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 * journalctl gained a new switch, --flush, that flushes
11 logs from /run/log/journal to /var/log/journal if
12 persistent storage is enabled.
14 * Services can notify the manager before they start a reload
15 (by sending RELOADING=1) or shutdown (by sending
16 STOPPING=1). This allows the manager to track and show the
17 internal state of daemons and closes a race condition when
18 the process is still running but has closed its D-Bus
21 * Services with Type=oneshot do not have to have any ExecStart
24 * User units are now loaded also from
25 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
26 /run/systemd/user directory that was already previously
27 supported, but is under the control of the user.
29 * Job timeouts (i.e. time-outs on the time a job that is
30 queued stays in the run queue) can now optionally result in
31 immediate reboot or power-off actions (JobTimeoutAction= and
32 JobTimeoutRebootArgument=). This is useful on ".target"
33 units, to limit the maximum time a target remains
34 undispatched in the run queue, and to trigger an emergency
35 operation in such a case. This is now used by default to
36 turn off the system if boot-up (as defined by everything in
37 basic.target) hangs and does not complete for at least
38 15min. Also, if power-off or reboot hang for at least 30min
39 an immediate power-off/reboot operation is triggered. This
40 functionality is particularly useful to increase reliability
41 on embedded devices, but also on laptops which might
42 accidentally get powered on when carried in a backpack and
43 whose boot stays stuck in a hard disk encryption passphrase
46 * systemd-logind can be configured to also handle lid switch
47 events even when the machine is docked or multiple displays
48 are attached (HandleLidSwitchDocked= option).
50 * A helper binary and a service have been added which can be
51 used to resume from hibernation in the initramfs. A
52 generator will parse the resume= option on the kernel
53 command-line to trigger resume.
55 * A user console daemon systemd-consoled has been
56 added. Currently, it is a preview, and will so far open a
57 single terminal on each session of the user marked as
58 Desktop=systemd-console.
60 * Route metrics can be specified for DHCP routes added by
63 * The SELinux context of socket-activated services can be set
64 from the information provided by the networking stack
65 (SELinuxContextFromNet= option).
67 * Userspace firmware loading support has been removed and
68 the minimum supported kernel version is thus bumped to 3.7.
70 * Timeout for udev workers has been increased from 1 to 3
71 minutes, but a warning will be printed after 1 minute to
72 help diagnose kernel modules that take a long time to load.
74 * Udev rules can now remove tags on devices with TAG-="foobar".
76 * systemd's readahead implementation has been removed. In many
77 circumstances it didn't give expected benefits even for
78 rotational disk drives and was becoming less relevant in the
79 age of SSDs. As none of the developers has been using
80 rotating media anymore, and nobody stepped up to actively
81 maintain this component of systemd it has now been removed.
83 * Swap units can use Options= to specify discard options.
84 Discard options specified for swaps in /etc/fstab are now
87 * Docker containers are now detected as a separate type of
90 * The Password Agent protocol gained support for queries where
91 the user input is shown, useful e.g. for user names.
92 systemd-ask-password gained a new --echo option to turn that
95 * The default sysctl.d/ snippets will now set:
97 net.core.default_qdisc = fq_codel
99 This selects Fair Queuing Controlled Delay as the default
100 queuing discipline for network interfaces. fq_codel helps
101 fight the network bufferbloat problem. It is believed to be
102 a good default with no tuning required for most workloads.
103 Downstream distributions may override this choice. On 10Gbit
104 servers that do not do forwarding, "fq" may perform better.
105 Systems without a good clocksource should use "pfifo_fast".
107 * If kdbus is enabled during build a new option BusPolicy= is
108 available for service units, that allows locking all service
109 processes into a stricter bus policy, in order to limit
110 access to various bus services, or even hide most of them
111 from the service's view entirely.
113 * networkctl will now show the .network and .link file
114 networkd has applied to a specific interface.
116 * sd-login gained a new API call sd_session_get_desktop() to
117 query which desktop environment has been selected for a
120 * UNIX utmp support is now compile-time optional to support
123 * systemctl gained two new commands "add-wants" and
124 "add-requires" for pulling in units from specific targets
127 * If the word "rescue" is specified on the kernel command line
128 the system will now boot into rescue mode (aka
129 rescue.target), which was previously available only by
130 specifying "1" or "systemd.unit=rescue.target" on the kernel
131 command line. This new kernel command line option nicely
132 mirrors the already existing "emergency" kernel command line
135 * New kernel command line options mount.usr=, mount.usrflags=,
136 mount.usrfstype= have been added that match root=, rootflags=,
137 rootfstype= but allow mounting a specific file system to
140 * The $NOTIFY_SOCKET is now also passed to control processes of
141 services, not only the main process.
143 * This version reenables support for fsck's -l switch. This
144 means at least version v2.25 of util-linux is required for
145 operation, otherwise dead-locks on device nodes may
146 occur. Again: you need to update util-linux to at least
147 v2.25 when updating systemd to v217.
149 * The "multi-seat-x" tool has been removed from systemd, as
150 its functionality has been integrated into X servers 1.16,
151 and the tool is hence redundant. It is recommended to update
152 display managers invoking this tool to simply invoke X
153 directly from now on, again.
155 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
156 message flag has been added for all of systemd's PolicyKit
157 authenticated method calls has been added. In particular
158 this now allows optional interactive authorization via
159 PolicyKit for many of PID1's privileged operations such as
160 unit file enabling and disabling.
162 * "udevadm hwdb --update" learnt a new switch "--usr" for
163 placing the rebuilt hardware database in /usr instead of
164 /etc. When used only hardware database entries stored in
165 /usr will be used, and any user database entries in /etc are
166 ignored. This functionality is useful for vendors to ship a
167 pre-built database on systems where local configuration is
168 unnecessary or unlikely.
170 * Calendar time specifications in .timer units now also
171 understand the strings "semi-annually", "quarterly" and
172 "minutely" as shortcuts (in addition to the preexisting
173 "anually", "hourly", ...).
175 * systemd-tmpfiles will now correctly create files in /dev
176 at boot which are marked for creation only at boot. It is
177 recommended to always create static device nodes with 'c!'
178 and 'b!', so that they are created only at boot and not
179 overwritten at runtime.
181 * When the watchdog logic is used for a service (WatchdogSec=)
182 and the watchdog timeout is hit the service will now be
183 terminated with SIGABRT (instead of just SIGTERM), in order
184 to make sure a proper coredump and backtrace is
185 generated. This ensures that hanging services will result in
186 similar coredump/backtrace behaviour as services that hit a
189 Contributions from: Andreas Henriksson, Andrei Borzenkov,
190 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
191 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
192 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
193 Herrmann, David Sommerseth, David Strauss, Emil Renner
194 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
195 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
196 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
197 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
198 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
199 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
200 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
201 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
202 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
203 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
204 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
205 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
206 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
207 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
208 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
209 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
212 -- Berlin, 2014-10-28
216 * timedated no longer reads NTP implementation unit names from
217 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
218 implementations should add a
220 Conflicts=systemd-timesyncd.service
222 to their unit files to take over and replace systemd's NTP
223 default functionality.
225 * systemd-sysusers gained a new line type "r" for configuring
226 which UID/GID ranges to allocate system users/groups
227 from. Lines of type "u" may now add an additional column
228 that specifies the home directory for the system user to be
229 created. Also, systemd-sysusers may now optionally read user
230 information from STDIN instead of a file. This is useful for
231 invoking it from RPM preinst scriptlets that need to create
232 users before the first RPM file is installed since these
233 files might need to be owned by them. A new
234 %sysusers_create_inline RPM macro has been introduced to do
235 just that. systemd-sysusers now updates the shadow files as
236 well as the user/group databases, which should enhance
237 compatibility with certain tools like grpck.
239 * A number of bus APIs of PID 1 now optionally consult
240 PolicyKit to permit access for otherwise unprivileged
241 clients under certain conditions. Note that this currently
242 doesn't support interactive authentication yet, but this is
243 expected to be added eventually, too.
245 * /etc/machine-info now has new fields for configuring the
246 deployment environment of the machine, as well as the
247 location of the machine. hostnamectl has been updated with
248 new command to update these fields.
250 * systemd-timesyncd has been updated to automatically acquire
251 NTP server information from systemd-networkd, which might
252 have been discovered via DHCP.
254 * systemd-resolved now includes a caching DNS stub resolver
255 and a complete LLMNR name resolution implementation. A new
256 NSS module "nss-resolve" has been added which can be used
257 instead of glibc's own "nss-dns" to resolve hostnames via
258 systemd-resolved. Hostnames, addresses and arbitrary RRs may
259 be resolved via systemd-resolved D-Bus APIs. In contrast to
260 the glibc internal resolver systemd-resolved is aware of
261 multi-homed system, and keeps DNS server and caches separate
262 and per-interface. Queries are sent simultaneously on all
263 interfaces that have DNS servers configured, in order to
264 properly handle VPNs and local LANs which might resolve
265 separate sets of domain names. systemd-resolved may acquire
266 DNS server information from systemd-networkd automatically,
267 which in turn might have discovered them via DHCP. A tool
268 "systemd-resolve-host" has been added that may be used to
269 query the DNS logic in resolved. systemd-resolved implements
270 IDNA and automatically uses IDNA or UTF-8 encoding depending
271 on whether classic DNS or LLMNR is used as transport. In the
272 next releases we intend to add a DNSSEC and mDNS/DNS-SD
273 implementation to systemd-resolved.
275 * A new NSS module nss-mymachines has been added, that
276 automatically resolves the names of all local registered
277 containers to their respective IP addresses.
279 * A new client tool "networkctl" for systemd-networkd has been
280 added. It currently is entirely passive and will query
281 networking configuration from udev, rtnetlink and networkd,
282 and present it to the user in a very friendly
283 way. Eventually, we hope to extend it to become a full
284 control utility for networkd.
286 * .socket units gained a new DeferAcceptSec= setting that
287 controls the kernels' TCP_DEFER_ACCEPT sockopt for
288 TCP. Similar, support for controlling TCP keep-alive
289 settings has been added (KeepAliveTimeSec=,
290 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
291 turning off Nagle's algorithm on TCP has been added
294 * logind learned a new session type "web", for use in projects
295 like Cockpit which register web clients as PAM sessions.
297 * timer units with at least one OnCalendar= setting will now
298 be started only after timer-sync.target has been
299 reached. This way they will not elapse before the system
300 clock has been corrected by a local NTP client or
301 similar. This is particular useful on RTC-less embedded
302 machines, that come up with an invalid system clock.
304 * systemd-nspawn's --network-veth= switch should now result in
305 stable MAC addresses for both the outer and the inner side
308 * systemd-nspawn gained a new --volatile= switch for running
309 container instances with /etc or /var unpopulated.
311 * The kdbus client code has been updated to use the new Linux
312 3.17 memfd subsystem instead of the old kdbus-specific one.
314 * systemd-networkd's DHCP client and server now support
315 FORCERENEW. There are also new configuration options to
316 configure the vendor client identifier and broadcast mode
319 * systemd will no longer inform the kernel about the current
320 timezone, as this is necessarily incorrect and racy as the
321 kernel has no understanding of DST and similar
322 concepts. This hence means FAT timestamps will be always
323 considered UTC, similar to what Android is already
324 doing. Also, when the RTC is configured to the local time
325 (rather than UTC) systemd will never synchronize back to it,
326 as this might confuse Windows at a later boot.
328 * systemd-analyze gained a new command "verify" for offline
329 validation of unit files.
331 * systemd-networkd gained support for a couple of additional
332 settings for bonding networking setups. Also, the metric for
333 statically configured routes may now be configured. For
334 network interfaces where this is appropriate the peer IP
335 address may now be configured.
337 * systemd-networkd's DHCP client will no longer request
338 broadcasting by default, as this tripped up some networks.
339 For hardware where broadcast is required the feature should
340 be switched back on using RequestBroadcast=yes.
342 * systemd-networkd will now set up IPv4LL addresses (when
343 enabled) even if DHCP is configured successfully.
345 * udev will now default to respect network device names given
346 by the kernel when the kernel indicates that these are
347 predictable. This behavior can be tweaked by changing
348 NamePolicy= in the relevant .link file.
350 * A new library systemd-terminal has been added that
351 implements full TTY stream parsing and rendering. This
352 library is supposed to be used later on for implementing a
353 full userspace VT subsystem, replacing the current kernel
356 * A new tool systemd-journal-upload has been added to push
357 journal data to a remote system running
358 systemd-journal-remote.
360 * journald will no longer forward all local data to another
361 running syslog daemon. This change has been made because
362 rsyslog (which appears to be the most commonly used syslog
363 implementation these days) no longer makes use of this, and
364 instead pulls the data out of the journal on its own. Since
365 forwarding the messages to a non-existent syslog server is
366 more expensive than we assumed we have now turned this
367 off. If you run a syslog server that is not a recent rsyslog
368 version, you have to turn this option on again
369 (ForwardToSyslog= in journald.conf).
371 * journald now optionally supports the LZ4 compressor for
372 larger journal fields. This compressor should perform much
373 better than XZ which was the previous default.
375 * machinectl now shows the IP addresses of local containers,
376 if it knows them, plus the interface name of the container.
378 * A new tool "systemd-escape" has been added that makes it
379 easy to escape strings to build unit names and similar.
381 * sd_notify() messages may now include a new ERRNO= field
382 which is parsed and collected by systemd and shown among the
383 "systemctl status" output for a service.
385 * A new component "systemd-firstboot" has been added that
386 queries the most basic systemd information (timezone,
387 hostname, root password) interactively on first
388 boot. Alternatively it may also be used to provision these
389 things offline on OS images installed into directories.
391 * The default sysctl.d/ snippets will now set
393 net.ipv4.conf.default.promote_secondaries=1
395 This has the benefit of no flushing secondary IP addresses
396 when primary addresses are removed.
398 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
399 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
400 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
401 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
402 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
403 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
404 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
405 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
406 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
407 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
408 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
409 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
410 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
411 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
412 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
414 -- Berlin, 2014-08-19
418 * A new tool systemd-sysusers has been added. This tool
419 creates system users and groups in /etc/passwd and
420 /etc/group, based on static declarative system user/group
421 definitions in /usr/lib/sysusers.d/. This is useful to
422 enable factory resets and volatile systems that boot up with
423 an empty /etc directory, and thus need system users and
424 groups created during early boot. systemd now also ships
425 with two default sysusers.d/ files for the most basic
426 users and groups systemd and the core operating system
429 * A new tmpfiles snippet has been added that rebuilds the
430 essential files in /etc on boot, should they be missing.
432 * A directive for ensuring automatic clean-up of
433 /var/cache/man/ has been removed from the default
434 configuration. This line should now be shipped by the man
435 implementation. The necessary change has been made to the
436 man-db implementation. Note that you need to update your man
437 implementation to one that ships this line, otherwise no
438 automatic clean-up of /var/cache/man will take place.
440 * A new condition ConditionNeedsUpdate= has been added that
441 may conditionalize services to only run when /etc or /var
442 are "older" than the vendor operating system resources in
443 /usr. This is useful for reconstructing or updating /etc
444 after an offline update of /usr or a factory reset, on the
445 next reboot. Services that want to run once after such an
446 update or reset should use this condition and order
447 themselves before the new systemd-update-done.service, which
448 will mark the two directories as fully updated. A number of
449 service files have been added making use of this, to rebuild
450 the udev hardware database, the journald message catalog and
451 dynamic loader cache (ldconfig). The systemd-sysusers tool
452 described above also makes use of this now. With this in
453 place it is now possible to start up a minimal operating
454 system with /etc empty cleanly. For more information on the
455 concepts involved see this recent blog story:
457 http://0pointer.de/blog/projects/stateless.html
459 * A new system group "input" has been introduced, and all
460 input device nodes get this group assigned. This is useful
461 for system-level software to get access to input devices. It
462 complements what is already done for "audio" and "video".
464 * systemd-networkd learnt minimal DHCPv4 server support in
465 addition to the existing DHCPv4 client support. It also
466 learnt DHCPv6 client and IPv6 Router Solicitation client
467 support. The DHCPv4 client gained support for static routes
468 passed in from the server. Note that the [DHCPv4] section
469 known in older systemd-networkd versions has been renamed to
470 [DHCP] and is now also used by the DHCPv6 client. Existing
471 .network files using settings of this section should be
472 updated, though compatibility is maintained. Optionally, the
473 client hostname may now be sent to the DHCP server.
475 * networkd gained support for vxlan virtual networks as well
476 as tun/tap and dummy devices.
478 * networkd gained support for automatic allocation of address
479 ranges for interfaces from a system-wide pool of
480 addresses. This is useful for dynamically managing a large
481 number of interfaces with a single network configuration
482 file. In particular this is useful to easily assign
483 appropriate IP addresses to the veth links of a large number
486 * RPM macros for processing sysusers, sysctl and binfmt
487 drop-in snippets at package installation time have been
490 * The /etc/os-release file should now be placed in
491 /usr/lib/os-release. The old location is automatically
492 created as symlink. /usr/lib is the more appropriate
493 location of this file, since it shall actually describe the
494 vendor operating system shipped in /usr, and not the
495 configuration stored in /etc.
497 * .mount units gained a new boolean SloppyOptions= setting
498 that maps to mount(8)'s -s option which enables permissive
499 parsing of unknown mount options.
501 * tmpfiles learnt a new "L+" directive which creates a symlink
502 but (unlike "L") deletes a pre-existing file first, should
503 it already exist and not already be the correct
504 symlink. Similar, "b+", "c+" and "p+" directives have been
505 added as well, which create block and character devices, as
506 well as fifos in the filesystem, possibly removing any
507 pre-existing files of different types.
509 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
510 'argument' field (which so far specified the source to
511 symlink/copy the files from) is now optional. If omitted the
512 same file os copied from /usr/share/factory/ suffixed by the
513 full destination path. This is useful for populating /etc
514 with essential files, by copying them from vendor defaults
515 shipped in /usr/share/factory/etc.
517 * A new command "systemctl preset-all" has been added that
518 applies the service preset settings to all installed unit
519 files. A new switch --preset-mode= has been added that
520 controls whether only enable or only disable operations
523 * A new command "systemctl is-system-running" has been added
524 that allows checking the overall state of the system, for
525 example whether it is fully up and running.
527 * When the system boots up with an empty /etc, the equivalent
528 to "systemctl preset-all" is executed during early boot, to
529 make sure all default services are enabled after a factory
532 * systemd now contains a minimal preset file that enables the
533 most basic services systemd ships by default.
535 * Unit files' [Install] section gained a new DefaultInstance=
536 field for defining the default instance to create if a
537 template unit is enabled with no instance specified.
539 * A new passive target cryptsetup-pre.target has been added
540 that may be used by services that need to make they run and
541 finish before the first LUKS cryptographic device is set up.
543 * The /dev/loop-control and /dev/btrfs-control device nodes
544 are now owned by the "disk" group by default, opening up
545 access to this group.
547 * systemd-coredump will now automatically generate a
548 stack trace of all core dumps taking place on the system,
549 based on elfutils' libdw library. This stack trace is logged
552 * systemd-coredump may now optionally store coredumps directly
553 on disk (in /var/lib/systemd/coredump, possibly compressed),
554 instead of storing them unconditionally in the journal. This
555 mode is the new default. A new configuration file
556 /etc/systemd/coredump.conf has been added to configure this
557 and other parameters of systemd-coredump.
559 * coredumpctl gained a new "info" verb to show details about a
560 specific coredump. A new switch "-1" has also been added
561 that makes sure to only show information about the most
562 recent entry instead of all entries. Also, as the tool is
563 generally useful now the "systemd-" prefix of the binary
564 name has been removed. Distributions that want to maintain
565 compatibility with the old name should add a symlink from
566 the old name to the new name.
568 * journald's SplitMode= now defaults to "uid". This makes sure
569 that unprivileged users can access their own coredumps with
570 coredumpctl without restrictions.
572 * New kernel command line options "systemd.wants=" (for
573 pulling an additional unit during boot), "systemd.mask="
574 (for masking a specific unit for the boot), and
575 "systemd.debug-shell" (for enabling the debug shell on tty9)
576 have been added. This is implemented in the new generator
577 "systemd-debug-generator".
579 * systemd-nspawn will now by default filter a couple of
580 syscalls for containers, among them those required for
581 kernel module loading, direct x86 IO port access, swap
582 management, and kexec. Most importantly though
583 open_by_handle_at() is now prohibited for containers,
584 closing a hole similar to a recently discussed vulnerability
585 in docker regarding access to files on file hierarchies the
586 container should normally not have access to. Note that for
587 nspawn we generally make no security claims anyway (and
588 this is explicitly documented in the man page), so this is
589 just a fix for one of the most obvious problems.
591 * A new man page file-hierarchy(7) has been added that
592 contains a minimized, modernized version of the file system
593 layout systemd expects, similar in style to the FHS
594 specification or hier(5). A new tool systemd-path(1) has
595 been added to query many of these paths for the local
598 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
599 longer done. Since the directory now has a per-user size
600 limit, and is cleaned on logout this appears unnecessary,
601 in particular since this now brings the lifecycle of this
602 directory closer in line with how IPC objects are handled.
604 * systemd.pc now exports a number of additional directories,
605 including $libdir (which is useful to identify the library
606 path for the primary architecture of the system), and a
607 couple of drop-in directories.
609 * udev's predictable network interface names now use the dev_port
610 sysfs attribute, introduced in linux 3.15 instead of dev_id to
611 distinguish between ports of the same PCI function. dev_id should
612 only be used for ports using the same HW address, hence the need
615 * machined has been updated to export the OS version of a
616 container (read from /etc/os-release and
617 /usr/lib/os-release) on the bus. This is now shown in
618 "machinectl status" for a machine.
620 * A new service setting RestartForceExitStatus= has been
621 added. If configured to a set of exit signals or process
622 return values, the service will be restarted when the main
623 daemon process exits with any of them, regardless of the
626 * systemctl's -H switch for connecting to remote systemd
627 machines has been extended so that it may be used to
628 directly connect to a specific container on the
629 host. "systemctl -H root@foobar:waldi" will now connect as
630 user "root" to host "foobar", and then proceed directly to
631 the container named "waldi". Note that currently you have to
632 authenticate as user "root" for this to work, as entering
633 containers is a privileged operation.
635 Contributions from: Andreas Henriksson, Benjamin Steinwender,
636 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
637 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
638 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
639 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
640 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
641 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
642 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
643 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
644 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
645 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
646 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
648 -- Berlin, 2014-07-03
652 * As an experimental feature, udev now tries to lock the
653 disk device node (flock(LOCK_SH|LOCK_NB)) while it
654 executes events for the disk or any of its partitions.
655 Applications like partitioning programs can lock the
656 disk device node (flock(LOCK_EX)) and claim temporary
657 device ownership that way; udev will entirely skip all event
658 handling for this disk and its partitions. If the disk
659 was opened for writing, the close will trigger a partition
660 table rescan in udev's "watch" facility, and if needed
661 synthesize "change" events for the disk and all its partitions.
662 This is now unconditionally enabled, and if it turns out to
663 cause major problems, we might turn it on only for specific
664 devices, or might need to disable it entirely. Device Mapper
665 devices are excluded from this logic.
667 * We temporarily dropped the "-l" switch for fsck invocations,
668 since they collide with the flock() logic above. util-linux
669 upstream has been changed already to avoid this conflict,
670 and we will readd "-l" as soon as util-linux with this
671 change has been released.
673 * The dependency on libattr has been removed. Since a long
674 time, the extended attribute calls have moved to glibc, and
675 libattr is thus unnecessary.
677 * Virtualization detection works without priviliges now. This
678 means the systemd-detect-virt binary no longer requires
679 CAP_SYS_PTRACE file capabilities, and our daemons can run
680 with fewer privileges.
682 * systemd-networkd now runs under its own "systemd-network"
683 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
684 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
685 loses the ability to write to files owned by root this way.
687 * Similar, systemd-resolved now runs under its own
688 "systemd-resolve" user with no capabilities remaining.
690 * Similar, systemd-bus-proxyd now runs under its own
691 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
693 * systemd-networkd gained support for setting up "veth"
694 virtual ethernet devices for container connectivity, as well
695 as GRE and VTI tunnels.
697 * systemd-networkd will no longer automatically attempt to
698 manually load kernel modules necessary for certain tunnel
699 transports. Instead, it is assumed the kernel loads them
700 automatically when required. This only works correctly on
701 very new kernels. On older kernels, please consider adding
702 the kernel modules to /etc/modules-load.d/ as a work-around.
704 * The resolv.conf file systemd-resolved generates has been
705 moved to /run/systemd/resolve/. If you have a symlink from
706 /etc/resolv.conf, it might be necessary to correct it.
708 * Two new service settings, ProtectHome= and ProtectSystem=,
709 have been added. When enabled, they will make the user data
710 (such as /home) inaccessible or read-only and the system
711 (such as /usr) read-only, for specific services. This allows
712 very light-weight per-service sandboxing to avoid
713 modifications of user data or system files from
714 services. These two new switches have been enabled for all
715 of systemd's long-running services, where appropriate.
717 * Socket units gained new SocketUser= and SocketGroup=
718 settings to set the owner user and group of AF_UNIX sockets
719 and FIFOs in the file system.
721 * Socket units gained a new RemoveOnStop= setting. If enabled,
722 all FIFOS and sockets in the file system will be removed
723 when the specific socket unit is stopped.
725 * Socket units gained a new Symlinks= setting. It takes a list
726 of symlinks to create to file system sockets or FIFOs
727 created by the specific Unix sockets. This is useful to
728 manage symlinks to socket nodes with the same life-cycle as
731 * The /dev/log socket and /dev/initctl FIFO have been moved to
732 /run, and have been replaced by symlinks. This allows
733 connecting to these facilities even if PrivateDevices=yes is
734 used for a service (which makes /dev/log itself unavailable,
735 but /run is left). This also has the benefit of ensuring
736 that /dev only contains device nodes, directories and
737 symlinks, and nothing else.
739 * sd-daemon gained two new calls sd_pid_notify() and
740 sd_pid_notifyf(). They are similar to sd_notify() and
741 sd_notifyf(), but allow overriding of the source PID of
742 notification messages if permissions permit this. This is
743 useful to send notify messages on behalf of a different
744 process (for example, the parent process). The
745 systemd-notify tool has been updated to make use of this
746 when sending messages (so that notification messages now
747 originate from the shell script invoking systemd-notify and
748 not the systemd-notify process itself. This should minimize
749 a race where systemd fails to associate notification
750 messages to services when the originating process already
753 * A new "on-abnormal" setting for Restart= has been added. If
754 set, it will result in automatic restarts on all "abnormal"
755 reasons for a process to exit, which includes unclean
756 signals, core dumps, timeouts and watchdog timeouts, but
757 does not include clean and unclean exit codes or clean
758 signals. Restart=on-abnormal is an alternative for
759 Restart=on-failure for services that shall be able to
760 terminate and avoid restarts on certain errors, by
761 indicating so with an unclean exit code. Restart=on-failure
762 or Restart=on-abnormal is now the recommended setting for
763 all long-running services.
765 * If the InaccessibleDirectories= service setting points to a
766 mount point (or if there are any submounts contained within
767 it), it is now attempted to completely unmount it, to make
768 the file systems truly unavailable for the respective
771 * The ReadOnlyDirectories= service setting and
772 systemd-nspawn's --read-only parameter are now recursively
773 applied to all submounts, too.
775 * Mount units may now be created transiently via the bus APIs.
777 * The support for SysV and LSB init scripts has been removed
778 from the systemd daemon itself. Instead, it is now
779 implemented as a generator that creates native systemd units
780 from these scripts when needed. This enables us to remove a
781 substantial amount of legacy code from PID 1, following the
782 fact that many distributions only ship a very small number
783 of LSB/SysV init scripts nowadays.
785 * Priviliged Xen (dom0) domains are not considered
786 virtualization anymore by the virtualization detection
787 logic. After all, they generally have unrestricted access to
788 the hardware and usually are used to manage the unprivileged
791 * systemd-tmpfiles gained a new "C" line type, for copying
792 files or entire directories.
794 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
795 lines. So far, they have been non-globbing versions of the
796 latter, and have thus been redundant. In future, it is
797 recommended to only use "z". "m" has hence been removed
798 from the documentation, even though it stays supported.
800 * A tmpfiles snippet to recreate the most basic structure in
801 /var has been added. This is enough to create the /var/run →
802 /run symlink and create a couple of structural
803 directories. This allows systems to boot up with an empty or
804 volatile /var. Of course, while with this change, the core OS
805 now is capable with dealing with a volatile /var, not all
806 user services are ready for it. However, we hope that sooner
807 or later, many service daemons will be changed upstream so
808 that they are able to automatically create their necessary
809 directories in /var at boot, should they be missing. This is
810 the first step to allow state-less systems that only require
811 the vendor image for /usr to boot.
813 * systemd-nspawn has gained a new --tmpfs= switch to mount an
814 empty tmpfs instance to a specific directory. This is
815 particularly useful for making use of the automatic
816 reconstruction of /var (see above), by passing --tmpfs=/var.
818 * Access modes specified in tmpfiles snippets may now be
819 prefixed with "~", which indicates that they shall be masked
820 by whether the existing file or directory is currently
821 writable, readable or executable at all. Also, if specified,
822 the sgid/suid/sticky bits will be masked for all
825 * A new passive target unit "network-pre.target" has been
826 added which is useful for services that shall run before any
827 network is configured, for example firewall scripts.
829 * The "floppy" group that previously owned the /dev/fd*
830 devices is no longer used. The "disk" group is now used
831 instead. Distributions should probably deprecate usage of
834 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
835 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
836 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
837 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
838 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
839 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
842 -- Berlin, 2014-06-11
846 * A new "systemd-timesyncd" daemon has been added for
847 synchronizing the system clock across the network. It
848 implements an SNTP client. In contrast to NTP
849 implementations such as chrony or the NTP reference server,
850 this only implements a client side, and does not bother with
851 the full NTP complexity, focusing only on querying time from
852 one remote server and synchronizing the local clock to
853 it. Unless you intend to serve NTP to networked clients or
854 want to connect to local hardware clocks, this simple NTP
855 client should be more than appropriate for most
856 installations. The daemon runs with minimal privileges, and
857 has been hooked up with networkd to only operate when
858 network connectivity is available. The daemon saves the
859 current clock to disk every time a new NTP sync has been
860 acquired, and uses this to possibly correct the system clock
861 early at bootup, in order to accommodate for systems that
862 lack an RTC such as the Raspberry Pi and embedded devices,
863 and to make sure that time monotonically progresses on these
864 systems, even if it is not always correct. To make use of
865 this daemon, a new system user and group "systemd-timesync"
866 needs to be created on installation of systemd.
868 * The queue "seqnum" interface of libudev has been disabled, as
869 it was generally incompatible with device namespacing as
870 sequence numbers of devices go "missing" if the devices are
871 part of a different namespace.
873 * "systemctl list-timers" and "systemctl list-sockets" gained
874 a --recursive switch for showing units of these types also
875 for all local containers, similar in style to the already
876 supported --recursive switch for "systemctl list-units".
878 * A new RebootArgument= setting has been added for service
879 units, which may be used to specify a kernel reboot argument
880 to use when triggering reboots with StartLimitAction=.
882 * A new FailureAction= setting has been added for service
883 units which may be used to specify an operation to trigger
884 when a service fails. This works similarly to
885 StartLimitAction=, but unlike it, controls what is done
886 immediately rather than only after several attempts to
887 restart the service in question.
889 * hostnamed got updated to also expose the kernel name,
890 release, and version on the bus. This is useful for
891 executing commands like hostnamectl with the -H switch.
892 systemd-analyze makes use of this to properly display
893 details when running non-locally.
895 * The bootchart tool can now show cgroup information in the
898 * The CFS CPU quota cgroup attribute is now exposed for
899 services. The new CPUQuota= switch has been added for this
900 which takes a percentage value. Setting this will have the
901 result that a service may never get more CPU time than the
902 specified percentage, even if the machine is otherwise idle.
904 * systemd-networkd learned IPIP and SIT tunnel support.
906 * LSB init scripts exposing a dependency on $network will now
907 get a dependency on network-online.target rather than simply
908 network.target. This should bring LSB handling closer to
909 what it was on SysV systems.
911 * A new fsck.repair= kernel option has been added to control
912 how fsck shall deal with unclean file systems at boot.
914 * The (.ini) configuration file parser will now silently
915 ignore sections whose name begins with "X-". This may be
916 used to maintain application-specific extension sections in unit
919 * machined gained a new API to query the IP addresses of
920 registered containers. "machinectl status" has been updated
921 to show these addresses in its output.
923 * A new call sd_uid_get_display() has been added to the
924 sd-login APIs for querying the "primary" session of a
925 user. The "primary" session of the user is elected from the
926 user's sessions and generally a graphical session is
927 preferred over a text one.
929 * A minimal systemd-resolved daemon has been added. It
930 currently simply acts as a companion to systemd-networkd and
931 manages resolv.conf based on per-interface DNS
932 configuration, possibly supplied via DHCP. In the long run
933 we hope to extend this into a local DNSSEC enabled DNS and
936 * The systemd-networkd-wait-online tool is now enabled by
937 default. It will delay network-online.target until a network
938 connection has been configured. The tool primarily integrates
939 with networkd, but will also make a best effort to make sense
940 of network configuration performed in some other way.
942 * Two new service options StartupCPUShares= and
943 StartupBlockIOWeight= have been added that work similarly to
944 CPUShares= and BlockIOWeight= however only apply during
945 system startup. This is useful to prioritize certain services
946 differently during bootup than during normal runtime.
948 * hostnamed has been changed to prefer the statically
949 configured hostname in /etc/hostname (unless set to
950 'localhost' or empty) over any dynamic one supplied by
951 dhcp. With this change, the rules for picking the hostname
952 match more closely the rules of other configuration settings
953 where the local administrator's configuration in /etc always
954 overrides any other settings.
956 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
957 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
958 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
959 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
960 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
961 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
962 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
963 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
964 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
965 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
966 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
967 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
968 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
969 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
970 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
971 Lindskog, WaLyong Cho, Will Woods, Zbigniew
974 -- Beijing, 2014-05-28
978 * When restoring the screen brightness at boot, stay away from
979 the darkest setting or from the lowest 5% of the available
980 range, depending on which is the larger value of both. This
981 should effectively protect the user from rebooting into a
982 black screen, should the brightness have been set to minimum
985 * sd-login gained a new sd_machine_get_class() call to
986 determine the class ("vm" or "container") of a machine
987 registered with machined.
989 * sd-login gained new calls
990 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
991 to query the identity of the peer of a local AF_UNIX
992 connection. They operate similarly to their sd_pid_get_xyz()
995 * PID 1 will now maintain a system-wide system state engine
996 with the states "starting", "running", "degraded",
997 "maintenance", "stopping". These states are bound to system
998 startup, normal runtime, runtime with at least one failed
999 service, rescue/emergency mode and system shutdown. This
1000 state is shown in the "systemctl status" output when no unit
1001 name is passed. It is useful to determine system state, in
1002 particularly when doing so for many systems or containers at
1005 * A new command "list-machines" has been added to "systemctl"
1006 that lists all local OS containers and shows their system
1007 state (see above), if systemd runs inside of them.
1009 * systemctl gained a new "-r" switch to recursively enumerate
1010 units on all local containers, when used with the
1011 "list-unit" command (which is the default one that is
1012 executed when no parameters are specified).
1014 * The GPT automatic partition discovery logic will now honour
1015 two GPT partition flags: one may be set on a partition to
1016 cause it to be mounted read-only, and the other may be set
1017 on a partition to ignore it during automatic discovery.
1019 * Two new GPT type UUIDs have been added for automatic root
1020 partition discovery, for 32-bit and 64-bit ARM. This is not
1021 particularly useful for discovering the root directory on
1022 these architectures during bare-metal boots (since UEFI is
1023 not common there), but still very useful to allow booting of
1024 ARM disk images in nspawn with the -i option.
1026 * MAC addresses of interfaces created with nspawn's
1027 --network-interface= switch will now be generated from the
1028 machine name, and thus be stable between multiple invocations
1031 * logind will now automatically remove all IPC objects owned
1032 by a user if she or he fully logs out. This makes sure that
1033 users who are logged out cannot continue to consume IPC
1034 resources. This covers SysV memory, semaphores and message
1035 queues as well as POSIX shared memory and message
1036 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1037 limits. With this functionality, that is corrected. This may
1038 be turned off by using the RemoveIPC= switch of logind.conf.
1040 * The systemd-machine-id-setup and tmpfiles tools gained a
1041 --root= switch to operate on a specific root directory,
1044 * journald can now forward logged messages to the TTYs of all
1045 logged in users ("wall"). This is the default for all
1046 emergency messages now.
1048 * A new tool systemd-journal-remote has been added to stream
1049 journal log messages across the network.
1051 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1052 controller trees are mounted into it. Note that the
1053 directories mounted beneath it are not read-only. This is a
1054 security measure and is particularly useful because glibc
1055 actually includes a search logic to pick any tmpfs it can
1056 find to implement shm_open() if /dev/shm is not available
1057 (which it might very well be in namespaced setups).
1059 * machinectl gained a new "poweroff" command to cleanly power
1060 down a local OS container.
1062 * The PrivateDevices= unit file setting will now also drop the
1063 CAP_MKNOD capability from the capability bound set, and
1064 imply DevicePolicy=closed.
1066 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1067 comprehensively on all long-running systemd services where
1068 this is appropriate.
1070 * systemd-udevd will now run in a disassociated mount
1071 namespace. To mount directories from udev rules, make sure to
1072 pull in mount units via SYSTEMD_WANTS properties.
1074 * The kdbus support gained support for uploading policy into
1075 the kernel. sd-bus gained support for creating "monitoring"
1076 connections that can eavesdrop into all bus communication
1077 for debugging purposes.
1079 * Timestamps may now be specified in seconds since the UNIX
1080 epoch Jan 1st, 1970 by specifying "@" followed by the value
1083 * Native tcpwrap support in systemd has been removed. tcpwrap
1084 is old code, not really maintained anymore and has serious
1085 shortcomings, and better options such as firewalls
1086 exist. For setups that require tcpwrap usage, please
1087 consider invoking your socket-activated service via tcpd,
1088 like on traditional inetd.
1090 * A new system.conf configuration option
1091 DefaultTimerAccuracySec= has been added that controls the
1092 default AccuracySec= setting of .timer units.
1094 * Timer units gained a new WakeSystem= switch. If enabled,
1095 timers configured this way will cause the system to resume
1096 from system suspend (if the system supports that, which most
1099 * Timer units gained a new Persistent= switch. If enabled,
1100 timers configured this way will save to disk when they have
1101 been last triggered. This information is then used on next
1102 reboot to possible execute overdue timer events, that
1103 could not take place because the system was powered off.
1104 This enables simple anacron-like behaviour for timer units.
1106 * systemctl's "list-timers" will now also list the time a
1107 timer unit was last triggered in addition to the next time
1108 it will be triggered.
1110 * systemd-networkd will now assign predictable IPv4LL
1111 addresses to its local interfaces.
1113 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1114 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1115 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1116 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1117 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1118 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1119 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1120 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1123 -- Berlin, 2014-03-25
1127 * A new unit file setting RestrictAddressFamilies= has been
1128 added to restrict which socket address families unit
1129 processes gain access to. This takes address family names
1130 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1131 attack surface of services via exotic protocol stacks. This
1132 is built on seccomp system call filters.
1134 * Two new unit file settings RuntimeDirectory= and
1135 RuntimeDirectoryMode= have been added that may be used to
1136 manage a per-daemon runtime directories below /run. This is
1137 an alternative for setting up directory permissions with
1138 tmpfiles snippets, and has the advantage that the runtime
1139 directory's lifetime is bound to the daemon runtime and that
1140 the daemon starts up with an empty directory each time. This
1141 is particularly useful when writing services that drop
1142 privileges using the User= or Group= setting.
1144 * The DeviceAllow= unit setting now supports globbing for
1145 matching against device group names.
1147 * The systemd configuration file system.conf gained new
1148 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1149 DefaultMemoryAccounting= to globally turn on/off accounting
1150 for specific resources (cgroups) for all units. These
1151 settings may still be overridden individually in each unit
1154 * systemd-gpt-auto-generator is now able to discover /srv and
1155 root partitions in addition to /home and swap partitions. It
1156 also supports LUKS-encrypted partitions now. With this in
1157 place, automatic discovery of partitions to mount following
1158 the Discoverable Partitions Specification
1159 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1160 is now a lot more complete. This allows booting without
1161 /etc/fstab and without root= on the kernel command line on
1162 systems prepared appropriately.
1164 * systemd-nspawn gained a new --image= switch which allows
1165 booting up disk images and Linux installations on any block
1166 device that follow the Discoverable Partitions Specification
1167 (see above). This means that installations made with
1168 appropriately updated installers may now be started and
1169 deployed using container managers, completely
1170 unmodified. (We hope that libvirt-lxc will add support for
1171 this feature soon, too.)
1173 * systemd-nspawn gained a new --network-macvlan= setting to
1174 set up a private macvlan interface for the
1175 container. Similarly, systemd-networkd gained a new
1176 Kind=macvlan setting in .netdev files.
1178 * systemd-networkd now supports configuring local addresses
1181 * A new tool systemd-network-wait-online has been added to
1182 synchronously wait for network connectivity using
1185 * The sd-bus.h bus API gained a new sd_bus_track object for
1186 tracking the life-cycle of bus peers. Note that sd-bus.h is
1187 still not a public API though (unless you specify
1188 --enable-kdbus on the configure command line, which however
1189 voids your warranty and you get no API stability guarantee).
1191 * The $XDG_RUNTIME_DIR runtime directories for each user are
1192 now individual tmpfs instances, which has the benefit of
1193 introducing separate pools for each user, with individual
1194 size limits, and thus making sure that unprivileged clients
1195 can no longer negatively impact the system or other users by
1196 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1197 RuntimeDirectorySize= has been introduced that allows
1198 controlling the default size limit for all users. It
1199 defaults to 10% of the available physical memory. This is no
1200 replacement for quotas on tmpfs though (which the kernel
1201 still does not support), as /dev/shm and /tmp are still
1202 shared resources used by both the system and unprivileged
1205 * logind will now automatically turn off automatic suspending
1206 on laptop lid close when more than one display is
1207 connected. This was previously expected to be implemented
1208 individually in desktop environments (such as GNOME),
1209 however has been added to logind now, in order to fix a
1210 boot-time race where a desktop environment might not have
1211 been started yet and thus not been able to take an inhibitor
1212 lock at the time where logind already suspends the system
1213 due to a closed lid.
1215 * logind will now wait at least 30s after each system
1216 suspend/resume cycle, and 3min after system boot before
1217 suspending the system due to a closed laptop lid. This
1218 should give USB docking stations and similar enough time to
1219 be probed and configured after system resume and boot in
1220 order to then act as suspend blocker.
1222 * systemd-run gained a new --property= setting which allows
1223 initialization of resource control properties (and others)
1224 for the created scope or service unit. Example: "systemd-run
1225 --property=BlockIOWeight=10 updatedb" may be used to run
1226 updatedb at a low block IO scheduling weight.
1228 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1229 now also work in --scope mode.
1231 * When systemd is compiled with kdbus support, basic support
1232 for enforced policies is now in place. (Note that enabling
1233 kdbus still voids your warranty and no API compatibility
1236 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1237 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1238 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1239 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1240 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1241 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1242 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1243 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1244 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1245 Zbigniew Jędrzejewski-Szmek
1247 -- Berlin, 2014-03-12
1251 * systemd will now relabel /dev after loading the SMACK policy
1252 according to SMACK rules.
1254 * A new unit file option AppArmorProfile= has been added to
1255 set the AppArmor profile for the processes of a unit.
1257 * A new condition check ConditionArchitecture= has been added
1258 to conditionalize units based on the system architecture, as
1259 reported by uname()'s "machine" field.
1261 * systemd-networkd now supports matching on the system
1262 virtualization, architecture, kernel command line, host name
1265 * logind is now a lot more aggressive when suspending the
1266 machine due to a closed laptop lid. Instead of acting only
1267 on the lid close action, it will continuously watch the lid
1268 status and act on it. This is useful for laptops where the
1269 power button is on the outside of the chassis so that it can
1270 be reached without opening the lid (such as the Lenovo
1271 Yoga). On those machines, logind will now immediately
1272 re-suspend the machine if the power button has been
1273 accidentally pressed while the laptop was suspended and in a
1274 backpack or similar.
1276 * logind will now watch SW_DOCK switches and inhibit reaction
1277 to the lid switch if it is pressed. This means that logind
1278 will not suspend the machine anymore if the lid is closed
1279 and the system is docked, if the laptop supports SW_DOCK
1280 notifications via the input layer. Note that ACPI docking
1281 stations do not generate this currently. Also note that this
1282 logic is usually not fully sufficient and Desktop
1283 Environments should take a lid switch inhibitor lock when an
1284 external display is connected, as systemd will not watch
1287 * nspawn will now make use of the devices cgroup controller by
1288 default, and only permit creation of and access to the usual
1289 API device nodes like /dev/null or /dev/random, as well as
1290 access to (but not creation of) the pty devices.
1292 * We will now ship a default .network file for
1293 systemd-networkd that automatically configures DHCP for
1294 network interfaces created by nspawn's --network-veth or
1295 --network-bridge= switches.
1297 * systemd will now understand the usual M, K, G, T suffixes
1298 according to SI conventions (i.e. to the base 1000) when
1299 referring to throughput and hardware metrics. It will stay
1300 with IEC conventions (i.e. to the base 1024) for software
1301 metrics, according to what is customary according to
1302 Wikipedia. We explicitly document which base applies for
1303 each configuration option.
1305 * The DeviceAllow= setting in unit files now supports a syntax
1306 to whitelist an entire group of devices node majors at once,
1307 based on the /proc/devices listing. For example, with the
1308 string "char-pts", it is now possible to whitelist all
1309 current and future pseudo-TTYs at once.
1311 * sd-event learned a new "post" event source. Event sources of
1312 this type are triggered by the dispatching of any event
1313 source of a type that is not "post". This is useful for
1314 implementing clean-up and check event sources that are
1315 triggered by other work being done in the program.
1317 * systemd-networkd is no longer statically enabled, but uses
1318 the usual [Install] sections so that it can be
1319 enabled/disabled using systemctl. It still is enabled by
1322 * When creating a veth interface pair with systemd-nspawn, the
1323 host side will now be prefixed with "vb-" if
1324 --network-bridge= is used, and with "ve-" if --network-veth
1325 is used. This way, it is easy to distinguish these cases on
1326 the host, for example to apply different configuration to
1327 them with systemd-networkd.
1329 * The compatibility libraries for libsystemd-journal.so,
1330 libsystem-id128.so, libsystemd-login.so and
1331 libsystemd-daemon.so do not make use of IFUNC
1332 anymore. Instead, we now build libsystemd.so multiple times
1333 under these alternative names. This means that the footprint
1334 is drastically increased, but given that these are
1335 transitional compatibility libraries, this should not matter
1336 much. This change has been made necessary to support the ARM
1337 platform for these compatibility libraries, as the ARM
1338 toolchain is not really at the same level as the toolchain
1339 for other architectures like x86 and does not support
1340 IFUNC. Please make sure to use --enable-compat-libs only
1341 during a transitional period!
1343 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1344 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1345 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1346 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1347 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1348 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1349 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1350 Zbigniew Jędrzejewski-Szmek
1352 -- Berlin, 2014-02-24
1356 * A new component "systemd-networkd" has been added that can
1357 be used to configure local network interfaces statically or
1358 via DHCP. It is capable of bringing up bridges, VLANs, and
1359 bonding. Currently, no hook-ups for interactive network
1360 configuration are provided. Use this for your initrd,
1361 container, embedded, or server setup if you need a simple,
1362 yet powerful, network configuration solution. This
1363 configuration subsystem is quite nifty, as it allows wildcard
1364 hotplug matching in interfaces. For example, with a single
1365 configuration snippet, you can configure that all Ethernet
1366 interfaces showing up are automatically added to a bridge,
1367 or similar. It supports link-sensing and more.
1369 * A new tool "systemd-socket-proxyd" has been added which can
1370 act as a bidirectional proxy for TCP sockets. This is
1371 useful for adding socket activation support to services that
1372 do not actually support socket activation, including virtual
1373 machines and the like.
1375 * Add a new tool to save/restore rfkill state on
1378 * Save/restore state of keyboard backlights in addition to
1379 display backlights on shutdown/boot.
1381 * udev learned a new SECLABEL{} construct to label device
1382 nodes with a specific security label when they appear. For
1383 now, only SECLABEL{selinux} is supported, but the syntax is
1384 prepared for additional security frameworks.
1386 * udev gained a new scheme to configure link-level attributes
1387 from files in /etc/systemd/network/*.link. These files can
1388 match against MAC address, device path, driver name and type,
1389 and will apply attributes like the naming policy, link speed,
1390 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1391 address assignment policy (randomized, ...).
1393 * The configuration of network interface naming rules for
1394 "permanent interface names" has changed: a new NamePolicy=
1395 setting in the [Link] section of .link files determines the
1396 priority of possible naming schemes (onboard, slot, mac,
1397 path). The default value of this setting is determined by
1398 /usr/lib/net/links/99-default.link. Old
1399 80-net-name-slot.rules udev configuration file has been
1400 removed, so local configuration overriding this file should
1401 be adapated to override 99-default.link instead.
1403 * When the User= switch is used in a unit file, also
1404 initialize $SHELL= based on the user database entry.
1406 * systemd no longer depends on libdbus. All communication is
1407 now done with sd-bus, systemd's low-level bus library
1410 * kdbus support has been added to PID 1 itself. When kdbus is
1411 enabled, this causes PID 1 to set up the system bus and
1412 enable support for a new ".busname" unit type that
1413 encapsulates bus name activation on kdbus. It works a little
1414 bit like ".socket" units, except for bus names. A new
1415 generator has been added that converts classic dbus1 service
1416 activation files automatically into native systemd .busname
1419 * sd-bus: add a light-weight vtable implementation that allows
1420 defining objects on the bus with a simple static const
1421 vtable array of its methods, signals and properties.
1423 * systemd will not generate or install static dbus
1424 introspection data anymore to /usr/share/dbus-1/interfaces,
1425 as the precise format of these files is unclear, and
1426 nothing makes use of it.
1428 * A proxy daemon is now provided to proxy clients connecting
1429 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1430 compatibility with classic D-Bus.
1432 * A bus driver implementation has been added that supports the
1433 classic D-Bus bus driver calls on kdbus, also for
1434 compatibility purposes.
1436 * A new API "sd-event.h" has been added that implements a
1437 minimal event loop API built around epoll. It provides a
1438 couple of features that direct epoll usage is lacking:
1439 prioritization of events, scales to large numbers of timer
1440 events, per-event timer slack (accuracy), system-wide
1441 coalescing of timer events, exit handlers, watchdog
1442 supervision support using systemd's sd_notify() API, child
1445 * A new API "sd-rntl.h" has been added that provides an API
1446 around the route netlink interface of the kernel, similar in
1447 style to "sd-bus.h".
1449 * A new API "sd-dhcp-client.h" has been added that provides a
1450 small DHCPv4 client-side implementation. This is used by
1453 * There is a new kernel command line option
1454 "systemd.restore_state=0|1". When set to "0", none of the
1455 systemd tools will restore saved runtime state to hardware
1456 devices. More specifically, the rfkill and backlight states
1459 * The FsckPassNo= compatibility option in mount/service units
1460 has been removed. The fstab generator will now add the
1461 necessary dependencies automatically, and does not require
1462 PID1's support for that anymore.
1464 * journalctl gained a new switch, --list-boots, that lists
1465 recent boots with their times and boot IDs.
1467 * The various tools like systemctl, loginctl, timedatectl,
1468 busctl, systemd-run, ... have gained a new switch "-M" to
1469 connect to a specific, local OS container (as direct
1470 connection, without requiring SSH). This works on any
1471 container that is registered with machined, such as those
1472 created by libvirt-lxc or nspawn.
1474 * systemd-run and systemd-analyze also gained support for "-H"
1475 to connect to remote hosts via SSH. This is particularly
1476 useful for systemd-run because it enables queuing of jobs
1477 onto remote systems.
1479 * machinectl gained a new command "login" to open a getty
1480 login in any local container. This works with any container
1481 that is registered with machined (such as those created by
1482 libvirt-lxc or nspawn), and which runs systemd inside.
1484 * machinectl gained a new "reboot" command that may be used to
1485 trigger a reboot on a specific container that is registered
1486 with machined. This works on any container that runs an init
1487 system of some kind.
1489 * systemctl gained a new "list-timers" command to print a nice
1490 listing of installed timer units with the times they elapse
1493 * Alternative reboot() parameters may now be specified on the
1494 "systemctl reboot" command line and are passed to the
1495 reboot() system call.
1497 * systemctl gained a new --job-mode= switch to configure the
1498 mode to queue a job with. This is a more generic version of
1499 --fail, --irreversible, and --ignore-dependencies, which are
1500 still available but not advertised anymore.
1502 * /etc/systemd/system.conf gained new settings to configure
1503 various default timeouts of units, as well as the default
1504 start limit interval and burst. These may still be overridden
1507 * PID1 will now export on the bus profile data of the security
1508 policy upload process (such as the SELinux policy upload to
1511 * journald: when forwarding logs to the console, include
1512 timestamps (following the setting in
1513 /sys/module/printk/parameters/time).
1515 * OnCalendar= in timer units now understands the special
1516 strings "yearly" and "annually". (Both are equivalent)
1518 * The accuracy of timer units is now configurable with the new
1519 AccuracySec= setting. It defaults to 1min.
1521 * A new dependency type JoinsNamespaceOf= has been added that
1522 allows running two services within the same /tmp and network
1523 namespace, if PrivateNetwork= or PrivateTmp= are used.
1525 * A new command "cat" has been added to systemctl. It outputs
1526 the original unit file of a unit, and concatenates the
1527 contents of additional "drop-in" unit file snippets, so that
1528 the full configuration is shown.
1530 * systemctl now supports globbing on the various "list-xyz"
1531 commands, like "list-units" or "list-sockets", as well as on
1532 those commands which take multiple unit names.
1534 * journalctl's --unit= switch gained support for globbing.
1536 * All systemd daemons now make use of the watchdog logic so
1537 that systemd automatically notices when they hang.
1539 * If the $container_ttys environment variable is set,
1540 getty-generator will automatically spawn a getty for each
1541 listed tty. This is useful for container managers to request
1542 login gettys to be spawned on as many ttys as needed.
1544 * %h, %s, %U specifier support is not available anymore when
1545 used in unit files for PID 1. This is because NSS calls are
1546 not safe from PID 1. They stay available for --user
1547 instances of systemd, and as special case for the root user.
1549 * loginctl gained a new "--no-legend" switch to turn off output
1552 * The "sd-login.h" API gained three new calls:
1553 sd_session_is_remote(), sd_session_get_remote_user(),
1554 sd_session_get_remote_host() to query information about
1557 * The udev hardware database now also carries vendor/product
1558 information of SDIO devices.
1560 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1561 determine whether watchdog notifications are requested by
1564 * Socket-activated per-connection services now include a
1565 short description of the connection parameters in the
1568 * tmpfiles gained a new "--boot" option. When this is not used,
1569 only lines where the command character is not suffixed with
1570 "!" are executed. When this option is specified, those
1571 options are executed too. This partitions tmpfiles
1572 directives into those that can be safely executed at any
1573 time, and those which should be run only at boot (for
1574 example, a line that creates /run/nologin).
1576 * A new API "sd-resolve.h" has been added which provides a simple
1577 asynchronous wrapper around glibc NSS host name resolution
1578 calls, such as getaddrinfo(). In contrast to glibc's
1579 getaddrinfo_a(), it does not use signals. In contrast to most
1580 other asynchronous name resolution libraries, this one does
1581 not reimplement DNS, but reuses NSS, so that alternate
1582 host name resolution systems continue to work, such as mDNS,
1583 LDAP, etc. This API is based on libasyncns, but it has been
1584 cleaned up for inclusion in systemd.
1586 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1587 "sd-daemon.h" are no longer found in individual libraries
1588 libsystemd-journal.so, libsystemd-login.so,
1589 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1590 merged them into a single library, libsystemd.so, which
1591 provides all symbols. The reason for this is cyclic
1592 dependencies, as these libraries tend to use each other's
1593 symbols. So far, we have managed to workaround that by linking
1594 a copy of a good part of our code into each of these
1595 libraries again and again, which, however, makes certain
1596 things hard to do, like sharing static variables. Also, it
1597 substantially increases footprint. With this change, there
1598 is only one library for the basic APIs systemd
1599 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1600 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1601 library as well, however are subject to the --enable-kdbus
1602 switch (see below). Note that "sd-dhcp-client.h" is not part
1603 of this library (this is because it only consumes, never
1604 provides, services of/to other APIs). To make the transition
1605 easy from the separate libraries to the unified one, we
1606 provide the --enable-compat-libs compile-time switch which
1607 will generate stub libraries that are compatible with the
1608 old ones but redirect all calls to the new one.
1610 * All of the kdbus logic and the new APIs "sd-bus.h",
1611 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1612 and "sd-utf8.h" are compile-time optional via the
1613 "--enable-kdbus" switch, and they are not compiled in by
1614 default. To make use of kdbus, you have to explicitly enable
1615 the switch. Note however, that neither the kernel nor the
1616 userspace API for all of this is considered stable yet. We
1617 want to maintain the freedom to still change the APIs for
1618 now. By specifying this build-time switch, you acknowledge
1619 that you are aware of the instability of the current
1622 * Also, note that while kdbus is pretty much complete,
1623 it lacks one thing: proper policy support. This means you
1624 can build a fully working system with all features; however,
1625 it will be highly insecure. Policy support will be added in
1626 one of the next releases, at the same time that we will
1627 declare the APIs stable.
1629 * When the kernel command-line argument "kdbus" is specified,
1630 systemd will automatically load the kdbus.ko kernel module. At
1631 this stage of development, it is only useful for testing kdbus
1632 and should not be used in production. Note: if "--enable-kdbus"
1633 is specified, and the kdbus.ko kernel module is available, and
1634 "kdbus" is added to the kernel command line, the entire system
1635 runs with kdbus instead of dbus-daemon, with the above mentioned
1636 problem of missing the system policy enforcement. Also a future
1637 version of kdbus.ko or a newer systemd will not be compatible with
1638 each other, and will unlikely be able to boot the machine if only
1639 one of them is updated.
1641 * systemctl gained a new "import-environment" command which
1642 uploads the caller's environment (or parts thereof) into the
1643 service manager so that it is inherited by services started
1644 by the manager. This is useful to upload variables like
1645 $DISPLAY into the user service manager.
1647 * A new PrivateDevices= switch has been added to service units
1648 which allows running a service with a namespaced /dev
1649 directory that does not contain any device nodes for
1650 physical devices. More specifically, it only includes devices
1651 such as /dev/null, /dev/urandom, and /dev/zero which are API
1654 * logind has been extended to support behaviour like VT
1655 switching on seats that do not support a VT. This makes
1656 multi-session available on seats that are not the first seat
1657 (seat0), and on systems where kernel support for VTs has
1658 been disabled at compile-time.
1660 * If a process holds a delay lock for system sleep or shutdown
1661 and fails to release it in time, we will now log its
1662 identity. This makes it easier to identify processes that
1663 cause slow suspends or power-offs.
1665 * When parsing /etc/crypttab, support for a new key-slot=
1666 option as supported by Debian is added. It allows indicating
1667 which LUKS slot to use on disk, speeding up key loading.
1669 * The sd_journald_sendv() API call has been checked and
1670 officially declared to be async-signal-safe so that it may
1671 be invoked from signal handlers for logging purposes.
1673 * Boot-time status output is now enabled automatically after a
1674 short timeout if boot does not progress, in order to give
1675 the user an indication what she or he is waiting for.
1677 * The boot-time output has been improved to show how much time
1678 remains until jobs expire.
1680 * The KillMode= switch in service units gained a new possible
1681 value "mixed". If set, and the unit is shut down, then the
1682 initial SIGTERM signal is sent only to the main daemon
1683 process, while the following SIGKILL signal is sent to
1684 all remaining processes of the service.
1686 * When a scope unit is registered, a new property "Controller"
1687 may be set. If set to a valid bus name, systemd will send a
1688 RequestStop() signal to this name when it would like to shut
1689 down the scope. This may be used to hook manager logic into
1690 the shutdown logic of scope units. Also, scope units may now
1691 be put in a special "abandoned" state, in which case the
1692 manager process which created them takes no further
1693 responsibilities for it.
1695 * When reading unit files, systemd will now verify
1696 the access mode of these files, and warn about certain
1697 suspicious combinations. This has been added to make it
1698 easier to track down packaging bugs where unit files are
1699 marked executable or world-writable.
1701 * systemd-nspawn gained a new "--setenv=" switch to set
1702 container-wide environment variables. The similar option in
1703 systemd-activate was renamed from "--environment=" to
1704 "--setenv=" for consistency.
1706 * systemd-nspawn has been updated to create a new kdbus domain
1707 for each container that is invoked, thus allowing each
1708 container to have its own set of system and user buses,
1709 independent of the host.
1711 * systemd-nspawn gained a new --drop-capability= switch to run
1712 the container with less capabilities than the default. Both
1713 --drop-capability= and --capability= now take the special
1714 string "all" for dropping or keeping all capabilities.
1716 * systemd-nspawn gained new switches for executing containers
1717 with specific SELinux labels set.
1719 * systemd-nspawn gained a new --quiet switch to not generate
1720 any additional output but the container's own console
1723 * systemd-nspawn gained a new --share-system switch to run a
1724 container without PID namespacing enabled.
1726 * systemd-nspawn gained a new --register= switch to control
1727 whether the container is registered with systemd-machined or
1728 not. This is useful for containers that do not run full
1729 OS images, but only specific apps.
1731 * systemd-nspawn gained a new --keep-unit which may be used
1732 when invoked as the only program from a service unit, and
1733 results in registration of the unit service itself in
1734 systemd-machined, instead of a newly opened scope unit.
1736 * systemd-nspawn gained a new --network-interface= switch for
1737 moving arbitrary interfaces to the container. The new
1738 --network-veth switch creates a virtual Ethernet connection
1739 between host and container. The new --network-bridge=
1740 switch then allows assigning the host side of this virtual
1741 Ethernet connection to a bridge device.
1743 * systemd-nspawn gained a new --personality= switch for
1744 setting the kernel personality for the container. This is
1745 useful when running a 32-bit container on a 64-bit host. A
1746 similar option Personality= is now also available for service
1749 * logind will now also track a "Desktop" identifier for each
1750 session which encodes the desktop environment of it. This is
1751 useful for desktop environments that want to identify
1752 multiple running sessions of itself easily.
1754 * A new SELinuxContext= setting for service units has been
1755 added that allows setting a specific SELinux execution
1756 context for a service.
1758 * Most systemd client tools will now honour $SYSTEMD_LESS for
1759 settings of the "less" pager. By default, these tools will
1760 override $LESS to allow certain operations to work, such as
1761 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1762 influence this logic.
1764 * systemd's "seccomp" hook-up has been changed to make use of
1765 the libseccomp library instead of using its own
1766 implementation. This has benefits for portability among
1769 * For usage together with SystemCallFilter=, a new
1770 SystemCallErrorNumber= setting has been introduced that
1771 allows configuration of a system error number to be returned
1772 on filtered system calls, instead of immediately killing the
1773 process. Also, SystemCallArchitectures= has been added to
1774 limit access to system calls of a particular architecture
1775 (in order to turn off support for unused secondary
1776 architectures). There is also a global
1777 SystemCallArchitectures= setting in system.conf now to turn
1778 off support for non-native system calls system-wide.
1780 * systemd requires a kernel with a working name_to_handle_at(),
1781 please see the kernel config requirements in the README file.
1783 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1784 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1785 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1786 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1787 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1788 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1789 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1790 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1791 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1792 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1793 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1794 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1795 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1796 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1797 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1798 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1799 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1800 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1801 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1802 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1803 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1804 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1805 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1806 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1808 -- Berlin, 2014-02-20
1812 * logind has gained support for facilitating privileged input
1813 and drm device access for unprivileged clients. This work is
1814 useful to allow Wayland display servers (and similar
1815 programs, such as kmscon) to run under the user's ID and
1816 access input and drm devices which are normally
1817 protected. When this is used (and the kernel is new enough)
1818 logind will "mute" IO on the file descriptors passed to
1819 Wayland as long as it is in the background and "unmute" it
1820 if it returns into the foreground. This allows secure
1821 session switching without allowing background sessions to
1822 eavesdrop on input and display data. This also introduces
1823 session switching support if VT support is turned off in the
1824 kernel, and on seats that are not seat0.
1826 * A new kernel command line option luks.options= is understood
1827 now which allows specifying LUKS options for usage for LUKS
1828 encrypted partitions specified with luks.uuid=.
1830 * tmpfiles.d(5) snippets may now use specifier expansion in
1831 path names. More specifically %m, %b, %H, %v, are now
1832 replaced by the local machine id, boot id, hostname, and
1833 kernel version number.
1835 * A new tmpfiles.d(5) command "m" has been introduced which
1836 may be used to change the owner/group/access mode of a file
1837 or directory if it exists, but do nothing if it does not.
1839 * This release removes high-level support for the
1840 MemorySoftLimit= cgroup setting. The underlying kernel
1841 cgroup attribute memory.soft_limit= is currently badly
1842 designed and likely to be removed from the kernel API in its
1843 current form, hence we should not expose it for now.
1845 * The memory.use_hierarchy cgroup attribute is now enabled for
1846 all cgroups systemd creates in the memory cgroup
1847 hierarchy. This option is likely to be come the built-in
1848 default in the kernel anyway, and the non-hierarchial mode
1849 never made much sense in the intrinsically hierarchial
1852 * A new field _SYSTEMD_SLICE= is logged along with all journal
1853 messages containing the slice a message was generated
1854 from. This is useful to allow easy per-customer filtering of
1855 logs among other things.
1857 * systemd-journald will no longer adjust the group of journal
1858 files it creates to the "systemd-journal" group. Instead we
1859 rely on the journal directory to be owned by the
1860 "systemd-journal" group, and its setgid bit set, so that the
1861 kernel file system layer will automatically enforce that
1862 journal files inherit this group assignment. The reason for
1863 this change is that we cannot allow NSS look-ups from
1864 journald which would be necessary to resolve
1865 "systemd-journal" to a numeric GID, because this might
1866 create deadlocks if NSS involves synchronous queries to
1867 other daemons (such as nscd, or sssd) which in turn are
1868 logging clients of journald and might block on it, which
1869 would then dead lock. A tmpfiles.d(5) snippet included in
1870 systemd will make sure the setgid bit and group are
1871 properly set on the journal directory if it exists on every
1872 boot. However, we recommend adjusting it manually after
1873 upgrades too (or from RPM scriptlets), so that the change is
1874 not delayed until next reboot.
1876 * Backlight and random seed files in /var/lib/ have moved into
1877 the /var/lib/systemd/ directory, in order to centralize all
1878 systemd generated files in one directory.
1880 * Boot time performance measurements (as displayed by
1881 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1882 performance information if that's available to determine how
1883 much time BIOS and boot loader initialization required. With
1884 a sufficiently new BIOS you hence no longer need to boot
1885 with Gummiboot to get access to such information.
1887 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1888 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1889 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1890 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1891 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1892 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1893 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1895 -- Berlin, 2013-10-02
1899 * The Restart= option for services now understands a new
1900 on-watchdog setting, which will restart the service
1901 automatically if the service stops sending out watchdog keep
1902 alive messages (as configured with WatchdogSec=).
1904 * The getty generator (which is responsible for bringing up a
1905 getty on configured serial consoles) will no longer only
1906 start a getty on the primary kernel console but on all
1907 others, too. This makes the order in which console= is
1908 specified on the kernel command line less important.
1910 * libsystemd-logind gained a new sd_session_get_vt() call to
1911 retrieve the VT number of a session.
1913 * If the option "tries=0" is set for an entry of /etc/crypttab
1914 its passphrase is queried indefinitely instead of any
1915 maximum number of tries.
1917 * If a service with a configure PID file terminates its PID
1918 file will now be removed automatically if it still exists
1919 afterwards. This should put an end to stale PID files.
1921 * systemd-run will now also take relative binary path names
1922 for execution and no longer insists on absolute paths.
1924 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1925 paths that are optionally prefixed with "-" to indicate that
1926 it should not be considered a failure if they do not exist.
1928 * journalctl -o (and similar commands) now understands a new
1929 output mode "short-precise", it is similar to "short" but
1930 shows timestamps with usec accuracy.
1932 * The option "discard" (as known from Debian) is now
1933 synonymous to "allow-discards" in /etc/crypttab. In fact,
1934 "discard" is preferred now (since it is easier to remember
1937 * Some licensing clean-ups were made, so that more code is now
1938 LGPL-2.1 licensed than before.
1940 * A minimal tool to save/restore the display backlight
1941 brightness across reboots has been added. It will store the
1942 backlight setting as late as possible at shutdown, and
1943 restore it as early as possible during reboot.
1945 * A logic to automatically discover and enable home and swap
1946 partitions on GPT disks has been added. With this in place
1947 /etc/fstab becomes optional for many setups as systemd can
1948 discover certain partitions located on the root disk
1949 automatically. Home partitions are recognized under their
1950 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1951 partitions are recognized under their GPT type ID
1952 0657fd6da4ab43c484e50933c84b4f4f.
1954 * systemd will no longer pass any environment from the kernel
1955 or initrd to system services. If you want to set an
1956 environment for all services, do so via the kernel command
1957 line systemd.setenv= assignment.
1959 * The systemd-sysctl tool no longer natively reads the file
1960 /etc/sysctl.conf. If desired, the file should be symlinked
1961 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1962 legacy support by a symlink rather than built-in code, it
1963 also makes the otherwise hidden order of application of the
1964 different files visible. (Note that this partly reverts to a
1965 pre-198 application order of sysctl knobs!)
1967 * The "systemctl set-log-level" and "systemctl dump" commands
1968 have been moved to systemd-analyze.
1970 * systemd-run learned the new --remain-after-exit switch,
1971 which causes the scope unit not to be cleaned up
1972 automatically after the process terminated.
1974 * tmpfiles learned a new --exclude-prefix= switch to exclude
1975 certain paths from operation.
1977 * journald will now automatically flush all messages to disk
1978 as soon as a message at the log level CRIT, ALERT or EMERG
1981 Contributions from: Andrew Cook, Brandon Philips, Christian
1982 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1983 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1984 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1985 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1986 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1987 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1988 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1989 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1990 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1991 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1992 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1993 William Giokas, Zbigniew Jędrzejewski-Szmek
1995 -- Berlin, 2013-09-13
1999 * The documentation has been updated to cover the various new
2000 concepts introduced with 205.
2002 * Unit files now understand the new %v specifier which
2003 resolves to the kernel version string as returned by "uname
2006 * systemctl now supports filtering the unit list output by
2007 load state, active state and sub state, using the new
2010 * "systemctl status" will now show the results of the
2011 condition checks (like ConditionPathExists= and similar) of
2012 the last start attempts of the unit. They are also logged to
2015 * "journalctl -b" may now be used to look for boot output of a
2016 specific boot. Try "journalctl -b -1" for the previous boot,
2017 but the syntax is substantially more powerful.
2019 * "journalctl --show-cursor" has been added which prints the
2020 cursor string the last shown log line. This may then be used
2021 with the new "journalctl --after-cursor=" switch to continue
2022 browsing logs from that point on.
2024 * "journalctl --force" may now be used to force regeneration
2027 * Creation of "dead" device nodes has been moved from udev
2028 into kmod and tmpfiles. Previously, udev would read the kmod
2029 databases to pre-generate dead device nodes based on meta
2030 information contained in kernel modules, so that these would
2031 be auto-loaded on access rather then at boot. As this
2032 does not really have much to do with the exposing actual
2033 kernel devices to userspace this has always been slightly
2034 alien in the udev codebase. Following the new scheme kmod
2035 will now generate a runtime snippet for tmpfiles from the
2036 module meta information and it now is tmpfiles' job to the
2037 create the nodes. This also allows overriding access and
2038 other parameters for the nodes using the usual tmpfiles
2039 facilities. As side effect this allows us to remove the
2040 CAP_SYS_MKNOD capability bit from udevd entirely.
2042 * logind's device ACLs may now be applied to these "dead"
2043 devices nodes too, thus finally allowing managed access to
2044 devices such as /dev/snd/sequencer whithout loading the
2045 backing module right-away.
2047 * A new RPM macro has been added that may be used to apply
2048 tmpfiles configuration during package installation.
2050 * systemd-detect-virt and ConditionVirtualization= now can
2051 detect User-Mode-Linux machines (UML).
2053 * journald will now implicitly log the effective capabilities
2054 set of processes in the message metadata.
2056 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2058 * The initrd interface has been simplified (more specifically,
2059 support for passing performance data via environment
2060 variables and fsck results via files in /run has been
2061 removed). These features were non-essential, and are
2062 nowadays available in a much nicer way by having systemd in
2063 the initrd serialize its state and have the hosts systemd
2064 deserialize it again.
2066 * The udev "keymap" data files and tools to apply keyboard
2067 specific mappings of scan to key codes, and force-release
2068 scan code lists have been entirely replaced by a udev
2069 "keyboard" builtin and a hwdb data file.
2071 * systemd will now honour the kernel's "quiet" command line
2072 argument also during late shutdown, resulting in a
2073 completely silent shutdown when used.
2075 * There's now an option to control the SO_REUSEPORT socket
2076 option in .socket units.
2078 * Instance units will now automatically get a per-template
2079 subslice of system.slice unless something else is explicitly
2080 configured. For example, instances of sshd@.service will now
2081 implicitly be placed in system-sshd.slice rather than
2082 system.slice as before.
2084 * Test coverage support may now be enabled at build time.
2086 Contributions from: Dave Reisner, Frederic Crozat, Harald
2087 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2088 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2089 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2090 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2091 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2092 Giokas, Zbigniew Jędrzejewski-Szmek
2094 -- Berlin, 2013-07-23
2098 * Two new unit types have been introduced:
2100 Scope units are very similar to service units, however, are
2101 created out of pre-existing processes -- instead of PID 1
2102 forking off the processes. By using scope units it is
2103 possible for system services and applications to group their
2104 own child processes (worker processes) in a powerful way
2105 which then maybe used to organize them, or kill them
2106 together, or apply resource limits on them.
2108 Slice units may be used to partition system resources in an
2109 hierarchial fashion and then assign other units to them. By
2110 default there are now three slices: system.slice (for all
2111 system services), user.slice (for all user sessions),
2112 machine.slice (for VMs and containers).
2114 Slices and scopes have been introduced primarily in
2115 context of the work to move cgroup handling to a
2116 single-writer scheme, where only PID 1
2117 creates/removes/manages cgroups.
2119 * There's a new concept of "transient" units. In contrast to
2120 normal units these units are created via an API at runtime,
2121 not from configuration from disk. More specifically this
2122 means it is now possible to run arbitrary programs as
2123 independent services, with all execution parameters passed
2124 in via bus APIs rather than read from disk. Transient units
2125 make systemd substantially more dynamic then it ever was,
2126 and useful as a general batch manager.
2128 * logind has been updated to make use of scope and slice units
2129 for managing user sessions. As a user logs in he will get
2130 his own private slice unit, to which all sessions are added
2131 as scope units. We also added support for automatically
2132 adding an instance of user@.service for the user into the
2133 slice. Effectively logind will no longer create cgroup
2134 hierarchies on its own now, it will defer entirely to PID 1
2135 for this by means of scope, service and slice units. Since
2136 user sessions this way become entities managed by PID 1
2137 the output of "systemctl" is now a lot more comprehensive.
2139 * A new mini-daemon "systemd-machined" has been added which
2140 may be used by virtualization managers to register local
2141 VMs/containers. nspawn has been updated accordingly, and
2142 libvirt will be updated shortly. machined will collect a bit
2143 of meta information about the VMs/containers, and assign
2144 them their own scope unit (see above). The collected
2145 meta-data is then made available via the "machinectl" tool,
2146 and exposed in "ps" and similar tools. machined/machinectl
2147 is compile-time optional.
2149 * As discussed earlier, the low-level cgroup configuration
2150 options ControlGroup=, ControlGroupModify=,
2151 ControlGroupPersistent=, ControlGroupAttribute= have been
2152 removed. Please use high-level attribute settings instead as
2153 well as slice units.
2155 * A new bus call SetUnitProperties() has been added to alter
2156 various runtime parameters of a unit. This is primarily
2157 useful to alter cgroup parameters dynamically in a nice way,
2158 but will be extended later on to make more properties
2159 modifiable at runtime. systemctl gained a new set-properties
2160 command that wraps this call.
2162 * A new tool "systemd-run" has been added which can be used to
2163 run arbitrary command lines as transient services or scopes,
2164 while configuring a number of settings via the command
2165 line. This tool is currently very basic, however already
2166 very useful. We plan to extend this tool to even allow
2167 queuing of execution jobs with time triggers from the
2168 command line, similar in fashion to "at".
2170 * nspawn will now inform the user explicitly that kernels with
2171 audit enabled break containers, and suggest the user to turn
2174 * Support for detecting the IMA and AppArmor security
2175 frameworks with ConditionSecurity= has been added.
2177 * journalctl gained a new "-k" switch for showing only kernel
2178 messages, mimicking dmesg output; in addition to "--user"
2179 and "--system" switches for showing only user's own logs
2182 * systemd-delta can now show information about drop-in
2183 snippets extending unit files.
2185 * libsystemd-bus has been substantially updated but is still
2186 not available as public API.
2188 * systemd will now look for the "debug" argument on the kernel
2189 command line and enable debug logging, similar to what
2190 "systemd.log_level=debug" already did before.
2192 * "systemctl set-default", "systemctl get-default" has been
2193 added to configure the default.target symlink, which
2194 controls what to boot into by default.
2196 * "systemctl set-log-level" has been added as a convenient
2197 way to raise and lower systemd logging threshold.
2199 * "systemd-analyze plot" will now show the time the various
2200 generators needed for execution, as well as information
2201 about the unit file loading.
2203 * libsystemd-journal gained a new sd_journal_open_files() call
2204 for opening specific journal files. journactl also gained a
2205 new switch to expose this new functionality. Previously we
2206 only supported opening all files from a directory, or all
2207 files from the system, as opening individual files only is
2208 racy due to journal file rotation.
2210 * systemd gained the new DefaultEnvironment= setting in
2211 /etc/systemd/system.conf to set environment variables for
2214 * If a privileged process logs a journal message with the
2215 OBJECT_PID= field set, then journald will automatically
2216 augment this with additional OBJECT_UID=, OBJECT_GID=,
2217 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2218 system services want to log events about specific client
2219 processes. journactl/systemctl has been updated to make use
2220 of this information if all log messages regarding a specific
2223 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2224 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2225 Reisner, David Coppa, David King, David Strauss, Eelco
2226 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2227 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2228 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2229 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2230 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2231 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2232 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2233 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2234 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2235 Łukasz Stelmach, 장동준
2239 * The Python bindings gained some minimal support for the APIs
2240 exposed by libsystemd-logind.
2242 * ConditionSecurity= gained support for detecting SMACK. Since
2243 this condition already supports SELinux and AppArmor we only
2244 miss IMA for this. Patches welcome!
2246 Contributions from: Karol Lewandowski, Lennart Poettering,
2247 Zbigniew Jędrzejewski-Szmek
2251 * systemd-nspawn will now create /etc/resolv.conf if
2252 necessary, before bind-mounting the host's file onto it.
2254 * systemd-nspawn will now store meta information about a
2255 container on the container's cgroup as extended attribute
2256 fields, including the root directory.
2258 * The cgroup hierarchy has been reworked in many ways. All
2259 objects any of the components systemd creates in the cgroup
2260 tree are now suffixed. More specifically, user sessions are
2261 now placed in cgroups suffixed with ".session", users in
2262 cgroups suffixed with ".user", and nspawn containers in
2263 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2264 names are now escaped in a simple scheme to avoid collision
2265 of userspace object names with kernel filenames. This work
2266 is preparation for making these objects relocatable in the
2267 cgroup tree, in order to allow easy resource partitioning of
2268 these objects without causing naming conflicts.
2270 * systemctl list-dependencies gained the new switches
2271 --plain, --reverse, --after and --before.
2273 * systemd-inhibit now shows the process name of processes that
2274 have taken an inhibitor lock.
2276 * nss-myhostname will now also resolve "localhost"
2277 implicitly. This makes /etc/hosts an optional file and
2278 nicely handles that on IPv6 ::1 maps to both "localhost" and
2281 * libsystemd-logind.so gained a new call
2282 sd_get_machine_names() to enumerate running containers and
2283 VMs (currently only supported by very new libvirt and
2284 nspawn). sd_login_monitor can now be used to watch
2285 VMs/containers coming and going.
2287 * .include is not allowed recursively anymore, and only in
2288 unit files. Usually it is better to use drop-in snippets in
2289 .d/*.conf anyway, as introduced with systemd 198.
2291 * systemd-analyze gained a new "critical-chain" command that
2292 determines the slowest chain of units run during system
2293 boot-up. It is very useful for tracking down where
2294 optimizing boot time is the most beneficial.
2296 * systemd will no longer allow manipulating service paths in
2297 the name=systemd:/system cgroup tree using ControlGroup= in
2298 units. (But is still fine with it in all other dirs.)
2300 * There's a new systemd-nspawn@.service service file that may
2301 be used to easily run nspawn containers as system
2302 services. With the container's root directory in
2303 /var/lib/container/foobar it is now sufficient to run
2304 "systemctl start systemd-nspawn@foobar.service" to boot it.
2306 * systemd-cgls gained a new parameter "--machine" to list only
2307 the processes within a certain container.
2309 * ConditionSecurity= now can check for "apparmor". We still
2310 are lacking checks for SMACK and IMA for this condition
2311 check though. Patches welcome!
2313 * A new configuration file /etc/systemd/sleep.conf has been
2314 added that may be used to configure which kernel operation
2315 systemd is supposed to execute when "suspend", "hibernate"
2316 or "hybrid-sleep" is requested. This makes the new kernel
2317 "freeze" state accessible to the user.
2319 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2320 the passed argument if applicable.
2322 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2323 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2324 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2325 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2326 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2327 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2328 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2333 * The output of 'systemctl list-jobs' got some polishing. The
2334 '--type=' argument may now be passed more than once. A new
2335 command 'systemctl list-sockets' has been added which shows
2336 a list of kernel sockets systemd is listening on with the
2337 socket units they belong to, plus the units these socket
2340 * The experimental libsystemd-bus library got substantial
2341 updates to work in conjunction with the (also experimental)
2342 kdbus kernel project. It works well enough to exchange
2343 messages with some sophistication. Note that kdbus is not
2344 ready yet, and the library is mostly an elaborate test case
2345 for now, and not installable.
2347 * systemd gained a new unit 'systemd-static-nodes.service'
2348 that generates static device nodes earlier during boot, and
2349 can run in conjunction with udev.
2351 * libsystemd-login gained a new call sd_pid_get_user_unit()
2352 to retrieve the user systemd unit a process is running
2353 in. This is useful for systems where systemd is used as
2356 * systemd-nspawn now places all containers in the new /machine
2357 top-level cgroup directory in the name=systemd
2358 hierarchy. libvirt will soon do the same, so that we get a
2359 uniform separation of /system, /user and /machine for system
2360 services, user processes and containers/virtual
2361 machines. This new cgroup hierarchy is also useful to stick
2362 stable names to specific container instances, which can be
2363 recognized later this way (this name may be controlled
2364 via systemd-nspawn's new -M switch). libsystemd-login also
2365 gained a new call sd_pid_get_machine_name() to retrieve the
2366 name of the container/VM a specific process belongs to.
2368 * bootchart can now store its data in the journal.
2370 * libsystemd-journal gained a new call
2371 sd_journal_add_conjunction() for AND expressions to the
2372 matching logic. This can be used to express more complex
2373 logical expressions.
2375 * journactl can now take multiple --unit= and --user-unit=
2378 * The cryptsetup logic now understands the "luks.key=" kernel
2379 command line switch for specifying a file to read the
2380 decryption key from. Also, if a configured key file is not
2381 found the tool will now automatically fall back to prompting
2384 * Python systemd.journal module was updated to wrap recently
2385 added functions from libsystemd-journal. The interface was
2386 changed to bring the low level interface in s.j._Reader
2387 closer to the C API, and the high level interface in
2388 s.j.Reader was updated to wrap and convert all data about
2391 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2392 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2393 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2394 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2395 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2396 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2400 * journalctl --update-catalog now understands a new --root=
2401 option to operate on catalogs found in a different root
2404 * During shutdown after systemd has terminated all running
2405 services a final killing loop kills all remaining left-over
2406 processes. We will now print the name of these processes
2407 when we send SIGKILL to them, since this usually indicates a
2410 * If /etc/crypttab refers to password files stored on
2411 configured mount points automatic dependencies will now be
2412 generated to ensure the specific mount is established first
2413 before the key file is attempted to be read.
2415 * 'systemctl status' will now show information about the
2416 network sockets a socket unit is listening on.
2418 * 'systemctl status' will also shown information about any
2419 drop-in configuration file for units. (Drop-In configuration
2420 files in this context are files such as
2421 /etc/systemd/systemd/foobar.service.d/*.conf)
2423 * systemd-cgtop now optionally shows summed up CPU times of
2424 cgroups. Press '%' while running cgtop to switch between
2425 percentage and absolute mode. This is useful to determine
2426 which cgroups use up the most CPU time over the entire
2427 runtime of the system. systemd-cgtop has also been updated
2428 to be 'pipeable' for processing with further shell tools.
2430 * 'hostnamectl set-hostname' will now allow setting of FQDN
2433 * The formatting and parsing of time span values has been
2434 changed. The parser now understands fractional expressions
2435 such as "5.5h". The formatter will now output fractional
2436 expressions for all time spans under 1min, i.e. "5.123456s"
2437 rather than "5s 123ms 456us". For time spans under 1s
2438 millisecond values are shown, for those under 1ms
2439 microsecond values are shown. This should greatly improve
2440 all time-related output of systemd.
2442 * libsystemd-login and libsystemd-journal gained new
2443 functions for querying the poll() events mask and poll()
2444 timeout value for integration into arbitrary event
2447 * localectl gained the ability to list available X11 keymaps
2448 (models, layouts, variants, options).
2450 * 'systemd-analyze dot' gained the ability to filter for
2451 specific units via shell-style globs, to create smaller,
2452 more useful graphs. I.e. it is now possible to create simple
2453 graphs of all the dependencies between only target units, or
2454 of all units that Avahi has dependencies with.
2456 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2457 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2458 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2459 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2460 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2461 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2462 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2466 * The boot-time readahead implementation for rotating media
2467 will now read the read-ahead data in multiple passes which
2468 consist of all read requests made in equidistant time
2469 intervals. This means instead of strictly reading read-ahead
2470 data in its physical order on disk we now try to find a
2471 middle ground between physical and access time order.
2473 * /etc/os-release files gained a new BUILD_ID= field for usage
2474 on operating systems that provide continuous builds of OS
2477 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2478 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2479 William Douglas, Zbigniew Jędrzejewski-Szmek
2483 * systemd-python gained an API exposing libsystemd-daemon.
2485 * The SMACK setup logic gained support for uploading CIPSO
2488 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2489 ReadOnlyDirectories= and InaccessibleDirectories= has
2490 changed. The private /tmp and /var/tmp directories are now
2491 shared by all processes of a service (which means
2492 ExecStartPre= may now leave data in /tmp that ExecStart= of
2493 the same service can still access). When a service is
2494 stopped its temporary directories are immediately deleted
2495 (normal clean-up with tmpfiles is still done in addition to
2498 * By default, systemd will now set a couple of sysctl
2499 variables in the kernel: the safe sysrq options are turned
2500 on, IP route verification is turned on, and source routing
2501 disabled. The recently added hardlink and softlink
2502 protection of the kernel is turned on. These settings should
2503 be reasonably safe, and good defaults for all new systems.
2505 * The predictable network naming logic may now be turned off
2506 with a new kernel command line switch: net.ifnames=0.
2508 * A new libsystemd-bus module has been added that implements a
2509 pretty complete D-Bus client library. For details see:
2511 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2513 * journald will now explicitly flush the journal files to disk
2514 at the latest 5min after each write. The file will then also
2515 be marked offline until the next write. This should increase
2516 reliability in case of a crash. The synchronization delay
2517 can be configured via SyncIntervalSec= in journald.conf.
2519 * There's a new remote-fs-setup.target unit that can be used
2520 to pull in specific services when at least one remote file
2521 system is to be mounted.
2523 * There are new targets timers.target and paths.target as
2524 canonical targets to pull user timer and path units in
2525 from. This complements sockets.target with a similar
2526 purpose for socket units.
2528 * libudev gained a new call udev_device_set_attribute_value()
2529 to set sysfs attributes of a device.
2531 * The udev daemon now sets the default number of worker
2532 processes executed in parallel based on the number of available
2533 CPUs instead of the amount of available RAM. This is supposed
2534 to provide a more reliable default and limit a too aggressive
2535 paralellism for setups with 1000s of devices connected.
2537 Contributions from: Auke Kok, Colin Walters, Cristian
2538 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2539 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2540 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2541 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2542 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2543 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2544 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2545 Zbigniew Jędrzejewski-Szmek
2549 * Configuration of unit files may now be extended via drop-in
2550 files without having to edit/override the unit files
2551 themselves. More specifically, if the administrator wants to
2552 change one value for a service file foobar.service he can
2553 now do so by dropping in a configuration snippet into
2554 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2555 will load all these snippets and apply them on top of the
2556 main unit configuration file, possibly extending or
2557 overriding its settings. Using these drop-in snippets is
2558 generally nicer than the two earlier options for changing
2559 unit files locally: copying the files from
2560 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2561 them there; or creating a new file in /etc/systemd/system/
2562 that incorporates the original one via ".include". Drop-in
2563 snippets into these .d/ directories can be placed in any
2564 directory systemd looks for units in, and the usual
2565 overriding semantics between /usr/lib, /etc and /run apply
2568 * Most unit file settings which take lists of items can now be
2569 reset by assigning the empty string to them. For example,
2570 normally, settings such as Environment=FOO=BAR append a new
2571 environment variable assignment to the environment block,
2572 each time they are used. By assigning Environment= the empty
2573 string the environment block can be reset to empty. This is
2574 particularly useful with the .d/*.conf drop-in snippets
2575 mentioned above, since this adds the ability to reset list
2576 settings from vendor unit files via these drop-ins.
2578 * systemctl gained a new "list-dependencies" command for
2579 listing the dependencies of a unit recursively.
2581 * Inhibitors are now honored and listed by "systemctl
2582 suspend", "systemctl poweroff" (and similar) too, not only
2583 GNOME. These commands will also list active sessions by
2586 * Resource limits (as exposed by the various control group
2587 controllers) can now be controlled dynamically at runtime
2588 for all units. More specifically, you can now use a command
2589 like "systemctl set-cgroup-attr foobar.service cpu.shares
2590 2000" to alter the CPU shares a specific service gets. These
2591 settings are stored persistently on disk, and thus allow the
2592 administrator to easily adjust the resource usage of
2593 services with a few simple commands. This dynamic resource
2594 management logic is also available to other programs via the
2595 bus. Almost any kernel cgroup attribute and controller is
2598 * systemd-vconsole-setup will now copy all font settings to
2599 all allocated VTs, where it previously applied them only to
2602 * libsystemd-login gained the new sd_session_get_tty() API
2605 * This release drops support for a few legacy or
2606 distribution-specific LSB facility names when parsing init
2607 scripts: $x-display-manager, $mail-transfer-agent,
2608 $mail-transport-agent, $mail-transfer-agent, $smtp,
2609 $null. Also, the mail-transfer-agent.target unit backing
2610 this has been removed. Distributions which want to retain
2611 compatibility with this should carry the burden for
2612 supporting this themselves and patch support for these back
2613 in, if they really need to. Also, the facilities $syslog and
2614 $local_fs are now ignored, since systemd does not support
2615 early-boot LSB init scripts anymore, and these facilities
2616 are implied anyway for normal services. syslog.target has
2619 * There are new bus calls on PID1's Manager object for
2620 cancelling jobs, and removing snapshot units. Previously,
2621 both calls were only available on the Job and Snapshot
2624 * systemd-journal-gatewayd gained SSL support.
2626 * The various "environment" files, such as /etc/locale.conf
2627 now support continuation lines with a backslash ("\") as
2628 last character in the line, similarly in style (but different)
2629 to how this is supported in shells.
2631 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2632 now implicitly appended to every log entry logged. systemctl
2633 has been updated to filter by this field when operating on a
2634 user systemd instance.
2636 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2637 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2638 the container. This makes it easier to boot unmodified
2639 Fedora systems in a container, which however still requires
2640 audit=0 to be passed on the kernel command line. Auditing in
2641 kernel and userspace is unfortunately still too broken in
2642 context of containers, hence we recommend compiling it out
2643 of the kernel or using audit=0. Hopefully this will be fixed
2644 one day for good in the kernel.
2646 * nspawn gained the new --bind= and --bind-ro= parameters to
2647 bind mount specific directories from the host into the
2650 * nspawn will now mount its own devpts file system instance
2651 into the container, in order not to leak pty devices from
2652 the host into the container.
2654 * systemd will now read the firmware boot time performance
2655 information from the EFI variables, if the used boot loader
2656 supports this, and takes it into account for boot performance
2657 analysis via "systemd-analyze". This is currently supported
2658 only in conjunction with Gummiboot, but could be supported
2659 by other boot loaders too. For details see:
2661 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2663 * A new generator has been added that automatically mounts the
2664 EFI System Partition (ESP) to /boot, if that directory
2665 exists, is empty, and no other file system has been
2666 configured to be mounted there.
2668 * logind will now send out PrepareForSleep(false) out
2669 unconditionally, after coming back from suspend. This may be
2670 used by applications as asynchronous notification for
2671 system resume events.
2673 * "systemctl unlock-sessions" has been added, that allows
2674 unlocking the screens of all user sessions at once, similar
2675 to how "systemctl lock-sessions" already locked all users
2676 sessions. This is backed by a new D-Bus call UnlockSessions().
2678 * "loginctl seat-status" will now show the master device of a
2679 seat. (i.e. the device of a seat that needs to be around for
2680 the seat to be considered available, usually the graphics
2683 * tmpfiles gained a new "X" line type, that allows
2684 configuration of files and directories (with wildcards) that
2685 shall be excluded from automatic cleanup ("aging").
2687 * udev default rules set the device node permissions now only
2688 at "add" events, and do not change them any longer with a
2689 later "change" event.
2691 * The log messages for lid events and power/sleep keypresses
2692 now carry a message ID.
2694 * We now have a substantially larger unit test suite, but this
2695 continues to be work in progress.
2697 * udevadm hwdb gained a new --root= parameter to change the
2698 root directory to operate relative to.
2700 * logind will now issue a background sync() request to the kernel
2701 early at shutdown, so that dirty buffers are flushed to disk early
2702 instead of at the last moment, in order to optimize shutdown
2705 * A new bootctl tool has been added that is an interface for
2706 certain boot loader operations. This is currently a preview
2707 and is likely to be extended into a small mechanism daemon
2708 like timedated, localed, hostnamed, and can be used by
2709 graphical UIs to enumerate available boot options, and
2710 request boot into firmware operations.
2712 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2713 the rest of the package. It also has been updated to work
2714 correctly in initrds.
2716 * Policykit previously has been runtime optional, and is now
2717 also compile time optional via a configure switch.
2719 * systemd-analyze has been reimplemented in C. Also "systemctl
2720 dot" has moved into systemd-analyze.
2722 * "systemctl status" with no further parameters will now print
2723 the status of all active or failed units.
2725 * Operations such as "systemctl start" can now be executed
2726 with a new mode "--irreversible" which may be used to queue
2727 operations that cannot accidentally be reversed by a later
2728 job queuing. This is by default used to make shutdown
2729 requests more robust.
2731 * The Python API of systemd now gained a new module for
2732 reading journal files.
2734 * A new tool kernel-install has been added that can install
2735 kernel images according to the Boot Loader Specification:
2737 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2739 * Boot time console output has been improved to provide
2740 animated boot time output for hanging jobs.
2742 * A new tool systemd-activate has been added which can be used
2743 to test socket activation with, directly from the command
2744 line. This should make it much easier to test and debug
2745 socket activation in daemons.
2747 * journalctl gained a new "--reverse" (or -r) option to show
2748 journal output in reverse order (i.e. newest line first).
2750 * journalctl gained a new "--pager-end" (or -e) option to jump
2751 to immediately jump to the end of the journal in the
2752 pager. This is only supported in conjunction with "less".
2754 * journalctl gained a new "--user-unit=" option, that works
2755 similarly to "--unit=" but filters for user units rather than
2758 * A number of unit files to ease adoption of systemd in
2759 initrds has been added. This moves some minimal logic from
2760 the various initrd implementations into systemd proper.
2762 * The journal files are now owned by a new group
2763 "systemd-journal", which exists specifically to allow access
2764 to the journal, and nothing else. Previously, we used the
2765 "adm" group for that, which however possibly covers more
2766 than just journal/log file access. This new group is now
2767 already used by systemd-journal-gatewayd to ensure this
2768 daemon gets access to the journal files and as little else
2769 as possible. Note that "make install" will also set FS ACLs
2770 up for /var/log/journal to give "adm" and "wheel" read
2771 access to it, in addition to "systemd-journal" which owns
2772 the journal files. We recommend that packaging scripts also
2773 add read access to "adm" + "wheel" to /var/log/journal, and
2774 all existing/future journal files. To normal users and
2775 administrators little changes, however packagers need to
2776 ensure to create the "systemd-journal" system group at
2777 package installation time.
2779 * The systemd-journal-gatewayd now runs as unprivileged user
2780 systemd-journal-gateway:systemd-journal-gateway. Packaging
2781 scripts need to create these system user/group at
2784 * timedated now exposes a new boolean property CanNTP that
2785 indicates whether a local NTP service is available or not.
2787 * systemd-detect-virt will now also detect xen PVs
2789 * The pstore file system is now mounted by default, if it is
2792 * In addition to the SELinux and IMA policies we will now also
2793 load SMACK policies at early boot.
2795 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2796 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2797 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2798 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2799 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2800 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2801 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2802 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2803 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2804 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2805 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2806 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2807 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2808 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2812 * Timer units now support calendar time events in addition to
2813 monotonic time events. That means you can now trigger a unit
2814 based on a calendar time specification such as "Thu,Fri
2815 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2816 or fifth day of any month of the year 2013, given that it is
2817 a thursday or friday. This brings timer event support
2818 considerably closer to cron's capabilities. For details on
2819 the supported calendar time specification language see
2822 * udev now supports a number of different naming policies for
2823 network interfaces for predictable names, and a combination
2824 of these policies is now the default. Please see this wiki
2825 document for details:
2827 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2829 * Auke Kok's bootchart implementation has been added to the
2830 systemd tree. It is an optional component that can graph the
2831 boot in quite some detail. It is one of the best bootchart
2832 implementations around and minimal in its code and
2835 * nss-myhostname has been integrated into the systemd source
2836 tree. nss-myhostname guarantees that the local hostname
2837 always stays resolvable via NSS. It has been a weak
2838 requirement of systemd-hostnamed since a long time, and
2839 since its code is actually trivial we decided to just
2840 include it in systemd's source tree. It can be turned off
2841 with a configure switch.
2843 * The read-ahead logic is now capable of properly detecting
2844 whether a btrfs file system is on SSD or rotating media, in
2845 order to optimize the read-ahead scheme. Previously, it was
2846 only capable of detecting this on traditional file systems
2849 * In udev, additional device properties are now read from the
2850 IAB in addition to the OUI database. Also, Bluetooth company
2851 identities are attached to the devices as well.
2853 * In service files %U may be used as specifier that is
2854 replaced by the configured user name of the service.
2856 * nspawn may now be invoked without a controlling TTY. This
2857 makes it suitable for invocation as its own service. This
2858 may be used to set up a simple containerized server system
2859 using only core OS tools.
2861 * systemd and nspawn can now accept socket file descriptors
2862 when they are started for socket activation. This enables
2863 implementation of socket activated nspawn
2864 containers. i.e. think about autospawning an entire OS image
2865 when the first SSH or HTTP connection is received. We expect
2866 that similar functionality will also be added to libvirt-lxc
2869 * journalctl will now suppress ANSI color codes when
2870 presenting log data.
2872 * systemctl will no longer show control group information for
2873 a unit if a the control group is empty anyway.
2875 * logind can now automatically suspend/hibernate/shutdown the
2878 * /etc/machine-info and hostnamed now also expose the chassis
2879 type of the system. This can be used to determine whether
2880 the local system is a laptop, desktop, handset or
2881 tablet. This information may either be configured by the
2882 user/vendor or is automatically determined from ACPI and DMI
2883 information if possible.
2885 * A number of PolicyKit actions are now bound together with
2886 "imply" rules. This should simplify creating UIs because
2887 many actions will now authenticate similar ones as well.
2889 * Unit files learnt a new condition ConditionACPower= which
2890 may be used to conditionalize a unit depending on whether an
2891 AC power source is connected or not, of whether the system
2892 is running on battery power.
2894 * systemctl gained a new "is-failed" verb that may be used in
2895 shell scripts and suchlike to check whether a specific unit
2896 is in the "failed" state.
2898 * The EnvironmentFile= setting in unit files now supports file
2899 globbing, and can hence be used to easily read a number of
2900 environment files at once.
2902 * systemd will no longer detect and recognize specific
2903 distributions. All distribution-specific #ifdeffery has been
2904 removed, systemd is now fully generic and
2905 distribution-agnostic. Effectively, not too much is lost as
2906 a lot of the code is still accessible via explicit configure
2907 switches. However, support for some distribution specific
2908 legacy configuration file formats has been dropped. We
2909 recommend distributions to simply adopt the configuration
2910 files everybody else uses now and convert the old
2911 configuration from packaging scripts. Most distributions
2912 already did that. If that's not possible or desirable,
2913 distributions are welcome to forward port the specific
2914 pieces of code locally from the git history.
2916 * When logging a message about a unit systemd will now always
2917 log the unit name in the message meta data.
2919 * localectl will now also discover system locale data that is
2920 not stored in locale archives, but directly unpacked.
2922 * logind will no longer unconditionally use framebuffer
2923 devices as seat masters, i.e. as devices that are required
2924 to be existing before a seat is considered preset. Instead,
2925 it will now look for all devices that are tagged as
2926 "seat-master" in udev. By default framebuffer devices will
2927 be marked as such, but depending on local systems other
2928 devices might be marked as well. This may be used to
2929 integrate graphics cards using closed source drivers (such
2930 as NVidia ones) more nicely into logind. Note however, that
2931 we recommend using the open source NVidia drivers instead,
2932 and no udev rules for the closed-source drivers will be
2933 shipped from us upstream.
2935 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2936 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2937 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2938 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2939 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2940 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2941 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2942 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2943 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2944 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2945 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2946 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2951 * udev gained support for loading additional device properties
2952 from an indexed database that is keyed by vendor/product IDs
2953 and similar device identifiers. For the beginning this
2954 "hwdb" is populated with data from the well-known PCI and
2955 USB database, but also includes PNP, ACPI and OID data. In
2956 the longer run this indexed database shall grow into
2957 becoming the one central database for non-essential
2958 userspace device metadata. Previously, data from the PCI/USB
2959 database was only attached to select devices, since the
2960 lookup was a relatively expensive operation due to O(n) time
2961 complexity (with n being the number of entries in the
2962 database). Since this is now O(1), we decided to add in this
2963 data for all devices where this is available, by
2964 default. Note that the indexed database needs to be rebuilt
2965 when new data files are installed. To achieve this you need
2966 to update your packaging scripts to invoke "udevadm hwdb
2967 --update" after installation of hwdb data files. For
2968 RPM-based distributions we introduced the new
2969 %udev_hwdb_update macro for this purpose.
2971 * The Journal gained support for the "Message Catalog", an
2972 indexed database to link up additional information with
2973 journal entries. For further details please check:
2975 http://www.freedesktop.org/wiki/Software/systemd/catalog
2977 The indexed message catalog database also needs to be
2978 rebuilt after installation of message catalog files. Use
2979 "journalctl --update-catalog" for this. For RPM-based
2980 distributions we introduced the %journal_catalog_update
2981 macro for this purpose.
2983 * The Python Journal bindings gained support for the standard
2984 Python logging framework.
2986 * The Journal API gained new functions for checking whether
2987 the underlying file system of a journal file is capable of
2988 properly reporting file change notifications, or whether
2989 applications that want to reflect journal changes "live"
2990 need to recheck journal files continuously in appropriate
2993 * It is now possible to set the "age" field for tmpfiles
2994 entries to 0, indicating that files matching this entry
2995 shall always be removed when the directories are cleaned up.
2997 * coredumpctl gained a new "gdb" verb which invokes gdb
2998 right-away on the selected coredump.
3000 * There's now support for "hybrid sleep" on kernels that
3001 support this, in addition to "suspend" and "hibernate". Use
3002 "systemctl hybrid-sleep" to make use of this.
3004 * logind's HandleSuspendKey= setting (and related settings)
3005 now gained support for a new "lock" setting to simply
3006 request the screen lock on all local sessions, instead of
3007 actually executing a suspend or hibernation.
3009 * systemd will now mount the EFI variables file system by
3012 * Socket units now gained support for configuration of the
3013 SMACK security label.
3015 * timedatectl will now output the time of the last and next
3016 daylight saving change.
3018 * We dropped support for various legacy and distro-specific
3019 concepts, such as insserv, early-boot SysV services
3020 (i.e. those for non-standard runlevels such as 'b' or 'S')
3021 or ArchLinux /etc/rc.conf support. We recommend the
3022 distributions who still need support this to either continue
3023 to maintain the necessary patches downstream, or find a
3024 different solution. (Talk to us if you have questions!)
3026 * Various systemd components will now bypass PolicyKit checks
3027 for root and otherwise handle properly if PolicyKit is not
3028 found to be around. This should fix most issues for
3029 PolicyKit-less systems. Quite frankly this should have been
3030 this way since day one. It is absolutely our intention to
3031 make systemd work fine on PolicyKit-less systems, and we
3032 consider it a bug if something does not work as it should if
3033 PolicyKit is not around.
3035 * For embedded systems it is now possible to build udev and
3036 systemd without blkid and/or kmod support.
3038 * "systemctl switch-root" is now capable of switching root
3039 more than once. I.e. in addition to transitions from the
3040 initrd to the host OS it is now possible to transition to
3041 further OS images from the host. This is useful to implement
3042 offline updating tools.
3044 * Various other additions have been made to the RPM macros
3045 shipped with systemd. Use %udev_rules_update() after
3046 installing new udev rules files. %_udevhwdbdir,
3047 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3048 %_sysctldir are now available which resolve to the right
3049 directories for packages to place various data files in.
3051 * journalctl gained the new --full switch (in addition to
3052 --all, to disable ellipsation for long messages.
3054 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3055 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3056 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3057 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3058 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3059 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3060 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3061 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3062 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3066 * journalctl gained new --since= and --until= switches to
3067 filter by time. It also now supports nice filtering for
3068 units via --unit=/-u.
3070 * Type=oneshot services may use ExecReload= and do the
3073 * The journal daemon now supports time-based rotation and
3074 vacuuming, in addition to the usual disk-space based
3077 * The journal will now index the available field values for
3078 each field name. This enables clients to show pretty drop
3079 downs of available match values when filtering. The bash
3080 completion of journalctl has been updated
3081 accordingly. journalctl gained a new switch -F to list all
3082 values a certain field takes in the journal database.
3084 * More service events are now written as structured messages
3085 to the journal, and made recognizable via message IDs.
3087 * The timedated, localed and hostnamed mini-services which
3088 previously only provided support for changing time, locale
3089 and hostname settings from graphical DEs such as GNOME now
3090 also have a minimal (but very useful) text-based client
3091 utility each. This is probably the nicest way to changing
3092 these settings from the command line now, especially since
3093 it lists available options and is fully integrated with bash
3096 * There's now a new tool "systemd-coredumpctl" to list and
3097 extract coredumps from the journal.
3099 * We now install a README each in /var/log/ and
3100 /etc/rc.d/init.d explaining where the system logs and init
3101 scripts went. This hopefully should help folks who go to
3102 that dirs and look into the otherwise now empty void and
3103 scratch their heads.
3105 * When user-services are invoked (by systemd --user) the
3106 $MANAGERPID env var is set to the PID of systemd.
3108 * SIGRTMIN+24 when sent to a --user instance will now result
3109 in immediate termination of systemd.
3111 * gatewayd received numerous feature additions such as a
3112 "follow" mode, for live syncing and filtering.
3114 * browse.html now allows filtering and showing detailed
3115 information on specific entries. Keyboard navigation and
3116 mouse screen support has been added.
3118 * gatewayd/journalctl now supports HTML5/JSON
3119 Server-Sent-Events as output.
3121 * The SysV init script compatibility logic will now
3122 heuristically determine whether a script supports the
3123 "reload" verb, and only then make this available as
3126 * "systemctl status --follow" has been removed, use "journalctl
3129 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3130 have been removed since they are hardly useful to be
3133 * And I'd like to take the opportunity to specifically mention
3134 Zbigniew for his great contributions. Zbigniew, you rock!
3136 Contributions from: Andrew Eikum, Christian Hesse, Colin
3137 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3138 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3139 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3140 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3141 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3142 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3146 * If /etc/vconsole.conf is non-existent or empty we will no
3147 longer load any console font or key map at boot by
3148 default. Instead the kernel defaults will be left
3149 intact. This is definitely the right thing to do, as no
3150 configuration should mean no configuration, and hard-coding
3151 font names that are different on all archs is probably a bad
3152 idea. Also, the kernel default key map and font should be
3153 good enough for most cases anyway, and mostly identical to
3154 the userspace fonts/key maps we previously overloaded them
3155 with. If distributions want to continue to default to a
3156 non-kernel font or key map they should ship a default
3157 /etc/vconsole.conf with the appropriate contents.
3159 Contributions from: Colin Walters, Daniel J Walsh, Dave
3160 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3161 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3165 * journalctl gained a new --cursor= switch to show entries
3166 starting from the specified location in the journal.
3168 * We now enforce a size limit on journal entry fields exported
3169 with "-o json" in journalctl. Fields larger than 4K will be
3170 assigned null. This can be turned off with --all.
3172 * An (optional) journal gateway daemon is now available as
3173 "systemd-journal-gatewayd.service". This service provides
3174 access to the journal via HTTP and JSON. This functionality
3175 will be used to implement live log synchronization in both
3176 pull and push modes, but has various other users too, such
3177 as easy log access for debugging of embedded devices. Right
3178 now it is already useful to retrieve the journal via HTTP:
3180 # systemctl start systemd-journal-gatewayd.service
3181 # wget http://localhost:19531/entries
3183 This will download the journal contents in a
3184 /var/log/messages compatible format. The same as JSON:
3186 # curl -H"Accept: application/json" http://localhost:19531/entries
3188 This service is also accessible via a web browser where a
3189 single static HTML5 app is served that uses the JSON logic
3190 to enable the user to do some basic browsing of the
3191 journal. This will be extended later on. Here's an example
3192 screenshot of this app in its current state:
3194 http://0pointer.de/public/journal-gatewayd
3196 Contributions from: Kay Sievers, Lennart Poettering, Robert
3197 Milasan, Tom Gundersen
3201 * The bash completion logic is now available for journalctl
3204 * We do not mount the "cpuset" controller anymore together with
3205 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3206 started if no parameters are assigned to it. "cpuset" hence
3207 broke code that assumed it it could create "cpu" groups and
3210 * journalctl -f will now subscribe to terminal size changes,
3211 and line break accordingly.
3213 Contributions from: Dave Reisner, Kay Sievers, Lennart
3214 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3218 * nspawn will now create a symlink /etc/localtime in the
3219 container environment, copying the host's timezone
3220 setting. Previously this has been done via a bind mount, but
3221 since symlinks cannot be bind mounted this has now been
3222 changed to create/update the appropriate symlink.
3224 * journalctl -n's line number argument is now optional, and
3225 will default to 10 if omitted.
3227 * journald will now log the maximum size the journal files may
3228 take up on disk. This is particularly useful if the default
3229 built-in logic of determining this parameter from the file
3230 system size is used. Use "systemctl status
3231 systemd-journald.service" to see this information.
3233 * The multi-seat X wrapper tool has been stripped down. As X
3234 is now capable of enumerating graphics devices via udev in a
3235 seat-aware way the wrapper is not strictly necessary
3236 anymore. A stripped down temporary stop-gap is still shipped
3237 until the upstream display managers have been updated to
3238 fully support the new X logic. Expect this wrapper to be
3239 removed entirely in one of the next releases.
3241 * HandleSleepKey= in logind.conf has been split up into
3242 HandleSuspendKey= and HandleHibernateKey=. The old setting
3243 is not available anymore. X11 and the kernel are
3244 distuingishing between these keys and we should too. This
3245 also means the inhibition lock for these keys has been split
3248 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3249 Poettering, Lukas Nykryn, Václav Pavlín
3253 * Whenever a unit changes state we will now log this to the
3254 journal and show along the unit's own log output in
3257 * ConditionPathIsMountPoint= can now properly detect bind
3258 mount points too. (Previously, a bind mount of one file
3259 system to another place in the same file system could not be
3260 detected as mount, since they shared struct stat's st_dev
3263 * We will now mount the cgroup controllers cpu, cpuacct,
3264 cpuset and the controllers net_cls, net_prio together by
3267 * nspawn containers will now have a virtualized boot
3268 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3269 over with a randomized ID at container initialization). This
3270 has the effect of making "journalctl -b" do the right thing
3273 * The JSON output journal serialization has been updated not
3274 to generate "endless" list objects anymore, but rather one
3275 JSON object per line. This is more in line how most JSON
3276 parsers expect JSON objects. The new output mode
3277 "json-pretty" has been added to provide similar output, but
3278 neatly aligned for readability by humans.
3280 * We dropped all explicit sync() invocations in the shutdown
3281 code. The kernel does this implicitly anyway in the kernel
3282 reboot() syscall. halt(8)'s -n option is now a compatibility
3285 * We now support virtualized reboot() in containers, as
3286 supported by newer kernels. We will fall back to exit() if
3287 CAP_SYS_REBOOT is not available to the container. Also,
3288 nspawn makes use of this now and will actually reboot the
3289 container if the containerized OS asks for that.
3291 * journalctl will only show local log output by default
3292 now. Use --merge (-m) to show remote log output, too.
3294 * libsystemd-journal gained the new sd_journal_get_usage()
3295 call to determine the current disk usage of all journal
3296 files. This is exposed in the new "journalctl --disk-usage"
3299 * journald gained a new configuration setting SplitMode= in
3300 journald.conf which may be used to control how user journals
3301 are split off. See journald.conf(5) for details.
3303 * A new condition type ConditionFileNotEmpty= has been added.
3305 * tmpfiles' "w" lines now support file globbing, to write
3306 multiple files at once.
3308 * We added Python bindings for the journal submission
3309 APIs. More Python APIs for a number of selected APIs will
3310 likely follow. Note that we intend to add native bindings
3311 only for the Python language, as we consider it common
3312 enough to deserve bindings shipped within systemd. There are
3313 various projects outside of systemd that provide bindings
3314 for languages such as PHP or Lua.
3316 * Many conditions will now resolve specifiers such as %i. In
3317 addition, PathChanged= and related directives of .path units
3318 now support specifiers as well.
3320 * There's now a new RPM macro definition for the system preset
3323 * journald will now warn if it ca not forward a message to the
3324 syslog daemon because its socket is full.
3326 * timedated will no longer write or process /etc/timezone,
3327 except on Debian. As we do not support late mounted /usr
3328 anymore /etc/localtime always being a symlink is now safe,
3329 and hence the information in /etc/timezone is not necessary
3332 * logind will now always reserve one VT for a text getty (VT6
3333 by default). Previously if more than 6 X sessions where
3334 started they took up all the VTs with auto-spawned gettys,
3335 so that no text gettys were available anymore.
3337 * udev will now automatically inform the btrfs kernel logic
3338 about btrfs RAID components showing up. This should make
3339 simple hotplug based btrfs RAID assembly work.
3341 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3342 (but not for its children which will stay at the kernel
3343 default). This should allow setups with a lot more listening
3346 * systemd will now always pass the configured timezone to the
3347 kernel at boot. timedated will do the same when the timezone
3350 * logind's inhibition logic has been updated. By default,
3351 logind will now handle the lid switch, the power and sleep
3352 keys all the time, even in graphical sessions. If DEs want
3353 to handle these events on their own they should take the new
3354 handle-power-key, handle-sleep-key and handle-lid-switch
3355 inhibitors during their runtime. A simple way to achiveve
3356 that is to invoke the DE wrapped in an invocation of:
3358 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3360 * Access to unit operations is now checked via SELinux taking
3361 the unit file label and client process label into account.
3363 * systemd will now notify the administrator in the journal
3364 when he over-mounts a non-empty directory.
3366 * There are new specifiers that are resolved in unit files,
3367 for the host name (%H), the machine ID (%m) and the boot ID
3370 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3371 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3372 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3373 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3374 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3375 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3376 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3380 * Support for reading structured kernel messages from
3381 /dev/kmsg has now been added and is enabled by default.
3383 * Support for reading kernel messages from /proc/kmsg has now
3384 been removed. If you want kernel messages in the journal
3385 make sure to run a recent kernel (>= 3.5) that supports
3386 reading structured messages from /dev/kmsg (see
3387 above). /proc/kmsg is now exclusive property of classic
3388 syslog daemons again.
3390 * The libudev API gained the new
3391 udev_device_new_from_device_id() call.
3393 * The logic for file system namespace (ReadOnlyDirectory=,
3394 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3395 require pivot_root() anymore. This means fewer temporary
3396 directories are created below /tmp for this feature.
3398 * nspawn containers will now see and receive all submounts
3399 made on the host OS below the root file system of the
3402 * Forward Secure Sealing is now supported for Journal files,
3403 which provide cryptographical sealing of journal files so
3404 that attackers cannot alter log history anymore without this
3405 being detectable. Lennart will soon post a blog story about
3406 this explaining it in more detail.
3408 * There are two new service settings RestartPreventExitStatus=
3409 and SuccessExitStatus= which allow configuration of exit
3410 status (exit code or signal) which will be excepted from the
3411 restart logic, resp. consider successful.
3413 * journalctl gained the new --verify switch that can be used
3414 to check the integrity of the structure of journal files and
3415 (if Forward Secure Sealing is enabled) the contents of
3418 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3419 and similar symlinks pre-created. This makes running shells
3420 as container init process a lot more fun.
3422 * The fstab support can now handle PARTUUID= and PARTLABEL=
3425 * A new ConditionHost= condition has been added to match
3426 against the hostname (with globs) and machine ID. This is
3427 useful for clusters where a single OS image is used to
3428 provision a large number of hosts which shall run slightly
3429 different sets of services.
3431 * Services which hit the restart limit will now be placed in a
3434 Contributions from: Bertram Poettering, Dave Reisner, Huang
3435 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3436 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3440 * When running in --user mode systemd will now become a
3441 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3442 tree a lot more organized.
3444 * A new PartOf= unit dependency type has been introduced that
3445 may be used to group services in a natural way.
3447 * "systemctl enable" may now be used to enable instances of
3450 * journalctl now prints error log levels in red, and
3451 warning/notice log levels in bright white. It also supports
3452 filtering by log level now.
3454 * cgtop gained a new -n switch (similar to top), to configure
3455 the maximum number of iterations to run for. It also gained
3456 -b, to run in batch mode (accepting no input).
3458 * The suffix ".service" may now be omitted on most systemctl
3459 command lines involving service unit names.
3461 * There's a new bus call in logind to lock all sessions, as
3462 well as a loginctl verb for it "lock-sessions".
3464 * libsystemd-logind.so gained a new call sd_journal_perror()
3465 that works similar to libc perror() but logs to the journal
3466 and encodes structured information about the error number.
3468 * /etc/crypttab entries now understand the new keyfile-size=
3471 * shutdown(8) now can send a (configurable) wall message when
3472 a shutdown is cancelled.
3474 * The mount propagation mode for the root file system will now
3475 default to "shared", which is useful to make containers work
3476 nicely out-of-the-box so that they receive new mounts from
3477 the host. This can be undone locally by running "mount
3478 --make-rprivate /" if needed.
3480 * The prefdm.service file has been removed. Distributions
3481 should maintain this unit downstream if they intend to keep
3482 it around. However, we recommend writing normal unit files
3483 for display managers instead.
3485 * Since systemd is a crucial part of the OS we will now
3486 default to a number of compiler switches that improve
3487 security (hardening) such as read-only relocations, stack
3488 protection, and suchlike.
3490 * The TimeoutSec= setting for services is now split into
3491 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3492 of individual time outs for the start and the stop phase of
3495 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3496 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3497 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3498 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3499 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3500 Gundersen, Zbigniew Jędrzejewski-Szmek
3504 * The journal and id128 C APIs are now fully documented as man
3507 * Extra safety checks have been added when transitioning from
3508 the initial RAM disk to the main system to avoid accidental
3511 * /etc/crypttab entries now understand the new keyfile-offset=
3514 * systemctl -t can now be used to filter by unit load state.
3516 * The journal C API gained the new sd_journal_wait() call to
3517 make writing synchronous journal clients easier.
3519 * journalctl gained the new -D switch to show journals from a
3522 * journalctl now displays a special marker between log
3523 messages of two different boots.
3525 * The journal is now explicitly flushed to /var via a service
3526 systemd-journal-flush.service, rather than implicitly simply
3527 by seeing /var/log/journal to be writable.
3529 * journalctl (and the journal C APIs) can now match for much
3530 more complex expressions, with alternatives and
3533 * When transitioning from the initial RAM disk to the main
3534 system we will now kill all processes in a killing spree to
3535 ensure no processes stay around by accident.
3537 * Three new specifiers may be used in unit files: %u, %h, %s
3538 resolve to the user name, user home directory resp. user
3539 shell. This is useful for running systemd user instances.
3541 * We now automatically rotate journal files if their data
3542 object hash table gets a fill level > 75%. We also size the
3543 hash table based on the configured maximum file size. This
3544 together should lower hash collisions drastically and thus
3545 speed things up a bit.
3547 * journalctl gained the new "--header" switch to introspect
3548 header data of journal files.
3550 * A new setting SystemCallFilters= has been added to services
3551 which may be used to apply blacklists or whitelists to
3552 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3554 * nspawn gained a new --link-journal= switch (and quicker: -j)
3555 to link the container journal with the host. This makes it
3556 very easy to centralize log viewing on the host for all
3557 guests while still keeping the journal files separated.
3559 * Many bugfixes and optimizations
3561 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3562 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3563 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3568 * Several tools now understand kernel command line arguments,
3569 which are only read when run in an initial RAM disk. They
3570 usually follow closely their normal counterparts, but are
3573 * There's a new tool to analyze the readahead files that are
3574 automatically generated at boot. Use:
3576 /usr/lib/systemd/systemd-readahead analyze /.readahead
3578 * We now provide an early debug shell on tty9 if this enabled. Use:
3580 systemctl enable debug-shell.service
3582 * All plymouth related units have been moved into the Plymouth
3583 package. Please make sure to upgrade your Plymouth version
3586 * systemd-tmpfiles now supports getting passed the basename of
3587 a configuration file only, in which case it will look for it
3588 in all appropriate directories automatically.
3590 * udevadm info now takes a /dev or /sys path as argument, and
3591 does the right thing. Example:
3593 udevadm info /dev/sda
3594 udevadm info /sys/class/block/sda
3596 * systemctl now prints a warning if a unit is stopped but a
3597 unit that might trigger it continues to run. Example: a
3598 service is stopped but the socket that activates it is left
3601 * "systemctl status" will now mention if the log output was
3602 shortened due to rotation since a service has been started.
3604 * The journal API now exposes functions to determine the
3605 "cutoff" times due to rotation.
3607 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3608 immediately flushing of runtime logs to /var if possible,
3609 resp. for triggering immediate rotation of the journal
3612 * It is now considered an error if a service is attempted to
3613 be stopped that is not loaded.
3615 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3617 * systemd-analyze now supports Python 3
3619 * tmpfiles now supports cleaning up directories via aging
3620 where the first level dirs are always kept around but
3621 directories beneath it automatically aged. This is enabled
3622 by prefixing the age field with '~'.
3624 * Seat objects now expose CanGraphical, CanTTY properties
3625 which is required to deal with very fast bootups where the
3626 display manager might be running before the graphics drivers
3627 completed initialization.
3629 * Seat objects now expose a State property.
3631 * We now include RPM macros for service enabling/disabling
3632 based on the preset logic. We recommend RPM based
3633 distributions to make use of these macros if possible. This
3634 makes it simpler to reuse RPM spec files across
3637 * We now make sure that the collected systemd unit name is
3638 always valid when services log to the journal via
3641 * There's a new man page kernel-command-line(7) detailing all
3642 command line options we understand.
3644 * The fstab generator may now be disabled at boot by passing
3645 fstab=0 on the kernel command line.
3647 * A new kernel command line option modules-load= is now understood
3648 to load a specific kernel module statically, early at boot.
3650 * Unit names specified on the systemctl command line are now
3651 automatically escaped as needed. Also, if file system or
3652 device paths are specified they are automatically turned
3653 into the appropriate mount or device unit names. Example:
3655 systemctl status /home
3656 systemctl status /dev/sda
3658 * The SysVConsole= configuration option has been removed from
3659 system.conf parsing.
3661 * The SysV search path is no longer exported on the D-Bus
3664 * The Names= option is been removed from unit file parsing.
3666 * There's a new man page bootup(7) detailing the boot process.
3668 * Every unit and every generator we ship with systemd now
3669 comes with full documentation. The self-explanatory boot is
3672 * A couple of services gained "systemd-" prefixes in their
3673 name if they wrap systemd code, rather than only external
3674 code. Among them fsck@.service which is now
3675 systemd-fsck@.service.
3677 * The HaveWatchdog property has been removed from the D-Bus
3680 * systemd.confirm_spawn= on the kernel command line should now
3683 * There's a new man page crypttab(5) which details all options
3684 we actually understand.
3686 * systemd-nspawn gained a new --capability= switch to pass
3687 additional capabilities to the container.
3689 * timedated will now read known NTP implementation unit names
3690 from /usr/lib/systemd/ntp-units.d/*.list,
3691 systemd-timedated-ntp.target has been removed.
3693 * journalctl gained a new switch "-b" that lists log data of
3694 the current boot only.
3696 * The notify socket is in the abstract namespace again, in
3697 order to support daemons which chroot() at start-up.
3699 * There is a new Storage= configuration option for journald
3700 which allows configuration of where log data should go. This
3701 also provides a way to disable journal logging entirely, so
3702 that data collected is only forwarded to the console, the
3703 kernel log buffer or another syslog implementation.
3705 * Many bugfixes and optimizations
3707 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3708 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3709 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3710 Shawn Landden, Tom Gundersen
3714 * "systemctl help <unit>" now shows the man page if one is
3717 * Several new man pages have been added.
3719 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3720 MaxLevelConsole= can now be specified in
3721 journald.conf. These options allow reducing the amount of
3722 data stored on disk or forwarded by the log level.
3724 * TimerSlackNSec= can now be specified in system.conf for
3725 PID1. This allows system-wide power savings.
3727 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3728 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3733 * logind is now capable of (optionally) handling power and
3734 sleep keys as well as the lid switch.
3736 * journalctl now understands the syntax "journalctl
3737 /usr/bin/avahi-daemon" to get all log output of a specific
3740 * CapabilityBoundingSet= in system.conf now also influences
3741 the capability bound set of usermode helpers of the kernel.
3743 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3744 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3745 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3746 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3750 * Note that we skipped 139 releases here in order to set the
3751 new version to something that is greater than both udev's
3752 and systemd's most recent version number.
3754 * udev: all udev sources are merged into the systemd source tree now.
3755 All future udev development will happen in the systemd tree. It
3756 is still fully supported to use the udev daemon and tools without
3757 systemd running, like in initramfs or other init systems. Building
3758 udev though, will require the *build* of the systemd tree, but
3759 udev can be properly *run* without systemd.
3761 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3762 should be used to create dead device nodes as workarounds for broken
3765 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3766 no longer supported. udev_monitor_new_from_netlink() needs to be
3767 used to subscribe to events.
3769 * udev: when udevd is started by systemd, processes which are left
3770 behind by forking them off of udev rules, are unconditionally cleaned
3771 up and killed now after the event handling has finished. Services or
3772 daemons must be started as systemd services. Services can be
3773 pulled-in by udev to get started, but they can no longer be directly
3774 forked by udev rules.
3776 * udev: the daemon binary is called systemd-udevd now and installed
3777 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3778 to adapt to that, create symlink, or rename the binary after building
3781 * libudev no longer provides these symbols:
3782 udev_monitor_from_socket()
3783 udev_queue_get_failed_list_entry()
3784 udev_get_{dev,sys,run}_path()
3785 The versions number was bumped and symbol versioning introduced.
3787 * systemd-loginctl and systemd-journalctl have been renamed
3788 to loginctl and journalctl to match systemctl.
3790 * The config files: /etc/systemd/systemd-logind.conf and
3791 /etc/systemd/systemd-journald.conf have been renamed to
3792 logind.conf and journald.conf. Package updates should rename
3793 the files to the new names on upgrade.
3795 * For almost all files the license is now LGPL2.1+, changed
3796 from the previous GPL2.0+. Exceptions are some minor stuff
3797 of udev (which will be changed to LGPL2.1 eventually, too),
3798 and the MIT licensed sd-daemon.[ch] library that is suitable
3799 to be used as drop-in files.
3801 * systemd and logind now handle system sleep states, in
3802 particular suspending and hibernating.
3804 * logind now implements a sleep/shutdown/idle inhibiting logic
3805 suitable for a variety of uses. Soonishly Lennart will blog
3806 about this in more detail.
3808 * var-run.mount and var-lock.mount are no longer provided
3809 (which prevously bind mounted these directories to their new
3810 places). Distributions which have not converted these
3811 directories to symlinks should consider stealing these files
3812 from git history and add them downstream.
3814 * We introduced the Documentation= field for units and added
3815 this to all our shipped units. This is useful to make it
3816 easier to explore the boot and the purpose of the various
3819 * All smaller setup units (such as
3820 systemd-vconsole-setup.service) now detect properly if they
3821 are run in a container and are skipped when
3822 appropriate. This guarantees an entirely noise-free boot in
3823 Linux container environments such as systemd-nspawn.
3825 * A framework for implementing offline system updates is now
3826 integrated, for details see:
3827 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3829 * A new service type Type=idle is available now which helps us
3830 avoiding ugly interleaving of getty output and boot status
3833 * There's now a system-wide CapabilityBoundingSet= option to
3834 globally reduce the set of capabilities for the
3835 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3836 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3837 even CAP_NET_ADMIN system-wide for secure systems.
3839 * There are now system-wide DefaultLimitXXX= options to
3840 globally change the defaults of the various resource limits
3841 for all units started by PID 1.
3843 * Harald Hoyer's systemd test suite has been integrated into
3844 systemd which allows easy testing of systemd builds in qemu
3845 and nspawn. (This is really awesome! Ask us for details!)
3847 * The fstab parser is now implemented as generator, not inside
3850 * systemctl will now warn you if .mount units generated from
3851 /etc/fstab are out of date due to changes in fstab that
3852 have not been read by systemd yet.
3854 * systemd is now suitable for usage in initrds. Dracut has
3855 already been updated to make use of this. With this in place
3856 initrds get a slight bit faster but primarily are much
3857 easier to introspect and debug since "systemctl status" in
3858 the host system can be used to introspect initrd services,
3859 and the journal from the initrd is kept around too.
3861 * systemd-delta has been added, a tool to explore differences
3862 between user/admin configuration and vendor defaults.
3864 * PrivateTmp= now affects both /tmp and /var/tmp.
3866 * Boot time status messages are now much prettier and feature
3867 proper english language. Booting up systemd has never been
3870 * Read-ahead pack files now include the inode number of all
3871 files to pre-cache. When the inode changes the pre-caching
3872 is not attempted. This should be nicer to deal with updated
3873 packages which might result in changes of read-ahead
3876 * We now temporaritly lower the kernel's read_ahead_kb variable
3877 when collecting read-ahead data to ensure the kernel's
3878 built-in read-ahead does not add noise to our measurements
3879 of necessary blocks to pre-cache.
3881 * There's now RequiresMountsFor= to add automatic dependencies
3882 for all mounts necessary for a specific file system path.
3884 * MountAuto= and SwapAuto= have been removed from
3885 system.conf. Mounting file systems at boot has to take place
3888 * nspawn now learned a new switch --uuid= to set the machine
3889 ID on the command line.
3891 * nspawn now learned the -b switch to automatically search
3894 * vt102 is now the default TERM for serial TTYs, upgraded from
3897 * systemd-logind now works on VT-less systems.
3899 * The build tree has been reorganized. The individual
3900 components now have directories of their own.
3902 * A new condition type ConditionPathIsReadWrite= is now available.
3904 * nspawn learned the new -C switch to create cgroups for the
3905 container in other hierarchies.
3907 * We now have support for hardware watchdogs, configurable in
3910 * The scheduled shutdown logic now has a public API.
3912 * We now mount /tmp as tmpfs by default, but this can be
3913 masked and /etc/fstab can override it.
3915 * Since udisks does not make use of /media anymore we are not
3916 mounting a tmpfs on it anymore.
3918 * journalctl gained a new --local switch to only interleave
3919 locally generated journal files.
3921 * We can now load the IMA policy at boot automatically.
3923 * The GTK tools have been split off into a systemd-ui.
3925 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3926 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3927 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3928 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3929 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3930 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3931 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3932 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3933 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3938 * This is mostly a bugfix release
3940 * Support optional initialization of the machine ID from the
3941 KVM or container configured UUID.
3943 * Support immediate reboots with "systemctl reboot -ff"
3945 * Show /etc/os-release data in systemd-analyze output
3947 * Many bugfixes for the journal, including endianness fixes and
3948 ensuring that disk space enforcement works
3950 * sd-login.h is C++ comptaible again
3952 * Extend the /etc/os-release format on request of the Debian
3955 * We now refuse non-UTF8 strings used in various configuration
3956 and unit files. This is done to ensure we do not pass invalid
3957 data over D-Bus or expose it elsewhere.
3959 * Register Mimo USB Screens as suitable for automatic seat
3962 * Read SELinux client context from journal clients in a race
3965 * Reorder configuration file lookup order. /etc now always
3966 overrides /run in order to allow the administrator to always
3967 and unconditionally override vendor supplied or
3968 automatically generated data.
3970 * The various user visible bits of the journal now have man
3971 pages. We still lack man pages for the journal API calls
3974 * We now ship all man pages in HTML format again in the
3977 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3978 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3979 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3984 * This is mostly a bugfix release
3986 * systems lacking /etc/os-release are no longer supported.
3988 * Various functionality updates to libsystemd-login.so
3990 * Track class of PAM logins to distuingish greeters from
3993 Contributions from: Kay Sievers, Lennart Poettering, Michael
3998 * This is an important bugfix release for v41.
4000 * Building man pages is now optional which should be useful
4001 for those building systemd from git but unwilling to install
4004 * Watchdog support for supervising services is now usable. In
4005 a future release support for hardware watchdogs
4006 (i.e. /dev/watchdog) will be added building on this.
4008 * Service start rate limiting is now configurable and can be
4009 turned off per service. When a start rate limit is hit a
4010 reboot can automatically be triggered.
4012 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4014 Contributions from: Benjamin Franzke, Bill Nottingham,
4015 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4016 Schmidt, Michał Górny, Piotr Drąg
4020 * The systemd binary is installed /usr/lib/systemd/systemd now;
4021 An existing /sbin/init symlink needs to be adapted with the
4024 * The code that loads kernel modules has been ported to invoke
4025 libkmod directly, instead of modprobe. This means we do not
4026 support systems with module-init-tools anymore.
4028 * Watchdog support is now already useful, but still not
4031 * A new kernel command line option systemd.setenv= is
4032 understood to set system wide environment variables
4033 dynamically at boot.
4035 * We now limit the set of capabilities of systemd-journald.
4037 * We now set SIGPIPE to ignore by default, since it only is
4038 useful in shell pipelines, and has little use in general
4039 code. This can be disabled with IgnoreSIPIPE=no in unit
4042 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4043 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4048 * This is mostly a bugfix release
4050 * We now expose the reason why a service failed in the
4051 "Result" D-Bus property.
4053 * Rudimentary service watchdog support (will be completed over
4054 the next few releases.)
4056 * When systemd forks off in order execute some service we will
4057 now immediately changes its argv[0] to reflect which process
4058 it will execute. This is useful to minimize the time window
4059 with a generic argv[0], which makes bootcharts more useful
4061 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4062 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4063 Mike Kazantsev, Ray Strode
4067 * This is mostly a test release, but incorporates many
4070 * New systemd-cgtop tool to show control groups by their
4073 * Linking against libacl for ACLs is optional again. If
4074 disabled, support tracking device access for active logins
4075 goes becomes unavailable, and so does access to the user
4076 journals by the respective users.
4078 * If a group "adm" exists, journal files are automatically
4079 owned by them, thus allow members of this group full access
4080 to the system journal as well as all user journals.
4082 * The journal now stores the SELinux context of the logging
4083 client for all entries.
4085 * Add C++ inclusion guards to all public headers
4087 * New output mode "cat" in the journal to print only text
4088 messages, without any meta data like date or time.
4090 * Include tiny X server wrapper as a temporary stop-gap to
4091 teach XOrg udev display enumeration. This is used by display
4092 managers such as gdm, and will go away as soon as XOrg
4093 learned native udev hotplugging for display devices.
4095 * Add new systemd-cat tool for executing arbitrary programs
4096 with STDERR/STDOUT connected to the journal. Can also act as
4097 BSD logger replacement, and does so by default.
4099 * Optionally store all locally generated coredumps in the
4100 journal along with meta data.
4102 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4103 writing short strings to files (for usage for /sys), and for
4104 creating symlinks, character and block device nodes.
4106 * New unit file option ControlGroupPersistent= to make cgroups
4107 persistent, following the mechanisms outlined in
4108 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4110 * Support multiple local RTCs in a sane way
4112 * No longer monopolize IO when replaying readahead data on
4113 rotating disks, since we might starve non-file-system IO to
4114 death, since fanotify() will not see accesses done by blkid,
4117 * Do not show kernel threads in systemd-cgls anymore, unless
4118 requested with new -k switch.
4120 Contributions from: Dan Horák, Kay Sievers, Lennart
4121 Poettering, Michal Schmidt
4125 * This is mostly a test release, but incorporates many
4128 * The git repository moved to:
4129 git://anongit.freedesktop.org/systemd/systemd
4130 ssh://git.freedesktop.org/git/systemd/systemd
4132 * First release with the journal
4133 http://0pointer.de/blog/projects/the-journal.html
4135 * The journal replaces both systemd-kmsg-syslogd and
4136 systemd-stdout-bridge.
4138 * New sd_pid_get_unit() API call in libsystemd-logind
4140 * Many systemadm clean-ups
4142 * Introduce remote-fs-pre.target which is ordered before all
4143 remote mounts and may be used to start services before all
4146 * Added Mageia support
4148 * Add bash completion for systemd-loginctl
4150 * Actively monitor PID file creation for daemons which exit in
4151 the parent process before having finished writing the PID
4152 file in the daemon process. Daemons which do this need to be
4153 fixed (i.e. PID file creation must have finished before the
4154 parent exits), but we now react a bit more gracefully to them.
4156 * Add colourful boot output, mimicking the well-known output
4157 of existing distributions.
4159 * New option PassCredentials= for socket units, for
4160 compatibility with a recent kernel ABI breakage.
4162 * /etc/rc.local is now hooked in via a generator binary, and
4163 thus will no longer act as synchronization point during
4166 * systemctl list-unit-files now supports --root=.
4168 * systemd-tmpfiles now understands two new commands: z, Z for
4169 relabelling files according to the SELinux database. This is
4170 useful to apply SELinux labels to specific files in /sys,
4173 * Output of SysV services is now forwarded to both the console
4174 and the journal by default, not only just the console.
4176 * New man pages for all APIs from libsystemd-login.
4178 * The build tree got reorganized and a the build system is a
4179 lot more modular allowing embedded setups to specifically
4180 select the components of systemd they are interested in.
4182 * Support for Linux systems lacking the kernel VT subsystem is
4185 * configure's --with-rootdir= got renamed to
4186 --with-rootprefix= to follow the naming used by udev and
4189 * Unless specified otherwise we will now install to /usr instead
4190 of /usr/local by default.
4192 * Processes with '@' in argv[0][0] are now excluded from the
4193 final shut-down killing spree, following the logic explained
4195 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4197 * All processes remaining in a service cgroup when we enter
4198 the START or START_PRE states are now killed with
4199 SIGKILL. That means it is no longer possible to spawn
4200 background processes from ExecStart= lines (which was never
4201 supported anyway, and bad style).
4203 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4204 reloading of units together.
4206 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4207 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4208 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4209 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4210 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek