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-activated 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 circumstances 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 Options= 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 user names.
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 Queuing Controlled Delay as the default
96 queuing 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 mount.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 processes of
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 privileged operations such as
156 unit file enabling and disabling.
158 * "udevadm hwdb --update" learnt a new switch "--usr" for
159 placing the rebuilt hardware database in /usr instead of
160 /etc. When used only hardware database entries stored in
161 /usr will be used, and any user database entries in /etc are
162 ignored. This functionality is useful for vendors to ship a
163 pre-built database on systems where local configuration is
164 unnecessary or unlikely.
166 * Calendar time specifications in .timer units now also
167 understand the strings "semi-annually", "quarterly" and
168 "minutely" as shortcuts (in addition to the preexisting
169 "anually", "hourly", ...).
171 * systemd-tmpfiles will now correctly create files in /dev
172 at boot which are marked for creation only at boot. It is
173 recommended to always create static device nodes with 'c!'
174 and 'b!', so that they are created only at boot and not
175 overwritten at runtime.
177 * When the watchdog logic is used for a service (WatchdogSec=)
178 and the watchdog timeout is hit the service will now be
179 terminated with SIGABRT (instead of just SIGTERM), in order
180 to make sure a proper coredump and backtrace is
181 generated. This ensures that hanging services will result in
182 similar coredump/backtrace behaviour as services that hit a
185 Contributions from: Andreas Henriksson, Andrei Borzenkov,
186 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
187 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
188 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
189 Herrmann, David Sommerseth, David Strauss, Emil Renner
190 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
191 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
192 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
193 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
194 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
195 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
196 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
197 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
198 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
199 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
200 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
201 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
202 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
203 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
204 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
205 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
208 -- Berlin, 2014-10-28
212 * timedated no longer reads NTP implementation unit names from
213 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
214 implementations should add a
216 Conflicts=systemd-timesyncd.service
218 to their unit files to take over and replace systemd's NTP
219 default functionality.
221 * systemd-sysusers gained a new line type "r" for configuring
222 which UID/GID ranges to allocate system users/groups
223 from. Lines of type "u" may now add an additional column
224 that specifies the home directory for the system user to be
225 created. Also, systemd-sysusers may now optionally read user
226 information from STDIN instead of a file. This is useful for
227 invoking it from RPM preinst scriptlets that need to create
228 users before the first RPM file is installed since these
229 files might need to be owned by them. A new
230 %sysusers_create_inline RPM macro has been introduced to do
231 just that. systemd-sysusers now updates the shadow files as
232 well as the user/group databases, which should enhance
233 compatibility with certain tools like grpck.
235 * A number of bus APIs of PID 1 now optionally consult
236 PolicyKit to permit access for otherwise unprivileged
237 clients under certain conditions. Note that this currently
238 doesn't support interactive authentication yet, but this is
239 expected to be added eventually, too.
241 * /etc/machine-info now has new fields for configuring the
242 deployment environment of the machine, as well as the
243 location of the machine. hostnamectl has been updated with
244 new command to update these fields.
246 * systemd-timesyncd has been updated to automatically acquire
247 NTP server information from systemd-networkd, which might
248 have been discovered via DHCP.
250 * systemd-resolved now includes a caching DNS stub resolver
251 and a complete LLMNR name resolution implementation. A new
252 NSS module "nss-resolve" has been added which can be used
253 instead of glibc's own "nss-dns" to resolve hostnames via
254 systemd-resolved. Hostnames, addresses and arbitrary RRs may
255 be resolved via systemd-resolved D-Bus APIs. In contrast to
256 the glibc internal resolver systemd-resolved is aware of
257 multi-homed system, and keeps DNS server and caches separate
258 and per-interface. Queries are sent simultaneously on all
259 interfaces that have DNS servers configured, in order to
260 properly handle VPNs and local LANs which might resolve
261 separate sets of domain names. systemd-resolved may acquire
262 DNS server information from systemd-networkd automatically,
263 which in turn might have discovered them via DHCP. A tool
264 "systemd-resolve-host" has been added that may be used to
265 query the DNS logic in resolved. systemd-resolved implements
266 IDNA and automatically uses IDNA or UTF-8 encoding depending
267 on whether classic DNS or LLMNR is used as transport. In the
268 next releases we intend to add a DNSSEC and mDNS/DNS-SD
269 implementation to systemd-resolved.
271 * A new NSS module nss-mymachines has been added, that
272 automatically resolves the names of all local registered
273 containers to their respective IP addresses.
275 * A new client tool "networkctl" for systemd-networkd has been
276 added. It currently is entirely passive and will query
277 networking configuration from udev, rtnetlink and networkd,
278 and present it to the user in a very friendly
279 way. Eventually, we hope to extend it to become a full
280 control utility for networkd.
282 * .socket units gained a new DeferAcceptSec= setting that
283 controls the kernels' TCP_DEFER_ACCEPT sockopt for
284 TCP. Similar, support for controlling TCP keep-alive
285 settings has been added (KeepAliveTimeSec=,
286 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
287 turning off Nagle's algorithm on TCP has been added
290 * logind learned a new session type "web", for use in projects
291 like Cockpit which register web clients as PAM sessions.
293 * timer units with at least one OnCalendar= setting will now
294 be started only after timer-sync.target has been
295 reached. This way they will not elapse before the system
296 clock has been corrected by a local NTP client or
297 similar. This is particular useful on RTC-less embedded
298 machines, that come up with an invalid system clock.
300 * systemd-nspawn's --network-veth= switch should now result in
301 stable MAC addresses for both the outer and the inner side
304 * systemd-nspawn gained a new --volatile= switch for running
305 container instances with /etc or /var unpopulated.
307 * The kdbus client code has been updated to use the new Linux
308 3.17 memfd subsystem instead of the old kdbus-specific one.
310 * systemd-networkd's DHCP client and server now support
311 FORCERENEW. There are also new configuration options to
312 configure the vendor client identifier and broadcast mode
315 * systemd will no longer inform the kernel about the current
316 timezone, as this is necessarily incorrect and racy as the
317 kernel has no understanding of DST and similar
318 concepts. This hence means FAT timestamps will be always
319 considered UTC, similar to what Android is already
320 doing. Also, when the RTC is configured to the local time
321 (rather than UTC) systemd will never synchronize back to it,
322 as this might confuse Windows at a later boot.
324 * systemd-analyze gained a new command "verify" for offline
325 validation of unit files.
327 * systemd-networkd gained support for a couple of additional
328 settings for bonding networking setups. Also, the metric for
329 statically configured routes may now be configured. For
330 network interfaces where this is appropriate the peer IP
331 address may now be configured.
333 * systemd-networkd's DHCP client will no longer request
334 broadcasting by default, as this tripped up some networks.
335 For hardware where broadcast is required the feature should
336 be switched back on using RequestBroadcast=yes.
338 * systemd-networkd will now set up IPv4LL addresses (when
339 enabled) even if DHCP is configured successfully.
341 * udev will now default to respect network device names given
342 by the kernel when the kernel indicates that these are
343 predictable. This behavior can be tweaked by changing
344 NamePolicy= in the relevant .link file.
346 * A new library systemd-terminal has been added that
347 implements full TTY stream parsing and rendering. This
348 library is supposed to be used later on for implementing a
349 full userspace VT subsystem, replacing the current kernel
352 * A new tool systemd-journal-upload has been added to push
353 journal data to a remote system running
354 systemd-journal-remote.
356 * journald will no longer forward all local data to another
357 running syslog daemon. This change has been made because
358 rsyslog (which appears to be the most commonly used syslog
359 implementation these days) no longer makes use of this, and
360 instead pulls the data out of the journal on its own. Since
361 forwarding the messages to a non-existent syslog server is
362 more expensive than we assumed we have now turned this
363 off. If you run a syslog server that is not a recent rsyslog
364 version, you have to turn this option on again
365 (ForwardToSyslog= in journald.conf).
367 * journald now optionally supports the LZ4 compressor for
368 larger journal fields. This compressor should perform much
369 better than XZ which was the previous default.
371 * machinectl now shows the IP addresses of local containers,
372 if it knows them, plus the interface name of the container.
374 * A new tool "systemd-escape" has been added that makes it
375 easy to escape strings to build unit names and similar.
377 * sd_notify() messages may now include a new ERRNO= field
378 which is parsed and collected by systemd and shown among the
379 "systemctl status" output for a service.
381 * A new component "systemd-firstboot" has been added that
382 queries the most basic systemd information (timezone,
383 hostname, root password) interactively on first
384 boot. Alternatively it may also be used to provision these
385 things offline on OS images installed into directories.
387 * The default sysctl.d/ snippets will now set
389 net.ipv4.conf.default.promote_secondaries=1
391 This has the benefit of no flushing secondary IP addresses
392 when primary addresses are removed.
394 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
395 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
396 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
397 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
398 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
399 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
400 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
401 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
402 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
403 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
404 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
405 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
406 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
407 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
408 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
410 -- Berlin, 2014-08-19
414 * A new tool systemd-sysusers has been added. This tool
415 creates system users and groups in /etc/passwd and
416 /etc/group, based on static declarative system user/group
417 definitions in /usr/lib/sysusers.d/. This is useful to
418 enable factory resets and volatile systems that boot up with
419 an empty /etc directory, and thus need system users and
420 groups created during early boot. systemd now also ships
421 with two default sysusers.d/ files for the most basic
422 users and groups systemd and the core operating system
425 * A new tmpfiles snippet has been added that rebuilds the
426 essential files in /etc on boot, should they be missing.
428 * A directive for ensuring automatic clean-up of
429 /var/cache/man/ has been removed from the default
430 configuration. This line should now be shipped by the man
431 implementation. The necessary change has been made to the
432 man-db implementation. Note that you need to update your man
433 implementation to one that ships this line, otherwise no
434 automatic clean-up of /var/cache/man will take place.
436 * A new condition ConditionNeedsUpdate= has been added that
437 may conditionalize services to only run when /etc or /var
438 are "older" than the vendor operating system resources in
439 /usr. This is useful for reconstructing or updating /etc
440 after an offline update of /usr or a factory reset, on the
441 next reboot. Services that want to run once after such an
442 update or reset should use this condition and order
443 themselves before the new systemd-update-done.service, which
444 will mark the two directories as fully updated. A number of
445 service files have been added making use of this, to rebuild
446 the udev hardware database, the journald message catalog and
447 dynamic loader cache (ldconfig). The systemd-sysusers tool
448 described above also makes use of this now. With this in
449 place it is now possible to start up a minimal operating
450 system with /etc empty cleanly. For more information on the
451 concepts involved see this recent blog story:
453 http://0pointer.de/blog/projects/stateless.html
455 * A new system group "input" has been introduced, and all
456 input device nodes get this group assigned. This is useful
457 for system-level software to get access to input devices. It
458 complements what is already done for "audio" and "video".
460 * systemd-networkd learnt minimal DHCPv4 server support in
461 addition to the existing DHCPv4 client support. It also
462 learnt DHCPv6 client and IPv6 Router Solicitation client
463 support. The DHCPv4 client gained support for static routes
464 passed in from the server. Note that the [DHCPv4] section
465 known in older systemd-networkd versions has been renamed to
466 [DHCP] and is now also used by the DHCPv6 client. Existing
467 .network files using settings of this section should be
468 updated, though compatibility is maintained. Optionally, the
469 client hostname may now be sent to the DHCP server.
471 * networkd gained support for vxlan virtual networks as well
472 as tun/tap and dummy devices.
474 * networkd gained support for automatic allocation of address
475 ranges for interfaces from a system-wide pool of
476 addresses. This is useful for dynamically managing a large
477 number of interfaces with a single network configuration
478 file. In particular this is useful to easily assign
479 appropriate IP addresses to the veth links of a large number
482 * RPM macros for processing sysusers, sysctl and binfmt
483 drop-in snippets at package installation time have been
486 * The /etc/os-release file should now be placed in
487 /usr/lib/os-release. The old location is automatically
488 created as symlink. /usr/lib is the more appropriate
489 location of this file, since it shall actually describe the
490 vendor operating system shipped in /usr, and not the
491 configuration stored in /etc.
493 * .mount units gained a new boolean SloppyOptions= setting
494 that maps to mount(8)'s -s option which enables permissive
495 parsing of unknown mount options.
497 * tmpfiles learnt a new "L+" directive which creates a symlink
498 but (unlike "L") deletes a pre-existing file first, should
499 it already exist and not already be the correct
500 symlink. Similar, "b+", "c+" and "p+" directives have been
501 added as well, which create block and character devices, as
502 well as fifos in the filesystem, possibly removing any
503 pre-existing files of different types.
505 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
506 'argument' field (which so far specified the source to
507 symlink/copy the files from) is now optional. If omitted the
508 same file os copied from /usr/share/factory/ suffixed by the
509 full destination path. This is useful for populating /etc
510 with essential files, by copying them from vendor defaults
511 shipped in /usr/share/factory/etc.
513 * A new command "systemctl preset-all" has been added that
514 applies the service preset settings to all installed unit
515 files. A new switch --preset-mode= has been added that
516 controls whether only enable or only disable operations
519 * A new command "systemctl is-system-running" has been added
520 that allows checking the overall state of the system, for
521 example whether it is fully up and running.
523 * When the system boots up with an empty /etc, the equivalent
524 to "systemctl preset-all" is executed during early boot, to
525 make sure all default services are enabled after a factory
528 * systemd now contains a minimal preset file that enables the
529 most basic services systemd ships by default.
531 * Unit files' [Install] section gained a new DefaultInstance=
532 field for defining the default instance to create if a
533 template unit is enabled with no instance specified.
535 * A new passive target cryptsetup-pre.target has been added
536 that may be used by services that need to make they run and
537 finish before the first LUKS cryptographic device is set up.
539 * The /dev/loop-control and /dev/btrfs-control device nodes
540 are now owned by the "disk" group by default, opening up
541 access to this group.
543 * systemd-coredump will now automatically generate a
544 stack trace of all core dumps taking place on the system,
545 based on elfutils' libdw library. This stack trace is logged
548 * systemd-coredump may now optionally store coredumps directly
549 on disk (in /var/lib/systemd/coredump, possibly compressed),
550 instead of storing them unconditionally in the journal. This
551 mode is the new default. A new configuration file
552 /etc/systemd/coredump.conf has been added to configure this
553 and other parameters of systemd-coredump.
555 * coredumpctl gained a new "info" verb to show details about a
556 specific coredump. A new switch "-1" has also been added
557 that makes sure to only show information about the most
558 recent entry instead of all entries. Also, as the tool is
559 generally useful now the "systemd-" prefix of the binary
560 name has been removed. Distributions that want to maintain
561 compatibility with the old name should add a symlink from
562 the old name to the new name.
564 * journald's SplitMode= now defaults to "uid". This makes sure
565 that unprivileged users can access their own coredumps with
566 coredumpctl without restrictions.
568 * New kernel command line options "systemd.wants=" (for
569 pulling an additional unit during boot), "systemd.mask="
570 (for masking a specific unit for the boot), and
571 "systemd.debug-shell" (for enabling the debug shell on tty9)
572 have been added. This is implemented in the new generator
573 "systemd-debug-generator".
575 * systemd-nspawn will now by default filter a couple of
576 syscalls for containers, among them those required for
577 kernel module loading, direct x86 IO port access, swap
578 management, and kexec. Most importantly though
579 open_by_handle_at() is now prohibited for containers,
580 closing a hole similar to a recently discussed vulnerability
581 in docker regarding access to files on file hierarchies the
582 container should normally not have access to. Note that for
583 nspawn we generally make no security claims anyway (and
584 this is explicitly documented in the man page), so this is
585 just a fix for one of the most obvious problems.
587 * A new man page file-hierarchy(7) has been added that
588 contains a minimized, modernized version of the file system
589 layout systemd expects, similar in style to the FHS
590 specification or hier(5). A new tool systemd-path(1) has
591 been added to query many of these paths for the local
594 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
595 longer done. Since the directory now has a per-user size
596 limit, and is cleaned on logout this appears unnecessary,
597 in particular since this now brings the lifecycle of this
598 directory closer in line with how IPC objects are handled.
600 * systemd.pc now exports a number of additional directories,
601 including $libdir (which is useful to identify the library
602 path for the primary architecture of the system), and a
603 couple of drop-in directories.
605 * udev's predictable network interface names now use the dev_port
606 sysfs attribute, introduced in linux 3.15 instead of dev_id to
607 distinguish between ports of the same PCI function. dev_id should
608 only be used for ports using the same HW address, hence the need
611 * machined has been updated to export the OS version of a
612 container (read from /etc/os-release and
613 /usr/lib/os-release) on the bus. This is now shown in
614 "machinectl status" for a machine.
616 * A new service setting RestartForceExitStatus= has been
617 added. If configured to a set of exit signals or process
618 return values, the service will be restarted when the main
619 daemon process exits with any of them, regardless of the
622 * systemctl's -H switch for connecting to remote systemd
623 machines has been extended so that it may be used to
624 directly connect to a specific container on the
625 host. "systemctl -H root@foobar:waldi" will now connect as
626 user "root" to host "foobar", and then proceed directly to
627 the container named "waldi". Note that currently you have to
628 authenticate as user "root" for this to work, as entering
629 containers is a privileged operation.
631 Contributions from: Andreas Henriksson, Benjamin Steinwender,
632 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
633 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
634 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
635 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
636 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
637 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
638 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
639 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
640 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
641 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
642 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
644 -- Berlin, 2014-07-03
648 * As an experimental feature, udev now tries to lock the
649 disk device node (flock(LOCK_SH|LOCK_NB)) while it
650 executes events for the disk or any of its partitions.
651 Applications like partitioning programs can lock the
652 disk device node (flock(LOCK_EX)) and claim temporary
653 device ownership that way; udev will entirely skip all event
654 handling for this disk and its partitions. If the disk
655 was opened for writing, the close will trigger a partition
656 table rescan in udev's "watch" facility, and if needed
657 synthesize "change" events for the disk and all its partitions.
658 This is now unconditionally enabled, and if it turns out to
659 cause major problems, we might turn it on only for specific
660 devices, or might need to disable it entirely. Device Mapper
661 devices are excluded from this logic.
663 * We temporarily dropped the "-l" switch for fsck invocations,
664 since they collide with the flock() logic above. util-linux
665 upstream has been changed already to avoid this conflict,
666 and we will readd "-l" as soon as util-linux with this
667 change has been released.
669 * The dependency on libattr has been removed. Since a long
670 time, the extended attribute calls have moved to glibc, and
671 libattr is thus unnecessary.
673 * Virtualization detection works without priviliges now. This
674 means the systemd-detect-virt binary no longer requires
675 CAP_SYS_PTRACE file capabilities, and our daemons can run
676 with fewer privileges.
678 * systemd-networkd now runs under its own "systemd-network"
679 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
680 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
681 loses the ability to write to files owned by root this way.
683 * Similar, systemd-resolved now runs under its own
684 "systemd-resolve" user with no capabilities remaining.
686 * Similar, systemd-bus-proxyd now runs under its own
687 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
689 * systemd-networkd gained support for setting up "veth"
690 virtual ethernet devices for container connectivity, as well
691 as GRE and VTI tunnels.
693 * systemd-networkd will no longer automatically attempt to
694 manually load kernel modules necessary for certain tunnel
695 transports. Instead, it is assumed the kernel loads them
696 automatically when required. This only works correctly on
697 very new kernels. On older kernels, please consider adding
698 the kernel modules to /etc/modules-load.d/ as a work-around.
700 * The resolv.conf file systemd-resolved generates has been
701 moved to /run/systemd/resolve/. If you have a symlink from
702 /etc/resolv.conf, it might be necessary to correct it.
704 * Two new service settings, ProtectHome= and ProtectSystem=,
705 have been added. When enabled, they will make the user data
706 (such as /home) inaccessible or read-only and the system
707 (such as /usr) read-only, for specific services. This allows
708 very light-weight per-service sandboxing to avoid
709 modifications of user data or system files from
710 services. These two new switches have been enabled for all
711 of systemd's long-running services, where appropriate.
713 * Socket units gained new SocketUser= and SocketGroup=
714 settings to set the owner user and group of AF_UNIX sockets
715 and FIFOs in the file system.
717 * Socket units gained a new RemoveOnStop= setting. If enabled,
718 all FIFOS and sockets in the file system will be removed
719 when the specific socket unit is stopped.
721 * Socket units gained a new Symlinks= setting. It takes a list
722 of symlinks to create to file system sockets or FIFOs
723 created by the specific Unix sockets. This is useful to
724 manage symlinks to socket nodes with the same life-cycle as
727 * The /dev/log socket and /dev/initctl FIFO have been moved to
728 /run, and have been replaced by symlinks. This allows
729 connecting to these facilities even if PrivateDevices=yes is
730 used for a service (which makes /dev/log itself unavailable,
731 but /run is left). This also has the benefit of ensuring
732 that /dev only contains device nodes, directories and
733 symlinks, and nothing else.
735 * sd-daemon gained two new calls sd_pid_notify() and
736 sd_pid_notifyf(). They are similar to sd_notify() and
737 sd_notifyf(), but allow overriding of the source PID of
738 notification messages if permissions permit this. This is
739 useful to send notify messages on behalf of a different
740 process (for example, the parent process). The
741 systemd-notify tool has been updated to make use of this
742 when sending messages (so that notification messages now
743 originate from the shell script invoking systemd-notify and
744 not the systemd-notify process itself. This should minimize
745 a race where systemd fails to associate notification
746 messages to services when the originating process already
749 * A new "on-abnormal" setting for Restart= has been added. If
750 set, it will result in automatic restarts on all "abnormal"
751 reasons for a process to exit, which includes unclean
752 signals, core dumps, timeouts and watchdog timeouts, but
753 does not include clean and unclean exit codes or clean
754 signals. Restart=on-abnormal is an alternative for
755 Restart=on-failure for services that shall be able to
756 terminate and avoid restarts on certain errors, by
757 indicating so with an unclean exit code. Restart=on-failure
758 or Restart=on-abnormal is now the recommended setting for
759 all long-running services.
761 * If the InaccessibleDirectories= service setting points to a
762 mount point (or if there are any submounts contained within
763 it), it is now attempted to completely unmount it, to make
764 the file systems truly unavailable for the respective
767 * The ReadOnlyDirectories= service setting and
768 systemd-nspawn's --read-only parameter are now recursively
769 applied to all submounts, too.
771 * Mount units may now be created transiently via the bus APIs.
773 * The support for SysV and LSB init scripts has been removed
774 from the systemd daemon itself. Instead, it is now
775 implemented as a generator that creates native systemd units
776 from these scripts when needed. This enables us to remove a
777 substantial amount of legacy code from PID 1, following the
778 fact that many distributions only ship a very small number
779 of LSB/SysV init scripts nowadays.
781 * Priviliged Xen (dom0) domains are not considered
782 virtualization anymore by the virtualization detection
783 logic. After all, they generally have unrestricted access to
784 the hardware and usually are used to manage the unprivileged
787 * systemd-tmpfiles gained a new "C" line type, for copying
788 files or entire directories.
790 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
791 lines. So far, they have been non-globbing versions of the
792 latter, and have thus been redundant. In future, it is
793 recommended to only use "z". "m" has hence been removed
794 from the documentation, even though it stays supported.
796 * A tmpfiles snippet to recreate the most basic structure in
797 /var has been added. This is enough to create the /var/run →
798 /run symlink and create a couple of structural
799 directories. This allows systems to boot up with an empty or
800 volatile /var. Of course, while with this change, the core OS
801 now is capable with dealing with a volatile /var, not all
802 user services are ready for it. However, we hope that sooner
803 or later, many service daemons will be changed upstream so
804 that they are able to automatically create their necessary
805 directories in /var at boot, should they be missing. This is
806 the first step to allow state-less systems that only require
807 the vendor image for /usr to boot.
809 * systemd-nspawn has gained a new --tmpfs= switch to mount an
810 empty tmpfs instance to a specific directory. This is
811 particularly useful for making use of the automatic
812 reconstruction of /var (see above), by passing --tmpfs=/var.
814 * Access modes specified in tmpfiles snippets may now be
815 prefixed with "~", which indicates that they shall be masked
816 by whether the existing file or directory is currently
817 writable, readable or executable at all. Also, if specified,
818 the sgid/suid/sticky bits will be masked for all
821 * A new passive target unit "network-pre.target" has been
822 added which is useful for services that shall run before any
823 network is configured, for example firewall scripts.
825 * The "floppy" group that previously owned the /dev/fd*
826 devices is no longer used. The "disk" group is now used
827 instead. Distributions should probably deprecate usage of
830 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
831 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
832 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
833 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
834 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
835 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
838 -- Berlin, 2014-06-11
842 * A new "systemd-timesyncd" daemon has been added for
843 synchronizing the system clock across the network. It
844 implements an SNTP client. In contrast to NTP
845 implementations such as chrony or the NTP reference server,
846 this only implements a client side, and does not bother with
847 the full NTP complexity, focusing only on querying time from
848 one remote server and synchronizing the local clock to
849 it. Unless you intend to serve NTP to networked clients or
850 want to connect to local hardware clocks, this simple NTP
851 client should be more than appropriate for most
852 installations. The daemon runs with minimal privileges, and
853 has been hooked up with networkd to only operate when
854 network connectivity is available. The daemon saves the
855 current clock to disk every time a new NTP sync has been
856 acquired, and uses this to possibly correct the system clock
857 early at bootup, in order to accommodate for systems that
858 lack an RTC such as the Raspberry Pi and embedded devices,
859 and to make sure that time monotonically progresses on these
860 systems, even if it is not always correct. To make use of
861 this daemon, a new system user and group "systemd-timesync"
862 needs to be created on installation of systemd.
864 * The queue "seqnum" interface of libudev has been disabled, as
865 it was generally incompatible with device namespacing as
866 sequence numbers of devices go "missing" if the devices are
867 part of a different namespace.
869 * "systemctl list-timers" and "systemctl list-sockets" gained
870 a --recursive switch for showing units of these types also
871 for all local containers, similar in style to the already
872 supported --recursive switch for "systemctl list-units".
874 * A new RebootArgument= setting has been added for service
875 units, which may be used to specify a kernel reboot argument
876 to use when triggering reboots with StartLimitAction=.
878 * A new FailureAction= setting has been added for service
879 units which may be used to specify an operation to trigger
880 when a service fails. This works similarly to
881 StartLimitAction=, but unlike it, controls what is done
882 immediately rather than only after several attempts to
883 restart the service in question.
885 * hostnamed got updated to also expose the kernel name,
886 release, and version on the bus. This is useful for
887 executing commands like hostnamectl with the -H switch.
888 systemd-analyze makes use of this to properly display
889 details when running non-locally.
891 * The bootchart tool can now show cgroup information in the
894 * The CFS CPU quota cgroup attribute is now exposed for
895 services. The new CPUQuota= switch has been added for this
896 which takes a percentage value. Setting this will have the
897 result that a service may never get more CPU time than the
898 specified percentage, even if the machine is otherwise idle.
900 * systemd-networkd learned IPIP and SIT tunnel support.
902 * LSB init scripts exposing a dependency on $network will now
903 get a dependency on network-online.target rather than simply
904 network.target. This should bring LSB handling closer to
905 what it was on SysV systems.
907 * A new fsck.repair= kernel option has been added to control
908 how fsck shall deal with unclean file systems at boot.
910 * The (.ini) configuration file parser will now silently
911 ignore sections whose name begins with "X-". This may be
912 used to maintain application-specific extension sections in unit
915 * machined gained a new API to query the IP addresses of
916 registered containers. "machinectl status" has been updated
917 to show these addresses in its output.
919 * A new call sd_uid_get_display() has been added to the
920 sd-login APIs for querying the "primary" session of a
921 user. The "primary" session of the user is elected from the
922 user's sessions and generally a graphical session is
923 preferred over a text one.
925 * A minimal systemd-resolved daemon has been added. It
926 currently simply acts as a companion to systemd-networkd and
927 manages resolv.conf based on per-interface DNS
928 configuration, possibly supplied via DHCP. In the long run
929 we hope to extend this into a local DNSSEC enabled DNS and
932 * The systemd-networkd-wait-online tool is now enabled by
933 default. It will delay network-online.target until a network
934 connection has been configured. The tool primarily integrates
935 with networkd, but will also make a best effort to make sense
936 of network configuration performed in some other way.
938 * Two new service options StartupCPUShares= and
939 StartupBlockIOWeight= have been added that work similarly to
940 CPUShares= and BlockIOWeight= however only apply during
941 system startup. This is useful to prioritize certain services
942 differently during bootup than during normal runtime.
944 * hostnamed has been changed to prefer the statically
945 configured hostname in /etc/hostname (unless set to
946 'localhost' or empty) over any dynamic one supplied by
947 dhcp. With this change, the rules for picking the hostname
948 match more closely the rules of other configuration settings
949 where the local administrator's configuration in /etc always
950 overrides any other settings.
952 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
953 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
954 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
955 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
956 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
957 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
958 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
959 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
960 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
961 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
962 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
963 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
964 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
965 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
966 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
967 Lindskog, WaLyong Cho, Will Woods, Zbigniew
970 -- Beijing, 2014-05-28
974 * When restoring the screen brightness at boot, stay away from
975 the darkest setting or from the lowest 5% of the available
976 range, depending on which is the larger value of both. This
977 should effectively protect the user from rebooting into a
978 black screen, should the brightness have been set to minimum
981 * sd-login gained a new sd_machine_get_class() call to
982 determine the class ("vm" or "container") of a machine
983 registered with machined.
985 * sd-login gained new calls
986 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
987 to query the identity of the peer of a local AF_UNIX
988 connection. They operate similarly to their sd_pid_get_xyz()
991 * PID 1 will now maintain a system-wide system state engine
992 with the states "starting", "running", "degraded",
993 "maintenance", "stopping". These states are bound to system
994 startup, normal runtime, runtime with at least one failed
995 service, rescue/emergency mode and system shutdown. This
996 state is shown in the "systemctl status" output when no unit
997 name is passed. It is useful to determine system state, in
998 particularly when doing so for many systems or containers at
1001 * A new command "list-machines" has been added to "systemctl"
1002 that lists all local OS containers and shows their system
1003 state (see above), if systemd runs inside of them.
1005 * systemctl gained a new "-r" switch to recursively enumerate
1006 units on all local containers, when used with the
1007 "list-unit" command (which is the default one that is
1008 executed when no parameters are specified).
1010 * The GPT automatic partition discovery logic will now honour
1011 two GPT partition flags: one may be set on a partition to
1012 cause it to be mounted read-only, and the other may be set
1013 on a partition to ignore it during automatic discovery.
1015 * Two new GPT type UUIDs have been added for automatic root
1016 partition discovery, for 32-bit and 64-bit ARM. This is not
1017 particularly useful for discovering the root directory on
1018 these architectures during bare-metal boots (since UEFI is
1019 not common there), but still very useful to allow booting of
1020 ARM disk images in nspawn with the -i option.
1022 * MAC addresses of interfaces created with nspawn's
1023 --network-interface= switch will now be generated from the
1024 machine name, and thus be stable between multiple invocations
1027 * logind will now automatically remove all IPC objects owned
1028 by a user if she or he fully logs out. This makes sure that
1029 users who are logged out cannot continue to consume IPC
1030 resources. This covers SysV memory, semaphores and message
1031 queues as well as POSIX shared memory and message
1032 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1033 limits. With this functionality, that is corrected. This may
1034 be turned off by using the RemoveIPC= switch of logind.conf.
1036 * The systemd-machine-id-setup and tmpfiles tools gained a
1037 --root= switch to operate on a specific root directory,
1040 * journald can now forward logged messages to the TTYs of all
1041 logged in users ("wall"). This is the default for all
1042 emergency messages now.
1044 * A new tool systemd-journal-remote has been added to stream
1045 journal log messages across the network.
1047 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1048 controller trees are mounted into it. Note that the
1049 directories mounted beneath it are not read-only. This is a
1050 security measure and is particularly useful because glibc
1051 actually includes a search logic to pick any tmpfs it can
1052 find to implement shm_open() if /dev/shm is not available
1053 (which it might very well be in namespaced setups).
1055 * machinectl gained a new "poweroff" command to cleanly power
1056 down a local OS container.
1058 * The PrivateDevices= unit file setting will now also drop the
1059 CAP_MKNOD capability from the capability bound set, and
1060 imply DevicePolicy=closed.
1062 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1063 comprehensively on all long-running systemd services where
1064 this is appropriate.
1066 * systemd-udevd will now run in a disassociated mount
1067 namespace. To mount directories from udev rules, make sure to
1068 pull in mount units via SYSTEMD_WANTS properties.
1070 * The kdbus support gained support for uploading policy into
1071 the kernel. sd-bus gained support for creating "monitoring"
1072 connections that can eavesdrop into all bus communication
1073 for debugging purposes.
1075 * Timestamps may now be specified in seconds since the UNIX
1076 epoch Jan 1st, 1970 by specifying "@" followed by the value
1079 * Native tcpwrap support in systemd has been removed. tcpwrap
1080 is old code, not really maintained anymore and has serious
1081 shortcomings, and better options such as firewalls
1082 exist. For setups that require tcpwrap usage, please
1083 consider invoking your socket-activated service via tcpd,
1084 like on traditional inetd.
1086 * A new system.conf configuration option
1087 DefaultTimerAccuracySec= has been added that controls the
1088 default AccuracySec= setting of .timer units.
1090 * Timer units gained a new WakeSystem= switch. If enabled,
1091 timers configured this way will cause the system to resume
1092 from system suspend (if the system supports that, which most
1095 * Timer units gained a new Persistent= switch. If enabled,
1096 timers configured this way will save to disk when they have
1097 been last triggered. This information is then used on next
1098 reboot to possible execute overdue timer events, that
1099 could not take place because the system was powered off.
1100 This enables simple anacron-like behaviour for timer units.
1102 * systemctl's "list-timers" will now also list the time a
1103 timer unit was last triggered in addition to the next time
1104 it will be triggered.
1106 * systemd-networkd will now assign predictable IPv4LL
1107 addresses to its local interfaces.
1109 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1110 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1111 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1112 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1113 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1114 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1115 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1116 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1119 -- Berlin, 2014-03-25
1123 * A new unit file setting RestrictAddressFamilies= has been
1124 added to restrict which socket address families unit
1125 processes gain access to. This takes address family names
1126 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1127 attack surface of services via exotic protocol stacks. This
1128 is built on seccomp system call filters.
1130 * Two new unit file settings RuntimeDirectory= and
1131 RuntimeDirectoryMode= have been added that may be used to
1132 manage a per-daemon runtime directories below /run. This is
1133 an alternative for setting up directory permissions with
1134 tmpfiles snippets, and has the advantage that the runtime
1135 directory's lifetime is bound to the daemon runtime and that
1136 the daemon starts up with an empty directory each time. This
1137 is particularly useful when writing services that drop
1138 privileges using the User= or Group= setting.
1140 * The DeviceAllow= unit setting now supports globbing for
1141 matching against device group names.
1143 * The systemd configuration file system.conf gained new
1144 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1145 DefaultMemoryAccounting= to globally turn on/off accounting
1146 for specific resources (cgroups) for all units. These
1147 settings may still be overridden individually in each unit
1150 * systemd-gpt-auto-generator is now able to discover /srv and
1151 root partitions in addition to /home and swap partitions. It
1152 also supports LUKS-encrypted partitions now. With this in
1153 place, automatic discovery of partitions to mount following
1154 the Discoverable Partitions Specification
1155 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1156 is now a lot more complete. This allows booting without
1157 /etc/fstab and without root= on the kernel command line on
1158 systems prepared appropriately.
1160 * systemd-nspawn gained a new --image= switch which allows
1161 booting up disk images and Linux installations on any block
1162 device that follow the Discoverable Partitions Specification
1163 (see above). This means that installations made with
1164 appropriately updated installers may now be started and
1165 deployed using container managers, completely
1166 unmodified. (We hope that libvirt-lxc will add support for
1167 this feature soon, too.)
1169 * systemd-nspawn gained a new --network-macvlan= setting to
1170 set up a private macvlan interface for the
1171 container. Similarly, systemd-networkd gained a new
1172 Kind=macvlan setting in .netdev files.
1174 * systemd-networkd now supports configuring local addresses
1177 * A new tool systemd-network-wait-online has been added to
1178 synchronously wait for network connectivity using
1181 * The sd-bus.h bus API gained a new sd_bus_track object for
1182 tracking the life-cycle of bus peers. Note that sd-bus.h is
1183 still not a public API though (unless you specify
1184 --enable-kdbus on the configure command line, which however
1185 voids your warranty and you get no API stability guarantee).
1187 * The $XDG_RUNTIME_DIR runtime directories for each user are
1188 now individual tmpfs instances, which has the benefit of
1189 introducing separate pools for each user, with individual
1190 size limits, and thus making sure that unprivileged clients
1191 can no longer negatively impact the system or other users by
1192 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1193 RuntimeDirectorySize= has been introduced that allows
1194 controlling the default size limit for all users. It
1195 defaults to 10% of the available physical memory. This is no
1196 replacement for quotas on tmpfs though (which the kernel
1197 still does not support), as /dev/shm and /tmp are still
1198 shared resources used by both the system and unprivileged
1201 * logind will now automatically turn off automatic suspending
1202 on laptop lid close when more than one display is
1203 connected. This was previously expected to be implemented
1204 individually in desktop environments (such as GNOME),
1205 however has been added to logind now, in order to fix a
1206 boot-time race where a desktop environment might not have
1207 been started yet and thus not been able to take an inhibitor
1208 lock at the time where logind already suspends the system
1209 due to a closed lid.
1211 * logind will now wait at least 30s after each system
1212 suspend/resume cycle, and 3min after system boot before
1213 suspending the system due to a closed laptop lid. This
1214 should give USB docking stations and similar enough time to
1215 be probed and configured after system resume and boot in
1216 order to then act as suspend blocker.
1218 * systemd-run gained a new --property= setting which allows
1219 initialization of resource control properties (and others)
1220 for the created scope or service unit. Example: "systemd-run
1221 --property=BlockIOWeight=10 updatedb" may be used to run
1222 updatedb at a low block IO scheduling weight.
1224 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1225 now also work in --scope mode.
1227 * When systemd is compiled with kdbus support, basic support
1228 for enforced policies is now in place. (Note that enabling
1229 kdbus still voids your warranty and no API compatibility
1232 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1233 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1234 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1235 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1236 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1237 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1238 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1239 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1240 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1241 Zbigniew Jędrzejewski-Szmek
1243 -- Berlin, 2014-03-12
1247 * systemd will now relabel /dev after loading the SMACK policy
1248 according to SMACK rules.
1250 * A new unit file option AppArmorProfile= has been added to
1251 set the AppArmor profile for the processes of a unit.
1253 * A new condition check ConditionArchitecture= has been added
1254 to conditionalize units based on the system architecture, as
1255 reported by uname()'s "machine" field.
1257 * systemd-networkd now supports matching on the system
1258 virtualization, architecture, kernel command line, host name
1261 * logind is now a lot more aggressive when suspending the
1262 machine due to a closed laptop lid. Instead of acting only
1263 on the lid close action, it will continuously watch the lid
1264 status and act on it. This is useful for laptops where the
1265 power button is on the outside of the chassis so that it can
1266 be reached without opening the lid (such as the Lenovo
1267 Yoga). On those machines, logind will now immediately
1268 re-suspend the machine if the power button has been
1269 accidentally pressed while the laptop was suspended and in a
1270 backpack or similar.
1272 * logind will now watch SW_DOCK switches and inhibit reaction
1273 to the lid switch if it is pressed. This means that logind
1274 will not suspend the machine anymore if the lid is closed
1275 and the system is docked, if the laptop supports SW_DOCK
1276 notifications via the input layer. Note that ACPI docking
1277 stations do not generate this currently. Also note that this
1278 logic is usually not fully sufficient and Desktop
1279 Environments should take a lid switch inhibitor lock when an
1280 external display is connected, as systemd will not watch
1283 * nspawn will now make use of the devices cgroup controller by
1284 default, and only permit creation of and access to the usual
1285 API device nodes like /dev/null or /dev/random, as well as
1286 access to (but not creation of) the pty devices.
1288 * We will now ship a default .network file for
1289 systemd-networkd that automatically configures DHCP for
1290 network interfaces created by nspawn's --network-veth or
1291 --network-bridge= switches.
1293 * systemd will now understand the usual M, K, G, T suffixes
1294 according to SI conventions (i.e. to the base 1000) when
1295 referring to throughput and hardware metrics. It will stay
1296 with IEC conventions (i.e. to the base 1024) for software
1297 metrics, according to what is customary according to
1298 Wikipedia. We explicitly document which base applies for
1299 each configuration option.
1301 * The DeviceAllow= setting in unit files now supports a syntax
1302 to whitelist an entire group of devices node majors at once,
1303 based on the /proc/devices listing. For example, with the
1304 string "char-pts", it is now possible to whitelist all
1305 current and future pseudo-TTYs at once.
1307 * sd-event learned a new "post" event source. Event sources of
1308 this type are triggered by the dispatching of any event
1309 source of a type that is not "post". This is useful for
1310 implementing clean-up and check event sources that are
1311 triggered by other work being done in the program.
1313 * systemd-networkd is no longer statically enabled, but uses
1314 the usual [Install] sections so that it can be
1315 enabled/disabled using systemctl. It still is enabled by
1318 * When creating a veth interface pair with systemd-nspawn, the
1319 host side will now be prefixed with "vb-" if
1320 --network-bridge= is used, and with "ve-" if --network-veth
1321 is used. This way, it is easy to distinguish these cases on
1322 the host, for example to apply different configuration to
1323 them with systemd-networkd.
1325 * The compatibility libraries for libsystemd-journal.so,
1326 libsystem-id128.so, libsystemd-login.so and
1327 libsystemd-daemon.so do not make use of IFUNC
1328 anymore. Instead, we now build libsystemd.so multiple times
1329 under these alternative names. This means that the footprint
1330 is drastically increased, but given that these are
1331 transitional compatibility libraries, this should not matter
1332 much. This change has been made necessary to support the ARM
1333 platform for these compatibility libraries, as the ARM
1334 toolchain is not really at the same level as the toolchain
1335 for other architectures like x86 and does not support
1336 IFUNC. Please make sure to use --enable-compat-libs only
1337 during a transitional period!
1339 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1340 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1341 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1342 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1343 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1344 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1345 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1346 Zbigniew Jędrzejewski-Szmek
1348 -- Berlin, 2014-02-24
1352 * A new component "systemd-networkd" has been added that can
1353 be used to configure local network interfaces statically or
1354 via DHCP. It is capable of bringing up bridges, VLANs, and
1355 bonding. Currently, no hook-ups for interactive network
1356 configuration are provided. Use this for your initrd,
1357 container, embedded, or server setup if you need a simple,
1358 yet powerful, network configuration solution. This
1359 configuration subsystem is quite nifty, as it allows wildcard
1360 hotplug matching in interfaces. For example, with a single
1361 configuration snippet, you can configure that all Ethernet
1362 interfaces showing up are automatically added to a bridge,
1363 or similar. It supports link-sensing and more.
1365 * A new tool "systemd-socket-proxyd" has been added which can
1366 act as a bidirectional proxy for TCP sockets. This is
1367 useful for adding socket activation support to services that
1368 do not actually support socket activation, including virtual
1369 machines and the like.
1371 * Add a new tool to save/restore rfkill state on
1374 * Save/restore state of keyboard backlights in addition to
1375 display backlights on shutdown/boot.
1377 * udev learned a new SECLABEL{} construct to label device
1378 nodes with a specific security label when they appear. For
1379 now, only SECLABEL{selinux} is supported, but the syntax is
1380 prepared for additional security frameworks.
1382 * udev gained a new scheme to configure link-level attributes
1383 from files in /etc/systemd/network/*.link. These files can
1384 match against MAC address, device path, driver name and type,
1385 and will apply attributes like the naming policy, link speed,
1386 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1387 address assignment policy (randomized, ...).
1389 * The configuration of network interface naming rules for
1390 "permanent interface names" has changed: a new NamePolicy=
1391 setting in the [Link] section of .link files determines the
1392 priority of possible naming schemes (onboard, slot, mac,
1393 path). The default value of this setting is determined by
1394 /usr/lib/net/links/99-default.link. Old
1395 80-net-name-slot.rules udev configuration file has been
1396 removed, so local configuration overriding this file should
1397 be adapated to override 99-default.link instead.
1399 * When the User= switch is used in a unit file, also
1400 initialize $SHELL= based on the user database entry.
1402 * systemd no longer depends on libdbus. All communication is
1403 now done with sd-bus, systemd's low-level bus library
1406 * kdbus support has been added to PID 1 itself. When kdbus is
1407 enabled, this causes PID 1 to set up the system bus and
1408 enable support for a new ".busname" unit type that
1409 encapsulates bus name activation on kdbus. It works a little
1410 bit like ".socket" units, except for bus names. A new
1411 generator has been added that converts classic dbus1 service
1412 activation files automatically into native systemd .busname
1415 * sd-bus: add a light-weight vtable implementation that allows
1416 defining objects on the bus with a simple static const
1417 vtable array of its methods, signals and properties.
1419 * systemd will not generate or install static dbus
1420 introspection data anymore to /usr/share/dbus-1/interfaces,
1421 as the precise format of these files is unclear, and
1422 nothing makes use of it.
1424 * A proxy daemon is now provided to proxy clients connecting
1425 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1426 compatibility with classic D-Bus.
1428 * A bus driver implementation has been added that supports the
1429 classic D-Bus bus driver calls on kdbus, also for
1430 compatibility purposes.
1432 * A new API "sd-event.h" has been added that implements a
1433 minimal event loop API built around epoll. It provides a
1434 couple of features that direct epoll usage is lacking:
1435 prioritization of events, scales to large numbers of timer
1436 events, per-event timer slack (accuracy), system-wide
1437 coalescing of timer events, exit handlers, watchdog
1438 supervision support using systemd's sd_notify() API, child
1441 * A new API "sd-rntl.h" has been added that provides an API
1442 around the route netlink interface of the kernel, similar in
1443 style to "sd-bus.h".
1445 * A new API "sd-dhcp-client.h" has been added that provides a
1446 small DHCPv4 client-side implementation. This is used by
1449 * There is a new kernel command line option
1450 "systemd.restore_state=0|1". When set to "0", none of the
1451 systemd tools will restore saved runtime state to hardware
1452 devices. More specifically, the rfkill and backlight states
1455 * The FsckPassNo= compatibility option in mount/service units
1456 has been removed. The fstab generator will now add the
1457 necessary dependencies automatically, and does not require
1458 PID1's support for that anymore.
1460 * journalctl gained a new switch, --list-boots, that lists
1461 recent boots with their times and boot IDs.
1463 * The various tools like systemctl, loginctl, timedatectl,
1464 busctl, systemd-run, ... have gained a new switch "-M" to
1465 connect to a specific, local OS container (as direct
1466 connection, without requiring SSH). This works on any
1467 container that is registered with machined, such as those
1468 created by libvirt-lxc or nspawn.
1470 * systemd-run and systemd-analyze also gained support for "-H"
1471 to connect to remote hosts via SSH. This is particularly
1472 useful for systemd-run because it enables queuing of jobs
1473 onto remote systems.
1475 * machinectl gained a new command "login" to open a getty
1476 login in any local container. This works with any container
1477 that is registered with machined (such as those created by
1478 libvirt-lxc or nspawn), and which runs systemd inside.
1480 * machinectl gained a new "reboot" command that may be used to
1481 trigger a reboot on a specific container that is registered
1482 with machined. This works on any container that runs an init
1483 system of some kind.
1485 * systemctl gained a new "list-timers" command to print a nice
1486 listing of installed timer units with the times they elapse
1489 * Alternative reboot() parameters may now be specified on the
1490 "systemctl reboot" command line and are passed to the
1491 reboot() system call.
1493 * systemctl gained a new --job-mode= switch to configure the
1494 mode to queue a job with. This is a more generic version of
1495 --fail, --irreversible, and --ignore-dependencies, which are
1496 still available but not advertised anymore.
1498 * /etc/systemd/system.conf gained new settings to configure
1499 various default timeouts of units, as well as the default
1500 start limit interval and burst. These may still be overridden
1503 * PID1 will now export on the bus profile data of the security
1504 policy upload process (such as the SELinux policy upload to
1507 * journald: when forwarding logs to the console, include
1508 timestamps (following the setting in
1509 /sys/module/printk/parameters/time).
1511 * OnCalendar= in timer units now understands the special
1512 strings "yearly" and "annually". (Both are equivalent)
1514 * The accuracy of timer units is now configurable with the new
1515 AccuracySec= setting. It defaults to 1min.
1517 * A new dependency type JoinsNamespaceOf= has been added that
1518 allows running two services within the same /tmp and network
1519 namespace, if PrivateNetwork= or PrivateTmp= are used.
1521 * A new command "cat" has been added to systemctl. It outputs
1522 the original unit file of a unit, and concatenates the
1523 contents of additional "drop-in" unit file snippets, so that
1524 the full configuration is shown.
1526 * systemctl now supports globbing on the various "list-xyz"
1527 commands, like "list-units" or "list-sockets", as well as on
1528 those commands which take multiple unit names.
1530 * journalctl's --unit= switch gained support for globbing.
1532 * All systemd daemons now make use of the watchdog logic so
1533 that systemd automatically notices when they hang.
1535 * If the $container_ttys environment variable is set,
1536 getty-generator will automatically spawn a getty for each
1537 listed tty. This is useful for container managers to request
1538 login gettys to be spawned on as many ttys as needed.
1540 * %h, %s, %U specifier support is not available anymore when
1541 used in unit files for PID 1. This is because NSS calls are
1542 not safe from PID 1. They stay available for --user
1543 instances of systemd, and as special case for the root user.
1545 * loginctl gained a new "--no-legend" switch to turn off output
1548 * The "sd-login.h" API gained three new calls:
1549 sd_session_is_remote(), sd_session_get_remote_user(),
1550 sd_session_get_remote_host() to query information about
1553 * The udev hardware database now also carries vendor/product
1554 information of SDIO devices.
1556 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1557 determine whether watchdog notifications are requested by
1560 * Socket-activated per-connection services now include a
1561 short description of the connection parameters in the
1564 * tmpfiles gained a new "--boot" option. When this is not used,
1565 only lines where the command character is not suffixed with
1566 "!" are executed. When this option is specified, those
1567 options are executed too. This partitions tmpfiles
1568 directives into those that can be safely executed at any
1569 time, and those which should be run only at boot (for
1570 example, a line that creates /run/nologin).
1572 * A new API "sd-resolve.h" has been added which provides a simple
1573 asynchronous wrapper around glibc NSS host name resolution
1574 calls, such as getaddrinfo(). In contrast to glibc's
1575 getaddrinfo_a(), it does not use signals. In contrast to most
1576 other asynchronous name resolution libraries, this one does
1577 not reimplement DNS, but reuses NSS, so that alternate
1578 host name resolution systems continue to work, such as mDNS,
1579 LDAP, etc. This API is based on libasyncns, but it has been
1580 cleaned up for inclusion in systemd.
1582 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1583 "sd-daemon.h" are no longer found in individual libraries
1584 libsystemd-journal.so, libsystemd-login.so,
1585 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1586 merged them into a single library, libsystemd.so, which
1587 provides all symbols. The reason for this is cyclic
1588 dependencies, as these libraries tend to use each other's
1589 symbols. So far, we have managed to workaround that by linking
1590 a copy of a good part of our code into each of these
1591 libraries again and again, which, however, makes certain
1592 things hard to do, like sharing static variables. Also, it
1593 substantially increases footprint. With this change, there
1594 is only one library for the basic APIs systemd
1595 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1596 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1597 library as well, however are subject to the --enable-kdbus
1598 switch (see below). Note that "sd-dhcp-client.h" is not part
1599 of this library (this is because it only consumes, never
1600 provides, services of/to other APIs). To make the transition
1601 easy from the separate libraries to the unified one, we
1602 provide the --enable-compat-libs compile-time switch which
1603 will generate stub libraries that are compatible with the
1604 old ones but redirect all calls to the new one.
1606 * All of the kdbus logic and the new APIs "sd-bus.h",
1607 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1608 and "sd-utf8.h" are compile-time optional via the
1609 "--enable-kdbus" switch, and they are not compiled in by
1610 default. To make use of kdbus, you have to explicitly enable
1611 the switch. Note however, that neither the kernel nor the
1612 userspace API for all of this is considered stable yet. We
1613 want to maintain the freedom to still change the APIs for
1614 now. By specifying this build-time switch, you acknowledge
1615 that you are aware of the instability of the current
1618 * Also, note that while kdbus is pretty much complete,
1619 it lacks one thing: proper policy support. This means you
1620 can build a fully working system with all features; however,
1621 it will be highly insecure. Policy support will be added in
1622 one of the next releases, at the same time that we will
1623 declare the APIs stable.
1625 * When the kernel command-line argument "kdbus" is specified,
1626 systemd will automatically load the kdbus.ko kernel module. At
1627 this stage of development, it is only useful for testing kdbus
1628 and should not be used in production. Note: if "--enable-kdbus"
1629 is specified, and the kdbus.ko kernel module is available, and
1630 "kdbus" is added to the kernel command line, the entire system
1631 runs with kdbus instead of dbus-daemon, with the above mentioned
1632 problem of missing the system policy enforcement. Also a future
1633 version of kdbus.ko or a newer systemd will not be compatible with
1634 each other, and will unlikely be able to boot the machine if only
1635 one of them is updated.
1637 * systemctl gained a new "import-environment" command which
1638 uploads the caller's environment (or parts thereof) into the
1639 service manager so that it is inherited by services started
1640 by the manager. This is useful to upload variables like
1641 $DISPLAY into the user service manager.
1643 * A new PrivateDevices= switch has been added to service units
1644 which allows running a service with a namespaced /dev
1645 directory that does not contain any device nodes for
1646 physical devices. More specifically, it only includes devices
1647 such as /dev/null, /dev/urandom, and /dev/zero which are API
1650 * logind has been extended to support behaviour like VT
1651 switching on seats that do not support a VT. This makes
1652 multi-session available on seats that are not the first seat
1653 (seat0), and on systems where kernel support for VTs has
1654 been disabled at compile-time.
1656 * If a process holds a delay lock for system sleep or shutdown
1657 and fails to release it in time, we will now log its
1658 identity. This makes it easier to identify processes that
1659 cause slow suspends or power-offs.
1661 * When parsing /etc/crypttab, support for a new key-slot=
1662 option as supported by Debian is added. It allows indicating
1663 which LUKS slot to use on disk, speeding up key loading.
1665 * The sd_journald_sendv() API call has been checked and
1666 officially declared to be async-signal-safe so that it may
1667 be invoked from signal handlers for logging purposes.
1669 * Boot-time status output is now enabled automatically after a
1670 short timeout if boot does not progress, in order to give
1671 the user an indication what she or he is waiting for.
1673 * The boot-time output has been improved to show how much time
1674 remains until jobs expire.
1676 * The KillMode= switch in service units gained a new possible
1677 value "mixed". If set, and the unit is shut down, then the
1678 initial SIGTERM signal is sent only to the main daemon
1679 process, while the following SIGKILL signal is sent to
1680 all remaining processes of the service.
1682 * When a scope unit is registered, a new property "Controller"
1683 may be set. If set to a valid bus name, systemd will send a
1684 RequestStop() signal to this name when it would like to shut
1685 down the scope. This may be used to hook manager logic into
1686 the shutdown logic of scope units. Also, scope units may now
1687 be put in a special "abandoned" state, in which case the
1688 manager process which created them takes no further
1689 responsibilities for it.
1691 * When reading unit files, systemd will now verify
1692 the access mode of these files, and warn about certain
1693 suspicious combinations. This has been added to make it
1694 easier to track down packaging bugs where unit files are
1695 marked executable or world-writable.
1697 * systemd-nspawn gained a new "--setenv=" switch to set
1698 container-wide environment variables. The similar option in
1699 systemd-activate was renamed from "--environment=" to
1700 "--setenv=" for consistency.
1702 * systemd-nspawn has been updated to create a new kdbus domain
1703 for each container that is invoked, thus allowing each
1704 container to have its own set of system and user buses,
1705 independent of the host.
1707 * systemd-nspawn gained a new --drop-capability= switch to run
1708 the container with less capabilities than the default. Both
1709 --drop-capability= and --capability= now take the special
1710 string "all" for dropping or keeping all capabilities.
1712 * systemd-nspawn gained new switches for executing containers
1713 with specific SELinux labels set.
1715 * systemd-nspawn gained a new --quiet switch to not generate
1716 any additional output but the container's own console
1719 * systemd-nspawn gained a new --share-system switch to run a
1720 container without PID namespacing enabled.
1722 * systemd-nspawn gained a new --register= switch to control
1723 whether the container is registered with systemd-machined or
1724 not. This is useful for containers that do not run full
1725 OS images, but only specific apps.
1727 * systemd-nspawn gained a new --keep-unit which may be used
1728 when invoked as the only program from a service unit, and
1729 results in registration of the unit service itself in
1730 systemd-machined, instead of a newly opened scope unit.
1732 * systemd-nspawn gained a new --network-interface= switch for
1733 moving arbitrary interfaces to the container. The new
1734 --network-veth switch creates a virtual Ethernet connection
1735 between host and container. The new --network-bridge=
1736 switch then allows assigning the host side of this virtual
1737 Ethernet connection to a bridge device.
1739 * systemd-nspawn gained a new --personality= switch for
1740 setting the kernel personality for the container. This is
1741 useful when running a 32-bit container on a 64-bit host. A
1742 similar option Personality= is now also available for service
1745 * logind will now also track a "Desktop" identifier for each
1746 session which encodes the desktop environment of it. This is
1747 useful for desktop environments that want to identify
1748 multiple running sessions of itself easily.
1750 * A new SELinuxContext= setting for service units has been
1751 added that allows setting a specific SELinux execution
1752 context for a service.
1754 * Most systemd client tools will now honour $SYSTEMD_LESS for
1755 settings of the "less" pager. By default, these tools will
1756 override $LESS to allow certain operations to work, such as
1757 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1758 influence this logic.
1760 * systemd's "seccomp" hook-up has been changed to make use of
1761 the libseccomp library instead of using its own
1762 implementation. This has benefits for portability among
1765 * For usage together with SystemCallFilter=, a new
1766 SystemCallErrorNumber= setting has been introduced that
1767 allows configuration of a system error number to be returned
1768 on filtered system calls, instead of immediately killing the
1769 process. Also, SystemCallArchitectures= has been added to
1770 limit access to system calls of a particular architecture
1771 (in order to turn off support for unused secondary
1772 architectures). There is also a global
1773 SystemCallArchitectures= setting in system.conf now to turn
1774 off support for non-native system calls system-wide.
1776 * systemd requires a kernel with a working name_to_handle_at(),
1777 please see the kernel config requirements in the README file.
1779 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1780 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1781 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1782 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1783 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1784 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1785 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1786 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1787 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1788 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1789 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1790 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1791 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1792 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1793 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1794 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1795 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1796 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1797 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1798 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1799 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1800 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1801 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1802 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1804 -- Berlin, 2014-02-20
1808 * logind has gained support for facilitating privileged input
1809 and drm device access for unprivileged clients. This work is
1810 useful to allow Wayland display servers (and similar
1811 programs, such as kmscon) to run under the user's ID and
1812 access input and drm devices which are normally
1813 protected. When this is used (and the kernel is new enough)
1814 logind will "mute" IO on the file descriptors passed to
1815 Wayland as long as it is in the background and "unmute" it
1816 if it returns into the foreground. This allows secure
1817 session switching without allowing background sessions to
1818 eavesdrop on input and display data. This also introduces
1819 session switching support if VT support is turned off in the
1820 kernel, and on seats that are not seat0.
1822 * A new kernel command line option luks.options= is understood
1823 now which allows specifying LUKS options for usage for LUKS
1824 encrypted partitions specified with luks.uuid=.
1826 * tmpfiles.d(5) snippets may now use specifier expansion in
1827 path names. More specifically %m, %b, %H, %v, are now
1828 replaced by the local machine id, boot id, hostname, and
1829 kernel version number.
1831 * A new tmpfiles.d(5) command "m" has been introduced which
1832 may be used to change the owner/group/access mode of a file
1833 or directory if it exists, but do nothing if it does not.
1835 * This release removes high-level support for the
1836 MemorySoftLimit= cgroup setting. The underlying kernel
1837 cgroup attribute memory.soft_limit= is currently badly
1838 designed and likely to be removed from the kernel API in its
1839 current form, hence we should not expose it for now.
1841 * The memory.use_hierarchy cgroup attribute is now enabled for
1842 all cgroups systemd creates in the memory cgroup
1843 hierarchy. This option is likely to be come the built-in
1844 default in the kernel anyway, and the non-hierarchial mode
1845 never made much sense in the intrinsically hierarchial
1848 * A new field _SYSTEMD_SLICE= is logged along with all journal
1849 messages containing the slice a message was generated
1850 from. This is useful to allow easy per-customer filtering of
1851 logs among other things.
1853 * systemd-journald will no longer adjust the group of journal
1854 files it creates to the "systemd-journal" group. Instead we
1855 rely on the journal directory to be owned by the
1856 "systemd-journal" group, and its setgid bit set, so that the
1857 kernel file system layer will automatically enforce that
1858 journal files inherit this group assignment. The reason for
1859 this change is that we cannot allow NSS look-ups from
1860 journald which would be necessary to resolve
1861 "systemd-journal" to a numeric GID, because this might
1862 create deadlocks if NSS involves synchronous queries to
1863 other daemons (such as nscd, or sssd) which in turn are
1864 logging clients of journald and might block on it, which
1865 would then dead lock. A tmpfiles.d(5) snippet included in
1866 systemd will make sure the setgid bit and group are
1867 properly set on the journal directory if it exists on every
1868 boot. However, we recommend adjusting it manually after
1869 upgrades too (or from RPM scriptlets), so that the change is
1870 not delayed until next reboot.
1872 * Backlight and random seed files in /var/lib/ have moved into
1873 the /var/lib/systemd/ directory, in order to centralize all
1874 systemd generated files in one directory.
1876 * Boot time performance measurements (as displayed by
1877 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1878 performance information if that's available to determine how
1879 much time BIOS and boot loader initialization required. With
1880 a sufficiently new BIOS you hence no longer need to boot
1881 with Gummiboot to get access to such information.
1883 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1884 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1885 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1886 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1887 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1888 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1889 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1891 -- Berlin, 2013-10-02
1895 * The Restart= option for services now understands a new
1896 on-watchdog setting, which will restart the service
1897 automatically if the service stops sending out watchdog keep
1898 alive messages (as configured with WatchdogSec=).
1900 * The getty generator (which is responsible for bringing up a
1901 getty on configured serial consoles) will no longer only
1902 start a getty on the primary kernel console but on all
1903 others, too. This makes the order in which console= is
1904 specified on the kernel command line less important.
1906 * libsystemd-logind gained a new sd_session_get_vt() call to
1907 retrieve the VT number of a session.
1909 * If the option "tries=0" is set for an entry of /etc/crypttab
1910 its passphrase is queried indefinitely instead of any
1911 maximum number of tries.
1913 * If a service with a configure PID file terminates its PID
1914 file will now be removed automatically if it still exists
1915 afterwards. This should put an end to stale PID files.
1917 * systemd-run will now also take relative binary path names
1918 for execution and no longer insists on absolute paths.
1920 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1921 paths that are optionally prefixed with "-" to indicate that
1922 it should not be considered a failure if they do not exist.
1924 * journalctl -o (and similar commands) now understands a new
1925 output mode "short-precise", it is similar to "short" but
1926 shows timestamps with usec accuracy.
1928 * The option "discard" (as known from Debian) is now
1929 synonymous to "allow-discards" in /etc/crypttab. In fact,
1930 "discard" is preferred now (since it is easier to remember
1933 * Some licensing clean-ups were made, so that more code is now
1934 LGPL-2.1 licensed than before.
1936 * A minimal tool to save/restore the display backlight
1937 brightness across reboots has been added. It will store the
1938 backlight setting as late as possible at shutdown, and
1939 restore it as early as possible during reboot.
1941 * A logic to automatically discover and enable home and swap
1942 partitions on GPT disks has been added. With this in place
1943 /etc/fstab becomes optional for many setups as systemd can
1944 discover certain partitions located on the root disk
1945 automatically. Home partitions are recognized under their
1946 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1947 partitions are recognized under their GPT type ID
1948 0657fd6da4ab43c484e50933c84b4f4f.
1950 * systemd will no longer pass any environment from the kernel
1951 or initrd to system services. If you want to set an
1952 environment for all services, do so via the kernel command
1953 line systemd.setenv= assignment.
1955 * The systemd-sysctl tool no longer natively reads the file
1956 /etc/sysctl.conf. If desired, the file should be symlinked
1957 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1958 legacy support by a symlink rather than built-in code, it
1959 also makes the otherwise hidden order of application of the
1960 different files visible. (Note that this partly reverts to a
1961 pre-198 application order of sysctl knobs!)
1963 * The "systemctl set-log-level" and "systemctl dump" commands
1964 have been moved to systemd-analyze.
1966 * systemd-run learned the new --remain-after-exit switch,
1967 which causes the scope unit not to be cleaned up
1968 automatically after the process terminated.
1970 * tmpfiles learned a new --exclude-prefix= switch to exclude
1971 certain paths from operation.
1973 * journald will now automatically flush all messages to disk
1974 as soon as a message of the log priorities CRIT, ALERT or
1977 Contributions from: Andrew Cook, Brandon Philips, Christian
1978 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1979 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1980 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1981 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1982 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1983 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1984 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1985 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1986 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1987 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1988 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1989 William Giokas, Zbigniew Jędrzejewski-Szmek
1991 -- Berlin, 2013-09-13
1995 * The documentation has been updated to cover the various new
1996 concepts introduced with 205.
1998 * Unit files now understand the new %v specifier which
1999 resolves to the kernel version string as returned by "uname
2002 * systemctl now supports filtering the unit list output by
2003 load state, active state and sub state, using the new
2006 * "systemctl status" will now show the results of the
2007 condition checks (like ConditionPathExists= and similar) of
2008 the last start attempts of the unit. They are also logged to
2011 * "journalctl -b" may now be used to look for boot output of a
2012 specific boot. Try "journalctl -b -1" for the previous boot,
2013 but the syntax is substantially more powerful.
2015 * "journalctl --show-cursor" has been added which prints the
2016 cursor string the last shown log line. This may then be used
2017 with the new "journalctl --after-cursor=" switch to continue
2018 browsing logs from that point on.
2020 * "journalctl --force" may now be used to force regeneration
2023 * Creation of "dead" device nodes has been moved from udev
2024 into kmod and tmpfiles. Previously, udev would read the kmod
2025 databases to pre-generate dead device nodes based on meta
2026 information contained in kernel modules, so that these would
2027 be auto-loaded on access rather then at boot. As this
2028 does not really have much to do with the exposing actual
2029 kernel devices to userspace this has always been slightly
2030 alien in the udev codebase. Following the new scheme kmod
2031 will now generate a runtime snippet for tmpfiles from the
2032 module meta information and it now is tmpfiles' job to the
2033 create the nodes. This also allows overriding access and
2034 other parameters for the nodes using the usual tmpfiles
2035 facilities. As side effect this allows us to remove the
2036 CAP_SYS_MKNOD capability bit from udevd entirely.
2038 * logind's device ACLs may now be applied to these "dead"
2039 devices nodes too, thus finally allowing managed access to
2040 devices such as /dev/snd/sequencer whithout loading the
2041 backing module right-away.
2043 * A new RPM macro has been added that may be used to apply
2044 tmpfiles configuration during package installation.
2046 * systemd-detect-virt and ConditionVirtualization= now can
2047 detect User-Mode-Linux machines (UML).
2049 * journald will now implicitly log the effective capabilities
2050 set of processes in the message metadata.
2052 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2054 * The initrd interface has been simplified (more specifically,
2055 support for passing performance data via environment
2056 variables and fsck results via files in /run has been
2057 removed). These features were non-essential, and are
2058 nowadays available in a much nicer way by having systemd in
2059 the initrd serialize its state and have the hosts systemd
2060 deserialize it again.
2062 * The udev "keymap" data files and tools to apply keyboard
2063 specific mappings of scan to key codes, and force-release
2064 scan code lists have been entirely replaced by a udev
2065 "keyboard" builtin and a hwdb data file.
2067 * systemd will now honour the kernel's "quiet" command line
2068 argument also during late shutdown, resulting in a
2069 completely silent shutdown when used.
2071 * There's now an option to control the SO_REUSEPORT socket
2072 option in .socket units.
2074 * Instance units will now automatically get a per-template
2075 subslice of system.slice unless something else is explicitly
2076 configured. For example, instances of sshd@.service will now
2077 implicitly be placed in system-sshd.slice rather than
2078 system.slice as before.
2080 * Test coverage support may now be enabled at build time.
2082 Contributions from: Dave Reisner, Frederic Crozat, Harald
2083 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2084 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2085 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2086 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2087 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2088 Giokas, Zbigniew Jędrzejewski-Szmek
2090 -- Berlin, 2013-07-23
2094 * Two new unit types have been introduced:
2096 Scope units are very similar to service units, however, are
2097 created out of pre-existing processes -- instead of PID 1
2098 forking off the processes. By using scope units it is
2099 possible for system services and applications to group their
2100 own child processes (worker processes) in a powerful way
2101 which then maybe used to organize them, or kill them
2102 together, or apply resource limits on them.
2104 Slice units may be used to partition system resources in an
2105 hierarchial fashion and then assign other units to them. By
2106 default there are now three slices: system.slice (for all
2107 system services), user.slice (for all user sessions),
2108 machine.slice (for VMs and containers).
2110 Slices and scopes have been introduced primarily in
2111 context of the work to move cgroup handling to a
2112 single-writer scheme, where only PID 1
2113 creates/removes/manages cgroups.
2115 * There's a new concept of "transient" units. In contrast to
2116 normal units these units are created via an API at runtime,
2117 not from configuration from disk. More specifically this
2118 means it is now possible to run arbitrary programs as
2119 independent services, with all execution parameters passed
2120 in via bus APIs rather than read from disk. Transient units
2121 make systemd substantially more dynamic then it ever was,
2122 and useful as a general batch manager.
2124 * logind has been updated to make use of scope and slice units
2125 for managing user sessions. As a user logs in he will get
2126 his own private slice unit, to which all sessions are added
2127 as scope units. We also added support for automatically
2128 adding an instance of user@.service for the user into the
2129 slice. Effectively logind will no longer create cgroup
2130 hierarchies on its own now, it will defer entirely to PID 1
2131 for this by means of scope, service and slice units. Since
2132 user sessions this way become entities managed by PID 1
2133 the output of "systemctl" is now a lot more comprehensive.
2135 * A new mini-daemon "systemd-machined" has been added which
2136 may be used by virtualization managers to register local
2137 VMs/containers. nspawn has been updated accordingly, and
2138 libvirt will be updated shortly. machined will collect a bit
2139 of meta information about the VMs/containers, and assign
2140 them their own scope unit (see above). The collected
2141 meta-data is then made available via the "machinectl" tool,
2142 and exposed in "ps" and similar tools. machined/machinectl
2143 is compile-time optional.
2145 * As discussed earlier, the low-level cgroup configuration
2146 options ControlGroup=, ControlGroupModify=,
2147 ControlGroupPersistent=, ControlGroupAttribute= have been
2148 removed. Please use high-level attribute settings instead as
2149 well as slice units.
2151 * A new bus call SetUnitProperties() has been added to alter
2152 various runtime parameters of a unit. This is primarily
2153 useful to alter cgroup parameters dynamically in a nice way,
2154 but will be extended later on to make more properties
2155 modifiable at runtime. systemctl gained a new set-properties
2156 command that wraps this call.
2158 * A new tool "systemd-run" has been added which can be used to
2159 run arbitrary command lines as transient services or scopes,
2160 while configuring a number of settings via the command
2161 line. This tool is currently very basic, however already
2162 very useful. We plan to extend this tool to even allow
2163 queuing of execution jobs with time triggers from the
2164 command line, similar in fashion to "at".
2166 * nspawn will now inform the user explicitly that kernels with
2167 audit enabled break containers, and suggest the user to turn
2170 * Support for detecting the IMA and AppArmor security
2171 frameworks with ConditionSecurity= has been added.
2173 * journalctl gained a new "-k" switch for showing only kernel
2174 messages, mimicking dmesg output; in addition to "--user"
2175 and "--system" switches for showing only user's own logs
2178 * systemd-delta can now show information about drop-in
2179 snippets extending unit files.
2181 * libsystemd-bus has been substantially updated but is still
2182 not available as public API.
2184 * systemd will now look for the "debug" argument on the kernel
2185 command line and enable debug logging, similar to what
2186 "systemd.log_level=debug" already did before.
2188 * "systemctl set-default", "systemctl get-default" has been
2189 added to configure the default.target symlink, which
2190 controls what to boot into by default.
2192 * "systemctl set-log-level" has been added as a convenient
2193 way to raise and lower systemd logging threshold.
2195 * "systemd-analyze plot" will now show the time the various
2196 generators needed for execution, as well as information
2197 about the unit file loading.
2199 * libsystemd-journal gained a new sd_journal_open_files() call
2200 for opening specific journal files. journactl also gained a
2201 new switch to expose this new functionality. Previously we
2202 only supported opening all files from a directory, or all
2203 files from the system, as opening individual files only is
2204 racy due to journal file rotation.
2206 * systemd gained the new DefaultEnvironment= setting in
2207 /etc/systemd/system.conf to set environment variables for
2210 * If a privileged process logs a journal message with the
2211 OBJECT_PID= field set, then journald will automatically
2212 augment this with additional OBJECT_UID=, OBJECT_GID=,
2213 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2214 system services want to log events about specific client
2215 processes. journactl/systemctl has been updated to make use
2216 of this information if all log messages regarding a specific
2219 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2220 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2221 Reisner, David Coppa, David King, David Strauss, Eelco
2222 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2223 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2224 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2225 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2226 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2227 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2228 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2229 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2230 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2231 Łukasz Stelmach, 장동준
2235 * The Python bindings gained some minimal support for the APIs
2236 exposed by libsystemd-logind.
2238 * ConditionSecurity= gained support for detecting SMACK. Since
2239 this condition already supports SELinux and AppArmor we only
2240 miss IMA for this. Patches welcome!
2242 Contributions from: Karol Lewandowski, Lennart Poettering,
2243 Zbigniew Jędrzejewski-Szmek
2247 * systemd-nspawn will now create /etc/resolv.conf if
2248 necessary, before bind-mounting the host's file onto it.
2250 * systemd-nspawn will now store meta information about a
2251 container on the container's cgroup as extended attribute
2252 fields, including the root directory.
2254 * The cgroup hierarchy has been reworked in many ways. All
2255 objects any of the components systemd creates in the cgroup
2256 tree are now suffixed. More specifically, user sessions are
2257 now placed in cgroups suffixed with ".session", users in
2258 cgroups suffixed with ".user", and nspawn containers in
2259 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2260 names are now escaped in a simple scheme to avoid collision
2261 of userspace object names with kernel filenames. This work
2262 is preparation for making these objects relocatable in the
2263 cgroup tree, in order to allow easy resource partitioning of
2264 these objects without causing naming conflicts.
2266 * systemctl list-dependencies gained the new switches
2267 --plain, --reverse, --after and --before.
2269 * systemd-inhibit now shows the process name of processes that
2270 have taken an inhibitor lock.
2272 * nss-myhostname will now also resolve "localhost"
2273 implicitly. This makes /etc/hosts an optional file and
2274 nicely handles that on IPv6 ::1 maps to both "localhost" and
2277 * libsystemd-logind.so gained a new call
2278 sd_get_machine_names() to enumerate running containers and
2279 VMs (currently only supported by very new libvirt and
2280 nspawn). sd_login_monitor can now be used to watch
2281 VMs/containers coming and going.
2283 * .include is not allowed recursively anymore, and only in
2284 unit files. Usually it is better to use drop-in snippets in
2285 .d/*.conf anyway, as introduced with systemd 198.
2287 * systemd-analyze gained a new "critical-chain" command that
2288 determines the slowest chain of units run during system
2289 boot-up. It is very useful for tracking down where
2290 optimizing boot time is the most beneficial.
2292 * systemd will no longer allow manipulating service paths in
2293 the name=systemd:/system cgroup tree using ControlGroup= in
2294 units. (But is still fine with it in all other dirs.)
2296 * There's a new systemd-nspawn@.service service file that may
2297 be used to easily run nspawn containers as system
2298 services. With the container's root directory in
2299 /var/lib/container/foobar it is now sufficient to run
2300 "systemctl start systemd-nspawn@foobar.service" to boot it.
2302 * systemd-cgls gained a new parameter "--machine" to list only
2303 the processes within a certain container.
2305 * ConditionSecurity= now can check for "apparmor". We still
2306 are lacking checks for SMACK and IMA for this condition
2307 check though. Patches welcome!
2309 * A new configuration file /etc/systemd/sleep.conf has been
2310 added that may be used to configure which kernel operation
2311 systemd is supposed to execute when "suspend", "hibernate"
2312 or "hybrid-sleep" is requested. This makes the new kernel
2313 "freeze" state accessible to the user.
2315 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2316 the passed argument if applicable.
2318 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2319 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2320 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2321 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2322 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2323 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2324 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2329 * The output of 'systemctl list-jobs' got some polishing. The
2330 '--type=' argument may now be passed more than once. A new
2331 command 'systemctl list-sockets' has been added which shows
2332 a list of kernel sockets systemd is listening on with the
2333 socket units they belong to, plus the units these socket
2336 * The experimental libsystemd-bus library got substantial
2337 updates to work in conjunction with the (also experimental)
2338 kdbus kernel project. It works well enough to exchange
2339 messages with some sophistication. Note that kdbus is not
2340 ready yet, and the library is mostly an elaborate test case
2341 for now, and not installable.
2343 * systemd gained a new unit 'systemd-static-nodes.service'
2344 that generates static device nodes earlier during boot, and
2345 can run in conjunction with udev.
2347 * libsystemd-login gained a new call sd_pid_get_user_unit()
2348 to retrieve the user systemd unit a process is running
2349 in. This is useful for systems where systemd is used as
2352 * systemd-nspawn now places all containers in the new /machine
2353 top-level cgroup directory in the name=systemd
2354 hierarchy. libvirt will soon do the same, so that we get a
2355 uniform separation of /system, /user and /machine for system
2356 services, user processes and containers/virtual
2357 machines. This new cgroup hierarchy is also useful to stick
2358 stable names to specific container instances, which can be
2359 recognized later this way (this name may be controlled
2360 via systemd-nspawn's new -M switch). libsystemd-login also
2361 gained a new call sd_pid_get_machine_name() to retrieve the
2362 name of the container/VM a specific process belongs to.
2364 * bootchart can now store its data in the journal.
2366 * libsystemd-journal gained a new call
2367 sd_journal_add_conjunction() for AND expressions to the
2368 matching logic. This can be used to express more complex
2369 logical expressions.
2371 * journactl can now take multiple --unit= and --user-unit=
2374 * The cryptsetup logic now understands the "luks.key=" kernel
2375 command line switch for specifying a file to read the
2376 decryption key from. Also, if a configured key file is not
2377 found the tool will now automatically fall back to prompting
2380 * Python systemd.journal module was updated to wrap recently
2381 added functions from libsystemd-journal. The interface was
2382 changed to bring the low level interface in s.j._Reader
2383 closer to the C API, and the high level interface in
2384 s.j.Reader was updated to wrap and convert all data about
2387 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2388 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2389 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2390 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2391 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2392 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2396 * journalctl --update-catalog now understands a new --root=
2397 option to operate on catalogs found in a different root
2400 * During shutdown after systemd has terminated all running
2401 services a final killing loop kills all remaining left-over
2402 processes. We will now print the name of these processes
2403 when we send SIGKILL to them, since this usually indicates a
2406 * If /etc/crypttab refers to password files stored on
2407 configured mount points automatic dependencies will now be
2408 generated to ensure the specific mount is established first
2409 before the key file is attempted to be read.
2411 * 'systemctl status' will now show information about the
2412 network sockets a socket unit is listening on.
2414 * 'systemctl status' will also shown information about any
2415 drop-in configuration file for units. (Drop-In configuration
2416 files in this context are files such as
2417 /etc/systemd/systemd/foobar.service.d/*.conf)
2419 * systemd-cgtop now optionally shows summed up CPU times of
2420 cgroups. Press '%' while running cgtop to switch between
2421 percentage and absolute mode. This is useful to determine
2422 which cgroups use up the most CPU time over the entire
2423 runtime of the system. systemd-cgtop has also been updated
2424 to be 'pipeable' for processing with further shell tools.
2426 * 'hostnamectl set-hostname' will now allow setting of FQDN
2429 * The formatting and parsing of time span values has been
2430 changed. The parser now understands fractional expressions
2431 such as "5.5h". The formatter will now output fractional
2432 expressions for all time spans under 1min, i.e. "5.123456s"
2433 rather than "5s 123ms 456us". For time spans under 1s
2434 millisecond values are shown, for those under 1ms
2435 microsecond values are shown. This should greatly improve
2436 all time-related output of systemd.
2438 * libsystemd-login and libsystemd-journal gained new
2439 functions for querying the poll() events mask and poll()
2440 timeout value for integration into arbitrary event
2443 * localectl gained the ability to list available X11 keymaps
2444 (models, layouts, variants, options).
2446 * 'systemd-analyze dot' gained the ability to filter for
2447 specific units via shell-style globs, to create smaller,
2448 more useful graphs. I.e. it is now possible to create simple
2449 graphs of all the dependencies between only target units, or
2450 of all units that Avahi has dependencies with.
2452 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2453 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2454 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2455 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2456 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2457 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2458 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2462 * The boot-time readahead implementation for rotating media
2463 will now read the read-ahead data in multiple passes which
2464 consist of all read requests made in equidistant time
2465 intervals. This means instead of strictly reading read-ahead
2466 data in its physical order on disk we now try to find a
2467 middle ground between physical and access time order.
2469 * /etc/os-release files gained a new BUILD_ID= field for usage
2470 on operating systems that provide continuous builds of OS
2473 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2474 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2475 William Douglas, Zbigniew Jędrzejewski-Szmek
2479 * systemd-python gained an API exposing libsystemd-daemon.
2481 * The SMACK setup logic gained support for uploading CIPSO
2484 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2485 ReadOnlyDirectories= and InaccessibleDirectories= has
2486 changed. The private /tmp and /var/tmp directories are now
2487 shared by all processes of a service (which means
2488 ExecStartPre= may now leave data in /tmp that ExecStart= of
2489 the same service can still access). When a service is
2490 stopped its temporary directories are immediately deleted
2491 (normal clean-up with tmpfiles is still done in addition to
2494 * By default, systemd will now set a couple of sysctl
2495 variables in the kernel: the safe sysrq options are turned
2496 on, IP route verification is turned on, and source routing
2497 disabled. The recently added hardlink and softlink
2498 protection of the kernel is turned on. These settings should
2499 be reasonably safe, and good defaults for all new systems.
2501 * The predictable network naming logic may now be turned off
2502 with a new kernel command line switch: net.ifnames=0.
2504 * A new libsystemd-bus module has been added that implements a
2505 pretty complete D-Bus client library. For details see:
2507 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2509 * journald will now explicitly flush the journal files to disk
2510 at the latest 5min after each write. The file will then also
2511 be marked offline until the next write. This should increase
2512 reliability in case of a crash. The synchronization delay
2513 can be configured via SyncIntervalSec= in journald.conf.
2515 * There's a new remote-fs-setup.target unit that can be used
2516 to pull in specific services when at least one remote file
2517 system is to be mounted.
2519 * There are new targets timers.target and paths.target as
2520 canonical targets to pull user timer and path units in
2521 from. This complements sockets.target with a similar
2522 purpose for socket units.
2524 * libudev gained a new call udev_device_set_attribute_value()
2525 to set sysfs attributes of a device.
2527 * The udev daemon now sets the default number of worker
2528 processes executed in parallel based on the number of available
2529 CPUs instead of the amount of available RAM. This is supposed
2530 to provide a more reliable default and limit a too aggressive
2531 paralellism for setups with 1000s of devices connected.
2533 Contributions from: Auke Kok, Colin Walters, Cristian
2534 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2535 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2536 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2537 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2538 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2539 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2540 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2541 Zbigniew Jędrzejewski-Szmek
2545 * Configuration of unit files may now be extended via drop-in
2546 files without having to edit/override the unit files
2547 themselves. More specifically, if the administrator wants to
2548 change one value for a service file foobar.service he can
2549 now do so by dropping in a configuration snippet into
2550 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2551 will load all these snippets and apply them on top of the
2552 main unit configuration file, possibly extending or
2553 overriding its settings. Using these drop-in snippets is
2554 generally nicer than the two earlier options for changing
2555 unit files locally: copying the files from
2556 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2557 them there; or creating a new file in /etc/systemd/system/
2558 that incorporates the original one via ".include". Drop-in
2559 snippets into these .d/ directories can be placed in any
2560 directory systemd looks for units in, and the usual
2561 overriding semantics between /usr/lib, /etc and /run apply
2564 * Most unit file settings which take lists of items can now be
2565 reset by assigning the empty string to them. For example,
2566 normally, settings such as Environment=FOO=BAR append a new
2567 environment variable assignment to the environment block,
2568 each time they are used. By assigning Environment= the empty
2569 string the environment block can be reset to empty. This is
2570 particularly useful with the .d/*.conf drop-in snippets
2571 mentioned above, since this adds the ability to reset list
2572 settings from vendor unit files via these drop-ins.
2574 * systemctl gained a new "list-dependencies" command for
2575 listing the dependencies of a unit recursively.
2577 * Inhibitors are now honored and listed by "systemctl
2578 suspend", "systemctl poweroff" (and similar) too, not only
2579 GNOME. These commands will also list active sessions by
2582 * Resource limits (as exposed by the various control group
2583 controllers) can now be controlled dynamically at runtime
2584 for all units. More specifically, you can now use a command
2585 like "systemctl set-cgroup-attr foobar.service cpu.shares
2586 2000" to alter the CPU shares a specific service gets. These
2587 settings are stored persistently on disk, and thus allow the
2588 administrator to easily adjust the resource usage of
2589 services with a few simple commands. This dynamic resource
2590 management logic is also available to other programs via the
2591 bus. Almost any kernel cgroup attribute and controller is
2594 * systemd-vconsole-setup will now copy all font settings to
2595 all allocated VTs, where it previously applied them only to
2598 * libsystemd-login gained the new sd_session_get_tty() API
2601 * This release drops support for a few legacy or
2602 distribution-specific LSB facility names when parsing init
2603 scripts: $x-display-manager, $mail-transfer-agent,
2604 $mail-transport-agent, $mail-transfer-agent, $smtp,
2605 $null. Also, the mail-transfer-agent.target unit backing
2606 this has been removed. Distributions which want to retain
2607 compatibility with this should carry the burden for
2608 supporting this themselves and patch support for these back
2609 in, if they really need to. Also, the facilities $syslog and
2610 $local_fs are now ignored, since systemd does not support
2611 early-boot LSB init scripts anymore, and these facilities
2612 are implied anyway for normal services. syslog.target has
2615 * There are new bus calls on PID1's Manager object for
2616 cancelling jobs, and removing snapshot units. Previously,
2617 both calls were only available on the Job and Snapshot
2620 * systemd-journal-gatewayd gained SSL support.
2622 * The various "environment" files, such as /etc/locale.conf
2623 now support continuation lines with a backslash ("\") as
2624 last character in the line, similarly in style (but different)
2625 to how this is supported in shells.
2627 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2628 now implicitly appended to every log entry logged. systemctl
2629 has been updated to filter by this field when operating on a
2630 user systemd instance.
2632 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2633 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2634 the container. This makes it easier to boot unmodified
2635 Fedora systems in a container, which however still requires
2636 audit=0 to be passed on the kernel command line. Auditing in
2637 kernel and userspace is unfortunately still too broken in
2638 context of containers, hence we recommend compiling it out
2639 of the kernel or using audit=0. Hopefully this will be fixed
2640 one day for good in the kernel.
2642 * nspawn gained the new --bind= and --bind-ro= parameters to
2643 bind mount specific directories from the host into the
2646 * nspawn will now mount its own devpts file system instance
2647 into the container, in order not to leak pty devices from
2648 the host into the container.
2650 * systemd will now read the firmware boot time performance
2651 information from the EFI variables, if the used boot loader
2652 supports this, and takes it into account for boot performance
2653 analysis via "systemd-analyze". This is currently supported
2654 only in conjunction with Gummiboot, but could be supported
2655 by other boot loaders too. For details see:
2657 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2659 * A new generator has been added that automatically mounts the
2660 EFI System Partition (ESP) to /boot, if that directory
2661 exists, is empty, and no other file system has been
2662 configured to be mounted there.
2664 * logind will now send out PrepareForSleep(false) out
2665 unconditionally, after coming back from suspend. This may be
2666 used by applications as asynchronous notification for
2667 system resume events.
2669 * "systemctl unlock-sessions" has been added, that allows
2670 unlocking the screens of all user sessions at once, similar
2671 to how "systemctl lock-sessions" already locked all users
2672 sessions. This is backed by a new D-Bus call UnlockSessions().
2674 * "loginctl seat-status" will now show the master device of a
2675 seat. (i.e. the device of a seat that needs to be around for
2676 the seat to be considered available, usually the graphics
2679 * tmpfiles gained a new "X" line type, that allows
2680 configuration of files and directories (with wildcards) that
2681 shall be excluded from automatic cleanup ("aging").
2683 * udev default rules set the device node permissions now only
2684 at "add" events, and do not change them any longer with a
2685 later "change" event.
2687 * The log messages for lid events and power/sleep keypresses
2688 now carry a message ID.
2690 * We now have a substantially larger unit test suite, but this
2691 continues to be work in progress.
2693 * udevadm hwdb gained a new --root= parameter to change the
2694 root directory to operate relative to.
2696 * logind will now issue a background sync() request to the kernel
2697 early at shutdown, so that dirty buffers are flushed to disk early
2698 instead of at the last moment, in order to optimize shutdown
2701 * A new bootctl tool has been added that is an interface for
2702 certain boot loader operations. This is currently a preview
2703 and is likely to be extended into a small mechanism daemon
2704 like timedated, localed, hostnamed, and can be used by
2705 graphical UIs to enumerate available boot options, and
2706 request boot into firmware operations.
2708 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2709 the rest of the package. It also has been updated to work
2710 correctly in initrds.
2712 * Policykit previously has been runtime optional, and is now
2713 also compile time optional via a configure switch.
2715 * systemd-analyze has been reimplemented in C. Also "systemctl
2716 dot" has moved into systemd-analyze.
2718 * "systemctl status" with no further parameters will now print
2719 the status of all active or failed units.
2721 * Operations such as "systemctl start" can now be executed
2722 with a new mode "--irreversible" which may be used to queue
2723 operations that cannot accidentally be reversed by a later
2724 job queuing. This is by default used to make shutdown
2725 requests more robust.
2727 * The Python API of systemd now gained a new module for
2728 reading journal files.
2730 * A new tool kernel-install has been added that can install
2731 kernel images according to the Boot Loader Specification:
2733 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2735 * Boot time console output has been improved to provide
2736 animated boot time output for hanging jobs.
2738 * A new tool systemd-activate has been added which can be used
2739 to test socket activation with, directly from the command
2740 line. This should make it much easier to test and debug
2741 socket activation in daemons.
2743 * journalctl gained a new "--reverse" (or -r) option to show
2744 journal output in reverse order (i.e. newest line first).
2746 * journalctl gained a new "--pager-end" (or -e) option to jump
2747 to immediately jump to the end of the journal in the
2748 pager. This is only supported in conjunction with "less".
2750 * journalctl gained a new "--user-unit=" option, that works
2751 similarly to "--unit=" but filters for user units rather than
2754 * A number of unit files to ease adoption of systemd in
2755 initrds has been added. This moves some minimal logic from
2756 the various initrd implementations into systemd proper.
2758 * The journal files are now owned by a new group
2759 "systemd-journal", which exists specifically to allow access
2760 to the journal, and nothing else. Previously, we used the
2761 "adm" group for that, which however possibly covers more
2762 than just journal/log file access. This new group is now
2763 already used by systemd-journal-gatewayd to ensure this
2764 daemon gets access to the journal files and as little else
2765 as possible. Note that "make install" will also set FS ACLs
2766 up for /var/log/journal to give "adm" and "wheel" read
2767 access to it, in addition to "systemd-journal" which owns
2768 the journal files. We recommend that packaging scripts also
2769 add read access to "adm" + "wheel" to /var/log/journal, and
2770 all existing/future journal files. To normal users and
2771 administrators little changes, however packagers need to
2772 ensure to create the "systemd-journal" system group at
2773 package installation time.
2775 * The systemd-journal-gatewayd now runs as unprivileged user
2776 systemd-journal-gateway:systemd-journal-gateway. Packaging
2777 scripts need to create these system user/group at
2780 * timedated now exposes a new boolean property CanNTP that
2781 indicates whether a local NTP service is available or not.
2783 * systemd-detect-virt will now also detect xen PVs
2785 * The pstore file system is now mounted by default, if it is
2788 * In addition to the SELinux and IMA policies we will now also
2789 load SMACK policies at early boot.
2791 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2792 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2793 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2794 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2795 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2796 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2797 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2798 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2799 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2800 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2801 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2802 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2803 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2804 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2808 * Timer units now support calendar time events in addition to
2809 monotonic time events. That means you can now trigger a unit
2810 based on a calendar time specification such as "Thu,Fri
2811 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2812 or fifth day of any month of the year 2013, given that it is
2813 a thursday or friday. This brings timer event support
2814 considerably closer to cron's capabilities. For details on
2815 the supported calendar time specification language see
2818 * udev now supports a number of different naming policies for
2819 network interfaces for predictable names, and a combination
2820 of these policies is now the default. Please see this wiki
2821 document for details:
2823 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2825 * Auke Kok's bootchart implementation has been added to the
2826 systemd tree. It is an optional component that can graph the
2827 boot in quite some detail. It is one of the best bootchart
2828 implementations around and minimal in its code and
2831 * nss-myhostname has been integrated into the systemd source
2832 tree. nss-myhostname guarantees that the local hostname
2833 always stays resolvable via NSS. It has been a weak
2834 requirement of systemd-hostnamed since a long time, and
2835 since its code is actually trivial we decided to just
2836 include it in systemd's source tree. It can be turned off
2837 with a configure switch.
2839 * The read-ahead logic is now capable of properly detecting
2840 whether a btrfs file system is on SSD or rotating media, in
2841 order to optimize the read-ahead scheme. Previously, it was
2842 only capable of detecting this on traditional file systems
2845 * In udev, additional device properties are now read from the
2846 IAB in addition to the OUI database. Also, Bluetooth company
2847 identities are attached to the devices as well.
2849 * In service files %U may be used as specifier that is
2850 replaced by the configured user name of the service.
2852 * nspawn may now be invoked without a controlling TTY. This
2853 makes it suitable for invocation as its own service. This
2854 may be used to set up a simple containerized server system
2855 using only core OS tools.
2857 * systemd and nspawn can now accept socket file descriptors
2858 when they are started for socket activation. This enables
2859 implementation of socket activated nspawn
2860 containers. i.e. think about autospawning an entire OS image
2861 when the first SSH or HTTP connection is received. We expect
2862 that similar functionality will also be added to libvirt-lxc
2865 * journalctl will now suppress ANSI color codes when
2866 presenting log data.
2868 * systemctl will no longer show control group information for
2869 a unit if a the control group is empty anyway.
2871 * logind can now automatically suspend/hibernate/shutdown the
2874 * /etc/machine-info and hostnamed now also expose the chassis
2875 type of the system. This can be used to determine whether
2876 the local system is a laptop, desktop, handset or
2877 tablet. This information may either be configured by the
2878 user/vendor or is automatically determined from ACPI and DMI
2879 information if possible.
2881 * A number of PolicyKit actions are now bound together with
2882 "imply" rules. This should simplify creating UIs because
2883 many actions will now authenticate similar ones as well.
2885 * Unit files learnt a new condition ConditionACPower= which
2886 may be used to conditionalize a unit depending on whether an
2887 AC power source is connected or not, of whether the system
2888 is running on battery power.
2890 * systemctl gained a new "is-failed" verb that may be used in
2891 shell scripts and suchlike to check whether a specific unit
2892 is in the "failed" state.
2894 * The EnvironmentFile= setting in unit files now supports file
2895 globbing, and can hence be used to easily read a number of
2896 environment files at once.
2898 * systemd will no longer detect and recognize specific
2899 distributions. All distribution-specific #ifdeffery has been
2900 removed, systemd is now fully generic and
2901 distribution-agnostic. Effectively, not too much is lost as
2902 a lot of the code is still accessible via explicit configure
2903 switches. However, support for some distribution specific
2904 legacy configuration file formats has been dropped. We
2905 recommend distributions to simply adopt the configuration
2906 files everybody else uses now and convert the old
2907 configuration from packaging scripts. Most distributions
2908 already did that. If that's not possible or desirable,
2909 distributions are welcome to forward port the specific
2910 pieces of code locally from the git history.
2912 * When logging a message about a unit systemd will now always
2913 log the unit name in the message meta data.
2915 * localectl will now also discover system locale data that is
2916 not stored in locale archives, but directly unpacked.
2918 * logind will no longer unconditionally use framebuffer
2919 devices as seat masters, i.e. as devices that are required
2920 to be existing before a seat is considered preset. Instead,
2921 it will now look for all devices that are tagged as
2922 "seat-master" in udev. By default framebuffer devices will
2923 be marked as such, but depending on local systems other
2924 devices might be marked as well. This may be used to
2925 integrate graphics cards using closed source drivers (such
2926 as NVidia ones) more nicely into logind. Note however, that
2927 we recommend using the open source NVidia drivers instead,
2928 and no udev rules for the closed-source drivers will be
2929 shipped from us upstream.
2931 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2932 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2933 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2934 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2935 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2936 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2937 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2938 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2939 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2940 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2941 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2942 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2947 * udev gained support for loading additional device properties
2948 from an indexed database that is keyed by vendor/product IDs
2949 and similar device identifiers. For the beginning this
2950 "hwdb" is populated with data from the well-known PCI and
2951 USB database, but also includes PNP, ACPI and OID data. In
2952 the longer run this indexed database shall grow into
2953 becoming the one central database for non-essential
2954 userspace device metadata. Previously, data from the PCI/USB
2955 database was only attached to select devices, since the
2956 lookup was a relatively expensive operation due to O(n) time
2957 complexity (with n being the number of entries in the
2958 database). Since this is now O(1), we decided to add in this
2959 data for all devices where this is available, by
2960 default. Note that the indexed database needs to be rebuilt
2961 when new data files are installed. To achieve this you need
2962 to update your packaging scripts to invoke "udevadm hwdb
2963 --update" after installation of hwdb data files. For
2964 RPM-based distributions we introduced the new
2965 %udev_hwdb_update macro for this purpose.
2967 * The Journal gained support for the "Message Catalog", an
2968 indexed database to link up additional information with
2969 journal entries. For further details please check:
2971 http://www.freedesktop.org/wiki/Software/systemd/catalog
2973 The indexed message catalog database also needs to be
2974 rebuilt after installation of message catalog files. Use
2975 "journalctl --update-catalog" for this. For RPM-based
2976 distributions we introduced the %journal_catalog_update
2977 macro for this purpose.
2979 * The Python Journal bindings gained support for the standard
2980 Python logging framework.
2982 * The Journal API gained new functions for checking whether
2983 the underlying file system of a journal file is capable of
2984 properly reporting file change notifications, or whether
2985 applications that want to reflect journal changes "live"
2986 need to recheck journal files continuously in appropriate
2989 * It is now possible to set the "age" field for tmpfiles
2990 entries to 0, indicating that files matching this entry
2991 shall always be removed when the directories are cleaned up.
2993 * coredumpctl gained a new "gdb" verb which invokes gdb
2994 right-away on the selected coredump.
2996 * There's now support for "hybrid sleep" on kernels that
2997 support this, in addition to "suspend" and "hibernate". Use
2998 "systemctl hybrid-sleep" to make use of this.
3000 * logind's HandleSuspendKey= setting (and related settings)
3001 now gained support for a new "lock" setting to simply
3002 request the screen lock on all local sessions, instead of
3003 actually executing a suspend or hibernation.
3005 * systemd will now mount the EFI variables file system by
3008 * Socket units now gained support for configuration of the
3009 SMACK security label.
3011 * timedatectl will now output the time of the last and next
3012 daylight saving change.
3014 * We dropped support for various legacy and distro-specific
3015 concepts, such as insserv, early-boot SysV services
3016 (i.e. those for non-standard runlevels such as 'b' or 'S')
3017 or ArchLinux /etc/rc.conf support. We recommend the
3018 distributions who still need support this to either continue
3019 to maintain the necessary patches downstream, or find a
3020 different solution. (Talk to us if you have questions!)
3022 * Various systemd components will now bypass PolicyKit checks
3023 for root and otherwise handle properly if PolicyKit is not
3024 found to be around. This should fix most issues for
3025 PolicyKit-less systems. Quite frankly this should have been
3026 this way since day one. It is absolutely our intention to
3027 make systemd work fine on PolicyKit-less systems, and we
3028 consider it a bug if something does not work as it should if
3029 PolicyKit is not around.
3031 * For embedded systems it is now possible to build udev and
3032 systemd without blkid and/or kmod support.
3034 * "systemctl switch-root" is now capable of switching root
3035 more than once. I.e. in addition to transitions from the
3036 initrd to the host OS it is now possible to transition to
3037 further OS images from the host. This is useful to implement
3038 offline updating tools.
3040 * Various other additions have been made to the RPM macros
3041 shipped with systemd. Use %udev_rules_update() after
3042 installing new udev rules files. %_udevhwdbdir,
3043 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3044 %_sysctldir are now available which resolve to the right
3045 directories for packages to place various data files in.
3047 * journalctl gained the new --full switch (in addition to
3048 --all, to disable ellipsation for long messages.
3050 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3051 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3052 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3053 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3054 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3055 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3056 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3057 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3058 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3062 * journalctl gained new --since= and --until= switches to
3063 filter by time. It also now supports nice filtering for
3064 units via --unit=/-u.
3066 * Type=oneshot services may use ExecReload= and do the
3069 * The journal daemon now supports time-based rotation and
3070 vacuuming, in addition to the usual disk-space based
3073 * The journal will now index the available field values for
3074 each field name. This enables clients to show pretty drop
3075 downs of available match values when filtering. The bash
3076 completion of journalctl has been updated
3077 accordingly. journalctl gained a new switch -F to list all
3078 values a certain field takes in the journal database.
3080 * More service events are now written as structured messages
3081 to the journal, and made recognizable via message IDs.
3083 * The timedated, localed and hostnamed mini-services which
3084 previously only provided support for changing time, locale
3085 and hostname settings from graphical DEs such as GNOME now
3086 also have a minimal (but very useful) text-based client
3087 utility each. This is probably the nicest way to changing
3088 these settings from the command line now, especially since
3089 it lists available options and is fully integrated with bash
3092 * There's now a new tool "systemd-coredumpctl" to list and
3093 extract coredumps from the journal.
3095 * We now install a README each in /var/log/ and
3096 /etc/rc.d/init.d explaining where the system logs and init
3097 scripts went. This hopefully should help folks who go to
3098 that dirs and look into the otherwise now empty void and
3099 scratch their heads.
3101 * When user-services are invoked (by systemd --user) the
3102 $MANAGERPID env var is set to the PID of systemd.
3104 * SIGRTMIN+24 when sent to a --user instance will now result
3105 in immediate termination of systemd.
3107 * gatewayd received numerous feature additions such as a
3108 "follow" mode, for live syncing and filtering.
3110 * browse.html now allows filtering and showing detailed
3111 information on specific entries. Keyboard navigation and
3112 mouse screen support has been added.
3114 * gatewayd/journalctl now supports HTML5/JSON
3115 Server-Sent-Events as output.
3117 * The SysV init script compatibility logic will now
3118 heuristically determine whether a script supports the
3119 "reload" verb, and only then make this available as
3122 * "systemctl status --follow" has been removed, use "journalctl
3125 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3126 have been removed since they are hardly useful to be
3129 * And I'd like to take the opportunity to specifically mention
3130 Zbigniew for his great contributions. Zbigniew, you rock!
3132 Contributions from: Andrew Eikum, Christian Hesse, Colin
3133 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3134 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3135 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3136 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3137 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3138 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3142 * If /etc/vconsole.conf is non-existent or empty we will no
3143 longer load any console font or key map at boot by
3144 default. Instead the kernel defaults will be left
3145 intact. This is definitely the right thing to do, as no
3146 configuration should mean no configuration, and hard-coding
3147 font names that are different on all archs is probably a bad
3148 idea. Also, the kernel default key map and font should be
3149 good enough for most cases anyway, and mostly identical to
3150 the userspace fonts/key maps we previously overloaded them
3151 with. If distributions want to continue to default to a
3152 non-kernel font or key map they should ship a default
3153 /etc/vconsole.conf with the appropriate contents.
3155 Contributions from: Colin Walters, Daniel J Walsh, Dave
3156 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3157 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3161 * journalctl gained a new --cursor= switch to show entries
3162 starting from the specified location in the journal.
3164 * We now enforce a size limit on journal entry fields exported
3165 with "-o json" in journalctl. Fields larger than 4K will be
3166 assigned null. This can be turned off with --all.
3168 * An (optional) journal gateway daemon is now available as
3169 "systemd-journal-gatewayd.service". This service provides
3170 access to the journal via HTTP and JSON. This functionality
3171 will be used to implement live log synchronization in both
3172 pull and push modes, but has various other users too, such
3173 as easy log access for debugging of embedded devices. Right
3174 now it is already useful to retrieve the journal via HTTP:
3176 # systemctl start systemd-journal-gatewayd.service
3177 # wget http://localhost:19531/entries
3179 This will download the journal contents in a
3180 /var/log/messages compatible format. The same as JSON:
3182 # curl -H"Accept: application/json" http://localhost:19531/entries
3184 This service is also accessible via a web browser where a
3185 single static HTML5 app is served that uses the JSON logic
3186 to enable the user to do some basic browsing of the
3187 journal. This will be extended later on. Here's an example
3188 screenshot of this app in its current state:
3190 http://0pointer.de/public/journal-gatewayd
3192 Contributions from: Kay Sievers, Lennart Poettering, Robert
3193 Milasan, Tom Gundersen
3197 * The bash completion logic is now available for journalctl
3200 * We do not mount the "cpuset" controller anymore together with
3201 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3202 started if no parameters are assigned to it. "cpuset" hence
3203 broke code that assumed it it could create "cpu" groups and
3206 * journalctl -f will now subscribe to terminal size changes,
3207 and line break accordingly.
3209 Contributions from: Dave Reisner, Kay Sievers, Lennart
3210 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3214 * nspawn will now create a symlink /etc/localtime in the
3215 container environment, copying the host's timezone
3216 setting. Previously this has been done via a bind mount, but
3217 since symlinks cannot be bind mounted this has now been
3218 changed to create/update the appropriate symlink.
3220 * journalctl -n's line number argument is now optional, and
3221 will default to 10 if omitted.
3223 * journald will now log the maximum size the journal files may
3224 take up on disk. This is particularly useful if the default
3225 built-in logic of determining this parameter from the file
3226 system size is used. Use "systemctl status
3227 systemd-journald.service" to see this information.
3229 * The multi-seat X wrapper tool has been stripped down. As X
3230 is now capable of enumerating graphics devices via udev in a
3231 seat-aware way the wrapper is not strictly necessary
3232 anymore. A stripped down temporary stop-gap is still shipped
3233 until the upstream display managers have been updated to
3234 fully support the new X logic. Expect this wrapper to be
3235 removed entirely in one of the next releases.
3237 * HandleSleepKey= in logind.conf has been split up into
3238 HandleSuspendKey= and HandleHibernateKey=. The old setting
3239 is not available anymore. X11 and the kernel are
3240 distuingishing between these keys and we should too. This
3241 also means the inhibition lock for these keys has been split
3244 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3245 Poettering, Lukas Nykryn, Václav Pavlín
3249 * Whenever a unit changes state we will now log this to the
3250 journal and show along the unit's own log output in
3253 * ConditionPathIsMountPoint= can now properly detect bind
3254 mount points too. (Previously, a bind mount of one file
3255 system to another place in the same file system could not be
3256 detected as mount, since they shared struct stat's st_dev
3259 * We will now mount the cgroup controllers cpu, cpuacct,
3260 cpuset and the controllers net_cls, net_prio together by
3263 * nspawn containers will now have a virtualized boot
3264 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3265 over with a randomized ID at container initialization). This
3266 has the effect of making "journalctl -b" do the right thing
3269 * The JSON output journal serialization has been updated not
3270 to generate "endless" list objects anymore, but rather one
3271 JSON object per line. This is more in line how most JSON
3272 parsers expect JSON objects. The new output mode
3273 "json-pretty" has been added to provide similar output, but
3274 neatly aligned for readability by humans.
3276 * We dropped all explicit sync() invocations in the shutdown
3277 code. The kernel does this implicitly anyway in the kernel
3278 reboot() syscall. halt(8)'s -n option is now a compatibility
3281 * We now support virtualized reboot() in containers, as
3282 supported by newer kernels. We will fall back to exit() if
3283 CAP_SYS_REBOOT is not available to the container. Also,
3284 nspawn makes use of this now and will actually reboot the
3285 container if the containerized OS asks for that.
3287 * journalctl will only show local log output by default
3288 now. Use --merge (-m) to show remote log output, too.
3290 * libsystemd-journal gained the new sd_journal_get_usage()
3291 call to determine the current disk usage of all journal
3292 files. This is exposed in the new "journalctl --disk-usage"
3295 * journald gained a new configuration setting SplitMode= in
3296 journald.conf which may be used to control how user journals
3297 are split off. See journald.conf(5) for details.
3299 * A new condition type ConditionFileNotEmpty= has been added.
3301 * tmpfiles' "w" lines now support file globbing, to write
3302 multiple files at once.
3304 * We added Python bindings for the journal submission
3305 APIs. More Python APIs for a number of selected APIs will
3306 likely follow. Note that we intend to add native bindings
3307 only for the Python language, as we consider it common
3308 enough to deserve bindings shipped within systemd. There are
3309 various projects outside of systemd that provide bindings
3310 for languages such as PHP or Lua.
3312 * Many conditions will now resolve specifiers such as %i. In
3313 addition, PathChanged= and related directives of .path units
3314 now support specifiers as well.
3316 * There's now a new RPM macro definition for the system preset
3319 * journald will now warn if it ca not forward a message to the
3320 syslog daemon because its socket is full.
3322 * timedated will no longer write or process /etc/timezone,
3323 except on Debian. As we do not support late mounted /usr
3324 anymore /etc/localtime always being a symlink is now safe,
3325 and hence the information in /etc/timezone is not necessary
3328 * logind will now always reserve one VT for a text getty (VT6
3329 by default). Previously if more than 6 X sessions where
3330 started they took up all the VTs with auto-spawned gettys,
3331 so that no text gettys were available anymore.
3333 * udev will now automatically inform the btrfs kernel logic
3334 about btrfs RAID components showing up. This should make
3335 simple hotplug based btrfs RAID assembly work.
3337 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3338 (but not for its children which will stay at the kernel
3339 default). This should allow setups with a lot more listening
3342 * systemd will now always pass the configured timezone to the
3343 kernel at boot. timedated will do the same when the timezone
3346 * logind's inhibition logic has been updated. By default,
3347 logind will now handle the lid switch, the power and sleep
3348 keys all the time, even in graphical sessions. If DEs want
3349 to handle these events on their own they should take the new
3350 handle-power-key, handle-sleep-key and handle-lid-switch
3351 inhibitors during their runtime. A simple way to achiveve
3352 that is to invoke the DE wrapped in an invocation of:
3354 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3356 * Access to unit operations is now checked via SELinux taking
3357 the unit file label and client process label into account.
3359 * systemd will now notify the administrator in the journal
3360 when he over-mounts a non-empty directory.
3362 * There are new specifiers that are resolved in unit files,
3363 for the host name (%H), the machine ID (%m) and the boot ID
3366 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3367 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3368 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3369 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3370 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3371 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3372 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3376 * Support for reading structured kernel messages from
3377 /dev/kmsg has now been added and is enabled by default.
3379 * Support for reading kernel messages from /proc/kmsg has now
3380 been removed. If you want kernel messages in the journal
3381 make sure to run a recent kernel (>= 3.5) that supports
3382 reading structured messages from /dev/kmsg (see
3383 above). /proc/kmsg is now exclusive property of classic
3384 syslog daemons again.
3386 * The libudev API gained the new
3387 udev_device_new_from_device_id() call.
3389 * The logic for file system namespace (ReadOnlyDirectory=,
3390 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3391 require pivot_root() anymore. This means fewer temporary
3392 directories are created below /tmp for this feature.
3394 * nspawn containers will now see and receive all submounts
3395 made on the host OS below the root file system of the
3398 * Forward Secure Sealing is now supported for Journal files,
3399 which provide cryptographical sealing of journal files so
3400 that attackers cannot alter log history anymore without this
3401 being detectable. Lennart will soon post a blog story about
3402 this explaining it in more detail.
3404 * There are two new service settings RestartPreventExitStatus=
3405 and SuccessExitStatus= which allow configuration of exit
3406 status (exit code or signal) which will be excepted from the
3407 restart logic, resp. consider successful.
3409 * journalctl gained the new --verify switch that can be used
3410 to check the integrity of the structure of journal files and
3411 (if Forward Secure Sealing is enabled) the contents of
3414 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3415 and similar symlinks pre-created. This makes running shells
3416 as container init process a lot more fun.
3418 * The fstab support can now handle PARTUUID= and PARTLABEL=
3421 * A new ConditionHost= condition has been added to match
3422 against the hostname (with globs) and machine ID. This is
3423 useful for clusters where a single OS image is used to
3424 provision a large number of hosts which shall run slightly
3425 different sets of services.
3427 * Services which hit the restart limit will now be placed in a
3430 Contributions from: Bertram Poettering, Dave Reisner, Huang
3431 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3432 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3436 * When running in --user mode systemd will now become a
3437 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3438 tree a lot more organized.
3440 * A new PartOf= unit dependency type has been introduced that
3441 may be used to group services in a natural way.
3443 * "systemctl enable" may now be used to enable instances of
3446 * journalctl now prints error log levels in red, and
3447 warning/notice log levels in bright white. It also supports
3448 filtering by log level now.
3450 * cgtop gained a new -n switch (similar to top), to configure
3451 the maximum number of iterations to run for. It also gained
3452 -b, to run in batch mode (accepting no input).
3454 * The suffix ".service" may now be omitted on most systemctl
3455 command lines involving service unit names.
3457 * There's a new bus call in logind to lock all sessions, as
3458 well as a loginctl verb for it "lock-sessions".
3460 * libsystemd-logind.so gained a new call sd_journal_perror()
3461 that works similar to libc perror() but logs to the journal
3462 and encodes structured information about the error number.
3464 * /etc/crypttab entries now understand the new keyfile-size=
3467 * shutdown(8) now can send a (configurable) wall message when
3468 a shutdown is cancelled.
3470 * The mount propagation mode for the root file system will now
3471 default to "shared", which is useful to make containers work
3472 nicely out-of-the-box so that they receive new mounts from
3473 the host. This can be undone locally by running "mount
3474 --make-rprivate /" if needed.
3476 * The prefdm.service file has been removed. Distributions
3477 should maintain this unit downstream if they intend to keep
3478 it around. However, we recommend writing normal unit files
3479 for display managers instead.
3481 * Since systemd is a crucial part of the OS we will now
3482 default to a number of compiler switches that improve
3483 security (hardening) such as read-only relocations, stack
3484 protection, and suchlike.
3486 * The TimeoutSec= setting for services is now split into
3487 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3488 of individual time outs for the start and the stop phase of
3491 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3492 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3493 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3494 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3495 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3496 Gundersen, Zbigniew Jędrzejewski-Szmek
3500 * The journal and id128 C APIs are now fully documented as man
3503 * Extra safety checks have been added when transitioning from
3504 the initial RAM disk to the main system to avoid accidental
3507 * /etc/crypttab entries now understand the new keyfile-offset=
3510 * systemctl -t can now be used to filter by unit load state.
3512 * The journal C API gained the new sd_journal_wait() call to
3513 make writing synchronous journal clients easier.
3515 * journalctl gained the new -D switch to show journals from a
3518 * journalctl now displays a special marker between log
3519 messages of two different boots.
3521 * The journal is now explicitly flushed to /var via a service
3522 systemd-journal-flush.service, rather than implicitly simply
3523 by seeing /var/log/journal to be writable.
3525 * journalctl (and the journal C APIs) can now match for much
3526 more complex expressions, with alternatives and
3529 * When transitioning from the initial RAM disk to the main
3530 system we will now kill all processes in a killing spree to
3531 ensure no processes stay around by accident.
3533 * Three new specifiers may be used in unit files: %u, %h, %s
3534 resolve to the user name, user home directory resp. user
3535 shell. This is useful for running systemd user instances.
3537 * We now automatically rotate journal files if their data
3538 object hash table gets a fill level > 75%. We also size the
3539 hash table based on the configured maximum file size. This
3540 together should lower hash collisions drastically and thus
3541 speed things up a bit.
3543 * journalctl gained the new "--header" switch to introspect
3544 header data of journal files.
3546 * A new setting SystemCallFilters= has been added to services
3547 which may be used to apply blacklists or whitelists to
3548 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3550 * nspawn gained a new --link-journal= switch (and quicker: -j)
3551 to link the container journal with the host. This makes it
3552 very easy to centralize log viewing on the host for all
3553 guests while still keeping the journal files separated.
3555 * Many bugfixes and optimizations
3557 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3558 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3559 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3564 * Several tools now understand kernel command line arguments,
3565 which are only read when run in an initial RAM disk. They
3566 usually follow closely their normal counterparts, but are
3569 * There's a new tool to analyze the readahead files that are
3570 automatically generated at boot. Use:
3572 /usr/lib/systemd/systemd-readahead analyze /.readahead
3574 * We now provide an early debug shell on tty9 if this enabled. Use:
3576 systemctl enable debug-shell.service
3578 * All plymouth related units have been moved into the Plymouth
3579 package. Please make sure to upgrade your Plymouth version
3582 * systemd-tmpfiles now supports getting passed the basename of
3583 a configuration file only, in which case it will look for it
3584 in all appropriate directories automatically.
3586 * udevadm info now takes a /dev or /sys path as argument, and
3587 does the right thing. Example:
3589 udevadm info /dev/sda
3590 udevadm info /sys/class/block/sda
3592 * systemctl now prints a warning if a unit is stopped but a
3593 unit that might trigger it continues to run. Example: a
3594 service is stopped but the socket that activates it is left
3597 * "systemctl status" will now mention if the log output was
3598 shortened due to rotation since a service has been started.
3600 * The journal API now exposes functions to determine the
3601 "cutoff" times due to rotation.
3603 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3604 immediately flushing of runtime logs to /var if possible,
3605 resp. for triggering immediate rotation of the journal
3608 * It is now considered an error if a service is attempted to
3609 be stopped that is not loaded.
3611 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3613 * systemd-analyze now supports Python 3
3615 * tmpfiles now supports cleaning up directories via aging
3616 where the first level dirs are always kept around but
3617 directories beneath it automatically aged. This is enabled
3618 by prefixing the age field with '~'.
3620 * Seat objects now expose CanGraphical, CanTTY properties
3621 which is required to deal with very fast bootups where the
3622 display manager might be running before the graphics drivers
3623 completed initialization.
3625 * Seat objects now expose a State property.
3627 * We now include RPM macros for service enabling/disabling
3628 based on the preset logic. We recommend RPM based
3629 distributions to make use of these macros if possible. This
3630 makes it simpler to reuse RPM spec files across
3633 * We now make sure that the collected systemd unit name is
3634 always valid when services log to the journal via
3637 * There's a new man page kernel-command-line(7) detailing all
3638 command line options we understand.
3640 * The fstab generator may now be disabled at boot by passing
3641 fstab=0 on the kernel command line.
3643 * A new kernel command line option modules-load= is now understood
3644 to load a specific kernel module statically, early at boot.
3646 * Unit names specified on the systemctl command line are now
3647 automatically escaped as needed. Also, if file system or
3648 device paths are specified they are automatically turned
3649 into the appropriate mount or device unit names. Example:
3651 systemctl status /home
3652 systemctl status /dev/sda
3654 * The SysVConsole= configuration option has been removed from
3655 system.conf parsing.
3657 * The SysV search path is no longer exported on the D-Bus
3660 * The Names= option is been removed from unit file parsing.
3662 * There's a new man page bootup(7) detailing the boot process.
3664 * Every unit and every generator we ship with systemd now
3665 comes with full documentation. The self-explanatory boot is
3668 * A couple of services gained "systemd-" prefixes in their
3669 name if they wrap systemd code, rather than only external
3670 code. Among them fsck@.service which is now
3671 systemd-fsck@.service.
3673 * The HaveWatchdog property has been removed from the D-Bus
3676 * systemd.confirm_spawn= on the kernel command line should now
3679 * There's a new man page crypttab(5) which details all options
3680 we actually understand.
3682 * systemd-nspawn gained a new --capability= switch to pass
3683 additional capabilities to the container.
3685 * timedated will now read known NTP implementation unit names
3686 from /usr/lib/systemd/ntp-units.d/*.list,
3687 systemd-timedated-ntp.target has been removed.
3689 * journalctl gained a new switch "-b" that lists log data of
3690 the current boot only.
3692 * The notify socket is in the abstract namespace again, in
3693 order to support daemons which chroot() at start-up.
3695 * There is a new Storage= configuration option for journald
3696 which allows configuration of where log data should go. This
3697 also provides a way to disable journal logging entirely, so
3698 that data collected is only forwarded to the console, the
3699 kernel log buffer or another syslog implementation.
3701 * Many bugfixes and optimizations
3703 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3704 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3705 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3706 Shawn Landden, Tom Gundersen
3710 * "systemctl help <unit>" now shows the man page if one is
3713 * Several new man pages have been added.
3715 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3716 MaxLevelConsole= can now be specified in
3717 journald.conf. These options allow reducing the amount of
3718 data stored on disk or forwarded by the log level.
3720 * TimerSlackNSec= can now be specified in system.conf for
3721 PID1. This allows system-wide power savings.
3723 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3724 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3729 * logind is now capable of (optionally) handling power and
3730 sleep keys as well as the lid switch.
3732 * journalctl now understands the syntax "journalctl
3733 /usr/bin/avahi-daemon" to get all log output of a specific
3736 * CapabilityBoundingSet= in system.conf now also influences
3737 the capability bound set of usermode helpers of the kernel.
3739 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3740 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3741 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3742 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3746 * Note that we skipped 139 releases here in order to set the
3747 new version to something that is greater than both udev's
3748 and systemd's most recent version number.
3750 * udev: all udev sources are merged into the systemd source tree now.
3751 All future udev development will happen in the systemd tree. It
3752 is still fully supported to use the udev daemon and tools without
3753 systemd running, like in initramfs or other init systems. Building
3754 udev though, will require the *build* of the systemd tree, but
3755 udev can be properly *run* without systemd.
3757 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3758 should be used to create dead device nodes as workarounds for broken
3761 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3762 no longer supported. udev_monitor_new_from_netlink() needs to be
3763 used to subscribe to events.
3765 * udev: when udevd is started by systemd, processes which are left
3766 behind by forking them off of udev rules, are unconditionally cleaned
3767 up and killed now after the event handling has finished. Services or
3768 daemons must be started as systemd services. Services can be
3769 pulled-in by udev to get started, but they can no longer be directly
3770 forked by udev rules.
3772 * udev: the daemon binary is called systemd-udevd now and installed
3773 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3774 to adapt to that, create symlink, or rename the binary after building
3777 * libudev no longer provides these symbols:
3778 udev_monitor_from_socket()
3779 udev_queue_get_failed_list_entry()
3780 udev_get_{dev,sys,run}_path()
3781 The versions number was bumped and symbol versioning introduced.
3783 * systemd-loginctl and systemd-journalctl have been renamed
3784 to loginctl and journalctl to match systemctl.
3786 * The config files: /etc/systemd/systemd-logind.conf and
3787 /etc/systemd/systemd-journald.conf have been renamed to
3788 logind.conf and journald.conf. Package updates should rename
3789 the files to the new names on upgrade.
3791 * For almost all files the license is now LGPL2.1+, changed
3792 from the previous GPL2.0+. Exceptions are some minor stuff
3793 of udev (which will be changed to LGPL2.1 eventually, too),
3794 and the MIT licensed sd-daemon.[ch] library that is suitable
3795 to be used as drop-in files.
3797 * systemd and logind now handle system sleep states, in
3798 particular suspending and hibernating.
3800 * logind now implements a sleep/shutdown/idle inhibiting logic
3801 suitable for a variety of uses. Soonishly Lennart will blog
3802 about this in more detail.
3804 * var-run.mount and var-lock.mount are no longer provided
3805 (which prevously bind mounted these directories to their new
3806 places). Distributions which have not converted these
3807 directories to symlinks should consider stealing these files
3808 from git history and add them downstream.
3810 * We introduced the Documentation= field for units and added
3811 this to all our shipped units. This is useful to make it
3812 easier to explore the boot and the purpose of the various
3815 * All smaller setup units (such as
3816 systemd-vconsole-setup.service) now detect properly if they
3817 are run in a container and are skipped when
3818 appropriate. This guarantees an entirely noise-free boot in
3819 Linux container environments such as systemd-nspawn.
3821 * A framework for implementing offline system updates is now
3822 integrated, for details see:
3823 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3825 * A new service type Type=idle is available now which helps us
3826 avoiding ugly interleaving of getty output and boot status
3829 * There's now a system-wide CapabilityBoundingSet= option to
3830 globally reduce the set of capabilities for the
3831 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3832 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3833 even CAP_NET_ADMIN system-wide for secure systems.
3835 * There are now system-wide DefaultLimitXXX= options to
3836 globally change the defaults of the various resource limits
3837 for all units started by PID 1.
3839 * Harald Hoyer's systemd test suite has been integrated into
3840 systemd which allows easy testing of systemd builds in qemu
3841 and nspawn. (This is really awesome! Ask us for details!)
3843 * The fstab parser is now implemented as generator, not inside
3846 * systemctl will now warn you if .mount units generated from
3847 /etc/fstab are out of date due to changes in fstab that
3848 have not been read by systemd yet.
3850 * systemd is now suitable for usage in initrds. Dracut has
3851 already been updated to make use of this. With this in place
3852 initrds get a slight bit faster but primarily are much
3853 easier to introspect and debug since "systemctl status" in
3854 the host system can be used to introspect initrd services,
3855 and the journal from the initrd is kept around too.
3857 * systemd-delta has been added, a tool to explore differences
3858 between user/admin configuration and vendor defaults.
3860 * PrivateTmp= now affects both /tmp and /var/tmp.
3862 * Boot time status messages are now much prettier and feature
3863 proper english language. Booting up systemd has never been
3866 * Read-ahead pack files now include the inode number of all
3867 files to pre-cache. When the inode changes the pre-caching
3868 is not attempted. This should be nicer to deal with updated
3869 packages which might result in changes of read-ahead
3872 * We now temporaritly lower the kernel's read_ahead_kb variable
3873 when collecting read-ahead data to ensure the kernel's
3874 built-in read-ahead does not add noise to our measurements
3875 of necessary blocks to pre-cache.
3877 * There's now RequiresMountsFor= to add automatic dependencies
3878 for all mounts necessary for a specific file system path.
3880 * MountAuto= and SwapAuto= have been removed from
3881 system.conf. Mounting file systems at boot has to take place
3884 * nspawn now learned a new switch --uuid= to set the machine
3885 ID on the command line.
3887 * nspawn now learned the -b switch to automatically search
3890 * vt102 is now the default TERM for serial TTYs, upgraded from
3893 * systemd-logind now works on VT-less systems.
3895 * The build tree has been reorganized. The individual
3896 components now have directories of their own.
3898 * A new condition type ConditionPathIsReadWrite= is now available.
3900 * nspawn learned the new -C switch to create cgroups for the
3901 container in other hierarchies.
3903 * We now have support for hardware watchdogs, configurable in
3906 * The scheduled shutdown logic now has a public API.
3908 * We now mount /tmp as tmpfs by default, but this can be
3909 masked and /etc/fstab can override it.
3911 * Since udisks does not make use of /media anymore we are not
3912 mounting a tmpfs on it anymore.
3914 * journalctl gained a new --local switch to only interleave
3915 locally generated journal files.
3917 * We can now load the IMA policy at boot automatically.
3919 * The GTK tools have been split off into a systemd-ui.
3921 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3922 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3923 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3924 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3925 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3926 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3927 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3928 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3929 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3934 * This is mostly a bugfix release
3936 * Support optional initialization of the machine ID from the
3937 KVM or container configured UUID.
3939 * Support immediate reboots with "systemctl reboot -ff"
3941 * Show /etc/os-release data in systemd-analyze output
3943 * Many bugfixes for the journal, including endianness fixes and
3944 ensuring that disk space enforcement works
3946 * sd-login.h is C++ comptaible again
3948 * Extend the /etc/os-release format on request of the Debian
3951 * We now refuse non-UTF8 strings used in various configuration
3952 and unit files. This is done to ensure we do not pass invalid
3953 data over D-Bus or expose it elsewhere.
3955 * Register Mimo USB Screens as suitable for automatic seat
3958 * Read SELinux client context from journal clients in a race
3961 * Reorder configuration file lookup order. /etc now always
3962 overrides /run in order to allow the administrator to always
3963 and unconditionally override vendor supplied or
3964 automatically generated data.
3966 * The various user visible bits of the journal now have man
3967 pages. We still lack man pages for the journal API calls
3970 * We now ship all man pages in HTML format again in the
3973 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3974 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3975 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3980 * This is mostly a bugfix release
3982 * systems lacking /etc/os-release are no longer supported.
3984 * Various functionality updates to libsystemd-login.so
3986 * Track class of PAM logins to distuingish greeters from
3989 Contributions from: Kay Sievers, Lennart Poettering, Michael
3994 * This is an important bugfix release for v41.
3996 * Building man pages is now optional which should be useful
3997 for those building systemd from git but unwilling to install
4000 * Watchdog support for supervising services is now usable. In
4001 a future release support for hardware watchdogs
4002 (i.e. /dev/watchdog) will be added building on this.
4004 * Service start rate limiting is now configurable and can be
4005 turned off per service. When a start rate limit is hit a
4006 reboot can automatically be triggered.
4008 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4010 Contributions from: Benjamin Franzke, Bill Nottingham,
4011 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4012 Schmidt, Michał Górny, Piotr Drąg
4016 * The systemd binary is installed /usr/lib/systemd/systemd now;
4017 An existing /sbin/init symlink needs to be adapted with the
4020 * The code that loads kernel modules has been ported to invoke
4021 libkmod directly, instead of modprobe. This means we do not
4022 support systems with module-init-tools anymore.
4024 * Watchdog support is now already useful, but still not
4027 * A new kernel command line option systemd.setenv= is
4028 understood to set system wide environment variables
4029 dynamically at boot.
4031 * We now limit the set of capabilities of systemd-journald.
4033 * We now set SIGPIPE to ignore by default, since it only is
4034 useful in shell pipelines, and has little use in general
4035 code. This can be disabled with IgnoreSIPIPE=no in unit
4038 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4039 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4044 * This is mostly a bugfix release
4046 * We now expose the reason why a service failed in the
4047 "Result" D-Bus property.
4049 * Rudimentary service watchdog support (will be completed over
4050 the next few releases.)
4052 * When systemd forks off in order execute some service we will
4053 now immediately changes its argv[0] to reflect which process
4054 it will execute. This is useful to minimize the time window
4055 with a generic argv[0], which makes bootcharts more useful
4057 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4058 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4059 Mike Kazantsev, Ray Strode
4063 * This is mostly a test release, but incorporates many
4066 * New systemd-cgtop tool to show control groups by their
4069 * Linking against libacl for ACLs is optional again. If
4070 disabled, support tracking device access for active logins
4071 goes becomes unavailable, and so does access to the user
4072 journals by the respective users.
4074 * If a group "adm" exists, journal files are automatically
4075 owned by them, thus allow members of this group full access
4076 to the system journal as well as all user journals.
4078 * The journal now stores the SELinux context of the logging
4079 client for all entries.
4081 * Add C++ inclusion guards to all public headers
4083 * New output mode "cat" in the journal to print only text
4084 messages, without any meta data like date or time.
4086 * Include tiny X server wrapper as a temporary stop-gap to
4087 teach XOrg udev display enumeration. This is used by display
4088 managers such as gdm, and will go away as soon as XOrg
4089 learned native udev hotplugging for display devices.
4091 * Add new systemd-cat tool for executing arbitrary programs
4092 with STDERR/STDOUT connected to the journal. Can also act as
4093 BSD logger replacement, and does so by default.
4095 * Optionally store all locally generated coredumps in the
4096 journal along with meta data.
4098 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4099 writing short strings to files (for usage for /sys), and for
4100 creating symlinks, character and block device nodes.
4102 * New unit file option ControlGroupPersistent= to make cgroups
4103 persistent, following the mechanisms outlined in
4104 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4106 * Support multiple local RTCs in a sane way
4108 * No longer monopolize IO when replaying readahead data on
4109 rotating disks, since we might starve non-file-system IO to
4110 death, since fanotify() will not see accesses done by blkid,
4113 * Do not show kernel threads in systemd-cgls anymore, unless
4114 requested with new -k switch.
4116 Contributions from: Dan Horák, Kay Sievers, Lennart
4117 Poettering, Michal Schmidt
4121 * This is mostly a test release, but incorporates many
4124 * The git repository moved to:
4125 git://anongit.freedesktop.org/systemd/systemd
4126 ssh://git.freedesktop.org/git/systemd/systemd
4128 * First release with the journal
4129 http://0pointer.de/blog/projects/the-journal.html
4131 * The journal replaces both systemd-kmsg-syslogd and
4132 systemd-stdout-bridge.
4134 * New sd_pid_get_unit() API call in libsystemd-logind
4136 * Many systemadm clean-ups
4138 * Introduce remote-fs-pre.target which is ordered before all
4139 remote mounts and may be used to start services before all
4142 * Added Mageia support
4144 * Add bash completion for systemd-loginctl
4146 * Actively monitor PID file creation for daemons which exit in
4147 the parent process before having finished writing the PID
4148 file in the daemon process. Daemons which do this need to be
4149 fixed (i.e. PID file creation must have finished before the
4150 parent exits), but we now react a bit more gracefully to them.
4152 * Add colourful boot output, mimicking the well-known output
4153 of existing distributions.
4155 * New option PassCredentials= for socket units, for
4156 compatibility with a recent kernel ABI breakage.
4158 * /etc/rc.local is now hooked in via a generator binary, and
4159 thus will no longer act as synchronization point during
4162 * systemctl list-unit-files now supports --root=.
4164 * systemd-tmpfiles now understands two new commands: z, Z for
4165 relabelling files according to the SELinux database. This is
4166 useful to apply SELinux labels to specific files in /sys,
4169 * Output of SysV services is now forwarded to both the console
4170 and the journal by default, not only just the console.
4172 * New man pages for all APIs from libsystemd-login.
4174 * The build tree got reorganized and a the build system is a
4175 lot more modular allowing embedded setups to specifically
4176 select the components of systemd they are interested in.
4178 * Support for Linux systems lacking the kernel VT subsystem is
4181 * configure's --with-rootdir= got renamed to
4182 --with-rootprefix= to follow the naming used by udev and
4185 * Unless specified otherwise we will now install to /usr instead
4186 of /usr/local by default.
4188 * Processes with '@' in argv[0][0] are now excluded from the
4189 final shut-down killing spree, following the logic explained
4191 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4193 * All processes remaining in a service cgroup when we enter
4194 the START or START_PRE states are now killed with
4195 SIGKILL. That means it is no longer possible to spawn
4196 background processes from ExecStart= lines (which was never
4197 supported anyway, and bad style).
4199 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4200 reloading of units together.
4202 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4203 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4204 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4205 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4206 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek