1 systemd System and Service Manager
5 * journalctl gained option -t/--identifier to match on the
6 syslog identifier and --utc option to show timestamps in UTC
7 timezone. journalctl now also accepts -n/--lines=all to
8 disable line cap under a pager.
10 * Services can notify the manager before they start a reload
11 (by sending RELOADING=1) or shutdown (by sending
12 STOPPING=1). This allows the manager to track and show the
13 internal state of daemons and closes a race condition when
14 the process is still running but has closed its d-bus
17 * Services with Type=oneshot do not have to have any
18 ExecStart commands anymore.
20 * User units are now loaded also from
21 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
22 /run/systemd/user directory that was already previously
23 supported, but is under the control of the user.
25 * A timeout for the start of the system can be configured. The
26 system can be configured to reboot or poweroff if the basic
27 system default target is not reached before the timeout (new
28 StartTimeoutSec=, StartTimeoutAction=,
29 StartTimeoutRebootArgument= options).
31 * systemd-logind can be configured to also handle lid switch
32 events even when the machine is docked or multiple displays
33 are attached (HandleLidSwitchDocked= option).
35 * A helper binary and a service have been added which can be
36 used to resume from hibernation in the initramfs. A
37 generator will parse the resume= option on the kernel
38 command-line to trigger resume.
40 * A user console daemon systemd-consoled has been added. It is
41 a preview, and will so far open a single terminal on each
42 session of the user marked as Desktop=SYSTEMD-CONSOLE.
44 * Route metrics can be specified for DHCP routes added by
47 * SELinux context of socket-actived services can be set from
48 the information provided by the remote peer
49 (SELinuxContextFromNet= option).
51 * Userspace firmware loading support has been removed and
52 the minimum supported kernel version is thus bumped to 3.7.
54 * Timeout for udev workers has been increased from 1 to 3
55 minutes, but a warning will be printed after 1 minute to
56 help diagnose kernel modules that take a long time to load.
58 * Udev rules can now remove tags with TAG-="foobar".
60 * systemd's readahead implementation has been removed. In many
61 circumstatances it didn't give expected benefits even for
62 rotational disk drives and was becoming less relevant in the
65 * Swap units can use Discard= to specify discard options.
66 Discard options specified for swaps in /etc/fstab are now
69 * Docker containers are now detected as a separate type of
72 * The Password Agent protocol gained support for queries where
73 the user input is shown, useful e.g. for usernames.
74 systemd-ask-password gained a new --echo option to turn
77 * The default sysctl.d/ snippets will now set:
79 net.core.default_qdisc = fq_codel
81 This selects Fair Queueing Controlled Delay as the default
82 queueing discipline for network interfaces. fq_codel helps
83 fight the network bufferbloat problem. It is believed to be
84 a good default with no tuning required for most workloads.
85 Downstream distributions may override this choice. On 10Gbit
86 servers that do not do forwarding, "fq" may perform better.
87 Systems without a good clocksource should use "pfifo_fast".
89 * If kdbus is enabled during build a new option BusPolicy= is
90 available for service units, that allows locking all service
91 processes into a stricter bus policy, in order to limit
92 access to various bus services, or even hide most of them
93 from the service's view entirely.
95 * networkctl will now show the .network and .link file
96 networkd has applied to a specific interface.
98 * sd-login gained a new API call sd_session_get_desktop() to
99 query which desktop environment has been selected for a
102 * UNIX utmp support is now compile-time optional to support
107 * timedated no longer reads NTP implementation unit names from
108 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
109 implementations should add a
111 Conflicts=systemd-timesyncd.service
113 to their unit files to take over and replace systemd's NTP
114 default functionality.
116 * systemd-sysusers gained a new line type "r" for configuring
117 which UID/GID ranges to allocate system users/groups
118 from. Lines of type "u" may now add an additional column
119 that specifies the home directory for the system user to be
120 created. Also, systemd-sysusers may now optionally read user
121 information from STDIN instead of a file. This is useful for
122 invoking it from RPM preinst scriptlets that need to create
123 users before the first RPM file is installed since these
124 files might need to be owned by them. A new
125 %sysusers_create_inline RPM macro has been introduced to do
126 just that. systemd-sysusers now updates the shadow files as
127 well as the user/group databases, which should enhance
128 compatibility with certain tools like grpck.
130 * A number of bus APIs of PID 1 now optionally consult
131 PolicyKit to permit access for otherwise unprivileged
132 clients under certain conditions. Note that this currently
133 doesn't support interactive authentication yet, but this is
134 expected to be added eventually, too.
136 * /etc/machine-info now has new fields for configuring the
137 deployment environment of the machine, as well as the
138 location of the machine. hostnamectl has been updated with
139 new command to update these fields.
141 * systemd-timesyncd has been updated to automatically acquire
142 NTP server information from systemd-networkd, which might
143 have been discovered via DHCP.
145 * systemd-resolved now includes a caching DNS stub resolver
146 and a complete LLMNR name resolution implementation. A new
147 NSS module "nss-resolve" has been added which can be used
148 instead of glibc's own "nss-dns" to resolve hostnames via
149 systemd-resolved. Hostnames, addresses and arbitrary RRs may
150 be resolved via systemd-resolved D-Bus APIs. In contrast to
151 the glibc internal resolver systemd-resolved is aware of
152 multi-homed system, and keeps DNS server and caches separate
153 and per-interface. Queries are sent simultaneously on all
154 interfaces that have DNS servers configured, in order to
155 properly handle VPNs and local LANs which might resolve
156 separate sets of domain names. systemd-resolved may acquire
157 DNS server information from systemd-networkd automatically,
158 which in turn might have discovered them via DHCP. A tool
159 "systemd-resolve-host" has been added that may be used to
160 query the DNS logic in resolved. systemd-resolved implements
161 IDNA and automatically uses IDNA or UTF-8 encoding depending
162 on whether classic DNS or LLMNR is used as transport. In the
163 next releases we intend to add a DNSSEC and mDNS/DNS-SD
164 implementation to systemd-resolved.
166 * A new NSS module nss-mymachines has been added, that
167 automatically resolves the names of all local registered
168 containers to their respective IP addresses.
170 * A new client tool "networkctl" for systemd-networkd has been
171 added. It currently is entirely passive and will query
172 networking configuration from udev, rtnetlink and networkd,
173 and present it to the user in a very friendly
174 way. Eventually, we hope to extend it to become a full
175 control utility for networkd.
177 * .socket units gained a new DeferAcceptSec= setting that
178 controls the kernels' TCP_DEFER_ACCEPT sockopt for
179 TCP. Similar, support for controlling TCP keep-alive
180 settings has been added (KeepAliveTimeSec=,
181 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
182 turning off Nagle's algorithm on TCP has been added
185 * logind learned a new session type "web", for use in projects
186 like Cockpit which register web clients as PAM sessions.
188 * timer units with at least one OnCalendar= setting will now
189 be started only after timer-sync.target has been
190 reached. This way they will not elapse before the system
191 clock has been corrected by a local NTP client or
192 similar. This is particular useful on RTC-less embedded
193 machines, that come up with an invalid system clock.
195 * systemd-nspawn's --network-veth= switch should now result in
196 stable MAC addresses for both the outer and the inner side
199 * systemd-nspawn gained a new --volatile= switch for running
200 container instances with /etc or /var unpopulated.
202 * The kdbus client code has been updated to use the new Linux
203 3.17 memfd subsystem instead of the old kdbus-specific one.
205 * systemd-networkd's DHCP client and server now support
206 FORCERENEW. There are also new configuration options to
207 configure the vendor client identifier and broadcast mode
210 * systemd will no longer inform the kernel about the current
211 timezone, as this is necessarily incorrect and racy as the
212 kernel has no understanding of DST and similar
213 concepts. This hence means FAT timestamps will be always
214 considered UTC, similar to what Android is already
215 doing. Also, when the RTC is configured to the local time
216 (rather than UTC) systemd will never synchronize back to it,
217 as this might confuse Windows at a later boot.
219 * systemd-analyze gained a new command "verify" for offline
220 validation of unit files.
222 * systemd-networkd gained support for a couple of additional
223 settings for bonding networking setups. Also, the metric for
224 statically configured routes may now be configured. For
225 network interfaces where this is appropriate the peer IP
226 address may now be configured.
228 * systemd-networkd's DHCP client will no longer request
229 broadcasting by default, as this tripped up some networks.
230 For hardware where broadcast is required the feature should
231 be switched back on using RequestBroadcast=yes.
233 * systemd-networkd will now set up IPv4LL addresses (when
234 enabled) even if DHCP is configured successfully.
236 * udev will now default to respect network device names given
237 by the kernel when the kernel indicates that these are
238 predictable. This behavior can be tweaked by changing
239 NamePolicy= in the relevant .link file.
241 * A new library systemd-terminal has been added that
242 implements full TTY stream parsing and rendering. This
243 library is supposed to be used later on for implementing a
244 full userspace VT subsystem, replacing the current kernel
247 * A new tool systemd-journal-upload has been added to push
248 journal data to a remote system running
249 systemd-journal-remote.
251 * journald will no longer forward all local data to another
252 running syslog daemon. This change has been made because
253 rsyslog (which appears to be the most commonly used syslog
254 implementation these days) no longer makes use of this, and
255 instead pulls the data out of the journal on its own. Since
256 forwarding the messages to a non-existent syslog server is
257 more expensive than we assumed we have now turned this
258 off. If you run a syslog server that is not a recent rsyslog
259 version, you have to turn this option on again
260 (ForwardToSyslog= in journald.conf).
262 * journald now optionally supports the LZ4 compressor for
263 larger journal fields. This compressor should perform much
264 better than XZ which was the previous default.
266 * machinectl now shows the IP addresses of local containers,
267 if it knows them, plus the interface name of the container.
269 * A new tool "systemd-escape" has been added that makes it
270 easy to escape strings to build unit names and similar.
272 * sd_notify() messages may now include a new ERRNO= field
273 which is parsed and collected by systemd and shown among the
274 "systemctl status" output for a service.
276 * A new component "systemd-firstboot" has been added that
277 queries the most basic systemd information (timezone,
278 hostname, root password) interactively on first
279 boot. Alternatively it may also be used to provision these
280 things offline on OS images installed into directories.
282 * The default sysctl.d/ snippets will now set
284 net.ipv4.conf.default.promote_secondaries=1
286 This has the benefit of no flushing secondary IP addresses
287 when primary addresses are removed.
289 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
290 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
291 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
292 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
293 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
294 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
295 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
296 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
297 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
298 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
299 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
300 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
301 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
302 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
303 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
305 -- Berlin, 2014-08-19
309 * A new tool systemd-sysusers has been added. This tool
310 creates system users and groups in /etc/passwd and
311 /etc/group, based on static declarative system user/group
312 definitions in /usr/lib/sysusers.d/. This is useful to
313 enable factory resets and volatile systems that boot up with
314 an empty /etc directory, and thus need system users and
315 groups created during early boot. systemd now also ships
316 with two default sysusers.d/ files for the most basic
317 users and groups systemd and the core operating system
320 * A new tmpfiles snippet has been added that rebuilds the
321 essential files in /etc on boot, should they be missing.
323 * A directive for ensuring automatic clean-up of
324 /var/cache/man/ has been removed from the default
325 configuration. This line should now be shipped by the man
326 implementation. The necessary change has been made to the
327 man-db implementation. Note that you need to update your man
328 implementation to one that ships this line, otherwise no
329 automatic clean-up of /var/cache/man will take place.
331 * A new condition ConditionNeedsUpdate= has been added that
332 may conditionalize services to only run when /etc or /var
333 are "older" than the vendor operating system resources in
334 /usr. This is useful for reconstructing or updating /etc
335 after an offline update of /usr or a factory reset, on the
336 next reboot. Services that want to run once after such an
337 update or reset should use this condition and order
338 themselves before the new systemd-update-done.service, which
339 will mark the two directories as fully updated. A number of
340 service files have been added making use of this, to rebuild
341 the udev hardware database, the journald message catalog and
342 dynamic loader cache (ldconfig). The systemd-sysusers tool
343 described above also makes use of this now. With this in
344 place it is now possible to start up a minimal operating
345 system with /etc empty cleanly. For more information on the
346 concepts involved see this recent blog story:
348 http://0pointer.de/blog/projects/stateless.html
350 * A new system group "input" has been introduced, and all
351 input device nodes get this group assigned. This is useful
352 for system-level software to get access to input devices. It
353 complements what is already done for "audio" and "video".
355 * systemd-networkd learnt minimal DHCPv4 server support in
356 addition to the existing DHCPv4 client support. It also
357 learnt DHCPv6 client and IPv6 Router Solicitation client
358 support. The DHCPv4 client gained support for static routes
359 passed in from the server. Note that the [DHCPv4] section
360 known in older systemd-networkd versions has been renamed to
361 [DHCP] and is now also used by the DHCPv6 client. Existing
362 .network files using settings of this section should be
363 updated, though compatibility is maintained. Optionally, the
364 client hostname may now be sent to the DHCP server.
366 * networkd gained support for vxlan virtual networks as well
367 as tun/tap and dummy devices.
369 * networkd gained support for automatic allocation of address
370 ranges for interfaces from a system-wide pool of
371 addresses. This is useful for dynamically managing a large
372 number of interfaces with a single network configuration
373 file. In particular this is useful to easily assign
374 appropriate IP addresses to the veth links of a large number
377 * RPM macros for processing sysusers, sysctl and binfmt
378 drop-in snippets at package installation time have been
381 * The /etc/os-release file should now be placed in
382 /usr/lib/os-release. The old location is automatically
383 created as symlink. /usr/lib is the more appropriate
384 location of this file, since it shall actually describe the
385 vendor operating system shipped in /usr, and not the
386 configuration stored in /etc.
388 * .mount units gained a new boolean SloppyOptions= setting
389 that maps to mount(8)'s -s option which enables permissive
390 parsing of unknown mount options.
392 * tmpfiles learnt a new "L+" directive which creates a symlink
393 but (unlike "L") deletes a pre-existing file first, should
394 it already exist and not already be the correct
395 symlink. Similar, "b+", "c+" and "p+" directives have been
396 added as well, which create block and character devices, as
397 well as fifos in the filesystem, possibly removing any
398 pre-existing files of different types.
400 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
401 'argument' field (which so far specified the source to
402 symlink/copy the files from) is now optional. If omitted the
403 same file os copied from /usr/share/factory/ suffixed by the
404 full destination path. This is useful for populating /etc
405 with essential files, by copying them from vendor defaults
406 shipped in /usr/share/factory/etc.
408 * A new command "systemctl preset-all" has been added that
409 applies the service preset settings to all installed unit
410 files. A new switch --preset-mode= has been added that
411 controls whether only enable or only disable operations
414 * A new command "systemctl is-system-running" has been added
415 that allows checking the overall state of the system, for
416 example whether it is fully up and running.
418 * When the system boots up with an empty /etc, the equivalent
419 to "systemctl preset-all" is executed during early boot, to
420 make sure all default services are enabled after a factory
423 * systemd now contains a minimal preset file that enables the
424 most basic services systemd ships by default.
426 * Unit files' [Install] section gained a new DefaultInstance=
427 field for defining the default instance to create if a
428 template unit is enabled with no instance specified.
430 * A new passive target cryptsetup-pre.target has been added
431 that may be used by services that need to make they run and
432 finish before the first LUKS cryptographic device is set up.
434 * The /dev/loop-control and /dev/btrfs-control device nodes
435 are now owned by the "disk" group by default, opening up
436 access to this group.
438 * systemd-coredump will now automatically generate a
439 stack trace of all core dumps taking place on the system,
440 based on elfutils' libdw library. This stack trace is logged
443 * systemd-coredump may now optionally store coredumps directly
444 on disk (in /var/lib/systemd/coredump, possibly compressed),
445 instead of storing them unconditionally in the journal. This
446 mode is the new default. A new configuration file
447 /etc/systemd/coredump.conf has been added to configure this
448 and other parameters of systemd-coredump.
450 * coredumpctl gained a new "info" verb to show details about a
451 specific coredump. A new switch "-1" has also been added
452 that makes sure to only show information about the most
453 recent entry instead of all entries. Also, as the tool is
454 generally useful now the "systemd-" prefix of the binary
455 name has been removed. Distributions that want to maintain
456 compatibility with the old name should add a symlink from
457 the old name to the new name.
459 * journald's SplitMode= now defaults to "uid". This makes sure
460 that unprivileged users can access their own coredumps with
461 coredumpctl without restrictions.
463 * New kernel command line options "systemd.wants=" (for
464 pulling an additional unit during boot), "systemd.mask="
465 (for masking a specific unit for the boot), and
466 "systemd.debug-shell" (for enabling the debug shell on tty9)
467 have been added. This is implemented in the new generator
468 "systemd-debug-generator".
470 * systemd-nspawn will now by default filter a couple of
471 syscalls for containers, among them those required for
472 kernel module loading, direct x86 IO port access, swap
473 management, and kexec. Most importantly though
474 open_by_handle_at() is now prohibited for containers,
475 closing a hole similar to a recently discussed vulnerability
476 in docker regarding access to files on file hierarchies the
477 container should normally not have access to. Note that for
478 nspawn we generally make no security claims anyway (and
479 this is explicitly documented in the man page), so this is
480 just a fix for one of the most obvious problems.
482 * A new man page file-hierarchy(7) has been added that
483 contains a minimized, modernized version of the file system
484 layout systemd expects, similar in style to the FHS
485 specification or hier(5). A new tool systemd-path(1) has
486 been added to query many of these paths for the local
489 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
490 longer done. Since the directory now has a per-user size
491 limit, and is cleaned on logout this appears unnecessary,
492 in particular since this now brings the lifecycle of this
493 directory closer in line with how IPC objects are handled.
495 * systemd.pc now exports a number of additional directories,
496 including $libdir (which is useful to identify the library
497 path for the primary architecture of the system), and a
498 couple of drop-in directories.
500 * udev's predictable network interface names now use the dev_port
501 sysfs attribute, introduced in linux 3.15 instead of dev_id to
502 distinguish between ports of the same PCI function. dev_id should
503 only be used for ports using the same HW address, hence the need
506 * machined has been updated to export the OS version of a
507 container (read from /etc/os-release and
508 /usr/lib/os-release) on the bus. This is now shown in
509 "machinectl status" for a machine.
511 * A new service setting RestartForceExitStatus= has been
512 added. If configured to a set of exit signals or process
513 return values, the service will be restarted when the main
514 daemon process exits with any of them, regardless of the
517 * systemctl's -H switch for connecting to remote systemd
518 machines has been extended so that it may be used to
519 directly connect to a specific container on the
520 host. "systemctl -H root@foobar:waldi" will now connect as
521 user "root" to host "foobar", and then proceed directly to
522 the container named "waldi". Note that currently you have to
523 authenticate as user "root" for this to work, as entering
524 containers is a privileged operation.
526 Contributions from: Andreas Henriksson, Benjamin Steinwender,
527 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
528 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
529 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
530 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
531 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
532 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
533 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
534 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
535 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
536 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
537 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
539 -- Berlin, 2014-07-03
543 * As an experimental feature, udev now tries to lock the
544 disk device node (flock(LOCK_SH|LOCK_NB)) while it
545 executes events for the disk or any of its partitions.
546 Applications like partitioning programs can lock the
547 disk device node (flock(LOCK_EX)) and claim temporary
548 device ownership that way; udev will entirely skip all event
549 handling for this disk and its partitions. If the disk
550 was opened for writing, the close will trigger a partition
551 table rescan in udev's "watch" facility, and if needed
552 synthesize "change" events for the disk and all its partitions.
553 This is now unconditionally enabled, and if it turns out to
554 cause major problems, we might turn it on only for specific
555 devices, or might need to disable it entirely. Device Mapper
556 devices are excluded from this logic.
558 * We temporarily dropped the "-l" switch for fsck invocations,
559 since they collide with the flock() logic above. util-linux
560 upstream has been changed already to avoid this conflict,
561 and we will readd "-l" as soon as util-linux with this
562 change has been released.
564 * The dependency on libattr has been removed. Since a long
565 time, the extended attribute calls have moved to glibc, and
566 libattr is thus unnecessary.
568 * Virtualization detection works without priviliges now. This
569 means the systemd-detect-virt binary no longer requires
570 CAP_SYS_PTRACE file capabilities, and our daemons can run
571 with fewer privileges.
573 * systemd-networkd now runs under its own "systemd-network"
574 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
575 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
576 loses the ability to write to files owned by root this way.
578 * Similar, systemd-resolved now runs under its own
579 "systemd-resolve" user with no capabilities remaining.
581 * Similar, systemd-bus-proxyd now runs under its own
582 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
584 * systemd-networkd gained support for setting up "veth"
585 virtual ethernet devices for container connectivity, as well
586 as GRE and VTI tunnels.
588 * systemd-networkd will no longer automatically attempt to
589 manually load kernel modules necessary for certain tunnel
590 transports. Instead, it is assumed the kernel loads them
591 automatically when required. This only works correctly on
592 very new kernels. On older kernels, please consider adding
593 the kernel modules to /etc/modules-load.d/ as a work-around.
595 * The resolv.conf file systemd-resolved generates has been
596 moved to /run/systemd/resolve/. If you have a symlink from
597 /etc/resolv.conf, it might be necessary to correct it.
599 * Two new service settings, ProtectHome= and ProtectSystem=,
600 have been added. When enabled, they will make the user data
601 (such as /home) inaccessible or read-only and the system
602 (such as /usr) read-only, for specific services. This allows
603 very light-weight per-service sandboxing to avoid
604 modifications of user data or system files from
605 services. These two new switches have been enabled for all
606 of systemd's long-running services, where appropriate.
608 * Socket units gained new SocketUser= and SocketGroup=
609 settings to set the owner user and group of AF_UNIX sockets
610 and FIFOs in the file system.
612 * Socket units gained a new RemoveOnStop= setting. If enabled,
613 all FIFOS and sockets in the file system will be removed
614 when the specific socket unit is stopped.
616 * Socket units gained a new Symlinks= setting. It takes a list
617 of symlinks to create to file system sockets or FIFOs
618 created by the specific Unix sockets. This is useful to
619 manage symlinks to socket nodes with the same life-cycle as
622 * The /dev/log socket and /dev/initctl FIFO have been moved to
623 /run, and have been replaced by symlinks. This allows
624 connecting to these facilities even if PrivateDevices=yes is
625 used for a service (which makes /dev/log itself unavailable,
626 but /run is left). This also has the benefit of ensuring
627 that /dev only contains device nodes, directories and
628 symlinks, and nothing else.
630 * sd-daemon gained two new calls sd_pid_notify() and
631 sd_pid_notifyf(). They are similar to sd_notify() and
632 sd_notifyf(), but allow overriding of the source PID of
633 notification messages if permissions permit this. This is
634 useful to send notify messages on behalf of a different
635 process (for example, the parent process). The
636 systemd-notify tool has been updated to make use of this
637 when sending messages (so that notification messages now
638 originate from the shell script invoking systemd-notify and
639 not the systemd-notify process itself. This should minimize
640 a race where systemd fails to associate notification
641 messages to services when the originating process already
644 * A new "on-abnormal" setting for Restart= has been added. If
645 set, it will result in automatic restarts on all "abnormal"
646 reasons for a process to exit, which includes unclean
647 signals, core dumps, timeouts and watchdog timeouts, but
648 does not include clean and unclean exit codes or clean
649 signals. Restart=on-abnormal is an alternative for
650 Restart=on-failure for services that shall be able to
651 terminate and avoid restarts on certain errors, by
652 indicating so with an unclean exit code. Restart=on-failure
653 or Restart=on-abnormal is now the recommended setting for
654 all long-running services.
656 * If the InaccessibleDirectories= service setting points to a
657 mount point (or if there are any submounts contained within
658 it), it is now attempted to completely unmount it, to make
659 the file systems truly unavailable for the respective
662 * The ReadOnlyDirectories= service setting and
663 systemd-nspawn's --read-only parameter are now recursively
664 applied to all submounts, too.
666 * Mount units may now be created transiently via the bus APIs.
668 * The support for SysV and LSB init scripts has been removed
669 from the systemd daemon itself. Instead, it is now
670 implemented as a generator that creates native systemd units
671 from these scripts when needed. This enables us to remove a
672 substantial amount of legacy code from PID 1, following the
673 fact that many distributions only ship a very small number
674 of LSB/SysV init scripts nowadays.
676 * Priviliged Xen (dom0) domains are not considered
677 virtualization anymore by the virtualization detection
678 logic. After all, they generally have unrestricted access to
679 the hardware and usually are used to manage the unprivileged
682 * systemd-tmpfiles gained a new "C" line type, for copying
683 files or entire directories.
685 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
686 lines. So far, they have been non-globbing versions of the
687 latter, and have thus been redundant. In future, it is
688 recommended to only use "z". "m" has hence been removed
689 from the documentation, even though it stays supported.
691 * A tmpfiles snippet to recreate the most basic structure in
692 /var has been added. This is enough to create the /var/run →
693 /run symlink and create a couple of structural
694 directories. This allows systems to boot up with an empty or
695 volatile /var. Of course, while with this change, the core OS
696 now is capable with dealing with a volatile /var, not all
697 user services are ready for it. However, we hope that sooner
698 or later, many service daemons will be changed upstream so
699 that they are able to automatically create their necessary
700 directories in /var at boot, should they be missing. This is
701 the first step to allow state-less systems that only require
702 the vendor image for /usr to boot.
704 * systemd-nspawn has gained a new --tmpfs= switch to mount an
705 empty tmpfs instance to a specific directory. This is
706 particularly useful for making use of the automatic
707 reconstruction of /var (see above), by passing --tmpfs=/var.
709 * Access modes specified in tmpfiles snippets may now be
710 prefixed with "~", which indicates that they shall be masked
711 by whether the existing file or directory is currently
712 writable, readable or executable at all. Also, if specified,
713 the sgid/suid/sticky bits will be masked for all
716 * A new passive target unit "network-pre.target" has been
717 added which is useful for services that shall run before any
718 network is configured, for example firewall scripts.
720 * The "floppy" group that previously owned the /dev/fd*
721 devices is no longer used. The "disk" group is now used
722 instead. Distributions should probably deprecate usage of
725 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
726 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
727 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
728 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
729 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
730 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
733 -- Berlin, 2014-06-11
737 * A new "systemd-timesyncd" daemon has been added for
738 synchronizing the system clock across the network. It
739 implements an SNTP client. In contrast to NTP
740 implementations such as chrony or the NTP reference server,
741 this only implements a client side, and does not bother with
742 the full NTP complexity, focusing only on querying time from
743 one remote server and synchronizing the local clock to
744 it. Unless you intend to serve NTP to networked clients or
745 want to connect to local hardware clocks, this simple NTP
746 client should be more than appropriate for most
747 installations. The daemon runs with minimal privileges, and
748 has been hooked up with networkd to only operate when
749 network connectivity is available. The daemon saves the
750 current clock to disk every time a new NTP sync has been
751 acquired, and uses this to possibly correct the system clock
752 early at bootup, in order to accommodate for systems that
753 lack an RTC such as the Raspberry Pi and embedded devices,
754 and to make sure that time monotonically progresses on these
755 systems, even if it is not always correct. To make use of
756 this daemon, a new system user and group "systemd-timesync"
757 needs to be created on installation of systemd.
759 * The queue "seqnum" interface of libudev has been disabled, as
760 it was generally incompatible with device namespacing as
761 sequence numbers of devices go "missing" if the devices are
762 part of a different namespace.
764 * "systemctl list-timers" and "systemctl list-sockets" gained
765 a --recursive switch for showing units of these types also
766 for all local containers, similar in style to the already
767 supported --recursive switch for "systemctl list-units".
769 * A new RebootArgument= setting has been added for service
770 units, which may be used to specify a kernel reboot argument
771 to use when triggering reboots with StartLimitAction=.
773 * A new FailureAction= setting has been added for service
774 units which may be used to specify an operation to trigger
775 when a service fails. This works similarly to
776 StartLimitAction=, but unlike it, controls what is done
777 immediately rather than only after several attempts to
778 restart the service in question.
780 * hostnamed got updated to also expose the kernel name,
781 release, and version on the bus. This is useful for
782 executing commands like hostnamectl with the -H switch.
783 systemd-analyze makes use of this to properly display
784 details when running non-locally.
786 * The bootchart tool can now show cgroup information in the
789 * The CFS CPU quota cgroup attribute is now exposed for
790 services. The new CPUQuota= switch has been added for this
791 which takes a percentage value. Setting this will have the
792 result that a service may never get more CPU time than the
793 specified percentage, even if the machine is otherwise idle.
795 * systemd-networkd learned IPIP and SIT tunnel support.
797 * LSB init scripts exposing a dependency on $network will now
798 get a dependency on network-online.target rather than simply
799 network.target. This should bring LSB handling closer to
800 what it was on SysV systems.
802 * A new fsck.repair= kernel option has been added to control
803 how fsck shall deal with unclean file systems at boot.
805 * The (.ini) configuration file parser will now silently
806 ignore sections whose name begins with "X-". This may be
807 used to maintain application-specific extension sections in unit
810 * machined gained a new API to query the IP addresses of
811 registered containers. "machinectl status" has been updated
812 to show these addresses in its output.
814 * A new call sd_uid_get_display() has been added to the
815 sd-login APIs for querying the "primary" session of a
816 user. The "primary" session of the user is elected from the
817 user's sessions and generally a graphical session is
818 preferred over a text one.
820 * A minimal systemd-resolved daemon has been added. It
821 currently simply acts as a companion to systemd-networkd and
822 manages resolv.conf based on per-interface DNS
823 configuration, possibly supplied via DHCP. In the long run
824 we hope to extend this into a local DNSSEC enabled DNS and
827 * The systemd-networkd-wait-online tool is now enabled by
828 default. It will delay network-online.target until a network
829 connection has been configured. The tool primarily integrates
830 with networkd, but will also make a best effort to make sense
831 of network configuration performed in some other way.
833 * Two new service options StartupCPUShares= and
834 StartupBlockIOWeight= have been added that work similarly to
835 CPUShares= and BlockIOWeight= however only apply during
836 system startup. This is useful to prioritize certain services
837 differently during bootup than during normal runtime.
839 * hostnamed has been changed to prefer the statically
840 configured hostname in /etc/hostname (unless set to
841 'localhost' or empty) over any dynamic one supplied by
842 dhcp. With this change, the rules for picking the hostname
843 match more closely the rules of other configuration settings
844 where the local administrator's configuration in /etc always
845 overrides any other settings.
847 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
848 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
849 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
850 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
851 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
852 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
853 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
854 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
855 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
856 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
857 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
858 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
859 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
860 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
861 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
862 Lindskog, WaLyong Cho, Will Woods, Zbigniew
865 -- Beijing, 2014-05-28
869 * When restoring the screen brightness at boot, stay away from
870 the darkest setting or from the lowest 5% of the available
871 range, depending on which is the larger value of both. This
872 should effectively protect the user from rebooting into a
873 black screen, should the brightness have been set to minimum
876 * sd-login gained a new sd_machine_get_class() call to
877 determine the class ("vm" or "container") of a machine
878 registered with machined.
880 * sd-login gained new calls
881 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
882 to query the identity of the peer of a local AF_UNIX
883 connection. They operate similarly to their sd_pid_get_xyz()
886 * PID 1 will now maintain a system-wide system state engine
887 with the states "starting", "running", "degraded",
888 "maintenance", "stopping". These states are bound to system
889 startup, normal runtime, runtime with at least one failed
890 service, rescue/emergency mode and system shutdown. This
891 state is shown in the "systemctl status" output when no unit
892 name is passed. It is useful to determine system state, in
893 particularly when doing so for many systems or containers at
896 * A new command "list-machines" has been added to "systemctl"
897 that lists all local OS containers and shows their system
898 state (see above), if systemd runs inside of them.
900 * systemctl gained a new "-r" switch to recursively enumerate
901 units on all local containers, when used with the
902 "list-unit" command (which is the default one that is
903 executed when no parameters are specified).
905 * The GPT automatic partition discovery logic will now honour
906 two GPT partition flags: one may be set on a partition to
907 cause it to be mounted read-only, and the other may be set
908 on a partition to ignore it during automatic discovery.
910 * Two new GPT type UUIDs have been added for automatic root
911 partition discovery, for 32-bit and 64-bit ARM. This is not
912 particularly useful for discovering the root directory on
913 these architectures during bare-metal boots (since UEFI is
914 not common there), but still very useful to allow booting of
915 ARM disk images in nspawn with the -i option.
917 * MAC addresses of interfaces created with nspawn's
918 --network-interface= switch will now be generated from the
919 machine name, and thus be stable between multiple invocations
922 * logind will now automatically remove all IPC objects owned
923 by a user if she or he fully logs out. This makes sure that
924 users who are logged out cannot continue to consume IPC
925 resources. This covers SysV memory, semaphores and message
926 queues as well as POSIX shared memory and message
927 queues. Traditionally, SysV and POSIX IPC had no life-cycle
928 limits. With this functionality, that is corrected. This may
929 be turned off by using the RemoveIPC= switch of logind.conf.
931 * The systemd-machine-id-setup and tmpfiles tools gained a
932 --root= switch to operate on a specific root directory,
935 * journald can now forward logged messages to the TTYs of all
936 logged in users ("wall"). This is the default for all
937 emergency messages now.
939 * A new tool systemd-journal-remote has been added to stream
940 journal log messages across the network.
942 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
943 controller trees are mounted into it. Note that the
944 directories mounted beneath it are not read-only. This is a
945 security measure and is particularly useful because glibc
946 actually includes a search logic to pick any tmpfs it can
947 find to implement shm_open() if /dev/shm is not available
948 (which it might very well be in namespaced setups).
950 * machinectl gained a new "poweroff" command to cleanly power
951 down a local OS container.
953 * The PrivateDevices= unit file setting will now also drop the
954 CAP_MKNOD capability from the capability bound set, and
955 imply DevicePolicy=closed.
957 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
958 comprehensively on all long-running systemd services where
961 * systemd-udevd will now run in a disassociated mount
962 namespace. To mount directories from udev rules, make sure to
963 pull in mount units via SYSTEMD_WANTS properties.
965 * The kdbus support gained support for uploading policy into
966 the kernel. sd-bus gained support for creating "monitoring"
967 connections that can eavesdrop into all bus communication
968 for debugging purposes.
970 * Timestamps may now be specified in seconds since the UNIX
971 epoch Jan 1st, 1970 by specifying "@" followed by the value
974 * Native tcpwrap support in systemd has been removed. tcpwrap
975 is old code, not really maintained anymore and has serious
976 shortcomings, and better options such as firewalls
977 exist. For setups that require tcpwrap usage, please
978 consider invoking your socket-activated service via tcpd,
979 like on traditional inetd.
981 * A new system.conf configuration option
982 DefaultTimerAccuracySec= has been added that controls the
983 default AccuracySec= setting of .timer units.
985 * Timer units gained a new WakeSystem= switch. If enabled,
986 timers configured this way will cause the system to resume
987 from system suspend (if the system supports that, which most
990 * Timer units gained a new Persistent= switch. If enabled,
991 timers configured this way will save to disk when they have
992 been last triggered. This information is then used on next
993 reboot to possible execute overdue timer events, that
994 could not take place because the system was powered off.
995 This enables simple anacron-like behaviour for timer units.
997 * systemctl's "list-timers" will now also list the time a
998 timer unit was last triggered in addition to the next time
999 it will be triggered.
1001 * systemd-networkd will now assign predictable IPv4LL
1002 addresses to its local interfaces.
1004 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1005 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1006 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1007 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1008 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1009 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1010 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1011 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1014 -- Berlin, 2014-03-25
1018 * A new unit file setting RestrictAddressFamilies= has been
1019 added to restrict which socket address families unit
1020 processes gain access to. This takes address family names
1021 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1022 attack surface of services via exotic protocol stacks. This
1023 is built on seccomp system call filters.
1025 * Two new unit file settings RuntimeDirectory= and
1026 RuntimeDirectoryMode= have been added that may be used to
1027 manage a per-daemon runtime directories below /run. This is
1028 an alternative for setting up directory permissions with
1029 tmpfiles snippets, and has the advantage that the runtime
1030 directory's lifetime is bound to the daemon runtime and that
1031 the daemon starts up with an empty directory each time. This
1032 is particularly useful when writing services that drop
1033 privileges using the User= or Group= setting.
1035 * The DeviceAllow= unit setting now supports globbing for
1036 matching against device group names.
1038 * The systemd configuration file system.conf gained new
1039 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1040 DefaultMemoryAccounting= to globally turn on/off accounting
1041 for specific resources (cgroups) for all units. These
1042 settings may still be overridden individually in each unit
1045 * systemd-gpt-auto-generator is now able to discover /srv and
1046 root partitions in addition to /home and swap partitions. It
1047 also supports LUKS-encrypted partitions now. With this in
1048 place, automatic discovery of partitions to mount following
1049 the Discoverable Partitions Specification
1050 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1051 is now a lot more complete. This allows booting without
1052 /etc/fstab and without root= on the kernel command line on
1053 systems prepared appropriately.
1055 * systemd-nspawn gained a new --image= switch which allows
1056 booting up disk images and Linux installations on any block
1057 device that follow the Discoverable Partitions Specification
1058 (see above). This means that installations made with
1059 appropriately updated installers may now be started and
1060 deployed using container managers, completely
1061 unmodified. (We hope that libvirt-lxc will add support for
1062 this feature soon, too.)
1064 * systemd-nspawn gained a new --network-macvlan= setting to
1065 set up a private macvlan interface for the
1066 container. Similarly, systemd-networkd gained a new
1067 Kind=macvlan setting in .netdev files.
1069 * systemd-networkd now supports configuring local addresses
1072 * A new tool systemd-network-wait-online has been added to
1073 synchronously wait for network connectivity using
1076 * The sd-bus.h bus API gained a new sd_bus_track object for
1077 tracking the life-cycle of bus peers. Note that sd-bus.h is
1078 still not a public API though (unless you specify
1079 --enable-kdbus on the configure command line, which however
1080 voids your warranty and you get no API stability guarantee).
1082 * The $XDG_RUNTIME_DIR runtime directories for each user are
1083 now individual tmpfs instances, which has the benefit of
1084 introducing separate pools for each user, with individual
1085 size limits, and thus making sure that unprivileged clients
1086 can no longer negatively impact the system or other users by
1087 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1088 RuntimeDirectorySize= has been introduced that allows
1089 controlling the default size limit for all users. It
1090 defaults to 10% of the available physical memory. This is no
1091 replacement for quotas on tmpfs though (which the kernel
1092 still does not support), as /dev/shm and /tmp are still
1093 shared resources used by both the system and unprivileged
1096 * logind will now automatically turn off automatic suspending
1097 on laptop lid close when more than one display is
1098 connected. This was previously expected to be implemented
1099 individually in desktop environments (such as GNOME),
1100 however has been added to logind now, in order to fix a
1101 boot-time race where a desktop environment might not have
1102 been started yet and thus not been able to take an inhibitor
1103 lock at the time where logind already suspends the system
1104 due to a closed lid.
1106 * logind will now wait at least 30s after each system
1107 suspend/resume cycle, and 3min after system boot before
1108 suspending the system due to a closed laptop lid. This
1109 should give USB docking stations and similar enough time to
1110 be probed and configured after system resume and boot in
1111 order to then act as suspend blocker.
1113 * systemd-run gained a new --property= setting which allows
1114 initialization of resource control properties (and others)
1115 for the created scope or service unit. Example: "systemd-run
1116 --property=BlockIOWeight=10 updatedb" may be used to run
1117 updatedb at a low block IO scheduling weight.
1119 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1120 now also work in --scope mode.
1122 * When systemd is compiled with kdbus support, basic support
1123 for enforced policies is now in place. (Note that enabling
1124 kdbus still voids your warranty and no API compatibility
1127 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1128 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1129 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1130 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1131 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1132 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1133 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1134 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1135 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1136 Zbigniew Jędrzejewski-Szmek
1138 -- Berlin, 2014-03-12
1142 * systemd will now relabel /dev after loading the SMACK policy
1143 according to SMACK rules.
1145 * A new unit file option AppArmorProfile= has been added to
1146 set the AppArmor profile for the processes of a unit.
1148 * A new condition check ConditionArchitecture= has been added
1149 to conditionalize units based on the system architecture, as
1150 reported by uname()'s "machine" field.
1152 * systemd-networkd now supports matching on the system
1153 virtualization, architecture, kernel command line, host name
1156 * logind is now a lot more aggressive when suspending the
1157 machine due to a closed laptop lid. Instead of acting only
1158 on the lid close action, it will continuously watch the lid
1159 status and act on it. This is useful for laptops where the
1160 power button is on the outside of the chassis so that it can
1161 be reached without opening the lid (such as the Lenovo
1162 Yoga). On those machines, logind will now immediately
1163 re-suspend the machine if the power button has been
1164 accidentally pressed while the laptop was suspended and in a
1165 backpack or similar.
1167 * logind will now watch SW_DOCK switches and inhibit reaction
1168 to the lid switch if it is pressed. This means that logind
1169 will not suspend the machine anymore if the lid is closed
1170 and the system is docked, if the laptop supports SW_DOCK
1171 notifications via the input layer. Note that ACPI docking
1172 stations do not generate this currently. Also note that this
1173 logic is usually not fully sufficient and Desktop
1174 Environments should take a lid switch inhibitor lock when an
1175 external display is connected, as systemd will not watch
1178 * nspawn will now make use of the devices cgroup controller by
1179 default, and only permit creation of and access to the usual
1180 API device nodes like /dev/null or /dev/random, as well as
1181 access to (but not creation of) the pty devices.
1183 * We will now ship a default .network file for
1184 systemd-networkd that automatically configures DHCP for
1185 network interfaces created by nspawn's --network-veth or
1186 --network-bridge= switches.
1188 * systemd will now understand the usual M, K, G, T suffixes
1189 according to SI conventions (i.e. to the base 1000) when
1190 referring to throughput and hardware metrics. It will stay
1191 with IEC conventions (i.e. to the base 1024) for software
1192 metrics, according to what is customary according to
1193 Wikipedia. We explicitly document which base applies for
1194 each configuration option.
1196 * The DeviceAllow= setting in unit files now supports a syntax
1197 to whitelist an entire group of devices node majors at once,
1198 based on the /proc/devices listing. For example, with the
1199 string "char-pts", it is now possible to whitelist all
1200 current and future pseudo-TTYs at once.
1202 * sd-event learned a new "post" event source. Event sources of
1203 this type are triggered by the dispatching of any event
1204 source of a type that is not "post". This is useful for
1205 implementing clean-up and check event sources that are
1206 triggered by other work being done in the program.
1208 * systemd-networkd is no longer statically enabled, but uses
1209 the usual [Install] sections so that it can be
1210 enabled/disabled using systemctl. It still is enabled by
1213 * When creating a veth interface pair with systemd-nspawn, the
1214 host side will now be prefixed with "vb-" if
1215 --network-bridge= is used, and with "ve-" if --network-veth
1216 is used. This way, it is easy to distinguish these cases on
1217 the host, for example to apply different configuration to
1218 them with systemd-networkd.
1220 * The compatibility libraries for libsystemd-journal.so,
1221 libsystem-id128.so, libsystemd-login.so and
1222 libsystemd-daemon.so do not make use of IFUNC
1223 anymore. Instead, we now build libsystemd.so multiple times
1224 under these alternative names. This means that the footprint
1225 is drastically increased, but given that these are
1226 transitional compatibility libraries, this should not matter
1227 much. This change has been made necessary to support the ARM
1228 platform for these compatibility libraries, as the ARM
1229 toolchain is not really at the same level as the toolchain
1230 for other architectures like x86 and does not support
1231 IFUNC. Please make sure to use --enable-compat-libs only
1232 during a transitional period!
1234 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1235 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1236 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1237 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1238 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1239 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1240 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1241 Zbigniew Jędrzejewski-Szmek
1243 -- Berlin, 2014-02-24
1247 * A new component "systemd-networkd" has been added that can
1248 be used to configure local network interfaces statically or
1249 via DHCP. It is capable of bringing up bridges, VLANs, and
1250 bonding. Currently, no hook-ups for interactive network
1251 configuration are provided. Use this for your initrd,
1252 container, embedded, or server setup if you need a simple,
1253 yet powerful, network configuration solution. This
1254 configuration subsystem is quite nifty, as it allows wildcard
1255 hotplug matching in interfaces. For example, with a single
1256 configuration snippet, you can configure that all Ethernet
1257 interfaces showing up are automatically added to a bridge,
1258 or similar. It supports link-sensing and more.
1260 * A new tool "systemd-socket-proxyd" has been added which can
1261 act as a bidirectional proxy for TCP sockets. This is
1262 useful for adding socket activation support to services that
1263 do not actually support socket activation, including virtual
1264 machines and the like.
1266 * Add a new tool to save/restore rfkill state on
1269 * Save/restore state of keyboard backlights in addition to
1270 display backlights on shutdown/boot.
1272 * udev learned a new SECLABEL{} construct to label device
1273 nodes with a specific security label when they appear. For
1274 now, only SECLABEL{selinux} is supported, but the syntax is
1275 prepared for additional security frameworks.
1277 * udev gained a new scheme to configure link-level attributes
1278 from files in /etc/systemd/network/*.link. These files can
1279 match against MAC address, device path, driver name and type,
1280 and will apply attributes like the naming policy, link speed,
1281 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1282 address assignment policy (randomized, ...).
1284 * The configuration of network interface naming rules for
1285 "permanent interface names" has changed: a new NamePolicy=
1286 setting in the [Link] section of .link files determines the
1287 priority of possible naming schemes (onboard, slot, mac,
1288 path). The default value of this setting is determined by
1289 /usr/lib/net/links/99-default.link. Old
1290 80-net-name-slot.rules udev configuration file has been
1291 removed, so local configuration overriding this file should
1292 be adapated to override 99-default.link instead.
1294 * When the User= switch is used in a unit file, also
1295 initialize $SHELL= based on the user database entry.
1297 * systemd no longer depends on libdbus. All communication is
1298 now done with sd-bus, systemd's low-level bus library
1301 * kdbus support has been added to PID 1 itself. When kdbus is
1302 enabled, this causes PID 1 to set up the system bus and
1303 enable support for a new ".busname" unit type that
1304 encapsulates bus name activation on kdbus. It works a little
1305 bit like ".socket" units, except for bus names. A new
1306 generator has been added that converts classic dbus1 service
1307 activation files automatically into native systemd .busname
1310 * sd-bus: add a light-weight vtable implementation that allows
1311 defining objects on the bus with a simple static const
1312 vtable array of its methods, signals and properties.
1314 * systemd will not generate or install static dbus
1315 introspection data anymore to /usr/share/dbus-1/interfaces,
1316 as the precise format of these files is unclear, and
1317 nothing makes use of it.
1319 * A proxy daemon is now provided to proxy clients connecting
1320 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1321 compatibility with classic D-Bus.
1323 * A bus driver implementation has been added that supports the
1324 classic D-Bus bus driver calls on kdbus, also for
1325 compatibility purposes.
1327 * A new API "sd-event.h" has been added that implements a
1328 minimal event loop API built around epoll. It provides a
1329 couple of features that direct epoll usage is lacking:
1330 prioritization of events, scales to large numbers of timer
1331 events, per-event timer slack (accuracy), system-wide
1332 coalescing of timer events, exit handlers, watchdog
1333 supervision support using systemd's sd_notify() API, child
1336 * A new API "sd-rntl.h" has been added that provides an API
1337 around the route netlink interface of the kernel, similar in
1338 style to "sd-bus.h".
1340 * A new API "sd-dhcp-client.h" has been added that provides a
1341 small DHCPv4 client-side implementation. This is used by
1344 * There is a new kernel command line option
1345 "systemd.restore_state=0|1". When set to "0", none of the
1346 systemd tools will restore saved runtime state to hardware
1347 devices. More specifically, the rfkill and backlight states
1350 * The FsckPassNo= compatibility option in mount/service units
1351 has been removed. The fstab generator will now add the
1352 necessary dependencies automatically, and does not require
1353 PID1's support for that anymore.
1355 * journalctl gained a new switch, --list-boots, that lists
1356 recent boots with their times and boot IDs.
1358 * The various tools like systemctl, loginctl, timedatectl,
1359 busctl, systemd-run, ... have gained a new switch "-M" to
1360 connect to a specific, local OS container (as direct
1361 connection, without requiring SSH). This works on any
1362 container that is registered with machined, such as those
1363 created by libvirt-lxc or nspawn.
1365 * systemd-run and systemd-analyze also gained support for "-H"
1366 to connect to remote hosts via SSH. This is particularly
1367 useful for systemd-run because it enables queuing of jobs
1368 onto remote systems.
1370 * machinectl gained a new command "login" to open a getty
1371 login in any local container. This works with any container
1372 that is registered with machined (such as those created by
1373 libvirt-lxc or nspawn), and which runs systemd inside.
1375 * machinectl gained a new "reboot" command that may be used to
1376 trigger a reboot on a specific container that is registered
1377 with machined. This works on any container that runs an init
1378 system of some kind.
1380 * systemctl gained a new "list-timers" command to print a nice
1381 listing of installed timer units with the times they elapse
1384 * Alternative reboot() parameters may now be specified on the
1385 "systemctl reboot" command line and are passed to the
1386 reboot() system call.
1388 * systemctl gained a new --job-mode= switch to configure the
1389 mode to queue a job with. This is a more generic version of
1390 --fail, --irreversible, and --ignore-dependencies, which are
1391 still available but not advertised anymore.
1393 * /etc/systemd/system.conf gained new settings to configure
1394 various default timeouts of units, as well as the default
1395 start limit interval and burst. These may still be overridden
1398 * PID1 will now export on the bus profile data of the security
1399 policy upload process (such as the SELinux policy upload to
1402 * journald: when forwarding logs to the console, include
1403 timestamps (following the setting in
1404 /sys/module/printk/parameters/time).
1406 * OnCalendar= in timer units now understands the special
1407 strings "yearly" and "annually". (Both are equivalent)
1409 * The accuracy of timer units is now configurable with the new
1410 AccuracySec= setting. It defaults to 1min.
1412 * A new dependency type JoinsNamespaceOf= has been added that
1413 allows running two services within the same /tmp and network
1414 namespace, if PrivateNetwork= or PrivateTmp= are used.
1416 * A new command "cat" has been added to systemctl. It outputs
1417 the original unit file of a unit, and concatenates the
1418 contents of additional "drop-in" unit file snippets, so that
1419 the full configuration is shown.
1421 * systemctl now supports globbing on the various "list-xyz"
1422 commands, like "list-units" or "list-sockets", as well as on
1423 those commands which take multiple unit names.
1425 * journalctl's --unit= switch gained support for globbing.
1427 * All systemd daemons now make use of the watchdog logic so
1428 that systemd automatically notices when they hang.
1430 * If the $container_ttys environment variable is set,
1431 getty-generator will automatically spawn a getty for each
1432 listed tty. This is useful for container managers to request
1433 login gettys to be spawned on as many ttys as needed.
1435 * %h, %s, %U specifier support is not available anymore when
1436 used in unit files for PID 1. This is because NSS calls are
1437 not safe from PID 1. They stay available for --user
1438 instances of systemd, and as special case for the root user.
1440 * loginctl gained a new "--no-legend" switch to turn off output
1443 * The "sd-login.h" API gained three new calls:
1444 sd_session_is_remote(), sd_session_get_remote_user(),
1445 sd_session_get_remote_host() to query information about
1448 * The udev hardware database now also carries vendor/product
1449 information of SDIO devices.
1451 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1452 determine whether watchdog notifications are requested by
1455 * Socket-activated per-connection services now include a
1456 short description of the connection parameters in the
1459 * tmpfiles gained a new "--boot" option. When this is not used,
1460 only lines where the command character is not suffixed with
1461 "!" are executed. When this option is specified, those
1462 options are executed too. This partitions tmpfiles
1463 directives into those that can be safely executed at any
1464 time, and those which should be run only at boot (for
1465 example, a line that creates /run/nologin).
1467 * A new API "sd-resolve.h" has been added which provides a simple
1468 asynchronous wrapper around glibc NSS host name resolution
1469 calls, such as getaddrinfo(). In contrast to glibc's
1470 getaddrinfo_a(), it does not use signals. In contrast to most
1471 other asynchronous name resolution libraries, this one does
1472 not reimplement DNS, but reuses NSS, so that alternate
1473 host name resolution systems continue to work, such as mDNS,
1474 LDAP, etc. This API is based on libasyncns, but it has been
1475 cleaned up for inclusion in systemd.
1477 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1478 "sd-daemon.h" are no longer found in individual libraries
1479 libsystemd-journal.so, libsystemd-login.so,
1480 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1481 merged them into a single library, libsystemd.so, which
1482 provides all symbols. The reason for this is cyclic
1483 dependencies, as these libraries tend to use each other's
1484 symbols. So far, we have managed to workaround that by linking
1485 a copy of a good part of our code into each of these
1486 libraries again and again, which, however, makes certain
1487 things hard to do, like sharing static variables. Also, it
1488 substantially increases footprint. With this change, there
1489 is only one library for the basic APIs systemd
1490 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1491 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1492 library as well, however are subject to the --enable-kdbus
1493 switch (see below). Note that "sd-dhcp-client.h" is not part
1494 of this library (this is because it only consumes, never
1495 provides, services of/to other APIs). To make the transition
1496 easy from the separate libraries to the unified one, we
1497 provide the --enable-compat-libs compile-time switch which
1498 will generate stub libraries that are compatible with the
1499 old ones but redirect all calls to the new one.
1501 * All of the kdbus logic and the new APIs "sd-bus.h",
1502 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1503 and "sd-utf8.h" are compile-time optional via the
1504 "--enable-kdbus" switch, and they are not compiled in by
1505 default. To make use of kdbus, you have to explicitly enable
1506 the switch. Note however, that neither the kernel nor the
1507 userspace API for all of this is considered stable yet. We
1508 want to maintain the freedom to still change the APIs for
1509 now. By specifying this build-time switch, you acknowledge
1510 that you are aware of the instability of the current
1513 * Also, note that while kdbus is pretty much complete,
1514 it lacks one thing: proper policy support. This means you
1515 can build a fully working system with all features; however,
1516 it will be highly insecure. Policy support will be added in
1517 one of the next releases, at the same time that we will
1518 declare the APIs stable.
1520 * When the kernel command-line argument "kdbus" is specified,
1521 systemd will automatically load the kdbus.ko kernel module. At
1522 this stage of development, it is only useful for testing kdbus
1523 and should not be used in production. Note: if "--enable-kdbus"
1524 is specified, and the kdbus.ko kernel module is available, and
1525 "kdbus" is added to the kernel command line, the entire system
1526 runs with kdbus instead of dbus-daemon, with the above mentioned
1527 problem of missing the system policy enforcement. Also a future
1528 version of kdbus.ko or a newer systemd will not be compatible with
1529 each other, and will unlikely be able to boot the machine if only
1530 one of them is updated.
1532 * systemctl gained a new "import-environment" command which
1533 uploads the caller's environment (or parts thereof) into the
1534 service manager so that it is inherited by services started
1535 by the manager. This is useful to upload variables like
1536 $DISPLAY into the user service manager.
1538 * A new PrivateDevices= switch has been added to service units
1539 which allows running a service with a namespaced /dev
1540 directory that does not contain any device nodes for
1541 physical devices. More specifically, it only includes devices
1542 such as /dev/null, /dev/urandom, and /dev/zero which are API
1545 * logind has been extended to support behaviour like VT
1546 switching on seats that do not support a VT. This makes
1547 multi-session available on seats that are not the first seat
1548 (seat0), and on systems where kernel support for VTs has
1549 been disabled at compile-time.
1551 * If a process holds a delay lock for system sleep or shutdown
1552 and fails to release it in time, we will now log its
1553 identity. This makes it easier to identify processes that
1554 cause slow suspends or power-offs.
1556 * When parsing /etc/crypttab, support for a new key-slot=
1557 option as supported by Debian is added. It allows indicating
1558 which LUKS slot to use on disk, speeding up key loading.
1560 * The sd_journald_sendv() API call has been checked and
1561 officially declared to be async-signal-safe so that it may
1562 be invoked from signal handlers for logging purposes.
1564 * Boot-time status output is now enabled automatically after a
1565 short timeout if boot does not progress, in order to give
1566 the user an indication what she or he is waiting for.
1568 * The boot-time output has been improved to show how much time
1569 remains until jobs expire.
1571 * The KillMode= switch in service units gained a new possible
1572 value "mixed". If set, and the unit is shut down, then the
1573 initial SIGTERM signal is sent only to the main daemon
1574 process, while the following SIGKILL signal is sent to
1575 all remaining processes of the service.
1577 * When a scope unit is registered, a new property "Controller"
1578 may be set. If set to a valid bus name, systemd will send a
1579 RequestStop() signal to this name when it would like to shut
1580 down the scope. This may be used to hook manager logic into
1581 the shutdown logic of scope units. Also, scope units may now
1582 be put in a special "abandoned" state, in which case the
1583 manager process which created them takes no further
1584 responsibilities for it.
1586 * When reading unit files, systemd will now verify
1587 the access mode of these files, and warn about certain
1588 suspicious combinations. This has been added to make it
1589 easier to track down packaging bugs where unit files are
1590 marked executable or world-writable.
1592 * systemd-nspawn gained a new "--setenv=" switch to set
1593 container-wide environment variables. The similar option in
1594 systemd-activate was renamed from "--environment=" to
1595 "--setenv=" for consistency.
1597 * systemd-nspawn has been updated to create a new kdbus domain
1598 for each container that is invoked, thus allowing each
1599 container to have its own set of system and user buses,
1600 independent of the host.
1602 * systemd-nspawn gained a new --drop-capability= switch to run
1603 the container with less capabilities than the default. Both
1604 --drop-capability= and --capability= now take the special
1605 string "all" for dropping or keeping all capabilities.
1607 * systemd-nspawn gained new switches for executing containers
1608 with specific SELinux labels set.
1610 * systemd-nspawn gained a new --quiet switch to not generate
1611 any additional output but the container's own console
1614 * systemd-nspawn gained a new --share-system switch to run a
1615 container without PID namespacing enabled.
1617 * systemd-nspawn gained a new --register= switch to control
1618 whether the container is registered with systemd-machined or
1619 not. This is useful for containers that do not run full
1620 OS images, but only specific apps.
1622 * systemd-nspawn gained a new --keep-unit which may be used
1623 when invoked as the only program from a service unit, and
1624 results in registration of the unit service itself in
1625 systemd-machined, instead of a newly opened scope unit.
1627 * systemd-nspawn gained a new --network-interface= switch for
1628 moving arbitrary interfaces to the container. The new
1629 --network-veth switch creates a virtual Ethernet connection
1630 between host and container. The new --network-bridge=
1631 switch then allows assigning the host side of this virtual
1632 Ethernet connection to a bridge device.
1634 * systemd-nspawn gained a new --personality= switch for
1635 setting the kernel personality for the container. This is
1636 useful when running a 32-bit container on a 64-bit host. A
1637 similar option Personality= is now also available for service
1640 * logind will now also track a "Desktop" identifier for each
1641 session which encodes the desktop environment of it. This is
1642 useful for desktop environments that want to identify
1643 multiple running sessions of itself easily.
1645 * A new SELinuxContext= setting for service units has been
1646 added that allows setting a specific SELinux execution
1647 context for a service.
1649 * Most systemd client tools will now honour $SYSTEMD_LESS for
1650 settings of the "less" pager. By default, these tools will
1651 override $LESS to allow certain operations to work, such as
1652 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1653 influence this logic.
1655 * systemd's "seccomp" hook-up has been changed to make use of
1656 the libseccomp library instead of using its own
1657 implementation. This has benefits for portability among
1660 * For usage together with SystemCallFilter=, a new
1661 SystemCallErrorNumber= setting has been introduced that
1662 allows configuration of a system error number to be returned
1663 on filtered system calls, instead of immediately killing the
1664 process. Also, SystemCallArchitectures= has been added to
1665 limit access to system calls of a particular architecture
1666 (in order to turn off support for unused secondary
1667 architectures). There is also a global
1668 SystemCallArchitectures= setting in system.conf now to turn
1669 off support for non-native system calls system-wide.
1671 * systemd requires a kernel with a working name_to_handle_at(),
1672 please see the kernel config requirements in the README file.
1674 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1675 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1676 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1677 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1678 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1679 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1680 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1681 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1682 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1683 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1684 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1685 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1686 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1687 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1688 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1689 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1690 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1691 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1692 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1693 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1694 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1695 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1696 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1697 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1699 -- Berlin, 2014-02-20
1703 * logind has gained support for facilitating privileged input
1704 and drm device access for unprivileged clients. This work is
1705 useful to allow Wayland display servers (and similar
1706 programs, such as kmscon) to run under the user's ID and
1707 access input and drm devices which are normally
1708 protected. When this is used (and the kernel is new enough)
1709 logind will "mute" IO on the file descriptors passed to
1710 Wayland as long as it is in the background and "unmute" it
1711 if it returns into the foreground. This allows secure
1712 session switching without allowing background sessions to
1713 eavesdrop on input and display data. This also introduces
1714 session switching support if VT support is turned off in the
1715 kernel, and on seats that are not seat0.
1717 * A new kernel command line option luks.options= is understood
1718 now which allows specifying LUKS options for usage for LUKS
1719 encrypted partitions specified with luks.uuid=.
1721 * tmpfiles.d(5) snippets may now use specifier expansion in
1722 path names. More specifically %m, %b, %H, %v, are now
1723 replaced by the local machine id, boot id, hostname, and
1724 kernel version number.
1726 * A new tmpfiles.d(5) command "m" has been introduced which
1727 may be used to change the owner/group/access mode of a file
1728 or directory if it exists, but do nothing if it does not.
1730 * This release removes high-level support for the
1731 MemorySoftLimit= cgroup setting. The underlying kernel
1732 cgroup attribute memory.soft_limit= is currently badly
1733 designed and likely to be removed from the kernel API in its
1734 current form, hence we should not expose it for now.
1736 * The memory.use_hierarchy cgroup attribute is now enabled for
1737 all cgroups systemd creates in the memory cgroup
1738 hierarchy. This option is likely to be come the built-in
1739 default in the kernel anyway, and the non-hierarchial mode
1740 never made much sense in the intrinsically hierarchial
1743 * A new field _SYSTEMD_SLICE= is logged along with all journal
1744 messages containing the slice a message was generated
1745 from. This is useful to allow easy per-customer filtering of
1746 logs among other things.
1748 * systemd-journald will no longer adjust the group of journal
1749 files it creates to the "systemd-journal" group. Instead we
1750 rely on the journal directory to be owned by the
1751 "systemd-journal" group, and its setgid bit set, so that the
1752 kernel file system layer will automatically enforce that
1753 journal files inherit this group assignment. The reason for
1754 this change is that we cannot allow NSS look-ups from
1755 journald which would be necessary to resolve
1756 "systemd-journal" to a numeric GID, because this might
1757 create deadlocks if NSS involves synchronous queries to
1758 other daemons (such as nscd, or sssd) which in turn are
1759 logging clients of journald and might block on it, which
1760 would then dead lock. A tmpfiles.d(5) snippet included in
1761 systemd will make sure the setgid bit and group are
1762 properly set on the journal directory if it exists on every
1763 boot. However, we recommend adjusting it manually after
1764 upgrades too (or from RPM scriptlets), so that the change is
1765 not delayed until next reboot.
1767 * Backlight and random seed files in /var/lib/ have moved into
1768 the /var/lib/systemd/ directory, in order to centralize all
1769 systemd generated files in one directory.
1771 * Boot time performance measurements (as displayed by
1772 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1773 performance information if that's available to determine how
1774 much time BIOS and boot loader initialization required. With
1775 a sufficiently new BIOS you hence no longer need to boot
1776 with Gummiboot to get access to such information.
1778 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1779 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1780 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1781 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1782 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1783 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1784 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1786 -- Berlin, 2013-10-02
1790 * The Restart= option for services now understands a new
1791 on-watchdog setting, which will restart the service
1792 automatically if the service stops sending out watchdog keep
1793 alive messages (as configured with WatchdogSec=).
1795 * The getty generator (which is responsible for bringing up a
1796 getty on configured serial consoles) will no longer only
1797 start a getty on the primary kernel console but on all
1798 others, too. This makes the order in which console= is
1799 specified on the kernel command line less important.
1801 * libsystemd-logind gained a new sd_session_get_vt() call to
1802 retrieve the VT number of a session.
1804 * If the option "tries=0" is set for an entry of /etc/crypttab
1805 its passphrase is queried indefinitely instead of any
1806 maximum number of tries.
1808 * If a service with a configure PID file terminates its PID
1809 file will now be removed automatically if it still exists
1810 afterwards. This should put an end to stale PID files.
1812 * systemd-run will now also take relative binary path names
1813 for execution and no longer insists on absolute paths.
1815 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1816 paths that are optionally prefixed with "-" to indicate that
1817 it should not be considered a failure if they do not exist.
1819 * journalctl -o (and similar commands) now understands a new
1820 output mode "short-precise", it is similar to "short" but
1821 shows timestamps with usec accuracy.
1823 * The option "discard" (as known from Debian) is now
1824 synonymous to "allow-discards" in /etc/crypttab. In fact,
1825 "discard" is preferred now (since it is easier to remember
1828 * Some licensing clean-ups were made, so that more code is now
1829 LGPL-2.1 licensed than before.
1831 * A minimal tool to save/restore the display backlight
1832 brightness across reboots has been added. It will store the
1833 backlight setting as late as possible at shutdown, and
1834 restore it as early as possible during reboot.
1836 * A logic to automatically discover and enable home and swap
1837 partitions on GPT disks has been added. With this in place
1838 /etc/fstab becomes optional for many setups as systemd can
1839 discover certain partitions located on the root disk
1840 automatically. Home partitions are recognized under their
1841 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1842 partitions are recognized under their GPT type ID
1843 0657fd6da4ab43c484e50933c84b4f4f.
1845 * systemd will no longer pass any environment from the kernel
1846 or initrd to system services. If you want to set an
1847 environment for all services, do so via the kernel command
1848 line systemd.setenv= assignment.
1850 * The systemd-sysctl tool no longer natively reads the file
1851 /etc/sysctl.conf. If desired, the file should be symlinked
1852 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1853 legacy support by a symlink rather than built-in code, it
1854 also makes the otherwise hidden order of application of the
1855 different files visible. (Note that this partly reverts to a
1856 pre-198 application order of sysctl knobs!)
1858 * The "systemctl set-log-level" and "systemctl dump" commands
1859 have been moved to systemd-analyze.
1861 * systemd-run learned the new --remain-after-exit switch,
1862 which causes the scope unit not to be cleaned up
1863 automatically after the process terminated.
1865 * tmpfiles learned a new --exclude-prefix= switch to exclude
1866 certain paths from operation.
1868 * journald will now automatically flush all messages to disk
1869 as soon as a message of the log priorities CRIT, ALERT or
1872 Contributions from: Andrew Cook, Brandon Philips, Christian
1873 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1874 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1875 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1876 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1877 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1878 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1879 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1880 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1881 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1882 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1883 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1884 William Giokas, Zbigniew Jędrzejewski-Szmek
1886 -- Berlin, 2013-09-13
1890 * The documentation has been updated to cover the various new
1891 concepts introduced with 205.
1893 * Unit files now understand the new %v specifier which
1894 resolves to the kernel version string as returned by "uname
1897 * systemctl now supports filtering the unit list output by
1898 load state, active state and sub state, using the new
1901 * "systemctl status" will now show the results of the
1902 condition checks (like ConditionPathExists= and similar) of
1903 the last start attempts of the unit. They are also logged to
1906 * "journalctl -b" may now be used to look for boot output of a
1907 specific boot. Try "journalctl -b -1" for the previous boot,
1908 but the syntax is substantially more powerful.
1910 * "journalctl --show-cursor" has been added which prints the
1911 cursor string the last shown log line. This may then be used
1912 with the new "journalctl --after-cursor=" switch to continue
1913 browsing logs from that point on.
1915 * "journalctl --force" may now be used to force regeneration
1918 * Creation of "dead" device nodes has been moved from udev
1919 into kmod and tmpfiles. Previously, udev would read the kmod
1920 databases to pre-generate dead device nodes based on meta
1921 information contained in kernel modules, so that these would
1922 be auto-loaded on access rather then at boot. As this
1923 does not really have much to do with the exposing actual
1924 kernel devices to userspace this has always been slightly
1925 alien in the udev codebase. Following the new scheme kmod
1926 will now generate a runtime snippet for tmpfiles from the
1927 module meta information and it now is tmpfiles' job to the
1928 create the nodes. This also allows overriding access and
1929 other parameters for the nodes using the usual tmpfiles
1930 facilities. As side effect this allows us to remove the
1931 CAP_SYS_MKNOD capability bit from udevd entirely.
1933 * logind's device ACLs may now be applied to these "dead"
1934 devices nodes too, thus finally allowing managed access to
1935 devices such as /dev/snd/sequencer whithout loading the
1936 backing module right-away.
1938 * A new RPM macro has been added that may be used to apply
1939 tmpfiles configuration during package installation.
1941 * systemd-detect-virt and ConditionVirtualization= now can
1942 detect User-Mode-Linux machines (UML).
1944 * journald will now implicitly log the effective capabilities
1945 set of processes in the message metadata.
1947 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1949 * The initrd interface has been simplified (more specifically,
1950 support for passing performance data via environment
1951 variables and fsck results via files in /run has been
1952 removed). These features were non-essential, and are
1953 nowadays available in a much nicer way by having systemd in
1954 the initrd serialize its state and have the hosts systemd
1955 deserialize it again.
1957 * The udev "keymap" data files and tools to apply keyboard
1958 specific mappings of scan to key codes, and force-release
1959 scan code lists have been entirely replaced by a udev
1960 "keyboard" builtin and a hwdb data file.
1962 * systemd will now honour the kernel's "quiet" command line
1963 argument also during late shutdown, resulting in a
1964 completely silent shutdown when used.
1966 * There's now an option to control the SO_REUSEPORT socket
1967 option in .socket units.
1969 * Instance units will now automatically get a per-template
1970 subslice of system.slice unless something else is explicitly
1971 configured. For example, instances of sshd@.service will now
1972 implicitly be placed in system-sshd.slice rather than
1973 system.slice as before.
1975 * Test coverage support may now be enabled at build time.
1977 Contributions from: Dave Reisner, Frederic Crozat, Harald
1978 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1979 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1980 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1981 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1982 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1983 Giokas, Zbigniew Jędrzejewski-Szmek
1985 -- Berlin, 2013-07-23
1989 * Two new unit types have been introduced:
1991 Scope units are very similar to service units, however, are
1992 created out of pre-existing processes -- instead of PID 1
1993 forking off the processes. By using scope units it is
1994 possible for system services and applications to group their
1995 own child processes (worker processes) in a powerful way
1996 which then maybe used to organize them, or kill them
1997 together, or apply resource limits on them.
1999 Slice units may be used to partition system resources in an
2000 hierarchial fashion and then assign other units to them. By
2001 default there are now three slices: system.slice (for all
2002 system services), user.slice (for all user sessions),
2003 machine.slice (for VMs and containers).
2005 Slices and scopes have been introduced primarily in
2006 context of the work to move cgroup handling to a
2007 single-writer scheme, where only PID 1
2008 creates/removes/manages cgroups.
2010 * There's a new concept of "transient" units. In contrast to
2011 normal units these units are created via an API at runtime,
2012 not from configuration from disk. More specifically this
2013 means it is now possible to run arbitrary programs as
2014 independent services, with all execution parameters passed
2015 in via bus APIs rather than read from disk. Transient units
2016 make systemd substantially more dynamic then it ever was,
2017 and useful as a general batch manager.
2019 * logind has been updated to make use of scope and slice units
2020 for managing user sessions. As a user logs in he will get
2021 his own private slice unit, to which all sessions are added
2022 as scope units. We also added support for automatically
2023 adding an instance of user@.service for the user into the
2024 slice. Effectively logind will no longer create cgroup
2025 hierarchies on its own now, it will defer entirely to PID 1
2026 for this by means of scope, service and slice units. Since
2027 user sessions this way become entities managed by PID 1
2028 the output of "systemctl" is now a lot more comprehensive.
2030 * A new mini-daemon "systemd-machined" has been added which
2031 may be used by virtualization managers to register local
2032 VMs/containers. nspawn has been updated accordingly, and
2033 libvirt will be updated shortly. machined will collect a bit
2034 of meta information about the VMs/containers, and assign
2035 them their own scope unit (see above). The collected
2036 meta-data is then made available via the "machinectl" tool,
2037 and exposed in "ps" and similar tools. machined/machinectl
2038 is compile-time optional.
2040 * As discussed earlier, the low-level cgroup configuration
2041 options ControlGroup=, ControlGroupModify=,
2042 ControlGroupPersistent=, ControlGroupAttribute= have been
2043 removed. Please use high-level attribute settings instead as
2044 well as slice units.
2046 * A new bus call SetUnitProperties() has been added to alter
2047 various runtime parameters of a unit. This is primarily
2048 useful to alter cgroup parameters dynamically in a nice way,
2049 but will be extended later on to make more properties
2050 modifiable at runtime. systemctl gained a new set-properties
2051 command that wraps this call.
2053 * A new tool "systemd-run" has been added which can be used to
2054 run arbitrary command lines as transient services or scopes,
2055 while configuring a number of settings via the command
2056 line. This tool is currently very basic, however already
2057 very useful. We plan to extend this tool to even allow
2058 queuing of execution jobs with time triggers from the
2059 command line, similar in fashion to "at".
2061 * nspawn will now inform the user explicitly that kernels with
2062 audit enabled break containers, and suggest the user to turn
2065 * Support for detecting the IMA and AppArmor security
2066 frameworks with ConditionSecurity= has been added.
2068 * journalctl gained a new "-k" switch for showing only kernel
2069 messages, mimicking dmesg output; in addition to "--user"
2070 and "--system" switches for showing only user's own logs
2073 * systemd-delta can now show information about drop-in
2074 snippets extending unit files.
2076 * libsystemd-bus has been substantially updated but is still
2077 not available as public API.
2079 * systemd will now look for the "debug" argument on the kernel
2080 command line and enable debug logging, similar to what
2081 "systemd.log_level=debug" already did before.
2083 * "systemctl set-default", "systemctl get-default" has been
2084 added to configure the default.target symlink, which
2085 controls what to boot into by default.
2087 * "systemctl set-log-level" has been added as a convenient
2088 way to raise and lower systemd logging threshold.
2090 * "systemd-analyze plot" will now show the time the various
2091 generators needed for execution, as well as information
2092 about the unit file loading.
2094 * libsystemd-journal gained a new sd_journal_open_files() call
2095 for opening specific journal files. journactl also gained a
2096 new switch to expose this new functionality. Previously we
2097 only supported opening all files from a directory, or all
2098 files from the system, as opening individual files only is
2099 racy due to journal file rotation.
2101 * systemd gained the new DefaultEnvironment= setting in
2102 /etc/systemd/system.conf to set environment variables for
2105 * If a privileged process logs a journal message with the
2106 OBJECT_PID= field set, then journald will automatically
2107 augment this with additional OBJECT_UID=, OBJECT_GID=,
2108 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2109 system services want to log events about specific client
2110 processes. journactl/systemctl has been updated to make use
2111 of this information if all log messages regarding a specific
2114 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2115 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2116 Reisner, David Coppa, David King, David Strauss, Eelco
2117 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2118 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2119 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2120 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2121 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2122 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2123 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2124 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2125 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2126 Łukasz Stelmach, 장동준
2130 * The Python bindings gained some minimal support for the APIs
2131 exposed by libsystemd-logind.
2133 * ConditionSecurity= gained support for detecting SMACK. Since
2134 this condition already supports SELinux and AppArmor we only
2135 miss IMA for this. Patches welcome!
2137 Contributions from: Karol Lewandowski, Lennart Poettering,
2138 Zbigniew Jędrzejewski-Szmek
2142 * systemd-nspawn will now create /etc/resolv.conf if
2143 necessary, before bind-mounting the host's file onto it.
2145 * systemd-nspawn will now store meta information about a
2146 container on the container's cgroup as extended attribute
2147 fields, including the root directory.
2149 * The cgroup hierarchy has been reworked in many ways. All
2150 objects any of the components systemd creates in the cgroup
2151 tree are now suffixed. More specifically, user sessions are
2152 now placed in cgroups suffixed with ".session", users in
2153 cgroups suffixed with ".user", and nspawn containers in
2154 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2155 names are now escaped in a simple scheme to avoid collision
2156 of userspace object names with kernel filenames. This work
2157 is preparation for making these objects relocatable in the
2158 cgroup tree, in order to allow easy resource partitioning of
2159 these objects without causing naming conflicts.
2161 * systemctl list-dependencies gained the new switches
2162 --plain, --reverse, --after and --before.
2164 * systemd-inhibit now shows the process name of processes that
2165 have taken an inhibitor lock.
2167 * nss-myhostname will now also resolve "localhost"
2168 implicitly. This makes /etc/hosts an optional file and
2169 nicely handles that on IPv6 ::1 maps to both "localhost" and
2172 * libsystemd-logind.so gained a new call
2173 sd_get_machine_names() to enumerate running containers and
2174 VMs (currently only supported by very new libvirt and
2175 nspawn). sd_login_monitor can now be used to watch
2176 VMs/containers coming and going.
2178 * .include is not allowed recursively anymore, and only in
2179 unit files. Usually it is better to use drop-in snippets in
2180 .d/*.conf anyway, as introduced with systemd 198.
2182 * systemd-analyze gained a new "critical-chain" command that
2183 determines the slowest chain of units run during system
2184 boot-up. It is very useful for tracking down where
2185 optimizing boot time is the most beneficial.
2187 * systemd will no longer allow manipulating service paths in
2188 the name=systemd:/system cgroup tree using ControlGroup= in
2189 units. (But is still fine with it in all other dirs.)
2191 * There's a new systemd-nspawn@.service service file that may
2192 be used to easily run nspawn containers as system
2193 services. With the container's root directory in
2194 /var/lib/container/foobar it is now sufficient to run
2195 "systemctl start systemd-nspawn@foobar.service" to boot it.
2197 * systemd-cgls gained a new parameter "--machine" to list only
2198 the processes within a certain container.
2200 * ConditionSecurity= now can check for "apparmor". We still
2201 are lacking checks for SMACK and IMA for this condition
2202 check though. Patches welcome!
2204 * A new configuration file /etc/systemd/sleep.conf has been
2205 added that may be used to configure which kernel operation
2206 systemd is supposed to execute when "suspend", "hibernate"
2207 or "hybrid-sleep" is requested. This makes the new kernel
2208 "freeze" state accessible to the user.
2210 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2211 the passed argument if applicable.
2213 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2214 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2215 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2216 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2217 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2218 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2219 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2224 * The output of 'systemctl list-jobs' got some polishing. The
2225 '--type=' argument may now be passed more than once. A new
2226 command 'systemctl list-sockets' has been added which shows
2227 a list of kernel sockets systemd is listening on with the
2228 socket units they belong to, plus the units these socket
2231 * The experimental libsystemd-bus library got substantial
2232 updates to work in conjunction with the (also experimental)
2233 kdbus kernel project. It works well enough to exchange
2234 messages with some sophistication. Note that kdbus is not
2235 ready yet, and the library is mostly an elaborate test case
2236 for now, and not installable.
2238 * systemd gained a new unit 'systemd-static-nodes.service'
2239 that generates static device nodes earlier during boot, and
2240 can run in conjunction with udev.
2242 * libsystemd-login gained a new call sd_pid_get_user_unit()
2243 to retrieve the user systemd unit a process is running
2244 in. This is useful for systems where systemd is used as
2247 * systemd-nspawn now places all containers in the new /machine
2248 top-level cgroup directory in the name=systemd
2249 hierarchy. libvirt will soon do the same, so that we get a
2250 uniform separation of /system, /user and /machine for system
2251 services, user processes and containers/virtual
2252 machines. This new cgroup hierarchy is also useful to stick
2253 stable names to specific container instances, which can be
2254 recognized later this way (this name may be controlled
2255 via systemd-nspawn's new -M switch). libsystemd-login also
2256 gained a new call sd_pid_get_machine_name() to retrieve the
2257 name of the container/VM a specific process belongs to.
2259 * bootchart can now store its data in the journal.
2261 * libsystemd-journal gained a new call
2262 sd_journal_add_conjunction() for AND expressions to the
2263 matching logic. This can be used to express more complex
2264 logical expressions.
2266 * journactl can now take multiple --unit= and --user-unit=
2269 * The cryptsetup logic now understands the "luks.key=" kernel
2270 command line switch for specifying a file to read the
2271 decryption key from. Also, if a configured key file is not
2272 found the tool will now automatically fall back to prompting
2275 * Python systemd.journal module was updated to wrap recently
2276 added functions from libsystemd-journal. The interface was
2277 changed to bring the low level interface in s.j._Reader
2278 closer to the C API, and the high level interface in
2279 s.j.Reader was updated to wrap and convert all data about
2282 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2283 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2284 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2285 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2286 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2287 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2291 * journalctl --update-catalog now understands a new --root=
2292 option to operate on catalogs found in a different root
2295 * During shutdown after systemd has terminated all running
2296 services a final killing loop kills all remaining left-over
2297 processes. We will now print the name of these processes
2298 when we send SIGKILL to them, since this usually indicates a
2301 * If /etc/crypttab refers to password files stored on
2302 configured mount points automatic dependencies will now be
2303 generated to ensure the specific mount is established first
2304 before the key file is attempted to be read.
2306 * 'systemctl status' will now show information about the
2307 network sockets a socket unit is listening on.
2309 * 'systemctl status' will also shown information about any
2310 drop-in configuration file for units. (Drop-In configuration
2311 files in this context are files such as
2312 /etc/systemd/systemd/foobar.service.d/*.conf)
2314 * systemd-cgtop now optionally shows summed up CPU times of
2315 cgroups. Press '%' while running cgtop to switch between
2316 percentage and absolute mode. This is useful to determine
2317 which cgroups use up the most CPU time over the entire
2318 runtime of the system. systemd-cgtop has also been updated
2319 to be 'pipeable' for processing with further shell tools.
2321 * 'hostnamectl set-hostname' will now allow setting of FQDN
2324 * The formatting and parsing of time span values has been
2325 changed. The parser now understands fractional expressions
2326 such as "5.5h". The formatter will now output fractional
2327 expressions for all time spans under 1min, i.e. "5.123456s"
2328 rather than "5s 123ms 456us". For time spans under 1s
2329 millisecond values are shown, for those under 1ms
2330 microsecond values are shown. This should greatly improve
2331 all time-related output of systemd.
2333 * libsystemd-login and libsystemd-journal gained new
2334 functions for querying the poll() events mask and poll()
2335 timeout value for integration into arbitrary event
2338 * localectl gained the ability to list available X11 keymaps
2339 (models, layouts, variants, options).
2341 * 'systemd-analyze dot' gained the ability to filter for
2342 specific units via shell-style globs, to create smaller,
2343 more useful graphs. I.e. it is now possible to create simple
2344 graphs of all the dependencies between only target units, or
2345 of all units that Avahi has dependencies with.
2347 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2348 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2349 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2350 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2351 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2352 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2353 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2357 * The boot-time readahead implementation for rotating media
2358 will now read the read-ahead data in multiple passes which
2359 consist of all read requests made in equidistant time
2360 intervals. This means instead of strictly reading read-ahead
2361 data in its physical order on disk we now try to find a
2362 middle ground between physical and access time order.
2364 * /etc/os-release files gained a new BUILD_ID= field for usage
2365 on operating systems that provide continuous builds of OS
2368 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2369 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2370 William Douglas, Zbigniew Jędrzejewski-Szmek
2374 * systemd-python gained an API exposing libsystemd-daemon.
2376 * The SMACK setup logic gained support for uploading CIPSO
2379 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2380 ReadOnlyDirectories= and InaccessibleDirectories= has
2381 changed. The private /tmp and /var/tmp directories are now
2382 shared by all processes of a service (which means
2383 ExecStartPre= may now leave data in /tmp that ExecStart= of
2384 the same service can still access). When a service is
2385 stopped its temporary directories are immediately deleted
2386 (normal clean-up with tmpfiles is still done in addition to
2389 * By default, systemd will now set a couple of sysctl
2390 variables in the kernel: the safe sysrq options are turned
2391 on, IP route verification is turned on, and source routing
2392 disabled. The recently added hardlink and softlink
2393 protection of the kernel is turned on. These settings should
2394 be reasonably safe, and good defaults for all new systems.
2396 * The predictable network naming logic may now be turned off
2397 with a new kernel command line switch: net.ifnames=0.
2399 * A new libsystemd-bus module has been added that implements a
2400 pretty complete D-Bus client library. For details see:
2402 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2404 * journald will now explicitly flush the journal files to disk
2405 at the latest 5min after each write. The file will then also
2406 be marked offline until the next write. This should increase
2407 reliability in case of a crash. The synchronization delay
2408 can be configured via SyncIntervalSec= in journald.conf.
2410 * There's a new remote-fs-setup.target unit that can be used
2411 to pull in specific services when at least one remote file
2412 system is to be mounted.
2414 * There are new targets timers.target and paths.target as
2415 canonical targets to pull user timer and path units in
2416 from. This complements sockets.target with a similar
2417 purpose for socket units.
2419 * libudev gained a new call udev_device_set_attribute_value()
2420 to set sysfs attributes of a device.
2422 * The udev daemon now sets the default number of worker
2423 processes executed in parallel based on the number of available
2424 CPUs instead of the amount of available RAM. This is supposed
2425 to provide a more reliable default and limit a too aggressive
2426 paralellism for setups with 1000s of devices connected.
2428 Contributions from: Auke Kok, Colin Walters, Cristian
2429 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2430 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2431 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2432 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2433 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2434 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2435 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2436 Zbigniew Jędrzejewski-Szmek
2440 * Configuration of unit files may now be extended via drop-in
2441 files without having to edit/override the unit files
2442 themselves. More specifically, if the administrator wants to
2443 change one value for a service file foobar.service he can
2444 now do so by dropping in a configuration snippet into
2445 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2446 will load all these snippets and apply them on top of the
2447 main unit configuration file, possibly extending or
2448 overriding its settings. Using these drop-in snippets is
2449 generally nicer than the two earlier options for changing
2450 unit files locally: copying the files from
2451 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2452 them there; or creating a new file in /etc/systemd/system/
2453 that incorporates the original one via ".include". Drop-in
2454 snippets into these .d/ directories can be placed in any
2455 directory systemd looks for units in, and the usual
2456 overriding semantics between /usr/lib, /etc and /run apply
2459 * Most unit file settings which take lists of items can now be
2460 reset by assigning the empty string to them. For example,
2461 normally, settings such as Environment=FOO=BAR append a new
2462 environment variable assignment to the environment block,
2463 each time they are used. By assigning Environment= the empty
2464 string the environment block can be reset to empty. This is
2465 particularly useful with the .d/*.conf drop-in snippets
2466 mentioned above, since this adds the ability to reset list
2467 settings from vendor unit files via these drop-ins.
2469 * systemctl gained a new "list-dependencies" command for
2470 listing the dependencies of a unit recursively.
2472 * Inhibitors are now honored and listed by "systemctl
2473 suspend", "systemctl poweroff" (and similar) too, not only
2474 GNOME. These commands will also list active sessions by
2477 * Resource limits (as exposed by the various control group
2478 controllers) can now be controlled dynamically at runtime
2479 for all units. More specifically, you can now use a command
2480 like "systemctl set-cgroup-attr foobar.service cpu.shares
2481 2000" to alter the CPU shares a specific service gets. These
2482 settings are stored persistently on disk, and thus allow the
2483 administrator to easily adjust the resource usage of
2484 services with a few simple commands. This dynamic resource
2485 management logic is also available to other programs via the
2486 bus. Almost any kernel cgroup attribute and controller is
2489 * systemd-vconsole-setup will now copy all font settings to
2490 all allocated VTs, where it previously applied them only to
2493 * libsystemd-login gained the new sd_session_get_tty() API
2496 * This release drops support for a few legacy or
2497 distribution-specific LSB facility names when parsing init
2498 scripts: $x-display-manager, $mail-transfer-agent,
2499 $mail-transport-agent, $mail-transfer-agent, $smtp,
2500 $null. Also, the mail-transfer-agent.target unit backing
2501 this has been removed. Distributions which want to retain
2502 compatibility with this should carry the burden for
2503 supporting this themselves and patch support for these back
2504 in, if they really need to. Also, the facilities $syslog and
2505 $local_fs are now ignored, since systemd does not support
2506 early-boot LSB init scripts anymore, and these facilities
2507 are implied anyway for normal services. syslog.target has
2510 * There are new bus calls on PID1's Manager object for
2511 cancelling jobs, and removing snapshot units. Previously,
2512 both calls were only available on the Job and Snapshot
2515 * systemd-journal-gatewayd gained SSL support.
2517 * The various "environment" files, such as /etc/locale.conf
2518 now support continuation lines with a backslash ("\") as
2519 last character in the line, similarly in style (but different)
2520 to how this is supported in shells.
2522 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2523 now implicitly appended to every log entry logged. systemctl
2524 has been updated to filter by this field when operating on a
2525 user systemd instance.
2527 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2528 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2529 the container. This makes it easier to boot unmodified
2530 Fedora systems in a container, which however still requires
2531 audit=0 to be passed on the kernel command line. Auditing in
2532 kernel and userspace is unfortunately still too broken in
2533 context of containers, hence we recommend compiling it out
2534 of the kernel or using audit=0. Hopefully this will be fixed
2535 one day for good in the kernel.
2537 * nspawn gained the new --bind= and --bind-ro= parameters to
2538 bind mount specific directories from the host into the
2541 * nspawn will now mount its own devpts file system instance
2542 into the container, in order not to leak pty devices from
2543 the host into the container.
2545 * systemd will now read the firmware boot time performance
2546 information from the EFI variables, if the used boot loader
2547 supports this, and takes it into account for boot performance
2548 analysis via "systemd-analyze". This is currently supported
2549 only in conjunction with Gummiboot, but could be supported
2550 by other boot loaders too. For details see:
2552 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2554 * A new generator has been added that automatically mounts the
2555 EFI System Partition (ESP) to /boot, if that directory
2556 exists, is empty, and no other file system has been
2557 configured to be mounted there.
2559 * logind will now send out PrepareForSleep(false) out
2560 unconditionally, after coming back from suspend. This may be
2561 used by applications as asynchronous notification for
2562 system resume events.
2564 * "systemctl unlock-sessions" has been added, that allows
2565 unlocking the screens of all user sessions at once, similar
2566 to how "systemctl lock-sessions" already locked all users
2567 sessions. This is backed by a new D-Bus call UnlockSessions().
2569 * "loginctl seat-status" will now show the master device of a
2570 seat. (i.e. the device of a seat that needs to be around for
2571 the seat to be considered available, usually the graphics
2574 * tmpfiles gained a new "X" line type, that allows
2575 configuration of files and directories (with wildcards) that
2576 shall be excluded from automatic cleanup ("aging").
2578 * udev default rules set the device node permissions now only
2579 at "add" events, and do not change them any longer with a
2580 later "change" event.
2582 * The log messages for lid events and power/sleep keypresses
2583 now carry a message ID.
2585 * We now have a substantially larger unit test suite, but this
2586 continues to be work in progress.
2588 * udevadm hwdb gained a new --root= parameter to change the
2589 root directory to operate relative to.
2591 * logind will now issue a background sync() request to the kernel
2592 early at shutdown, so that dirty buffers are flushed to disk early
2593 instead of at the last moment, in order to optimize shutdown
2596 * A new bootctl tool has been added that is an interface for
2597 certain boot loader operations. This is currently a preview
2598 and is likely to be extended into a small mechanism daemon
2599 like timedated, localed, hostnamed, and can be used by
2600 graphical UIs to enumerate available boot options, and
2601 request boot into firmware operations.
2603 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2604 the rest of the package. It also has been updated to work
2605 correctly in initrds.
2607 * Policykit previously has been runtime optional, and is now
2608 also compile time optional via a configure switch.
2610 * systemd-analyze has been reimplemented in C. Also "systemctl
2611 dot" has moved into systemd-analyze.
2613 * "systemctl status" with no further parameters will now print
2614 the status of all active or failed units.
2616 * Operations such as "systemctl start" can now be executed
2617 with a new mode "--irreversible" which may be used to queue
2618 operations that cannot accidentally be reversed by a later
2619 job queuing. This is by default used to make shutdown
2620 requests more robust.
2622 * The Python API of systemd now gained a new module for
2623 reading journal files.
2625 * A new tool kernel-install has been added that can install
2626 kernel images according to the Boot Loader Specification:
2628 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2630 * Boot time console output has been improved to provide
2631 animated boot time output for hanging jobs.
2633 * A new tool systemd-activate has been added which can be used
2634 to test socket activation with, directly from the command
2635 line. This should make it much easier to test and debug
2636 socket activation in daemons.
2638 * journalctl gained a new "--reverse" (or -r) option to show
2639 journal output in reverse order (i.e. newest line first).
2641 * journalctl gained a new "--pager-end" (or -e) option to jump
2642 to immediately jump to the end of the journal in the
2643 pager. This is only supported in conjunction with "less".
2645 * journalctl gained a new "--user-unit=" option, that works
2646 similarly to "--unit=" but filters for user units rather than
2649 * A number of unit files to ease adoption of systemd in
2650 initrds has been added. This moves some minimal logic from
2651 the various initrd implementations into systemd proper.
2653 * The journal files are now owned by a new group
2654 "systemd-journal", which exists specifically to allow access
2655 to the journal, and nothing else. Previously, we used the
2656 "adm" group for that, which however possibly covers more
2657 than just journal/log file access. This new group is now
2658 already used by systemd-journal-gatewayd to ensure this
2659 daemon gets access to the journal files and as little else
2660 as possible. Note that "make install" will also set FS ACLs
2661 up for /var/log/journal to give "adm" and "wheel" read
2662 access to it, in addition to "systemd-journal" which owns
2663 the journal files. We recommend that packaging scripts also
2664 add read access to "adm" + "wheel" to /var/log/journal, and
2665 all existing/future journal files. To normal users and
2666 administrators little changes, however packagers need to
2667 ensure to create the "systemd-journal" system group at
2668 package installation time.
2670 * The systemd-journal-gatewayd now runs as unprivileged user
2671 systemd-journal-gateway:systemd-journal-gateway. Packaging
2672 scripts need to create these system user/group at
2675 * timedated now exposes a new boolean property CanNTP that
2676 indicates whether a local NTP service is available or not.
2678 * systemd-detect-virt will now also detect xen PVs
2680 * The pstore file system is now mounted by default, if it is
2683 * In addition to the SELinux and IMA policies we will now also
2684 load SMACK policies at early boot.
2686 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2687 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2688 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2689 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2690 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2691 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2692 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2693 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2694 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2695 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2696 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2697 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2698 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2699 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2703 * Timer units now support calendar time events in addition to
2704 monotonic time events. That means you can now trigger a unit
2705 based on a calendar time specification such as "Thu,Fri
2706 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2707 or fifth day of any month of the year 2013, given that it is
2708 a thursday or friday. This brings timer event support
2709 considerably closer to cron's capabilities. For details on
2710 the supported calendar time specification language see
2713 * udev now supports a number of different naming policies for
2714 network interfaces for predictable names, and a combination
2715 of these policies is now the default. Please see this wiki
2716 document for details:
2718 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2720 * Auke Kok's bootchart implementation has been added to the
2721 systemd tree. It is an optional component that can graph the
2722 boot in quite some detail. It is one of the best bootchart
2723 implementations around and minimal in its code and
2726 * nss-myhostname has been integrated into the systemd source
2727 tree. nss-myhostname guarantees that the local hostname
2728 always stays resolvable via NSS. It has been a weak
2729 requirement of systemd-hostnamed since a long time, and
2730 since its code is actually trivial we decided to just
2731 include it in systemd's source tree. It can be turned off
2732 with a configure switch.
2734 * The read-ahead logic is now capable of properly detecting
2735 whether a btrfs file system is on SSD or rotating media, in
2736 order to optimize the read-ahead scheme. Previously, it was
2737 only capable of detecting this on traditional file systems
2740 * In udev, additional device properties are now read from the
2741 IAB in addition to the OUI database. Also, Bluetooth company
2742 identities are attached to the devices as well.
2744 * In service files %U may be used as specifier that is
2745 replaced by the configured user name of the service.
2747 * nspawn may now be invoked without a controlling TTY. This
2748 makes it suitable for invocation as its own service. This
2749 may be used to set up a simple containerized server system
2750 using only core OS tools.
2752 * systemd and nspawn can now accept socket file descriptors
2753 when they are started for socket activation. This enables
2754 implementation of socket activated nspawn
2755 containers. i.e. think about autospawning an entire OS image
2756 when the first SSH or HTTP connection is received. We expect
2757 that similar functionality will also be added to libvirt-lxc
2760 * journalctl will now suppress ANSI color codes when
2761 presenting log data.
2763 * systemctl will no longer show control group information for
2764 a unit if a the control group is empty anyway.
2766 * logind can now automatically suspend/hibernate/shutdown the
2769 * /etc/machine-info and hostnamed now also expose the chassis
2770 type of the system. This can be used to determine whether
2771 the local system is a laptop, desktop, handset or
2772 tablet. This information may either be configured by the
2773 user/vendor or is automatically determined from ACPI and DMI
2774 information if possible.
2776 * A number of PolicyKit actions are now bound together with
2777 "imply" rules. This should simplify creating UIs because
2778 many actions will now authenticate similar ones as well.
2780 * Unit files learnt a new condition ConditionACPower= which
2781 may be used to conditionalize a unit depending on whether an
2782 AC power source is connected or not, of whether the system
2783 is running on battery power.
2785 * systemctl gained a new "is-failed" verb that may be used in
2786 shell scripts and suchlike to check whether a specific unit
2787 is in the "failed" state.
2789 * The EnvironmentFile= setting in unit files now supports file
2790 globbing, and can hence be used to easily read a number of
2791 environment files at once.
2793 * systemd will no longer detect and recognize specific
2794 distributions. All distribution-specific #ifdeffery has been
2795 removed, systemd is now fully generic and
2796 distribution-agnostic. Effectively, not too much is lost as
2797 a lot of the code is still accessible via explicit configure
2798 switches. However, support for some distribution specific
2799 legacy configuration file formats has been dropped. We
2800 recommend distributions to simply adopt the configuration
2801 files everybody else uses now and convert the old
2802 configuration from packaging scripts. Most distributions
2803 already did that. If that's not possible or desirable,
2804 distributions are welcome to forward port the specific
2805 pieces of code locally from the git history.
2807 * When logging a message about a unit systemd will now always
2808 log the unit name in the message meta data.
2810 * localectl will now also discover system locale data that is
2811 not stored in locale archives, but directly unpacked.
2813 * logind will no longer unconditionally use framebuffer
2814 devices as seat masters, i.e. as devices that are required
2815 to be existing before a seat is considered preset. Instead,
2816 it will now look for all devices that are tagged as
2817 "seat-master" in udev. By default framebuffer devices will
2818 be marked as such, but depending on local systems other
2819 devices might be marked as well. This may be used to
2820 integrate graphics cards using closed source drivers (such
2821 as NVidia ones) more nicely into logind. Note however, that
2822 we recommend using the open source NVidia drivers instead,
2823 and no udev rules for the closed-source drivers will be
2824 shipped from us upstream.
2826 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2827 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2828 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2829 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2830 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2831 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2832 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2833 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2834 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2835 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2836 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2837 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2842 * udev gained support for loading additional device properties
2843 from an indexed database that is keyed by vendor/product IDs
2844 and similar device identifiers. For the beginning this
2845 "hwdb" is populated with data from the well-known PCI and
2846 USB database, but also includes PNP, ACPI and OID data. In
2847 the longer run this indexed database shall grow into
2848 becoming the one central database for non-essential
2849 userspace device metadata. Previously, data from the PCI/USB
2850 database was only attached to select devices, since the
2851 lookup was a relatively expensive operation due to O(n) time
2852 complexity (with n being the number of entries in the
2853 database). Since this is now O(1), we decided to add in this
2854 data for all devices where this is available, by
2855 default. Note that the indexed database needs to be rebuilt
2856 when new data files are installed. To achieve this you need
2857 to update your packaging scripts to invoke "udevadm hwdb
2858 --update" after installation of hwdb data files. For
2859 RPM-based distributions we introduced the new
2860 %udev_hwdb_update macro for this purpose.
2862 * The Journal gained support for the "Message Catalog", an
2863 indexed database to link up additional information with
2864 journal entries. For further details please check:
2866 http://www.freedesktop.org/wiki/Software/systemd/catalog
2868 The indexed message catalog database also needs to be
2869 rebuilt after installation of message catalog files. Use
2870 "journalctl --update-catalog" for this. For RPM-based
2871 distributions we introduced the %journal_catalog_update
2872 macro for this purpose.
2874 * The Python Journal bindings gained support for the standard
2875 Python logging framework.
2877 * The Journal API gained new functions for checking whether
2878 the underlying file system of a journal file is capable of
2879 properly reporting file change notifications, or whether
2880 applications that want to reflect journal changes "live"
2881 need to recheck journal files continuously in appropriate
2884 * It is now possible to set the "age" field for tmpfiles
2885 entries to 0, indicating that files matching this entry
2886 shall always be removed when the directories are cleaned up.
2888 * coredumpctl gained a new "gdb" verb which invokes gdb
2889 right-away on the selected coredump.
2891 * There's now support for "hybrid sleep" on kernels that
2892 support this, in addition to "suspend" and "hibernate". Use
2893 "systemctl hybrid-sleep" to make use of this.
2895 * logind's HandleSuspendKey= setting (and related settings)
2896 now gained support for a new "lock" setting to simply
2897 request the screen lock on all local sessions, instead of
2898 actually executing a suspend or hibernation.
2900 * systemd will now mount the EFI variables file system by
2903 * Socket units now gained support for configuration of the
2904 SMACK security label.
2906 * timedatectl will now output the time of the last and next
2907 daylight saving change.
2909 * We dropped support for various legacy and distro-specific
2910 concepts, such as insserv, early-boot SysV services
2911 (i.e. those for non-standard runlevels such as 'b' or 'S')
2912 or ArchLinux /etc/rc.conf support. We recommend the
2913 distributions who still need support this to either continue
2914 to maintain the necessary patches downstream, or find a
2915 different solution. (Talk to us if you have questions!)
2917 * Various systemd components will now bypass PolicyKit checks
2918 for root and otherwise handle properly if PolicyKit is not
2919 found to be around. This should fix most issues for
2920 PolicyKit-less systems. Quite frankly this should have been
2921 this way since day one. It is absolutely our intention to
2922 make systemd work fine on PolicyKit-less systems, and we
2923 consider it a bug if something does not work as it should if
2924 PolicyKit is not around.
2926 * For embedded systems it is now possible to build udev and
2927 systemd without blkid and/or kmod support.
2929 * "systemctl switch-root" is now capable of switching root
2930 more than once. I.e. in addition to transitions from the
2931 initrd to the host OS it is now possible to transition to
2932 further OS images from the host. This is useful to implement
2933 offline updating tools.
2935 * Various other additions have been made to the RPM macros
2936 shipped with systemd. Use %udev_rules_update() after
2937 installing new udev rules files. %_udevhwdbdir,
2938 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2939 %_sysctldir are now available which resolve to the right
2940 directories for packages to place various data files in.
2942 * journalctl gained the new --full switch (in addition to
2943 --all, to disable ellipsation for long messages.
2945 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2946 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2947 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2948 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2949 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2950 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2951 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2952 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2953 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2957 * journalctl gained new --since= and --until= switches to
2958 filter by time. It also now supports nice filtering for
2959 units via --unit=/-u.
2961 * Type=oneshot services may use ExecReload= and do the
2964 * The journal daemon now supports time-based rotation and
2965 vacuuming, in addition to the usual disk-space based
2968 * The journal will now index the available field values for
2969 each field name. This enables clients to show pretty drop
2970 downs of available match values when filtering. The bash
2971 completion of journalctl has been updated
2972 accordingly. journalctl gained a new switch -F to list all
2973 values a certain field takes in the journal database.
2975 * More service events are now written as structured messages
2976 to the journal, and made recognizable via message IDs.
2978 * The timedated, localed and hostnamed mini-services which
2979 previously only provided support for changing time, locale
2980 and hostname settings from graphical DEs such as GNOME now
2981 also have a minimal (but very useful) text-based client
2982 utility each. This is probably the nicest way to changing
2983 these settings from the command line now, especially since
2984 it lists available options and is fully integrated with bash
2987 * There's now a new tool "systemd-coredumpctl" to list and
2988 extract coredumps from the journal.
2990 * We now install a README each in /var/log/ and
2991 /etc/rc.d/init.d explaining where the system logs and init
2992 scripts went. This hopefully should help folks who go to
2993 that dirs and look into the otherwise now empty void and
2994 scratch their heads.
2996 * When user-services are invoked (by systemd --user) the
2997 $MANAGERPID env var is set to the PID of systemd.
2999 * SIGRTMIN+24 when sent to a --user instance will now result
3000 in immediate termination of systemd.
3002 * gatewayd received numerous feature additions such as a
3003 "follow" mode, for live syncing and filtering.
3005 * browse.html now allows filtering and showing detailed
3006 information on specific entries. Keyboard navigation and
3007 mouse screen support has been added.
3009 * gatewayd/journalctl now supports HTML5/JSON
3010 Server-Sent-Events as output.
3012 * The SysV init script compatibility logic will now
3013 heuristically determine whether a script supports the
3014 "reload" verb, and only then make this available as
3017 * "systemctl status --follow" has been removed, use "journalctl
3020 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3021 have been removed since they are hardly useful to be
3024 * And I'd like to take the opportunity to specifically mention
3025 Zbigniew for his great contributions. Zbigniew, you rock!
3027 Contributions from: Andrew Eikum, Christian Hesse, Colin
3028 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3029 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3030 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3031 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3032 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3033 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3037 * If /etc/vconsole.conf is non-existent or empty we will no
3038 longer load any console font or key map at boot by
3039 default. Instead the kernel defaults will be left
3040 intact. This is definitely the right thing to do, as no
3041 configuration should mean no configuration, and hard-coding
3042 font names that are different on all archs is probably a bad
3043 idea. Also, the kernel default key map and font should be
3044 good enough for most cases anyway, and mostly identical to
3045 the userspace fonts/key maps we previously overloaded them
3046 with. If distributions want to continue to default to a
3047 non-kernel font or key map they should ship a default
3048 /etc/vconsole.conf with the appropriate contents.
3050 Contributions from: Colin Walters, Daniel J Walsh, Dave
3051 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3052 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3056 * journalctl gained a new --cursor= switch to show entries
3057 starting from the specified location in the journal.
3059 * We now enforce a size limit on journal entry fields exported
3060 with "-o json" in journalctl. Fields larger than 4K will be
3061 assigned null. This can be turned off with --all.
3063 * An (optional) journal gateway daemon is now available as
3064 "systemd-journal-gatewayd.service". This service provides
3065 access to the journal via HTTP and JSON. This functionality
3066 will be used to implement live log synchronization in both
3067 pull and push modes, but has various other users too, such
3068 as easy log access for debugging of embedded devices. Right
3069 now it is already useful to retrieve the journal via HTTP:
3071 # systemctl start systemd-journal-gatewayd.service
3072 # wget http://localhost:19531/entries
3074 This will download the journal contents in a
3075 /var/log/messages compatible format. The same as JSON:
3077 # curl -H"Accept: application/json" http://localhost:19531/entries
3079 This service is also accessible via a web browser where a
3080 single static HTML5 app is served that uses the JSON logic
3081 to enable the user to do some basic browsing of the
3082 journal. This will be extended later on. Here's an example
3083 screenshot of this app in its current state:
3085 http://0pointer.de/public/journal-gatewayd
3087 Contributions from: Kay Sievers, Lennart Poettering, Robert
3088 Milasan, Tom Gundersen
3092 * The bash completion logic is now available for journalctl
3095 * We do not mount the "cpuset" controller anymore together with
3096 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3097 started if no parameters are assigned to it. "cpuset" hence
3098 broke code that assumed it it could create "cpu" groups and
3101 * journalctl -f will now subscribe to terminal size changes,
3102 and line break accordingly.
3104 Contributions from: Dave Reisner, Kay Sievers, Lennart
3105 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3109 * nspawn will now create a symlink /etc/localtime in the
3110 container environment, copying the host's timezone
3111 setting. Previously this has been done via a bind mount, but
3112 since symlinks cannot be bind mounted this has now been
3113 changed to create/update the appropriate symlink.
3115 * journalctl -n's line number argument is now optional, and
3116 will default to 10 if omitted.
3118 * journald will now log the maximum size the journal files may
3119 take up on disk. This is particularly useful if the default
3120 built-in logic of determining this parameter from the file
3121 system size is used. Use "systemctl status
3122 systemd-journald.service" to see this information.
3124 * The multi-seat X wrapper tool has been stripped down. As X
3125 is now capable of enumerating graphics devices via udev in a
3126 seat-aware way the wrapper is not strictly necessary
3127 anymore. A stripped down temporary stop-gap is still shipped
3128 until the upstream display managers have been updated to
3129 fully support the new X logic. Expect this wrapper to be
3130 removed entirely in one of the next releases.
3132 * HandleSleepKey= in logind.conf has been split up into
3133 HandleSuspendKey= and HandleHibernateKey=. The old setting
3134 is not available anymore. X11 and the kernel are
3135 distuingishing between these keys and we should too. This
3136 also means the inhibition lock for these keys has been split
3139 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3140 Poettering, Lukas Nykryn, Václav Pavlín
3144 * Whenever a unit changes state we will now log this to the
3145 journal and show along the unit's own log output in
3148 * ConditionPathIsMountPoint= can now properly detect bind
3149 mount points too. (Previously, a bind mount of one file
3150 system to another place in the same file system could not be
3151 detected as mount, since they shared struct stat's st_dev
3154 * We will now mount the cgroup controllers cpu, cpuacct,
3155 cpuset and the controllers net_cls, net_prio together by
3158 * nspawn containers will now have a virtualized boot
3159 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3160 over with a randomized ID at container initialization). This
3161 has the effect of making "journalctl -b" do the right thing
3164 * The JSON output journal serialization has been updated not
3165 to generate "endless" list objects anymore, but rather one
3166 JSON object per line. This is more in line how most JSON
3167 parsers expect JSON objects. The new output mode
3168 "json-pretty" has been added to provide similar output, but
3169 neatly aligned for readability by humans.
3171 * We dropped all explicit sync() invocations in the shutdown
3172 code. The kernel does this implicitly anyway in the kernel
3173 reboot() syscall. halt(8)'s -n option is now a compatibility
3176 * We now support virtualized reboot() in containers, as
3177 supported by newer kernels. We will fall back to exit() if
3178 CAP_SYS_REBOOT is not available to the container. Also,
3179 nspawn makes use of this now and will actually reboot the
3180 container if the containerized OS asks for that.
3182 * journalctl will only show local log output by default
3183 now. Use --merge (-m) to show remote log output, too.
3185 * libsystemd-journal gained the new sd_journal_get_usage()
3186 call to determine the current disk usage of all journal
3187 files. This is exposed in the new "journalctl --disk-usage"
3190 * journald gained a new configuration setting SplitMode= in
3191 journald.conf which may be used to control how user journals
3192 are split off. See journald.conf(5) for details.
3194 * A new condition type ConditionFileNotEmpty= has been added.
3196 * tmpfiles' "w" lines now support file globbing, to write
3197 multiple files at once.
3199 * We added Python bindings for the journal submission
3200 APIs. More Python APIs for a number of selected APIs will
3201 likely follow. Note that we intend to add native bindings
3202 only for the Python language, as we consider it common
3203 enough to deserve bindings shipped within systemd. There are
3204 various projects outside of systemd that provide bindings
3205 for languages such as PHP or Lua.
3207 * Many conditions will now resolve specifiers such as %i. In
3208 addition, PathChanged= and related directives of .path units
3209 now support specifiers as well.
3211 * There's now a new RPM macro definition for the system preset
3214 * journald will now warn if it ca not forward a message to the
3215 syslog daemon because its socket is full.
3217 * timedated will no longer write or process /etc/timezone,
3218 except on Debian. As we do not support late mounted /usr
3219 anymore /etc/localtime always being a symlink is now safe,
3220 and hence the information in /etc/timezone is not necessary
3223 * logind will now always reserve one VT for a text getty (VT6
3224 by default). Previously if more than 6 X sessions where
3225 started they took up all the VTs with auto-spawned gettys,
3226 so that no text gettys were available anymore.
3228 * udev will now automatically inform the btrfs kernel logic
3229 about btrfs RAID components showing up. This should make
3230 simple hotplug based btrfs RAID assembly work.
3232 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3233 (but not for its children which will stay at the kernel
3234 default). This should allow setups with a lot more listening
3237 * systemd will now always pass the configured timezone to the
3238 kernel at boot. timedated will do the same when the timezone
3241 * logind's inhibition logic has been updated. By default,
3242 logind will now handle the lid switch, the power and sleep
3243 keys all the time, even in graphical sessions. If DEs want
3244 to handle these events on their own they should take the new
3245 handle-power-key, handle-sleep-key and handle-lid-switch
3246 inhibitors during their runtime. A simple way to achiveve
3247 that is to invoke the DE wrapped in an invocation of:
3249 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3251 * Access to unit operations is now checked via SELinux taking
3252 the unit file label and client process label into account.
3254 * systemd will now notify the administrator in the journal
3255 when he over-mounts a non-empty directory.
3257 * There are new specifiers that are resolved in unit files,
3258 for the host name (%H), the machine ID (%m) and the boot ID
3261 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3262 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3263 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3264 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3265 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3266 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3267 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3271 * Support for reading structured kernel messages from
3272 /dev/kmsg has now been added and is enabled by default.
3274 * Support for reading kernel messages from /proc/kmsg has now
3275 been removed. If you want kernel messages in the journal
3276 make sure to run a recent kernel (>= 3.5) that supports
3277 reading structured messages from /dev/kmsg (see
3278 above). /proc/kmsg is now exclusive property of classic
3279 syslog daemons again.
3281 * The libudev API gained the new
3282 udev_device_new_from_device_id() call.
3284 * The logic for file system namespace (ReadOnlyDirectory=,
3285 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3286 require pivot_root() anymore. This means fewer temporary
3287 directories are created below /tmp for this feature.
3289 * nspawn containers will now see and receive all submounts
3290 made on the host OS below the root file system of the
3293 * Forward Secure Sealing is now supported for Journal files,
3294 which provide cryptographical sealing of journal files so
3295 that attackers cannot alter log history anymore without this
3296 being detectable. Lennart will soon post a blog story about
3297 this explaining it in more detail.
3299 * There are two new service settings RestartPreventExitStatus=
3300 and SuccessExitStatus= which allow configuration of exit
3301 status (exit code or signal) which will be excepted from the
3302 restart logic, resp. consider successful.
3304 * journalctl gained the new --verify switch that can be used
3305 to check the integrity of the structure of journal files and
3306 (if Forward Secure Sealing is enabled) the contents of
3309 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3310 and similar symlinks pre-created. This makes running shells
3311 as container init process a lot more fun.
3313 * The fstab support can now handle PARTUUID= and PARTLABEL=
3316 * A new ConditionHost= condition has been added to match
3317 against the hostname (with globs) and machine ID. This is
3318 useful for clusters where a single OS image is used to
3319 provision a large number of hosts which shall run slightly
3320 different sets of services.
3322 * Services which hit the restart limit will now be placed in a
3325 Contributions from: Bertram Poettering, Dave Reisner, Huang
3326 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3327 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3331 * When running in --user mode systemd will now become a
3332 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3333 tree a lot more organized.
3335 * A new PartOf= unit dependency type has been introduced that
3336 may be used to group services in a natural way.
3338 * "systemctl enable" may now be used to enable instances of
3341 * journalctl now prints error log levels in red, and
3342 warning/notice log levels in bright white. It also supports
3343 filtering by log level now.
3345 * cgtop gained a new -n switch (similar to top), to configure
3346 the maximum number of iterations to run for. It also gained
3347 -b, to run in batch mode (accepting no input).
3349 * The suffix ".service" may now be omitted on most systemctl
3350 command lines involving service unit names.
3352 * There's a new bus call in logind to lock all sessions, as
3353 well as a loginctl verb for it "lock-sessions".
3355 * libsystemd-logind.so gained a new call sd_journal_perror()
3356 that works similar to libc perror() but logs to the journal
3357 and encodes structured information about the error number.
3359 * /etc/crypttab entries now understand the new keyfile-size=
3362 * shutdown(8) now can send a (configurable) wall message when
3363 a shutdown is cancelled.
3365 * The mount propagation mode for the root file system will now
3366 default to "shared", which is useful to make containers work
3367 nicely out-of-the-box so that they receive new mounts from
3368 the host. This can be undone locally by running "mount
3369 --make-rprivate /" if needed.
3371 * The prefdm.service file has been removed. Distributions
3372 should maintain this unit downstream if they intend to keep
3373 it around. However, we recommend writing normal unit files
3374 for display managers instead.
3376 * Since systemd is a crucial part of the OS we will now
3377 default to a number of compiler switches that improve
3378 security (hardening) such as read-only relocations, stack
3379 protection, and suchlike.
3381 * The TimeoutSec= setting for services is now split into
3382 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3383 of individual time outs for the start and the stop phase of
3386 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3387 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3388 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3389 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3390 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3391 Gundersen, Zbigniew Jędrzejewski-Szmek
3395 * The journal and id128 C APIs are now fully documented as man
3398 * Extra safety checks have been added when transitioning from
3399 the initial RAM disk to the main system to avoid accidental
3402 * /etc/crypttab entries now understand the new keyfile-offset=
3405 * systemctl -t can now be used to filter by unit load state.
3407 * The journal C API gained the new sd_journal_wait() call to
3408 make writing synchronous journal clients easier.
3410 * journalctl gained the new -D switch to show journals from a
3413 * journalctl now displays a special marker between log
3414 messages of two different boots.
3416 * The journal is now explicitly flushed to /var via a service
3417 systemd-journal-flush.service, rather than implicitly simply
3418 by seeing /var/log/journal to be writable.
3420 * journalctl (and the journal C APIs) can now match for much
3421 more complex expressions, with alternatives and
3424 * When transitioning from the initial RAM disk to the main
3425 system we will now kill all processes in a killing spree to
3426 ensure no processes stay around by accident.
3428 * Three new specifiers may be used in unit files: %u, %h, %s
3429 resolve to the user name, user home directory resp. user
3430 shell. This is useful for running systemd user instances.
3432 * We now automatically rotate journal files if their data
3433 object hash table gets a fill level > 75%. We also size the
3434 hash table based on the configured maximum file size. This
3435 together should lower hash collisions drastically and thus
3436 speed things up a bit.
3438 * journalctl gained the new "--header" switch to introspect
3439 header data of journal files.
3441 * A new setting SystemCallFilters= has been added to services
3442 which may be used to apply blacklists or whitelists to
3443 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3445 * nspawn gained a new --link-journal= switch (and quicker: -j)
3446 to link the container journal with the host. This makes it
3447 very easy to centralize log viewing on the host for all
3448 guests while still keeping the journal files separated.
3450 * Many bugfixes and optimizations
3452 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3453 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3454 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3459 * Several tools now understand kernel command line arguments,
3460 which are only read when run in an initial RAM disk. They
3461 usually follow closely their normal counterparts, but are
3464 * There's a new tool to analyze the readahead files that are
3465 automatically generated at boot. Use:
3467 /usr/lib/systemd/systemd-readahead analyze /.readahead
3469 * We now provide an early debug shell on tty9 if this enabled. Use:
3471 systemctl enable debug-shell.service
3473 * All plymouth related units have been moved into the Plymouth
3474 package. Please make sure to upgrade your Plymouth version
3477 * systemd-tmpfiles now supports getting passed the basename of
3478 a configuration file only, in which case it will look for it
3479 in all appropriate directories automatically.
3481 * udevadm info now takes a /dev or /sys path as argument, and
3482 does the right thing. Example:
3484 udevadm info /dev/sda
3485 udevadm info /sys/class/block/sda
3487 * systemctl now prints a warning if a unit is stopped but a
3488 unit that might trigger it continues to run. Example: a
3489 service is stopped but the socket that activates it is left
3492 * "systemctl status" will now mention if the log output was
3493 shortened due to rotation since a service has been started.
3495 * The journal API now exposes functions to determine the
3496 "cutoff" times due to rotation.
3498 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3499 immediately flushing of runtime logs to /var if possible,
3500 resp. for triggering immediate rotation of the journal
3503 * It is now considered an error if a service is attempted to
3504 be stopped that is not loaded.
3506 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3508 * systemd-analyze now supports Python 3
3510 * tmpfiles now supports cleaning up directories via aging
3511 where the first level dirs are always kept around but
3512 directories beneath it automatically aged. This is enabled
3513 by prefixing the age field with '~'.
3515 * Seat objects now expose CanGraphical, CanTTY properties
3516 which is required to deal with very fast bootups where the
3517 display manager might be running before the graphics drivers
3518 completed initialization.
3520 * Seat objects now expose a State property.
3522 * We now include RPM macros for service enabling/disabling
3523 based on the preset logic. We recommend RPM based
3524 distributions to make use of these macros if possible. This
3525 makes it simpler to reuse RPM spec files across
3528 * We now make sure that the collected systemd unit name is
3529 always valid when services log to the journal via
3532 * There's a new man page kernel-command-line(7) detailing all
3533 command line options we understand.
3535 * The fstab generator may now be disabled at boot by passing
3536 fstab=0 on the kernel command line.
3538 * A new kernel command line option modules-load= is now understood
3539 to load a specific kernel module statically, early at boot.
3541 * Unit names specified on the systemctl command line are now
3542 automatically escaped as needed. Also, if file system or
3543 device paths are specified they are automatically turned
3544 into the appropriate mount or device unit names. Example:
3546 systemctl status /home
3547 systemctl status /dev/sda
3549 * The SysVConsole= configuration option has been removed from
3550 system.conf parsing.
3552 * The SysV search path is no longer exported on the D-Bus
3555 * The Names= option is been removed from unit file parsing.
3557 * There's a new man page bootup(7) detailing the boot process.
3559 * Every unit and every generator we ship with systemd now
3560 comes with full documentation. The self-explanatory boot is
3563 * A couple of services gained "systemd-" prefixes in their
3564 name if they wrap systemd code, rather than only external
3565 code. Among them fsck@.service which is now
3566 systemd-fsck@.service.
3568 * The HaveWatchdog property has been removed from the D-Bus
3571 * systemd.confirm_spawn= on the kernel command line should now
3574 * There's a new man page crypttab(5) which details all options
3575 we actually understand.
3577 * systemd-nspawn gained a new --capability= switch to pass
3578 additional capabilities to the container.
3580 * timedated will now read known NTP implementation unit names
3581 from /usr/lib/systemd/ntp-units.d/*.list,
3582 systemd-timedated-ntp.target has been removed.
3584 * journalctl gained a new switch "-b" that lists log data of
3585 the current boot only.
3587 * The notify socket is in the abstract namespace again, in
3588 order to support daemons which chroot() at start-up.
3590 * There is a new Storage= configuration option for journald
3591 which allows configuration of where log data should go. This
3592 also provides a way to disable journal logging entirely, so
3593 that data collected is only forwarded to the console, the
3594 kernel log buffer or another syslog implementation.
3596 * Many bugfixes and optimizations
3598 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3599 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3600 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3601 Shawn Landden, Tom Gundersen
3605 * "systemctl help <unit>" now shows the man page if one is
3608 * Several new man pages have been added.
3610 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3611 MaxLevelConsole= can now be specified in
3612 journald.conf. These options allow reducing the amount of
3613 data stored on disk or forwarded by the log level.
3615 * TimerSlackNSec= can now be specified in system.conf for
3616 PID1. This allows system-wide power savings.
3618 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3619 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3624 * logind is now capable of (optionally) handling power and
3625 sleep keys as well as the lid switch.
3627 * journalctl now understands the syntax "journalctl
3628 /usr/bin/avahi-daemon" to get all log output of a specific
3631 * CapabilityBoundingSet= in system.conf now also influences
3632 the capability bound set of usermode helpers of the kernel.
3634 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3635 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3636 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3637 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3641 * Note that we skipped 139 releases here in order to set the
3642 new version to something that is greater than both udev's
3643 and systemd's most recent version number.
3645 * udev: all udev sources are merged into the systemd source tree now.
3646 All future udev development will happen in the systemd tree. It
3647 is still fully supported to use the udev daemon and tools without
3648 systemd running, like in initramfs or other init systems. Building
3649 udev though, will require the *build* of the systemd tree, but
3650 udev can be properly *run* without systemd.
3652 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3653 should be used to create dead device nodes as workarounds for broken
3656 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3657 no longer supported. udev_monitor_new_from_netlink() needs to be
3658 used to subscribe to events.
3660 * udev: when udevd is started by systemd, processes which are left
3661 behind by forking them off of udev rules, are unconditionally cleaned
3662 up and killed now after the event handling has finished. Services or
3663 daemons must be started as systemd services. Services can be
3664 pulled-in by udev to get started, but they can no longer be directly
3665 forked by udev rules.
3667 * udev: the daemon binary is called systemd-udevd now and installed
3668 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3669 to adapt to that, create symlink, or rename the binary after building
3672 * libudev no longer provides these symbols:
3673 udev_monitor_from_socket()
3674 udev_queue_get_failed_list_entry()
3675 udev_get_{dev,sys,run}_path()
3676 The versions number was bumped and symbol versioning introduced.
3678 * systemd-loginctl and systemd-journalctl have been renamed
3679 to loginctl and journalctl to match systemctl.
3681 * The config files: /etc/systemd/systemd-logind.conf and
3682 /etc/systemd/systemd-journald.conf have been renamed to
3683 logind.conf and journald.conf. Package updates should rename
3684 the files to the new names on upgrade.
3686 * For almost all files the license is now LGPL2.1+, changed
3687 from the previous GPL2.0+. Exceptions are some minor stuff
3688 of udev (which will be changed to LGPL2.1 eventually, too),
3689 and the MIT licensed sd-daemon.[ch] library that is suitable
3690 to be used as drop-in files.
3692 * systemd and logind now handle system sleep states, in
3693 particular suspending and hibernating.
3695 * logind now implements a sleep/shutdown/idle inhibiting logic
3696 suitable for a variety of uses. Soonishly Lennart will blog
3697 about this in more detail.
3699 * var-run.mount and var-lock.mount are no longer provided
3700 (which prevously bind mounted these directories to their new
3701 places). Distributions which have not converted these
3702 directories to symlinks should consider stealing these files
3703 from git history and add them downstream.
3705 * We introduced the Documentation= field for units and added
3706 this to all our shipped units. This is useful to make it
3707 easier to explore the boot and the purpose of the various
3710 * All smaller setup units (such as
3711 systemd-vconsole-setup.service) now detect properly if they
3712 are run in a container and are skipped when
3713 appropriate. This guarantees an entirely noise-free boot in
3714 Linux container environments such as systemd-nspawn.
3716 * A framework for implementing offline system updates is now
3717 integrated, for details see:
3718 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3720 * A new service type Type=idle is available now which helps us
3721 avoiding ugly interleaving of getty output and boot status
3724 * There's now a system-wide CapabilityBoundingSet= option to
3725 globally reduce the set of capabilities for the
3726 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3727 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3728 even CAP_NET_ADMIN system-wide for secure systems.
3730 * There are now system-wide DefaultLimitXXX= options to
3731 globally change the defaults of the various resource limits
3732 for all units started by PID 1.
3734 * Harald Hoyer's systemd test suite has been integrated into
3735 systemd which allows easy testing of systemd builds in qemu
3736 and nspawn. (This is really awesome! Ask us for details!)
3738 * The fstab parser is now implemented as generator, not inside
3741 * systemctl will now warn you if .mount units generated from
3742 /etc/fstab are out of date due to changes in fstab that
3743 have not been read by systemd yet.
3745 * systemd is now suitable for usage in initrds. Dracut has
3746 already been updated to make use of this. With this in place
3747 initrds get a slight bit faster but primarily are much
3748 easier to introspect and debug since "systemctl status" in
3749 the host system can be used to introspect initrd services,
3750 and the journal from the initrd is kept around too.
3752 * systemd-delta has been added, a tool to explore differences
3753 between user/admin configuration and vendor defaults.
3755 * PrivateTmp= now affects both /tmp and /var/tmp.
3757 * Boot time status messages are now much prettier and feature
3758 proper english language. Booting up systemd has never been
3761 * Read-ahead pack files now include the inode number of all
3762 files to pre-cache. When the inode changes the pre-caching
3763 is not attempted. This should be nicer to deal with updated
3764 packages which might result in changes of read-ahead
3767 * We now temporaritly lower the kernel's read_ahead_kb variable
3768 when collecting read-ahead data to ensure the kernel's
3769 built-in read-ahead does not add noise to our measurements
3770 of necessary blocks to pre-cache.
3772 * There's now RequiresMountsFor= to add automatic dependencies
3773 for all mounts necessary for a specific file system path.
3775 * MountAuto= and SwapAuto= have been removed from
3776 system.conf. Mounting file systems at boot has to take place
3779 * nspawn now learned a new switch --uuid= to set the machine
3780 ID on the command line.
3782 * nspawn now learned the -b switch to automatically search
3785 * vt102 is now the default TERM for serial TTYs, upgraded from
3788 * systemd-logind now works on VT-less systems.
3790 * The build tree has been reorganized. The individual
3791 components now have directories of their own.
3793 * A new condition type ConditionPathIsReadWrite= is now available.
3795 * nspawn learned the new -C switch to create cgroups for the
3796 container in other hierarchies.
3798 * We now have support for hardware watchdogs, configurable in
3801 * The scheduled shutdown logic now has a public API.
3803 * We now mount /tmp as tmpfs by default, but this can be
3804 masked and /etc/fstab can override it.
3806 * Since udisks does not make use of /media anymore we are not
3807 mounting a tmpfs on it anymore.
3809 * journalctl gained a new --local switch to only interleave
3810 locally generated journal files.
3812 * We can now load the IMA policy at boot automatically.
3814 * The GTK tools have been split off into a systemd-ui.
3816 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3817 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3818 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3819 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3820 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3821 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3822 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3823 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3824 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3829 * This is mostly a bugfix release
3831 * Support optional initialization of the machine ID from the
3832 KVM or container configured UUID.
3834 * Support immediate reboots with "systemctl reboot -ff"
3836 * Show /etc/os-release data in systemd-analyze output
3838 * Many bugfixes for the journal, including endianness fixes and
3839 ensuring that disk space enforcement works
3841 * sd-login.h is C++ comptaible again
3843 * Extend the /etc/os-release format on request of the Debian
3846 * We now refuse non-UTF8 strings used in various configuration
3847 and unit files. This is done to ensure we do not pass invalid
3848 data over D-Bus or expose it elsewhere.
3850 * Register Mimo USB Screens as suitable for automatic seat
3853 * Read SELinux client context from journal clients in a race
3856 * Reorder configuration file lookup order. /etc now always
3857 overrides /run in order to allow the administrator to always
3858 and unconditionally override vendor supplied or
3859 automatically generated data.
3861 * The various user visible bits of the journal now have man
3862 pages. We still lack man pages for the journal API calls
3865 * We now ship all man pages in HTML format again in the
3868 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3869 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3870 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3875 * This is mostly a bugfix release
3877 * systems lacking /etc/os-release are no longer supported.
3879 * Various functionality updates to libsystemd-login.so
3881 * Track class of PAM logins to distuingish greeters from
3884 Contributions from: Kay Sievers, Lennart Poettering, Michael
3889 * This is an important bugfix release for v41.
3891 * Building man pages is now optional which should be useful
3892 for those building systemd from git but unwilling to install
3895 * Watchdog support for supervising services is now usable. In
3896 a future release support for hardware watchdogs
3897 (i.e. /dev/watchdog) will be added building on this.
3899 * Service start rate limiting is now configurable and can be
3900 turned off per service. When a start rate limit is hit a
3901 reboot can automatically be triggered.
3903 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3905 Contributions from: Benjamin Franzke, Bill Nottingham,
3906 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3907 Schmidt, Michał Górny, Piotr Drąg
3911 * The systemd binary is installed /usr/lib/systemd/systemd now;
3912 An existing /sbin/init symlink needs to be adapted with the
3915 * The code that loads kernel modules has been ported to invoke
3916 libkmod directly, instead of modprobe. This means we do not
3917 support systems with module-init-tools anymore.
3919 * Watchdog support is now already useful, but still not
3922 * A new kernel command line option systemd.setenv= is
3923 understood to set system wide environment variables
3924 dynamically at boot.
3926 * We now limit the set of capabilities of systemd-journald.
3928 * We now set SIGPIPE to ignore by default, since it only is
3929 useful in shell pipelines, and has little use in general
3930 code. This can be disabled with IgnoreSIPIPE=no in unit
3933 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3934 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3939 * This is mostly a bugfix release
3941 * We now expose the reason why a service failed in the
3942 "Result" D-Bus property.
3944 * Rudimentary service watchdog support (will be completed over
3945 the next few releases.)
3947 * When systemd forks off in order execute some service we will
3948 now immediately changes its argv[0] to reflect which process
3949 it will execute. This is useful to minimize the time window
3950 with a generic argv[0], which makes bootcharts more useful
3952 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3953 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3954 Mike Kazantsev, Ray Strode
3958 * This is mostly a test release, but incorporates many
3961 * New systemd-cgtop tool to show control groups by their
3964 * Linking against libacl for ACLs is optional again. If
3965 disabled, support tracking device access for active logins
3966 goes becomes unavailable, and so does access to the user
3967 journals by the respective users.
3969 * If a group "adm" exists, journal files are automatically
3970 owned by them, thus allow members of this group full access
3971 to the system journal as well as all user journals.
3973 * The journal now stores the SELinux context of the logging
3974 client for all entries.
3976 * Add C++ inclusion guards to all public headers
3978 * New output mode "cat" in the journal to print only text
3979 messages, without any meta data like date or time.
3981 * Include tiny X server wrapper as a temporary stop-gap to
3982 teach XOrg udev display enumeration. This is used by display
3983 managers such as gdm, and will go away as soon as XOrg
3984 learned native udev hotplugging for display devices.
3986 * Add new systemd-cat tool for executing arbitrary programs
3987 with STDERR/STDOUT connected to the journal. Can also act as
3988 BSD logger replacement, and does so by default.
3990 * Optionally store all locally generated coredumps in the
3991 journal along with meta data.
3993 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3994 writing short strings to files (for usage for /sys), and for
3995 creating symlinks, character and block device nodes.
3997 * New unit file option ControlGroupPersistent= to make cgroups
3998 persistent, following the mechanisms outlined in
3999 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4001 * Support multiple local RTCs in a sane way
4003 * No longer monopolize IO when replaying readahead data on
4004 rotating disks, since we might starve non-file-system IO to
4005 death, since fanotify() will not see accesses done by blkid,
4008 * Do not show kernel threads in systemd-cgls anymore, unless
4009 requested with new -k switch.
4011 Contributions from: Dan Horák, Kay Sievers, Lennart
4012 Poettering, Michal Schmidt
4016 * This is mostly a test release, but incorporates many
4019 * The git repository moved to:
4020 git://anongit.freedesktop.org/systemd/systemd
4021 ssh://git.freedesktop.org/git/systemd/systemd
4023 * First release with the journal
4024 http://0pointer.de/blog/projects/the-journal.html
4026 * The journal replaces both systemd-kmsg-syslogd and
4027 systemd-stdout-bridge.
4029 * New sd_pid_get_unit() API call in libsystemd-logind
4031 * Many systemadm clean-ups
4033 * Introduce remote-fs-pre.target which is ordered before all
4034 remote mounts and may be used to start services before all
4037 * Added Mageia support
4039 * Add bash completion for systemd-loginctl
4041 * Actively monitor PID file creation for daemons which exit in
4042 the parent process before having finished writing the PID
4043 file in the daemon process. Daemons which do this need to be
4044 fixed (i.e. PID file creation must have finished before the
4045 parent exits), but we now react a bit more gracefully to them.
4047 * Add colourful boot output, mimicking the well-known output
4048 of existing distributions.
4050 * New option PassCredentials= for socket units, for
4051 compatibility with a recent kernel ABI breakage.
4053 * /etc/rc.local is now hooked in via a generator binary, and
4054 thus will no longer act as synchronization point during
4057 * systemctl list-unit-files now supports --root=.
4059 * systemd-tmpfiles now understands two new commands: z, Z for
4060 relabelling files according to the SELinux database. This is
4061 useful to apply SELinux labels to specific files in /sys,
4064 * Output of SysV services is now forwarded to both the console
4065 and the journal by default, not only just the console.
4067 * New man pages for all APIs from libsystemd-login.
4069 * The build tree got reorganized and a the build system is a
4070 lot more modular allowing embedded setups to specifically
4071 select the components of systemd they are interested in.
4073 * Support for Linux systems lacking the kernel VT subsystem is
4076 * configure's --with-rootdir= got renamed to
4077 --with-rootprefix= to follow the naming used by udev and
4080 * Unless specified otherwise we will now install to /usr instead
4081 of /usr/local by default.
4083 * Processes with '@' in argv[0][0] are now excluded from the
4084 final shut-down killing spree, following the logic explained
4086 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4088 * All processes remaining in a service cgroup when we enter
4089 the START or START_PRE states are now killed with
4090 SIGKILL. That means it is no longer possible to spawn
4091 background processes from ExecStart= lines (which was never
4092 supported anyway, and bad style).
4094 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4095 reloading of units together.
4097 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4098 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4099 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4100 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4101 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek