1 systemd System and Service Manager
5 * When querying unit file enablement status (for example via
6 "systemctl is-enabled"), a new state "indirect" is now known
7 which indicates that a unit might not be enabled itself, but
8 another unit listed in its Alias= setting might be.
10 * Similar to the various existing ConditionXYZ= settings for
11 units there are now matching AssertXYZ= settings. While
12 failing conditions cause a unit to be skipped, but its job
13 to succeed, failing assertions declared like this will cause
14 a unit start operation and its job to fail.
16 * hostnamed now knows a new chassis type "embedded".
18 * systemctl gained a new "edit" command. When used on a unit
19 file this allows extending unit files with .d/ drop-in
20 configuration snippets or editing the full file (after
21 copying it from /usr/lib to /etc). This will invoke the
22 user's editor (as configured with $EDITOR), and reload the
23 modified configuration after editing.
25 * "systemctl status" now shows the suggested enablement state
26 for a unit, as declared in the (usually vendor-supplied)
29 * nss-myhostname will now resolve the single-label host name
30 "gateway" to the locally configured default IP routing
31 gateways, ordered by their metrics. This assigns a stable
32 name to the used gateways, regardless which ones are
33 currently configured. Note that the name will only be
34 resolved after all other name sources (if nss-myhostname is
35 configured properly) and should hence not negatively impact
36 systems that use the single-label host name "gateway" in
39 * systemd-inhibit now allows filtering by mode when listing
42 * Units with resource management gained a new "Delegate"
43 boolean property, which when set allows processes running
44 inside the unit to further partition resources. This is
45 primarily useful for systemd user instances as well as
48 * journald will now pick up audit messages directly from
49 the kernel, and log them like any other log message. The
50 audit fields are split up and fully indexed. This means that
51 journalctl in many ways is now a (nicer!) alternative to
52 ausearch, the traditional audit client. Note that this
53 implements only a minimal audit client, if you want the
54 special audit modes like reboot-on-log-overflow, please use
55 the traditional auditd instead, which can be used in
58 * The ConditionSecurity= unit file option now understands the
59 special string "audit" to check whether auditing is
62 * journalctl gained two new commands --vacuum-size= and
63 --vacuum-time= to delete old journal files until the
64 remaining ones take up no more the specified size on disk,
65 or are not older than the specified time.
67 * A new, native PPPoE library has been added to sd-network,
68 systemd's library of light-weight networking protocols. This
69 library will be used in a future version of networkd to
70 enable PPPoE communication without an external pppd daemon.
72 * The busctl tool now understands a new "capture" verb that
73 works similar to "monitor", but writes a packet capture
74 trace to STDOUT that can be redirected to a file which is
75 compatible with libcap's capture file format. This can then
76 be loaded in Wireshark and similar tools to inspect bus
79 * The busctl tool now understands a new "tree" verb that shows
80 the object trees of a specific service on the bus, or of all
83 * The busctl tool now understands a new "introspect" verb that
84 shows all interfaces and members of objects on the bus,
85 including their signature and values. This is particularly
86 useful to get more information about bus objects shown by
87 the new "busctl tree" command.
89 * The busctl tool now understands new verbs "call",
90 "set-property" and "get-property" for invoking bus method
91 calls, setting and getting bus object properties in a
94 * busctl gained a new --augment-creds= argument that controls
95 whether the tool shall augment credential information it
96 gets from the bus with data from /proc, in a possibly
99 * nspawn's --link-journal= switch gained two new values
100 "try-guest" and "try-host" that work like "guest" and
101 "host", but do not fail if the host has no persistant
102 journalling enabled. -j is now equivalent to
103 --link-journal=try-guest.
105 * macvlan network devices created by nspawn will now have
106 stable MAC addresses.
108 * A new SmackProcessLabel= unit setting has been added, which
109 controls the SMACK security label processes forked off by
110 the respective unit shall use.
112 * If compiled with --enable-xkbcommon, systemd-localed will
113 verify x11 keymap settings by compiling the given keymap. It
114 will spew out warnings if the compilation fails. This
115 requires libxkbcommon to be installed.
117 * When a coredump is collected a larger number of metadata
118 fields is now collected and included in the journal records
119 created for it. More specifically control group membership,
120 environment variables, memory maps, working directory,
121 chroot directory, /proc/$PID/status, and a list of open file
122 descriptors is now stored in the log entry.
124 * The udev hwdb now containes DPI information for mice. For
127 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
129 * All systemd programs that read standalone configuration
130 files in /etc now also support a corresponding series of
131 .conf.d configuration directories in /etc/, /run/,
132 /usr/local/lib/, /usr/lib/, and (if configured with
133 --enable-split-usr) /lib/. In particular, the following
134 configuration files now have corresponding configuration
135 directories: system.conf user.conf, logind.conf,
136 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
137 resolved.conf, timesyncd.conf, journal-remote.conf, and
138 journal-upload.conf. Note that distributions should use the
139 configuration directories in /usr/lib/; the directories in
140 /etc/ are reserved for the system administrator.
142 * systemd-rfkill will no longer take the rfkill device name
143 into account when storing rfkill state on disk, as the name
144 might be dynamically assigned and not stable. Instead, the
145 ID_PATH udev variable combined with the rfkill type (wlan,
146 bluetooth, ...) is used.
148 * A new service systemd-machine-id-commit.service has been
149 added. When used on systems where /etc is read-only during
150 boot, and /etc/machine-id is not initialized (but an empty
151 file), this service will copy the temporary machine ID
152 created as replacement into /etc after the system is fully
153 booted up. This is useful for systems that are freshly
154 installed with a non-initialized machine ID, but should get
155 a fixed machine ID for subsequent boots.
157 * networkd's .netdev files now provide a large set of
158 configuration parameters for VXLAN devices. Similar, the
159 bridge port cost parameter is now configurable in .network
160 files. There's also new support for configuring IP source
161 routing. networkd .link files gained support for a new
162 OriginalName= match that is useful to match against the
163 original interface name the kernel assigned. .network files
164 may include MTU= and MACAddress= fields for altering the MTU
165 and MAC address while being connected to a specific network
168 * The LUKS logic gained supported for configuring
169 UUID-specific key files. There's also new support for naming
170 LUKS device from the kernel command line, using the new
173 * Timer units may now be transiently created via the bus API
174 (this was previously already available for scope and service
175 units). In addition it is now possible to create multiple
176 transient units at the same time with a single bus call. The
177 "systemd-run" tool has been updated to make use of this for
178 running commands on a specified time, in at(1)-style.
180 * tmpfiles gained support for "t" lines, for assigning
181 extended attributes to files. Among other uses this may be
182 used to assign SMACK labels to files.
187 * journalctl gained the new options -t/--identifier= to match
188 on the syslog identifier (aka "tag"), as well as --utc to
189 show log timestamps in the UTC timezone. journalctl now also
190 accepts -n/--lines=all to disable line capping in a pager.
192 * journalctl gained a new switch, --flush, that synchronously
193 flushes logs from /run/log/journal to /var/log/journal if
194 persistent storage is enabled. systemd-journal-flush.service
195 now waits until the operation is complete.
197 * Services can notify the manager before they start a reload
198 (by sending RELOADING=1) or shutdown (by sending
199 STOPPING=1). This allows the manager to track and show the
200 internal state of daemons and closes a race condition when
201 the process is still running but has closed its D-Bus
204 * Services with Type=oneshot do not have to have any ExecStart
207 * User units are now loaded also from
208 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
209 /run/systemd/user directory that was already previously
210 supported, but is under the control of the user.
212 * Job timeouts (i.e. time-outs on the time a job that is
213 queued stays in the run queue) can now optionally result in
214 immediate reboot or power-off actions (JobTimeoutAction= and
215 JobTimeoutRebootArgument=). This is useful on ".target"
216 units, to limit the maximum time a target remains
217 undispatched in the run queue, and to trigger an emergency
218 operation in such a case. This is now used by default to
219 turn off the system if boot-up (as defined by everything in
220 basic.target) hangs and does not complete for at least
221 15min. Also, if power-off or reboot hang for at least 30min
222 an immediate power-off/reboot operation is triggered. This
223 functionality is particularly useful to increase reliability
224 on embedded devices, but also on laptops which might
225 accidentally get powered on when carried in a backpack and
226 whose boot stays stuck in a hard disk encryption passphrase
229 * systemd-logind can be configured to also handle lid switch
230 events even when the machine is docked or multiple displays
231 are attached (HandleLidSwitchDocked= option).
233 * A helper binary and a service have been added which can be
234 used to resume from hibernation in the initramfs. A
235 generator will parse the resume= option on the kernel
236 command line to trigger resume.
238 * A user console daemon systemd-consoled has been
239 added. Currently, it is a preview, and will so far open a
240 single terminal on each session of the user marked as
241 Desktop=systemd-console.
243 * Route metrics can be specified for DHCP routes added by
246 * The SELinux context of socket-activated services can be set
247 from the information provided by the networking stack
248 (SELinuxContextFromNet= option).
250 * Userspace firmware loading support has been removed and
251 the minimum supported kernel version is thus bumped to 3.7.
253 * Timeout for udev workers has been increased from 1 to 3
254 minutes, but a warning will be printed after 1 minute to
255 help diagnose kernel modules that take a long time to load.
257 * Udev rules can now remove tags on devices with TAG-="foobar".
259 * systemd's readahead implementation has been removed. In many
260 circumstances it didn't give expected benefits even for
261 rotational disk drives and was becoming less relevant in the
262 age of SSDs. As none of the developers has been using
263 rotating media anymore, and nobody stepped up to actively
264 maintain this component of systemd it has now been removed.
266 * Swap units can use Options= to specify discard options.
267 Discard options specified for swaps in /etc/fstab are now
270 * Docker containers are now detected as a separate type of
273 * The Password Agent protocol gained support for queries where
274 the user input is shown, useful e.g. for user names.
275 systemd-ask-password gained a new --echo option to turn that
278 * The default sysctl.d/ snippets will now set:
280 net.core.default_qdisc = fq_codel
282 This selects Fair Queuing Controlled Delay as the default
283 queuing discipline for network interfaces. fq_codel helps
284 fight the network bufferbloat problem. It is believed to be
285 a good default with no tuning required for most workloads.
286 Downstream distributions may override this choice. On 10Gbit
287 servers that do not do forwarding, "fq" may perform better.
288 Systems without a good clocksource should use "pfifo_fast".
290 * If kdbus is enabled during build a new option BusPolicy= is
291 available for service units, that allows locking all service
292 processes into a stricter bus policy, in order to limit
293 access to various bus services, or even hide most of them
294 from the service's view entirely.
296 * networkctl will now show the .network and .link file
297 networkd has applied to a specific interface.
299 * sd-login gained a new API call sd_session_get_desktop() to
300 query which desktop environment has been selected for a
303 * UNIX utmp support is now compile-time optional to support
306 * systemctl gained two new commands "add-wants" and
307 "add-requires" for pulling in units from specific targets
310 * If the word "rescue" is specified on the kernel command line
311 the system will now boot into rescue mode (aka
312 rescue.target), which was previously available only by
313 specifying "1" or "systemd.unit=rescue.target" on the kernel
314 command line. This new kernel command line option nicely
315 mirrors the already existing "emergency" kernel command line
318 * New kernel command line options mount.usr=, mount.usrflags=,
319 mount.usrfstype= have been added that match root=, rootflags=,
320 rootfstype= but allow mounting a specific file system to
323 * The $NOTIFY_SOCKET is now also passed to control processes of
324 services, not only the main process.
326 * This version reenables support for fsck's -l switch. This
327 means at least version v2.25 of util-linux is required for
328 operation, otherwise dead-locks on device nodes may
329 occur. Again: you need to update util-linux to at least
330 v2.25 when updating systemd to v217.
332 * The "multi-seat-x" tool has been removed from systemd, as
333 its functionality has been integrated into X servers 1.16,
334 and the tool is hence redundant. It is recommended to update
335 display managers invoking this tool to simply invoke X
336 directly from now on, again.
338 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
339 message flag has been added for all of systemd's PolicyKit
340 authenticated method calls has been added. In particular
341 this now allows optional interactive authorization via
342 PolicyKit for many of PID1's privileged operations such as
343 unit file enabling and disabling.
345 * "udevadm hwdb --update" learnt a new switch "--usr" for
346 placing the rebuilt hardware database in /usr instead of
347 /etc. When used only hardware database entries stored in
348 /usr will be used, and any user database entries in /etc are
349 ignored. This functionality is useful for vendors to ship a
350 pre-built database on systems where local configuration is
351 unnecessary or unlikely.
353 * Calendar time specifications in .timer units now also
354 understand the strings "semi-annually", "quarterly" and
355 "minutely" as shortcuts (in addition to the preexisting
356 "anually", "hourly", ...).
358 * systemd-tmpfiles will now correctly create files in /dev
359 at boot which are marked for creation only at boot. It is
360 recommended to always create static device nodes with 'c!'
361 and 'b!', so that they are created only at boot and not
362 overwritten at runtime.
364 * When the watchdog logic is used for a service (WatchdogSec=)
365 and the watchdog timeout is hit the service will now be
366 terminated with SIGABRT (instead of just SIGTERM), in order
367 to make sure a proper coredump and backtrace is
368 generated. This ensures that hanging services will result in
369 similar coredump/backtrace behaviour as services that hit a
372 Contributions from: Andreas Henriksson, Andrei Borzenkov,
373 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
374 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
375 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
376 Herrmann, David Sommerseth, David Strauss, Emil Renner
377 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
378 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
379 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
380 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
381 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
382 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
383 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
384 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
385 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
386 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
387 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
388 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
389 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
390 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
391 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
392 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
395 -- Berlin, 2014-10-28
399 * timedated no longer reads NTP implementation unit names from
400 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
401 implementations should add a
403 Conflicts=systemd-timesyncd.service
405 to their unit files to take over and replace systemd's NTP
406 default functionality.
408 * systemd-sysusers gained a new line type "r" for configuring
409 which UID/GID ranges to allocate system users/groups
410 from. Lines of type "u" may now add an additional column
411 that specifies the home directory for the system user to be
412 created. Also, systemd-sysusers may now optionally read user
413 information from STDIN instead of a file. This is useful for
414 invoking it from RPM preinst scriptlets that need to create
415 users before the first RPM file is installed since these
416 files might need to be owned by them. A new
417 %sysusers_create_inline RPM macro has been introduced to do
418 just that. systemd-sysusers now updates the shadow files as
419 well as the user/group databases, which should enhance
420 compatibility with certain tools like grpck.
422 * A number of bus APIs of PID 1 now optionally consult
423 PolicyKit to permit access for otherwise unprivileged
424 clients under certain conditions. Note that this currently
425 doesn't support interactive authentication yet, but this is
426 expected to be added eventually, too.
428 * /etc/machine-info now has new fields for configuring the
429 deployment environment of the machine, as well as the
430 location of the machine. hostnamectl has been updated with
431 new command to update these fields.
433 * systemd-timesyncd has been updated to automatically acquire
434 NTP server information from systemd-networkd, which might
435 have been discovered via DHCP.
437 * systemd-resolved now includes a caching DNS stub resolver
438 and a complete LLMNR name resolution implementation. A new
439 NSS module "nss-resolve" has been added which can be used
440 instead of glibc's own "nss-dns" to resolve hostnames via
441 systemd-resolved. Hostnames, addresses and arbitrary RRs may
442 be resolved via systemd-resolved D-Bus APIs. In contrast to
443 the glibc internal resolver systemd-resolved is aware of
444 multi-homed system, and keeps DNS server and caches separate
445 and per-interface. Queries are sent simultaneously on all
446 interfaces that have DNS servers configured, in order to
447 properly handle VPNs and local LANs which might resolve
448 separate sets of domain names. systemd-resolved may acquire
449 DNS server information from systemd-networkd automatically,
450 which in turn might have discovered them via DHCP. A tool
451 "systemd-resolve-host" has been added that may be used to
452 query the DNS logic in resolved. systemd-resolved implements
453 IDNA and automatically uses IDNA or UTF-8 encoding depending
454 on whether classic DNS or LLMNR is used as transport. In the
455 next releases we intend to add a DNSSEC and mDNS/DNS-SD
456 implementation to systemd-resolved.
458 * A new NSS module nss-mymachines has been added, that
459 automatically resolves the names of all local registered
460 containers to their respective IP addresses.
462 * A new client tool "networkctl" for systemd-networkd has been
463 added. It currently is entirely passive and will query
464 networking configuration from udev, rtnetlink and networkd,
465 and present it to the user in a very friendly
466 way. Eventually, we hope to extend it to become a full
467 control utility for networkd.
469 * .socket units gained a new DeferAcceptSec= setting that
470 controls the kernels' TCP_DEFER_ACCEPT sockopt for
471 TCP. Similar, support for controlling TCP keep-alive
472 settings has been added (KeepAliveTimeSec=,
473 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
474 turning off Nagle's algorithm on TCP has been added
477 * logind learned a new session type "web", for use in projects
478 like Cockpit which register web clients as PAM sessions.
480 * timer units with at least one OnCalendar= setting will now
481 be started only after timer-sync.target has been
482 reached. This way they will not elapse before the system
483 clock has been corrected by a local NTP client or
484 similar. This is particular useful on RTC-less embedded
485 machines, that come up with an invalid system clock.
487 * systemd-nspawn's --network-veth= switch should now result in
488 stable MAC addresses for both the outer and the inner side
491 * systemd-nspawn gained a new --volatile= switch for running
492 container instances with /etc or /var unpopulated.
494 * The kdbus client code has been updated to use the new Linux
495 3.17 memfd subsystem instead of the old kdbus-specific one.
497 * systemd-networkd's DHCP client and server now support
498 FORCERENEW. There are also new configuration options to
499 configure the vendor client identifier and broadcast mode
502 * systemd will no longer inform the kernel about the current
503 timezone, as this is necessarily incorrect and racy as the
504 kernel has no understanding of DST and similar
505 concepts. This hence means FAT timestamps will be always
506 considered UTC, similar to what Android is already
507 doing. Also, when the RTC is configured to the local time
508 (rather than UTC) systemd will never synchronize back to it,
509 as this might confuse Windows at a later boot.
511 * systemd-analyze gained a new command "verify" for offline
512 validation of unit files.
514 * systemd-networkd gained support for a couple of additional
515 settings for bonding networking setups. Also, the metric for
516 statically configured routes may now be configured. For
517 network interfaces where this is appropriate the peer IP
518 address may now be configured.
520 * systemd-networkd's DHCP client will no longer request
521 broadcasting by default, as this tripped up some networks.
522 For hardware where broadcast is required the feature should
523 be switched back on using RequestBroadcast=yes.
525 * systemd-networkd will now set up IPv4LL addresses (when
526 enabled) even if DHCP is configured successfully.
528 * udev will now default to respect network device names given
529 by the kernel when the kernel indicates that these are
530 predictable. This behavior can be tweaked by changing
531 NamePolicy= in the relevant .link file.
533 * A new library systemd-terminal has been added that
534 implements full TTY stream parsing and rendering. This
535 library is supposed to be used later on for implementing a
536 full userspace VT subsystem, replacing the current kernel
539 * A new tool systemd-journal-upload has been added to push
540 journal data to a remote system running
541 systemd-journal-remote.
543 * journald will no longer forward all local data to another
544 running syslog daemon. This change has been made because
545 rsyslog (which appears to be the most commonly used syslog
546 implementation these days) no longer makes use of this, and
547 instead pulls the data out of the journal on its own. Since
548 forwarding the messages to a non-existent syslog server is
549 more expensive than we assumed we have now turned this
550 off. If you run a syslog server that is not a recent rsyslog
551 version, you have to turn this option on again
552 (ForwardToSyslog= in journald.conf).
554 * journald now optionally supports the LZ4 compressor for
555 larger journal fields. This compressor should perform much
556 better than XZ which was the previous default.
558 * machinectl now shows the IP addresses of local containers,
559 if it knows them, plus the interface name of the container.
561 * A new tool "systemd-escape" has been added that makes it
562 easy to escape strings to build unit names and similar.
564 * sd_notify() messages may now include a new ERRNO= field
565 which is parsed and collected by systemd and shown among the
566 "systemctl status" output for a service.
568 * A new component "systemd-firstboot" has been added that
569 queries the most basic systemd information (timezone,
570 hostname, root password) interactively on first
571 boot. Alternatively it may also be used to provision these
572 things offline on OS images installed into directories.
574 * The default sysctl.d/ snippets will now set
576 net.ipv4.conf.default.promote_secondaries=1
578 This has the benefit of no flushing secondary IP addresses
579 when primary addresses are removed.
581 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
582 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
583 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
584 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
585 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
586 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
587 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
588 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
589 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
590 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
591 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
592 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
593 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
594 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
595 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
597 -- Berlin, 2014-08-19
601 * A new tool systemd-sysusers has been added. This tool
602 creates system users and groups in /etc/passwd and
603 /etc/group, based on static declarative system user/group
604 definitions in /usr/lib/sysusers.d/. This is useful to
605 enable factory resets and volatile systems that boot up with
606 an empty /etc directory, and thus need system users and
607 groups created during early boot. systemd now also ships
608 with two default sysusers.d/ files for the most basic
609 users and groups systemd and the core operating system
612 * A new tmpfiles snippet has been added that rebuilds the
613 essential files in /etc on boot, should they be missing.
615 * A directive for ensuring automatic clean-up of
616 /var/cache/man/ has been removed from the default
617 configuration. This line should now be shipped by the man
618 implementation. The necessary change has been made to the
619 man-db implementation. Note that you need to update your man
620 implementation to one that ships this line, otherwise no
621 automatic clean-up of /var/cache/man will take place.
623 * A new condition ConditionNeedsUpdate= has been added that
624 may conditionalize services to only run when /etc or /var
625 are "older" than the vendor operating system resources in
626 /usr. This is useful for reconstructing or updating /etc
627 after an offline update of /usr or a factory reset, on the
628 next reboot. Services that want to run once after such an
629 update or reset should use this condition and order
630 themselves before the new systemd-update-done.service, which
631 will mark the two directories as fully updated. A number of
632 service files have been added making use of this, to rebuild
633 the udev hardware database, the journald message catalog and
634 dynamic loader cache (ldconfig). The systemd-sysusers tool
635 described above also makes use of this now. With this in
636 place it is now possible to start up a minimal operating
637 system with /etc empty cleanly. For more information on the
638 concepts involved see this recent blog story:
640 http://0pointer.de/blog/projects/stateless.html
642 * A new system group "input" has been introduced, and all
643 input device nodes get this group assigned. This is useful
644 for system-level software to get access to input devices. It
645 complements what is already done for "audio" and "video".
647 * systemd-networkd learnt minimal DHCPv4 server support in
648 addition to the existing DHCPv4 client support. It also
649 learnt DHCPv6 client and IPv6 Router Solicitation client
650 support. The DHCPv4 client gained support for static routes
651 passed in from the server. Note that the [DHCPv4] section
652 known in older systemd-networkd versions has been renamed to
653 [DHCP] and is now also used by the DHCPv6 client. Existing
654 .network files using settings of this section should be
655 updated, though compatibility is maintained. Optionally, the
656 client hostname may now be sent to the DHCP server.
658 * networkd gained support for vxlan virtual networks as well
659 as tun/tap and dummy devices.
661 * networkd gained support for automatic allocation of address
662 ranges for interfaces from a system-wide pool of
663 addresses. This is useful for dynamically managing a large
664 number of interfaces with a single network configuration
665 file. In particular this is useful to easily assign
666 appropriate IP addresses to the veth links of a large number
669 * RPM macros for processing sysusers, sysctl and binfmt
670 drop-in snippets at package installation time have been
673 * The /etc/os-release file should now be placed in
674 /usr/lib/os-release. The old location is automatically
675 created as symlink. /usr/lib is the more appropriate
676 location of this file, since it shall actually describe the
677 vendor operating system shipped in /usr, and not the
678 configuration stored in /etc.
680 * .mount units gained a new boolean SloppyOptions= setting
681 that maps to mount(8)'s -s option which enables permissive
682 parsing of unknown mount options.
684 * tmpfiles learnt a new "L+" directive which creates a symlink
685 but (unlike "L") deletes a pre-existing file first, should
686 it already exist and not already be the correct
687 symlink. Similar, "b+", "c+" and "p+" directives have been
688 added as well, which create block and character devices, as
689 well as fifos in the filesystem, possibly removing any
690 pre-existing files of different types.
692 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
693 'argument' field (which so far specified the source to
694 symlink/copy the files from) is now optional. If omitted the
695 same file os copied from /usr/share/factory/ suffixed by the
696 full destination path. This is useful for populating /etc
697 with essential files, by copying them from vendor defaults
698 shipped in /usr/share/factory/etc.
700 * A new command "systemctl preset-all" has been added that
701 applies the service preset settings to all installed unit
702 files. A new switch --preset-mode= has been added that
703 controls whether only enable or only disable operations
706 * A new command "systemctl is-system-running" has been added
707 that allows checking the overall state of the system, for
708 example whether it is fully up and running.
710 * When the system boots up with an empty /etc, the equivalent
711 to "systemctl preset-all" is executed during early boot, to
712 make sure all default services are enabled after a factory
715 * systemd now contains a minimal preset file that enables the
716 most basic services systemd ships by default.
718 * Unit files' [Install] section gained a new DefaultInstance=
719 field for defining the default instance to create if a
720 template unit is enabled with no instance specified.
722 * A new passive target cryptsetup-pre.target has been added
723 that may be used by services that need to make they run and
724 finish before the first LUKS cryptographic device is set up.
726 * The /dev/loop-control and /dev/btrfs-control device nodes
727 are now owned by the "disk" group by default, opening up
728 access to this group.
730 * systemd-coredump will now automatically generate a
731 stack trace of all core dumps taking place on the system,
732 based on elfutils' libdw library. This stack trace is logged
735 * systemd-coredump may now optionally store coredumps directly
736 on disk (in /var/lib/systemd/coredump, possibly compressed),
737 instead of storing them unconditionally in the journal. This
738 mode is the new default. A new configuration file
739 /etc/systemd/coredump.conf has been added to configure this
740 and other parameters of systemd-coredump.
742 * coredumpctl gained a new "info" verb to show details about a
743 specific coredump. A new switch "-1" has also been added
744 that makes sure to only show information about the most
745 recent entry instead of all entries. Also, as the tool is
746 generally useful now the "systemd-" prefix of the binary
747 name has been removed. Distributions that want to maintain
748 compatibility with the old name should add a symlink from
749 the old name to the new name.
751 * journald's SplitMode= now defaults to "uid". This makes sure
752 that unprivileged users can access their own coredumps with
753 coredumpctl without restrictions.
755 * New kernel command line options "systemd.wants=" (for
756 pulling an additional unit during boot), "systemd.mask="
757 (for masking a specific unit for the boot), and
758 "systemd.debug-shell" (for enabling the debug shell on tty9)
759 have been added. This is implemented in the new generator
760 "systemd-debug-generator".
762 * systemd-nspawn will now by default filter a couple of
763 syscalls for containers, among them those required for
764 kernel module loading, direct x86 IO port access, swap
765 management, and kexec. Most importantly though
766 open_by_handle_at() is now prohibited for containers,
767 closing a hole similar to a recently discussed vulnerability
768 in docker regarding access to files on file hierarchies the
769 container should normally not have access to. Note that for
770 nspawn we generally make no security claims anyway (and
771 this is explicitly documented in the man page), so this is
772 just a fix for one of the most obvious problems.
774 * A new man page file-hierarchy(7) has been added that
775 contains a minimized, modernized version of the file system
776 layout systemd expects, similar in style to the FHS
777 specification or hier(5). A new tool systemd-path(1) has
778 been added to query many of these paths for the local
781 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
782 longer done. Since the directory now has a per-user size
783 limit, and is cleaned on logout this appears unnecessary,
784 in particular since this now brings the lifecycle of this
785 directory closer in line with how IPC objects are handled.
787 * systemd.pc now exports a number of additional directories,
788 including $libdir (which is useful to identify the library
789 path for the primary architecture of the system), and a
790 couple of drop-in directories.
792 * udev's predictable network interface names now use the dev_port
793 sysfs attribute, introduced in linux 3.15 instead of dev_id to
794 distinguish between ports of the same PCI function. dev_id should
795 only be used for ports using the same HW address, hence the need
798 * machined has been updated to export the OS version of a
799 container (read from /etc/os-release and
800 /usr/lib/os-release) on the bus. This is now shown in
801 "machinectl status" for a machine.
803 * A new service setting RestartForceExitStatus= has been
804 added. If configured to a set of exit signals or process
805 return values, the service will be restarted when the main
806 daemon process exits with any of them, regardless of the
809 * systemctl's -H switch for connecting to remote systemd
810 machines has been extended so that it may be used to
811 directly connect to a specific container on the
812 host. "systemctl -H root@foobar:waldi" will now connect as
813 user "root" to host "foobar", and then proceed directly to
814 the container named "waldi". Note that currently you have to
815 authenticate as user "root" for this to work, as entering
816 containers is a privileged operation.
818 Contributions from: Andreas Henriksson, Benjamin Steinwender,
819 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
820 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
821 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
822 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
823 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
824 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
825 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
826 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
827 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
828 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
829 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
831 -- Berlin, 2014-07-03
835 * As an experimental feature, udev now tries to lock the
836 disk device node (flock(LOCK_SH|LOCK_NB)) while it
837 executes events for the disk or any of its partitions.
838 Applications like partitioning programs can lock the
839 disk device node (flock(LOCK_EX)) and claim temporary
840 device ownership that way; udev will entirely skip all event
841 handling for this disk and its partitions. If the disk
842 was opened for writing, the close will trigger a partition
843 table rescan in udev's "watch" facility, and if needed
844 synthesize "change" events for the disk and all its partitions.
845 This is now unconditionally enabled, and if it turns out to
846 cause major problems, we might turn it on only for specific
847 devices, or might need to disable it entirely. Device Mapper
848 devices are excluded from this logic.
850 * We temporarily dropped the "-l" switch for fsck invocations,
851 since they collide with the flock() logic above. util-linux
852 upstream has been changed already to avoid this conflict,
853 and we will readd "-l" as soon as util-linux with this
854 change has been released.
856 * The dependency on libattr has been removed. Since a long
857 time, the extended attribute calls have moved to glibc, and
858 libattr is thus unnecessary.
860 * Virtualization detection works without priviliges now. This
861 means the systemd-detect-virt binary no longer requires
862 CAP_SYS_PTRACE file capabilities, and our daemons can run
863 with fewer privileges.
865 * systemd-networkd now runs under its own "systemd-network"
866 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
867 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
868 loses the ability to write to files owned by root this way.
870 * Similar, systemd-resolved now runs under its own
871 "systemd-resolve" user with no capabilities remaining.
873 * Similar, systemd-bus-proxyd now runs under its own
874 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
876 * systemd-networkd gained support for setting up "veth"
877 virtual ethernet devices for container connectivity, as well
878 as GRE and VTI tunnels.
880 * systemd-networkd will no longer automatically attempt to
881 manually load kernel modules necessary for certain tunnel
882 transports. Instead, it is assumed the kernel loads them
883 automatically when required. This only works correctly on
884 very new kernels. On older kernels, please consider adding
885 the kernel modules to /etc/modules-load.d/ as a work-around.
887 * The resolv.conf file systemd-resolved generates has been
888 moved to /run/systemd/resolve/. If you have a symlink from
889 /etc/resolv.conf, it might be necessary to correct it.
891 * Two new service settings, ProtectHome= and ProtectSystem=,
892 have been added. When enabled, they will make the user data
893 (such as /home) inaccessible or read-only and the system
894 (such as /usr) read-only, for specific services. This allows
895 very light-weight per-service sandboxing to avoid
896 modifications of user data or system files from
897 services. These two new switches have been enabled for all
898 of systemd's long-running services, where appropriate.
900 * Socket units gained new SocketUser= and SocketGroup=
901 settings to set the owner user and group of AF_UNIX sockets
902 and FIFOs in the file system.
904 * Socket units gained a new RemoveOnStop= setting. If enabled,
905 all FIFOS and sockets in the file system will be removed
906 when the specific socket unit is stopped.
908 * Socket units gained a new Symlinks= setting. It takes a list
909 of symlinks to create to file system sockets or FIFOs
910 created by the specific Unix sockets. This is useful to
911 manage symlinks to socket nodes with the same life-cycle as
914 * The /dev/log socket and /dev/initctl FIFO have been moved to
915 /run, and have been replaced by symlinks. This allows
916 connecting to these facilities even if PrivateDevices=yes is
917 used for a service (which makes /dev/log itself unavailable,
918 but /run is left). This also has the benefit of ensuring
919 that /dev only contains device nodes, directories and
920 symlinks, and nothing else.
922 * sd-daemon gained two new calls sd_pid_notify() and
923 sd_pid_notifyf(). They are similar to sd_notify() and
924 sd_notifyf(), but allow overriding of the source PID of
925 notification messages if permissions permit this. This is
926 useful to send notify messages on behalf of a different
927 process (for example, the parent process). The
928 systemd-notify tool has been updated to make use of this
929 when sending messages (so that notification messages now
930 originate from the shell script invoking systemd-notify and
931 not the systemd-notify process itself. This should minimize
932 a race where systemd fails to associate notification
933 messages to services when the originating process already
936 * A new "on-abnormal" setting for Restart= has been added. If
937 set, it will result in automatic restarts on all "abnormal"
938 reasons for a process to exit, which includes unclean
939 signals, core dumps, timeouts and watchdog timeouts, but
940 does not include clean and unclean exit codes or clean
941 signals. Restart=on-abnormal is an alternative for
942 Restart=on-failure for services that shall be able to
943 terminate and avoid restarts on certain errors, by
944 indicating so with an unclean exit code. Restart=on-failure
945 or Restart=on-abnormal is now the recommended setting for
946 all long-running services.
948 * If the InaccessibleDirectories= service setting points to a
949 mount point (or if there are any submounts contained within
950 it), it is now attempted to completely unmount it, to make
951 the file systems truly unavailable for the respective
954 * The ReadOnlyDirectories= service setting and
955 systemd-nspawn's --read-only parameter are now recursively
956 applied to all submounts, too.
958 * Mount units may now be created transiently via the bus APIs.
960 * The support for SysV and LSB init scripts has been removed
961 from the systemd daemon itself. Instead, it is now
962 implemented as a generator that creates native systemd units
963 from these scripts when needed. This enables us to remove a
964 substantial amount of legacy code from PID 1, following the
965 fact that many distributions only ship a very small number
966 of LSB/SysV init scripts nowadays.
968 * Priviliged Xen (dom0) domains are not considered
969 virtualization anymore by the virtualization detection
970 logic. After all, they generally have unrestricted access to
971 the hardware and usually are used to manage the unprivileged
974 * systemd-tmpfiles gained a new "C" line type, for copying
975 files or entire directories.
977 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
978 lines. So far, they have been non-globbing versions of the
979 latter, and have thus been redundant. In future, it is
980 recommended to only use "z". "m" has hence been removed
981 from the documentation, even though it stays supported.
983 * A tmpfiles snippet to recreate the most basic structure in
984 /var has been added. This is enough to create the /var/run →
985 /run symlink and create a couple of structural
986 directories. This allows systems to boot up with an empty or
987 volatile /var. Of course, while with this change, the core OS
988 now is capable with dealing with a volatile /var, not all
989 user services are ready for it. However, we hope that sooner
990 or later, many service daemons will be changed upstream so
991 that they are able to automatically create their necessary
992 directories in /var at boot, should they be missing. This is
993 the first step to allow state-less systems that only require
994 the vendor image for /usr to boot.
996 * systemd-nspawn has gained a new --tmpfs= switch to mount an
997 empty tmpfs instance to a specific directory. This is
998 particularly useful for making use of the automatic
999 reconstruction of /var (see above), by passing --tmpfs=/var.
1001 * Access modes specified in tmpfiles snippets may now be
1002 prefixed with "~", which indicates that they shall be masked
1003 by whether the existing file or directory is currently
1004 writable, readable or executable at all. Also, if specified,
1005 the sgid/suid/sticky bits will be masked for all
1008 * A new passive target unit "network-pre.target" has been
1009 added which is useful for services that shall run before any
1010 network is configured, for example firewall scripts.
1012 * The "floppy" group that previously owned the /dev/fd*
1013 devices is no longer used. The "disk" group is now used
1014 instead. Distributions should probably deprecate usage of
1017 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
1018 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
1019 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
1020 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
1021 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
1022 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
1025 -- Berlin, 2014-06-11
1029 * A new "systemd-timesyncd" daemon has been added for
1030 synchronizing the system clock across the network. It
1031 implements an SNTP client. In contrast to NTP
1032 implementations such as chrony or the NTP reference server,
1033 this only implements a client side, and does not bother with
1034 the full NTP complexity, focusing only on querying time from
1035 one remote server and synchronizing the local clock to
1036 it. Unless you intend to serve NTP to networked clients or
1037 want to connect to local hardware clocks, this simple NTP
1038 client should be more than appropriate for most
1039 installations. The daemon runs with minimal privileges, and
1040 has been hooked up with networkd to only operate when
1041 network connectivity is available. The daemon saves the
1042 current clock to disk every time a new NTP sync has been
1043 acquired, and uses this to possibly correct the system clock
1044 early at bootup, in order to accommodate for systems that
1045 lack an RTC such as the Raspberry Pi and embedded devices,
1046 and to make sure that time monotonically progresses on these
1047 systems, even if it is not always correct. To make use of
1048 this daemon, a new system user and group "systemd-timesync"
1049 needs to be created on installation of systemd.
1051 * The queue "seqnum" interface of libudev has been disabled, as
1052 it was generally incompatible with device namespacing as
1053 sequence numbers of devices go "missing" if the devices are
1054 part of a different namespace.
1056 * "systemctl list-timers" and "systemctl list-sockets" gained
1057 a --recursive switch for showing units of these types also
1058 for all local containers, similar in style to the already
1059 supported --recursive switch for "systemctl list-units".
1061 * A new RebootArgument= setting has been added for service
1062 units, which may be used to specify a kernel reboot argument
1063 to use when triggering reboots with StartLimitAction=.
1065 * A new FailureAction= setting has been added for service
1066 units which may be used to specify an operation to trigger
1067 when a service fails. This works similarly to
1068 StartLimitAction=, but unlike it, controls what is done
1069 immediately rather than only after several attempts to
1070 restart the service in question.
1072 * hostnamed got updated to also expose the kernel name,
1073 release, and version on the bus. This is useful for
1074 executing commands like hostnamectl with the -H switch.
1075 systemd-analyze makes use of this to properly display
1076 details when running non-locally.
1078 * The bootchart tool can now show cgroup information in the
1079 graphs it generates.
1081 * The CFS CPU quota cgroup attribute is now exposed for
1082 services. The new CPUQuota= switch has been added for this
1083 which takes a percentage value. Setting this will have the
1084 result that a service may never get more CPU time than the
1085 specified percentage, even if the machine is otherwise idle.
1087 * systemd-networkd learned IPIP and SIT tunnel support.
1089 * LSB init scripts exposing a dependency on $network will now
1090 get a dependency on network-online.target rather than simply
1091 network.target. This should bring LSB handling closer to
1092 what it was on SysV systems.
1094 * A new fsck.repair= kernel option has been added to control
1095 how fsck shall deal with unclean file systems at boot.
1097 * The (.ini) configuration file parser will now silently
1098 ignore sections whose name begins with "X-". This may be
1099 used to maintain application-specific extension sections in unit
1102 * machined gained a new API to query the IP addresses of
1103 registered containers. "machinectl status" has been updated
1104 to show these addresses in its output.
1106 * A new call sd_uid_get_display() has been added to the
1107 sd-login APIs for querying the "primary" session of a
1108 user. The "primary" session of the user is elected from the
1109 user's sessions and generally a graphical session is
1110 preferred over a text one.
1112 * A minimal systemd-resolved daemon has been added. It
1113 currently simply acts as a companion to systemd-networkd and
1114 manages resolv.conf based on per-interface DNS
1115 configuration, possibly supplied via DHCP. In the long run
1116 we hope to extend this into a local DNSSEC enabled DNS and
1119 * The systemd-networkd-wait-online tool is now enabled by
1120 default. It will delay network-online.target until a network
1121 connection has been configured. The tool primarily integrates
1122 with networkd, but will also make a best effort to make sense
1123 of network configuration performed in some other way.
1125 * Two new service options StartupCPUShares= and
1126 StartupBlockIOWeight= have been added that work similarly to
1127 CPUShares= and BlockIOWeight= however only apply during
1128 system startup. This is useful to prioritize certain services
1129 differently during bootup than during normal runtime.
1131 * hostnamed has been changed to prefer the statically
1132 configured hostname in /etc/hostname (unless set to
1133 'localhost' or empty) over any dynamic one supplied by
1134 dhcp. With this change, the rules for picking the hostname
1135 match more closely the rules of other configuration settings
1136 where the local administrator's configuration in /etc always
1137 overrides any other settings.
1139 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
1140 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
1141 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
1142 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
1143 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
1144 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
1145 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
1146 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
1147 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
1148 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
1149 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
1150 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
1151 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
1152 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
1153 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
1154 Lindskog, WaLyong Cho, Will Woods, Zbigniew
1157 -- Beijing, 2014-05-28
1161 * When restoring the screen brightness at boot, stay away from
1162 the darkest setting or from the lowest 5% of the available
1163 range, depending on which is the larger value of both. This
1164 should effectively protect the user from rebooting into a
1165 black screen, should the brightness have been set to minimum
1168 * sd-login gained a new sd_machine_get_class() call to
1169 determine the class ("vm" or "container") of a machine
1170 registered with machined.
1172 * sd-login gained new calls
1173 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
1174 to query the identity of the peer of a local AF_UNIX
1175 connection. They operate similarly to their sd_pid_get_xyz()
1178 * PID 1 will now maintain a system-wide system state engine
1179 with the states "starting", "running", "degraded",
1180 "maintenance", "stopping". These states are bound to system
1181 startup, normal runtime, runtime with at least one failed
1182 service, rescue/emergency mode and system shutdown. This
1183 state is shown in the "systemctl status" output when no unit
1184 name is passed. It is useful to determine system state, in
1185 particularly when doing so for many systems or containers at
1188 * A new command "list-machines" has been added to "systemctl"
1189 that lists all local OS containers and shows their system
1190 state (see above), if systemd runs inside of them.
1192 * systemctl gained a new "-r" switch to recursively enumerate
1193 units on all local containers, when used with the
1194 "list-unit" command (which is the default one that is
1195 executed when no parameters are specified).
1197 * The GPT automatic partition discovery logic will now honour
1198 two GPT partition flags: one may be set on a partition to
1199 cause it to be mounted read-only, and the other may be set
1200 on a partition to ignore it during automatic discovery.
1202 * Two new GPT type UUIDs have been added for automatic root
1203 partition discovery, for 32-bit and 64-bit ARM. This is not
1204 particularly useful for discovering the root directory on
1205 these architectures during bare-metal boots (since UEFI is
1206 not common there), but still very useful to allow booting of
1207 ARM disk images in nspawn with the -i option.
1209 * MAC addresses of interfaces created with nspawn's
1210 --network-interface= switch will now be generated from the
1211 machine name, and thus be stable between multiple invocations
1214 * logind will now automatically remove all IPC objects owned
1215 by a user if she or he fully logs out. This makes sure that
1216 users who are logged out cannot continue to consume IPC
1217 resources. This covers SysV memory, semaphores and message
1218 queues as well as POSIX shared memory and message
1219 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1220 limits. With this functionality, that is corrected. This may
1221 be turned off by using the RemoveIPC= switch of logind.conf.
1223 * The systemd-machine-id-setup and tmpfiles tools gained a
1224 --root= switch to operate on a specific root directory,
1227 * journald can now forward logged messages to the TTYs of all
1228 logged in users ("wall"). This is the default for all
1229 emergency messages now.
1231 * A new tool systemd-journal-remote has been added to stream
1232 journal log messages across the network.
1234 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1235 controller trees are mounted into it. Note that the
1236 directories mounted beneath it are not read-only. This is a
1237 security measure and is particularly useful because glibc
1238 actually includes a search logic to pick any tmpfs it can
1239 find to implement shm_open() if /dev/shm is not available
1240 (which it might very well be in namespaced setups).
1242 * machinectl gained a new "poweroff" command to cleanly power
1243 down a local OS container.
1245 * The PrivateDevices= unit file setting will now also drop the
1246 CAP_MKNOD capability from the capability bound set, and
1247 imply DevicePolicy=closed.
1249 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1250 comprehensively on all long-running systemd services where
1251 this is appropriate.
1253 * systemd-udevd will now run in a disassociated mount
1254 namespace. To mount directories from udev rules, make sure to
1255 pull in mount units via SYSTEMD_WANTS properties.
1257 * The kdbus support gained support for uploading policy into
1258 the kernel. sd-bus gained support for creating "monitoring"
1259 connections that can eavesdrop into all bus communication
1260 for debugging purposes.
1262 * Timestamps may now be specified in seconds since the UNIX
1263 epoch Jan 1st, 1970 by specifying "@" followed by the value
1266 * Native tcpwrap support in systemd has been removed. tcpwrap
1267 is old code, not really maintained anymore and has serious
1268 shortcomings, and better options such as firewalls
1269 exist. For setups that require tcpwrap usage, please
1270 consider invoking your socket-activated service via tcpd,
1271 like on traditional inetd.
1273 * A new system.conf configuration option
1274 DefaultTimerAccuracySec= has been added that controls the
1275 default AccuracySec= setting of .timer units.
1277 * Timer units gained a new WakeSystem= switch. If enabled,
1278 timers configured this way will cause the system to resume
1279 from system suspend (if the system supports that, which most
1282 * Timer units gained a new Persistent= switch. If enabled,
1283 timers configured this way will save to disk when they have
1284 been last triggered. This information is then used on next
1285 reboot to possible execute overdue timer events, that
1286 could not take place because the system was powered off.
1287 This enables simple anacron-like behaviour for timer units.
1289 * systemctl's "list-timers" will now also list the time a
1290 timer unit was last triggered in addition to the next time
1291 it will be triggered.
1293 * systemd-networkd will now assign predictable IPv4LL
1294 addresses to its local interfaces.
1296 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1297 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1298 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1299 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1300 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1301 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1302 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1303 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1306 -- Berlin, 2014-03-25
1310 * A new unit file setting RestrictAddressFamilies= has been
1311 added to restrict which socket address families unit
1312 processes gain access to. This takes address family names
1313 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1314 attack surface of services via exotic protocol stacks. This
1315 is built on seccomp system call filters.
1317 * Two new unit file settings RuntimeDirectory= and
1318 RuntimeDirectoryMode= have been added that may be used to
1319 manage a per-daemon runtime directories below /run. This is
1320 an alternative for setting up directory permissions with
1321 tmpfiles snippets, and has the advantage that the runtime
1322 directory's lifetime is bound to the daemon runtime and that
1323 the daemon starts up with an empty directory each time. This
1324 is particularly useful when writing services that drop
1325 privileges using the User= or Group= setting.
1327 * The DeviceAllow= unit setting now supports globbing for
1328 matching against device group names.
1330 * The systemd configuration file system.conf gained new
1331 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1332 DefaultMemoryAccounting= to globally turn on/off accounting
1333 for specific resources (cgroups) for all units. These
1334 settings may still be overridden individually in each unit
1337 * systemd-gpt-auto-generator is now able to discover /srv and
1338 root partitions in addition to /home and swap partitions. It
1339 also supports LUKS-encrypted partitions now. With this in
1340 place, automatic discovery of partitions to mount following
1341 the Discoverable Partitions Specification
1342 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1343 is now a lot more complete. This allows booting without
1344 /etc/fstab and without root= on the kernel command line on
1345 systems prepared appropriately.
1347 * systemd-nspawn gained a new --image= switch which allows
1348 booting up disk images and Linux installations on any block
1349 device that follow the Discoverable Partitions Specification
1350 (see above). This means that installations made with
1351 appropriately updated installers may now be started and
1352 deployed using container managers, completely
1353 unmodified. (We hope that libvirt-lxc will add support for
1354 this feature soon, too.)
1356 * systemd-nspawn gained a new --network-macvlan= setting to
1357 set up a private macvlan interface for the
1358 container. Similarly, systemd-networkd gained a new
1359 Kind=macvlan setting in .netdev files.
1361 * systemd-networkd now supports configuring local addresses
1364 * A new tool systemd-network-wait-online has been added to
1365 synchronously wait for network connectivity using
1368 * The sd-bus.h bus API gained a new sd_bus_track object for
1369 tracking the life-cycle of bus peers. Note that sd-bus.h is
1370 still not a public API though (unless you specify
1371 --enable-kdbus on the configure command line, which however
1372 voids your warranty and you get no API stability guarantee).
1374 * The $XDG_RUNTIME_DIR runtime directories for each user are
1375 now individual tmpfs instances, which has the benefit of
1376 introducing separate pools for each user, with individual
1377 size limits, and thus making sure that unprivileged clients
1378 can no longer negatively impact the system or other users by
1379 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1380 RuntimeDirectorySize= has been introduced that allows
1381 controlling the default size limit for all users. It
1382 defaults to 10% of the available physical memory. This is no
1383 replacement for quotas on tmpfs though (which the kernel
1384 still does not support), as /dev/shm and /tmp are still
1385 shared resources used by both the system and unprivileged
1388 * logind will now automatically turn off automatic suspending
1389 on laptop lid close when more than one display is
1390 connected. This was previously expected to be implemented
1391 individually in desktop environments (such as GNOME),
1392 however has been added to logind now, in order to fix a
1393 boot-time race where a desktop environment might not have
1394 been started yet and thus not been able to take an inhibitor
1395 lock at the time where logind already suspends the system
1396 due to a closed lid.
1398 * logind will now wait at least 30s after each system
1399 suspend/resume cycle, and 3min after system boot before
1400 suspending the system due to a closed laptop lid. This
1401 should give USB docking stations and similar enough time to
1402 be probed and configured after system resume and boot in
1403 order to then act as suspend blocker.
1405 * systemd-run gained a new --property= setting which allows
1406 initialization of resource control properties (and others)
1407 for the created scope or service unit. Example: "systemd-run
1408 --property=BlockIOWeight=10 updatedb" may be used to run
1409 updatedb at a low block IO scheduling weight.
1411 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1412 now also work in --scope mode.
1414 * When systemd is compiled with kdbus support, basic support
1415 for enforced policies is now in place. (Note that enabling
1416 kdbus still voids your warranty and no API compatibility
1419 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1420 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1421 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1422 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1423 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1424 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1425 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1426 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1427 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1428 Zbigniew Jędrzejewski-Szmek
1430 -- Berlin, 2014-03-12
1434 * systemd will now relabel /dev after loading the SMACK policy
1435 according to SMACK rules.
1437 * A new unit file option AppArmorProfile= has been added to
1438 set the AppArmor profile for the processes of a unit.
1440 * A new condition check ConditionArchitecture= has been added
1441 to conditionalize units based on the system architecture, as
1442 reported by uname()'s "machine" field.
1444 * systemd-networkd now supports matching on the system
1445 virtualization, architecture, kernel command line, host name
1448 * logind is now a lot more aggressive when suspending the
1449 machine due to a closed laptop lid. Instead of acting only
1450 on the lid close action, it will continuously watch the lid
1451 status and act on it. This is useful for laptops where the
1452 power button is on the outside of the chassis so that it can
1453 be reached without opening the lid (such as the Lenovo
1454 Yoga). On those machines, logind will now immediately
1455 re-suspend the machine if the power button has been
1456 accidentally pressed while the laptop was suspended and in a
1457 backpack or similar.
1459 * logind will now watch SW_DOCK switches and inhibit reaction
1460 to the lid switch if it is pressed. This means that logind
1461 will not suspend the machine anymore if the lid is closed
1462 and the system is docked, if the laptop supports SW_DOCK
1463 notifications via the input layer. Note that ACPI docking
1464 stations do not generate this currently. Also note that this
1465 logic is usually not fully sufficient and Desktop
1466 Environments should take a lid switch inhibitor lock when an
1467 external display is connected, as systemd will not watch
1470 * nspawn will now make use of the devices cgroup controller by
1471 default, and only permit creation of and access to the usual
1472 API device nodes like /dev/null or /dev/random, as well as
1473 access to (but not creation of) the pty devices.
1475 * We will now ship a default .network file for
1476 systemd-networkd that automatically configures DHCP for
1477 network interfaces created by nspawn's --network-veth or
1478 --network-bridge= switches.
1480 * systemd will now understand the usual M, K, G, T suffixes
1481 according to SI conventions (i.e. to the base 1000) when
1482 referring to throughput and hardware metrics. It will stay
1483 with IEC conventions (i.e. to the base 1024) for software
1484 metrics, according to what is customary according to
1485 Wikipedia. We explicitly document which base applies for
1486 each configuration option.
1488 * The DeviceAllow= setting in unit files now supports a syntax
1489 to whitelist an entire group of devices node majors at once,
1490 based on the /proc/devices listing. For example, with the
1491 string "char-pts", it is now possible to whitelist all
1492 current and future pseudo-TTYs at once.
1494 * sd-event learned a new "post" event source. Event sources of
1495 this type are triggered by the dispatching of any event
1496 source of a type that is not "post". This is useful for
1497 implementing clean-up and check event sources that are
1498 triggered by other work being done in the program.
1500 * systemd-networkd is no longer statically enabled, but uses
1501 the usual [Install] sections so that it can be
1502 enabled/disabled using systemctl. It still is enabled by
1505 * When creating a veth interface pair with systemd-nspawn, the
1506 host side will now be prefixed with "vb-" if
1507 --network-bridge= is used, and with "ve-" if --network-veth
1508 is used. This way, it is easy to distinguish these cases on
1509 the host, for example to apply different configuration to
1510 them with systemd-networkd.
1512 * The compatibility libraries for libsystemd-journal.so,
1513 libsystem-id128.so, libsystemd-login.so and
1514 libsystemd-daemon.so do not make use of IFUNC
1515 anymore. Instead, we now build libsystemd.so multiple times
1516 under these alternative names. This means that the footprint
1517 is drastically increased, but given that these are
1518 transitional compatibility libraries, this should not matter
1519 much. This change has been made necessary to support the ARM
1520 platform for these compatibility libraries, as the ARM
1521 toolchain is not really at the same level as the toolchain
1522 for other architectures like x86 and does not support
1523 IFUNC. Please make sure to use --enable-compat-libs only
1524 during a transitional period!
1526 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1527 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1528 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1529 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1530 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1531 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1532 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1533 Zbigniew Jędrzejewski-Szmek
1535 -- Berlin, 2014-02-24
1539 * A new component "systemd-networkd" has been added that can
1540 be used to configure local network interfaces statically or
1541 via DHCP. It is capable of bringing up bridges, VLANs, and
1542 bonding. Currently, no hook-ups for interactive network
1543 configuration are provided. Use this for your initrd,
1544 container, embedded, or server setup if you need a simple,
1545 yet powerful, network configuration solution. This
1546 configuration subsystem is quite nifty, as it allows wildcard
1547 hotplug matching in interfaces. For example, with a single
1548 configuration snippet, you can configure that all Ethernet
1549 interfaces showing up are automatically added to a bridge,
1550 or similar. It supports link-sensing and more.
1552 * A new tool "systemd-socket-proxyd" has been added which can
1553 act as a bidirectional proxy for TCP sockets. This is
1554 useful for adding socket activation support to services that
1555 do not actually support socket activation, including virtual
1556 machines and the like.
1558 * Add a new tool to save/restore rfkill state on
1561 * Save/restore state of keyboard backlights in addition to
1562 display backlights on shutdown/boot.
1564 * udev learned a new SECLABEL{} construct to label device
1565 nodes with a specific security label when they appear. For
1566 now, only SECLABEL{selinux} is supported, but the syntax is
1567 prepared for additional security frameworks.
1569 * udev gained a new scheme to configure link-level attributes
1570 from files in /etc/systemd/network/*.link. These files can
1571 match against MAC address, device path, driver name and type,
1572 and will apply attributes like the naming policy, link speed,
1573 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1574 address assignment policy (randomized, ...).
1576 * The configuration of network interface naming rules for
1577 "permanent interface names" has changed: a new NamePolicy=
1578 setting in the [Link] section of .link files determines the
1579 priority of possible naming schemes (onboard, slot, mac,
1580 path). The default value of this setting is determined by
1581 /usr/lib/net/links/99-default.link. Old
1582 80-net-name-slot.rules udev configuration file has been
1583 removed, so local configuration overriding this file should
1584 be adapated to override 99-default.link instead.
1586 * When the User= switch is used in a unit file, also
1587 initialize $SHELL= based on the user database entry.
1589 * systemd no longer depends on libdbus. All communication is
1590 now done with sd-bus, systemd's low-level bus library
1593 * kdbus support has been added to PID 1 itself. When kdbus is
1594 enabled, this causes PID 1 to set up the system bus and
1595 enable support for a new ".busname" unit type that
1596 encapsulates bus name activation on kdbus. It works a little
1597 bit like ".socket" units, except for bus names. A new
1598 generator has been added that converts classic dbus1 service
1599 activation files automatically into native systemd .busname
1602 * sd-bus: add a light-weight vtable implementation that allows
1603 defining objects on the bus with a simple static const
1604 vtable array of its methods, signals and properties.
1606 * systemd will not generate or install static dbus
1607 introspection data anymore to /usr/share/dbus-1/interfaces,
1608 as the precise format of these files is unclear, and
1609 nothing makes use of it.
1611 * A proxy daemon is now provided to proxy clients connecting
1612 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1613 compatibility with classic D-Bus.
1615 * A bus driver implementation has been added that supports the
1616 classic D-Bus bus driver calls on kdbus, also for
1617 compatibility purposes.
1619 * A new API "sd-event.h" has been added that implements a
1620 minimal event loop API built around epoll. It provides a
1621 couple of features that direct epoll usage is lacking:
1622 prioritization of events, scales to large numbers of timer
1623 events, per-event timer slack (accuracy), system-wide
1624 coalescing of timer events, exit handlers, watchdog
1625 supervision support using systemd's sd_notify() API, child
1628 * A new API "sd-rntl.h" has been added that provides an API
1629 around the route netlink interface of the kernel, similar in
1630 style to "sd-bus.h".
1632 * A new API "sd-dhcp-client.h" has been added that provides a
1633 small DHCPv4 client-side implementation. This is used by
1636 * There is a new kernel command line option
1637 "systemd.restore_state=0|1". When set to "0", none of the
1638 systemd tools will restore saved runtime state to hardware
1639 devices. More specifically, the rfkill and backlight states
1642 * The FsckPassNo= compatibility option in mount/service units
1643 has been removed. The fstab generator will now add the
1644 necessary dependencies automatically, and does not require
1645 PID1's support for that anymore.
1647 * journalctl gained a new switch, --list-boots, that lists
1648 recent boots with their times and boot IDs.
1650 * The various tools like systemctl, loginctl, timedatectl,
1651 busctl, systemd-run, ... have gained a new switch "-M" to
1652 connect to a specific, local OS container (as direct
1653 connection, without requiring SSH). This works on any
1654 container that is registered with machined, such as those
1655 created by libvirt-lxc or nspawn.
1657 * systemd-run and systemd-analyze also gained support for "-H"
1658 to connect to remote hosts via SSH. This is particularly
1659 useful for systemd-run because it enables queuing of jobs
1660 onto remote systems.
1662 * machinectl gained a new command "login" to open a getty
1663 login in any local container. This works with any container
1664 that is registered with machined (such as those created by
1665 libvirt-lxc or nspawn), and which runs systemd inside.
1667 * machinectl gained a new "reboot" command that may be used to
1668 trigger a reboot on a specific container that is registered
1669 with machined. This works on any container that runs an init
1670 system of some kind.
1672 * systemctl gained a new "list-timers" command to print a nice
1673 listing of installed timer units with the times they elapse
1676 * Alternative reboot() parameters may now be specified on the
1677 "systemctl reboot" command line and are passed to the
1678 reboot() system call.
1680 * systemctl gained a new --job-mode= switch to configure the
1681 mode to queue a job with. This is a more generic version of
1682 --fail, --irreversible, and --ignore-dependencies, which are
1683 still available but not advertised anymore.
1685 * /etc/systemd/system.conf gained new settings to configure
1686 various default timeouts of units, as well as the default
1687 start limit interval and burst. These may still be overridden
1690 * PID1 will now export on the bus profile data of the security
1691 policy upload process (such as the SELinux policy upload to
1694 * journald: when forwarding logs to the console, include
1695 timestamps (following the setting in
1696 /sys/module/printk/parameters/time).
1698 * OnCalendar= in timer units now understands the special
1699 strings "yearly" and "annually". (Both are equivalent)
1701 * The accuracy of timer units is now configurable with the new
1702 AccuracySec= setting. It defaults to 1min.
1704 * A new dependency type JoinsNamespaceOf= has been added that
1705 allows running two services within the same /tmp and network
1706 namespace, if PrivateNetwork= or PrivateTmp= are used.
1708 * A new command "cat" has been added to systemctl. It outputs
1709 the original unit file of a unit, and concatenates the
1710 contents of additional "drop-in" unit file snippets, so that
1711 the full configuration is shown.
1713 * systemctl now supports globbing on the various "list-xyz"
1714 commands, like "list-units" or "list-sockets", as well as on
1715 those commands which take multiple unit names.
1717 * journalctl's --unit= switch gained support for globbing.
1719 * All systemd daemons now make use of the watchdog logic so
1720 that systemd automatically notices when they hang.
1722 * If the $container_ttys environment variable is set,
1723 getty-generator will automatically spawn a getty for each
1724 listed tty. This is useful for container managers to request
1725 login gettys to be spawned on as many ttys as needed.
1727 * %h, %s, %U specifier support is not available anymore when
1728 used in unit files for PID 1. This is because NSS calls are
1729 not safe from PID 1. They stay available for --user
1730 instances of systemd, and as special case for the root user.
1732 * loginctl gained a new "--no-legend" switch to turn off output
1735 * The "sd-login.h" API gained three new calls:
1736 sd_session_is_remote(), sd_session_get_remote_user(),
1737 sd_session_get_remote_host() to query information about
1740 * The udev hardware database now also carries vendor/product
1741 information of SDIO devices.
1743 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1744 determine whether watchdog notifications are requested by
1747 * Socket-activated per-connection services now include a
1748 short description of the connection parameters in the
1751 * tmpfiles gained a new "--boot" option. When this is not used,
1752 only lines where the command character is not suffixed with
1753 "!" are executed. When this option is specified, those
1754 options are executed too. This partitions tmpfiles
1755 directives into those that can be safely executed at any
1756 time, and those which should be run only at boot (for
1757 example, a line that creates /run/nologin).
1759 * A new API "sd-resolve.h" has been added which provides a simple
1760 asynchronous wrapper around glibc NSS host name resolution
1761 calls, such as getaddrinfo(). In contrast to glibc's
1762 getaddrinfo_a(), it does not use signals. In contrast to most
1763 other asynchronous name resolution libraries, this one does
1764 not reimplement DNS, but reuses NSS, so that alternate
1765 host name resolution systems continue to work, such as mDNS,
1766 LDAP, etc. This API is based on libasyncns, but it has been
1767 cleaned up for inclusion in systemd.
1769 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1770 "sd-daemon.h" are no longer found in individual libraries
1771 libsystemd-journal.so, libsystemd-login.so,
1772 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1773 merged them into a single library, libsystemd.so, which
1774 provides all symbols. The reason for this is cyclic
1775 dependencies, as these libraries tend to use each other's
1776 symbols. So far, we have managed to workaround that by linking
1777 a copy of a good part of our code into each of these
1778 libraries again and again, which, however, makes certain
1779 things hard to do, like sharing static variables. Also, it
1780 substantially increases footprint. With this change, there
1781 is only one library for the basic APIs systemd
1782 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1783 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1784 library as well, however are subject to the --enable-kdbus
1785 switch (see below). Note that "sd-dhcp-client.h" is not part
1786 of this library (this is because it only consumes, never
1787 provides, services of/to other APIs). To make the transition
1788 easy from the separate libraries to the unified one, we
1789 provide the --enable-compat-libs compile-time switch which
1790 will generate stub libraries that are compatible with the
1791 old ones but redirect all calls to the new one.
1793 * All of the kdbus logic and the new APIs "sd-bus.h",
1794 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1795 and "sd-utf8.h" are compile-time optional via the
1796 "--enable-kdbus" switch, and they are not compiled in by
1797 default. To make use of kdbus, you have to explicitly enable
1798 the switch. Note however, that neither the kernel nor the
1799 userspace API for all of this is considered stable yet. We
1800 want to maintain the freedom to still change the APIs for
1801 now. By specifying this build-time switch, you acknowledge
1802 that you are aware of the instability of the current
1805 * Also, note that while kdbus is pretty much complete,
1806 it lacks one thing: proper policy support. This means you
1807 can build a fully working system with all features; however,
1808 it will be highly insecure. Policy support will be added in
1809 one of the next releases, at the same time that we will
1810 declare the APIs stable.
1812 * When the kernel command line argument "kdbus" is specified,
1813 systemd will automatically load the kdbus.ko kernel module. At
1814 this stage of development, it is only useful for testing kdbus
1815 and should not be used in production. Note: if "--enable-kdbus"
1816 is specified, and the kdbus.ko kernel module is available, and
1817 "kdbus" is added to the kernel command line, the entire system
1818 runs with kdbus instead of dbus-daemon, with the above mentioned
1819 problem of missing the system policy enforcement. Also a future
1820 version of kdbus.ko or a newer systemd will not be compatible with
1821 each other, and will unlikely be able to boot the machine if only
1822 one of them is updated.
1824 * systemctl gained a new "import-environment" command which
1825 uploads the caller's environment (or parts thereof) into the
1826 service manager so that it is inherited by services started
1827 by the manager. This is useful to upload variables like
1828 $DISPLAY into the user service manager.
1830 * A new PrivateDevices= switch has been added to service units
1831 which allows running a service with a namespaced /dev
1832 directory that does not contain any device nodes for
1833 physical devices. More specifically, it only includes devices
1834 such as /dev/null, /dev/urandom, and /dev/zero which are API
1837 * logind has been extended to support behaviour like VT
1838 switching on seats that do not support a VT. This makes
1839 multi-session available on seats that are not the first seat
1840 (seat0), and on systems where kernel support for VTs has
1841 been disabled at compile-time.
1843 * If a process holds a delay lock for system sleep or shutdown
1844 and fails to release it in time, we will now log its
1845 identity. This makes it easier to identify processes that
1846 cause slow suspends or power-offs.
1848 * When parsing /etc/crypttab, support for a new key-slot=
1849 option as supported by Debian is added. It allows indicating
1850 which LUKS slot to use on disk, speeding up key loading.
1852 * The sd_journald_sendv() API call has been checked and
1853 officially declared to be async-signal-safe so that it may
1854 be invoked from signal handlers for logging purposes.
1856 * Boot-time status output is now enabled automatically after a
1857 short timeout if boot does not progress, in order to give
1858 the user an indication what she or he is waiting for.
1860 * The boot-time output has been improved to show how much time
1861 remains until jobs expire.
1863 * The KillMode= switch in service units gained a new possible
1864 value "mixed". If set, and the unit is shut down, then the
1865 initial SIGTERM signal is sent only to the main daemon
1866 process, while the following SIGKILL signal is sent to
1867 all remaining processes of the service.
1869 * When a scope unit is registered, a new property "Controller"
1870 may be set. If set to a valid bus name, systemd will send a
1871 RequestStop() signal to this name when it would like to shut
1872 down the scope. This may be used to hook manager logic into
1873 the shutdown logic of scope units. Also, scope units may now
1874 be put in a special "abandoned" state, in which case the
1875 manager process which created them takes no further
1876 responsibilities for it.
1878 * When reading unit files, systemd will now verify
1879 the access mode of these files, and warn about certain
1880 suspicious combinations. This has been added to make it
1881 easier to track down packaging bugs where unit files are
1882 marked executable or world-writable.
1884 * systemd-nspawn gained a new "--setenv=" switch to set
1885 container-wide environment variables. The similar option in
1886 systemd-activate was renamed from "--environment=" to
1887 "--setenv=" for consistency.
1889 * systemd-nspawn has been updated to create a new kdbus domain
1890 for each container that is invoked, thus allowing each
1891 container to have its own set of system and user buses,
1892 independent of the host.
1894 * systemd-nspawn gained a new --drop-capability= switch to run
1895 the container with less capabilities than the default. Both
1896 --drop-capability= and --capability= now take the special
1897 string "all" for dropping or keeping all capabilities.
1899 * systemd-nspawn gained new switches for executing containers
1900 with specific SELinux labels set.
1902 * systemd-nspawn gained a new --quiet switch to not generate
1903 any additional output but the container's own console
1906 * systemd-nspawn gained a new --share-system switch to run a
1907 container without PID namespacing enabled.
1909 * systemd-nspawn gained a new --register= switch to control
1910 whether the container is registered with systemd-machined or
1911 not. This is useful for containers that do not run full
1912 OS images, but only specific apps.
1914 * systemd-nspawn gained a new --keep-unit which may be used
1915 when invoked as the only program from a service unit, and
1916 results in registration of the unit service itself in
1917 systemd-machined, instead of a newly opened scope unit.
1919 * systemd-nspawn gained a new --network-interface= switch for
1920 moving arbitrary interfaces to the container. The new
1921 --network-veth switch creates a virtual Ethernet connection
1922 between host and container. The new --network-bridge=
1923 switch then allows assigning the host side of this virtual
1924 Ethernet connection to a bridge device.
1926 * systemd-nspawn gained a new --personality= switch for
1927 setting the kernel personality for the container. This is
1928 useful when running a 32-bit container on a 64-bit host. A
1929 similar option Personality= is now also available for service
1932 * logind will now also track a "Desktop" identifier for each
1933 session which encodes the desktop environment of it. This is
1934 useful for desktop environments that want to identify
1935 multiple running sessions of itself easily.
1937 * A new SELinuxContext= setting for service units has been
1938 added that allows setting a specific SELinux execution
1939 context for a service.
1941 * Most systemd client tools will now honour $SYSTEMD_LESS for
1942 settings of the "less" pager. By default, these tools will
1943 override $LESS to allow certain operations to work, such as
1944 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1945 influence this logic.
1947 * systemd's "seccomp" hook-up has been changed to make use of
1948 the libseccomp library instead of using its own
1949 implementation. This has benefits for portability among
1952 * For usage together with SystemCallFilter=, a new
1953 SystemCallErrorNumber= setting has been introduced that
1954 allows configuration of a system error number to be returned
1955 on filtered system calls, instead of immediately killing the
1956 process. Also, SystemCallArchitectures= has been added to
1957 limit access to system calls of a particular architecture
1958 (in order to turn off support for unused secondary
1959 architectures). There is also a global
1960 SystemCallArchitectures= setting in system.conf now to turn
1961 off support for non-native system calls system-wide.
1963 * systemd requires a kernel with a working name_to_handle_at(),
1964 please see the kernel config requirements in the README file.
1966 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1967 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1968 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1969 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1970 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1971 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1972 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1973 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1974 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1975 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1976 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1977 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1978 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1979 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1980 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1981 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1982 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1983 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1984 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1985 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1986 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1987 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1988 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1989 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1991 -- Berlin, 2014-02-20
1995 * logind has gained support for facilitating privileged input
1996 and drm device access for unprivileged clients. This work is
1997 useful to allow Wayland display servers (and similar
1998 programs, such as kmscon) to run under the user's ID and
1999 access input and drm devices which are normally
2000 protected. When this is used (and the kernel is new enough)
2001 logind will "mute" IO on the file descriptors passed to
2002 Wayland as long as it is in the background and "unmute" it
2003 if it returns into the foreground. This allows secure
2004 session switching without allowing background sessions to
2005 eavesdrop on input and display data. This also introduces
2006 session switching support if VT support is turned off in the
2007 kernel, and on seats that are not seat0.
2009 * A new kernel command line option luks.options= is understood
2010 now which allows specifying LUKS options for usage for LUKS
2011 encrypted partitions specified with luks.uuid=.
2013 * tmpfiles.d(5) snippets may now use specifier expansion in
2014 path names. More specifically %m, %b, %H, %v, are now
2015 replaced by the local machine id, boot id, hostname, and
2016 kernel version number.
2018 * A new tmpfiles.d(5) command "m" has been introduced which
2019 may be used to change the owner/group/access mode of a file
2020 or directory if it exists, but do nothing if it does not.
2022 * This release removes high-level support for the
2023 MemorySoftLimit= cgroup setting. The underlying kernel
2024 cgroup attribute memory.soft_limit= is currently badly
2025 designed and likely to be removed from the kernel API in its
2026 current form, hence we should not expose it for now.
2028 * The memory.use_hierarchy cgroup attribute is now enabled for
2029 all cgroups systemd creates in the memory cgroup
2030 hierarchy. This option is likely to be come the built-in
2031 default in the kernel anyway, and the non-hierarchial mode
2032 never made much sense in the intrinsically hierarchial
2035 * A new field _SYSTEMD_SLICE= is logged along with all journal
2036 messages containing the slice a message was generated
2037 from. This is useful to allow easy per-customer filtering of
2038 logs among other things.
2040 * systemd-journald will no longer adjust the group of journal
2041 files it creates to the "systemd-journal" group. Instead we
2042 rely on the journal directory to be owned by the
2043 "systemd-journal" group, and its setgid bit set, so that the
2044 kernel file system layer will automatically enforce that
2045 journal files inherit this group assignment. The reason for
2046 this change is that we cannot allow NSS look-ups from
2047 journald which would be necessary to resolve
2048 "systemd-journal" to a numeric GID, because this might
2049 create deadlocks if NSS involves synchronous queries to
2050 other daemons (such as nscd, or sssd) which in turn are
2051 logging clients of journald and might block on it, which
2052 would then dead lock. A tmpfiles.d(5) snippet included in
2053 systemd will make sure the setgid bit and group are
2054 properly set on the journal directory if it exists on every
2055 boot. However, we recommend adjusting it manually after
2056 upgrades too (or from RPM scriptlets), so that the change is
2057 not delayed until next reboot.
2059 * Backlight and random seed files in /var/lib/ have moved into
2060 the /var/lib/systemd/ directory, in order to centralize all
2061 systemd generated files in one directory.
2063 * Boot time performance measurements (as displayed by
2064 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
2065 performance information if that's available to determine how
2066 much time BIOS and boot loader initialization required. With
2067 a sufficiently new BIOS you hence no longer need to boot
2068 with Gummiboot to get access to such information.
2070 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
2071 Cristian Rodríguez, Dave Reisner, David Herrmann, David
2072 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
2073 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
2074 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
2075 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
2076 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2078 -- Berlin, 2013-10-02
2082 * The Restart= option for services now understands a new
2083 on-watchdog setting, which will restart the service
2084 automatically if the service stops sending out watchdog keep
2085 alive messages (as configured with WatchdogSec=).
2087 * The getty generator (which is responsible for bringing up a
2088 getty on configured serial consoles) will no longer only
2089 start a getty on the primary kernel console but on all
2090 others, too. This makes the order in which console= is
2091 specified on the kernel command line less important.
2093 * libsystemd-logind gained a new sd_session_get_vt() call to
2094 retrieve the VT number of a session.
2096 * If the option "tries=0" is set for an entry of /etc/crypttab
2097 its passphrase is queried indefinitely instead of any
2098 maximum number of tries.
2100 * If a service with a configure PID file terminates its PID
2101 file will now be removed automatically if it still exists
2102 afterwards. This should put an end to stale PID files.
2104 * systemd-run will now also take relative binary path names
2105 for execution and no longer insists on absolute paths.
2107 * InaccessibleDirectories= and ReadOnlyDirectories= now take
2108 paths that are optionally prefixed with "-" to indicate that
2109 it should not be considered a failure if they do not exist.
2111 * journalctl -o (and similar commands) now understands a new
2112 output mode "short-precise", it is similar to "short" but
2113 shows timestamps with usec accuracy.
2115 * The option "discard" (as known from Debian) is now
2116 synonymous to "allow-discards" in /etc/crypttab. In fact,
2117 "discard" is preferred now (since it is easier to remember
2120 * Some licensing clean-ups were made, so that more code is now
2121 LGPL-2.1 licensed than before.
2123 * A minimal tool to save/restore the display backlight
2124 brightness across reboots has been added. It will store the
2125 backlight setting as late as possible at shutdown, and
2126 restore it as early as possible during reboot.
2128 * A logic to automatically discover and enable home and swap
2129 partitions on GPT disks has been added. With this in place
2130 /etc/fstab becomes optional for many setups as systemd can
2131 discover certain partitions located on the root disk
2132 automatically. Home partitions are recognized under their
2133 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
2134 partitions are recognized under their GPT type ID
2135 0657fd6da4ab43c484e50933c84b4f4f.
2137 * systemd will no longer pass any environment from the kernel
2138 or initrd to system services. If you want to set an
2139 environment for all services, do so via the kernel command
2140 line systemd.setenv= assignment.
2142 * The systemd-sysctl tool no longer natively reads the file
2143 /etc/sysctl.conf. If desired, the file should be symlinked
2144 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
2145 legacy support by a symlink rather than built-in code, it
2146 also makes the otherwise hidden order of application of the
2147 different files visible. (Note that this partly reverts to a
2148 pre-198 application order of sysctl knobs!)
2150 * The "systemctl set-log-level" and "systemctl dump" commands
2151 have been moved to systemd-analyze.
2153 * systemd-run learned the new --remain-after-exit switch,
2154 which causes the scope unit not to be cleaned up
2155 automatically after the process terminated.
2157 * tmpfiles learned a new --exclude-prefix= switch to exclude
2158 certain paths from operation.
2160 * journald will now automatically flush all messages to disk
2161 as soon as a message at the log level CRIT, ALERT or EMERG
2164 Contributions from: Andrew Cook, Brandon Philips, Christian
2165 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
2166 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
2167 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
2168 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
2169 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
2170 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
2171 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
2172 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
2173 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
2174 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
2175 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
2176 William Giokas, Zbigniew Jędrzejewski-Szmek
2178 -- Berlin, 2013-09-13
2182 * The documentation has been updated to cover the various new
2183 concepts introduced with 205.
2185 * Unit files now understand the new %v specifier which
2186 resolves to the kernel version string as returned by "uname
2189 * systemctl now supports filtering the unit list output by
2190 load state, active state and sub state, using the new
2193 * "systemctl status" will now show the results of the
2194 condition checks (like ConditionPathExists= and similar) of
2195 the last start attempts of the unit. They are also logged to
2198 * "journalctl -b" may now be used to look for boot output of a
2199 specific boot. Try "journalctl -b -1" for the previous boot,
2200 but the syntax is substantially more powerful.
2202 * "journalctl --show-cursor" has been added which prints the
2203 cursor string the last shown log line. This may then be used
2204 with the new "journalctl --after-cursor=" switch to continue
2205 browsing logs from that point on.
2207 * "journalctl --force" may now be used to force regeneration
2210 * Creation of "dead" device nodes has been moved from udev
2211 into kmod and tmpfiles. Previously, udev would read the kmod
2212 databases to pre-generate dead device nodes based on meta
2213 information contained in kernel modules, so that these would
2214 be auto-loaded on access rather then at boot. As this
2215 does not really have much to do with the exposing actual
2216 kernel devices to userspace this has always been slightly
2217 alien in the udev codebase. Following the new scheme kmod
2218 will now generate a runtime snippet for tmpfiles from the
2219 module meta information and it now is tmpfiles' job to the
2220 create the nodes. This also allows overriding access and
2221 other parameters for the nodes using the usual tmpfiles
2222 facilities. As side effect this allows us to remove the
2223 CAP_SYS_MKNOD capability bit from udevd entirely.
2225 * logind's device ACLs may now be applied to these "dead"
2226 devices nodes too, thus finally allowing managed access to
2227 devices such as /dev/snd/sequencer whithout loading the
2228 backing module right-away.
2230 * A new RPM macro has been added that may be used to apply
2231 tmpfiles configuration during package installation.
2233 * systemd-detect-virt and ConditionVirtualization= now can
2234 detect User-Mode-Linux machines (UML).
2236 * journald will now implicitly log the effective capabilities
2237 set of processes in the message metadata.
2239 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2241 * The initrd interface has been simplified (more specifically,
2242 support for passing performance data via environment
2243 variables and fsck results via files in /run has been
2244 removed). These features were non-essential, and are
2245 nowadays available in a much nicer way by having systemd in
2246 the initrd serialize its state and have the hosts systemd
2247 deserialize it again.
2249 * The udev "keymap" data files and tools to apply keyboard
2250 specific mappings of scan to key codes, and force-release
2251 scan code lists have been entirely replaced by a udev
2252 "keyboard" builtin and a hwdb data file.
2254 * systemd will now honour the kernel's "quiet" command line
2255 argument also during late shutdown, resulting in a
2256 completely silent shutdown when used.
2258 * There's now an option to control the SO_REUSEPORT socket
2259 option in .socket units.
2261 * Instance units will now automatically get a per-template
2262 subslice of system.slice unless something else is explicitly
2263 configured. For example, instances of sshd@.service will now
2264 implicitly be placed in system-sshd.slice rather than
2265 system.slice as before.
2267 * Test coverage support may now be enabled at build time.
2269 Contributions from: Dave Reisner, Frederic Crozat, Harald
2270 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2271 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2272 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2273 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2274 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2275 Giokas, Zbigniew Jędrzejewski-Szmek
2277 -- Berlin, 2013-07-23
2281 * Two new unit types have been introduced:
2283 Scope units are very similar to service units, however, are
2284 created out of pre-existing processes -- instead of PID 1
2285 forking off the processes. By using scope units it is
2286 possible for system services and applications to group their
2287 own child processes (worker processes) in a powerful way
2288 which then maybe used to organize them, or kill them
2289 together, or apply resource limits on them.
2291 Slice units may be used to partition system resources in an
2292 hierarchial fashion and then assign other units to them. By
2293 default there are now three slices: system.slice (for all
2294 system services), user.slice (for all user sessions),
2295 machine.slice (for VMs and containers).
2297 Slices and scopes have been introduced primarily in
2298 context of the work to move cgroup handling to a
2299 single-writer scheme, where only PID 1
2300 creates/removes/manages cgroups.
2302 * There's a new concept of "transient" units. In contrast to
2303 normal units these units are created via an API at runtime,
2304 not from configuration from disk. More specifically this
2305 means it is now possible to run arbitrary programs as
2306 independent services, with all execution parameters passed
2307 in via bus APIs rather than read from disk. Transient units
2308 make systemd substantially more dynamic then it ever was,
2309 and useful as a general batch manager.
2311 * logind has been updated to make use of scope and slice units
2312 for managing user sessions. As a user logs in he will get
2313 his own private slice unit, to which all sessions are added
2314 as scope units. We also added support for automatically
2315 adding an instance of user@.service for the user into the
2316 slice. Effectively logind will no longer create cgroup
2317 hierarchies on its own now, it will defer entirely to PID 1
2318 for this by means of scope, service and slice units. Since
2319 user sessions this way become entities managed by PID 1
2320 the output of "systemctl" is now a lot more comprehensive.
2322 * A new mini-daemon "systemd-machined" has been added which
2323 may be used by virtualization managers to register local
2324 VMs/containers. nspawn has been updated accordingly, and
2325 libvirt will be updated shortly. machined will collect a bit
2326 of meta information about the VMs/containers, and assign
2327 them their own scope unit (see above). The collected
2328 meta-data is then made available via the "machinectl" tool,
2329 and exposed in "ps" and similar tools. machined/machinectl
2330 is compile-time optional.
2332 * As discussed earlier, the low-level cgroup configuration
2333 options ControlGroup=, ControlGroupModify=,
2334 ControlGroupPersistent=, ControlGroupAttribute= have been
2335 removed. Please use high-level attribute settings instead as
2336 well as slice units.
2338 * A new bus call SetUnitProperties() has been added to alter
2339 various runtime parameters of a unit. This is primarily
2340 useful to alter cgroup parameters dynamically in a nice way,
2341 but will be extended later on to make more properties
2342 modifiable at runtime. systemctl gained a new set-properties
2343 command that wraps this call.
2345 * A new tool "systemd-run" has been added which can be used to
2346 run arbitrary command lines as transient services or scopes,
2347 while configuring a number of settings via the command
2348 line. This tool is currently very basic, however already
2349 very useful. We plan to extend this tool to even allow
2350 queuing of execution jobs with time triggers from the
2351 command line, similar in fashion to "at".
2353 * nspawn will now inform the user explicitly that kernels with
2354 audit enabled break containers, and suggest the user to turn
2357 * Support for detecting the IMA and AppArmor security
2358 frameworks with ConditionSecurity= has been added.
2360 * journalctl gained a new "-k" switch for showing only kernel
2361 messages, mimicking dmesg output; in addition to "--user"
2362 and "--system" switches for showing only user's own logs
2365 * systemd-delta can now show information about drop-in
2366 snippets extending unit files.
2368 * libsystemd-bus has been substantially updated but is still
2369 not available as public API.
2371 * systemd will now look for the "debug" argument on the kernel
2372 command line and enable debug logging, similar to what
2373 "systemd.log_level=debug" already did before.
2375 * "systemctl set-default", "systemctl get-default" has been
2376 added to configure the default.target symlink, which
2377 controls what to boot into by default.
2379 * "systemctl set-log-level" has been added as a convenient
2380 way to raise and lower systemd logging threshold.
2382 * "systemd-analyze plot" will now show the time the various
2383 generators needed for execution, as well as information
2384 about the unit file loading.
2386 * libsystemd-journal gained a new sd_journal_open_files() call
2387 for opening specific journal files. journactl also gained a
2388 new switch to expose this new functionality. Previously we
2389 only supported opening all files from a directory, or all
2390 files from the system, as opening individual files only is
2391 racy due to journal file rotation.
2393 * systemd gained the new DefaultEnvironment= setting in
2394 /etc/systemd/system.conf to set environment variables for
2397 * If a privileged process logs a journal message with the
2398 OBJECT_PID= field set, then journald will automatically
2399 augment this with additional OBJECT_UID=, OBJECT_GID=,
2400 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2401 system services want to log events about specific client
2402 processes. journactl/systemctl has been updated to make use
2403 of this information if all log messages regarding a specific
2406 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2407 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2408 Reisner, David Coppa, David King, David Strauss, Eelco
2409 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2410 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2411 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2412 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2413 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2414 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2415 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2416 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2417 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2418 Łukasz Stelmach, 장동준
2422 * The Python bindings gained some minimal support for the APIs
2423 exposed by libsystemd-logind.
2425 * ConditionSecurity= gained support for detecting SMACK. Since
2426 this condition already supports SELinux and AppArmor we only
2427 miss IMA for this. Patches welcome!
2429 Contributions from: Karol Lewandowski, Lennart Poettering,
2430 Zbigniew Jędrzejewski-Szmek
2434 * systemd-nspawn will now create /etc/resolv.conf if
2435 necessary, before bind-mounting the host's file onto it.
2437 * systemd-nspawn will now store meta information about a
2438 container on the container's cgroup as extended attribute
2439 fields, including the root directory.
2441 * The cgroup hierarchy has been reworked in many ways. All
2442 objects any of the components systemd creates in the cgroup
2443 tree are now suffixed. More specifically, user sessions are
2444 now placed in cgroups suffixed with ".session", users in
2445 cgroups suffixed with ".user", and nspawn containers in
2446 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2447 names are now escaped in a simple scheme to avoid collision
2448 of userspace object names with kernel filenames. This work
2449 is preparation for making these objects relocatable in the
2450 cgroup tree, in order to allow easy resource partitioning of
2451 these objects without causing naming conflicts.
2453 * systemctl list-dependencies gained the new switches
2454 --plain, --reverse, --after and --before.
2456 * systemd-inhibit now shows the process name of processes that
2457 have taken an inhibitor lock.
2459 * nss-myhostname will now also resolve "localhost"
2460 implicitly. This makes /etc/hosts an optional file and
2461 nicely handles that on IPv6 ::1 maps to both "localhost" and
2464 * libsystemd-logind.so gained a new call
2465 sd_get_machine_names() to enumerate running containers and
2466 VMs (currently only supported by very new libvirt and
2467 nspawn). sd_login_monitor can now be used to watch
2468 VMs/containers coming and going.
2470 * .include is not allowed recursively anymore, and only in
2471 unit files. Usually it is better to use drop-in snippets in
2472 .d/*.conf anyway, as introduced with systemd 198.
2474 * systemd-analyze gained a new "critical-chain" command that
2475 determines the slowest chain of units run during system
2476 boot-up. It is very useful for tracking down where
2477 optimizing boot time is the most beneficial.
2479 * systemd will no longer allow manipulating service paths in
2480 the name=systemd:/system cgroup tree using ControlGroup= in
2481 units. (But is still fine with it in all other dirs.)
2483 * There's a new systemd-nspawn@.service service file that may
2484 be used to easily run nspawn containers as system
2485 services. With the container's root directory in
2486 /var/lib/container/foobar it is now sufficient to run
2487 "systemctl start systemd-nspawn@foobar.service" to boot it.
2489 * systemd-cgls gained a new parameter "--machine" to list only
2490 the processes within a certain container.
2492 * ConditionSecurity= now can check for "apparmor". We still
2493 are lacking checks for SMACK and IMA for this condition
2494 check though. Patches welcome!
2496 * A new configuration file /etc/systemd/sleep.conf has been
2497 added that may be used to configure which kernel operation
2498 systemd is supposed to execute when "suspend", "hibernate"
2499 or "hybrid-sleep" is requested. This makes the new kernel
2500 "freeze" state accessible to the user.
2502 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2503 the passed argument if applicable.
2505 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2506 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2507 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2508 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2509 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2510 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2511 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2516 * The output of 'systemctl list-jobs' got some polishing. The
2517 '--type=' argument may now be passed more than once. A new
2518 command 'systemctl list-sockets' has been added which shows
2519 a list of kernel sockets systemd is listening on with the
2520 socket units they belong to, plus the units these socket
2523 * The experimental libsystemd-bus library got substantial
2524 updates to work in conjunction with the (also experimental)
2525 kdbus kernel project. It works well enough to exchange
2526 messages with some sophistication. Note that kdbus is not
2527 ready yet, and the library is mostly an elaborate test case
2528 for now, and not installable.
2530 * systemd gained a new unit 'systemd-static-nodes.service'
2531 that generates static device nodes earlier during boot, and
2532 can run in conjunction with udev.
2534 * libsystemd-login gained a new call sd_pid_get_user_unit()
2535 to retrieve the user systemd unit a process is running
2536 in. This is useful for systems where systemd is used as
2539 * systemd-nspawn now places all containers in the new /machine
2540 top-level cgroup directory in the name=systemd
2541 hierarchy. libvirt will soon do the same, so that we get a
2542 uniform separation of /system, /user and /machine for system
2543 services, user processes and containers/virtual
2544 machines. This new cgroup hierarchy is also useful to stick
2545 stable names to specific container instances, which can be
2546 recognized later this way (this name may be controlled
2547 via systemd-nspawn's new -M switch). libsystemd-login also
2548 gained a new call sd_pid_get_machine_name() to retrieve the
2549 name of the container/VM a specific process belongs to.
2551 * bootchart can now store its data in the journal.
2553 * libsystemd-journal gained a new call
2554 sd_journal_add_conjunction() for AND expressions to the
2555 matching logic. This can be used to express more complex
2556 logical expressions.
2558 * journactl can now take multiple --unit= and --user-unit=
2561 * The cryptsetup logic now understands the "luks.key=" kernel
2562 command line switch for specifying a file to read the
2563 decryption key from. Also, if a configured key file is not
2564 found the tool will now automatically fall back to prompting
2567 * Python systemd.journal module was updated to wrap recently
2568 added functions from libsystemd-journal. The interface was
2569 changed to bring the low level interface in s.j._Reader
2570 closer to the C API, and the high level interface in
2571 s.j.Reader was updated to wrap and convert all data about
2574 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2575 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2576 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2577 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2578 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2579 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2583 * journalctl --update-catalog now understands a new --root=
2584 option to operate on catalogs found in a different root
2587 * During shutdown after systemd has terminated all running
2588 services a final killing loop kills all remaining left-over
2589 processes. We will now print the name of these processes
2590 when we send SIGKILL to them, since this usually indicates a
2593 * If /etc/crypttab refers to password files stored on
2594 configured mount points automatic dependencies will now be
2595 generated to ensure the specific mount is established first
2596 before the key file is attempted to be read.
2598 * 'systemctl status' will now show information about the
2599 network sockets a socket unit is listening on.
2601 * 'systemctl status' will also shown information about any
2602 drop-in configuration file for units. (Drop-In configuration
2603 files in this context are files such as
2604 /etc/systemd/systemd/foobar.service.d/*.conf)
2606 * systemd-cgtop now optionally shows summed up CPU times of
2607 cgroups. Press '%' while running cgtop to switch between
2608 percentage and absolute mode. This is useful to determine
2609 which cgroups use up the most CPU time over the entire
2610 runtime of the system. systemd-cgtop has also been updated
2611 to be 'pipeable' for processing with further shell tools.
2613 * 'hostnamectl set-hostname' will now allow setting of FQDN
2616 * The formatting and parsing of time span values has been
2617 changed. The parser now understands fractional expressions
2618 such as "5.5h". The formatter will now output fractional
2619 expressions for all time spans under 1min, i.e. "5.123456s"
2620 rather than "5s 123ms 456us". For time spans under 1s
2621 millisecond values are shown, for those under 1ms
2622 microsecond values are shown. This should greatly improve
2623 all time-related output of systemd.
2625 * libsystemd-login and libsystemd-journal gained new
2626 functions for querying the poll() events mask and poll()
2627 timeout value for integration into arbitrary event
2630 * localectl gained the ability to list available X11 keymaps
2631 (models, layouts, variants, options).
2633 * 'systemd-analyze dot' gained the ability to filter for
2634 specific units via shell-style globs, to create smaller,
2635 more useful graphs. I.e. it is now possible to create simple
2636 graphs of all the dependencies between only target units, or
2637 of all units that Avahi has dependencies with.
2639 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2640 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2641 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2642 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2643 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2644 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2645 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2649 * The boot-time readahead implementation for rotating media
2650 will now read the read-ahead data in multiple passes which
2651 consist of all read requests made in equidistant time
2652 intervals. This means instead of strictly reading read-ahead
2653 data in its physical order on disk we now try to find a
2654 middle ground between physical and access time order.
2656 * /etc/os-release files gained a new BUILD_ID= field for usage
2657 on operating systems that provide continuous builds of OS
2660 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2661 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2662 William Douglas, Zbigniew Jędrzejewski-Szmek
2666 * systemd-python gained an API exposing libsystemd-daemon.
2668 * The SMACK setup logic gained support for uploading CIPSO
2671 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2672 ReadOnlyDirectories= and InaccessibleDirectories= has
2673 changed. The private /tmp and /var/tmp directories are now
2674 shared by all processes of a service (which means
2675 ExecStartPre= may now leave data in /tmp that ExecStart= of
2676 the same service can still access). When a service is
2677 stopped its temporary directories are immediately deleted
2678 (normal clean-up with tmpfiles is still done in addition to
2681 * By default, systemd will now set a couple of sysctl
2682 variables in the kernel: the safe sysrq options are turned
2683 on, IP route verification is turned on, and source routing
2684 disabled. The recently added hardlink and softlink
2685 protection of the kernel is turned on. These settings should
2686 be reasonably safe, and good defaults for all new systems.
2688 * The predictable network naming logic may now be turned off
2689 with a new kernel command line switch: net.ifnames=0.
2691 * A new libsystemd-bus module has been added that implements a
2692 pretty complete D-Bus client library. For details see:
2694 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2696 * journald will now explicitly flush the journal files to disk
2697 at the latest 5min after each write. The file will then also
2698 be marked offline until the next write. This should increase
2699 reliability in case of a crash. The synchronization delay
2700 can be configured via SyncIntervalSec= in journald.conf.
2702 * There's a new remote-fs-setup.target unit that can be used
2703 to pull in specific services when at least one remote file
2704 system is to be mounted.
2706 * There are new targets timers.target and paths.target as
2707 canonical targets to pull user timer and path units in
2708 from. This complements sockets.target with a similar
2709 purpose for socket units.
2711 * libudev gained a new call udev_device_set_attribute_value()
2712 to set sysfs attributes of a device.
2714 * The udev daemon now sets the default number of worker
2715 processes executed in parallel based on the number of available
2716 CPUs instead of the amount of available RAM. This is supposed
2717 to provide a more reliable default and limit a too aggressive
2718 paralellism for setups with 1000s of devices connected.
2720 Contributions from: Auke Kok, Colin Walters, Cristian
2721 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2722 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2723 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2724 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2725 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2726 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2727 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2728 Zbigniew Jędrzejewski-Szmek
2732 * Configuration of unit files may now be extended via drop-in
2733 files without having to edit/override the unit files
2734 themselves. More specifically, if the administrator wants to
2735 change one value for a service file foobar.service he can
2736 now do so by dropping in a configuration snippet into
2737 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2738 will load all these snippets and apply them on top of the
2739 main unit configuration file, possibly extending or
2740 overriding its settings. Using these drop-in snippets is
2741 generally nicer than the two earlier options for changing
2742 unit files locally: copying the files from
2743 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2744 them there; or creating a new file in /etc/systemd/system/
2745 that incorporates the original one via ".include". Drop-in
2746 snippets into these .d/ directories can be placed in any
2747 directory systemd looks for units in, and the usual
2748 overriding semantics between /usr/lib, /etc and /run apply
2751 * Most unit file settings which take lists of items can now be
2752 reset by assigning the empty string to them. For example,
2753 normally, settings such as Environment=FOO=BAR append a new
2754 environment variable assignment to the environment block,
2755 each time they are used. By assigning Environment= the empty
2756 string the environment block can be reset to empty. This is
2757 particularly useful with the .d/*.conf drop-in snippets
2758 mentioned above, since this adds the ability to reset list
2759 settings from vendor unit files via these drop-ins.
2761 * systemctl gained a new "list-dependencies" command for
2762 listing the dependencies of a unit recursively.
2764 * Inhibitors are now honored and listed by "systemctl
2765 suspend", "systemctl poweroff" (and similar) too, not only
2766 GNOME. These commands will also list active sessions by
2769 * Resource limits (as exposed by the various control group
2770 controllers) can now be controlled dynamically at runtime
2771 for all units. More specifically, you can now use a command
2772 like "systemctl set-cgroup-attr foobar.service cpu.shares
2773 2000" to alter the CPU shares a specific service gets. These
2774 settings are stored persistently on disk, and thus allow the
2775 administrator to easily adjust the resource usage of
2776 services with a few simple commands. This dynamic resource
2777 management logic is also available to other programs via the
2778 bus. Almost any kernel cgroup attribute and controller is
2781 * systemd-vconsole-setup will now copy all font settings to
2782 all allocated VTs, where it previously applied them only to
2785 * libsystemd-login gained the new sd_session_get_tty() API
2788 * This release drops support for a few legacy or
2789 distribution-specific LSB facility names when parsing init
2790 scripts: $x-display-manager, $mail-transfer-agent,
2791 $mail-transport-agent, $mail-transfer-agent, $smtp,
2792 $null. Also, the mail-transfer-agent.target unit backing
2793 this has been removed. Distributions which want to retain
2794 compatibility with this should carry the burden for
2795 supporting this themselves and patch support for these back
2796 in, if they really need to. Also, the facilities $syslog and
2797 $local_fs are now ignored, since systemd does not support
2798 early-boot LSB init scripts anymore, and these facilities
2799 are implied anyway for normal services. syslog.target has
2802 * There are new bus calls on PID1's Manager object for
2803 cancelling jobs, and removing snapshot units. Previously,
2804 both calls were only available on the Job and Snapshot
2807 * systemd-journal-gatewayd gained SSL support.
2809 * The various "environment" files, such as /etc/locale.conf
2810 now support continuation lines with a backslash ("\") as
2811 last character in the line, similarly in style (but different)
2812 to how this is supported in shells.
2814 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2815 now implicitly appended to every log entry logged. systemctl
2816 has been updated to filter by this field when operating on a
2817 user systemd instance.
2819 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2820 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2821 the container. This makes it easier to boot unmodified
2822 Fedora systems in a container, which however still requires
2823 audit=0 to be passed on the kernel command line. Auditing in
2824 kernel and userspace is unfortunately still too broken in
2825 context of containers, hence we recommend compiling it out
2826 of the kernel or using audit=0. Hopefully this will be fixed
2827 one day for good in the kernel.
2829 * nspawn gained the new --bind= and --bind-ro= parameters to
2830 bind mount specific directories from the host into the
2833 * nspawn will now mount its own devpts file system instance
2834 into the container, in order not to leak pty devices from
2835 the host into the container.
2837 * systemd will now read the firmware boot time performance
2838 information from the EFI variables, if the used boot loader
2839 supports this, and takes it into account for boot performance
2840 analysis via "systemd-analyze". This is currently supported
2841 only in conjunction with Gummiboot, but could be supported
2842 by other boot loaders too. For details see:
2844 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2846 * A new generator has been added that automatically mounts the
2847 EFI System Partition (ESP) to /boot, if that directory
2848 exists, is empty, and no other file system has been
2849 configured to be mounted there.
2851 * logind will now send out PrepareForSleep(false) out
2852 unconditionally, after coming back from suspend. This may be
2853 used by applications as asynchronous notification for
2854 system resume events.
2856 * "systemctl unlock-sessions" has been added, that allows
2857 unlocking the screens of all user sessions at once, similar
2858 to how "systemctl lock-sessions" already locked all users
2859 sessions. This is backed by a new D-Bus call UnlockSessions().
2861 * "loginctl seat-status" will now show the master device of a
2862 seat. (i.e. the device of a seat that needs to be around for
2863 the seat to be considered available, usually the graphics
2866 * tmpfiles gained a new "X" line type, that allows
2867 configuration of files and directories (with wildcards) that
2868 shall be excluded from automatic cleanup ("aging").
2870 * udev default rules set the device node permissions now only
2871 at "add" events, and do not change them any longer with a
2872 later "change" event.
2874 * The log messages for lid events and power/sleep keypresses
2875 now carry a message ID.
2877 * We now have a substantially larger unit test suite, but this
2878 continues to be work in progress.
2880 * udevadm hwdb gained a new --root= parameter to change the
2881 root directory to operate relative to.
2883 * logind will now issue a background sync() request to the kernel
2884 early at shutdown, so that dirty buffers are flushed to disk early
2885 instead of at the last moment, in order to optimize shutdown
2888 * A new bootctl tool has been added that is an interface for
2889 certain boot loader operations. This is currently a preview
2890 and is likely to be extended into a small mechanism daemon
2891 like timedated, localed, hostnamed, and can be used by
2892 graphical UIs to enumerate available boot options, and
2893 request boot into firmware operations.
2895 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2896 the rest of the package. It also has been updated to work
2897 correctly in initrds.
2899 * Policykit previously has been runtime optional, and is now
2900 also compile time optional via a configure switch.
2902 * systemd-analyze has been reimplemented in C. Also "systemctl
2903 dot" has moved into systemd-analyze.
2905 * "systemctl status" with no further parameters will now print
2906 the status of all active or failed units.
2908 * Operations such as "systemctl start" can now be executed
2909 with a new mode "--irreversible" which may be used to queue
2910 operations that cannot accidentally be reversed by a later
2911 job queuing. This is by default used to make shutdown
2912 requests more robust.
2914 * The Python API of systemd now gained a new module for
2915 reading journal files.
2917 * A new tool kernel-install has been added that can install
2918 kernel images according to the Boot Loader Specification:
2920 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2922 * Boot time console output has been improved to provide
2923 animated boot time output for hanging jobs.
2925 * A new tool systemd-activate has been added which can be used
2926 to test socket activation with, directly from the command
2927 line. This should make it much easier to test and debug
2928 socket activation in daemons.
2930 * journalctl gained a new "--reverse" (or -r) option to show
2931 journal output in reverse order (i.e. newest line first).
2933 * journalctl gained a new "--pager-end" (or -e) option to jump
2934 to immediately jump to the end of the journal in the
2935 pager. This is only supported in conjunction with "less".
2937 * journalctl gained a new "--user-unit=" option, that works
2938 similarly to "--unit=" but filters for user units rather than
2941 * A number of unit files to ease adoption of systemd in
2942 initrds has been added. This moves some minimal logic from
2943 the various initrd implementations into systemd proper.
2945 * The journal files are now owned by a new group
2946 "systemd-journal", which exists specifically to allow access
2947 to the journal, and nothing else. Previously, we used the
2948 "adm" group for that, which however possibly covers more
2949 than just journal/log file access. This new group is now
2950 already used by systemd-journal-gatewayd to ensure this
2951 daemon gets access to the journal files and as little else
2952 as possible. Note that "make install" will also set FS ACLs
2953 up for /var/log/journal to give "adm" and "wheel" read
2954 access to it, in addition to "systemd-journal" which owns
2955 the journal files. We recommend that packaging scripts also
2956 add read access to "adm" + "wheel" to /var/log/journal, and
2957 all existing/future journal files. To normal users and
2958 administrators little changes, however packagers need to
2959 ensure to create the "systemd-journal" system group at
2960 package installation time.
2962 * The systemd-journal-gatewayd now runs as unprivileged user
2963 systemd-journal-gateway:systemd-journal-gateway. Packaging
2964 scripts need to create these system user/group at
2967 * timedated now exposes a new boolean property CanNTP that
2968 indicates whether a local NTP service is available or not.
2970 * systemd-detect-virt will now also detect xen PVs
2972 * The pstore file system is now mounted by default, if it is
2975 * In addition to the SELinux and IMA policies we will now also
2976 load SMACK policies at early boot.
2978 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2979 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2980 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2981 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2982 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2983 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2984 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2985 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2986 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2987 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2988 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2989 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2990 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2991 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2995 * Timer units now support calendar time events in addition to
2996 monotonic time events. That means you can now trigger a unit
2997 based on a calendar time specification such as "Thu,Fri
2998 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2999 or fifth day of any month of the year 2013, given that it is
3000 a thursday or friday. This brings timer event support
3001 considerably closer to cron's capabilities. For details on
3002 the supported calendar time specification language see
3005 * udev now supports a number of different naming policies for
3006 network interfaces for predictable names, and a combination
3007 of these policies is now the default. Please see this wiki
3008 document for details:
3010 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
3012 * Auke Kok's bootchart implementation has been added to the
3013 systemd tree. It is an optional component that can graph the
3014 boot in quite some detail. It is one of the best bootchart
3015 implementations around and minimal in its code and
3018 * nss-myhostname has been integrated into the systemd source
3019 tree. nss-myhostname guarantees that the local hostname
3020 always stays resolvable via NSS. It has been a weak
3021 requirement of systemd-hostnamed since a long time, and
3022 since its code is actually trivial we decided to just
3023 include it in systemd's source tree. It can be turned off
3024 with a configure switch.
3026 * The read-ahead logic is now capable of properly detecting
3027 whether a btrfs file system is on SSD or rotating media, in
3028 order to optimize the read-ahead scheme. Previously, it was
3029 only capable of detecting this on traditional file systems
3032 * In udev, additional device properties are now read from the
3033 IAB in addition to the OUI database. Also, Bluetooth company
3034 identities are attached to the devices as well.
3036 * In service files %U may be used as specifier that is
3037 replaced by the configured user name of the service.
3039 * nspawn may now be invoked without a controlling TTY. This
3040 makes it suitable for invocation as its own service. This
3041 may be used to set up a simple containerized server system
3042 using only core OS tools.
3044 * systemd and nspawn can now accept socket file descriptors
3045 when they are started for socket activation. This enables
3046 implementation of socket activated nspawn
3047 containers. i.e. think about autospawning an entire OS image
3048 when the first SSH or HTTP connection is received. We expect
3049 that similar functionality will also be added to libvirt-lxc
3052 * journalctl will now suppress ANSI color codes when
3053 presenting log data.
3055 * systemctl will no longer show control group information for
3056 a unit if a the control group is empty anyway.
3058 * logind can now automatically suspend/hibernate/shutdown the
3061 * /etc/machine-info and hostnamed now also expose the chassis
3062 type of the system. This can be used to determine whether
3063 the local system is a laptop, desktop, handset or
3064 tablet. This information may either be configured by the
3065 user/vendor or is automatically determined from ACPI and DMI
3066 information if possible.
3068 * A number of PolicyKit actions are now bound together with
3069 "imply" rules. This should simplify creating UIs because
3070 many actions will now authenticate similar ones as well.
3072 * Unit files learnt a new condition ConditionACPower= which
3073 may be used to conditionalize a unit depending on whether an
3074 AC power source is connected or not, of whether the system
3075 is running on battery power.
3077 * systemctl gained a new "is-failed" verb that may be used in
3078 shell scripts and suchlike to check whether a specific unit
3079 is in the "failed" state.
3081 * The EnvironmentFile= setting in unit files now supports file
3082 globbing, and can hence be used to easily read a number of
3083 environment files at once.
3085 * systemd will no longer detect and recognize specific
3086 distributions. All distribution-specific #ifdeffery has been
3087 removed, systemd is now fully generic and
3088 distribution-agnostic. Effectively, not too much is lost as
3089 a lot of the code is still accessible via explicit configure
3090 switches. However, support for some distribution specific
3091 legacy configuration file formats has been dropped. We
3092 recommend distributions to simply adopt the configuration
3093 files everybody else uses now and convert the old
3094 configuration from packaging scripts. Most distributions
3095 already did that. If that's not possible or desirable,
3096 distributions are welcome to forward port the specific
3097 pieces of code locally from the git history.
3099 * When logging a message about a unit systemd will now always
3100 log the unit name in the message meta data.
3102 * localectl will now also discover system locale data that is
3103 not stored in locale archives, but directly unpacked.
3105 * logind will no longer unconditionally use framebuffer
3106 devices as seat masters, i.e. as devices that are required
3107 to be existing before a seat is considered preset. Instead,
3108 it will now look for all devices that are tagged as
3109 "seat-master" in udev. By default framebuffer devices will
3110 be marked as such, but depending on local systems other
3111 devices might be marked as well. This may be used to
3112 integrate graphics cards using closed source drivers (such
3113 as NVidia ones) more nicely into logind. Note however, that
3114 we recommend using the open source NVidia drivers instead,
3115 and no udev rules for the closed-source drivers will be
3116 shipped from us upstream.
3118 Contributions from: Adam Williamson, Alessandro Crismani, Auke
3119 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
3120 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
3121 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
3122 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
3123 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
3124 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
3125 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
3126 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
3127 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
3128 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
3129 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
3134 * udev gained support for loading additional device properties
3135 from an indexed database that is keyed by vendor/product IDs
3136 and similar device identifiers. For the beginning this
3137 "hwdb" is populated with data from the well-known PCI and
3138 USB database, but also includes PNP, ACPI and OID data. In
3139 the longer run this indexed database shall grow into
3140 becoming the one central database for non-essential
3141 userspace device metadata. Previously, data from the PCI/USB
3142 database was only attached to select devices, since the
3143 lookup was a relatively expensive operation due to O(n) time
3144 complexity (with n being the number of entries in the
3145 database). Since this is now O(1), we decided to add in this
3146 data for all devices where this is available, by
3147 default. Note that the indexed database needs to be rebuilt
3148 when new data files are installed. To achieve this you need
3149 to update your packaging scripts to invoke "udevadm hwdb
3150 --update" after installation of hwdb data files. For
3151 RPM-based distributions we introduced the new
3152 %udev_hwdb_update macro for this purpose.
3154 * The Journal gained support for the "Message Catalog", an
3155 indexed database to link up additional information with
3156 journal entries. For further details please check:
3158 http://www.freedesktop.org/wiki/Software/systemd/catalog
3160 The indexed message catalog database also needs to be
3161 rebuilt after installation of message catalog files. Use
3162 "journalctl --update-catalog" for this. For RPM-based
3163 distributions we introduced the %journal_catalog_update
3164 macro for this purpose.
3166 * The Python Journal bindings gained support for the standard
3167 Python logging framework.
3169 * The Journal API gained new functions for checking whether
3170 the underlying file system of a journal file is capable of
3171 properly reporting file change notifications, or whether
3172 applications that want to reflect journal changes "live"
3173 need to recheck journal files continuously in appropriate
3176 * It is now possible to set the "age" field for tmpfiles
3177 entries to 0, indicating that files matching this entry
3178 shall always be removed when the directories are cleaned up.
3180 * coredumpctl gained a new "gdb" verb which invokes gdb
3181 right-away on the selected coredump.
3183 * There's now support for "hybrid sleep" on kernels that
3184 support this, in addition to "suspend" and "hibernate". Use
3185 "systemctl hybrid-sleep" to make use of this.
3187 * logind's HandleSuspendKey= setting (and related settings)
3188 now gained support for a new "lock" setting to simply
3189 request the screen lock on all local sessions, instead of
3190 actually executing a suspend or hibernation.
3192 * systemd will now mount the EFI variables file system by
3195 * Socket units now gained support for configuration of the
3196 SMACK security label.
3198 * timedatectl will now output the time of the last and next
3199 daylight saving change.
3201 * We dropped support for various legacy and distro-specific
3202 concepts, such as insserv, early-boot SysV services
3203 (i.e. those for non-standard runlevels such as 'b' or 'S')
3204 or ArchLinux /etc/rc.conf support. We recommend the
3205 distributions who still need support this to either continue
3206 to maintain the necessary patches downstream, or find a
3207 different solution. (Talk to us if you have questions!)
3209 * Various systemd components will now bypass PolicyKit checks
3210 for root and otherwise handle properly if PolicyKit is not
3211 found to be around. This should fix most issues for
3212 PolicyKit-less systems. Quite frankly this should have been
3213 this way since day one. It is absolutely our intention to
3214 make systemd work fine on PolicyKit-less systems, and we
3215 consider it a bug if something does not work as it should if
3216 PolicyKit is not around.
3218 * For embedded systems it is now possible to build udev and
3219 systemd without blkid and/or kmod support.
3221 * "systemctl switch-root" is now capable of switching root
3222 more than once. I.e. in addition to transitions from the
3223 initrd to the host OS it is now possible to transition to
3224 further OS images from the host. This is useful to implement
3225 offline updating tools.
3227 * Various other additions have been made to the RPM macros
3228 shipped with systemd. Use %udev_rules_update() after
3229 installing new udev rules files. %_udevhwdbdir,
3230 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3231 %_sysctldir are now available which resolve to the right
3232 directories for packages to place various data files in.
3234 * journalctl gained the new --full switch (in addition to
3235 --all, to disable ellipsation for long messages.
3237 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3238 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3239 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3240 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3241 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3242 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3243 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3244 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3245 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3249 * journalctl gained new --since= and --until= switches to
3250 filter by time. It also now supports nice filtering for
3251 units via --unit=/-u.
3253 * Type=oneshot services may use ExecReload= and do the
3256 * The journal daemon now supports time-based rotation and
3257 vacuuming, in addition to the usual disk-space based
3260 * The journal will now index the available field values for
3261 each field name. This enables clients to show pretty drop
3262 downs of available match values when filtering. The bash
3263 completion of journalctl has been updated
3264 accordingly. journalctl gained a new switch -F to list all
3265 values a certain field takes in the journal database.
3267 * More service events are now written as structured messages
3268 to the journal, and made recognizable via message IDs.
3270 * The timedated, localed and hostnamed mini-services which
3271 previously only provided support for changing time, locale
3272 and hostname settings from graphical DEs such as GNOME now
3273 also have a minimal (but very useful) text-based client
3274 utility each. This is probably the nicest way to changing
3275 these settings from the command line now, especially since
3276 it lists available options and is fully integrated with bash
3279 * There's now a new tool "systemd-coredumpctl" to list and
3280 extract coredumps from the journal.
3282 * We now install a README each in /var/log/ and
3283 /etc/rc.d/init.d explaining where the system logs and init
3284 scripts went. This hopefully should help folks who go to
3285 that dirs and look into the otherwise now empty void and
3286 scratch their heads.
3288 * When user-services are invoked (by systemd --user) the
3289 $MANAGERPID env var is set to the PID of systemd.
3291 * SIGRTMIN+24 when sent to a --user instance will now result
3292 in immediate termination of systemd.
3294 * gatewayd received numerous feature additions such as a
3295 "follow" mode, for live syncing and filtering.
3297 * browse.html now allows filtering and showing detailed
3298 information on specific entries. Keyboard navigation and
3299 mouse screen support has been added.
3301 * gatewayd/journalctl now supports HTML5/JSON
3302 Server-Sent-Events as output.
3304 * The SysV init script compatibility logic will now
3305 heuristically determine whether a script supports the
3306 "reload" verb, and only then make this available as
3309 * "systemctl status --follow" has been removed, use "journalctl
3312 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3313 have been removed since they are hardly useful to be
3316 * And I'd like to take the opportunity to specifically mention
3317 Zbigniew for his great contributions. Zbigniew, you rock!
3319 Contributions from: Andrew Eikum, Christian Hesse, Colin
3320 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3321 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3322 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3323 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3324 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3325 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3329 * If /etc/vconsole.conf is non-existent or empty we will no
3330 longer load any console font or key map at boot by
3331 default. Instead the kernel defaults will be left
3332 intact. This is definitely the right thing to do, as no
3333 configuration should mean no configuration, and hard-coding
3334 font names that are different on all archs is probably a bad
3335 idea. Also, the kernel default key map and font should be
3336 good enough for most cases anyway, and mostly identical to
3337 the userspace fonts/key maps we previously overloaded them
3338 with. If distributions want to continue to default to a
3339 non-kernel font or key map they should ship a default
3340 /etc/vconsole.conf with the appropriate contents.
3342 Contributions from: Colin Walters, Daniel J Walsh, Dave
3343 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3344 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3348 * journalctl gained a new --cursor= switch to show entries
3349 starting from the specified location in the journal.
3351 * We now enforce a size limit on journal entry fields exported
3352 with "-o json" in journalctl. Fields larger than 4K will be
3353 assigned null. This can be turned off with --all.
3355 * An (optional) journal gateway daemon is now available as
3356 "systemd-journal-gatewayd.service". This service provides
3357 access to the journal via HTTP and JSON. This functionality
3358 will be used to implement live log synchronization in both
3359 pull and push modes, but has various other users too, such
3360 as easy log access for debugging of embedded devices. Right
3361 now it is already useful to retrieve the journal via HTTP:
3363 # systemctl start systemd-journal-gatewayd.service
3364 # wget http://localhost:19531/entries
3366 This will download the journal contents in a
3367 /var/log/messages compatible format. The same as JSON:
3369 # curl -H"Accept: application/json" http://localhost:19531/entries
3371 This service is also accessible via a web browser where a
3372 single static HTML5 app is served that uses the JSON logic
3373 to enable the user to do some basic browsing of the
3374 journal. This will be extended later on. Here's an example
3375 screenshot of this app in its current state:
3377 http://0pointer.de/public/journal-gatewayd
3379 Contributions from: Kay Sievers, Lennart Poettering, Robert
3380 Milasan, Tom Gundersen
3384 * The bash completion logic is now available for journalctl
3387 * We do not mount the "cpuset" controller anymore together with
3388 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3389 started if no parameters are assigned to it. "cpuset" hence
3390 broke code that assumed it it could create "cpu" groups and
3393 * journalctl -f will now subscribe to terminal size changes,
3394 and line break accordingly.
3396 Contributions from: Dave Reisner, Kay Sievers, Lennart
3397 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3401 * nspawn will now create a symlink /etc/localtime in the
3402 container environment, copying the host's timezone
3403 setting. Previously this has been done via a bind mount, but
3404 since symlinks cannot be bind mounted this has now been
3405 changed to create/update the appropriate symlink.
3407 * journalctl -n's line number argument is now optional, and
3408 will default to 10 if omitted.
3410 * journald will now log the maximum size the journal files may
3411 take up on disk. This is particularly useful if the default
3412 built-in logic of determining this parameter from the file
3413 system size is used. Use "systemctl status
3414 systemd-journald.service" to see this information.
3416 * The multi-seat X wrapper tool has been stripped down. As X
3417 is now capable of enumerating graphics devices via udev in a
3418 seat-aware way the wrapper is not strictly necessary
3419 anymore. A stripped down temporary stop-gap is still shipped
3420 until the upstream display managers have been updated to
3421 fully support the new X logic. Expect this wrapper to be
3422 removed entirely in one of the next releases.
3424 * HandleSleepKey= in logind.conf has been split up into
3425 HandleSuspendKey= and HandleHibernateKey=. The old setting
3426 is not available anymore. X11 and the kernel are
3427 distuingishing between these keys and we should too. This
3428 also means the inhibition lock for these keys has been split
3431 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3432 Poettering, Lukas Nykryn, Václav Pavlín
3436 * Whenever a unit changes state we will now log this to the
3437 journal and show along the unit's own log output in
3440 * ConditionPathIsMountPoint= can now properly detect bind
3441 mount points too. (Previously, a bind mount of one file
3442 system to another place in the same file system could not be
3443 detected as mount, since they shared struct stat's st_dev
3446 * We will now mount the cgroup controllers cpu, cpuacct,
3447 cpuset and the controllers net_cls, net_prio together by
3450 * nspawn containers will now have a virtualized boot
3451 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3452 over with a randomized ID at container initialization). This
3453 has the effect of making "journalctl -b" do the right thing
3456 * The JSON output journal serialization has been updated not
3457 to generate "endless" list objects anymore, but rather one
3458 JSON object per line. This is more in line how most JSON
3459 parsers expect JSON objects. The new output mode
3460 "json-pretty" has been added to provide similar output, but
3461 neatly aligned for readability by humans.
3463 * We dropped all explicit sync() invocations in the shutdown
3464 code. The kernel does this implicitly anyway in the kernel
3465 reboot() syscall. halt(8)'s -n option is now a compatibility
3468 * We now support virtualized reboot() in containers, as
3469 supported by newer kernels. We will fall back to exit() if
3470 CAP_SYS_REBOOT is not available to the container. Also,
3471 nspawn makes use of this now and will actually reboot the
3472 container if the containerized OS asks for that.
3474 * journalctl will only show local log output by default
3475 now. Use --merge (-m) to show remote log output, too.
3477 * libsystemd-journal gained the new sd_journal_get_usage()
3478 call to determine the current disk usage of all journal
3479 files. This is exposed in the new "journalctl --disk-usage"
3482 * journald gained a new configuration setting SplitMode= in
3483 journald.conf which may be used to control how user journals
3484 are split off. See journald.conf(5) for details.
3486 * A new condition type ConditionFileNotEmpty= has been added.
3488 * tmpfiles' "w" lines now support file globbing, to write
3489 multiple files at once.
3491 * We added Python bindings for the journal submission
3492 APIs. More Python APIs for a number of selected APIs will
3493 likely follow. Note that we intend to add native bindings
3494 only for the Python language, as we consider it common
3495 enough to deserve bindings shipped within systemd. There are
3496 various projects outside of systemd that provide bindings
3497 for languages such as PHP or Lua.
3499 * Many conditions will now resolve specifiers such as %i. In
3500 addition, PathChanged= and related directives of .path units
3501 now support specifiers as well.
3503 * There's now a new RPM macro definition for the system preset
3506 * journald will now warn if it ca not forward a message to the
3507 syslog daemon because its socket is full.
3509 * timedated will no longer write or process /etc/timezone,
3510 except on Debian. As we do not support late mounted /usr
3511 anymore /etc/localtime always being a symlink is now safe,
3512 and hence the information in /etc/timezone is not necessary
3515 * logind will now always reserve one VT for a text getty (VT6
3516 by default). Previously if more than 6 X sessions where
3517 started they took up all the VTs with auto-spawned gettys,
3518 so that no text gettys were available anymore.
3520 * udev will now automatically inform the btrfs kernel logic
3521 about btrfs RAID components showing up. This should make
3522 simple hotplug based btrfs RAID assembly work.
3524 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3525 (but not for its children which will stay at the kernel
3526 default). This should allow setups with a lot more listening
3529 * systemd will now always pass the configured timezone to the
3530 kernel at boot. timedated will do the same when the timezone
3533 * logind's inhibition logic has been updated. By default,
3534 logind will now handle the lid switch, the power and sleep
3535 keys all the time, even in graphical sessions. If DEs want
3536 to handle these events on their own they should take the new
3537 handle-power-key, handle-sleep-key and handle-lid-switch
3538 inhibitors during their runtime. A simple way to achiveve
3539 that is to invoke the DE wrapped in an invocation of:
3541 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3543 * Access to unit operations is now checked via SELinux taking
3544 the unit file label and client process label into account.
3546 * systemd will now notify the administrator in the journal
3547 when he over-mounts a non-empty directory.
3549 * There are new specifiers that are resolved in unit files,
3550 for the host name (%H), the machine ID (%m) and the boot ID
3553 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3554 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3555 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3556 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3557 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3558 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3559 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3563 * Support for reading structured kernel messages from
3564 /dev/kmsg has now been added and is enabled by default.
3566 * Support for reading kernel messages from /proc/kmsg has now
3567 been removed. If you want kernel messages in the journal
3568 make sure to run a recent kernel (>= 3.5) that supports
3569 reading structured messages from /dev/kmsg (see
3570 above). /proc/kmsg is now exclusive property of classic
3571 syslog daemons again.
3573 * The libudev API gained the new
3574 udev_device_new_from_device_id() call.
3576 * The logic for file system namespace (ReadOnlyDirectory=,
3577 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3578 require pivot_root() anymore. This means fewer temporary
3579 directories are created below /tmp for this feature.
3581 * nspawn containers will now see and receive all submounts
3582 made on the host OS below the root file system of the
3585 * Forward Secure Sealing is now supported for Journal files,
3586 which provide cryptographical sealing of journal files so
3587 that attackers cannot alter log history anymore without this
3588 being detectable. Lennart will soon post a blog story about
3589 this explaining it in more detail.
3591 * There are two new service settings RestartPreventExitStatus=
3592 and SuccessExitStatus= which allow configuration of exit
3593 status (exit code or signal) which will be excepted from the
3594 restart logic, resp. consider successful.
3596 * journalctl gained the new --verify switch that can be used
3597 to check the integrity of the structure of journal files and
3598 (if Forward Secure Sealing is enabled) the contents of
3601 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3602 and similar symlinks pre-created. This makes running shells
3603 as container init process a lot more fun.
3605 * The fstab support can now handle PARTUUID= and PARTLABEL=
3608 * A new ConditionHost= condition has been added to match
3609 against the hostname (with globs) and machine ID. This is
3610 useful for clusters where a single OS image is used to
3611 provision a large number of hosts which shall run slightly
3612 different sets of services.
3614 * Services which hit the restart limit will now be placed in a
3617 Contributions from: Bertram Poettering, Dave Reisner, Huang
3618 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3619 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3623 * When running in --user mode systemd will now become a
3624 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3625 tree a lot more organized.
3627 * A new PartOf= unit dependency type has been introduced that
3628 may be used to group services in a natural way.
3630 * "systemctl enable" may now be used to enable instances of
3633 * journalctl now prints error log levels in red, and
3634 warning/notice log levels in bright white. It also supports
3635 filtering by log level now.
3637 * cgtop gained a new -n switch (similar to top), to configure
3638 the maximum number of iterations to run for. It also gained
3639 -b, to run in batch mode (accepting no input).
3641 * The suffix ".service" may now be omitted on most systemctl
3642 command lines involving service unit names.
3644 * There's a new bus call in logind to lock all sessions, as
3645 well as a loginctl verb for it "lock-sessions".
3647 * libsystemd-logind.so gained a new call sd_journal_perror()
3648 that works similar to libc perror() but logs to the journal
3649 and encodes structured information about the error number.
3651 * /etc/crypttab entries now understand the new keyfile-size=
3654 * shutdown(8) now can send a (configurable) wall message when
3655 a shutdown is cancelled.
3657 * The mount propagation mode for the root file system will now
3658 default to "shared", which is useful to make containers work
3659 nicely out-of-the-box so that they receive new mounts from
3660 the host. This can be undone locally by running "mount
3661 --make-rprivate /" if needed.
3663 * The prefdm.service file has been removed. Distributions
3664 should maintain this unit downstream if they intend to keep
3665 it around. However, we recommend writing normal unit files
3666 for display managers instead.
3668 * Since systemd is a crucial part of the OS we will now
3669 default to a number of compiler switches that improve
3670 security (hardening) such as read-only relocations, stack
3671 protection, and suchlike.
3673 * The TimeoutSec= setting for services is now split into
3674 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3675 of individual time outs for the start and the stop phase of
3678 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3679 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3680 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3681 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3682 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3683 Gundersen, Zbigniew Jędrzejewski-Szmek
3687 * The journal and id128 C APIs are now fully documented as man
3690 * Extra safety checks have been added when transitioning from
3691 the initial RAM disk to the main system to avoid accidental
3694 * /etc/crypttab entries now understand the new keyfile-offset=
3697 * systemctl -t can now be used to filter by unit load state.
3699 * The journal C API gained the new sd_journal_wait() call to
3700 make writing synchronous journal clients easier.
3702 * journalctl gained the new -D switch to show journals from a
3705 * journalctl now displays a special marker between log
3706 messages of two different boots.
3708 * The journal is now explicitly flushed to /var via a service
3709 systemd-journal-flush.service, rather than implicitly simply
3710 by seeing /var/log/journal to be writable.
3712 * journalctl (and the journal C APIs) can now match for much
3713 more complex expressions, with alternatives and
3716 * When transitioning from the initial RAM disk to the main
3717 system we will now kill all processes in a killing spree to
3718 ensure no processes stay around by accident.
3720 * Three new specifiers may be used in unit files: %u, %h, %s
3721 resolve to the user name, user home directory resp. user
3722 shell. This is useful for running systemd user instances.
3724 * We now automatically rotate journal files if their data
3725 object hash table gets a fill level > 75%. We also size the
3726 hash table based on the configured maximum file size. This
3727 together should lower hash collisions drastically and thus
3728 speed things up a bit.
3730 * journalctl gained the new "--header" switch to introspect
3731 header data of journal files.
3733 * A new setting SystemCallFilters= has been added to services
3734 which may be used to apply blacklists or whitelists to
3735 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3737 * nspawn gained a new --link-journal= switch (and quicker: -j)
3738 to link the container journal with the host. This makes it
3739 very easy to centralize log viewing on the host for all
3740 guests while still keeping the journal files separated.
3742 * Many bugfixes and optimizations
3744 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3745 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3746 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3751 * Several tools now understand kernel command line arguments,
3752 which are only read when run in an initial RAM disk. They
3753 usually follow closely their normal counterparts, but are
3756 * There's a new tool to analyze the readahead files that are
3757 automatically generated at boot. Use:
3759 /usr/lib/systemd/systemd-readahead analyze /.readahead
3761 * We now provide an early debug shell on tty9 if this enabled. Use:
3763 systemctl enable debug-shell.service
3765 * All plymouth related units have been moved into the Plymouth
3766 package. Please make sure to upgrade your Plymouth version
3769 * systemd-tmpfiles now supports getting passed the basename of
3770 a configuration file only, in which case it will look for it
3771 in all appropriate directories automatically.
3773 * udevadm info now takes a /dev or /sys path as argument, and
3774 does the right thing. Example:
3776 udevadm info /dev/sda
3777 udevadm info /sys/class/block/sda
3779 * systemctl now prints a warning if a unit is stopped but a
3780 unit that might trigger it continues to run. Example: a
3781 service is stopped but the socket that activates it is left
3784 * "systemctl status" will now mention if the log output was
3785 shortened due to rotation since a service has been started.
3787 * The journal API now exposes functions to determine the
3788 "cutoff" times due to rotation.
3790 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3791 immediately flushing of runtime logs to /var if possible,
3792 resp. for triggering immediate rotation of the journal
3795 * It is now considered an error if a service is attempted to
3796 be stopped that is not loaded.
3798 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3800 * systemd-analyze now supports Python 3
3802 * tmpfiles now supports cleaning up directories via aging
3803 where the first level dirs are always kept around but
3804 directories beneath it automatically aged. This is enabled
3805 by prefixing the age field with '~'.
3807 * Seat objects now expose CanGraphical, CanTTY properties
3808 which is required to deal with very fast bootups where the
3809 display manager might be running before the graphics drivers
3810 completed initialization.
3812 * Seat objects now expose a State property.
3814 * We now include RPM macros for service enabling/disabling
3815 based on the preset logic. We recommend RPM based
3816 distributions to make use of these macros if possible. This
3817 makes it simpler to reuse RPM spec files across
3820 * We now make sure that the collected systemd unit name is
3821 always valid when services log to the journal via
3824 * There's a new man page kernel-command-line(7) detailing all
3825 command line options we understand.
3827 * The fstab generator may now be disabled at boot by passing
3828 fstab=0 on the kernel command line.
3830 * A new kernel command line option modules-load= is now understood
3831 to load a specific kernel module statically, early at boot.
3833 * Unit names specified on the systemctl command line are now
3834 automatically escaped as needed. Also, if file system or
3835 device paths are specified they are automatically turned
3836 into the appropriate mount or device unit names. Example:
3838 systemctl status /home
3839 systemctl status /dev/sda
3841 * The SysVConsole= configuration option has been removed from
3842 system.conf parsing.
3844 * The SysV search path is no longer exported on the D-Bus
3847 * The Names= option is been removed from unit file parsing.
3849 * There's a new man page bootup(7) detailing the boot process.
3851 * Every unit and every generator we ship with systemd now
3852 comes with full documentation. The self-explanatory boot is
3855 * A couple of services gained "systemd-" prefixes in their
3856 name if they wrap systemd code, rather than only external
3857 code. Among them fsck@.service which is now
3858 systemd-fsck@.service.
3860 * The HaveWatchdog property has been removed from the D-Bus
3863 * systemd.confirm_spawn= on the kernel command line should now
3866 * There's a new man page crypttab(5) which details all options
3867 we actually understand.
3869 * systemd-nspawn gained a new --capability= switch to pass
3870 additional capabilities to the container.
3872 * timedated will now read known NTP implementation unit names
3873 from /usr/lib/systemd/ntp-units.d/*.list,
3874 systemd-timedated-ntp.target has been removed.
3876 * journalctl gained a new switch "-b" that lists log data of
3877 the current boot only.
3879 * The notify socket is in the abstract namespace again, in
3880 order to support daemons which chroot() at start-up.
3882 * There is a new Storage= configuration option for journald
3883 which allows configuration of where log data should go. This
3884 also provides a way to disable journal logging entirely, so
3885 that data collected is only forwarded to the console, the
3886 kernel log buffer or another syslog implementation.
3888 * Many bugfixes and optimizations
3890 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3891 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3892 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3893 Shawn Landden, Tom Gundersen
3897 * "systemctl help <unit>" now shows the man page if one is
3900 * Several new man pages have been added.
3902 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3903 MaxLevelConsole= can now be specified in
3904 journald.conf. These options allow reducing the amount of
3905 data stored on disk or forwarded by the log level.
3907 * TimerSlackNSec= can now be specified in system.conf for
3908 PID1. This allows system-wide power savings.
3910 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3911 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3916 * logind is now capable of (optionally) handling power and
3917 sleep keys as well as the lid switch.
3919 * journalctl now understands the syntax "journalctl
3920 /usr/bin/avahi-daemon" to get all log output of a specific
3923 * CapabilityBoundingSet= in system.conf now also influences
3924 the capability bound set of usermode helpers of the kernel.
3926 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3927 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3928 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3929 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3933 * Note that we skipped 139 releases here in order to set the
3934 new version to something that is greater than both udev's
3935 and systemd's most recent version number.
3937 * udev: all udev sources are merged into the systemd source tree now.
3938 All future udev development will happen in the systemd tree. It
3939 is still fully supported to use the udev daemon and tools without
3940 systemd running, like in initramfs or other init systems. Building
3941 udev though, will require the *build* of the systemd tree, but
3942 udev can be properly *run* without systemd.
3944 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3945 should be used to create dead device nodes as workarounds for broken
3948 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3949 no longer supported. udev_monitor_new_from_netlink() needs to be
3950 used to subscribe to events.
3952 * udev: when udevd is started by systemd, processes which are left
3953 behind by forking them off of udev rules, are unconditionally cleaned
3954 up and killed now after the event handling has finished. Services or
3955 daemons must be started as systemd services. Services can be
3956 pulled-in by udev to get started, but they can no longer be directly
3957 forked by udev rules.
3959 * udev: the daemon binary is called systemd-udevd now and installed
3960 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3961 to adapt to that, create symlink, or rename the binary after building
3964 * libudev no longer provides these symbols:
3965 udev_monitor_from_socket()
3966 udev_queue_get_failed_list_entry()
3967 udev_get_{dev,sys,run}_path()
3968 The versions number was bumped and symbol versioning introduced.
3970 * systemd-loginctl and systemd-journalctl have been renamed
3971 to loginctl and journalctl to match systemctl.
3973 * The config files: /etc/systemd/systemd-logind.conf and
3974 /etc/systemd/systemd-journald.conf have been renamed to
3975 logind.conf and journald.conf. Package updates should rename
3976 the files to the new names on upgrade.
3978 * For almost all files the license is now LGPL2.1+, changed
3979 from the previous GPL2.0+. Exceptions are some minor stuff
3980 of udev (which will be changed to LGPL2.1 eventually, too),
3981 and the MIT licensed sd-daemon.[ch] library that is suitable
3982 to be used as drop-in files.
3984 * systemd and logind now handle system sleep states, in
3985 particular suspending and hibernating.
3987 * logind now implements a sleep/shutdown/idle inhibiting logic
3988 suitable for a variety of uses. Soonishly Lennart will blog
3989 about this in more detail.
3991 * var-run.mount and var-lock.mount are no longer provided
3992 (which prevously bind mounted these directories to their new
3993 places). Distributions which have not converted these
3994 directories to symlinks should consider stealing these files
3995 from git history and add them downstream.
3997 * We introduced the Documentation= field for units and added
3998 this to all our shipped units. This is useful to make it
3999 easier to explore the boot and the purpose of the various
4002 * All smaller setup units (such as
4003 systemd-vconsole-setup.service) now detect properly if they
4004 are run in a container and are skipped when
4005 appropriate. This guarantees an entirely noise-free boot in
4006 Linux container environments such as systemd-nspawn.
4008 * A framework for implementing offline system updates is now
4009 integrated, for details see:
4010 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
4012 * A new service type Type=idle is available now which helps us
4013 avoiding ugly interleaving of getty output and boot status
4016 * There's now a system-wide CapabilityBoundingSet= option to
4017 globally reduce the set of capabilities for the
4018 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
4019 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
4020 even CAP_NET_ADMIN system-wide for secure systems.
4022 * There are now system-wide DefaultLimitXXX= options to
4023 globally change the defaults of the various resource limits
4024 for all units started by PID 1.
4026 * Harald Hoyer's systemd test suite has been integrated into
4027 systemd which allows easy testing of systemd builds in qemu
4028 and nspawn. (This is really awesome! Ask us for details!)
4030 * The fstab parser is now implemented as generator, not inside
4033 * systemctl will now warn you if .mount units generated from
4034 /etc/fstab are out of date due to changes in fstab that
4035 have not been read by systemd yet.
4037 * systemd is now suitable for usage in initrds. Dracut has
4038 already been updated to make use of this. With this in place
4039 initrds get a slight bit faster but primarily are much
4040 easier to introspect and debug since "systemctl status" in
4041 the host system can be used to introspect initrd services,
4042 and the journal from the initrd is kept around too.
4044 * systemd-delta has been added, a tool to explore differences
4045 between user/admin configuration and vendor defaults.
4047 * PrivateTmp= now affects both /tmp and /var/tmp.
4049 * Boot time status messages are now much prettier and feature
4050 proper english language. Booting up systemd has never been
4053 * Read-ahead pack files now include the inode number of all
4054 files to pre-cache. When the inode changes the pre-caching
4055 is not attempted. This should be nicer to deal with updated
4056 packages which might result in changes of read-ahead
4059 * We now temporaritly lower the kernel's read_ahead_kb variable
4060 when collecting read-ahead data to ensure the kernel's
4061 built-in read-ahead does not add noise to our measurements
4062 of necessary blocks to pre-cache.
4064 * There's now RequiresMountsFor= to add automatic dependencies
4065 for all mounts necessary for a specific file system path.
4067 * MountAuto= and SwapAuto= have been removed from
4068 system.conf. Mounting file systems at boot has to take place
4071 * nspawn now learned a new switch --uuid= to set the machine
4072 ID on the command line.
4074 * nspawn now learned the -b switch to automatically search
4077 * vt102 is now the default TERM for serial TTYs, upgraded from
4080 * systemd-logind now works on VT-less systems.
4082 * The build tree has been reorganized. The individual
4083 components now have directories of their own.
4085 * A new condition type ConditionPathIsReadWrite= is now available.
4087 * nspawn learned the new -C switch to create cgroups for the
4088 container in other hierarchies.
4090 * We now have support for hardware watchdogs, configurable in
4093 * The scheduled shutdown logic now has a public API.
4095 * We now mount /tmp as tmpfs by default, but this can be
4096 masked and /etc/fstab can override it.
4098 * Since udisks does not make use of /media anymore we are not
4099 mounting a tmpfs on it anymore.
4101 * journalctl gained a new --local switch to only interleave
4102 locally generated journal files.
4104 * We can now load the IMA policy at boot automatically.
4106 * The GTK tools have been split off into a systemd-ui.
4108 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
4109 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
4110 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
4111 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
4112 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
4113 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
4114 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
4115 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
4116 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
4121 * This is mostly a bugfix release
4123 * Support optional initialization of the machine ID from the
4124 KVM or container configured UUID.
4126 * Support immediate reboots with "systemctl reboot -ff"
4128 * Show /etc/os-release data in systemd-analyze output
4130 * Many bugfixes for the journal, including endianness fixes and
4131 ensuring that disk space enforcement works
4133 * sd-login.h is C++ comptaible again
4135 * Extend the /etc/os-release format on request of the Debian
4138 * We now refuse non-UTF8 strings used in various configuration
4139 and unit files. This is done to ensure we do not pass invalid
4140 data over D-Bus or expose it elsewhere.
4142 * Register Mimo USB Screens as suitable for automatic seat
4145 * Read SELinux client context from journal clients in a race
4148 * Reorder configuration file lookup order. /etc now always
4149 overrides /run in order to allow the administrator to always
4150 and unconditionally override vendor supplied or
4151 automatically generated data.
4153 * The various user visible bits of the journal now have man
4154 pages. We still lack man pages for the journal API calls
4157 * We now ship all man pages in HTML format again in the
4160 Contributions from: Dave Reisner, Dirk Eibach, Frederic
4161 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
4162 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
4167 * This is mostly a bugfix release
4169 * systems lacking /etc/os-release are no longer supported.
4171 * Various functionality updates to libsystemd-login.so
4173 * Track class of PAM logins to distuingish greeters from
4176 Contributions from: Kay Sievers, Lennart Poettering, Michael
4181 * This is an important bugfix release for v41.
4183 * Building man pages is now optional which should be useful
4184 for those building systemd from git but unwilling to install
4187 * Watchdog support for supervising services is now usable. In
4188 a future release support for hardware watchdogs
4189 (i.e. /dev/watchdog) will be added building on this.
4191 * Service start rate limiting is now configurable and can be
4192 turned off per service. When a start rate limit is hit a
4193 reboot can automatically be triggered.
4195 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4197 Contributions from: Benjamin Franzke, Bill Nottingham,
4198 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4199 Schmidt, Michał Górny, Piotr Drąg
4203 * The systemd binary is installed /usr/lib/systemd/systemd now;
4204 An existing /sbin/init symlink needs to be adapted with the
4207 * The code that loads kernel modules has been ported to invoke
4208 libkmod directly, instead of modprobe. This means we do not
4209 support systems with module-init-tools anymore.
4211 * Watchdog support is now already useful, but still not
4214 * A new kernel command line option systemd.setenv= is
4215 understood to set system wide environment variables
4216 dynamically at boot.
4218 * We now limit the set of capabilities of systemd-journald.
4220 * We now set SIGPIPE to ignore by default, since it only is
4221 useful in shell pipelines, and has little use in general
4222 code. This can be disabled with IgnoreSIPIPE=no in unit
4225 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4226 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4231 * This is mostly a bugfix release
4233 * We now expose the reason why a service failed in the
4234 "Result" D-Bus property.
4236 * Rudimentary service watchdog support (will be completed over
4237 the next few releases.)
4239 * When systemd forks off in order execute some service we will
4240 now immediately changes its argv[0] to reflect which process
4241 it will execute. This is useful to minimize the time window
4242 with a generic argv[0], which makes bootcharts more useful
4244 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4245 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4246 Mike Kazantsev, Ray Strode
4250 * This is mostly a test release, but incorporates many
4253 * New systemd-cgtop tool to show control groups by their
4256 * Linking against libacl for ACLs is optional again. If
4257 disabled, support tracking device access for active logins
4258 goes becomes unavailable, and so does access to the user
4259 journals by the respective users.
4261 * If a group "adm" exists, journal files are automatically
4262 owned by them, thus allow members of this group full access
4263 to the system journal as well as all user journals.
4265 * The journal now stores the SELinux context of the logging
4266 client for all entries.
4268 * Add C++ inclusion guards to all public headers
4270 * New output mode "cat" in the journal to print only text
4271 messages, without any meta data like date or time.
4273 * Include tiny X server wrapper as a temporary stop-gap to
4274 teach XOrg udev display enumeration. This is used by display
4275 managers such as gdm, and will go away as soon as XOrg
4276 learned native udev hotplugging for display devices.
4278 * Add new systemd-cat tool for executing arbitrary programs
4279 with STDERR/STDOUT connected to the journal. Can also act as
4280 BSD logger replacement, and does so by default.
4282 * Optionally store all locally generated coredumps in the
4283 journal along with meta data.
4285 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4286 writing short strings to files (for usage for /sys), and for
4287 creating symlinks, character and block device nodes.
4289 * New unit file option ControlGroupPersistent= to make cgroups
4290 persistent, following the mechanisms outlined in
4291 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4293 * Support multiple local RTCs in a sane way
4295 * No longer monopolize IO when replaying readahead data on
4296 rotating disks, since we might starve non-file-system IO to
4297 death, since fanotify() will not see accesses done by blkid,
4300 * Do not show kernel threads in systemd-cgls anymore, unless
4301 requested with new -k switch.
4303 Contributions from: Dan Horák, Kay Sievers, Lennart
4304 Poettering, Michal Schmidt
4308 * This is mostly a test release, but incorporates many
4311 * The git repository moved to:
4312 git://anongit.freedesktop.org/systemd/systemd
4313 ssh://git.freedesktop.org/git/systemd/systemd
4315 * First release with the journal
4316 http://0pointer.de/blog/projects/the-journal.html
4318 * The journal replaces both systemd-kmsg-syslogd and
4319 systemd-stdout-bridge.
4321 * New sd_pid_get_unit() API call in libsystemd-logind
4323 * Many systemadm clean-ups
4325 * Introduce remote-fs-pre.target which is ordered before all
4326 remote mounts and may be used to start services before all
4329 * Added Mageia support
4331 * Add bash completion for systemd-loginctl
4333 * Actively monitor PID file creation for daemons which exit in
4334 the parent process before having finished writing the PID
4335 file in the daemon process. Daemons which do this need to be
4336 fixed (i.e. PID file creation must have finished before the
4337 parent exits), but we now react a bit more gracefully to them.
4339 * Add colourful boot output, mimicking the well-known output
4340 of existing distributions.
4342 * New option PassCredentials= for socket units, for
4343 compatibility with a recent kernel ABI breakage.
4345 * /etc/rc.local is now hooked in via a generator binary, and
4346 thus will no longer act as synchronization point during
4349 * systemctl list-unit-files now supports --root=.
4351 * systemd-tmpfiles now understands two new commands: z, Z for
4352 relabelling files according to the SELinux database. This is
4353 useful to apply SELinux labels to specific files in /sys,
4356 * Output of SysV services is now forwarded to both the console
4357 and the journal by default, not only just the console.
4359 * New man pages for all APIs from libsystemd-login.
4361 * The build tree got reorganized and a the build system is a
4362 lot more modular allowing embedded setups to specifically
4363 select the components of systemd they are interested in.
4365 * Support for Linux systems lacking the kernel VT subsystem is
4368 * configure's --with-rootdir= got renamed to
4369 --with-rootprefix= to follow the naming used by udev and
4372 * Unless specified otherwise we will now install to /usr instead
4373 of /usr/local by default.
4375 * Processes with '@' in argv[0][0] are now excluded from the
4376 final shut-down killing spree, following the logic explained
4378 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4380 * All processes remaining in a service cgroup when we enter
4381 the START or START_PRE states are now killed with
4382 SIGKILL. That means it is no longer possible to spawn
4383 background processes from ExecStart= lines (which was never
4384 supported anyway, and bad style).
4386 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4387 reloading of units together.
4389 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4390 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4391 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4392 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4393 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek