option.
1094 * The DeviceAllow= setting in unit files now supports a syntax
1095 to whitelist an entire group of devices node majors at once,
1096 based on the /proc/devices listing. For example, with the
1097 string "char-pts", it is now possible to whitelist all
1098 current and future pseudo-TTYs at once.
1100 * sd-event learned a new "post" event source. Event sources of
1101 this type are triggered by the dispatching of any event
1102 source of a type that is not "post". This is useful for
1103 implementing clean-up and check event sources that are
1104 triggered by other work being done in the program.
1106 * systemd-networkd is no longer statically enabled, but uses
1107 the usual [Install] sections so that it can be
1108 enabled/disabled using systemctl. It still is enabled by
1111 * When creating a veth interface pair with systemd-nspawn, the
1112 host side will now be prefixed with "vb-" if
1113 --network-bridge= is used, and with "ve-" if --network-veth
1114 is used. This way, it is easy to distinguish these cases on
1115 the host, for example to apply different configuration to
1116 them with systemd-networkd.
1118 * The compatibility libraries for libsystemd-journal.so,
1119 libsystem-id128.so, libsystemd-login.so and
1120 libsystemd-daemon.so do not make use of IFUNC
1121 anymore. Instead, we now build libsystemd.so multiple times
1122 under these alternative names. This means that the footprint
1123 is drastically increased, but given that these are
1124 transitional compatibility libraries, this should not matter
1125 much. This change has been made necessary to support the ARM
1126 platform for these compatibility libraries, as the ARM
1127 toolchain is not really at the same level as the toolchain
1128 for other architectures like x86 and does not support
1129 IFUNC. Please make sure to use --enable-compat-libs only
1130 during a transitional period!
1132 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1133 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1134 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1135 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1136 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1137 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1138 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1139 Zbigniew Jędrzejewski-Szmek
1141 -- Berlin, 2014-02-24
1145 * A new component "systemd-networkd" has been added that can
1146 be used to configure local network interfaces statically or
1147 via DHCP. It is capable of bringing up bridges, VLANs, and
1148 bonding. Currently, no hook-ups for interactive network
1149 configuration are provided. Use this for your initrd,
1150 container, embedded, or server setup if you need a simple,
1151 yet powerful, network configuration solution. This
1152 configuration subsystem is quite nifty, as it allows wildcard
1153 hotplug matching in interfaces. For example, with a single
1154 configuration snippet, you can configure that all Ethernet
1155 interfaces showing up are automatically added to a bridge,
1156 or similar. It supports link-sensing and more.
1158 * A new tool "systemd-socket-proxyd" has been added which can
1159 act as a bidirectional proxy for TCP sockets. This is
1160 useful for adding socket activation support to services that
1161 do not actually support socket activation, including virtual
1162 machines and the like.
1164 * Add a new tool to save/restore rfkill state on
1167 * Save/restore state of keyboard backlights in addition to
1168 display backlights on shutdown/boot.
1170 * udev learned a new SECLABEL{} construct to label device
1171 nodes with a specific security label when they appear. For
1172 now, only SECLABEL{selinux} is supported, but the syntax is
1173 prepared for additional security frameworks.
1175 * udev gained a new scheme to configure link-level attributes
1176 from files in /etc/systemd/network/*.link. These files can
1177 match against MAC address, device path, driver name and type,
1178 and will apply attributes like the naming policy, link speed,
1179 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1180 address assignment policy (randomized, ...).
1182 * The configuration of network interface naming rules for
1183 "permanent interface names" has changed: a new NamePolicy=
1184 setting in the [Link] section of .link files determines the
1185 priority of possible naming schemes (onboard, slot, mac,
1186 path). The default value of this setting is determined by
1187 /usr/lib/net/links/99-default.link. Old
1188 80-net-name-slot.rules udev configuration file has been
1189 removed, so local configuration overriding this file should
1190 be adapated to override 99-default.link instead.
1192 * When the User= switch is used in a unit file, also
1193 initialize $SHELL= based on the user database entry.
1195 * systemd no longer depends on libdbus. All communication is
1196 now done with sd-bus, systemd's low-level bus library
1199 * kdbus support has been added to PID 1 itself. When kdbus is
1200 enabled, this causes PID 1 to set up the system bus and
1201 enable support for a new ".busname" unit type that
1202 encapsulates bus name activation on kdbus. It works a little
1203 bit like ".socket" units, except for bus names. A new
1204 generator has been added that converts classic dbus1 service
1205 activation files automatically into native systemd .busname
1208 * sd-bus: add a light-weight vtable implementation that allows
1209 defining objects on the bus with a simple static const
1210 vtable array of its methods, signals and properties.
1212 * systemd will not generate or install static dbus
1213 introspection data anymore to /usr/share/dbus-1/interfaces,
1214 as the precise format of these files is unclear, and
1215 nothing makes use of it.
1217 * A proxy daemon is now provided to proxy clients connecting
1218 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1219 compatibility with classic D-Bus.
1221 * A bus driver implementation has been added that supports the
1222 classic D-Bus bus driver calls on kdbus, also for
1223 compatibility purposes.
1225 * A new API "sd-event.h" has been added that implements a
1226 minimal event loop API built around epoll. It provides a
1227 couple of features that direct epoll usage is lacking:
1228 prioritization of events, scales to large numbers of timer
1229 events, per-event timer slack (accuracy), system-wide
1230 coalescing of timer events, exit handlers, watchdog
1231 supervision support using systemd's sd_notify() API, child
1234 * A new API "sd-rntl.h" has been added that provides an API
1235 around the route netlink interface of the kernel, similar in
1236 style to "sd-bus.h".
1238 * A new API "sd-dhcp-client.h" has been added that provides a
1239 small DHCPv4 client-side implementation. This is used by
1242 * There is a new kernel command line option
1243 "systemd.restore_state=0|1". When set to "0", none of the
1244 systemd tools will restore saved runtime state to hardware
1245 devices. More specifically, the rfkill and backlight states
1248 * The FsckPassNo= compatibility option in mount/service units
1249 has been removed. The fstab generator will now add the
1250 necessary dependencies automatically, and does not require
1251 PID1's support for that anymore.
1253 * journalctl gained a new switch, --list-boots, that lists
1254 recent boots with their times and boot IDs.
1256 * The various tools like systemctl, loginctl, timedatectl,
1257 busctl, systemd-run, ... have gained a new switch "-M" to
1258 connect to a specific, local OS container (as direct
1259 connection, without requiring SSH). This works on any
1260 container that is registered with machined, such as those
1261 created by libvirt-lxc or nspawn.
1263 * systemd-run and systemd-analyze also gained support for "-H"
1264 to connect to remote hosts via SSH. This is particularly
1265 useful for systemd-run because it enables queuing of jobs
1266 onto remote systems.
1268 * machinectl gained a new command "login" to open a getty
1269 login in any local container. This works with any container
1270 that is registered with machined (such as those created by
1271 libvirt-lxc or nspawn), and which runs systemd inside.
1273 * machinectl gained a new "reboot" command that may be used to
1274 trigger a reboot on a specific container that is registered
1275 with machined. This works on any container that runs an init
1276 system of some kind.
1278 * systemctl gained a new "list-timers" command to print a nice
1279 listing of installed timer units with the times they elapse
1282 * Alternative reboot() parameters may now be specified on the
1283 "systemctl reboot" command line and are passed to the
1284 reboot() system call.
1286 * systemctl gained a new --job-mode= switch to configure the
1287 mode to queue a job with. This is a more generic version of
1288 --fail, --irreversible, and --ignore-dependencies, which are
1289 still available but not advertised anymore.
1291 * /etc/systemd/system.conf gained new settings to configure
1292 various default timeouts of units, as well as the default
1293 start limit interval and burst. These may still be overridden
1296 * PID1 will now export on the bus profile data of the security
1297 policy upload process (such as the SELinux policy upload to
1300 * journald: when forwarding logs to the console, include
1301 timestamps (following the setting in
1302 /sys/module/printk/parameters/time).
1304 * OnCalendar= in timer units now understands the special
1305 strings "yearly" and "annually". (Both are equivalent)
1307 * The accuracy of timer units is now configurable with the new
1308 AccuracySec= setting. It defaults to 1min.
1310 * A new dependency type JoinsNamespaceOf= has been added that
1311 allows running two services within the same /tmp and network
1312 namespace, if PrivateNetwork= or PrivateTmp= are used.
1314 * A new command "cat" has been added to systemctl. It outputs
1315 the original unit file of a unit, and concatenates the
1316 contents of additional "drop-in" unit file snippets, so that
1317 the full configuration is shown.
1319 * systemctl now supports globbing on the various "list-xyz"
1320 commands, like "list-units" or "list-sockets", as well as on
1321 those commands which take multiple unit names.
1323 * journalctl's --unit= switch gained support for globbing.
1325 * All systemd daemons now make use of the watchdog logic so
1326 that systemd automatically notices when they hang.
1328 * If the $container_ttys environment variable is set,
1329 getty-generator will automatically spawn a getty for each
1330 listed tty. This is useful for container managers to request
1331 login gettys to be spawned on as many ttys as needed.
1333 * %h, %s, %U specifier support is not available anymore when
1334 used in unit files for PID 1. This is because NSS calls are
1335 not safe from PID 1. They stay available for --user
1336 instances of systemd, and as special case for the root user.
1338 * loginctl gained a new "--no-legend" switch to turn off output
1341 * The "sd-login.h" API gained three new calls:
1342 sd_session_is_remote(), sd_session_get_remote_user(),
1343 sd_session_get_remote_host() to query information about
1346 * The udev hardware database now also carries vendor/product
1347 information of SDIO devices.
1349 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1350 determine whether watchdog notifications are requested by
1353 * Socket-activated per-connection services now include a
1354 short description of the connection parameters in the
1357 * tmpfiles gained a new "--boot" option. When this is not used,
1358 only lines where the command character is not suffixed with
1359 "!" are executed. When this option is specified, those
1360 options are executed too. This partitions tmpfiles
1361 directives into those that can be safely executed at any
1362 time, and those which should be run only at boot (for
1363 example, a line that creates /run/nologin).
1365 * A new API "sd-resolve.h" has been added which provides a simple
1366 asynchronous wrapper around glibc NSS host name resolution
1367 calls, such as getaddrinfo(). In contrast to glibc's
1368 getaddrinfo_a(), it does not use signals. In contrast to most
1369 other asynchronous name resolution libraries, this one does
1370 not reimplement DNS, but reuses NSS, so that alternate
1371 host name resolution systems continue to work, such as mDNS,
1372 LDAP, etc. This API is based on libasyncns, but it has been
1373 cleaned up for inclusion in systemd.
1375 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1376 "sd-daemon.h" are no longer found in individual libraries
1377 libsystemd-journal.so, libsystemd-login.so,
1378 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1379 merged them into a single library, libsystemd.so, which
1380 provides all symbols. The reason for this is cyclic
1381 dependencies, as these libraries tend to use each other's
1382 symbols. So far, we have managed to workaround that by linking
1383 a copy of a good part of our code into each of these
1384 libraries again and again, which, however, makes certain
1385 things hard to do, like sharing static variables. Also, it
1386 substantially increases footprint. With this change, there
1387 is only one library for the basic APIs systemd
1388 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1389 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1390 library as well, however are subject to the --enable-kdbus
1391 switch (see below). Note that "sd-dhcp-client.h" is not part
1392 of this library (this is because it only consumes, never
1393 provides, services of/to other APIs). To make the transition
1394 easy from the separate libraries to the unified one, we
1395 provide the --enable-compat-libs compile-time switch which
1396 will generate stub libraries that are compatible with the
1397 old ones but redirect all calls to the new one.
1399 * All of the kdbus logic and the new APIs "sd-bus.h",
1400 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1401 and "sd-utf8.h" are compile-time optional via the
1402 "--enable-kdbus" switch, and they are not compiled in by
1403 default. To make use of kdbus, you have to explicitly enable
1404 the switch. Note however, that neither the kernel nor the
1405 userspace API for all of this is considered stable yet. We
1406 want to maintain the freedom to still change the APIs for
1407 now. By specifying this build-time switch, you acknowledge
1408 that you are aware of the instability of the current
1411 * Also, note that while kdbus is pretty much complete,
1412 it lacks one thing: proper policy support. This means you
1413 can build a fully working system with all features; however,
1414 it will be highly insecure. Policy support will be added in
1415 one of the next releases, at the same time that we will
1416 declare the APIs stable.
1418 * When the kernel command-line argument "kdbus" is specified,
1419 systemd will automatically load the kdbus.ko kernel module. At
1420 this stage of development, it is only useful for testing kdbus
1421 and should not be used in production. Note: if "--enable-kdbus"
1422 is specified, and the kdbus.ko kernel module is available, and
1423 "kdbus" is added to the kernel command line, the entire system
1424 runs with kdbus instead of dbus-daemon, with the above mentioned
1425 problem of missing the system policy enforcement. Also a future
1426 version of kdbus.ko or a newer systemd will not be compatible with
1427 each other, and will unlikely be able to boot the machine if only
1428 one of them is updated.
1430 * systemctl gained a new "import-environment" command which
1431 uploads the caller's environment (or parts thereof) into the
1432 service manager so that it is inherited by services started
1433 by the manager. This is useful to upload variables like
1434 $DISPLAY into the user service manager.
1436 * A new PrivateDevices= switch has been added to service units
1437 which allows running a service with a namespaced /dev
1438 directory that does not contain any device nodes for
1439 physical devices. More specifically, it only includes devices
1440 such as /dev/null, /dev/urandom, and /dev/zero which are API
1443 * logind has been extended to support behaviour like VT
1444 switching on seats that do not support a VT. This makes
1445 multi-session available on seats that are not the first seat
1446 (seat0), and on systems where kernel support for VTs has
1447 been disabled at compile-time.
1449 * If a process holds a delay lock for system sleep or shutdown
1450 and fails to release it in time, we will now log its
1451 identity. This makes it easier to identify processes that
1452 cause slow suspends or power-offs.
1454 * When parsing /etc/crypttab, support for a new key-slot=
1455 option as supported by Debian is added. It allows indicating
1456 which LUKS slot to use on disk, speeding up key loading.
1458 * The sd_journald_sendv() API call has been checked and
1459 officially declared to be async-signal-safe so that it may
1460 be invoked from signal handlers for logging purposes.
1462 * Boot-time status output is now enabled automatically after a
1463 short timeout if boot does not progress, in order to give
1464 the user an indication what she or he is waiting for.
1466 * The boot-time output has been improved to show how much time
1467 remains until jobs expire.
1469 * The KillMode= switch in service units gained a new possible
1470 value "mixed". If set, and the unit is shut down, then the
1471 initial SIGTERM signal is sent only to the main daemon
1472 process, while the following SIGKILL signal is sent to
1473 all remaining processes of the service.
1475 * When a scope unit is registered, a new property "Controller"
1476 may be set. If set to a valid bus name, systemd will send a
1477 RequestStop() signal to this name when it would like to shut
1478 down the scope. This may be used to hook manager logic into
1479 the shutdown logic of scope units. Also, scope units may now
1480 be put in a special "abandoned" state, in which case the
1481 manager process which created them takes no further
1482 responsibilities for it.
1484 * When reading unit files, systemd will now verify
1485 the access mode of these files, and warn about certain
1486 suspicious combinations. This has been added to make it
1487 easier to track down packaging bugs where unit files are
1488 marked executable or world-writable.
1490 * systemd-nspawn gained a new "--setenv=" switch to set
1491 container-wide environment variables. The similar option in
1492 systemd-activate was renamed from "--environment=" to
1493 "--setenv=" for consistency.
1495 * systemd-nspawn has been updated to create a new kdbus domain
1496 for each container that is invoked, thus allowing each
1497 container to have its own set of system and user buses,
1498 independent of the host.
1500 * systemd-nspawn gained a new --drop-capability= switch to run
1501 the container with less capabilities than the default. Both
1502 --drop-capability= and --capability= now take the special
1503 string "all" for dropping or keeping all capabilities.
1505 * systemd-nspawn gained new switches for executing containers
1506 with specific SELinux labels set.
1508 * systemd-nspawn gained a new --quiet switch to not generate
1509 any additional output but the container's own console
1512 * systemd-nspawn gained a new --share-system switch to run a
1513 container without PID namespacing enabled.
1515 * systemd-nspawn gained a new --register= switch to control
1516 whether the container is registered with systemd-machined or
1517 not. This is useful for containers that do not run full
1518 OS images, but only specific apps.
1520 * systemd-nspawn gained a new --keep-unit which may be used
1521 when invoked as the only program from a service unit, and
1522 results in registration of the unit service itself in
1523 systemd-machined, instead of a newly opened scope unit.
1525 * systemd-nspawn gained a new --network-interface= switch for
1526 moving arbitrary interfaces to the container. The new
1527 --network-veth switch creates a virtual Ethernet connection
1528 between host and container. The new --network-bridge=
1529 switch then allows assigning the host side of this virtual
1530 Ethernet connection to a bridge device.
1532 * systemd-nspawn gained a new --personality= switch for
1533 setting the kernel personality for the container. This is
1534 useful when running a 32-bit container on a 64-bit host. A
1535 similar option Personality= is now also available for service
1538 * logind will now also track a "Desktop" identifier for each
1539 session which encodes the desktop environment of it. This is
1540 useful for desktop environments that want to identify
1541 multiple running sessions of itself easily.
1543 * A new SELinuxContext= setting for service units has been
1544 added that allows setting a specific SELinux execution
1545 context for a service.
1547 * Most systemd client tools will now honour $SYSTEMD_LESS for
1548 settings of the "less" pager. By default, these tools will
1549 override $LESS to allow certain operations to work, such as
1550 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1551 influence this logic.
1553 * systemd's "seccomp" hook-up has been changed to make use of
1554 the libseccomp library instead of using its own
1555 implementation. This has benefits for portability among
1558 * For usage together with SystemCallFilter=, a new
1559 SystemCallErrorNumber= setting has been introduced that
1560 allows configuration of a system error number to be returned
1561 on filtered system calls, instead of immediately killing the
1562 process. Also, SystemCallArchitectures= has been added to
1563 limit access to system calls of a particular architecture
1564 (in order to turn off support for unused secondary
1565 architectures). There is also a global
1566 SystemCallArchitectures= setting in system.conf now to turn
1567 off support for non-native system calls system-wide.
1569 * systemd requires a kernel with a working name_to_handle_at(),
1570 please see the kernel config requirements in the README file.
1572 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1573 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1574 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1575 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1576 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1577 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1578 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1579 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1580 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1581 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1582 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1583 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1584 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1585 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1586 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1587 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1588 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1589 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1590 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1591 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1592 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1593 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1594 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1595 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1597 -- Berlin, 2014-02-20
1601 * logind has gained support for facilitating privileged input
1602 and drm device access for unprivileged clients. This work is
1603 useful to allow Wayland display servers (and similar
1604 programs, such as kmscon) to run under the user's ID and
1605 access input and drm devices which are normally
1606 protected. When this is used (and the kernel is new enough)
1607 logind will "mute" IO on the file descriptors passed to
1608 Wayland as long as it is in the background and "unmute" it
1609 if it returns into the foreground. This allows secure
1610 session switching without allowing background sessions to
1611 eavesdrop on input and display data. This also introduces
1612 session switching support if VT support is turned off in the
1613 kernel, and on seats that are not seat0.
1615 * A new kernel command line option luks.options= is understood
1616 now which allows specifying LUKS options for usage for LUKS
1617 encrypted partitions specified with luks.uuid=.
1619 * tmpfiles.d(5) snippets may now use specifier expansion in
1620 path names. More specifically %m, %b, %H, %v, are now
1621 replaced by the local machine id, boot id, hostname, and
1622 kernel version number.
1624 * A new tmpfiles.d(5) command "m" has been introduced which
1625 may be used to change the owner/group/access mode of a file
1626 or directory if it exists, but do nothing if it does not.
1628 * This release removes high-level support for the
1629 MemorySoftLimit= cgroup setting. The underlying kernel
1630 cgroup attribute memory.soft_limit= is currently badly
1631 designed and likely to be removed from the kernel API in its
1632 current form, hence we should not expose it for now.
1634 * The memory.use_hierarchy cgroup attribute is now enabled for
1635 all cgroups systemd creates in the memory cgroup
1636 hierarchy. This option is likely to be come the built-in
1637 default in the kernel anyway, and the non-hierarchial mode
1638 never made much sense in the intrinsically hierarchial
1641 * A new field _SYSTEMD_SLICE= is logged along with all journal
1642 messages containing the slice a message was generated
1643 from. This is useful to allow easy per-customer filtering of
1644 logs among other things.
1646 * systemd-journald will no longer adjust the group of journal
1647 files it creates to the "systemd-journal" group. Instead we
1648 rely on the journal directory to be owned by the
1649 "systemd-journal" group, and its setgid bit set, so that the
1650 kernel file system layer will automatically enforce that
1651 journal files inherit this group assignment. The reason for
1652 this change is that we cannot allow NSS look-ups from
1653 journald which would be necessary to resolve
1654 "systemd-journal" to a numeric GID, because this might
1655 create deadlocks if NSS involves synchronous queries to
1656 other daemons (such as nscd, or sssd) which in turn are
1657 logging clients of journald and might block on it, which
1658 would then dead lock. A tmpfiles.d(5) snippet included in
1659 systemd will make sure the setgid bit and group are
1660 properly set on the journal directory if it exists on every
1661 boot. However, we recommend adjusting it manually after
1662 upgrades too (or from RPM scriptlets), so that the change is
1663 not delayed until next reboot.
1665 * Backlight and random seed files in /var/lib/ have moved into
1666 the /var/lib/systemd/ directory, in order to centralize all
1667 systemd generated files in one directory.
1669 * Boot time performance measurements (as displayed by
1670 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1671 performance information if that's available to determine how
1672 much time BIOS and boot loader initialization required. With
1673 a sufficiently new BIOS you hence no longer need to boot
1674 with Gummiboot to get access to such information.
1676 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1677 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1678 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1679 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1680 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1681 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1682 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1684 -- Berlin, 2013-10-02
1688 * The Restart= option for services now understands a new
1689 on-watchdog setting, which will restart the service
1690 automatically if the service stops sending out watchdog keep
1691 alive messages (as configured with WatchdogSec=).
1693 * The getty generator (which is responsible for bringing up a
1694 getty on configured serial consoles) will no longer only
1695 start a getty on the primary kernel console but on all
1696 others, too. This makes the order in which console= is
1697 specified on the kernel command line less important.
1699 * libsystemd-logind gained a new sd_session_get_vt() call to
1700 retrieve the VT number of a session.
1702 * If the option "tries=0" is set for an entry of /etc/crypttab
1703 its passphrase is queried indefinitely instead of any
1704 maximum number of tries.
1706 * If a service with a configure PID file terminates its PID
1707 file will now be removed automatically if it still exists
1708 afterwards. This should put an end to stale PID files.
1710 * systemd-run will now also take relative binary path names
1711 for execution and no longer insists on absolute paths.
1713 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1714 paths that are optionally prefixed with "-" to indicate that
1715 it should not be considered a failure if they do not exist.
1717 * journalctl -o (and similar commands) now understands a new
1718 output mode "short-precise", it is similar to "short" but
1719 shows timestamps with usec accuracy.
1721 * The option "discard" (as known from Debian) is now
1722 synonymous to "allow-discards" in /etc/crypttab. In fact,
1723 "discard" is preferred now (since it is easier to remember
1726 * Some licensing clean-ups were made, so that more code is now
1727 LGPL-2.1 licensed than before.
1729 * A minimal tool to save/restore the display backlight
1730 brightness across reboots has been added. It will store the
1731 backlight setting as late as possible at shutdown, and
1732 restore it as early as possible during reboot.
1734 * A logic to automatically discover and enable home and swap
1735 partitions on GPT disks has been added. With this in place
1736 /etc/fstab becomes optional for many setups as systemd can
1737 discover certain partitions located on the root disk
1738 automatically. Home partitions are recognized under their
1739 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1740 partitions are recognized under their GPT type ID
1741 0657fd6da4ab43c484e50933c84b4f4f.
1743 * systemd will no longer pass any environment from the kernel
1744 or initrd to system services. If you want to set an
1745 environment for all services, do so via the kernel command
1746 line systemd.setenv= assignment.
1748 * The systemd-sysctl tool no longer natively reads the file
1749 /etc/sysctl.conf. If desired, the file should be symlinked
1750 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1751 legacy support by a symlink rather than built-in code, it
1752 also makes the otherwise hidden order of application of the
1753 different files visible. (Note that this partly reverts to a
1754 pre-198 application order of sysctl knobs!)
1756 * The "systemctl set-log-level" and "systemctl dump" commands
1757 have been moved to systemd-analyze.
1759 * systemd-run learned the new --remain-after-exit switch,
1760 which causes the scope unit not to be cleaned up
1761 automatically after the process terminated.
1763 * tmpfiles learned a new --exclude-prefix= switch to exclude
1764 certain paths from operation.
1766 * journald will now automatically flush all messages to disk
1767 as soon as a message of the log priorities CRIT, ALERT or
1770 Contributions from: Andrew Cook, Brandon Philips, Christian
1771 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1772 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1773 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1774 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1775 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1776 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1777 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1778 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1779 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1780 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1781 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1782 William Giokas, Zbigniew Jędrzejewski-Szmek
1784 -- Berlin, 2013-09-13
1788 * The documentation has been updated to cover the various new
1789 concepts introduced with 205.
1791 * Unit files now understand the new %v specifier which
1792 resolves to the kernel version string as returned by "uname
1795 * systemctl now supports filtering the unit list output by
1796 load state, active state and sub state, using the new
1799 * "systemctl status" will now show the results of the
1800 condition checks (like ConditionPathExists= and similar) of
1801 the last start attempts of the unit. They are also logged to
1804 * "journalctl -b" may now be used to look for boot output of a
1805 specific boot. Try "journalctl -b -1" for the previous boot,
1806 but the syntax is substantially more powerful.
1808 * "journalctl --show-cursor" has been added which prints the
1809 cursor string the last shown log line. This may then be used
1810 with the new "journalctl --after-cursor=" switch to continue
1811 browsing logs from that point on.
1813 * "journalctl --force" may now be used to force regeneration
1816 * Creation of "dead" device nodes has been moved from udev
1817 into kmod and tmpfiles. Previously, udev would read the kmod
1818 databases to pre-generate dead device nodes based on meta
1819 information contained in kernel modules, so that these would
1820 be auto-loaded on access rather then at boot. As this
1821 does not really have much to do with the exposing actual
1822 kernel devices to userspace this has always been slightly
1823 alien in the udev codebase. Following the new scheme kmod
1824 will now generate a runtime snippet for tmpfiles from the
1825 module meta information and it now is tmpfiles' job to the
1826 create the nodes. This also allows overriding access and
1827 other parameters for the nodes using the usual tmpfiles
1828 facilities. As side effect this allows us to remove the
1829 CAP_SYS_MKNOD capability bit from udevd entirely.
1831 * logind's device ACLs may now be applied to these "dead"
1832 devices nodes too, thus finally allowing managed access to
1833 devices such as /dev/snd/sequencer whithout loading the
1834 backing module right-away.
1836 * A new RPM macro has been added that may be used to apply
1837 tmpfiles configuration during package installation.
1839 * systemd-detect-virt and ConditionVirtualization= now can
1840 detect User-Mode-Linux machines (UML).
1842 * journald will now implicitly log the effective capabilities
1843 set of processes in the message metadata.
1845 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1847 * The initrd interface has been simplified (more specifically,
1848 support for passing performance data via environment
1849 variables and fsck results via files in /run has been
1850 removed). These features were non-essential, and are
1851 nowadays available in a much nicer way by having systemd in
1852 the initrd serialize its state and have the hosts systemd
1853 deserialize it again.
1855 * The udev "keymap" data files and tools to apply keyboard
1856 specific mappings of scan to key codes, and force-release
1857 scan code lists have been entirely replaced by a udev
1858 "keyboard" builtin and a hwdb data file.
1860 * systemd will now honour the kernel's "quiet" command line
1861 argument also during late shutdown, resulting in a
1862 completely silent shutdown when used.
1864 * There's now an option to control the SO_REUSEPORT socket
1865 option in .socket units.
1867 * Instance units will now automatically get a per-template
1868 subslice of system.slice unless something else is explicitly
1869 configured. For example, instances of sshd@.service will now
1870 implicitly be placed in system-sshd.slice rather than
1871 system.slice as before.
1873 * Test coverage support may now be enabled at build time.
1875 Contributions from: Dave Reisner, Frederic Crozat, Harald
1876 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1877 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1878 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1879 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1880 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1881 Giokas, Zbigniew Jędrzejewski-Szmek
1883 -- Berlin, 2013-07-23
1887 * Two new unit types have been introduced:
1889 Scope units are very similar to service units, however, are
1890 created out of pre-existing processes -- instead of PID 1
1891 forking off the processes. By using scope units it is
1892 possible for system services and applications to group their
1893 own child processes (worker processes) in a powerful way
1894 which then maybe used to organize them, or kill them
1895 together, or apply resource limits on them.
1897 Slice units may be used to partition system resources in an
1898 hierarchial fashion and then assign other units to them. By
1899 default there are now three slices: system.slice (for all
1900 system services), user.slice (for all user sessions),
1901 machine.slice (for VMs and containers).
1903 Slices and scopes have been introduced primarily in
1904 context of the work to move cgroup handling to a
1905 single-writer scheme, where only PID 1
1906 creates/removes/manages cgroups.
1908 * There's a new concept of "transient" units. In contrast to
1909 normal units these units are created via an API at runtime,
1910 not from configuration from disk. More specifically this
1911 means it is now possible to run arbitrary programs as
1912 independent services, with all execution parameters passed
1913 in via bus APIs rather than read from disk. Transient units
1914 make systemd substantially more dynamic then it ever was,
1915 and useful as a general batch manager.
1917 * logind has been updated to make use of scope and slice units
1918 for managing user sessions. As a user logs in he will get
1919 his own private slice unit, to which all sessions are added
1920 as scope units. We also added support for automatically
1921 adding an instance of user@.service for the user into the
1922 slice. Effectively logind will no longer create cgroup
1923 hierarchies on its own now, it will defer entirely to PID 1
1924 for this by means of scope, service and slice units. Since
1925 user sessions this way become entities managed by PID 1
1926 the output of "systemctl" is now a lot more comprehensive.
1928 * A new mini-daemon "systemd-machined" has been added which
1929 may be used by virtualization managers to register local
1930 VMs/containers. nspawn has been updated accordingly, and
1931 libvirt will be updated shortly. machined will collect a bit
1932 of meta information about the VMs/containers, and assign
1933 them their own scope unit (see above). The collected
1934 meta-data is then made available via the "machinectl" tool,
1935 and exposed in "ps" and similar tools. machined/machinectl
1936 is compile-time optional.
1938 * As discussed earlier, the low-level cgroup configuration
1939 options ControlGroup=, ControlGroupModify=,
1940 ControlGroupPersistent=, ControlGroupAttribute= have been
1941 removed. Please use high-level attribute settings instead as
1942 well as slice units.
1944 * A new bus call SetUnitProperties() has been added to alter
1945 various runtime parameters of a unit. This is primarily
1946 useful to alter cgroup parameters dynamically in a nice way,
1947 but will be extended later on to make more properties
1948 modifiable at runtime. systemctl gained a new set-properties
1949 command that wraps this call.
1951 * A new tool "systemd-run" has been added which can be used to
1952 run arbitrary command lines as transient services or scopes,
1953 while configuring a number of settings via the command
1954 line. This tool is currently very basic, however already
1955 very useful. We plan to extend this tool to even allow
1956 queuing of execution jobs with time triggers from the
1957 command line, similar in fashion to "at".
1959 * nspawn will now inform the user explicitly that kernels with
1960 audit enabled break containers, and suggest the user to turn
1963 * Support for detecting the IMA and AppArmor security
1964 frameworks with ConditionSecurity= has been added.
1966 * journalctl gained a new "-k" switch for showing only kernel
1967 messages, mimicking dmesg output; in addition to "--user"
1968 and "--system" switches for showing only user's own logs
1971 * systemd-delta can now show information about drop-in
1972 snippets extending unit files.
1974 * libsystemd-bus has been substantially updated but is still
1975 not available as public API.
1977 * systemd will now look for the "debug" argument on the kernel
1978 command line and enable debug logging, similar to what
1979 "systemd.log_level=debug" already did before.
1981 * "systemctl set-default", "systemctl get-default" has been
1982 added to configure the default.target symlink, which
1983 controls what to boot into by default.
1985 * "systemctl set-log-level" has been added as a convenient
1986 way to raise and lower systemd logging threshold.
1988 * "systemd-analyze plot" will now show the time the various
1989 generators needed for execution, as well as information
1990 about the unit file loading.
1992 * libsystemd-journal gained a new sd_journal_open_files() call
1993 for opening specific journal files. journactl also gained a
1994 new switch to expose this new functionality. Previously we
1995 only supported opening all files from a directory, or all
1996 files from the system, as opening individual files only is
1997 racy due to journal file rotation.
1999 * systemd gained the new DefaultEnvironment= setting in
2000 /etc/systemd/system.conf to set environment variables for
2003 * If a privileged process logs a journal message with the
2004 OBJECT_PID= field set, then journald will automatically
2005 augment this with additional OBJECT_UID=, OBJECT_GID=,
2006 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2007 system services want to log events about specific client
2008 processes. journactl/systemctl has been updated to make use
2009 of this information if all log messages regarding a specific
2012 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2013 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2014 Reisner, David Coppa, David King, David Strauss, Eelco
2015 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2016 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2017 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2018 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2019 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2020 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2021 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2022 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2023 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2024 Łukasz Stelmach, 장동준
2028 * The Python bindings gained some minimal support for the APIs
2029 exposed by libsystemd-logind.
2031 * ConditionSecurity= gained support for detecting SMACK. Since
2032 this condition already supports SELinux and AppArmor we only
2033 miss IMA for this. Patches welcome!
2035 Contributions from: Karol Lewandowski, Lennart Poettering,
2036 Zbigniew Jędrzejewski-Szmek
2040 * systemd-nspawn will now create /etc/resolv.conf if
2041 necessary, before bind-mounting the host's file onto it.
2043 * systemd-nspawn will now store meta information about a
2044 container on the container's cgroup as extended attribute
2045 fields, including the root directory.
2047 * The cgroup hierarchy has been reworked in many ways. All
2048 objects any of the components systemd creates in the cgroup
2049 tree are now suffixed. More specifically, user sessions are
2050 now placed in cgroups suffixed with ".session", users in
2051 cgroups suffixed with ".user", and nspawn containers in
2052 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2053 names are now escaped in a simple scheme to avoid collision
2054 of userspace object names with kernel filenames. This work
2055 is preparation for making these objects relocatable in the
2056 cgroup tree, in order to allow easy resource partitioning of
2057 these objects without causing naming conflicts.
2059 * systemctl list-dependencies gained the new switches
2060 --plain, --reverse, --after and --before.
2062 * systemd-inhibit now shows the process name of processes that
2063 have taken an inhibitor lock.
2065 * nss-myhostname will now also resolve "localhost"
2066 implicitly. This makes /etc/hosts an optional file and
2067 nicely handles that on IPv6 ::1 maps to both "localhost" and
2070 * libsystemd-logind.so gained a new call
2071 sd_get_machine_names() to enumerate running containers and
2072 VMs (currently only supported by very new libvirt and
2073 nspawn). sd_login_monitor can now be used to watch
2074 VMs/containers coming and going.
2076 * .include is not allowed recursively anymore, and only in
2077 unit files. Usually it is better to use drop-in snippets in
2078 .d/*.conf anyway, as introduced with systemd 198.
2080 * systemd-analyze gained a new "critical-chain" command that
2081 determines the slowest chain of units run during system
2082 boot-up. It is very useful for tracking down where
2083 optimizing boot time is the most beneficial.
2085 * systemd will no longer allow manipulating service paths in
2086 the name=systemd:/system cgroup tree using ControlGroup= in
2087 units. (But is still fine with it in all other dirs.)
2089 * There's a new systemd-nspawn@.service service file that may
2090 be used to easily run nspawn containers as system
2091 services. With the container's root directory in
2092 /var/lib/container/foobar it is now sufficient to run
2093 "systemctl start systemd-nspawn@foobar.service" to boot it.
2095 * systemd-cgls gained a new parameter "--machine" to list only
2096 the processes within a certain container.
2098 * ConditionSecurity= now can check for "apparmor". We still
2099 are lacking checks for SMACK and IMA for this condition
2100 check though. Patches welcome!
2102 * A new configuration file /etc/systemd/sleep.conf has been
2103 added that may be used to configure which kernel operation
2104 systemd is supposed to execute when "suspend", "hibernate"
2105 or "hybrid-sleep" is requested. This makes the new kernel
2106 "freeze" state accessible to the user.
2108 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2109 the passed argument if applicable.
2111 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2112 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2113 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2114 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2115 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2116 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2117 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2122 * The output of 'systemctl list-jobs' got some polishing. The
2123 '--type=' argument may now be passed more than once. A new
2124 command 'systemctl list-sockets' has been added which shows
2125 a list of kernel sockets systemd is listening on with the
2126 socket units they belong to, plus the units these socket
2129 * The experimental libsystemd-bus library got substantial
2130 updates to work in conjunction with the (also experimental)
2131 kdbus kernel project. It works well enough to exchange
2132 messages with some sophistication. Note that kdbus is not
2133 ready yet, and the library is mostly an elaborate test case
2134 for now, and not installable.
2136 * systemd gained a new unit 'systemd-static-nodes.service'
2137 that generates static device nodes earlier during boot, and
2138 can run in conjunction with udev.
2140 * libsystemd-login gained a new call sd_pid_get_user_unit()
2141 to retrieve the user systemd unit a process is running
2142 in. This is useful for systems where systemd is used as
2145 * systemd-nspawn now places all containers in the new /machine
2146 top-level cgroup directory in the name=systemd
2147 hierarchy. libvirt will soon do the same, so that we get a
2148 uniform separation of /system, /user and /machine for system
2149 services, user processes and containers/virtual
2150 machines. This new cgroup hierarchy is also useful to stick
2151 stable names to specific container instances, which can be
2152 recognized later this way (this name may be controlled
2153 via systemd-nspawn's new -M switch). libsystemd-login also
2154 gained a new call sd_pid_get_machine_name() to retrieve the
2155 name of the container/VM a specific process belongs to.
2157 * bootchart can now store its data in the journal.
2159 * libsystemd-journal gained a new call
2160 sd_journal_add_conjunction() for AND expressions to the
2161 matching logic. This can be used to express more complex
2162 logical expressions.
2164 * journactl can now take multiple --unit= and --user-unit=
2167 * The cryptsetup logic now understands the "luks.key=" kernel
2168 command line switch for specifying a file to read the
2169 decryption key from. Also, if a configured key file is not
2170 found the tool will now automatically fall back to prompting
2173 * Python systemd.journal module was updated to wrap recently
2174 added functions from libsystemd-journal. The interface was
2175 changed to bring the low level interface in s.j._Reader
2176 closer to the C API, and the high level interface in
2177 s.j.Reader was updated to wrap and convert all data about
2180 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2181 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2182 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2183 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2184 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2185 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2189 * journalctl --update-catalog now understands a new --root=
2190 option to operate on catalogs found in a different root
2193 * During shutdown after systemd has terminated all running
2194 services a final killing loop kills all remaining left-over
2195 processes. We will now print the name of these processes
2196 when we send SIGKILL to them, since this usually indicates a
2199 * If /etc/crypttab refers to password files stored on
2200 configured mount points automatic dependencies will now be
2201 generated to ensure the specific mount is established first
2202 before the key file is attempted to be read.
2204 * 'systemctl status' will now show information about the
2205 network sockets a socket unit is listening on.
2207 * 'systemctl status' will also shown information about any
2208 drop-in configuration file for units. (Drop-In configuration
2209 files in this context are files such as
2210 /etc/systemd/systemd/foobar.service.d/*.conf)
2212 * systemd-cgtop now optionally shows summed up CPU times of
2213 cgroups. Press '%' while running cgtop to switch between
2214 percentage and absolute mode. This is useful to determine
2215 which cgroups use up the most CPU time over the entire
2216 runtime of the system. systemd-cgtop has also been updated
2217 to be 'pipeable' for processing with further shell tools.
2219 * 'hostnamectl set-hostname' will now allow setting of FQDN
2222 * The formatting and parsing of time span values has been
2223 changed. The parser now understands fractional expressions
2224 such as "5.5h". The formatter will now output fractional
2225 expressions for all time spans under 1min, i.e. "5.123456s"
2226 rather than "5s 123ms 456us". For time spans under 1s
2227 millisecond values are shown, for those under 1ms
2228 microsecond values are shown. This should greatly improve
2229 all time-related output of systemd.
2231 * libsystemd-login and libsystemd-journal gained new
2232 functions for querying the poll() events mask and poll()
2233 timeout value for integration into arbitrary event
2236 * localectl gained the ability to list available X11 keymaps
2237 (models, layouts, variants, options).
2239 * 'systemd-analyze dot' gained the ability to filter for
2240 specific units via shell-style globs, to create smaller,
2241 more useful graphs. I.e. it is now possible to create simple
2242 graphs of all the dependencies between only target units, or
2243 of all units that Avahi has dependencies with.
2245 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2246 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2247 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2248 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2249 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2250 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2251 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2255 * The boot-time readahead implementation for rotating media
2256 will now read the read-ahead data in multiple passes which
2257 consist of all read requests made in equidistant time
2258 intervals. This means instead of strictly reading read-ahead
2259 data in its physical order on disk we now try to find a
2260 middle ground between physical and access time order.
2262 * /etc/os-release files gained a new BUILD_ID= field for usage
2263 on operating systems that provide continuous builds of OS
2266 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2267 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2268 William Douglas, Zbigniew Jędrzejewski-Szmek
2272 * systemd-python gained an API exposing libsystemd-daemon.
2274 * The SMACK setup logic gained support for uploading CIPSO
2277 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2278 ReadOnlyDirectories= and InaccessibleDirectories= has
2279 changed. The private /tmp and /var/tmp directories are now
2280 shared by all processes of a service (which means
2281 ExecStartPre= may now leave data in /tmp that ExecStart= of
2282 the same service can still access). When a service is
2283 stopped its temporary directories are immediately deleted
2284 (normal clean-up with tmpfiles is still done in addition to
2287 * By default, systemd will now set a couple of sysctl
2288 variables in the kernel: the safe sysrq options are turned
2289 on, IP route verification is turned on, and source routing
2290 disabled. The recently added hardlink and softlink
2291 protection of the kernel is turned on. These settings should
2292 be reasonably safe, and good defaults for all new systems.
2294 * The predictable network naming logic may now be turned off
2295 with a new kernel command line switch: net.ifnames=0.
2297 * A new libsystemd-bus module has been added that implements a
2298 pretty complete D-Bus client library. For details see:
2300 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2302 * journald will now explicitly flush the journal files to disk
2303 at the latest 5min after each write. The file will then also
2304 be marked offline until the next write. This should increase
2305 reliability in case of a crash. The synchronization delay
2306 can be configured via SyncIntervalSec= in journald.conf.
2308 * There's a new remote-fs-setup.target unit that can be used
2309 to pull in specific services when at least one remote file
2310 system is to be mounted.
2312 * There are new targets timers.target and paths.target as
2313 canonical targets to pull user timer and path units in
2314 from. This complements sockets.target with a similar
2315 purpose for socket units.
2317 * libudev gained a new call udev_device_set_attribute_value()
2318 to set sysfs attributes of a device.
2320 * The udev daemon now sets the default number of worker
2321 processes executed in parallel based on the number of available
2322 CPUs instead of the amount of available RAM. This is supposed
2323 to provide a more reliable default and limit a too aggressive
2324 paralellism for setups with 1000s of devices connected.
2326 Contributions from: Auke Kok, Colin Walters, Cristian
2327 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2328 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2329 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2330 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2331 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2332 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2333 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2334 Zbigniew Jędrzejewski-Szmek
2338 * Configuration of unit files may now be extended via drop-in
2339 files without having to edit/override the unit files
2340 themselves. More specifically, if the administrator wants to
2341 change one value for a service file foobar.service he can
2342 now do so by dropping in a configuration snippet into
2343 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2344 will load all these snippets and apply them on top of the
2345 main unit configuration file, possibly extending or
2346 overriding its settings. Using these drop-in snippets is
2347 generally nicer than the two earlier options for changing
2348 unit files locally: copying the files from
2349 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2350 them there; or creating a new file in /etc/systemd/system/
2351 that incorporates the original one via ".include". Drop-in
2352 snippets into these .d/ directories can be placed in any
2353 directory systemd looks for units in, and the usual
2354 overriding semantics between /usr/lib, /etc and /run apply
2357 * Most unit file settings which take lists of items can now be
2358 reset by assigning the empty string to them. For example,
2359 normally, settings such as Environment=FOO=BAR append a new
2360 environment variable assignment to the environment block,
2361 each time they are used. By assigning Environment= the empty
2362 string the environment block can be reset to empty. This is
2363 particularly useful with the .d/*.conf drop-in snippets
2364 mentioned above, since this adds the ability to reset list
2365 settings from vendor unit files via these drop-ins.
2367 * systemctl gained a new "list-dependencies" command for
2368 listing the dependencies of a unit recursively.
2370 * Inhibitors are now honored and listed by "systemctl
2371 suspend", "systemctl poweroff" (and similar) too, not only
2372 GNOME. These commands will also list active sessions by
2375 * Resource limits (as exposed by the various control group
2376 controllers) can now be controlled dynamically at runtime
2377 for all units. More specifically, you can now use a command
2378 like "systemctl set-cgroup-attr foobar.service cpu.shares
2379 2000" to alter the CPU shares a specific service gets. These
2380 settings are stored persistently on disk, and thus allow the
2381 administrator to easily adjust the resource usage of
2382 services with a few simple commands. This dynamic resource
2383 management logic is also available to other programs via the
2384 bus. Almost any kernel cgroup attribute and controller is
2387 * systemd-vconsole-setup will now copy all font settings to
2388 all allocated VTs, where it previously applied them only to
2391 * libsystemd-login gained the new sd_session_get_tty() API
2394 * This release drops support for a few legacy or
2395 distribution-specific LSB facility names when parsing init
2396 scripts: $x-display-manager, $mail-transfer-agent,
2397 $mail-transport-agent, $mail-transfer-agent, $smtp,
2398 $null. Also, the mail-transfer-agent.target unit backing
2399 this has been removed. Distributions which want to retain
2400 compatibility with this should carry the burden for
2401 supporting this themselves and patch support for these back
2402 in, if they really need to. Also, the facilities $syslog and
2403 $local_fs are now ignored, since systemd does not support
2404 early-boot LSB init scripts anymore, and these facilities
2405 are implied anyway for normal services. syslog.target has
2408 * There are new bus calls on PID1's Manager object for
2409 cancelling jobs, and removing snapshot units. Previously,
2410 both calls were only available on the Job and Snapshot
2413 * systemd-journal-gatewayd gained SSL support.
2415 * The various "environment" files, such as /etc/locale.conf
2416 now support continuation lines with a backslash ("\") as
2417 last character in the line, similarly in style (but different)
2418 to how this is supported in shells.
2420 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2421 now implicitly appended to every log entry logged. systemctl
2422 has been updated to filter by this field when operating on a
2423 user systemd instance.
2425 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2426 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2427 the container. This makes it easier to boot unmodified
2428 Fedora systems in a container, which however still requires
2429 audit=0 to be passed on the kernel command line. Auditing in
2430 kernel and userspace is unfortunately still too broken in
2431 context of containers, hence we recommend compiling it out
2432 of the kernel or using audit=0. Hopefully this will be fixed
2433 one day for good in the kernel.
2435 * nspawn gained the new --bind= and --bind-ro= parameters to
2436 bind mount specific directories from the host into the
2439 * nspawn will now mount its own devpts file system instance
2440 into the container, in order not to leak pty devices from
2441 the host into the container.
2443 * systemd will now read the firmware boot time performance
2444 information from the EFI variables, if the used boot loader
2445 supports this, and takes it into account for boot performance
2446 analysis via "systemd-analyze". This is currently supported
2447 only in conjunction with Gummiboot, but could be supported
2448 by other boot loaders too. For details see:
2450 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2452 * A new generator has been added that automatically mounts the
2453 EFI System Partition (ESP) to /boot, if that directory
2454 exists, is empty, and no other file system has been
2455 configured to be mounted there.
2457 * logind will now send out PrepareForSleep(false) out
2458 unconditionally, after coming back from suspend. This may be
2459 used by applications as asynchronous notification for
2460 system resume events.
2462 * "systemctl unlock-sessions" has been added, that allows
2463 unlocking the screens of all user sessions at once, similar
2464 to how "systemctl lock-sessions" already locked all users
2465 sessions. This is backed by a new D-Bus call UnlockSessions().
2467 * "loginctl seat-status" will now show the master device of a
2468 seat. (i.e. the device of a seat that needs to be around for
2469 the seat to be considered available, usually the graphics
2472 * tmpfiles gained a new "X" line type, that allows
2473 configuration of files and directories (with wildcards) that
2474 shall be excluded from automatic cleanup ("aging").
2476 * udev default rules set the device node permissions now only
2477 at "add" events, and do not change them any longer with a
2478 later "change" event.
2480 * The log messages for lid events and power/sleep keypresses
2481 now carry a message ID.
2483 * We now have a substantially larger unit test suite, but this
2484 continues to be work in progress.
2486 * udevadm hwdb gained a new --root= parameter to change the
2487 root directory to operate relative to.
2489 * logind will now issue a background sync() request to the kernel
2490 early at shutdown, so that dirty buffers are flushed to disk early
2491 instead of at the last moment, in order to optimize shutdown
2494 * A new bootctl tool has been added that is an interface for
2495 certain boot loader operations. This is currently a preview
2496 and is likely to be extended into a small mechanism daemon
2497 like timedated, localed, hostnamed, and can be used by
2498 graphical UIs to enumerate available boot options, and
2499 request boot into firmware operations.
2501 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2502 the rest of the package. It also has been updated to work
2503 correctly in initrds.
2505 * Policykit previously has been runtime optional, and is now
2506 also compile time optional via a configure switch.
2508 * systemd-analyze has been reimplemented in C. Also "systemctl
2509 dot" has moved into systemd-analyze.
2511 * "systemctl status" with no further parameters will now print
2512 the status of all active or failed units.
2514 * Operations such as "systemctl start" can now be executed
2515 with a new mode "--irreversible" which may be used to queue
2516 operations that cannot accidentally be reversed by a later
2517 job queuing. This is by default used to make shutdown
2518 requests more robust.
2520 * The Python API of systemd now gained a new module for
2521 reading journal files.
2523 * A new tool kernel-install has been added that can install
2524 kernel images according to the Boot Loader Specification:
2526 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2528 * Boot time console output has been improved to provide
2529 animated boot time output for hanging jobs.
2531 * A new tool systemd-activate has been added which can be used
2532 to test socket activation with, directly from the command
2533 line. This should make it much easier to test and debug
2534 socket activation in daemons.
2536 * journalctl gained a new "--reverse" (or -r) option to show
2537 journal output in reverse order (i.e. newest line first).
2539 * journalctl gained a new "--pager-end" (or -e) option to jump
2540 to immediately jump to the end of the journal in the
2541 pager. This is only supported in conjunction with "less".
2543 * journalctl gained a new "--user-unit=" option, that works
2544 similarly to "--unit=" but filters for user units rather than
2547 * A number of unit files to ease adoption of systemd in
2548 initrds has been added. This moves some minimal logic from
2549 the various initrd implementations into systemd proper.
2551 * The journal files are now owned by a new group
2552 "systemd-journal", which exists specifically to allow access
2553 to the journal, and nothing else. Previously, we used the
2554 "adm" group for that, which however possibly covers more
2555 than just journal/log file access. This new group is now
2556 already used by systemd-journal-gatewayd to ensure this
2557 daemon gets access to the journal files and as little else
2558 as possible. Note that "make install" will also set FS ACLs
2559 up for /var/log/journal to give "adm" and "wheel" read
2560 access to it, in addition to "systemd-journal" which owns
2561 the journal files. We recommend that packaging scripts also
2562 add read access to "adm" + "wheel" to /var/log/journal, and
2563 all existing/future journal files. To normal users and
2564 administrators little changes, however packagers need to
2565 ensure to create the "systemd-journal" system group at
2566 package installation time.
2568 * The systemd-journal-gatewayd now runs as unprivileged user
2569 systemd-journal-gateway:systemd-journal-gateway. Packaging
2570 scripts need to create these system user/group at
2573 * timedated now exposes a new boolean property CanNTP that
2574 indicates whether a local NTP service is available or not.
2576 * systemd-detect-virt will now also detect xen PVs
2578 * The pstore file system is now mounted by default, if it is
2581 * In addition to the SELinux and IMA policies we will now also
2582 load SMACK policies at early boot.
2584 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2585 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2586 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2587 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2588 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2589 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2590 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2591 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2592 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2593 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2594 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2595 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2596 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2597 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2601 * Timer units now support calendar time events in addition to
2602 monotonic time events. That means you can now trigger a unit
2603 based on a calendar time specification such as "Thu,Fri
2604 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2605 or fifth day of any month of the year 2013, given that it is
2606 a thursday or friday. This brings timer event support
2607 considerably closer to cron's capabilities. For details on
2608 the supported calendar time specification language see
2611 * udev now supports a number of different naming policies for
2612 network interfaces for predictable names, and a combination
2613 of these policies is now the default. Please see this wiki
2614 document for details:
2616 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2618 * Auke Kok's bootchart implementation has been added to the
2619 systemd tree. It is an optional component that can graph the
2620 boot in quite some detail. It is one of the best bootchart
2621 implementations around and minimal in its code and
2624 * nss-myhostname has been integrated into the systemd source
2625 tree. nss-myhostname guarantees that the local hostname
2626 always stays resolvable via NSS. It has been a weak
2627 requirement of systemd-hostnamed since a long time, and
2628 since its code is actually trivial we decided to just
2629 include it in systemd's source tree. It can be turned off
2630 with a configure switch.
2632 * The read-ahead logic is now capable of properly detecting
2633 whether a btrfs file system is on SSD or rotating media, in
2634 order to optimize the read-ahead scheme. Previously, it was
2635 only capable of detecting this on traditional file systems
2638 * In udev, additional device properties are now read from the
2639 IAB in addition to the OUI database. Also, Bluetooth company
2640 identities are attached to the devices as well.
2642 * In service files %U may be used as specifier that is
2643 replaced by the configured user name of the service.
2645 * nspawn may now be invoked without a controlling TTY. This
2646 makes it suitable for invocation as its own service. This
2647 may be used to set up a simple containerized server system
2648 using only core OS tools.
2650 * systemd and nspawn can now accept socket file descriptors
2651 when they are started for socket activation. This enables
2652 implementation of socket activated nspawn
2653 containers. i.e. think about autospawning an entire OS image
2654 when the first SSH or HTTP connection is received. We expect
2655 that similar functionality will also be added to libvirt-lxc
2658 * journalctl will now suppress ANSI color codes when
2659 presenting log data.
2661 * systemctl will no longer show control group information for
2662 a unit if a the control group is empty anyway.
2664 * logind can now automatically suspend/hibernate/shutdown the
2667 * /etc/machine-info and hostnamed now also expose the chassis
2668 type of the system. This can be used to determine whether
2669 the local system is a laptop, desktop, handset or
2670 tablet. This information may either be configured by the
2671 user/vendor or is automatically determined from ACPI and DMI
2672 information if possible.
2674 * A number of PolicyKit actions are now bound together with
2675 "imply" rules. This should simplify creating UIs because
2676 many actions will now authenticate similar ones as well.
2678 * Unit files learnt a new condition ConditionACPower= which
2679 may be used to conditionalize a unit depending on whether an
2680 AC power source is connected or not, of whether the system
2681 is running on battery power.
2683 * systemctl gained a new "is-failed" verb that may be used in
2684 shell scripts and suchlike to check whether a specific unit
2685 is in the "failed" state.
2687 * The EnvironmentFile= setting in unit files now supports file
2688 globbing, and can hence be used to easily read a number of
2689 environment files at once.
2691 * systemd will no longer detect and recognize specific
2692 distributions. All distribution-specific #ifdeffery has been
2693 removed, systemd is now fully generic and
2694 distribution-agnostic. Effectively, not too much is lost as
2695 a lot of the code is still accessible via explicit configure
2696 switches. However, support for some distribution specific
2697 legacy configuration file formats has been dropped. We
2698 recommend distributions to simply adopt the configuration
2699 files everybody else uses now and convert the old
2700 configuration from packaging scripts. Most distributions
2701 already did that. If that's not possible or desirable,
2702 distributions are welcome to forward port the specific
2703 pieces of code locally from the git history.
2705 * When logging a message about a unit systemd will now always
2706 log the unit name in the message meta data.
2708 * localectl will now also discover system locale data that is
2709 not stored in locale archives, but directly unpacked.
2711 * logind will no longer unconditionally use framebuffer
2712 devices as seat masters, i.e. as devices that are required
2713 to be existing before a seat is considered preset. Instead,
2714 it will now look for all devices that are tagged as
2715 "seat-master" in udev. By default framebuffer devices will
2716 be marked as such, but depending on local systems other
2717 devices might be marked as well. This may be used to
2718 integrate graphics cards using closed source drivers (such
2719 as NVidia ones) more nicely into logind. Note however, that
2720 we recommend using the open source NVidia drivers instead,
2721 and no udev rules for the closed-source drivers will be
2722 shipped from us upstream.
2724 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2725 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2726 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2727 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2728 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2729 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2730 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2731 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2732 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2733 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2734 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2735 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2740 * udev gained support for loading additional device properties
2741 from an indexed database that is keyed by vendor/product IDs
2742 and similar device identifiers. For the beginning this
2743 "hwdb" is populated with data from the well-known PCI and
2744 USB database, but also includes PNP, ACPI and OID data. In
2745 the longer run this indexed database shall grow into
2746 becoming the one central database for non-essential
2747 userspace device metadata. Previously, data from the PCI/USB
2748 database was only attached to select devices, since the
2749 lookup was a relatively expensive operation due to O(n) time
2750 complexity (with n being the number of entries in the
2751 database). Since this is now O(1), we decided to add in this
2752 data for all devices where this is available, by
2753 default. Note that the indexed database needs to be rebuilt
2754 when new data files are installed. To achieve this you need
2755 to update your packaging scripts to invoke "udevadm hwdb
2756 --update" after installation of hwdb data files. For
2757 RPM-based distributions we introduced the new
2758 %udev_hwdb_update macro for this purpose.
2760 * The Journal gained support for the "Message Catalog", an
2761 indexed database to link up additional information with
2762 journal entries. For further details please check:
2764 http://www.freedesktop.org/wiki/Software/systemd/catalog
2766 The indexed message catalog database also needs to be
2767 rebuilt after installation of message catalog files. Use
2768 "journalctl --update-catalog" for this. For RPM-based
2769 distributions we introduced the %journal_catalog_update
2770 macro for this purpose.
2772 * The Python Journal bindings gained support for the standard
2773 Python logging framework.
2775 * The Journal API gained new functions for checking whether
2776 the underlying file system of a journal file is capable of
2777 properly reporting file change notifications, or whether
2778 applications that want to reflect journal changes "live"
2779 need to recheck journal files continuously in appropriate
2782 * It is now possible to set the "age" field for tmpfiles
2783 entries to 0, indicating that files matching this entry
2784 shall always be removed when the directories are cleaned up.
2786 * coredumpctl gained a new "gdb" verb which invokes gdb
2787 right-away on the selected coredump.
2789 * There's now support for "hybrid sleep" on kernels that
2790 support this, in addition to "suspend" and "hibernate". Use
2791 "systemctl hybrid-sleep" to make use of this.
2793 * logind's HandleSuspendKey= setting (and related settings)
2794 now gained support for a new "lock" setting to simply
2795 request the screen lock on all local sessions, instead of
2796 actually executing a suspend or hibernation.
2798 * systemd will now mount the EFI variables file system by
2801 * Socket units now gained support for configuration of the
2802 SMACK security label.
2804 * timedatectl will now output the time of the last and next
2805 daylight saving change.
2807 * We dropped support for various legacy and distro-specific
2808 concepts, such as insserv, early-boot SysV services
2809 (i.e. those for non-standard runlevels such as 'b' or 'S')
2810 or ArchLinux /etc/rc.conf support. We recommend the
2811 distributions who still need support this to either continue
2812 to maintain the necessary patches downstream, or find a
2813 different solution. (Talk to us if you have questions!)
2815 * Various systemd components will now bypass PolicyKit checks
2816 for root and otherwise handle properly if PolicyKit is not
2817 found to be around. This should fix most issues for
2818 PolicyKit-less systems. Quite frankly this should have been
2819 this way since day one. It is absolutely our intention to
2820 make systemd work fine on PolicyKit-less systems, and we
2821 consider it a bug if something does not work as it should if
2822 PolicyKit is not around.
2824 * For embedded systems it is now possible to build udev and
2825 systemd without blkid and/or kmod support.
2827 * "systemctl switch-root" is now capable of switching root
2828 more than once. I.e. in addition to transitions from the
2829 initrd to the host OS it is now possible to transition to
2830 further OS images from the host. This is useful to implement
2831 offline updating tools.
2833 * Various other additions have been made to the RPM macros
2834 shipped with systemd. Use %udev_rules_update() after
2835 installing new udev rules files. %_udevhwdbdir,
2836 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2837 %_sysctldir are now available which resolve to the right
2838 directories for packages to place various data files in.
2840 * journalctl gained the new --full switch (in addition to
2841 --all, to disable ellipsation for long messages.
2843 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2844 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2845 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2846 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2847 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2848 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2849 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2850 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2851 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2855 * journalctl gained new --since= and --until= switches to
2856 filter by time. It also now supports nice filtering for
2857 units via --unit=/-u.
2859 * Type=oneshot services may use ExecReload= and do the
2862 * The journal daemon now supports time-based rotation and
2863 vacuuming, in addition to the usual disk-space based
2866 * The journal will now index the available field values for
2867 each field name. This enables clients to show pretty drop
2868 downs of available match values when filtering. The bash
2869 completion of journalctl has been updated
2870 accordingly. journalctl gained a new switch -F to list all
2871 values a certain field takes in the journal database.
2873 * More service events are now written as structured messages
2874 to the journal, and made recognizable via message IDs.
2876 * The timedated, localed and hostnamed mini-services which
2877 previously only provided support for changing time, locale
2878 and hostname settings from graphical DEs such as GNOME now
2879 also have a minimal (but very useful) text-based client
2880 utility each. This is probably the nicest way to changing
2881 these settings from the command line now, especially since
2882 it lists available options and is fully integrated with bash
2885 * There's now a new tool "systemd-coredumpctl" to list and
2886 extract coredumps from the journal.
2888 * We now install a README each in /var/log/ and
2889 /etc/rc.d/init.d explaining where the system logs and init
2890 scripts went. This hopefully should help folks who go to
2891 that dirs and look into the otherwise now empty void and
2892 scratch their heads.
2894 * When user-services are invoked (by systemd --user) the
2895 $MANAGERPID env var is set to the PID of systemd.
2897 * SIGRTMIN+24 when sent to a --user instance will now result
2898 in immediate termination of systemd.
2900 * gatewayd received numerous feature additions such as a
2901 "follow" mode, for live syncing and filtering.
2903 * browse.html now allows filtering and showing detailed
2904 information on specific entries. Keyboard navigation and
2905 mouse screen support has been added.
2907 * gatewayd/journalctl now supports HTML5/JSON
2908 Server-Sent-Events as output.
2910 * The SysV init script compatibility logic will now
2911 heuristically determine whether a script supports the
2912 "reload" verb, and only then make this available as
2915 * "systemctl status --follow" has been removed, use "journalctl
2918 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
2919 have been removed since they are hardly useful to be
2922 * And I'd like to take the opportunity to specifically mention
2923 Zbigniew for his great contributions. Zbigniew, you rock!
2925 Contributions from: Andrew Eikum, Christian Hesse, Colin
2926 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
2927 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
2928 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
2929 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
2930 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
2931 Jędrzejewski-Szmek, Сковорода Никита Андреевич
2935 * If /etc/vconsole.conf is non-existent or empty we will no
2936 longer load any console font or key map at boot by
2937 default. Instead the kernel defaults will be left
2938 intact. This is definitely the right thing to do, as no
2939 configuration should mean no configuration, and hard-coding
2940 font names that are different on all archs is probably a bad
2941 idea. Also, the kernel default key map and font should be
2942 good enough for most cases anyway, and mostly identical to
2943 the userspace fonts/key maps we previously overloaded them
2944 with. If distributions want to continue to default to a
2945 non-kernel font or key map they should ship a default
2946 /etc/vconsole.conf with the appropriate contents.
2948 Contributions from: Colin Walters, Daniel J Walsh, Dave
2949 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
2950 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2954 * journalctl gained a new --cursor= switch to show entries
2955 starting from the specified location in the journal.
2957 * We now enforce a size limit on journal entry fields exported
2958 with "-o json" in journalctl. Fields larger than 4K will be
2959 assigned null. This can be turned off with --all.
2961 * An (optional) journal gateway daemon is now available as
2962 "systemd-journal-gatewayd.service". This service provides
2963 access to the journal via HTTP and JSON. This functionality
2964 will be used to implement live log synchronization in both
2965 pull and push modes, but has various other users too, such
2966 as easy log access for debugging of embedded devices. Right
2967 now it is already useful to retrieve the journal via HTTP:
2969 # systemctl start systemd-journal-gatewayd.service
2970 # wget http://localhost:19531/entries
2972 This will download the journal contents in a
2973 /var/log/messages compatible format. The same as JSON:
2975 # curl -H"Accept: application/json" http://localhost:19531/entries
2977 This service is also accessible via a web browser where a
2978 single static HTML5 app is served that uses the JSON logic
2979 to enable the user to do some basic browsing of the
2980 journal. This will be extended later on. Here's an example
2981 screenshot of this app in its current state:
2983 http://0pointer.de/public/journal-gatewayd
2985 Contributions from: Kay Sievers, Lennart Poettering, Robert
2986 Milasan, Tom Gundersen
2990 * The bash completion logic is now available for journalctl
2993 * We do not mount the "cpuset" controller anymore together with
2994 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
2995 started if no parameters are assigned to it. "cpuset" hence
2996 broke code that assumed it it could create "cpu" groups and
2999 * journalctl -f will now subscribe to terminal size changes,
3000 and line break accordingly.
3002 Contributions from: Dave Reisner, Kay Sievers, Lennart
3003 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3007 * nspawn will now create a symlink /etc/localtime in the
3008 container environment, copying the host's timezone
3009 setting. Previously this has been done via a bind mount, but
3010 since symlinks cannot be bind mounted this has now been
3011 changed to create/update the appropriate symlink.
3013 * journalctl -n's line number argument is now optional, and
3014 will default to 10 if omitted.
3016 * journald will now log the maximum size the journal files may
3017 take up on disk. This is particularly useful if the default
3018 built-in logic of determining this parameter from the file
3019 system size is used. Use "systemctl status
3020 systemd-journald.service" to see this information.
3022 * The multi-seat X wrapper tool has been stripped down. As X
3023 is now capable of enumerating graphics devices via udev in a
3024 seat-aware way the wrapper is not strictly necessary
3025 anymore. A stripped down temporary stop-gap is still shipped
3026 until the upstream display managers have been updated to
3027 fully support the new X logic. Expect this wrapper to be
3028 removed entirely in one of the next releases.
3030 * HandleSleepKey= in logind.conf has been split up into
3031 HandleSuspendKey= and HandleHibernateKey=. The old setting
3032 is not available anymore. X11 and the kernel are
3033 distuingishing between these keys and we should too. This
3034 also means the inhibition lock for these keys has been split
3037 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3038 Poettering, Lukas Nykryn, Václav Pavlín
3042 * Whenever a unit changes state we will now log this to the
3043 journal and show along the unit's own log output in
3046 * ConditionPathIsMountPoint= can now properly detect bind
3047 mount points too. (Previously, a bind mount of one file
3048 system to another place in the same file system could not be
3049 detected as mount, since they shared struct stat's st_dev
3052 * We will now mount the cgroup controllers cpu, cpuacct,
3053 cpuset and the controllers net_cls, net_prio together by
3056 * nspawn containers will now have a virtualized boot
3057 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3058 over with a randomized ID at container initialization). This
3059 has the effect of making "journalctl -b" do the right thing
3062 * The JSON output journal serialization has been updated not
3063 to generate "endless" list objects anymore, but rather one
3064 JSON object per line. This is more in line how most JSON
3065 parsers expect JSON objects. The new output mode
3066 "json-pretty" has been added to provide similar output, but
3067 neatly aligned for readability by humans.
3069 * We dropped all explicit sync() invocations in the shutdown
3070 code. The kernel does this implicitly anyway in the kernel
3071 reboot() syscall. halt(8)'s -n option is now a compatibility
3074 * We now support virtualized reboot() in containers, as
3075 supported by newer kernels. We will fall back to exit() if
3076 CAP_SYS_REBOOT is not available to the container. Also,
3077 nspawn makes use of this now and will actually reboot the
3078 container if the containerized OS asks for that.
3080 * journalctl will only show local log output by default
3081 now. Use --merge (-m) to show remote log output, too.
3083 * libsystemd-journal gained the new sd_journal_get_usage()
3084 call to determine the current disk usage of all journal
3085 files. This is exposed in the new "journalctl --disk-usage"
3088 * journald gained a new configuration setting SplitMode= in
3089 journald.conf which may be used to control how user journals
3090 are split off. See journald.conf(5) for details.
3092 * A new condition type ConditionFileNotEmpty= has been added.
3094 * tmpfiles' "w" lines now support file globbing, to write
3095 multiple files at once.
3097 * We added Python bindings for the journal submission
3098 APIs. More Python APIs for a number of selected APIs will
3099 likely follow. Note that we intend to add native bindings
3100 only for the Python language, as we consider it common
3101 enough to deserve bindings shipped within systemd. There are
3102 various projects outside of systemd that provide bindings
3103 for languages such as PHP or Lua.
3105 * Many conditions will now resolve specifiers such as %i. In
3106 addition, PathChanged= and related directives of .path units
3107 now support specifiers as well.
3109 * There's now a new RPM macro definition for the system preset
3112 * journald will now warn if it ca not forward a message to the
3113 syslog daemon because its socket is full.
3115 * timedated will no longer write or process /etc/timezone,
3116 except on Debian. As we do not support late mounted /usr
3117 anymore /etc/localtime always being a symlink is now safe,
3118 and hence the information in /etc/timezone is not necessary
3121 * logind will now always reserve one VT for a text getty (VT6
3122 by default). Previously if more than 6 X sessions where
3123 started they took up all the VTs with auto-spawned gettys,
3124 so that no text gettys were available anymore.
3126 * udev will now automatically inform the btrfs kernel logic
3127 about btrfs RAID components showing up. This should make
3128 simple hotplug based btrfs RAID assembly work.
3130 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3131 (but not for its children which will stay at the kernel
3132 default). This should allow setups with a lot more listening
3135 * systemd will now always pass the configured timezone to the
3136 kernel at boot. timedated will do the same when the timezone
3139 * logind's inhibition logic has been updated. By default,
3140 logind will now handle the lid switch, the power and sleep
3141 keys all the time, even in graphical sessions. If DEs want
3142 to handle these events on their own they should take the new
3143 handle-power-key, handle-sleep-key and handle-lid-switch
3144 inhibitors during their runtime. A simple way to achiveve
3145 that is to invoke the DE wrapped in an invocation of:
3147 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3149 * Access to unit operations is now checked via SELinux taking
3150 the unit file label and client process label into account.
3152 * systemd will now notify the administrator in the journal
3153 when he over-mounts a non-empty directory.
3155 * There are new specifiers that are resolved in unit files,
3156 for the host name (%H), the machine ID (%m) and the boot ID
3159 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3160 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3161 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3162 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3163 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3164 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3165 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3169 * Support for reading structured kernel messages from
3170 /dev/kmsg has now been added and is enabled by default.
3172 * Support for reading kernel messages from /proc/kmsg has now
3173 been removed. If you want kernel messages in the journal
3174 make sure to run a recent kernel (>= 3.5) that supports
3175 reading structured messages from /dev/kmsg (see
3176 above). /proc/kmsg is now exclusive property of classic
3177 syslog daemons again.
3179 * The libudev API gained the new
3180 udev_device_new_from_device_id() call.
3182 * The logic for file system namespace (ReadOnlyDirectory=,
3183 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3184 require pivot_root() anymore. This means fewer temporary
3185 directories are created below /tmp for this feature.
3187 * nspawn containers will now see and receive all submounts
3188 made on the host OS below the root file system of the
3191 * Forward Secure Sealing is now supported for Journal files,
3192 which provide cryptographical sealing of journal files so
3193 that attackers cannot alter log history anymore without this
3194 being detectable. Lennart will soon post a blog story about
3195 this explaining it in more detail.
3197 * There are two new service settings RestartPreventExitStatus=
3198 and SuccessExitStatus= which allow configuration of exit
3199 status (exit code or signal) which will be excepted from the
3200 restart logic, resp. consider successful.
3202 * journalctl gained the new --verify switch that can be used
3203 to check the integrity of the structure of journal files and
3204 (if Forward Secure Sealing is enabled) the contents of
3207 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3208 and similar symlinks pre-created. This makes running shells
3209 as container init process a lot more fun.
3211 * The fstab support can now handle PARTUUID= and PARTLABEL=
3214 * A new ConditionHost= condition has been added to match
3215 against the hostname (with globs) and machine ID. This is
3216 useful for clusters where a single OS image is used to
3217 provision a large number of hosts which shall run slightly
3218 different sets of services.
3220 * Services which hit the restart limit will now be placed in a
3223 Contributions from: Bertram Poettering, Dave Reisner, Huang
3224 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3225 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3229 * When running in --user mode systemd will now become a
3230 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3231 tree a lot more organized.
3233 * A new PartOf= unit dependency type has been introduced that
3234 may be used to group services in a natural way.
3236 * "systemctl enable" may now be used to enable instances of
3239 * journalctl now prints error log levels in red, and
3240 warning/notice log levels in bright white. It also supports
3241 filtering by log level now.
3243 * cgtop gained a new -n switch (similar to top), to configure
3244 the maximum number of iterations to run for. It also gained
3245 -b, to run in batch mode (accepting no input).
3247 * The suffix ".service" may now be omitted on most systemctl
3248 command lines involving service unit names.
3250 * There's a new bus call in logind to lock all sessions, as
3251 well as a loginctl verb for it "lock-sessions".
3253 * libsystemd-logind.so gained a new call sd_journal_perror()
3254 that works similar to libc perror() but logs to the journal
3255 and encodes structured information about the error number.
3257 * /etc/crypttab entries now understand the new keyfile-size=
3260 * shutdown(8) now can send a (configurable) wall message when
3261 a shutdown is cancelled.
3263 * The mount propagation mode for the root file system will now
3264 default to "shared", which is useful to make containers work
3265 nicely out-of-the-box so that they receive new mounts from
3266 the host. This can be undone locally by running "mount
3267 --make-rprivate /" if needed.
3269 * The prefdm.service file has been removed. Distributions
3270 should maintain this unit downstream if they intend to keep
3271 it around. However, we recommend writing normal unit files
3272 for display managers instead.
3274 * Since systemd is a crucial part of the OS we will now
3275 default to a number of compiler switches that improve
3276 security (hardening) such as read-only relocations, stack
3277 protection, and suchlike.
3279 * The TimeoutSec= setting for services is now split into
3280 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3281 of individual time outs for the start and the stop phase of
3284 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3285 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3286 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3287 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3288 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3289 Gundersen, Zbigniew Jędrzejewski-Szmek
3293 * The journal and id128 C APIs are now fully documented as man
3296 * Extra safety checks have been added when transitioning from
3297 the initial RAM disk to the main system to avoid accidental
3300 * /etc/crypttab entries now understand the new keyfile-offset=
3303 * systemctl -t can now be used to filter by unit load state.
3305 * The journal C API gained the new sd_journal_wait() call to
3306 make writing synchronous journal clients easier.
3308 * journalctl gained the new -D switch to show journals from a
3311 * journalctl now displays a special marker between log
3312 messages of two different boots.
3314 * The journal is now explicitly flushed to /var via a service
3315 systemd-journal-flush.service, rather than implicitly simply
3316 by seeing /var/log/journal to be writable.
3318 * journalctl (and the journal C APIs) can now match for much
3319 more complex expressions, with alternatives and
3322 * When transitioning from the initial RAM disk to the main
3323 system we will now kill all processes in a killing spree to
3324 ensure no processes stay around by accident.
3326 * Three new specifiers may be used in unit files: %u, %h, %s
3327 resolve to the user name, user home directory resp. user
3328 shell. This is useful for running systemd user instances.
3330 * We now automatically rotate journal files if their data
3331 object hash table gets a fill level > 75%. We also size the
3332 hash table based on the configured maximum file size. This
3333 together should lower hash collisions drastically and thus
3334 speed things up a bit.
3336 * journalctl gained the new "--header" switch to introspect
3337 header data of journal files.
3339 * A new setting SystemCallFilters= has been added to services
3340 which may be used to apply blacklists or whitelists to
3341 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3343 * nspawn gained a new --link-journal= switch (and quicker: -j)
3344 to link the container journal with the host. This makes it
3345 very easy to centralize log viewing on the host for all
3346 guests while still keeping the journal files separated.
3348 * Many bugfixes and optimizations
3350 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3351 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3352 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3357 * Several tools now understand kernel command line arguments,
3358 which are only read when run in an initial RAM disk. They
3359 usually follow closely their normal counterparts, but are
3362 * There's a new tool to analyze the readahead files that are
3363 automatically generated at boot. Use:
3365 /usr/lib/systemd/systemd-readahead analyze /.readahead
3367 * We now provide an early debug shell on tty9 if this enabled. Use:
3369 systemctl enable debug-shell.service
3371 * All plymouth related units have been moved into the Plymouth
3372 package. Please make sure to upgrade your Plymouth version
3375 * systemd-tmpfiles now supports getting passed the basename of
3376 a configuration file only, in which case it will look for it
3377 in all appropriate directories automatically.
3379 * udevadm info now takes a /dev or /sys path as argument, and
3380 does the right thing. Example:
3382 udevadm info /dev/sda
3383 udevadm info /sys/class/block/sda
3385 * systemctl now prints a warning if a unit is stopped but a
3386 unit that might trigger it continues to run. Example: a
3387 service is stopped but the socket that activates it is left
3390 * "systemctl status" will now mention if the log output was
3391 shortened due to rotation since a service has been started.
3393 * The journal API now exposes functions to determine the
3394 "cutoff" times due to rotation.
3396 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3397 immediately flushing of runtime logs to /var if possible,
3398 resp. for triggering immediate rotation of the journal
3401 * It is now considered an error if a service is attempted to
3402 be stopped that is not loaded.
3404 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3406 * systemd-analyze now supports Python 3
3408 * tmpfiles now supports cleaning up directories via aging
3409 where the first level dirs are always kept around but
3410 directories beneath it automatically aged. This is enabled
3411 by prefixing the age field with '~'.
3413 * Seat objects now expose CanGraphical, CanTTY properties
3414 which is required to deal with very fast bootups where the
3415 display manager might be running before the graphics drivers
3416 completed initialization.
3418 * Seat objects now expose a State property.
3420 * We now include RPM macros for service enabling/disabling
3421 based on the preset logic. We recommend RPM based
3422 distributions to make use of these macros if possible. This
3423 makes it simpler to reuse RPM spec files across
3426 * We now make sure that the collected systemd unit name is
3427 always valid when services log to the journal via
3430 * There's a new man page kernel-command-line(7) detailing all
3431 command line options we understand.
3433 * The fstab generator may now be disabled at boot by passing
3434 fstab=0 on the kernel command line.
3436 * A new kernel command line option modules-load= is now understood
3437 to load a specific kernel module statically, early at boot.
3439 * Unit names specified on the systemctl command line are now
3440 automatically escaped as needed. Also, if file system or
3441 device paths are specified they are automatically turned
3442 into the appropriate mount or device unit names. Example:
3444 systemctl status /home
3445 systemctl status /dev/sda
3447 * The SysVConsole= configuration option has been removed from
3448 system.conf parsing.
3450 * The SysV search path is no longer exported on the D-Bus
3453 * The Names= option is been removed from unit file parsing.
3455 * There's a new man page bootup(7) detailing the boot process.
3457 * Every unit and every generator we ship with systemd now
3458 comes with full documentation. The self-explanatory boot is
3461 * A couple of services gained "systemd-" prefixes in their
3462 name if they wrap systemd code, rather than only external
3463 code. Among them fsck@.service which is now
3464 systemd-fsck@.service.
3466 * The HaveWatchdog property has been removed from the D-Bus
3469 * systemd.confirm_spawn= on the kernel command line should now
3472 * There's a new man page crypttab(5) which details all options
3473 we actually understand.
3475 * systemd-nspawn gained a new --capability= switch to pass
3476 additional capabilities to the container.
3478 * timedated will now read known NTP implementation unit names
3479 from /usr/lib/systemd/ntp-units.d/*.list,
3480 systemd-timedated-ntp.target has been removed.
3482 * journalctl gained a new switch "-b" that lists log data of
3483 the current boot only.
3485 * The notify socket is in the abstract namespace again, in
3486 order to support daemons which chroot() at start-up.
3488 * There is a new Storage= configuration option for journald
3489 which allows configuration of where log data should go. This
3490 also provides a way to disable journal logging entirely, so
3491 that data collected is only forwarded to the console, the
3492 kernel log buffer or another syslog implementation.
3494 * Many bugfixes and optimizations
3496 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3497 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3498 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3499 Shawn Landden, Tom Gundersen
3503 * "systemctl help <unit>" now shows the man page if one is
3506 * Several new man pages have been added.
3508 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3509 MaxLevelConsole= can now be specified in
3510 journald.conf. These options allow reducing the amount of
3511 data stored on disk or forwarded by the log level.
3513 * TimerSlackNSec= can now be specified in system.conf for
3514 PID1. This allows system-wide power savings.
3516 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3517 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3522 * logind is now capable of (optionally) handling power and
3523 sleep keys as well as the lid switch.
3525 * journalctl now understands the syntax "journalctl
3526 /usr/bin/avahi-daemon" to get all log output of a specific
3529 * CapabilityBoundingSet= in system.conf now also influences
3530 the capability bound set of usermode helpers of the kernel.
3532 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3533 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3534 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3535 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3539 * Note that we skipped 139 releases here in order to set the
3540 new version to something that is greater than both udev's
3541 and systemd's most recent version number.
3543 * udev: all udev sources are merged into the systemd source tree now.
3544 All future udev development will happen in the systemd tree. It
3545 is still fully supported to use the udev daemon and tools without
3546 systemd running, like in initramfs or other init systems. Building
3547 udev though, will require the *build* of the systemd tree, but
3548 udev can be properly *run* without systemd.
3550 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3551 should be used to create dead device nodes as workarounds for broken
3554 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3555 no longer supported. udev_monitor_new_from_netlink() needs to be
3556 used to subscribe to events.
3558 * udev: when udevd is started by systemd, processes which are left
3559 behind by forking them off of udev rules, are unconditionally cleaned
3560 up and killed now after the event handling has finished. Services or
3561 daemons must be started as systemd services. Services can be
3562 pulled-in by udev to get started, but they can no longer be directly
3563 forked by udev rules.
3565 * udev: the daemon binary is called systemd-udevd now and installed
3566 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3567 to adapt to that, create symlink, or rename the binary after building
3570 * libudev no longer provides these symbols:
3571 udev_monitor_from_socket()
3572 udev_queue_get_failed_list_entry()
3573 udev_get_{dev,sys,run}_path()
3574 The versions number was bumped and symbol versioning introduced.
3576 * systemd-loginctl and systemd-journalctl have been renamed
3577 to loginctl and journalctl to match systemctl.
3579 * The config files: /etc/systemd/systemd-logind.conf and
3580 /etc/systemd/systemd-journald.conf have been renamed to
3581 logind.conf and journald.conf. Package updates should rename
3582 the files to the new names on upgrade.
3584 * For almost all files the license is now LGPL2.1+, changed
3585 from the previous GPL2.0+. Exceptions are some minor stuff
3586 of udev (which will be changed to LGPL2.1 eventually, too),
3587 and the MIT licensed sd-daemon.[ch] library that is suitable
3588 to be used as drop-in files.
3590 * systemd and logind now handle system sleep states, in
3591 particular suspending and hibernating.
3593 * logind now implements a sleep/shutdown/idle inhibiting logic
3594 suitable for a variety of uses. Soonishly Lennart will blog
3595 about this in more detail.
3597 * var-run.mount and var-lock.mount are no longer provided
3598 (which prevously bind mounted these directories to their new
3599 places). Distributions which have not converted these
3600 directories to symlinks should consider stealing these files
3601 from git history and add them downstream.
3603 * We introduced the Documentation= field for units and added
3604 this to all our shipped units. This is useful to make it
3605 easier to explore the boot and the purpose of the various
3608 * All smaller setup units (such as
3609 systemd-vconsole-setup.service) now detect properly if they
3610 are run in a container and are skipped when
3611 appropriate. This guarantees an entirely noise-free boot in
3612 Linux container environments such as systemd-nspawn.
3614 * A framework for implementing offline system updates is now
3615 integrated, for details see:
3616 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3618 * A new service type Type=idle is available now which helps us
3619 avoiding ugly interleaving of getty output and boot status
3622 * There's now a system-wide CapabilityBoundingSet= option to
3623 globally reduce the set of capabilities for the
3624 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3625 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3626 even CAP_NET_ADMIN system-wide for secure systems.
3628 * There are now system-wide DefaultLimitXXX= options to
3629 globally change the defaults of the various resource limits
3630 for all units started by PID 1.
3632 * Harald Hoyer's systemd test suite has been integrated into
3633 systemd which allows easy testing of systemd builds in qemu
3634 and nspawn. (This is really awesome! Ask us for details!)
3636 * The fstab parser is now implemented as generator, not inside
3639 * systemctl will now warn you if .mount units generated from
3640 /etc/fstab are out of date due to changes in fstab that
3641 have not been read by systemd yet.
3643 * systemd is now suitable for usage in initrds. Dracut has
3644 already been updated to make use of this. With this in place
3645 initrds get a slight bit faster but primarily are much
3646 easier to introspect and debug since "systemctl status" in
3647 the host system can be used to introspect initrd services,
3648 and the journal from the initrd is kept around too.
3650 * systemd-delta has been added, a tool to explore differences
3651 between user/admin configuration and vendor defaults.
3653 * PrivateTmp= now affects both /tmp and /var/tmp.
3655 * Boot time status messages are now much prettier and feature
3656 proper english language. Booting up systemd has never been
3659 * Read-ahead pack files now include the inode number of all
3660 files to pre-cache. When the inode changes the pre-caching
3661 is not attempted. This should be nicer to deal with updated
3662 packages which might result in changes of read-ahead
3665 * We now temporaritly lower the kernel's read_ahead_kb variable
3666 when collecting read-ahead data to ensure the kernel's
3667 built-in read-ahead does not add noise to our measurements
3668 of necessary blocks to pre-cache.
3670 * There's now RequiresMountsFor= to add automatic dependencies
3671 for all mounts necessary for a specific file system path.
3673 * MountAuto= and SwapAuto= have been removed from
3674 system.conf. Mounting file systems at boot has to take place
3677 * nspawn now learned a new switch --uuid= to set the machine
3678 ID on the command line.
3680 * nspawn now learned the -b switch to automatically search
3683 * vt102 is now the default TERM for serial TTYs, upgraded from
3686 * systemd-logind now works on VT-less systems.
3688 * The build tree has been reorganized. The individual
3689 components now have directories of their own.
3691 * A new condition type ConditionPathIsReadWrite= is now available.
3693 * nspawn learned the new -C switch to create cgroups for the
3694 container in other hierarchies.
3696 * We now have support for hardware watchdogs, configurable in
3699 * The scheduled shutdown logic now has a public API.
3701 * We now mount /tmp as tmpfs by default, but this can be
3702 masked and /etc/fstab can override it.
3704 * Since udisks does not make use of /media anymore we are not
3705 mounting a tmpfs on it anymore.
3707 * journalctl gained a new --local switch to only interleave
3708 locally generated journal files.
3710 * We can now load the IMA policy at boot automatically.
3712 * The GTK tools have been split off into a systemd-ui.
3714 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3715 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3716 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3717 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3718 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3719 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3720 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3721 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3722 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3727 * This is mostly a bugfix release
3729 * Support optional initialization of the machine ID from the
3730 KVM or container configured UUID.
3732 * Support immediate reboots with "systemctl reboot -ff"
3734 * Show /etc/os-release data in systemd-analyze output
3736 * Many bugfixes for the journal, including endianness fixes and
3737 ensuring that disk space enforcement works
3739 * sd-login.h is C++ comptaible again
3741 * Extend the /etc/os-release format on request of the Debian
3744 * We now refuse non-UTF8 strings used in various configuration
3745 and unit files. This is done to ensure we do not pass invalid
3746 data over D-Bus or expose it elsewhere.
3748 * Register Mimo USB Screens as suitable for automatic seat
3751 * Read SELinux client context from journal clients in a race
3754 * Reorder configuration file lookup order. /etc now always
3755 overrides /run in order to allow the administrator to always
3756 and unconditionally override vendor supplied or
3757 automatically generated data.
3759 * The various user visible bits of the journal now have man
3760 pages. We still lack man pages for the journal API calls
3763 * We now ship all man pages in HTML format again in the
3766 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3767 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3768 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3773 * This is mostly a bugfix release
3775 * systems lacking /etc/os-release are no longer supported.
3777 * Various functionality updates to libsystemd-login.so
3779 * Track class of PAM logins to distuingish greeters from
3782 Contributions from: Kay Sievers, Lennart Poettering, Michael
3787 * This is an important bugfix release for v41.
3789 * Building man pages is now optional which should be useful
3790 for those building systemd from git but unwilling to install
3793 * Watchdog support for supervising services is now usable. In
3794 a future release support for hardware watchdogs
3795 (i.e. /dev/watchdog) will be added building on this.
3797 * Service start rate limiting is now configurable and can be
3798 turned off per service. When a start rate limit is hit a
3799 reboot can automatically be triggered.
3801 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3803 Contributions from: Benjamin Franzke, Bill Nottingham,
3804 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3805 Schmidt, Michał Górny, Piotr Drąg
3809 * The systemd binary is installed /usr/lib/systemd/systemd now;
3810 An existing /sbin/init symlink needs to be adapted with the
3813 * The code that loads kernel modules has been ported to invoke
3814 libkmod directly, instead of modprobe. This means we do not
3815 support systems with module-init-tools anymore.
3817 * Watchdog support is now already useful, but still not
3820 * A new kernel command line option systemd.setenv= is
3821 understood to set system wide environment variables
3822 dynamically at boot.
3824 * We now limit the set of capabilities of systemd-journald.
3826 * We now set SIGPIPE to ignore by default, since it only is
3827 useful in shell pipelines, and has little use in general
3828 code. This can be disabled with IgnoreSIPIPE=no in unit
3831 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3832 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3837 * This is mostly a bugfix release
3839 * We now expose the reason why a service failed in the
3840 "Result" D-Bus property.
3842 * Rudimentary service watchdog support (will be completed over
3843 the next few releases.)
3845 * When systemd forks off in order execute some service we will
3846 now immediately changes its argv[0] to reflect which process
3847 it will execute. This is useful to minimize the time window
3848 with a generic argv[0], which makes bootcharts more useful
3850 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3851 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3852 Mike Kazantsev, Ray Strode
3856 * This is mostly a test release, but incorporates many
3859 * New systemd-cgtop tool to show control groups by their
3862 * Linking against libacl for ACLs is optional again. If
3863 disabled, support tracking device access for active logins
3864 goes becomes unavailable, and so does access to the user
3865 journals by the respective users.
3867 * If a group "adm" exists, journal files are automatically
3868 owned by them, thus allow members of this group full access
3869 to the system journal as well as all user journals.
3871 * The journal now stores the SELinux context of the logging
3872 client for all entries.
3874 * Add C++ inclusion guards to all public headers
3876 * New output mode "cat" in the journal to print only text
3877 messages, without any meta data like date or time.
3879 * Include tiny X server wrapper as a temporary stop-gap to
3880 teach XOrg udev display enumeration. This is used by display
3881 managers such as gdm, and will go away as soon as XOrg
3882 learned native udev hotplugging for display devices.
3884 * Add new systemd-cat tool for executing arbitrary programs
3885 with STDERR/STDOUT connected to the journal. Can also act as
3886 BSD logger replacement, and does so by default.
3888 * Optionally store all locally generated coredumps in the
3889 journal along with meta data.
3891 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3892 writing short strings to files (for usage for /sys), and for
3893 creating symlinks, character and block device nodes.
3895 * New unit file option ControlGroupPersistent= to make cgroups
3896 persistent, following the mechanisms outlined in
3897 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3899 * Support multiple local RTCs in a sane way
3901 * No longer monopolize IO when replaying readahead data on
3902 rotating disks, since we might starve non-file-system IO to
3903 death, since fanotify() will not see accesses done by blkid,
3906 * Do not show kernel threads in systemd-cgls anymore, unless
3907 requested with new -k switch.
3909 Contributions from: Dan Horák, Kay Sievers, Lennart
3910 Poettering, Michal Schmidt
3914 * This is mostly a test release, but incorporates many
3917 * The git repository moved to:
3918 git://anongit.freedesktop.org/systemd/systemd
3919 ssh://git.freedesktop.org/git/systemd/systemd
3921 * First release with the journal
3922 http://0pointer.de/blog/projects/the-journal.html
3924 * The journal replaces both systemd-kmsg-syslogd and
3925 systemd-stdout-bridge.
3927 * New sd_pid_get_unit() API call in libsystemd-logind
3929 * Many systemadm clean-ups
3931 * Introduce remote-fs-pre.target which is ordered before all
3932 remote mounts and may be used to start services before all
3935 * Added Mageia support
3937 * Add bash completion for systemd-loginctl
3939 * Actively monitor PID file creation for daemons which exit in
3940 the parent process before having finished writing the PID
3941 file in the daemon process. Daemons which do this need to be
3942 fixed (i.e. PID file creation must have finished before the
3943 parent exits), but we now react a bit more gracefully to them.
3945 * Add colourful boot output, mimicking the well-known output
3946 of existing distributions.
3948 * New option PassCredentials= for socket units, for
3949 compatibility with a recent kernel ABI breakage.
3951 * /etc/rc.local is now hooked in via a generator binary, and
3952 thus will no longer act as synchronization point during
3955 * systemctl list-unit-files now supports --root=.
3957 * systemd-tmpfiles now understands two new commands: z, Z for
3958 relabelling files according to the SELinux database. This is
3959 useful to apply SELinux labels to specific files in /sys,
3962 * Output of SysV services is now forwarded to both the console
3963 and the journal by default, not only just the console.
3965 * New man pages for all APIs from libsystemd-login.
3967 * The build tree got reorganized and a the build system is a
3968 lot more modular allowing embedded setups to specifically
3969 select the components of systemd they are interested in.
3971 * Support for Linux systems lacking the kernel VT subsystem is
3974 * configure's --with-rootdir= got renamed to
3975 --with-rootprefix= to follow the naming used by udev and
3978 * Unless specified otherwise we will now install to /usr instead
3979 of /usr/local by default.
3981 * Processes with '@' in argv[0][0] are now excluded from the
3982 final shut-down killing spree, following the logic explained
3984 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
3986 * All processes remaining in a service cgroup when we enter
3987 the START or START_PRE states are now killed with
3988 SIGKILL. That means it is no longer possible to spawn
3989 background processes from ExecStart= lines (which was never
3990 supported anyway, and bad style).
3992 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
3993 reloading of units together.
3995 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
3996 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
3997 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
3998 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
3999 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek