1 systemd System and Service Manager
5 * journalctl gained the new options -t/--identifier= to match
6 on the syslog identifier (aka "tag"), as well as --utc to
7 show log timestamps in the UTC timezone. journalctl now also
8 accepts -n/--lines=all to disable line capping in 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 ExecStart
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 * Job timeouts (i.e. time-outs on the time a job that is
26 queued stays in the run queue) can now optionally result in
27 immediate reboot or power-off actions (JobTimeoutAction= and
28 JobTimeoutRebootArgument=). This is useful on ".target"
29 units, to limit the maximum time a target remains
30 undispatched in the run queue, and to trigger an emergency
31 operation in such a case. This is now used by default to
32 turn off the system if boot-up (as defined by everything in
33 basic.target) hangs and does not complete for at least
34 15min. Also, if power-off or reboot hang for at least 30min
35 an immediate power-off/reboot operation is triggered. This
36 functionality is particularly useful to increase reliability
37 on embedded devices, but also on laptops which might
38 accidentally get powered on when carried in a backpack and
39 whose boot stays stuck in a hard disk encryption passphrase
42 * systemd-logind can be configured to also handle lid switch
43 events even when the machine is docked or multiple displays
44 are attached (HandleLidSwitchDocked= option).
46 * A helper binary and a service have been added which can be
47 used to resume from hibernation in the initramfs. A
48 generator will parse the resume= option on the kernel
49 command-line to trigger resume.
51 * A user console daemon systemd-consoled has been
52 added. Currently, it is a preview, and will so far open a
53 single terminal on each session of the user marked as
54 Desktop=systemd-console.
56 * Route metrics can be specified for DHCP routes added by
59 * The SELinux context of socket-actived services can be set
60 from the information provided by the networking stack
61 (SELinuxContextFromNet= option).
63 * Userspace firmware loading support has been removed and
64 the minimum supported kernel version is thus bumped to 3.7.
66 * Timeout for udev workers has been increased from 1 to 3
67 minutes, but a warning will be printed after 1 minute to
68 help diagnose kernel modules that take a long time to load.
70 * Udev rules can now remove tags on devices with TAG-="foobar".
72 * systemd's readahead implementation has been removed. In many
73 circumstatances it didn't give expected benefits even for
74 rotational disk drives and was becoming less relevant in the
75 age of SSDs. As none of the developers has been using
76 rotating media anymore, and nobody stepped up to actively
77 maintain this component of systemd it has now been removed.
79 * Swap units can use Discard= to specify discard options.
80 Discard options specified for swaps in /etc/fstab are now
83 * Docker containers are now detected as a separate type of
86 * The Password Agent protocol gained support for queries where
87 the user input is shown, useful e.g. for usernames.
88 systemd-ask-password gained a new --echo option to turn that
91 * The default sysctl.d/ snippets will now set:
93 net.core.default_qdisc = fq_codel
95 This selects Fair Queueing Controlled Delay as the default
96 queueing discipline for network interfaces. fq_codel helps
97 fight the network bufferbloat problem. It is believed to be
98 a good default with no tuning required for most workloads.
99 Downstream distributions may override this choice. On 10Gbit
100 servers that do not do forwarding, "fq" may perform better.
101 Systems without a good clocksource should use "pfifo_fast".
103 * If kdbus is enabled during build a new option BusPolicy= is
104 available for service units, that allows locking all service
105 processes into a stricter bus policy, in order to limit
106 access to various bus services, or even hide most of them
107 from the service's view entirely.
109 * networkctl will now show the .network and .link file
110 networkd has applied to a specific interface.
112 * sd-login gained a new API call sd_session_get_desktop() to
113 query which desktop environment has been selected for a
116 * UNIX utmp support is now compile-time optional to support
119 * systemctl gained two new commands "add-wants" and
120 "add-requires" for pulling in units from specific targets
123 * If the word "rescue" is specified on the kernel command line
124 the system will now boot into rescue mode (aka
125 rescue.target), which was previously available only by
126 specifying "1" or "systemd.unit=rescue.target" on the kernel
127 command line. This new kernel command line option nicely
128 mirrors the already existing "emergency" kernel command line
131 * New kernel command line options mount.usr=, mount.usrflags=,
132 usrfstype= have been added that match root=, rootflags=,
133 rootfstype= but allow mounting a specific file system to
136 * The $NOTIFY_SOCKET is now also passed to control processesof
137 services, not only the main process.
139 * This version reenables support for fsck's -l switch. This
140 means at least version v2.25 of util-linux is required for
141 operation, otherwise dead-locks on device nodes may
142 occur. Again: you need to update util-linux to at least
143 v2.25 when updating systemd to v217.
145 * The "multi-seat-x" tool has been removed from systemd, as
146 its functionality has been integrated into X servers 1.16,
147 and the tool is hence redundant. It is recommended to update
148 display managers invoking this tool to simply invoke X
149 directly from now on, again.
151 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
152 message flag has been added for all of systemd's PolicyKit
153 authenticated method calls has been added. In particular
154 this now allows optional interactive authorization via
155 PolicyKit for many of PID1's priviliged operations such as
156 unit file enabling and disabling.
160 * timedated no longer reads NTP implementation unit names from
161 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
162 implementations should add a
164 Conflicts=systemd-timesyncd.service
166 to their unit files to take over and replace systemd's NTP
167 default functionality.
169 * systemd-sysusers gained a new line type "r" for configuring
170 which UID/GID ranges to allocate system users/groups
171 from. Lines of type "u" may now add an additional column
172 that specifies the home directory for the system user to be
173 created. Also, systemd-sysusers may now optionally read user
174 information from STDIN instead of a file. This is useful for
175 invoking it from RPM preinst scriptlets that need to create
176 users before the first RPM file is installed since these
177 files might need to be owned by them. A new
178 %sysusers_create_inline RPM macro has been introduced to do
179 just that. systemd-sysusers now updates the shadow files as
180 well as the user/group databases, which should enhance
181 compatibility with certain tools like grpck.
183 * A number of bus APIs of PID 1 now optionally consult
184 PolicyKit to permit access for otherwise unprivileged
185 clients under certain conditions. Note that this currently
186 doesn't support interactive authentication yet, but this is
187 expected to be added eventually, too.
189 * /etc/machine-info now has new fields for configuring the
190 deployment environment of the machine, as well as the
191 location of the machine. hostnamectl has been updated with
192 new command to update these fields.
194 * systemd-timesyncd has been updated to automatically acquire
195 NTP server information from systemd-networkd, which might
196 have been discovered via DHCP.
198 * systemd-resolved now includes a caching DNS stub resolver
199 and a complete LLMNR name resolution implementation. A new
200 NSS module "nss-resolve" has been added which can be used
201 instead of glibc's own "nss-dns" to resolve hostnames via
202 systemd-resolved. Hostnames, addresses and arbitrary RRs may
203 be resolved via systemd-resolved D-Bus APIs. In contrast to
204 the glibc internal resolver systemd-resolved is aware of
205 multi-homed system, and keeps DNS server and caches separate
206 and per-interface. Queries are sent simultaneously on all
207 interfaces that have DNS servers configured, in order to
208 properly handle VPNs and local LANs which might resolve
209 separate sets of domain names. systemd-resolved may acquire
210 DNS server information from systemd-networkd automatically,
211 which in turn might have discovered them via DHCP. A tool
212 "systemd-resolve-host" has been added that may be used to
213 query the DNS logic in resolved. systemd-resolved implements
214 IDNA and automatically uses IDNA or UTF-8 encoding depending
215 on whether classic DNS or LLMNR is used as transport. In the
216 next releases we intend to add a DNSSEC and mDNS/DNS-SD
217 implementation to systemd-resolved.
219 * A new NSS module nss-mymachines has been added, that
220 automatically resolves the names of all local registered
221 containers to their respective IP addresses.
223 * A new client tool "networkctl" for systemd-networkd has been
224 added. It currently is entirely passive and will query
225 networking configuration from udev, rtnetlink and networkd,
226 and present it to the user in a very friendly
227 way. Eventually, we hope to extend it to become a full
228 control utility for networkd.
230 * .socket units gained a new DeferAcceptSec= setting that
231 controls the kernels' TCP_DEFER_ACCEPT sockopt for
232 TCP. Similar, support for controlling TCP keep-alive
233 settings has been added (KeepAliveTimeSec=,
234 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
235 turning off Nagle's algorithm on TCP has been added
238 * logind learned a new session type "web", for use in projects
239 like Cockpit which register web clients as PAM sessions.
241 * timer units with at least one OnCalendar= setting will now
242 be started only after timer-sync.target has been
243 reached. This way they will not elapse before the system
244 clock has been corrected by a local NTP client or
245 similar. This is particular useful on RTC-less embedded
246 machines, that come up with an invalid system clock.
248 * systemd-nspawn's --network-veth= switch should now result in
249 stable MAC addresses for both the outer and the inner side
252 * systemd-nspawn gained a new --volatile= switch for running
253 container instances with /etc or /var unpopulated.
255 * The kdbus client code has been updated to use the new Linux
256 3.17 memfd subsystem instead of the old kdbus-specific one.
258 * systemd-networkd's DHCP client and server now support
259 FORCERENEW. There are also new configuration options to
260 configure the vendor client identifier and broadcast mode
263 * systemd will no longer inform the kernel about the current
264 timezone, as this is necessarily incorrect and racy as the
265 kernel has no understanding of DST and similar
266 concepts. This hence means FAT timestamps will be always
267 considered UTC, similar to what Android is already
268 doing. Also, when the RTC is configured to the local time
269 (rather than UTC) systemd will never synchronize back to it,
270 as this might confuse Windows at a later boot.
272 * systemd-analyze gained a new command "verify" for offline
273 validation of unit files.
275 * systemd-networkd gained support for a couple of additional
276 settings for bonding networking setups. Also, the metric for
277 statically configured routes may now be configured. For
278 network interfaces where this is appropriate the peer IP
279 address may now be configured.
281 * systemd-networkd's DHCP client will no longer request
282 broadcasting by default, as this tripped up some networks.
283 For hardware where broadcast is required the feature should
284 be switched back on using RequestBroadcast=yes.
286 * systemd-networkd will now set up IPv4LL addresses (when
287 enabled) even if DHCP is configured successfully.
289 * udev will now default to respect network device names given
290 by the kernel when the kernel indicates that these are
291 predictable. This behavior can be tweaked by changing
292 NamePolicy= in the relevant .link file.
294 * A new library systemd-terminal has been added that
295 implements full TTY stream parsing and rendering. This
296 library is supposed to be used later on for implementing a
297 full userspace VT subsystem, replacing the current kernel
300 * A new tool systemd-journal-upload has been added to push
301 journal data to a remote system running
302 systemd-journal-remote.
304 * journald will no longer forward all local data to another
305 running syslog daemon. This change has been made because
306 rsyslog (which appears to be the most commonly used syslog
307 implementation these days) no longer makes use of this, and
308 instead pulls the data out of the journal on its own. Since
309 forwarding the messages to a non-existent syslog server is
310 more expensive than we assumed we have now turned this
311 off. If you run a syslog server that is not a recent rsyslog
312 version, you have to turn this option on again
313 (ForwardToSyslog= in journald.conf).
315 * journald now optionally supports the LZ4 compressor for
316 larger journal fields. This compressor should perform much
317 better than XZ which was the previous default.
319 * machinectl now shows the IP addresses of local containers,
320 if it knows them, plus the interface name of the container.
322 * A new tool "systemd-escape" has been added that makes it
323 easy to escape strings to build unit names and similar.
325 * sd_notify() messages may now include a new ERRNO= field
326 which is parsed and collected by systemd and shown among the
327 "systemctl status" output for a service.
329 * A new component "systemd-firstboot" has been added that
330 queries the most basic systemd information (timezone,
331 hostname, root password) interactively on first
332 boot. Alternatively it may also be used to provision these
333 things offline on OS images installed into directories.
335 * The default sysctl.d/ snippets will now set
337 net.ipv4.conf.default.promote_secondaries=1
339 This has the benefit of no flushing secondary IP addresses
340 when primary addresses are removed.
342 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
343 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
344 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
345 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
346 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
347 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
348 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
349 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
350 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
351 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
352 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
353 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
354 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
355 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
356 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
358 -- Berlin, 2014-08-19
362 * A new tool systemd-sysusers has been added. This tool
363 creates system users and groups in /etc/passwd and
364 /etc/group, based on static declarative system user/group
365 definitions in /usr/lib/sysusers.d/. This is useful to
366 enable factory resets and volatile systems that boot up with
367 an empty /etc directory, and thus need system users and
368 groups created during early boot. systemd now also ships
369 with two default sysusers.d/ files for the most basic
370 users and groups systemd and the core operating system
373 * A new tmpfiles snippet has been added that rebuilds the
374 essential files in /etc on boot, should they be missing.
376 * A directive for ensuring automatic clean-up of
377 /var/cache/man/ has been removed from the default
378 configuration. This line should now be shipped by the man
379 implementation. The necessary change has been made to the
380 man-db implementation. Note that you need to update your man
381 implementation to one that ships this line, otherwise no
382 automatic clean-up of /var/cache/man will take place.
384 * A new condition ConditionNeedsUpdate= has been added that
385 may conditionalize services to only run when /etc or /var
386 are "older" than the vendor operating system resources in
387 /usr. This is useful for reconstructing or updating /etc
388 after an offline update of /usr or a factory reset, on the
389 next reboot. Services that want to run once after such an
390 update or reset should use this condition and order
391 themselves before the new systemd-update-done.service, which
392 will mark the two directories as fully updated. A number of
393 service files have been added making use of this, to rebuild
394 the udev hardware database, the journald message catalog and
395 dynamic loader cache (ldconfig). The systemd-sysusers tool
396 described above also makes use of this now. With this in
397 place it is now possible to start up a minimal operating
398 system with /etc empty cleanly. For more information on the
399 concepts involved see this recent blog story:
401 http://0pointer.de/blog/projects/stateless.html
403 * A new system group "input" has been introduced, and all
404 input device nodes get this group assigned. This is useful
405 for system-level software to get access to input devices. It
406 complements what is already done for "audio" and "video".
408 * systemd-networkd learnt minimal DHCPv4 server support in
409 addition to the existing DHCPv4 client support. It also
410 learnt DHCPv6 client and IPv6 Router Solicitation client
411 support. The DHCPv4 client gained support for static routes
412 passed in from the server. Note that the [DHCPv4] section
413 known in older systemd-networkd versions has been renamed to
414 [DHCP] and is now also used by the DHCPv6 client. Existing
415 .network files using settings of this section should be
416 updated, though compatibility is maintained. Optionally, the
417 client hostname may now be sent to the DHCP server.
419 * networkd gained support for vxlan virtual networks as well
420 as tun/tap and dummy devices.
422 * networkd gained support for automatic allocation of address
423 ranges for interfaces from a system-wide pool of
424 addresses. This is useful for dynamically managing a large
425 number of interfaces with a single network configuration
426 file. In particular this is useful to easily assign
427 appropriate IP addresses to the veth links of a large number
430 * RPM macros for processing sysusers, sysctl and binfmt
431 drop-in snippets at package installation time have been
434 * The /etc/os-release file should now be placed in
435 /usr/lib/os-release. The old location is automatically
436 created as symlink. /usr/lib is the more appropriate
437 location of this file, since it shall actually describe the
438 vendor operating system shipped in /usr, and not the
439 configuration stored in /etc.
441 * .mount units gained a new boolean SloppyOptions= setting
442 that maps to mount(8)'s -s option which enables permissive
443 parsing of unknown mount options.
445 * tmpfiles learnt a new "L+" directive which creates a symlink
446 but (unlike "L") deletes a pre-existing file first, should
447 it already exist and not already be the correct
448 symlink. Similar, "b+", "c+" and "p+" directives have been
449 added as well, which create block and character devices, as
450 well as fifos in the filesystem, possibly removing any
451 pre-existing files of different types.
453 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
454 'argument' field (which so far specified the source to
455 symlink/copy the files from) is now optional. If omitted the
456 same file os copied from /usr/share/factory/ suffixed by the
457 full destination path. This is useful for populating /etc
458 with essential files, by copying them from vendor defaults
459 shipped in /usr/share/factory/etc.
461 * A new command "systemctl preset-all" has been added that
462 applies the service preset settings to all installed unit
463 files. A new switch --preset-mode= has been added that
464 controls whether only enable or only disable operations
467 * A new command "systemctl is-system-running" has been added
468 that allows checking the overall state of the system, for
469 example whether it is fully up and running.
471 * When the system boots up with an empty /etc, the equivalent
472 to "systemctl preset-all" is executed during early boot, to
473 make sure all default services are enabled after a factory
476 * systemd now contains a minimal preset file that enables the
477 most basic services systemd ships by default.
479 * Unit files' [Install] section gained a new DefaultInstance=
480 field for defining the default instance to create if a
481 template unit is enabled with no instance specified.
483 * A new passive target cryptsetup-pre.target has been added
484 that may be used by services that need to make they run and
485 finish before the first LUKS cryptographic device is set up.
487 * The /dev/loop-control and /dev/btrfs-control device nodes
488 are now owned by the "disk" group by default, opening up
489 access to this group.
491 * systemd-coredump will now automatically generate a
492 stack trace of all core dumps taking place on the system,
493 based on elfutils' libdw library. This stack trace is logged
496 * systemd-coredump may now optionally store coredumps directly
497 on disk (in /var/lib/systemd/coredump, possibly compressed),
498 instead of storing them unconditionally in the journal. This
499 mode is the new default. A new configuration file
500 /etc/systemd/coredump.conf has been added to configure this
501 and other parameters of systemd-coredump.
503 * coredumpctl gained a new "info" verb to show details about a
504 specific coredump. A new switch "-1" has also been added
505 that makes sure to only show information about the most
506 recent entry instead of all entries. Also, as the tool is
507 generally useful now the "systemd-" prefix of the binary
508 name has been removed. Distributions that want to maintain
509 compatibility with the old name should add a symlink from
510 the old name to the new name.
512 * journald's SplitMode= now defaults to "uid". This makes sure
513 that unprivileged users can access their own coredumps with
514 coredumpctl without restrictions.
516 * New kernel command line options "systemd.wants=" (for
517 pulling an additional unit during boot), "systemd.mask="
518 (for masking a specific unit for the boot), and
519 "systemd.debug-shell" (for enabling the debug shell on tty9)
520 have been added. This is implemented in the new generator
521 "systemd-debug-generator".
523 * systemd-nspawn will now by default filter a couple of
524 syscalls for containers, among them those required for
525 kernel module loading, direct x86 IO port access, swap
526 management, and kexec. Most importantly though
527 open_by_handle_at() is now prohibited for containers,
528 closing a hole similar to a recently discussed vulnerability
529 in docker regarding access to files on file hierarchies the
530 container should normally not have access to. Note that for
531 nspawn we generally make no security claims anyway (and
532 this is explicitly documented in the man page), so this is
533 just a fix for one of the most obvious problems.
535 * A new man page file-hierarchy(7) has been added that
536 contains a minimized, modernized version of the file system
537 layout systemd expects, similar in style to the FHS
538 specification or hier(5). A new tool systemd-path(1) has
539 been added to query many of these paths for the local
542 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
543 longer done. Since the directory now has a per-user size
544 limit, and is cleaned on logout this appears unnecessary,
545 in particular since this now brings the lifecycle of this
546 directory closer in line with how IPC objects are handled.
548 * systemd.pc now exports a number of additional directories,
549 including $libdir (which is useful to identify the library
550 path for the primary architecture of the system), and a
551 couple of drop-in directories.
553 * udev's predictable network interface names now use the dev_port
554 sysfs attribute, introduced in linux 3.15 instead of dev_id to
555 distinguish between ports of the same PCI function. dev_id should
556 only be used for ports using the same HW address, hence the need
559 * machined has been updated to export the OS version of a
560 container (read from /etc/os-release and
561 /usr/lib/os-release) on the bus. This is now shown in
562 "machinectl status" for a machine.
564 * A new service setting RestartForceExitStatus= has been
565 added. If configured to a set of exit signals or process
566 return values, the service will be restarted when the main
567 daemon process exits with any of them, regardless of the
570 * systemctl's -H switch for connecting to remote systemd
571 machines has been extended so that it may be used to
572 directly connect to a specific container on the
573 host. "systemctl -H root@foobar:waldi" will now connect as
574 user "root" to host "foobar", and then proceed directly to
575 the container named "waldi". Note that currently you have to
576 authenticate as user "root" for this to work, as entering
577 containers is a privileged operation.
579 Contributions from: Andreas Henriksson, Benjamin Steinwender,
580 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
581 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
582 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
583 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
584 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
585 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
586 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
587 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
588 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
589 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
590 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
592 -- Berlin, 2014-07-03
596 * As an experimental feature, udev now tries to lock the
597 disk device node (flock(LOCK_SH|LOCK_NB)) while it
598 executes events for the disk or any of its partitions.
599 Applications like partitioning programs can lock the
600 disk device node (flock(LOCK_EX)) and claim temporary
601 device ownership that way; udev will entirely skip all event
602 handling for this disk and its partitions. If the disk
603 was opened for writing, the close will trigger a partition
604 table rescan in udev's "watch" facility, and if needed
605 synthesize "change" events for the disk and all its partitions.
606 This is now unconditionally enabled, and if it turns out to
607 cause major problems, we might turn it on only for specific
608 devices, or might need to disable it entirely. Device Mapper
609 devices are excluded from this logic.
611 * We temporarily dropped the "-l" switch for fsck invocations,
612 since they collide with the flock() logic above. util-linux
613 upstream has been changed already to avoid this conflict,
614 and we will readd "-l" as soon as util-linux with this
615 change has been released.
617 * The dependency on libattr has been removed. Since a long
618 time, the extended attribute calls have moved to glibc, and
619 libattr is thus unnecessary.
621 * Virtualization detection works without priviliges now. This
622 means the systemd-detect-virt binary no longer requires
623 CAP_SYS_PTRACE file capabilities, and our daemons can run
624 with fewer privileges.
626 * systemd-networkd now runs under its own "systemd-network"
627 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
628 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
629 loses the ability to write to files owned by root this way.
631 * Similar, systemd-resolved now runs under its own
632 "systemd-resolve" user with no capabilities remaining.
634 * Similar, systemd-bus-proxyd now runs under its own
635 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
637 * systemd-networkd gained support for setting up "veth"
638 virtual ethernet devices for container connectivity, as well
639 as GRE and VTI tunnels.
641 * systemd-networkd will no longer automatically attempt to
642 manually load kernel modules necessary for certain tunnel
643 transports. Instead, it is assumed the kernel loads them
644 automatically when required. This only works correctly on
645 very new kernels. On older kernels, please consider adding
646 the kernel modules to /etc/modules-load.d/ as a work-around.
648 * The resolv.conf file systemd-resolved generates has been
649 moved to /run/systemd/resolve/. If you have a symlink from
650 /etc/resolv.conf, it might be necessary to correct it.
652 * Two new service settings, ProtectHome= and ProtectSystem=,
653 have been added. When enabled, they will make the user data
654 (such as /home) inaccessible or read-only and the system
655 (such as /usr) read-only, for specific services. This allows
656 very light-weight per-service sandboxing to avoid
657 modifications of user data or system files from
658 services. These two new switches have been enabled for all
659 of systemd's long-running services, where appropriate.
661 * Socket units gained new SocketUser= and SocketGroup=
662 settings to set the owner user and group of AF_UNIX sockets
663 and FIFOs in the file system.
665 * Socket units gained a new RemoveOnStop= setting. If enabled,
666 all FIFOS and sockets in the file system will be removed
667 when the specific socket unit is stopped.
669 * Socket units gained a new Symlinks= setting. It takes a list
670 of symlinks to create to file system sockets or FIFOs
671 created by the specific Unix sockets. This is useful to
672 manage symlinks to socket nodes with the same life-cycle as
675 * The /dev/log socket and /dev/initctl FIFO have been moved to
676 /run, and have been replaced by symlinks. This allows
677 connecting to these facilities even if PrivateDevices=yes is
678 used for a service (which makes /dev/log itself unavailable,
679 but /run is left). This also has the benefit of ensuring
680 that /dev only contains device nodes, directories and
681 symlinks, and nothing else.
683 * sd-daemon gained two new calls sd_pid_notify() and
684 sd_pid_notifyf(). They are similar to sd_notify() and
685 sd_notifyf(), but allow overriding of the source PID of
686 notification messages if permissions permit this. This is
687 useful to send notify messages on behalf of a different
688 process (for example, the parent process). The
689 systemd-notify tool has been updated to make use of this
690 when sending messages (so that notification messages now
691 originate from the shell script invoking systemd-notify and
692 not the systemd-notify process itself. This should minimize
693 a race where systemd fails to associate notification
694 messages to services when the originating process already
697 * A new "on-abnormal" setting for Restart= has been added. If
698 set, it will result in automatic restarts on all "abnormal"
699 reasons for a process to exit, which includes unclean
700 signals, core dumps, timeouts and watchdog timeouts, but
701 does not include clean and unclean exit codes or clean
702 signals. Restart=on-abnormal is an alternative for
703 Restart=on-failure for services that shall be able to
704 terminate and avoid restarts on certain errors, by
705 indicating so with an unclean exit code. Restart=on-failure
706 or Restart=on-abnormal is now the recommended setting for
707 all long-running services.
709 * If the InaccessibleDirectories= service setting points to a
710 mount point (or if there are any submounts contained within
711 it), it is now attempted to completely unmount it, to make
712 the file systems truly unavailable for the respective
715 * The ReadOnlyDirectories= service setting and
716 systemd-nspawn's --read-only parameter are now recursively
717 applied to all submounts, too.
719 * Mount units may now be created transiently via the bus APIs.
721 * The support for SysV and LSB init scripts has been removed
722 from the systemd daemon itself. Instead, it is now
723 implemented as a generator that creates native systemd units
724 from these scripts when needed. This enables us to remove a
725 substantial amount of legacy code from PID 1, following the
726 fact that many distributions only ship a very small number
727 of LSB/SysV init scripts nowadays.
729 * Priviliged Xen (dom0) domains are not considered
730 virtualization anymore by the virtualization detection
731 logic. After all, they generally have unrestricted access to
732 the hardware and usually are used to manage the unprivileged
735 * systemd-tmpfiles gained a new "C" line type, for copying
736 files or entire directories.
738 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
739 lines. So far, they have been non-globbing versions of the
740 latter, and have thus been redundant. In future, it is
741 recommended to only use "z". "m" has hence been removed
742 from the documentation, even though it stays supported.
744 * A tmpfiles snippet to recreate the most basic structure in
745 /var has been added. This is enough to create the /var/run →
746 /run symlink and create a couple of structural
747 directories. This allows systems to boot up with an empty or
748 volatile /var. Of course, while with this change, the core OS
749 now is capable with dealing with a volatile /var, not all
750 user services are ready for it. However, we hope that sooner
751 or later, many service daemons will be changed upstream so
752 that they are able to automatically create their necessary
753 directories in /var at boot, should they be missing. This is
754 the first step to allow state-less systems that only require
755 the vendor image for /usr to boot.
757 * systemd-nspawn has gained a new --tmpfs= switch to mount an
758 empty tmpfs instance to a specific directory. This is
759 particularly useful for making use of the automatic
760 reconstruction of /var (see above), by passing --tmpfs=/var.
762 * Access modes specified in tmpfiles snippets may now be
763 prefixed with "~", which indicates that they shall be masked
764 by whether the existing file or directory is currently
765 writable, readable or executable at all. Also, if specified,
766 the sgid/suid/sticky bits will be masked for all
769 * A new passive target unit "network-pre.target" has been
770 added which is useful for services that shall run before any
771 network is configured, for example firewall scripts.
773 * The "floppy" group that previously owned the /dev/fd*
774 devices is no longer used. The "disk" group is now used
775 instead. Distributions should probably deprecate usage of
778 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
779 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
780 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
781 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
782 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
783 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
786 -- Berlin, 2014-06-11
790 * A new "systemd-timesyncd" daemon has been added for
791 synchronizing the system clock across the network. It
792 implements an SNTP client. In contrast to NTP
793 implementations such as chrony or the NTP reference server,
794 this only implements a client side, and does not bother with
795 the full NTP complexity, focusing only on querying time from
796 one remote server and synchronizing the local clock to
797 it. Unless you intend to serve NTP to networked clients or
798 want to connect to local hardware clocks, this simple NTP
799 client should be more than appropriate for most
800 installations. The daemon runs with minimal privileges, and
801 has been hooked up with networkd to only operate when
802 network connectivity is available. The daemon saves the
803 current clock to disk every time a new NTP sync has been
804 acquired, and uses this to possibly correct the system clock
805 early at bootup, in order to accommodate for systems that
806 lack an RTC such as the Raspberry Pi and embedded devices,
807 and to make sure that time monotonically progresses on these
808 systems, even if it is not always correct. To make use of
809 this daemon, a new system user and group "systemd-timesync"
810 needs to be created on installation of systemd.
812 * The queue "seqnum" interface of libudev has been disabled, as
813 it was generally incompatible with device namespacing as
814 sequence numbers of devices go "missing" if the devices are
815 part of a different namespace.
817 * "systemctl list-timers" and "systemctl list-sockets" gained
818 a --recursive switch for showing units of these types also
819 for all local containers, similar in style to the already
820 supported --recursive switch for "systemctl list-units".
822 * A new RebootArgument= setting has been added for service
823 units, which may be used to specify a kernel reboot argument
824 to use when triggering reboots with StartLimitAction=.
826 * A new FailureAction= setting has been added for service
827 units which may be used to specify an operation to trigger
828 when a service fails. This works similarly to
829 StartLimitAction=, but unlike it, controls what is done
830 immediately rather than only after several attempts to
831 restart the service in question.
833 * hostnamed got updated to also expose the kernel name,
834 release, and version on the bus. This is useful for
835 executing commands like hostnamectl with the -H switch.
836 systemd-analyze makes use of this to properly display
837 details when running non-locally.
839 * The bootchart tool can now show cgroup information in the
842 * The CFS CPU quota cgroup attribute is now exposed for
843 services. The new CPUQuota= switch has been added for this
844 which takes a percentage value. Setting this will have the
845 result that a service may never get more CPU time than the
846 specified percentage, even if the machine is otherwise idle.
848 * systemd-networkd learned IPIP and SIT tunnel support.
850 * LSB init scripts exposing a dependency on $network will now
851 get a dependency on network-online.target rather than simply
852 network.target. This should bring LSB handling closer to
853 what it was on SysV systems.
855 * A new fsck.repair= kernel option has been added to control
856 how fsck shall deal with unclean file systems at boot.
858 * The (.ini) configuration file parser will now silently
859 ignore sections whose name begins with "X-". This may be
860 used to maintain application-specific extension sections in unit
863 * machined gained a new API to query the IP addresses of
864 registered containers. "machinectl status" has been updated
865 to show these addresses in its output.
867 * A new call sd_uid_get_display() has been added to the
868 sd-login APIs for querying the "primary" session of a
869 user. The "primary" session of the user is elected from the
870 user's sessions and generally a graphical session is
871 preferred over a text one.
873 * A minimal systemd-resolved daemon has been added. It
874 currently simply acts as a companion to systemd-networkd and
875 manages resolv.conf based on per-interface DNS
876 configuration, possibly supplied via DHCP. In the long run
877 we hope to extend this into a local DNSSEC enabled DNS and
880 * The systemd-networkd-wait-online tool is now enabled by
881 default. It will delay network-online.target until a network
882 connection has been configured. The tool primarily integrates
883 with networkd, but will also make a best effort to make sense
884 of network configuration performed in some other way.
886 * Two new service options StartupCPUShares= and
887 StartupBlockIOWeight= have been added that work similarly to
888 CPUShares= and BlockIOWeight= however only apply during
889 system startup. This is useful to prioritize certain services
890 differently during bootup than during normal runtime.
892 * hostnamed has been changed to prefer the statically
893 configured hostname in /etc/hostname (unless set to
894 'localhost' or empty) over any dynamic one supplied by
895 dhcp. With this change, the rules for picking the hostname
896 match more closely the rules of other configuration settings
897 where the local administrator's configuration in /etc always
898 overrides any other settings.
900 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
901 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
902 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
903 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
904 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
905 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
906 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
907 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
908 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
909 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
910 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
911 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
912 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
913 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
914 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
915 Lindskog, WaLyong Cho, Will Woods, Zbigniew
918 -- Beijing, 2014-05-28
922 * When restoring the screen brightness at boot, stay away from
923 the darkest setting or from the lowest 5% of the available
924 range, depending on which is the larger value of both. This
925 should effectively protect the user from rebooting into a
926 black screen, should the brightness have been set to minimum
929 * sd-login gained a new sd_machine_get_class() call to
930 determine the class ("vm" or "container") of a machine
931 registered with machined.
933 * sd-login gained new calls
934 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
935 to query the identity of the peer of a local AF_UNIX
936 connection. They operate similarly to their sd_pid_get_xyz()
939 * PID 1 will now maintain a system-wide system state engine
940 with the states "starting", "running", "degraded",
941 "maintenance", "stopping". These states are bound to system
942 startup, normal runtime, runtime with at least one failed
943 service, rescue/emergency mode and system shutdown. This
944 state is shown in the "systemctl status" output when no unit
945 name is passed. It is useful to determine system state, in
946 particularly when doing so for many systems or containers at
949 * A new command "list-machines" has been added to "systemctl"
950 that lists all local OS containers and shows their system
951 state (see above), if systemd runs inside of them.
953 * systemctl gained a new "-r" switch to recursively enumerate
954 units on all local containers, when used with the
955 "list-unit" command (which is the default one that is
956 executed when no parameters are specified).
958 * The GPT automatic partition discovery logic will now honour
959 two GPT partition flags: one may be set on a partition to
960 cause it to be mounted read-only, and the other may be set
961 on a partition to ignore it during automatic discovery.
963 * Two new GPT type UUIDs have been added for automatic root
964 partition discovery, for 32-bit and 64-bit ARM. This is not
965 particularly useful for discovering the root directory on
966 these architectures during bare-metal boots (since UEFI is
967 not common there), but still very useful to allow booting of
968 ARM disk images in nspawn with the -i option.
970 * MAC addresses of interfaces created with nspawn's
971 --network-interface= switch will now be generated from the
972 machine name, and thus be stable between multiple invocations
975 * logind will now automatically remove all IPC objects owned
976 by a user if she or he fully logs out. This makes sure that
977 users who are logged out cannot continue to consume IPC
978 resources. This covers SysV memory, semaphores and message
979 queues as well as POSIX shared memory and message
980 queues. Traditionally, SysV and POSIX IPC had no life-cycle
981 limits. With this functionality, that is corrected. This may
982 be turned off by using the RemoveIPC= switch of logind.conf.
984 * The systemd-machine-id-setup and tmpfiles tools gained a
985 --root= switch to operate on a specific root directory,
988 * journald can now forward logged messages to the TTYs of all
989 logged in users ("wall"). This is the default for all
990 emergency messages now.
992 * A new tool systemd-journal-remote has been added to stream
993 journal log messages across the network.
995 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
996 controller trees are mounted into it. Note that the
997 directories mounted beneath it are not read-only. This is a
998 security measure and is particularly useful because glibc
999 actually includes a search logic to pick any tmpfs it can
1000 find to implement shm_open() if /dev/shm is not available
1001 (which it might very well be in namespaced setups).
1003 * machinectl gained a new "poweroff" command to cleanly power
1004 down a local OS container.
1006 * The PrivateDevices= unit file setting will now also drop the
1007 CAP_MKNOD capability from the capability bound set, and
1008 imply DevicePolicy=closed.
1010 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1011 comprehensively on all long-running systemd services where
1012 this is appropriate.
1014 * systemd-udevd will now run in a disassociated mount
1015 namespace. To mount directories from udev rules, make sure to
1016 pull in mount units via SYSTEMD_WANTS properties.
1018 * The kdbus support gained support for uploading policy into
1019 the kernel. sd-bus gained support for creating "monitoring"
1020 connections that can eavesdrop into all bus communication
1021 for debugging purposes.
1023 * Timestamps may now be specified in seconds since the UNIX
1024 epoch Jan 1st, 1970 by specifying "@" followed by the value
1027 * Native tcpwrap support in systemd has been removed. tcpwrap
1028 is old code, not really maintained anymore and has serious
1029 shortcomings, and better options such as firewalls
1030 exist. For setups that require tcpwrap usage, please
1031 consider invoking your socket-activated service via tcpd,
1032 like on traditional inetd.
1034 * A new system.conf configuration option
1035 DefaultTimerAccuracySec= has been added that controls the
1036 default AccuracySec= setting of .timer units.
1038 * Timer units gained a new WakeSystem= switch. If enabled,
1039 timers configured this way will cause the system to resume
1040 from system suspend (if the system supports that, which most
1043 * Timer units gained a new Persistent= switch. If enabled,
1044 timers configured this way will save to disk when they have
1045 been last triggered. This information is then used on next
1046 reboot to possible execute overdue timer events, that
1047 could not take place because the system was powered off.
1048 This enables simple anacron-like behaviour for timer units.
1050 * systemctl's "list-timers" will now also list the time a
1051 timer unit was last triggered in addition to the next time
1052 it will be triggered.
1054 * systemd-networkd will now assign predictable IPv4LL
1055 addresses to its local interfaces.
1057 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1058 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1059 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1060 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1061 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1062 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1063 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1064 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1067 -- Berlin, 2014-03-25
1071 * A new unit file setting RestrictAddressFamilies= has been
1072 added to restrict which socket address families unit
1073 processes gain access to. This takes address family names
1074 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1075 attack surface of services via exotic protocol stacks. This
1076 is built on seccomp system call filters.
1078 * Two new unit file settings RuntimeDirectory= and
1079 RuntimeDirectoryMode= have been added that may be used to
1080 manage a per-daemon runtime directories below /run. This is
1081 an alternative for setting up directory permissions with
1082 tmpfiles snippets, and has the advantage that the runtime
1083 directory's lifetime is bound to the daemon runtime and that
1084 the daemon starts up with an empty directory each time. This
1085 is particularly useful when writing services that drop
1086 privileges using the User= or Group= setting.
1088 * The DeviceAllow= unit setting now supports globbing for
1089 matching against device group names.
1091 * The systemd configuration file system.conf gained new
1092 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1093 DefaultMemoryAccounting= to globally turn on/off accounting
1094 for specific resources (cgroups) for all units. These
1095 settings may still be overridden individually in each unit
1098 * systemd-gpt-auto-generator is now able to discover /srv and
1099 root partitions in addition to /home and swap partitions. It
1100 also supports LUKS-encrypted partitions now. With this in
1101 place, automatic discovery of partitions to mount following
1102 the Discoverable Partitions Specification
1103 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1104 is now a lot more complete. This allows booting without
1105 /etc/fstab and without root= on the kernel command line on
1106 systems prepared appropriately.
1108 * systemd-nspawn gained a new --image= switch which allows
1109 booting up disk images and Linux installations on any block
1110 device that follow the Discoverable Partitions Specification
1111 (see above). This means that installations made with
1112 appropriately updated installers may now be started and
1113 deployed using container managers, completely
1114 unmodified. (We hope that libvirt-lxc will add support for
1115 this feature soon, too.)
1117 * systemd-nspawn gained a new --network-macvlan= setting to
1118 set up a private macvlan interface for the
1119 container. Similarly, systemd-networkd gained a new
1120 Kind=macvlan setting in .netdev files.
1122 * systemd-networkd now supports configuring local addresses
1125 * A new tool systemd-network-wait-online has been added to
1126 synchronously wait for network connectivity using
1129 * The sd-bus.h bus API gained a new sd_bus_track object for
1130 tracking the life-cycle of bus peers. Note that sd-bus.h is
1131 still not a public API though (unless you specify
1132 --enable-kdbus on the configure command line, which however
1133 voids your warranty and you get no API stability guarantee).
1135 * The $XDG_RUNTIME_DIR runtime directories for each user are
1136 now individual tmpfs instances, which has the benefit of
1137 introducing separate pools for each user, with individual
1138 size limits, and thus making sure that unprivileged clients
1139 can no longer negatively impact the system or other users by
1140 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1141 RuntimeDirectorySize= has been introduced that allows
1142 controlling the default size limit for all users. It
1143 defaults to 10% of the available physical memory. This is no
1144 replacement for quotas on tmpfs though (which the kernel
1145 still does not support), as /dev/shm and /tmp are still
1146 shared resources used by both the system and unprivileged
1149 * logind will now automatically turn off automatic suspending
1150 on laptop lid close when more than one display is
1151 connected. This was previously expected to be implemented
1152 individually in desktop environments (such as GNOME),
1153 however has been added to logind now, in order to fix a
1154 boot-time race where a desktop environment might not have
1155 been started yet and thus not been able to take an inhibitor
1156 lock at the time where logind already suspends the system
1157 due to a closed lid.
1159 * logind will now wait at least 30s after each system
1160 suspend/resume cycle, and 3min after system boot before
1161 suspending the system due to a closed laptop lid. This
1162 should give USB docking stations and similar enough time to
1163 be probed and configured after system resume and boot in
1164 order to then act as suspend blocker.
1166 * systemd-run gained a new --property= setting which allows
1167 initialization of resource control properties (and others)
1168 for the created scope or service unit. Example: "systemd-run
1169 --property=BlockIOWeight=10 updatedb" may be used to run
1170 updatedb at a low block IO scheduling weight.
1172 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1173 now also work in --scope mode.
1175 * When systemd is compiled with kdbus support, basic support
1176 for enforced policies is now in place. (Note that enabling
1177 kdbus still voids your warranty and no API compatibility
1180 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1181 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1182 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1183 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1184 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1185 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1186 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1187 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1188 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1189 Zbigniew Jędrzejewski-Szmek
1191 -- Berlin, 2014-03-12
1195 * systemd will now relabel /dev after loading the SMACK policy
1196 according to SMACK rules.
1198 * A new unit file option AppArmorProfile= has been added to
1199 set the AppArmor profile for the processes of a unit.
1201 * A new condition check ConditionArchitecture= has been added
1202 to conditionalize units based on the system architecture, as
1203 reported by uname()'s "machine" field.
1205 * systemd-networkd now supports matching on the system
1206 virtualization, architecture, kernel command line, host name
1209 * logind is now a lot more aggressive when suspending the
1210 machine due to a closed laptop lid. Instead of acting only
1211 on the lid close action, it will continuously watch the lid
1212 status and act on it. This is useful for laptops where the
1213 power button is on the outside of the chassis so that it can
1214 be reached without opening the lid (such as the Lenovo
1215 Yoga). On those machines, logind will now immediately
1216 re-suspend the machine if the power button has been
1217 accidentally pressed while the laptop was suspended and in a
1218 backpack or similar.
1220 * logind will now watch SW_DOCK switches and inhibit reaction
1221 to the lid switch if it is pressed. This means that logind
1222 will not suspend the machine anymore if the lid is closed
1223 and the system is docked, if the laptop supports SW_DOCK
1224 notifications via the input layer. Note that ACPI docking
1225 stations do not generate this currently. Also note that this
1226 logic is usually not fully sufficient and Desktop
1227 Environments should take a lid switch inhibitor lock when an
1228 external display is connected, as systemd will not watch
1231 * nspawn will now make use of the devices cgroup controller by
1232 default, and only permit creation of and access to the usual
1233 API device nodes like /dev/null or /dev/random, as well as
1234 access to (but not creation of) the pty devices.
1236 * We will now ship a default .network file for
1237 systemd-networkd that automatically configures DHCP for
1238 network interfaces created by nspawn's --network-veth or
1239 --network-bridge= switches.
1241 * systemd will now understand the usual M, K, G, T suffixes
1242 according to SI conventions (i.e. to the base 1000) when
1243 referring to throughput and hardware metrics. It will stay
1244 with IEC conventions (i.e. to the base 1024) for software
1245 metrics, according to what is customary according to
1246 Wikipedia. We explicitly document which base applies for
1247 each configuration option.
1249 * The DeviceAllow= setting in unit files now supports a syntax
1250 to whitelist an entire group of devices node majors at once,
1251 based on the /proc/devices listing. For example, with the
1252 string "char-pts", it is now possible to whitelist all
1253 current and future pseudo-TTYs at once.
1255 * sd-event learned a new "post" event source. Event sources of
1256 this type are triggered by the dispatching of any event
1257 source of a type that is not "post". This is useful for
1258 implementing clean-up and check event sources that are
1259 triggered by other work being done in the program.
1261 * systemd-networkd is no longer statically enabled, but uses
1262 the usual [Install] sections so that it can be
1263 enabled/disabled using systemctl. It still is enabled by
1266 * When creating a veth interface pair with systemd-nspawn, the
1267 host side will now be prefixed with "vb-" if
1268 --network-bridge= is used, and with "ve-" if --network-veth
1269 is used. This way, it is easy to distinguish these cases on
1270 the host, for example to apply different configuration to
1271 them with systemd-networkd.
1273 * The compatibility libraries for libsystemd-journal.so,
1274 libsystem-id128.so, libsystemd-login.so and
1275 libsystemd-daemon.so do not make use of IFUNC
1276 anymore. Instead, we now build libsystemd.so multiple times
1277 under these alternative names. This means that the footprint
1278 is drastically increased, but given that these are
1279 transitional compatibility libraries, this should not matter
1280 much. This change has been made necessary to support the ARM
1281 platform for these compatibility libraries, as the ARM
1282 toolchain is not really at the same level as the toolchain
1283 for other architectures like x86 and does not support
1284 IFUNC. Please make sure to use --enable-compat-libs only
1285 during a transitional period!
1287 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1288 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1289 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1290 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1291 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1292 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1293 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1294 Zbigniew Jędrzejewski-Szmek
1296 -- Berlin, 2014-02-24
1300 * A new component "systemd-networkd" has been added that can
1301 be used to configure local network interfaces statically or
1302 via DHCP. It is capable of bringing up bridges, VLANs, and
1303 bonding. Currently, no hook-ups for interactive network
1304 configuration are provided. Use this for your initrd,
1305 container, embedded, or server setup if you need a simple,
1306 yet powerful, network configuration solution. This
1307 configuration subsystem is quite nifty, as it allows wildcard
1308 hotplug matching in interfaces. For example, with a single
1309 configuration snippet, you can configure that all Ethernet
1310 interfaces showing up are automatically added to a bridge,
1311 or similar. It supports link-sensing and more.
1313 * A new tool "systemd-socket-proxyd" has been added which can
1314 act as a bidirectional proxy for TCP sockets. This is
1315 useful for adding socket activation support to services that
1316 do not actually support socket activation, including virtual
1317 machines and the like.
1319 * Add a new tool to save/restore rfkill state on
1322 * Save/restore state of keyboard backlights in addition to
1323 display backlights on shutdown/boot.
1325 * udev learned a new SECLABEL{} construct to label device
1326 nodes with a specific security label when they appear. For
1327 now, only SECLABEL{selinux} is supported, but the syntax is
1328 prepared for additional security frameworks.
1330 * udev gained a new scheme to configure link-level attributes
1331 from files in /etc/systemd/network/*.link. These files can
1332 match against MAC address, device path, driver name and type,
1333 and will apply attributes like the naming policy, link speed,
1334 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1335 address assignment policy (randomized, ...).
1337 * The configuration of network interface naming rules for
1338 "permanent interface names" has changed: a new NamePolicy=
1339 setting in the [Link] section of .link files determines the
1340 priority of possible naming schemes (onboard, slot, mac,
1341 path). The default value of this setting is determined by
1342 /usr/lib/net/links/99-default.link. Old
1343 80-net-name-slot.rules udev configuration file has been
1344 removed, so local configuration overriding this file should
1345 be adapated to override 99-default.link instead.
1347 * When the User= switch is used in a unit file, also
1348 initialize $SHELL= based on the user database entry.
1350 * systemd no longer depends on libdbus. All communication is
1351 now done with sd-bus, systemd's low-level bus library
1354 * kdbus support has been added to PID 1 itself. When kdbus is
1355 enabled, this causes PID 1 to set up the system bus and
1356 enable support for a new ".busname" unit type that
1357 encapsulates bus name activation on kdbus. It works a little
1358 bit like ".socket" units, except for bus names. A new
1359 generator has been added that converts classic dbus1 service
1360 activation files automatically into native systemd .busname
1363 * sd-bus: add a light-weight vtable implementation that allows
1364 defining objects on the bus with a simple static const
1365 vtable array of its methods, signals and properties.
1367 * systemd will not generate or install static dbus
1368 introspection data anymore to /usr/share/dbus-1/interfaces,
1369 as the precise format of these files is unclear, and
1370 nothing makes use of it.
1372 * A proxy daemon is now provided to proxy clients connecting
1373 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1374 compatibility with classic D-Bus.
1376 * A bus driver implementation has been added that supports the
1377 classic D-Bus bus driver calls on kdbus, also for
1378 compatibility purposes.
1380 * A new API "sd-event.h" has been added that implements a
1381 minimal event loop API built around epoll. It provides a
1382 couple of features that direct epoll usage is lacking:
1383 prioritization of events, scales to large numbers of timer
1384 events, per-event timer slack (accuracy), system-wide
1385 coalescing of timer events, exit handlers, watchdog
1386 supervision support using systemd's sd_notify() API, child
1389 * A new API "sd-rntl.h" has been added that provides an API
1390 around the route netlink interface of the kernel, similar in
1391 style to "sd-bus.h".
1393 * A new API "sd-dhcp-client.h" has been added that provides a
1394 small DHCPv4 client-side implementation. This is used by
1397 * There is a new kernel command line option
1398 "systemd.restore_state=0|1". When set to "0", none of the
1399 systemd tools will restore saved runtime state to hardware
1400 devices. More specifically, the rfkill and backlight states
1403 * The FsckPassNo= compatibility option in mount/service units
1404 has been removed. The fstab generator will now add the
1405 necessary dependencies automatically, and does not require
1406 PID1's support for that anymore.
1408 * journalctl gained a new switch, --list-boots, that lists
1409 recent boots with their times and boot IDs.
1411 * The various tools like systemctl, loginctl, timedatectl,
1412 busctl, systemd-run, ... have gained a new switch "-M" to
1413 connect to a specific, local OS container (as direct
1414 connection, without requiring SSH). This works on any
1415 container that is registered with machined, such as those
1416 created by libvirt-lxc or nspawn.
1418 * systemd-run and systemd-analyze also gained support for "-H"
1419 to connect to remote hosts via SSH. This is particularly
1420 useful for systemd-run because it enables queuing of jobs
1421 onto remote systems.
1423 * machinectl gained a new command "login" to open a getty
1424 login in any local container. This works with any container
1425 that is registered with machined (such as those created by
1426 libvirt-lxc or nspawn), and which runs systemd inside.
1428 * machinectl gained a new "reboot" command that may be used to
1429 trigger a reboot on a specific container that is registered
1430 with machined. This works on any container that runs an init
1431 system of some kind.
1433 * systemctl gained a new "list-timers" command to print a nice
1434 listing of installed timer units with the times they elapse
1437 * Alternative reboot() parameters may now be specified on the
1438 "systemctl reboot" command line and are passed to the
1439 reboot() system call.
1441 * systemctl gained a new --job-mode= switch to configure the
1442 mode to queue a job with. This is a more generic version of
1443 --fail, --irreversible, and --ignore-dependencies, which are
1444 still available but not advertised anymore.
1446 * /etc/systemd/system.conf gained new settings to configure
1447 various default timeouts of units, as well as the default
1448 start limit interval and burst. These may still be overridden
1451 * PID1 will now export on the bus profile data of the security
1452 policy upload process (such as the SELinux policy upload to
1455 * journald: when forwarding logs to the console, include
1456 timestamps (following the setting in
1457 /sys/module/printk/parameters/time).
1459 * OnCalendar= in timer units now understands the special
1460 strings "yearly" and "annually". (Both are equivalent)
1462 * The accuracy of timer units is now configurable with the new
1463 AccuracySec= setting. It defaults to 1min.
1465 * A new dependency type JoinsNamespaceOf= has been added that
1466 allows running two services within the same /tmp and network
1467 namespace, if PrivateNetwork= or PrivateTmp= are used.
1469 * A new command "cat" has been added to systemctl. It outputs
1470 the original unit file of a unit, and concatenates the
1471 contents of additional "drop-in" unit file snippets, so that
1472 the full configuration is shown.
1474 * systemctl now supports globbing on the various "list-xyz"
1475 commands, like "list-units" or "list-sockets", as well as on
1476 those commands which take multiple unit names.
1478 * journalctl's --unit= switch gained support for globbing.
1480 * All systemd daemons now make use of the watchdog logic so
1481 that systemd automatically notices when they hang.
1483 * If the $container_ttys environment variable is set,
1484 getty-generator will automatically spawn a getty for each
1485 listed tty. This is useful for container managers to request
1486 login gettys to be spawned on as many ttys as needed.
1488 * %h, %s, %U specifier support is not available anymore when
1489 used in unit files for PID 1. This is because NSS calls are
1490 not safe from PID 1. They stay available for --user
1491 instances of systemd, and as special case for the root user.
1493 * loginctl gained a new "--no-legend" switch to turn off output
1496 * The "sd-login.h" API gained three new calls:
1497 sd_session_is_remote(), sd_session_get_remote_user(),
1498 sd_session_get_remote_host() to query information about
1501 * The udev hardware database now also carries vendor/product
1502 information of SDIO devices.
1504 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1505 determine whether watchdog notifications are requested by
1508 * Socket-activated per-connection services now include a
1509 short description of the connection parameters in the
1512 * tmpfiles gained a new "--boot" option. When this is not used,
1513 only lines where the command character is not suffixed with
1514 "!" are executed. When this option is specified, those
1515 options are executed too. This partitions tmpfiles
1516 directives into those that can be safely executed at any
1517 time, and those which should be run only at boot (for
1518 example, a line that creates /run/nologin).
1520 * A new API "sd-resolve.h" has been added which provides a simple
1521 asynchronous wrapper around glibc NSS host name resolution
1522 calls, such as getaddrinfo(). In contrast to glibc's
1523 getaddrinfo_a(), it does not use signals. In contrast to most
1524 other asynchronous name resolution libraries, this one does
1525 not reimplement DNS, but reuses NSS, so that alternate
1526 host name resolution systems continue to work, such as mDNS,
1527 LDAP, etc. This API is based on libasyncns, but it has been
1528 cleaned up for inclusion in systemd.
1530 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1531 "sd-daemon.h" are no longer found in individual libraries
1532 libsystemd-journal.so, libsystemd-login.so,
1533 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1534 merged them into a single library, libsystemd.so, which
1535 provides all symbols. The reason for this is cyclic
1536 dependencies, as these libraries tend to use each other's
1537 symbols. So far, we have managed to workaround that by linking
1538 a copy of a good part of our code into each of these
1539 libraries again and again, which, however, makes certain
1540 things hard to do, like sharing static variables. Also, it
1541 substantially increases footprint. With this change, there
1542 is only one library for the basic APIs systemd
1543 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1544 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1545 library as well, however are subject to the --enable-kdbus
1546 switch (see below). Note that "sd-dhcp-client.h" is not part
1547 of this library (this is because it only consumes, never
1548 provides, services of/to other APIs). To make the transition
1549 easy from the separate libraries to the unified one, we
1550 provide the --enable-compat-libs compile-time switch which
1551 will generate stub libraries that are compatible with the
1552 old ones but redirect all calls to the new one.
1554 * All of the kdbus logic and the new APIs "sd-bus.h",
1555 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1556 and "sd-utf8.h" are compile-time optional via the
1557 "--enable-kdbus" switch, and they are not compiled in by
1558 default. To make use of kdbus, you have to explicitly enable
1559 the switch. Note however, that neither the kernel nor the
1560 userspace API for all of this is considered stable yet. We
1561 want to maintain the freedom to still change the APIs for
1562 now. By specifying this build-time switch, you acknowledge
1563 that you are aware of the instability of the current
1566 * Also, note that while kdbus is pretty much complete,
1567 it lacks one thing: proper policy support. This means you
1568 can build a fully working system with all features; however,
1569 it will be highly insecure. Policy support will be added in
1570 one of the next releases, at the same time that we will
1571 declare the APIs stable.
1573 * When the kernel command-line argument "kdbus" is specified,
1574 systemd will automatically load the kdbus.ko kernel module. At
1575 this stage of development, it is only useful for testing kdbus
1576 and should not be used in production. Note: if "--enable-kdbus"
1577 is specified, and the kdbus.ko kernel module is available, and
1578 "kdbus" is added to the kernel command line, the entire system
1579 runs with kdbus instead of dbus-daemon, with the above mentioned
1580 problem of missing the system policy enforcement. Also a future
1581 version of kdbus.ko or a newer systemd will not be compatible with
1582 each other, and will unlikely be able to boot the machine if only
1583 one of them is updated.
1585 * systemctl gained a new "import-environment" command which
1586 uploads the caller's environment (or parts thereof) into the
1587 service manager so that it is inherited by services started
1588 by the manager. This is useful to upload variables like
1589 $DISPLAY into the user service manager.
1591 * A new PrivateDevices= switch has been added to service units
1592 which allows running a service with a namespaced /dev
1593 directory that does not contain any device nodes for
1594 physical devices. More specifically, it only includes devices
1595 such as /dev/null, /dev/urandom, and /dev/zero which are API
1598 * logind has been extended to support behaviour like VT
1599 switching on seats that do not support a VT. This makes
1600 multi-session available on seats that are not the first seat
1601 (seat0), and on systems where kernel support for VTs has
1602 been disabled at compile-time.
1604 * If a process holds a delay lock for system sleep or shutdown
1605 and fails to release it in time, we will now log its
1606 identity. This makes it easier to identify processes that
1607 cause slow suspends or power-offs.
1609 * When parsing /etc/crypttab, support for a new key-slot=
1610 option as supported by Debian is added. It allows indicating
1611 which LUKS slot to use on disk, speeding up key loading.
1613 * The sd_journald_sendv() API call has been checked and
1614 officially declared to be async-signal-safe so that it may
1615 be invoked from signal handlers for logging purposes.
1617 * Boot-time status output is now enabled automatically after a
1618 short timeout if boot does not progress, in order to give
1619 the user an indication what she or he is waiting for.
1621 * The boot-time output has been improved to show how much time
1622 remains until jobs expire.
1624 * The KillMode= switch in service units gained a new possible
1625 value "mixed". If set, and the unit is shut down, then the
1626 initial SIGTERM signal is sent only to the main daemon
1627 process, while the following SIGKILL signal is sent to
1628 all remaining processes of the service.
1630 * When a scope unit is registered, a new property "Controller"
1631 may be set. If set to a valid bus name, systemd will send a
1632 RequestStop() signal to this name when it would like to shut
1633 down the scope. This may be used to hook manager logic into
1634 the shutdown logic of scope units. Also, scope units may now
1635 be put in a special "abandoned" state, in which case the
1636 manager process which created them takes no further
1637 responsibilities for it.
1639 * When reading unit files, systemd will now verify
1640 the access mode of these files, and warn about certain
1641 suspicious combinations. This has been added to make it
1642 easier to track down packaging bugs where unit files are
1643 marked executable or world-writable.
1645 * systemd-nspawn gained a new "--setenv=" switch to set
1646 container-wide environment variables. The similar option in
1647 systemd-activate was renamed from "--environment=" to
1648 "--setenv=" for consistency.
1650 * systemd-nspawn has been updated to create a new kdbus domain
1651 for each container that is invoked, thus allowing each
1652 container to have its own set of system and user buses,
1653 independent of the host.
1655 * systemd-nspawn gained a new --drop-capability= switch to run
1656 the container with less capabilities than the default. Both
1657 --drop-capability= and --capability= now take the special
1658 string "all" for dropping or keeping all capabilities.
1660 * systemd-nspawn gained new switches for executing containers
1661 with specific SELinux labels set.
1663 * systemd-nspawn gained a new --quiet switch to not generate
1664 any additional output but the container's own console
1667 * systemd-nspawn gained a new --share-system switch to run a
1668 container without PID namespacing enabled.
1670 * systemd-nspawn gained a new --register= switch to control
1671 whether the container is registered with systemd-machined or
1672 not. This is useful for containers that do not run full
1673 OS images, but only specific apps.
1675 * systemd-nspawn gained a new --keep-unit which may be used
1676 when invoked as the only program from a service unit, and
1677 results in registration of the unit service itself in
1678 systemd-machined, instead of a newly opened scope unit.
1680 * systemd-nspawn gained a new --network-interface= switch for
1681 moving arbitrary interfaces to the container. The new
1682 --network-veth switch creates a virtual Ethernet connection
1683 between host and container. The new --network-bridge=
1684 switch then allows assigning the host side of this virtual
1685 Ethernet connection to a bridge device.
1687 * systemd-nspawn gained a new --personality= switch for
1688 setting the kernel personality for the container. This is
1689 useful when running a 32-bit container on a 64-bit host. A
1690 similar option Personality= is now also available for service
1693 * logind will now also track a "Desktop" identifier for each
1694 session which encodes the desktop environment of it. This is
1695 useful for desktop environments that want to identify
1696 multiple running sessions of itself easily.
1698 * A new SELinuxContext= setting for service units has been
1699 added that allows setting a specific SELinux execution
1700 context for a service.
1702 * Most systemd client tools will now honour $SYSTEMD_LESS for
1703 settings of the "less" pager. By default, these tools will
1704 override $LESS to allow certain operations to work, such as
1705 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1706 influence this logic.
1708 * systemd's "seccomp" hook-up has been changed to make use of
1709 the libseccomp library instead of using its own
1710 implementation. This has benefits for portability among
1713 * For usage together with SystemCallFilter=, a new
1714 SystemCallErrorNumber= setting has been introduced that
1715 allows configuration of a system error number to be returned
1716 on filtered system calls, instead of immediately killing the
1717 process. Also, SystemCallArchitectures= has been added to
1718 limit access to system calls of a particular architecture
1719 (in order to turn off support for unused secondary
1720 architectures). There is also a global
1721 SystemCallArchitectures= setting in system.conf now to turn
1722 off support for non-native system calls system-wide.
1724 * systemd requires a kernel with a working name_to_handle_at(),
1725 please see the kernel config requirements in the README file.
1727 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1728 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1729 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1730 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1731 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1732 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1733 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1734 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1735 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1736 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1737 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1738 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1739 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1740 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1741 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1742 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1743 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1744 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1745 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1746 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1747 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1748 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1749 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1750 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1752 -- Berlin, 2014-02-20
1756 * logind has gained support for facilitating privileged input
1757 and drm device access for unprivileged clients. This work is
1758 useful to allow Wayland display servers (and similar
1759 programs, such as kmscon) to run under the user's ID and
1760 access input and drm devices which are normally
1761 protected. When this is used (and the kernel is new enough)
1762 logind will "mute" IO on the file descriptors passed to
1763 Wayland as long as it is in the background and "unmute" it
1764 if it returns into the foreground. This allows secure
1765 session switching without allowing background sessions to
1766 eavesdrop on input and display data. This also introduces
1767 session switching support if VT support is turned off in the
1768 kernel, and on seats that are not seat0.
1770 * A new kernel command line option luks.options= is understood
1771 now which allows specifying LUKS options for usage for LUKS
1772 encrypted partitions specified with luks.uuid=.
1774 * tmpfiles.d(5) snippets may now use specifier expansion in
1775 path names. More specifically %m, %b, %H, %v, are now
1776 replaced by the local machine id, boot id, hostname, and
1777 kernel version number.
1779 * A new tmpfiles.d(5) command "m" has been introduced which
1780 may be used to change the owner/group/access mode of a file
1781 or directory if it exists, but do nothing if it does not.
1783 * This release removes high-level support for the
1784 MemorySoftLimit= cgroup setting. The underlying kernel
1785 cgroup attribute memory.soft_limit= is currently badly
1786 designed and likely to be removed from the kernel API in its
1787 current form, hence we should not expose it for now.
1789 * The memory.use_hierarchy cgroup attribute is now enabled for
1790 all cgroups systemd creates in the memory cgroup
1791 hierarchy. This option is likely to be come the built-in
1792 default in the kernel anyway, and the non-hierarchial mode
1793 never made much sense in the intrinsically hierarchial
1796 * A new field _SYSTEMD_SLICE= is logged along with all journal
1797 messages containing the slice a message was generated
1798 from. This is useful to allow easy per-customer filtering of
1799 logs among other things.
1801 * systemd-journald will no longer adjust the group of journal
1802 files it creates to the "systemd-journal" group. Instead we
1803 rely on the journal directory to be owned by the
1804 "systemd-journal" group, and its setgid bit set, so that the
1805 kernel file system layer will automatically enforce that
1806 journal files inherit this group assignment. The reason for
1807 this change is that we cannot allow NSS look-ups from
1808 journald which would be necessary to resolve
1809 "systemd-journal" to a numeric GID, because this might
1810 create deadlocks if NSS involves synchronous queries to
1811 other daemons (such as nscd, or sssd) which in turn are
1812 logging clients of journald and might block on it, which
1813 would then dead lock. A tmpfiles.d(5) snippet included in
1814 systemd will make sure the setgid bit and group are
1815 properly set on the journal directory if it exists on every
1816 boot. However, we recommend adjusting it manually after
1817 upgrades too (or from RPM scriptlets), so that the change is
1818 not delayed until next reboot.
1820 * Backlight and random seed files in /var/lib/ have moved into
1821 the /var/lib/systemd/ directory, in order to centralize all
1822 systemd generated files in one directory.
1824 * Boot time performance measurements (as displayed by
1825 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1826 performance information if that's available to determine how
1827 much time BIOS and boot loader initialization required. With
1828 a sufficiently new BIOS you hence no longer need to boot
1829 with Gummiboot to get access to such information.
1831 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1832 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1833 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1834 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1835 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1836 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1837 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1839 -- Berlin, 2013-10-02
1843 * The Restart= option for services now understands a new
1844 on-watchdog setting, which will restart the service
1845 automatically if the service stops sending out watchdog keep
1846 alive messages (as configured with WatchdogSec=).
1848 * The getty generator (which is responsible for bringing up a
1849 getty on configured serial consoles) will no longer only
1850 start a getty on the primary kernel console but on all
1851 others, too. This makes the order in which console= is
1852 specified on the kernel command line less important.
1854 * libsystemd-logind gained a new sd_session_get_vt() call to
1855 retrieve the VT number of a session.
1857 * If the option "tries=0" is set for an entry of /etc/crypttab
1858 its passphrase is queried indefinitely instead of any
1859 maximum number of tries.
1861 * If a service with a configure PID file terminates its PID
1862 file will now be removed automatically if it still exists
1863 afterwards. This should put an end to stale PID files.
1865 * systemd-run will now also take relative binary path names
1866 for execution and no longer insists on absolute paths.
1868 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1869 paths that are optionally prefixed with "-" to indicate that
1870 it should not be considered a failure if they do not exist.
1872 * journalctl -o (and similar commands) now understands a new
1873 output mode "short-precise", it is similar to "short" but
1874 shows timestamps with usec accuracy.
1876 * The option "discard" (as known from Debian) is now
1877 synonymous to "allow-discards" in /etc/crypttab. In fact,
1878 "discard" is preferred now (since it is easier to remember
1881 * Some licensing clean-ups were made, so that more code is now
1882 LGPL-2.1 licensed than before.
1884 * A minimal tool to save/restore the display backlight
1885 brightness across reboots has been added. It will store the
1886 backlight setting as late as possible at shutdown, and
1887 restore it as early as possible during reboot.
1889 * A logic to automatically discover and enable home and swap
1890 partitions on GPT disks has been added. With this in place
1891 /etc/fstab becomes optional for many setups as systemd can
1892 discover certain partitions located on the root disk
1893 automatically. Home partitions are recognized under their
1894 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1895 partitions are recognized under their GPT type ID
1896 0657fd6da4ab43c484e50933c84b4f4f.
1898 * systemd will no longer pass any environment from the kernel
1899 or initrd to system services. If you want to set an
1900 environment for all services, do so via the kernel command
1901 line systemd.setenv= assignment.
1903 * The systemd-sysctl tool no longer natively reads the file
1904 /etc/sysctl.conf. If desired, the file should be symlinked
1905 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1906 legacy support by a symlink rather than built-in code, it
1907 also makes the otherwise hidden order of application of the
1908 different files visible. (Note that this partly reverts to a
1909 pre-198 application order of sysctl knobs!)
1911 * The "systemctl set-log-level" and "systemctl dump" commands
1912 have been moved to systemd-analyze.
1914 * systemd-run learned the new --remain-after-exit switch,
1915 which causes the scope unit not to be cleaned up
1916 automatically after the process terminated.
1918 * tmpfiles learned a new --exclude-prefix= switch to exclude
1919 certain paths from operation.
1921 * journald will now automatically flush all messages to disk
1922 as soon as a message of the log priorities CRIT, ALERT or
1925 Contributions from: Andrew Cook, Brandon Philips, Christian
1926 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1927 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1928 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1929 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1930 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1931 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1932 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1933 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1934 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1935 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1936 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1937 William Giokas, Zbigniew Jędrzejewski-Szmek
1939 -- Berlin, 2013-09-13
1943 * The documentation has been updated to cover the various new
1944 concepts introduced with 205.
1946 * Unit files now understand the new %v specifier which
1947 resolves to the kernel version string as returned by "uname
1950 * systemctl now supports filtering the unit list output by
1951 load state, active state and sub state, using the new
1954 * "systemctl status" will now show the results of the
1955 condition checks (like ConditionPathExists= and similar) of
1956 the last start attempts of the unit. They are also logged to
1959 * "journalctl -b" may now be used to look for boot output of a
1960 specific boot. Try "journalctl -b -1" for the previous boot,
1961 but the syntax is substantially more powerful.
1963 * "journalctl --show-cursor" has been added which prints the
1964 cursor string the last shown log line. This may then be used
1965 with the new "journalctl --after-cursor=" switch to continue
1966 browsing logs from that point on.
1968 * "journalctl --force" may now be used to force regeneration
1971 * Creation of "dead" device nodes has been moved from udev
1972 into kmod and tmpfiles. Previously, udev would read the kmod
1973 databases to pre-generate dead device nodes based on meta
1974 information contained in kernel modules, so that these would
1975 be auto-loaded on access rather then at boot. As this
1976 does not really have much to do with the exposing actual
1977 kernel devices to userspace this has always been slightly
1978 alien in the udev codebase. Following the new scheme kmod
1979 will now generate a runtime snippet for tmpfiles from the
1980 module meta information and it now is tmpfiles' job to the
1981 create the nodes. This also allows overriding access and
1982 other parameters for the nodes using the usual tmpfiles
1983 facilities. As side effect this allows us to remove the
1984 CAP_SYS_MKNOD capability bit from udevd entirely.
1986 * logind's device ACLs may now be applied to these "dead"
1987 devices nodes too, thus finally allowing managed access to
1988 devices such as /dev/snd/sequencer whithout loading the
1989 backing module right-away.
1991 * A new RPM macro has been added that may be used to apply
1992 tmpfiles configuration during package installation.
1994 * systemd-detect-virt and ConditionVirtualization= now can
1995 detect User-Mode-Linux machines (UML).
1997 * journald will now implicitly log the effective capabilities
1998 set of processes in the message metadata.
2000 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2002 * The initrd interface has been simplified (more specifically,
2003 support for passing performance data via environment
2004 variables and fsck results via files in /run has been
2005 removed). These features were non-essential, and are
2006 nowadays available in a much nicer way by having systemd in
2007 the initrd serialize its state and have the hosts systemd
2008 deserialize it again.
2010 * The udev "keymap" data files and tools to apply keyboard
2011 specific mappings of scan to key codes, and force-release
2012 scan code lists have been entirely replaced by a udev
2013 "keyboard" builtin and a hwdb data file.
2015 * systemd will now honour the kernel's "quiet" command line
2016 argument also during late shutdown, resulting in a
2017 completely silent shutdown when used.
2019 * There's now an option to control the SO_REUSEPORT socket
2020 option in .socket units.
2022 * Instance units will now automatically get a per-template
2023 subslice of system.slice unless something else is explicitly
2024 configured. For example, instances of sshd@.service will now
2025 implicitly be placed in system-sshd.slice rather than
2026 system.slice as before.
2028 * Test coverage support may now be enabled at build time.
2030 Contributions from: Dave Reisner, Frederic Crozat, Harald
2031 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2032 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2033 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2034 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2035 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2036 Giokas, Zbigniew Jędrzejewski-Szmek
2038 -- Berlin, 2013-07-23
2042 * Two new unit types have been introduced:
2044 Scope units are very similar to service units, however, are
2045 created out of pre-existing processes -- instead of PID 1
2046 forking off the processes. By using scope units it is
2047 possible for system services and applications to group their
2048 own child processes (worker processes) in a powerful way
2049 which then maybe used to organize them, or kill them
2050 together, or apply resource limits on them.
2052 Slice units may be used to partition system resources in an
2053 hierarchial fashion and then assign other units to them. By
2054 default there are now three slices: system.slice (for all
2055 system services), user.slice (for all user sessions),
2056 machine.slice (for VMs and containers).
2058 Slices and scopes have been introduced primarily in
2059 context of the work to move cgroup handling to a
2060 single-writer scheme, where only PID 1
2061 creates/removes/manages cgroups.
2063 * There's a new concept of "transient" units. In contrast to
2064 normal units these units are created via an API at runtime,
2065 not from configuration from disk. More specifically this
2066 means it is now possible to run arbitrary programs as
2067 independent services, with all execution parameters passed
2068 in via bus APIs rather than read from disk. Transient units
2069 make systemd substantially more dynamic then it ever was,
2070 and useful as a general batch manager.
2072 * logind has been updated to make use of scope and slice units
2073 for managing user sessions. As a user logs in he will get
2074 his own private slice unit, to which all sessions are added
2075 as scope units. We also added support for automatically
2076 adding an instance of user@.service for the user into the
2077 slice. Effectively logind will no longer create cgroup
2078 hierarchies on its own now, it will defer entirely to PID 1
2079 for this by means of scope, service and slice units. Since
2080 user sessions this way become entities managed by PID 1
2081 the output of "systemctl" is now a lot more comprehensive.
2083 * A new mini-daemon "systemd-machined" has been added which
2084 may be used by virtualization managers to register local
2085 VMs/containers. nspawn has been updated accordingly, and
2086 libvirt will be updated shortly. machined will collect a bit
2087 of meta information about the VMs/containers, and assign
2088 them their own scope unit (see above). The collected
2089 meta-data is then made available via the "machinectl" tool,
2090 and exposed in "ps" and similar tools. machined/machinectl
2091 is compile-time optional.
2093 * As discussed earlier, the low-level cgroup configuration
2094 options ControlGroup=, ControlGroupModify=,
2095 ControlGroupPersistent=, ControlGroupAttribute= have been
2096 removed. Please use high-level attribute settings instead as
2097 well as slice units.
2099 * A new bus call SetUnitProperties() has been added to alter
2100 various runtime parameters of a unit. This is primarily
2101 useful to alter cgroup parameters dynamically in a nice way,
2102 but will be extended later on to make more properties
2103 modifiable at runtime. systemctl gained a new set-properties
2104 command that wraps this call.
2106 * A new tool "systemd-run" has been added which can be used to
2107 run arbitrary command lines as transient services or scopes,
2108 while configuring a number of settings via the command
2109 line. This tool is currently very basic, however already
2110 very useful. We plan to extend this tool to even allow
2111 queuing of execution jobs with time triggers from the
2112 command line, similar in fashion to "at".
2114 * nspawn will now inform the user explicitly that kernels with
2115 audit enabled break containers, and suggest the user to turn
2118 * Support for detecting the IMA and AppArmor security
2119 frameworks with ConditionSecurity= has been added.
2121 * journalctl gained a new "-k" switch for showing only kernel
2122 messages, mimicking dmesg output; in addition to "--user"
2123 and "--system" switches for showing only user's own logs
2126 * systemd-delta can now show information about drop-in
2127 snippets extending unit files.
2129 * libsystemd-bus has been substantially updated but is still
2130 not available as public API.
2132 * systemd will now look for the "debug" argument on the kernel
2133 command line and enable debug logging, similar to what
2134 "systemd.log_level=debug" already did before.
2136 * "systemctl set-default", "systemctl get-default" has been
2137 added to configure the default.target symlink, which
2138 controls what to boot into by default.
2140 * "systemctl set-log-level" has been added as a convenient
2141 way to raise and lower systemd logging threshold.
2143 * "systemd-analyze plot" will now show the time the various
2144 generators needed for execution, as well as information
2145 about the unit file loading.
2147 * libsystemd-journal gained a new sd_journal_open_files() call
2148 for opening specific journal files. journactl also gained a
2149 new switch to expose this new functionality. Previously we
2150 only supported opening all files from a directory, or all
2151 files from the system, as opening individual files only is
2152 racy due to journal file rotation.
2154 * systemd gained the new DefaultEnvironment= setting in
2155 /etc/systemd/system.conf to set environment variables for
2158 * If a privileged process logs a journal message with the
2159 OBJECT_PID= field set, then journald will automatically
2160 augment this with additional OBJECT_UID=, OBJECT_GID=,
2161 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2162 system services want to log events about specific client
2163 processes. journactl/systemctl has been updated to make use
2164 of this information if all log messages regarding a specific
2167 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2168 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2169 Reisner, David Coppa, David King, David Strauss, Eelco
2170 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2171 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2172 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2173 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2174 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2175 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2176 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2177 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2178 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2179 Łukasz Stelmach, 장동준
2183 * The Python bindings gained some minimal support for the APIs
2184 exposed by libsystemd-logind.
2186 * ConditionSecurity= gained support for detecting SMACK. Since
2187 this condition already supports SELinux and AppArmor we only
2188 miss IMA for this. Patches welcome!
2190 Contributions from: Karol Lewandowski, Lennart Poettering,
2191 Zbigniew Jędrzejewski-Szmek
2195 * systemd-nspawn will now create /etc/resolv.conf if
2196 necessary, before bind-mounting the host's file onto it.
2198 * systemd-nspawn will now store meta information about a
2199 container on the container's cgroup as extended attribute
2200 fields, including the root directory.
2202 * The cgroup hierarchy has been reworked in many ways. All
2203 objects any of the components systemd creates in the cgroup
2204 tree are now suffixed. More specifically, user sessions are
2205 now placed in cgroups suffixed with ".session", users in
2206 cgroups suffixed with ".user", and nspawn containers in
2207 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2208 names are now escaped in a simple scheme to avoid collision
2209 of userspace object names with kernel filenames. This work
2210 is preparation for making these objects relocatable in the
2211 cgroup tree, in order to allow easy resource partitioning of
2212 these objects without causing naming conflicts.
2214 * systemctl list-dependencies gained the new switches
2215 --plain, --reverse, --after and --before.
2217 * systemd-inhibit now shows the process name of processes that
2218 have taken an inhibitor lock.
2220 * nss-myhostname will now also resolve "localhost"
2221 implicitly. This makes /etc/hosts an optional file and
2222 nicely handles that on IPv6 ::1 maps to both "localhost" and
2225 * libsystemd-logind.so gained a new call
2226 sd_get_machine_names() to enumerate running containers and
2227 VMs (currently only supported by very new libvirt and
2228 nspawn). sd_login_monitor can now be used to watch
2229 VMs/containers coming and going.
2231 * .include is not allowed recursively anymore, and only in
2232 unit files. Usually it is better to use drop-in snippets in
2233 .d/*.conf anyway, as introduced with systemd 198.
2235 * systemd-analyze gained a new "critical-chain" command that
2236 determines the slowest chain of units run during system
2237 boot-up. It is very useful for tracking down where
2238 optimizing boot time is the most beneficial.
2240 * systemd will no longer allow manipulating service paths in
2241 the name=systemd:/system cgroup tree using ControlGroup= in
2242 units. (But is still fine with it in all other dirs.)
2244 * There's a new systemd-nspawn@.service service file that may
2245 be used to easily run nspawn containers as system
2246 services. With the container's root directory in
2247 /var/lib/container/foobar it is now sufficient to run
2248 "systemctl start systemd-nspawn@foobar.service" to boot it.
2250 * systemd-cgls gained a new parameter "--machine" to list only
2251 the processes within a certain container.
2253 * ConditionSecurity= now can check for "apparmor". We still
2254 are lacking checks for SMACK and IMA for this condition
2255 check though. Patches welcome!
2257 * A new configuration file /etc/systemd/sleep.conf has been
2258 added that may be used to configure which kernel operation
2259 systemd is supposed to execute when "suspend", "hibernate"
2260 or "hybrid-sleep" is requested. This makes the new kernel
2261 "freeze" state accessible to the user.
2263 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2264 the passed argument if applicable.
2266 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2267 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2268 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2269 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2270 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2271 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2272 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2277 * The output of 'systemctl list-jobs' got some polishing. The
2278 '--type=' argument may now be passed more than once. A new
2279 command 'systemctl list-sockets' has been added which shows
2280 a list of kernel sockets systemd is listening on with the
2281 socket units they belong to, plus the units these socket
2284 * The experimental libsystemd-bus library got substantial
2285 updates to work in conjunction with the (also experimental)
2286 kdbus kernel project. It works well enough to exchange
2287 messages with some sophistication. Note that kdbus is not
2288 ready yet, and the library is mostly an elaborate test case
2289 for now, and not installable.
2291 * systemd gained a new unit 'systemd-static-nodes.service'
2292 that generates static device nodes earlier during boot, and
2293 can run in conjunction with udev.
2295 * libsystemd-login gained a new call sd_pid_get_user_unit()
2296 to retrieve the user systemd unit a process is running
2297 in. This is useful for systems where systemd is used as
2300 * systemd-nspawn now places all containers in the new /machine
2301 top-level cgroup directory in the name=systemd
2302 hierarchy. libvirt will soon do the same, so that we get a
2303 uniform separation of /system, /user and /machine for system
2304 services, user processes and containers/virtual
2305 machines. This new cgroup hierarchy is also useful to stick
2306 stable names to specific container instances, which can be
2307 recognized later this way (this name may be controlled
2308 via systemd-nspawn's new -M switch). libsystemd-login also
2309 gained a new call sd_pid_get_machine_name() to retrieve the
2310 name of the container/VM a specific process belongs to.
2312 * bootchart can now store its data in the journal.
2314 * libsystemd-journal gained a new call
2315 sd_journal_add_conjunction() for AND expressions to the
2316 matching logic. This can be used to express more complex
2317 logical expressions.
2319 * journactl can now take multiple --unit= and --user-unit=
2322 * The cryptsetup logic now understands the "luks.key=" kernel
2323 command line switch for specifying a file to read the
2324 decryption key from. Also, if a configured key file is not
2325 found the tool will now automatically fall back to prompting
2328 * Python systemd.journal module was updated to wrap recently
2329 added functions from libsystemd-journal. The interface was
2330 changed to bring the low level interface in s.j._Reader
2331 closer to the C API, and the high level interface in
2332 s.j.Reader was updated to wrap and convert all data about
2335 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2336 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2337 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2338 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2339 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2340 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2344 * journalctl --update-catalog now understands a new --root=
2345 option to operate on catalogs found in a different root
2348 * During shutdown after systemd has terminated all running
2349 services a final killing loop kills all remaining left-over
2350 processes. We will now print the name of these processes
2351 when we send SIGKILL to them, since this usually indicates a
2354 * If /etc/crypttab refers to password files stored on
2355 configured mount points automatic dependencies will now be
2356 generated to ensure the specific mount is established first
2357 before the key file is attempted to be read.
2359 * 'systemctl status' will now show information about the
2360 network sockets a socket unit is listening on.
2362 * 'systemctl status' will also shown information about any
2363 drop-in configuration file for units. (Drop-In configuration
2364 files in this context are files such as
2365 /etc/systemd/systemd/foobar.service.d/*.conf)
2367 * systemd-cgtop now optionally shows summed up CPU times of
2368 cgroups. Press '%' while running cgtop to switch between
2369 percentage and absolute mode. This is useful to determine
2370 which cgroups use up the most CPU time over the entire
2371 runtime of the system. systemd-cgtop has also been updated
2372 to be 'pipeable' for processing with further shell tools.
2374 * 'hostnamectl set-hostname' will now allow setting of FQDN
2377 * The formatting and parsing of time span values has been
2378 changed. The parser now understands fractional expressions
2379 such as "5.5h". The formatter will now output fractional
2380 expressions for all time spans under 1min, i.e. "5.123456s"
2381 rather than "5s 123ms 456us". For time spans under 1s
2382 millisecond values are shown, for those under 1ms
2383 microsecond values are shown. This should greatly improve
2384 all time-related output of systemd.
2386 * libsystemd-login and libsystemd-journal gained new
2387 functions for querying the poll() events mask and poll()
2388 timeout value for integration into arbitrary event
2391 * localectl gained the ability to list available X11 keymaps
2392 (models, layouts, variants, options).
2394 * 'systemd-analyze dot' gained the ability to filter for
2395 specific units via shell-style globs, to create smaller,
2396 more useful graphs. I.e. it is now possible to create simple
2397 graphs of all the dependencies between only target units, or
2398 of all units that Avahi has dependencies with.
2400 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2401 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2402 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2403 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2404 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2405 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2406 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2410 * The boot-time readahead implementation for rotating media
2411 will now read the read-ahead data in multiple passes which
2412 consist of all read requests made in equidistant time
2413 intervals. This means instead of strictly reading read-ahead
2414 data in its physical order on disk we now try to find a
2415 middle ground between physical and access time order.
2417 * /etc/os-release files gained a new BUILD_ID= field for usage
2418 on operating systems that provide continuous builds of OS
2421 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2422 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2423 William Douglas, Zbigniew Jędrzejewski-Szmek
2427 * systemd-python gained an API exposing libsystemd-daemon.
2429 * The SMACK setup logic gained support for uploading CIPSO
2432 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2433 ReadOnlyDirectories= and InaccessibleDirectories= has
2434 changed. The private /tmp and /var/tmp directories are now
2435 shared by all processes of a service (which means
2436 ExecStartPre= may now leave data in /tmp that ExecStart= of
2437 the same service can still access). When a service is
2438 stopped its temporary directories are immediately deleted
2439 (normal clean-up with tmpfiles is still done in addition to
2442 * By default, systemd will now set a couple of sysctl
2443 variables in the kernel: the safe sysrq options are turned
2444 on, IP route verification is turned on, and source routing
2445 disabled. The recently added hardlink and softlink
2446 protection of the kernel is turned on. These settings should
2447 be reasonably safe, and good defaults for all new systems.
2449 * The predictable network naming logic may now be turned off
2450 with a new kernel command line switch: net.ifnames=0.
2452 * A new libsystemd-bus module has been added that implements a
2453 pretty complete D-Bus client library. For details see:
2455 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2457 * journald will now explicitly flush the journal files to disk
2458 at the latest 5min after each write. The file will then also
2459 be marked offline until the next write. This should increase
2460 reliability in case of a crash. The synchronization delay
2461 can be configured via SyncIntervalSec= in journald.conf.
2463 * There's a new remote-fs-setup.target unit that can be used
2464 to pull in specific services when at least one remote file
2465 system is to be mounted.
2467 * There are new targets timers.target and paths.target as
2468 canonical targets to pull user timer and path units in
2469 from. This complements sockets.target with a similar
2470 purpose for socket units.
2472 * libudev gained a new call udev_device_set_attribute_value()
2473 to set sysfs attributes of a device.
2475 * The udev daemon now sets the default number of worker
2476 processes executed in parallel based on the number of available
2477 CPUs instead of the amount of available RAM. This is supposed
2478 to provide a more reliable default and limit a too aggressive
2479 paralellism for setups with 1000s of devices connected.
2481 Contributions from: Auke Kok, Colin Walters, Cristian
2482 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2483 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2484 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2485 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2486 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2487 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2488 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2489 Zbigniew Jędrzejewski-Szmek
2493 * Configuration of unit files may now be extended via drop-in
2494 files without having to edit/override the unit files
2495 themselves. More specifically, if the administrator wants to
2496 change one value for a service file foobar.service he can
2497 now do so by dropping in a configuration snippet into
2498 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2499 will load all these snippets and apply them on top of the
2500 main unit configuration file, possibly extending or
2501 overriding its settings. Using these drop-in snippets is
2502 generally nicer than the two earlier options for changing
2503 unit files locally: copying the files from
2504 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2505 them there; or creating a new file in /etc/systemd/system/
2506 that incorporates the original one via ".include". Drop-in
2507 snippets into these .d/ directories can be placed in any
2508 directory systemd looks for units in, and the usual
2509 overriding semantics between /usr/lib, /etc and /run apply
2512 * Most unit file settings which take lists of items can now be
2513 reset by assigning the empty string to them. For example,
2514 normally, settings such as Environment=FOO=BAR append a new
2515 environment variable assignment to the environment block,
2516 each time they are used. By assigning Environment= the empty
2517 string the environment block can be reset to empty. This is
2518 particularly useful with the .d/*.conf drop-in snippets
2519 mentioned above, since this adds the ability to reset list
2520 settings from vendor unit files via these drop-ins.
2522 * systemctl gained a new "list-dependencies" command for
2523 listing the dependencies of a unit recursively.
2525 * Inhibitors are now honored and listed by "systemctl
2526 suspend", "systemctl poweroff" (and similar) too, not only
2527 GNOME. These commands will also list active sessions by
2530 * Resource limits (as exposed by the various control group
2531 controllers) can now be controlled dynamically at runtime
2532 for all units. More specifically, you can now use a command
2533 like "systemctl set-cgroup-attr foobar.service cpu.shares
2534 2000" to alter the CPU shares a specific service gets. These
2535 settings are stored persistently on disk, and thus allow the
2536 administrator to easily adjust the resource usage of
2537 services with a few simple commands. This dynamic resource
2538 management logic is also available to other programs via the
2539 bus. Almost any kernel cgroup attribute and controller is
2542 * systemd-vconsole-setup will now copy all font settings to
2543 all allocated VTs, where it previously applied them only to
2546 * libsystemd-login gained the new sd_session_get_tty() API
2549 * This release drops support for a few legacy or
2550 distribution-specific LSB facility names when parsing init
2551 scripts: $x-display-manager, $mail-transfer-agent,
2552 $mail-transport-agent, $mail-transfer-agent, $smtp,
2553 $null. Also, the mail-transfer-agent.target unit backing
2554 this has been removed. Distributions which want to retain
2555 compatibility with this should carry the burden for
2556 supporting this themselves and patch support for these back
2557 in, if they really need to. Also, the facilities $syslog and
2558 $local_fs are now ignored, since systemd does not support
2559 early-boot LSB init scripts anymore, and these facilities
2560 are implied anyway for normal services. syslog.target has
2563 * There are new bus calls on PID1's Manager object for
2564 cancelling jobs, and removing snapshot units. Previously,
2565 both calls were only available on the Job and Snapshot
2568 * systemd-journal-gatewayd gained SSL support.
2570 * The various "environment" files, such as /etc/locale.conf
2571 now support continuation lines with a backslash ("\") as
2572 last character in the line, similarly in style (but different)
2573 to how this is supported in shells.
2575 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2576 now implicitly appended to every log entry logged. systemctl
2577 has been updated to filter by this field when operating on a
2578 user systemd instance.
2580 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2581 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2582 the container. This makes it easier to boot unmodified
2583 Fedora systems in a container, which however still requires
2584 audit=0 to be passed on the kernel command line. Auditing in
2585 kernel and userspace is unfortunately still too broken in
2586 context of containers, hence we recommend compiling it out
2587 of the kernel or using audit=0. Hopefully this will be fixed
2588 one day for good in the kernel.
2590 * nspawn gained the new --bind= and --bind-ro= parameters to
2591 bind mount specific directories from the host into the
2594 * nspawn will now mount its own devpts file system instance
2595 into the container, in order not to leak pty devices from
2596 the host into the container.
2598 * systemd will now read the firmware boot time performance
2599 information from the EFI variables, if the used boot loader
2600 supports this, and takes it into account for boot performance
2601 analysis via "systemd-analyze". This is currently supported
2602 only in conjunction with Gummiboot, but could be supported
2603 by other boot loaders too. For details see:
2605 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2607 * A new generator has been added that automatically mounts the
2608 EFI System Partition (ESP) to /boot, if that directory
2609 exists, is empty, and no other file system has been
2610 configured to be mounted there.
2612 * logind will now send out PrepareForSleep(false) out
2613 unconditionally, after coming back from suspend. This may be
2614 used by applications as asynchronous notification for
2615 system resume events.
2617 * "systemctl unlock-sessions" has been added, that allows
2618 unlocking the screens of all user sessions at once, similar
2619 to how "systemctl lock-sessions" already locked all users
2620 sessions. This is backed by a new D-Bus call UnlockSessions().
2622 * "loginctl seat-status" will now show the master device of a
2623 seat. (i.e. the device of a seat that needs to be around for
2624 the seat to be considered available, usually the graphics
2627 * tmpfiles gained a new "X" line type, that allows
2628 configuration of files and directories (with wildcards) that
2629 shall be excluded from automatic cleanup ("aging").
2631 * udev default rules set the device node permissions now only
2632 at "add" events, and do not change them any longer with a
2633 later "change" event.
2635 * The log messages for lid events and power/sleep keypresses
2636 now carry a message ID.
2638 * We now have a substantially larger unit test suite, but this
2639 continues to be work in progress.
2641 * udevadm hwdb gained a new --root= parameter to change the
2642 root directory to operate relative to.
2644 * logind will now issue a background sync() request to the kernel
2645 early at shutdown, so that dirty buffers are flushed to disk early
2646 instead of at the last moment, in order to optimize shutdown
2649 * A new bootctl tool has been added that is an interface for
2650 certain boot loader operations. This is currently a preview
2651 and is likely to be extended into a small mechanism daemon
2652 like timedated, localed, hostnamed, and can be used by
2653 graphical UIs to enumerate available boot options, and
2654 request boot into firmware operations.
2656 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2657 the rest of the package. It also has been updated to work
2658 correctly in initrds.
2660 * Policykit previously has been runtime optional, and is now
2661 also compile time optional via a configure switch.
2663 * systemd-analyze has been reimplemented in C. Also "systemctl
2664 dot" has moved into systemd-analyze.
2666 * "systemctl status" with no further parameters will now print
2667 the status of all active or failed units.
2669 * Operations such as "systemctl start" can now be executed
2670 with a new mode "--irreversible" which may be used to queue
2671 operations that cannot accidentally be reversed by a later
2672 job queuing. This is by default used to make shutdown
2673 requests more robust.
2675 * The Python API of systemd now gained a new module for
2676 reading journal files.
2678 * A new tool kernel-install has been added that can install
2679 kernel images according to the Boot Loader Specification:
2681 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2683 * Boot time console output has been improved to provide
2684 animated boot time output for hanging jobs.
2686 * A new tool systemd-activate has been added which can be used
2687 to test socket activation with, directly from the command
2688 line. This should make it much easier to test and debug
2689 socket activation in daemons.
2691 * journalctl gained a new "--reverse" (or -r) option to show
2692 journal output in reverse order (i.e. newest line first).
2694 * journalctl gained a new "--pager-end" (or -e) option to jump
2695 to immediately jump to the end of the journal in the
2696 pager. This is only supported in conjunction with "less".
2698 * journalctl gained a new "--user-unit=" option, that works
2699 similarly to "--unit=" but filters for user units rather than
2702 * A number of unit files to ease adoption of systemd in
2703 initrds has been added. This moves some minimal logic from
2704 the various initrd implementations into systemd proper.
2706 * The journal files are now owned by a new group
2707 "systemd-journal", which exists specifically to allow access
2708 to the journal, and nothing else. Previously, we used the
2709 "adm" group for that, which however possibly covers more
2710 than just journal/log file access. This new group is now
2711 already used by systemd-journal-gatewayd to ensure this
2712 daemon gets access to the journal files and as little else
2713 as possible. Note that "make install" will also set FS ACLs
2714 up for /var/log/journal to give "adm" and "wheel" read
2715 access to it, in addition to "systemd-journal" which owns
2716 the journal files. We recommend that packaging scripts also
2717 add read access to "adm" + "wheel" to /var/log/journal, and
2718 all existing/future journal files. To normal users and
2719 administrators little changes, however packagers need to
2720 ensure to create the "systemd-journal" system group at
2721 package installation time.
2723 * The systemd-journal-gatewayd now runs as unprivileged user
2724 systemd-journal-gateway:systemd-journal-gateway. Packaging
2725 scripts need to create these system user/group at
2728 * timedated now exposes a new boolean property CanNTP that
2729 indicates whether a local NTP service is available or not.
2731 * systemd-detect-virt will now also detect xen PVs
2733 * The pstore file system is now mounted by default, if it is
2736 * In addition to the SELinux and IMA policies we will now also
2737 load SMACK policies at early boot.
2739 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2740 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2741 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2742 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2743 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2744 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2745 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2746 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2747 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2748 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2749 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2750 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2751 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2752 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2756 * Timer units now support calendar time events in addition to
2757 monotonic time events. That means you can now trigger a unit
2758 based on a calendar time specification such as "Thu,Fri
2759 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2760 or fifth day of any month of the year 2013, given that it is
2761 a thursday or friday. This brings timer event support
2762 considerably closer to cron's capabilities. For details on
2763 the supported calendar time specification language see
2766 * udev now supports a number of different naming policies for
2767 network interfaces for predictable names, and a combination
2768 of these policies is now the default. Please see this wiki
2769 document for details:
2771 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2773 * Auke Kok's bootchart implementation has been added to the
2774 systemd tree. It is an optional component that can graph the
2775 boot in quite some detail. It is one of the best bootchart
2776 implementations around and minimal in its code and
2779 * nss-myhostname has been integrated into the systemd source
2780 tree. nss-myhostname guarantees that the local hostname
2781 always stays resolvable via NSS. It has been a weak
2782 requirement of systemd-hostnamed since a long time, and
2783 since its code is actually trivial we decided to just
2784 include it in systemd's source tree. It can be turned off
2785 with a configure switch.
2787 * The read-ahead logic is now capable of properly detecting
2788 whether a btrfs file system is on SSD or rotating media, in
2789 order to optimize the read-ahead scheme. Previously, it was
2790 only capable of detecting this on traditional file systems
2793 * In udev, additional device properties are now read from the
2794 IAB in addition to the OUI database. Also, Bluetooth company
2795 identities are attached to the devices as well.
2797 * In service files %U may be used as specifier that is
2798 replaced by the configured user name of the service.
2800 * nspawn may now be invoked without a controlling TTY. This
2801 makes it suitable for invocation as its own service. This
2802 may be used to set up a simple containerized server system
2803 using only core OS tools.
2805 * systemd and nspawn can now accept socket file descriptors
2806 when they are started for socket activation. This enables
2807 implementation of socket activated nspawn
2808 containers. i.e. think about autospawning an entire OS image
2809 when the first SSH or HTTP connection is received. We expect
2810 that similar functionality will also be added to libvirt-lxc
2813 * journalctl will now suppress ANSI color codes when
2814 presenting log data.
2816 * systemctl will no longer show control group information for
2817 a unit if a the control group is empty anyway.
2819 * logind can now automatically suspend/hibernate/shutdown the
2822 * /etc/machine-info and hostnamed now also expose the chassis
2823 type of the system. This can be used to determine whether
2824 the local system is a laptop, desktop, handset or
2825 tablet. This information may either be configured by the
2826 user/vendor or is automatically determined from ACPI and DMI
2827 information if possible.
2829 * A number of PolicyKit actions are now bound together with
2830 "imply" rules. This should simplify creating UIs because
2831 many actions will now authenticate similar ones as well.
2833 * Unit files learnt a new condition ConditionACPower= which
2834 may be used to conditionalize a unit depending on whether an
2835 AC power source is connected or not, of whether the system
2836 is running on battery power.
2838 * systemctl gained a new "is-failed" verb that may be used in
2839 shell scripts and suchlike to check whether a specific unit
2840 is in the "failed" state.
2842 * The EnvironmentFile= setting in unit files now supports file
2843 globbing, and can hence be used to easily read a number of
2844 environment files at once.
2846 * systemd will no longer detect and recognize specific
2847 distributions. All distribution-specific #ifdeffery has been
2848 removed, systemd is now fully generic and
2849 distribution-agnostic. Effectively, not too much is lost as
2850 a lot of the code is still accessible via explicit configure
2851 switches. However, support for some distribution specific
2852 legacy configuration file formats has been dropped. We
2853 recommend distributions to simply adopt the configuration
2854 files everybody else uses now and convert the old
2855 configuration from packaging scripts. Most distributions
2856 already did that. If that's not possible or desirable,
2857 distributions are welcome to forward port the specific
2858 pieces of code locally from the git history.
2860 * When logging a message about a unit systemd will now always
2861 log the unit name in the message meta data.
2863 * localectl will now also discover system locale data that is
2864 not stored in locale archives, but directly unpacked.
2866 * logind will no longer unconditionally use framebuffer
2867 devices as seat masters, i.e. as devices that are required
2868 to be existing before a seat is considered preset. Instead,
2869 it will now look for all devices that are tagged as
2870 "seat-master" in udev. By default framebuffer devices will
2871 be marked as such, but depending on local systems other
2872 devices might be marked as well. This may be used to
2873 integrate graphics cards using closed source drivers (such
2874 as NVidia ones) more nicely into logind. Note however, that
2875 we recommend using the open source NVidia drivers instead,
2876 and no udev rules for the closed-source drivers will be
2877 shipped from us upstream.
2879 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2880 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2881 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2882 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2883 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2884 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2885 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2886 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2887 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2888 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2889 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2890 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2895 * udev gained support for loading additional device properties
2896 from an indexed database that is keyed by vendor/product IDs
2897 and similar device identifiers. For the beginning this
2898 "hwdb" is populated with data from the well-known PCI and
2899 USB database, but also includes PNP, ACPI and OID data. In
2900 the longer run this indexed database shall grow into
2901 becoming the one central database for non-essential
2902 userspace device metadata. Previously, data from the PCI/USB
2903 database was only attached to select devices, since the
2904 lookup was a relatively expensive operation due to O(n) time
2905 complexity (with n being the number of entries in the
2906 database). Since this is now O(1), we decided to add in this
2907 data for all devices where this is available, by
2908 default. Note that the indexed database needs to be rebuilt
2909 when new data files are installed. To achieve this you need
2910 to update your packaging scripts to invoke "udevadm hwdb
2911 --update" after installation of hwdb data files. For
2912 RPM-based distributions we introduced the new
2913 %udev_hwdb_update macro for this purpose.
2915 * The Journal gained support for the "Message Catalog", an
2916 indexed database to link up additional information with
2917 journal entries. For further details please check:
2919 http://www.freedesktop.org/wiki/Software/systemd/catalog
2921 The indexed message catalog database also needs to be
2922 rebuilt after installation of message catalog files. Use
2923 "journalctl --update-catalog" for this. For RPM-based
2924 distributions we introduced the %journal_catalog_update
2925 macro for this purpose.
2927 * The Python Journal bindings gained support for the standard
2928 Python logging framework.
2930 * The Journal API gained new functions for checking whether
2931 the underlying file system of a journal file is capable of
2932 properly reporting file change notifications, or whether
2933 applications that want to reflect journal changes "live"
2934 need to recheck journal files continuously in appropriate
2937 * It is now possible to set the "age" field for tmpfiles
2938 entries to 0, indicating that files matching this entry
2939 shall always be removed when the directories are cleaned up.
2941 * coredumpctl gained a new "gdb" verb which invokes gdb
2942 right-away on the selected coredump.
2944 * There's now support for "hybrid sleep" on kernels that
2945 support this, in addition to "suspend" and "hibernate". Use
2946 "systemctl hybrid-sleep" to make use of this.
2948 * logind's HandleSuspendKey= setting (and related settings)
2949 now gained support for a new "lock" setting to simply
2950 request the screen lock on all local sessions, instead of
2951 actually executing a suspend or hibernation.
2953 * systemd will now mount the EFI variables file system by
2956 * Socket units now gained support for configuration of the
2957 SMACK security label.
2959 * timedatectl will now output the time of the last and next
2960 daylight saving change.
2962 * We dropped support for various legacy and distro-specific
2963 concepts, such as insserv, early-boot SysV services
2964 (i.e. those for non-standard runlevels such as 'b' or 'S')
2965 or ArchLinux /etc/rc.conf support. We recommend the
2966 distributions who still need support this to either continue
2967 to maintain the necessary patches downstream, or find a
2968 different solution. (Talk to us if you have questions!)
2970 * Various systemd components will now bypass PolicyKit checks
2971 for root and otherwise handle properly if PolicyKit is not
2972 found to be around. This should fix most issues for
2973 PolicyKit-less systems. Quite frankly this should have been
2974 this way since day one. It is absolutely our intention to
2975 make systemd work fine on PolicyKit-less systems, and we
2976 consider it a bug if something does not work as it should if
2977 PolicyKit is not around.
2979 * For embedded systems it is now possible to build udev and
2980 systemd without blkid and/or kmod support.
2982 * "systemctl switch-root" is now capable of switching root
2983 more than once. I.e. in addition to transitions from the
2984 initrd to the host OS it is now possible to transition to
2985 further OS images from the host. This is useful to implement
2986 offline updating tools.
2988 * Various other additions have been made to the RPM macros
2989 shipped with systemd. Use %udev_rules_update() after
2990 installing new udev rules files. %_udevhwdbdir,
2991 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2992 %_sysctldir are now available which resolve to the right
2993 directories for packages to place various data files in.
2995 * journalctl gained the new --full switch (in addition to
2996 --all, to disable ellipsation for long messages.
2998 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2999 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3000 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3001 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3002 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3003 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3004 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3005 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3006 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3010 * journalctl gained new --since= and --until= switches to
3011 filter by time. It also now supports nice filtering for
3012 units via --unit=/-u.
3014 * Type=oneshot services may use ExecReload= and do the
3017 * The journal daemon now supports time-based rotation and
3018 vacuuming, in addition to the usual disk-space based
3021 * The journal will now index the available field values for
3022 each field name. This enables clients to show pretty drop
3023 downs of available match values when filtering. The bash
3024 completion of journalctl has been updated
3025 accordingly. journalctl gained a new switch -F to list all
3026 values a certain field takes in the journal database.
3028 * More service events are now written as structured messages
3029 to the journal, and made recognizable via message IDs.
3031 * The timedated, localed and hostnamed mini-services which
3032 previously only provided support for changing time, locale
3033 and hostname settings from graphical DEs such as GNOME now
3034 also have a minimal (but very useful) text-based client
3035 utility each. This is probably the nicest way to changing
3036 these settings from the command line now, especially since
3037 it lists available options and is fully integrated with bash
3040 * There's now a new tool "systemd-coredumpctl" to list and
3041 extract coredumps from the journal.
3043 * We now install a README each in /var/log/ and
3044 /etc/rc.d/init.d explaining where the system logs and init
3045 scripts went. This hopefully should help folks who go to
3046 that dirs and look into the otherwise now empty void and
3047 scratch their heads.
3049 * When user-services are invoked (by systemd --user) the
3050 $MANAGERPID env var is set to the PID of systemd.
3052 * SIGRTMIN+24 when sent to a --user instance will now result
3053 in immediate termination of systemd.
3055 * gatewayd received numerous feature additions such as a
3056 "follow" mode, for live syncing and filtering.
3058 * browse.html now allows filtering and showing detailed
3059 information on specific entries. Keyboard navigation and
3060 mouse screen support has been added.
3062 * gatewayd/journalctl now supports HTML5/JSON
3063 Server-Sent-Events as output.
3065 * The SysV init script compatibility logic will now
3066 heuristically determine whether a script supports the
3067 "reload" verb, and only then make this available as
3070 * "systemctl status --follow" has been removed, use "journalctl
3073 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3074 have been removed since they are hardly useful to be
3077 * And I'd like to take the opportunity to specifically mention
3078 Zbigniew for his great contributions. Zbigniew, you rock!
3080 Contributions from: Andrew Eikum, Christian Hesse, Colin
3081 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3082 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3083 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3084 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3085 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3086 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3090 * If /etc/vconsole.conf is non-existent or empty we will no
3091 longer load any console font or key map at boot by
3092 default. Instead the kernel defaults will be left
3093 intact. This is definitely the right thing to do, as no
3094 configuration should mean no configuration, and hard-coding
3095 font names that are different on all archs is probably a bad
3096 idea. Also, the kernel default key map and font should be
3097 good enough for most cases anyway, and mostly identical to
3098 the userspace fonts/key maps we previously overloaded them
3099 with. If distributions want to continue to default to a
3100 non-kernel font or key map they should ship a default
3101 /etc/vconsole.conf with the appropriate contents.
3103 Contributions from: Colin Walters, Daniel J Walsh, Dave
3104 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3105 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3109 * journalctl gained a new --cursor= switch to show entries
3110 starting from the specified location in the journal.
3112 * We now enforce a size limit on journal entry fields exported
3113 with "-o json" in journalctl. Fields larger than 4K will be
3114 assigned null. This can be turned off with --all.
3116 * An (optional) journal gateway daemon is now available as
3117 "systemd-journal-gatewayd.service". This service provides
3118 access to the journal via HTTP and JSON. This functionality
3119 will be used to implement live log synchronization in both
3120 pull and push modes, but has various other users too, such
3121 as easy log access for debugging of embedded devices. Right
3122 now it is already useful to retrieve the journal via HTTP:
3124 # systemctl start systemd-journal-gatewayd.service
3125 # wget http://localhost:19531/entries
3127 This will download the journal contents in a
3128 /var/log/messages compatible format. The same as JSON:
3130 # curl -H"Accept: application/json" http://localhost:19531/entries
3132 This service is also accessible via a web browser where a
3133 single static HTML5 app is served that uses the JSON logic
3134 to enable the user to do some basic browsing of the
3135 journal. This will be extended later on. Here's an example
3136 screenshot of this app in its current state:
3138 http://0pointer.de/public/journal-gatewayd
3140 Contributions from: Kay Sievers, Lennart Poettering, Robert
3141 Milasan, Tom Gundersen
3145 * The bash completion logic is now available for journalctl
3148 * We do not mount the "cpuset" controller anymore together with
3149 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3150 started if no parameters are assigned to it. "cpuset" hence
3151 broke code that assumed it it could create "cpu" groups and
3154 * journalctl -f will now subscribe to terminal size changes,
3155 and line break accordingly.
3157 Contributions from: Dave Reisner, Kay Sievers, Lennart
3158 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3162 * nspawn will now create a symlink /etc/localtime in the
3163 container environment, copying the host's timezone
3164 setting. Previously this has been done via a bind mount, but
3165 since symlinks cannot be bind mounted this has now been
3166 changed to create/update the appropriate symlink.
3168 * journalctl -n's line number argument is now optional, and
3169 will default to 10 if omitted.
3171 * journald will now log the maximum size the journal files may
3172 take up on disk. This is particularly useful if the default
3173 built-in logic of determining this parameter from the file
3174 system size is used. Use "systemctl status
3175 systemd-journald.service" to see this information.
3177 * The multi-seat X wrapper tool has been stripped down. As X
3178 is now capable of enumerating graphics devices via udev in a
3179 seat-aware way the wrapper is not strictly necessary
3180 anymore. A stripped down temporary stop-gap is still shipped
3181 until the upstream display managers have been updated to
3182 fully support the new X logic. Expect this wrapper to be
3183 removed entirely in one of the next releases.
3185 * HandleSleepKey= in logind.conf has been split up into
3186 HandleSuspendKey= and HandleHibernateKey=. The old setting
3187 is not available anymore. X11 and the kernel are
3188 distuingishing between these keys and we should too. This
3189 also means the inhibition lock for these keys has been split
3192 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3193 Poettering, Lukas Nykryn, Václav Pavlín
3197 * Whenever a unit changes state we will now log this to the
3198 journal and show along the unit's own log output in
3201 * ConditionPathIsMountPoint= can now properly detect bind
3202 mount points too. (Previously, a bind mount of one file
3203 system to another place in the same file system could not be
3204 detected as mount, since they shared struct stat's st_dev
3207 * We will now mount the cgroup controllers cpu, cpuacct,
3208 cpuset and the controllers net_cls, net_prio together by
3211 * nspawn containers will now have a virtualized boot
3212 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3213 over with a randomized ID at container initialization). This
3214 has the effect of making "journalctl -b" do the right thing
3217 * The JSON output journal serialization has been updated not
3218 to generate "endless" list objects anymore, but rather one
3219 JSON object per line. This is more in line how most JSON
3220 parsers expect JSON objects. The new output mode
3221 "json-pretty" has been added to provide similar output, but
3222 neatly aligned for readability by humans.
3224 * We dropped all explicit sync() invocations in the shutdown
3225 code. The kernel does this implicitly anyway in the kernel
3226 reboot() syscall. halt(8)'s -n option is now a compatibility
3229 * We now support virtualized reboot() in containers, as
3230 supported by newer kernels. We will fall back to exit() if
3231 CAP_SYS_REBOOT is not available to the container. Also,
3232 nspawn makes use of this now and will actually reboot the
3233 container if the containerized OS asks for that.
3235 * journalctl will only show local log output by default
3236 now. Use --merge (-m) to show remote log output, too.
3238 * libsystemd-journal gained the new sd_journal_get_usage()
3239 call to determine the current disk usage of all journal
3240 files. This is exposed in the new "journalctl --disk-usage"
3243 * journald gained a new configuration setting SplitMode= in
3244 journald.conf which may be used to control how user journals
3245 are split off. See journald.conf(5) for details.
3247 * A new condition type ConditionFileNotEmpty= has been added.
3249 * tmpfiles' "w" lines now support file globbing, to write
3250 multiple files at once.
3252 * We added Python bindings for the journal submission
3253 APIs. More Python APIs for a number of selected APIs will
3254 likely follow. Note that we intend to add native bindings
3255 only for the Python language, as we consider it common
3256 enough to deserve bindings shipped within systemd. There are
3257 various projects outside of systemd that provide bindings
3258 for languages such as PHP or Lua.
3260 * Many conditions will now resolve specifiers such as %i. In
3261 addition, PathChanged= and related directives of .path units
3262 now support specifiers as well.
3264 * There's now a new RPM macro definition for the system preset
3267 * journald will now warn if it ca not forward a message to the
3268 syslog daemon because its socket is full.
3270 * timedated will no longer write or process /etc/timezone,
3271 except on Debian. As we do not support late mounted /usr
3272 anymore /etc/localtime always being a symlink is now safe,
3273 and hence the information in /etc/timezone is not necessary
3276 * logind will now always reserve one VT for a text getty (VT6
3277 by default). Previously if more than 6 X sessions where
3278 started they took up all the VTs with auto-spawned gettys,
3279 so that no text gettys were available anymore.
3281 * udev will now automatically inform the btrfs kernel logic
3282 about btrfs RAID components showing up. This should make
3283 simple hotplug based btrfs RAID assembly work.
3285 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3286 (but not for its children which will stay at the kernel
3287 default). This should allow setups with a lot more listening
3290 * systemd will now always pass the configured timezone to the
3291 kernel at boot. timedated will do the same when the timezone
3294 * logind's inhibition logic has been updated. By default,
3295 logind will now handle the lid switch, the power and sleep
3296 keys all the time, even in graphical sessions. If DEs want
3297 to handle these events on their own they should take the new
3298 handle-power-key, handle-sleep-key and handle-lid-switch
3299 inhibitors during their runtime. A simple way to achiveve
3300 that is to invoke the DE wrapped in an invocation of:
3302 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3304 * Access to unit operations is now checked via SELinux taking
3305 the unit file label and client process label into account.
3307 * systemd will now notify the administrator in the journal
3308 when he over-mounts a non-empty directory.
3310 * There are new specifiers that are resolved in unit files,
3311 for the host name (%H), the machine ID (%m) and the boot ID
3314 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3315 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3316 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3317 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3318 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3319 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3320 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3324 * Support for reading structured kernel messages from
3325 /dev/kmsg has now been added and is enabled by default.
3327 * Support for reading kernel messages from /proc/kmsg has now
3328 been removed. If you want kernel messages in the journal
3329 make sure to run a recent kernel (>= 3.5) that supports
3330 reading structured messages from /dev/kmsg (see
3331 above). /proc/kmsg is now exclusive property of classic
3332 syslog daemons again.
3334 * The libudev API gained the new
3335 udev_device_new_from_device_id() call.
3337 * The logic for file system namespace (ReadOnlyDirectory=,
3338 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3339 require pivot_root() anymore. This means fewer temporary
3340 directories are created below /tmp for this feature.
3342 * nspawn containers will now see and receive all submounts
3343 made on the host OS below the root file system of the
3346 * Forward Secure Sealing is now supported for Journal files,
3347 which provide cryptographical sealing of journal files so
3348 that attackers cannot alter log history anymore without this
3349 being detectable. Lennart will soon post a blog story about
3350 this explaining it in more detail.
3352 * There are two new service settings RestartPreventExitStatus=
3353 and SuccessExitStatus= which allow configuration of exit
3354 status (exit code or signal) which will be excepted from the
3355 restart logic, resp. consider successful.
3357 * journalctl gained the new --verify switch that can be used
3358 to check the integrity of the structure of journal files and
3359 (if Forward Secure Sealing is enabled) the contents of
3362 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3363 and similar symlinks pre-created. This makes running shells
3364 as container init process a lot more fun.
3366 * The fstab support can now handle PARTUUID= and PARTLABEL=
3369 * A new ConditionHost= condition has been added to match
3370 against the hostname (with globs) and machine ID. This is
3371 useful for clusters where a single OS image is used to
3372 provision a large number of hosts which shall run slightly
3373 different sets of services.
3375 * Services which hit the restart limit will now be placed in a
3378 Contributions from: Bertram Poettering, Dave Reisner, Huang
3379 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3380 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3384 * When running in --user mode systemd will now become a
3385 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3386 tree a lot more organized.
3388 * A new PartOf= unit dependency type has been introduced that
3389 may be used to group services in a natural way.
3391 * "systemctl enable" may now be used to enable instances of
3394 * journalctl now prints error log levels in red, and
3395 warning/notice log levels in bright white. It also supports
3396 filtering by log level now.
3398 * cgtop gained a new -n switch (similar to top), to configure
3399 the maximum number of iterations to run for. It also gained
3400 -b, to run in batch mode (accepting no input).
3402 * The suffix ".service" may now be omitted on most systemctl
3403 command lines involving service unit names.
3405 * There's a new bus call in logind to lock all sessions, as
3406 well as a loginctl verb for it "lock-sessions".
3408 * libsystemd-logind.so gained a new call sd_journal_perror()
3409 that works similar to libc perror() but logs to the journal
3410 and encodes structured information about the error number.
3412 * /etc/crypttab entries now understand the new keyfile-size=
3415 * shutdown(8) now can send a (configurable) wall message when
3416 a shutdown is cancelled.
3418 * The mount propagation mode for the root file system will now
3419 default to "shared", which is useful to make containers work
3420 nicely out-of-the-box so that they receive new mounts from
3421 the host. This can be undone locally by running "mount
3422 --make-rprivate /" if needed.
3424 * The prefdm.service file has been removed. Distributions
3425 should maintain this unit downstream if they intend to keep
3426 it around. However, we recommend writing normal unit files
3427 for display managers instead.
3429 * Since systemd is a crucial part of the OS we will now
3430 default to a number of compiler switches that improve
3431 security (hardening) such as read-only relocations, stack
3432 protection, and suchlike.
3434 * The TimeoutSec= setting for services is now split into
3435 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3436 of individual time outs for the start and the stop phase of
3439 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3440 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3441 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3442 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3443 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3444 Gundersen, Zbigniew Jędrzejewski-Szmek
3448 * The journal and id128 C APIs are now fully documented as man
3451 * Extra safety checks have been added when transitioning from
3452 the initial RAM disk to the main system to avoid accidental
3455 * /etc/crypttab entries now understand the new keyfile-offset=
3458 * systemctl -t can now be used to filter by unit load state.
3460 * The journal C API gained the new sd_journal_wait() call to
3461 make writing synchronous journal clients easier.
3463 * journalctl gained the new -D switch to show journals from a
3466 * journalctl now displays a special marker between log
3467 messages of two different boots.
3469 * The journal is now explicitly flushed to /var via a service
3470 systemd-journal-flush.service, rather than implicitly simply
3471 by seeing /var/log/journal to be writable.
3473 * journalctl (and the journal C APIs) can now match for much
3474 more complex expressions, with alternatives and
3477 * When transitioning from the initial RAM disk to the main
3478 system we will now kill all processes in a killing spree to
3479 ensure no processes stay around by accident.
3481 * Three new specifiers may be used in unit files: %u, %h, %s
3482 resolve to the user name, user home directory resp. user
3483 shell. This is useful for running systemd user instances.
3485 * We now automatically rotate journal files if their data
3486 object hash table gets a fill level > 75%. We also size the
3487 hash table based on the configured maximum file size. This
3488 together should lower hash collisions drastically and thus
3489 speed things up a bit.
3491 * journalctl gained the new "--header" switch to introspect
3492 header data of journal files.
3494 * A new setting SystemCallFilters= has been added to services
3495 which may be used to apply blacklists or whitelists to
3496 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3498 * nspawn gained a new --link-journal= switch (and quicker: -j)
3499 to link the container journal with the host. This makes it
3500 very easy to centralize log viewing on the host for all
3501 guests while still keeping the journal files separated.
3503 * Many bugfixes and optimizations
3505 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3506 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3507 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3512 * Several tools now understand kernel command line arguments,
3513 which are only read when run in an initial RAM disk. They
3514 usually follow closely their normal counterparts, but are
3517 * There's a new tool to analyze the readahead files that are
3518 automatically generated at boot. Use:
3520 /usr/lib/systemd/systemd-readahead analyze /.readahead
3522 * We now provide an early debug shell on tty9 if this enabled. Use:
3524 systemctl enable debug-shell.service
3526 * All plymouth related units have been moved into the Plymouth
3527 package. Please make sure to upgrade your Plymouth version
3530 * systemd-tmpfiles now supports getting passed the basename of
3531 a configuration file only, in which case it will look for it
3532 in all appropriate directories automatically.
3534 * udevadm info now takes a /dev or /sys path as argument, and
3535 does the right thing. Example:
3537 udevadm info /dev/sda
3538 udevadm info /sys/class/block/sda
3540 * systemctl now prints a warning if a unit is stopped but a
3541 unit that might trigger it continues to run. Example: a
3542 service is stopped but the socket that activates it is left
3545 * "systemctl status" will now mention if the log output was
3546 shortened due to rotation since a service has been started.
3548 * The journal API now exposes functions to determine the
3549 "cutoff" times due to rotation.
3551 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3552 immediately flushing of runtime logs to /var if possible,
3553 resp. for triggering immediate rotation of the journal
3556 * It is now considered an error if a service is attempted to
3557 be stopped that is not loaded.
3559 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3561 * systemd-analyze now supports Python 3
3563 * tmpfiles now supports cleaning up directories via aging
3564 where the first level dirs are always kept around but
3565 directories beneath it automatically aged. This is enabled
3566 by prefixing the age field with '~'.
3568 * Seat objects now expose CanGraphical, CanTTY properties
3569 which is required to deal with very fast bootups where the
3570 display manager might be running before the graphics drivers
3571 completed initialization.
3573 * Seat objects now expose a State property.
3575 * We now include RPM macros for service enabling/disabling
3576 based on the preset logic. We recommend RPM based
3577 distributions to make use of these macros if possible. This
3578 makes it simpler to reuse RPM spec files across
3581 * We now make sure that the collected systemd unit name is
3582 always valid when services log to the journal via
3585 * There's a new man page kernel-command-line(7) detailing all
3586 command line options we understand.
3588 * The fstab generator may now be disabled at boot by passing
3589 fstab=0 on the kernel command line.
3591 * A new kernel command line option modules-load= is now understood
3592 to load a specific kernel module statically, early at boot.
3594 * Unit names specified on the systemctl command line are now
3595 automatically escaped as needed. Also, if file system or
3596 device paths are specified they are automatically turned
3597 into the appropriate mount or device unit names. Example:
3599 systemctl status /home
3600 systemctl status /dev/sda
3602 * The SysVConsole= configuration option has been removed from
3603 system.conf parsing.
3605 * The SysV search path is no longer exported on the D-Bus
3608 * The Names= option is been removed from unit file parsing.
3610 * There's a new man page bootup(7) detailing the boot process.
3612 * Every unit and every generator we ship with systemd now
3613 comes with full documentation. The self-explanatory boot is
3616 * A couple of services gained "systemd-" prefixes in their
3617 name if they wrap systemd code, rather than only external
3618 code. Among them fsck@.service which is now
3619 systemd-fsck@.service.
3621 * The HaveWatchdog property has been removed from the D-Bus
3624 * systemd.confirm_spawn= on the kernel command line should now
3627 * There's a new man page crypttab(5) which details all options
3628 we actually understand.
3630 * systemd-nspawn gained a new --capability= switch to pass
3631 additional capabilities to the container.
3633 * timedated will now read known NTP implementation unit names
3634 from /usr/lib/systemd/ntp-units.d/*.list,
3635 systemd-timedated-ntp.target has been removed.
3637 * journalctl gained a new switch "-b" that lists log data of
3638 the current boot only.
3640 * The notify socket is in the abstract namespace again, in
3641 order to support daemons which chroot() at start-up.
3643 * There is a new Storage= configuration option for journald
3644 which allows configuration of where log data should go. This
3645 also provides a way to disable journal logging entirely, so
3646 that data collected is only forwarded to the console, the
3647 kernel log buffer or another syslog implementation.
3649 * Many bugfixes and optimizations
3651 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3652 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3653 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3654 Shawn Landden, Tom Gundersen
3658 * "systemctl help <unit>" now shows the man page if one is
3661 * Several new man pages have been added.
3663 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3664 MaxLevelConsole= can now be specified in
3665 journald.conf. These options allow reducing the amount of
3666 data stored on disk or forwarded by the log level.
3668 * TimerSlackNSec= can now be specified in system.conf for
3669 PID1. This allows system-wide power savings.
3671 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3672 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3677 * logind is now capable of (optionally) handling power and
3678 sleep keys as well as the lid switch.
3680 * journalctl now understands the syntax "journalctl
3681 /usr/bin/avahi-daemon" to get all log output of a specific
3684 * CapabilityBoundingSet= in system.conf now also influences
3685 the capability bound set of usermode helpers of the kernel.
3687 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3688 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3689 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3690 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3694 * Note that we skipped 139 releases here in order to set the
3695 new version to something that is greater than both udev's
3696 and systemd's most recent version number.
3698 * udev: all udev sources are merged into the systemd source tree now.
3699 All future udev development will happen in the systemd tree. It
3700 is still fully supported to use the udev daemon and tools without
3701 systemd running, like in initramfs or other init systems. Building
3702 udev though, will require the *build* of the systemd tree, but
3703 udev can be properly *run* without systemd.
3705 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3706 should be used to create dead device nodes as workarounds for broken
3709 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3710 no longer supported. udev_monitor_new_from_netlink() needs to be
3711 used to subscribe to events.
3713 * udev: when udevd is started by systemd, processes which are left
3714 behind by forking them off of udev rules, are unconditionally cleaned
3715 up and killed now after the event handling has finished. Services or
3716 daemons must be started as systemd services. Services can be
3717 pulled-in by udev to get started, but they can no longer be directly
3718 forked by udev rules.
3720 * udev: the daemon binary is called systemd-udevd now and installed
3721 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3722 to adapt to that, create symlink, or rename the binary after building
3725 * libudev no longer provides these symbols:
3726 udev_monitor_from_socket()
3727 udev_queue_get_failed_list_entry()
3728 udev_get_{dev,sys,run}_path()
3729 The versions number was bumped and symbol versioning introduced.
3731 * systemd-loginctl and systemd-journalctl have been renamed
3732 to loginctl and journalctl to match systemctl.
3734 * The config files: /etc/systemd/systemd-logind.conf and
3735 /etc/systemd/systemd-journald.conf have been renamed to
3736 logind.conf and journald.conf. Package updates should rename
3737 the files to the new names on upgrade.
3739 * For almost all files the license is now LGPL2.1+, changed
3740 from the previous GPL2.0+. Exceptions are some minor stuff
3741 of udev (which will be changed to LGPL2.1 eventually, too),
3742 and the MIT licensed sd-daemon.[ch] library that is suitable
3743 to be used as drop-in files.
3745 * systemd and logind now handle system sleep states, in
3746 particular suspending and hibernating.
3748 * logind now implements a sleep/shutdown/idle inhibiting logic
3749 suitable for a variety of uses. Soonishly Lennart will blog
3750 about this in more detail.
3752 * var-run.mount and var-lock.mount are no longer provided
3753 (which prevously bind mounted these directories to their new
3754 places). Distributions which have not converted these
3755 directories to symlinks should consider stealing these files
3756 from git history and add them downstream.
3758 * We introduced the Documentation= field for units and added
3759 this to all our shipped units. This is useful to make it
3760 easier to explore the boot and the purpose of the various
3763 * All smaller setup units (such as
3764 systemd-vconsole-setup.service) now detect properly if they
3765 are run in a container and are skipped when
3766 appropriate. This guarantees an entirely noise-free boot in
3767 Linux container environments such as systemd-nspawn.
3769 * A framework for implementing offline system updates is now
3770 integrated, for details see:
3771 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3773 * A new service type Type=idle is available now which helps us
3774 avoiding ugly interleaving of getty output and boot status
3777 * There's now a system-wide CapabilityBoundingSet= option to
3778 globally reduce the set of capabilities for the
3779 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3780 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3781 even CAP_NET_ADMIN system-wide for secure systems.
3783 * There are now system-wide DefaultLimitXXX= options to
3784 globally change the defaults of the various resource limits
3785 for all units started by PID 1.
3787 * Harald Hoyer's systemd test suite has been integrated into
3788 systemd which allows easy testing of systemd builds in qemu
3789 and nspawn. (This is really awesome! Ask us for details!)
3791 * The fstab parser is now implemented as generator, not inside
3794 * systemctl will now warn you if .mount units generated from
3795 /etc/fstab are out of date due to changes in fstab that
3796 have not been read by systemd yet.
3798 * systemd is now suitable for usage in initrds. Dracut has
3799 already been updated to make use of this. With this in place
3800 initrds get a slight bit faster but primarily are much
3801 easier to introspect and debug since "systemctl status" in
3802 the host system can be used to introspect initrd services,
3803 and the journal from the initrd is kept around too.
3805 * systemd-delta has been added, a tool to explore differences
3806 between user/admin configuration and vendor defaults.
3808 * PrivateTmp= now affects both /tmp and /var/tmp.
3810 * Boot time status messages are now much prettier and feature
3811 proper english language. Booting up systemd has never been
3814 * Read-ahead pack files now include the inode number of all
3815 files to pre-cache. When the inode changes the pre-caching
3816 is not attempted. This should be nicer to deal with updated
3817 packages which might result in changes of read-ahead
3820 * We now temporaritly lower the kernel's read_ahead_kb variable
3821 when collecting read-ahead data to ensure the kernel's
3822 built-in read-ahead does not add noise to our measurements
3823 of necessary blocks to pre-cache.
3825 * There's now RequiresMountsFor= to add automatic dependencies
3826 for all mounts necessary for a specific file system path.
3828 * MountAuto= and SwapAuto= have been removed from
3829 system.conf. Mounting file systems at boot has to take place
3832 * nspawn now learned a new switch --uuid= to set the machine
3833 ID on the command line.
3835 * nspawn now learned the -b switch to automatically search
3838 * vt102 is now the default TERM for serial TTYs, upgraded from
3841 * systemd-logind now works on VT-less systems.
3843 * The build tree has been reorganized. The individual
3844 components now have directories of their own.
3846 * A new condition type ConditionPathIsReadWrite= is now available.
3848 * nspawn learned the new -C switch to create cgroups for the
3849 container in other hierarchies.
3851 * We now have support for hardware watchdogs, configurable in
3854 * The scheduled shutdown logic now has a public API.
3856 * We now mount /tmp as tmpfs by default, but this can be
3857 masked and /etc/fstab can override it.
3859 * Since udisks does not make use of /media anymore we are not
3860 mounting a tmpfs on it anymore.
3862 * journalctl gained a new --local switch to only interleave
3863 locally generated journal files.
3865 * We can now load the IMA policy at boot automatically.
3867 * The GTK tools have been split off into a systemd-ui.
3869 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3870 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3871 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3872 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3873 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3874 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3875 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3876 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3877 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3882 * This is mostly a bugfix release
3884 * Support optional initialization of the machine ID from the
3885 KVM or container configured UUID.
3887 * Support immediate reboots with "systemctl reboot -ff"
3889 * Show /etc/os-release data in systemd-analyze output
3891 * Many bugfixes for the journal, including endianness fixes and
3892 ensuring that disk space enforcement works
3894 * sd-login.h is C++ comptaible again
3896 * Extend the /etc/os-release format on request of the Debian
3899 * We now refuse non-UTF8 strings used in various configuration
3900 and unit files. This is done to ensure we do not pass invalid
3901 data over D-Bus or expose it elsewhere.
3903 * Register Mimo USB Screens as suitable for automatic seat
3906 * Read SELinux client context from journal clients in a race
3909 * Reorder configuration file lookup order. /etc now always
3910 overrides /run in order to allow the administrator to always
3911 and unconditionally override vendor supplied or
3912 automatically generated data.
3914 * The various user visible bits of the journal now have man
3915 pages. We still lack man pages for the journal API calls
3918 * We now ship all man pages in HTML format again in the
3921 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3922 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3923 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3928 * This is mostly a bugfix release
3930 * systems lacking /etc/os-release are no longer supported.
3932 * Various functionality updates to libsystemd-login.so
3934 * Track class of PAM logins to distuingish greeters from
3937 Contributions from: Kay Sievers, Lennart Poettering, Michael
3942 * This is an important bugfix release for v41.
3944 * Building man pages is now optional which should be useful
3945 for those building systemd from git but unwilling to install
3948 * Watchdog support for supervising services is now usable. In
3949 a future release support for hardware watchdogs
3950 (i.e. /dev/watchdog) will be added building on this.
3952 * Service start rate limiting is now configurable and can be
3953 turned off per service. When a start rate limit is hit a
3954 reboot can automatically be triggered.
3956 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3958 Contributions from: Benjamin Franzke, Bill Nottingham,
3959 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3960 Schmidt, Michał Górny, Piotr Drąg
3964 * The systemd binary is installed /usr/lib/systemd/systemd now;
3965 An existing /sbin/init symlink needs to be adapted with the
3968 * The code that loads kernel modules has been ported to invoke
3969 libkmod directly, instead of modprobe. This means we do not
3970 support systems with module-init-tools anymore.
3972 * Watchdog support is now already useful, but still not
3975 * A new kernel command line option systemd.setenv= is
3976 understood to set system wide environment variables
3977 dynamically at boot.
3979 * We now limit the set of capabilities of systemd-journald.
3981 * We now set SIGPIPE to ignore by default, since it only is
3982 useful in shell pipelines, and has little use in general
3983 code. This can be disabled with IgnoreSIPIPE=no in unit
3986 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3987 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3992 * This is mostly a bugfix release
3994 * We now expose the reason why a service failed in the
3995 "Result" D-Bus property.
3997 * Rudimentary service watchdog support (will be completed over
3998 the next few releases.)
4000 * When systemd forks off in order execute some service we will
4001 now immediately changes its argv[0] to reflect which process
4002 it will execute. This is useful to minimize the time window
4003 with a generic argv[0], which makes bootcharts more useful
4005 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4006 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4007 Mike Kazantsev, Ray Strode
4011 * This is mostly a test release, but incorporates many
4014 * New systemd-cgtop tool to show control groups by their
4017 * Linking against libacl for ACLs is optional again. If
4018 disabled, support tracking device access for active logins
4019 goes becomes unavailable, and so does access to the user
4020 journals by the respective users.
4022 * If a group "adm" exists, journal files are automatically
4023 owned by them, thus allow members of this group full access
4024 to the system journal as well as all user journals.
4026 * The journal now stores the SELinux context of the logging
4027 client for all entries.
4029 * Add C++ inclusion guards to all public headers
4031 * New output mode "cat" in the journal to print only text
4032 messages, without any meta data like date or time.
4034 * Include tiny X server wrapper as a temporary stop-gap to
4035 teach XOrg udev display enumeration. This is used by display
4036 managers such as gdm, and will go away as soon as XOrg
4037 learned native udev hotplugging for display devices.
4039 * Add new systemd-cat tool for executing arbitrary programs
4040 with STDERR/STDOUT connected to the journal. Can also act as
4041 BSD logger replacement, and does so by default.
4043 * Optionally store all locally generated coredumps in the
4044 journal along with meta data.
4046 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4047 writing short strings to files (for usage for /sys), and for
4048 creating symlinks, character and block device nodes.
4050 * New unit file option ControlGroupPersistent= to make cgroups
4051 persistent, following the mechanisms outlined in
4052 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4054 * Support multiple local RTCs in a sane way
4056 * No longer monopolize IO when replaying readahead data on
4057 rotating disks, since we might starve non-file-system IO to
4058 death, since fanotify() will not see accesses done by blkid,
4061 * Do not show kernel threads in systemd-cgls anymore, unless
4062 requested with new -k switch.
4064 Contributions from: Dan Horák, Kay Sievers, Lennart
4065 Poettering, Michal Schmidt
4069 * This is mostly a test release, but incorporates many
4072 * The git repository moved to:
4073 git://anongit.freedesktop.org/systemd/systemd
4074 ssh://git.freedesktop.org/git/systemd/systemd
4076 * First release with the journal
4077 http://0pointer.de/blog/projects/the-journal.html
4079 * The journal replaces both systemd-kmsg-syslogd and
4080 systemd-stdout-bridge.
4082 * New sd_pid_get_unit() API call in libsystemd-logind
4084 * Many systemadm clean-ups
4086 * Introduce remote-fs-pre.target which is ordered before all
4087 remote mounts and may be used to start services before all
4090 * Added Mageia support
4092 * Add bash completion for systemd-loginctl
4094 * Actively monitor PID file creation for daemons which exit in
4095 the parent process before having finished writing the PID
4096 file in the daemon process. Daemons which do this need to be
4097 fixed (i.e. PID file creation must have finished before the
4098 parent exits), but we now react a bit more gracefully to them.
4100 * Add colourful boot output, mimicking the well-known output
4101 of existing distributions.
4103 * New option PassCredentials= for socket units, for
4104 compatibility with a recent kernel ABI breakage.
4106 * /etc/rc.local is now hooked in via a generator binary, and
4107 thus will no longer act as synchronization point during
4110 * systemctl list-unit-files now supports --root=.
4112 * systemd-tmpfiles now understands two new commands: z, Z for
4113 relabelling files according to the SELinux database. This is
4114 useful to apply SELinux labels to specific files in /sys,
4117 * Output of SysV services is now forwarded to both the console
4118 and the journal by default, not only just the console.
4120 * New man pages for all APIs from libsystemd-login.
4122 * The build tree got reorganized and a the build system is a
4123 lot more modular allowing embedded setups to specifically
4124 select the components of systemd they are interested in.
4126 * Support for Linux systems lacking the kernel VT subsystem is
4129 * configure's --with-rootdir= got renamed to
4130 --with-rootprefix= to follow the naming used by udev and
4133 * Unless specified otherwise we will now install to /usr instead
4134 of /usr/local by default.
4136 * Processes with '@' in argv[0][0] are now excluded from the
4137 final shut-down killing spree, following the logic explained
4139 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4141 * All processes remaining in a service cgroup when we enter
4142 the START or START_PRE states are now killed with
4143 SIGKILL. That means it is no longer possible to spawn
4144 background processes from ExecStart= lines (which was never
4145 supported anyway, and bad style).
4147 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4148 reloading of units together.
4150 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4151 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4152 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4153 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4154 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek