1 systemd System and Service Manager
5 * journalctl gained the new options -t/--identifier= to match
6 on the syslog identifier (aka "tag"), as well as --utc to
7 show log timestamps in the UTC timezone. journalctl now also
8 accepts -n/--lines=all to disable line capping in a pager.
10 * Services can notify the manager before they start a reload
11 (by sending RELOADING=1) or shutdown (by sending
12 STOPPING=1). This allows the manager to track and show the
13 internal state of daemons and closes a race condition when
14 the process is still running but has closed its D-Bus
17 * Services with Type=oneshot do not have to have any ExecStart
20 * User units are now loaded also from
21 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
22 /run/systemd/user directory that was already previously
23 supported, but is under the control of the user.
25 * Job timeouts (i.e. time-outs on the time a job that is
26 queued stays in the run queue) can now optionally result in
27 immediate reboot or power-off actions (JobTimeoutAction= and
28 JobTimeoutRebootArgument=). This is useful on ".target"
29 units, to limit the maximum time a target remains
30 undispatched in the run queue, and to trigger an emergency
31 operation in such a case. This is now used by default to
32 turn off the system if boot-up (as defined by everything in
33 basic.target) hangs and does not complete for at least
34 15min. Also, if power-off or reboot hang for at least 30min
35 an immediate power-off/reboot operation is triggered. This
36 functionality is particularly useful to increase reliability
37 on embedded devices, but also on laptops which might
38 accidentally get powered on when carried in a backpack and
39 whose boot stays stuck in a hard disk encryption passphrase
42 * systemd-logind can be configured to also handle lid switch
43 events even when the machine is docked or multiple displays
44 are attached (HandleLidSwitchDocked= option).
46 * A helper binary and a service have been added which can be
47 used to resume from hibernation in the initramfs. A
48 generator will parse the resume= option on the kernel
49 command-line to trigger resume.
51 * A user console daemon systemd-consoled has been
52 added. Currently, it is a preview, and will so far open a
53 single terminal on each session of the user marked as
54 Desktop=systemd-console.
56 * Route metrics can be specified for DHCP routes added by
59 * The SELinux context of socket-actived services can be set
60 from the information provided by the networking stack
61 (SELinuxContextFromNet= option).
63 * Userspace firmware loading support has been removed and
64 the minimum supported kernel version is thus bumped to 3.7.
66 * Timeout for udev workers has been increased from 1 to 3
67 minutes, but a warning will be printed after 1 minute to
68 help diagnose kernel modules that take a long time to load.
70 * Udev rules can now remove tags on devices with TAG-="foobar".
72 * systemd's readahead implementation has been removed. In many
73 circumstatances it didn't give expected benefits even for
74 rotational disk drives and was becoming less relevant in the
75 age of SSDs. As none of the developers has been using
76 rotating media anymore, and nobody stepped up to actively
77 maintain this component of systemd it has now been removed.
79 * Swap units can use Discard= to specify discard options.
80 Discard options specified for swaps in /etc/fstab are now
83 * Docker containers are now detected as a separate type of
86 * The Password Agent protocol gained support for queries where
87 the user input is shown, useful e.g. for usernames.
88 systemd-ask-password gained a new --echo option to turn that
91 * The default sysctl.d/ snippets will now set:
93 net.core.default_qdisc = fq_codel
95 This selects Fair Queueing Controlled Delay as the default
96 queueing discipline for network interfaces. fq_codel helps
97 fight the network bufferbloat problem. It is believed to be
98 a good default with no tuning required for most workloads.
99 Downstream distributions may override this choice. On 10Gbit
100 servers that do not do forwarding, "fq" may perform better.
101 Systems without a good clocksource should use "pfifo_fast".
103 * If kdbus is enabled during build a new option BusPolicy= is
104 available for service units, that allows locking all service
105 processes into a stricter bus policy, in order to limit
106 access to various bus services, or even hide most of them
107 from the service's view entirely.
109 * networkctl will now show the .network and .link file
110 networkd has applied to a specific interface.
112 * sd-login gained a new API call sd_session_get_desktop() to
113 query which desktop environment has been selected for a
116 * UNIX utmp support is now compile-time optional to support
119 * systemctl gained two new commands "add-wants" and
120 "add-requires" for pulling in units from specific targets
123 * If the word "rescue" is specified on the kernel command line
124 the system will now boot into rescue mode (aka
125 rescue.target), which was previously available only by
126 specifying "1" or "systemd.unit=rescue.target" on the kernel
127 command line. This new kernel command line option nicely
128 mirrors the already existing "emergency" kernel command line
131 * New kernel command line options mount.usr=, mount.usrflags=,
132 usrfstype= have been added that match root=, rootflags=,
133 rootfstype= but allow mounting a specific file system to
136 * The $NOTIFY_SOCKET is now also passed to control processesof
137 services, not only the main process.
139 * This version reenables support for fsck's -l switch. This
140 means at least version v2.25 of util-linux is required for
141 operation, otherwise dead-locks on device nodes may
142 occur. Again: you need to update util-linux to at least
143 v2.25 when updating systemd to v217.
145 * The "multi-seat-x" tool has been removed from systemd, as
146 its functionality has been integrated into X servers 1.16,
147 and the tool is hence redundant. It is recommended to update
148 display managers invoking this tool to simply invoke X
149 directly from now on, again.
151 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
152 message flag has been added for all of systemd's PolicyKit
153 authenticated method calls has been added. In particular
154 this now allows optional interactive authorization via
155 PolicyKit for many of PID1's priviliged operations such as
156 unit file enabling and disabling.
158 * "udevadm hwdb --update" learnt a new switch "--usr" for
159 placing the rebuilt hardware database in /usr instead of
160 /etc. When used only hardware database entries stored in
161 /usr will be used, and any user database entries in /etc are
162 ignored. This functionality is useful for vendors to ship a
163 pre-built database on systems where local configuration is
164 unnecessary or unlikely.
166 Contributions from: Andreas Henriksson, Andrei Borzenkov,
167 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
168 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
169 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
170 Herrmann, David Sommerseth, David Strauss, Emil Renner
171 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
172 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
173 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
174 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
175 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
176 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
177 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
178 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
179 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
180 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
181 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
182 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
183 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
184 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
185 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
186 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
189 -- Berlin, 2014-10-28
193 * timedated no longer reads NTP implementation unit names from
194 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
195 implementations should add a
197 Conflicts=systemd-timesyncd.service
199 to their unit files to take over and replace systemd's NTP
200 default functionality.
202 * systemd-sysusers gained a new line type "r" for configuring
203 which UID/GID ranges to allocate system users/groups
204 from. Lines of type "u" may now add an additional column
205 that specifies the home directory for the system user to be
206 created. Also, systemd-sysusers may now optionally read user
207 information from STDIN instead of a file. This is useful for
208 invoking it from RPM preinst scriptlets that need to create
209 users before the first RPM file is installed since these
210 files might need to be owned by them. A new
211 %sysusers_create_inline RPM macro has been introduced to do
212 just that. systemd-sysusers now updates the shadow files as
213 well as the user/group databases, which should enhance
214 compatibility with certain tools like grpck.
216 * A number of bus APIs of PID 1 now optionally consult
217 PolicyKit to permit access for otherwise unprivileged
218 clients under certain conditions. Note that this currently
219 doesn't support interactive authentication yet, but this is
220 expected to be added eventually, too.
222 * /etc/machine-info now has new fields for configuring the
223 deployment environment of the machine, as well as the
224 location of the machine. hostnamectl has been updated with
225 new command to update these fields.
227 * systemd-timesyncd has been updated to automatically acquire
228 NTP server information from systemd-networkd, which might
229 have been discovered via DHCP.
231 * systemd-resolved now includes a caching DNS stub resolver
232 and a complete LLMNR name resolution implementation. A new
233 NSS module "nss-resolve" has been added which can be used
234 instead of glibc's own "nss-dns" to resolve hostnames via
235 systemd-resolved. Hostnames, addresses and arbitrary RRs may
236 be resolved via systemd-resolved D-Bus APIs. In contrast to
237 the glibc internal resolver systemd-resolved is aware of
238 multi-homed system, and keeps DNS server and caches separate
239 and per-interface. Queries are sent simultaneously on all
240 interfaces that have DNS servers configured, in order to
241 properly handle VPNs and local LANs which might resolve
242 separate sets of domain names. systemd-resolved may acquire
243 DNS server information from systemd-networkd automatically,
244 which in turn might have discovered them via DHCP. A tool
245 "systemd-resolve-host" has been added that may be used to
246 query the DNS logic in resolved. systemd-resolved implements
247 IDNA and automatically uses IDNA or UTF-8 encoding depending
248 on whether classic DNS or LLMNR is used as transport. In the
249 next releases we intend to add a DNSSEC and mDNS/DNS-SD
250 implementation to systemd-resolved.
252 * A new NSS module nss-mymachines has been added, that
253 automatically resolves the names of all local registered
254 containers to their respective IP addresses.
256 * A new client tool "networkctl" for systemd-networkd has been
257 added. It currently is entirely passive and will query
258 networking configuration from udev, rtnetlink and networkd,
259 and present it to the user in a very friendly
260 way. Eventually, we hope to extend it to become a full
261 control utility for networkd.
263 * .socket units gained a new DeferAcceptSec= setting that
264 controls the kernels' TCP_DEFER_ACCEPT sockopt for
265 TCP. Similar, support for controlling TCP keep-alive
266 settings has been added (KeepAliveTimeSec=,
267 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
268 turning off Nagle's algorithm on TCP has been added
271 * logind learned a new session type "web", for use in projects
272 like Cockpit which register web clients as PAM sessions.
274 * timer units with at least one OnCalendar= setting will now
275 be started only after timer-sync.target has been
276 reached. This way they will not elapse before the system
277 clock has been corrected by a local NTP client or
278 similar. This is particular useful on RTC-less embedded
279 machines, that come up with an invalid system clock.
281 * systemd-nspawn's --network-veth= switch should now result in
282 stable MAC addresses for both the outer and the inner side
285 * systemd-nspawn gained a new --volatile= switch for running
286 container instances with /etc or /var unpopulated.
288 * The kdbus client code has been updated to use the new Linux
289 3.17 memfd subsystem instead of the old kdbus-specific one.
291 * systemd-networkd's DHCP client and server now support
292 FORCERENEW. There are also new configuration options to
293 configure the vendor client identifier and broadcast mode
296 * systemd will no longer inform the kernel about the current
297 timezone, as this is necessarily incorrect and racy as the
298 kernel has no understanding of DST and similar
299 concepts. This hence means FAT timestamps will be always
300 considered UTC, similar to what Android is already
301 doing. Also, when the RTC is configured to the local time
302 (rather than UTC) systemd will never synchronize back to it,
303 as this might confuse Windows at a later boot.
305 * systemd-analyze gained a new command "verify" for offline
306 validation of unit files.
308 * systemd-networkd gained support for a couple of additional
309 settings for bonding networking setups. Also, the metric for
310 statically configured routes may now be configured. For
311 network interfaces where this is appropriate the peer IP
312 address may now be configured.
314 * systemd-networkd's DHCP client will no longer request
315 broadcasting by default, as this tripped up some networks.
316 For hardware where broadcast is required the feature should
317 be switched back on using RequestBroadcast=yes.
319 * systemd-networkd will now set up IPv4LL addresses (when
320 enabled) even if DHCP is configured successfully.
322 * udev will now default to respect network device names given
323 by the kernel when the kernel indicates that these are
324 predictable. This behavior can be tweaked by changing
325 NamePolicy= in the relevant .link file.
327 * A new library systemd-terminal has been added that
328 implements full TTY stream parsing and rendering. This
329 library is supposed to be used later on for implementing a
330 full userspace VT subsystem, replacing the current kernel
333 * A new tool systemd-journal-upload has been added to push
334 journal data to a remote system running
335 systemd-journal-remote.
337 * journald will no longer forward all local data to another
338 running syslog daemon. This change has been made because
339 rsyslog (which appears to be the most commonly used syslog
340 implementation these days) no longer makes use of this, and
341 instead pulls the data out of the journal on its own. Since
342 forwarding the messages to a non-existent syslog server is
343 more expensive than we assumed we have now turned this
344 off. If you run a syslog server that is not a recent rsyslog
345 version, you have to turn this option on again
346 (ForwardToSyslog= in journald.conf).
348 * journald now optionally supports the LZ4 compressor for
349 larger journal fields. This compressor should perform much
350 better than XZ which was the previous default.
352 * machinectl now shows the IP addresses of local containers,
353 if it knows them, plus the interface name of the container.
355 * A new tool "systemd-escape" has been added that makes it
356 easy to escape strings to build unit names and similar.
358 * sd_notify() messages may now include a new ERRNO= field
359 which is parsed and collected by systemd and shown among the
360 "systemctl status" output for a service.
362 * A new component "systemd-firstboot" has been added that
363 queries the most basic systemd information (timezone,
364 hostname, root password) interactively on first
365 boot. Alternatively it may also be used to provision these
366 things offline on OS images installed into directories.
368 * The default sysctl.d/ snippets will now set
370 net.ipv4.conf.default.promote_secondaries=1
372 This has the benefit of no flushing secondary IP addresses
373 when primary addresses are removed.
375 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
376 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
377 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
378 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
379 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
380 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
381 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
382 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
383 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
384 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
385 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
386 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
387 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
388 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
389 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
391 -- Berlin, 2014-08-19
395 * A new tool systemd-sysusers has been added. This tool
396 creates system users and groups in /etc/passwd and
397 /etc/group, based on static declarative system user/group
398 definitions in /usr/lib/sysusers.d/. This is useful to
399 enable factory resets and volatile systems that boot up with
400 an empty /etc directory, and thus need system users and
401 groups created during early boot. systemd now also ships
402 with two default sysusers.d/ files for the most basic
403 users and groups systemd and the core operating system
406 * A new tmpfiles snippet has been added that rebuilds the
407 essential files in /etc on boot, should they be missing.
409 * A directive for ensuring automatic clean-up of
410 /var/cache/man/ has been removed from the default
411 configuration. This line should now be shipped by the man
412 implementation. The necessary change has been made to the
413 man-db implementation. Note that you need to update your man
414 implementation to one that ships this line, otherwise no
415 automatic clean-up of /var/cache/man will take place.
417 * A new condition ConditionNeedsUpdate= has been added that
418 may conditionalize services to only run when /etc or /var
419 are "older" than the vendor operating system resources in
420 /usr. This is useful for reconstructing or updating /etc
421 after an offline update of /usr or a factory reset, on the
422 next reboot. Services that want to run once after such an
423 update or reset should use this condition and order
424 themselves before the new systemd-update-done.service, which
425 will mark the two directories as fully updated. A number of
426 service files have been added making use of this, to rebuild
427 the udev hardware database, the journald message catalog and
428 dynamic loader cache (ldconfig). The systemd-sysusers tool
429 described above also makes use of this now. With this in
430 place it is now possible to start up a minimal operating
431 system with /etc empty cleanly. For more information on the
432 concepts involved see this recent blog story:
434 http://0pointer.de/blog/projects/stateless.html
436 * A new system group "input" has been introduced, and all
437 input device nodes get this group assigned. This is useful
438 for system-level software to get access to input devices. It
439 complements what is already done for "audio" and "video".
441 * systemd-networkd learnt minimal DHCPv4 server support in
442 addition to the existing DHCPv4 client support. It also
443 learnt DHCPv6 client and IPv6 Router Solicitation client
444 support. The DHCPv4 client gained support for static routes
445 passed in from the server. Note that the [DHCPv4] section
446 known in older systemd-networkd versions has been renamed to
447 [DHCP] and is now also used by the DHCPv6 client. Existing
448 .network files using settings of this section should be
449 updated, though compatibility is maintained. Optionally, the
450 client hostname may now be sent to the DHCP server.
452 * networkd gained support for vxlan virtual networks as well
453 as tun/tap and dummy devices.
455 * networkd gained support for automatic allocation of address
456 ranges for interfaces from a system-wide pool of
457 addresses. This is useful for dynamically managing a large
458 number of interfaces with a single network configuration
459 file. In particular this is useful to easily assign
460 appropriate IP addresses to the veth links of a large number
463 * RPM macros for processing sysusers, sysctl and binfmt
464 drop-in snippets at package installation time have been
467 * The /etc/os-release file should now be placed in
468 /usr/lib/os-release. The old location is automatically
469 created as symlink. /usr/lib is the more appropriate
470 location of this file, since it shall actually describe the
471 vendor operating system shipped in /usr, and not the
472 configuration stored in /etc.
474 * .mount units gained a new boolean SloppyOptions= setting
475 that maps to mount(8)'s -s option which enables permissive
476 parsing of unknown mount options.
478 * tmpfiles learnt a new "L+" directive which creates a symlink
479 but (unlike "L") deletes a pre-existing file first, should
480 it already exist and not already be the correct
481 symlink. Similar, "b+", "c+" and "p+" directives have been
482 added as well, which create block and character devices, as
483 well as fifos in the filesystem, possibly removing any
484 pre-existing files of different types.
486 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
487 'argument' field (which so far specified the source to
488 symlink/copy the files from) is now optional. If omitted the
489 same file os copied from /usr/share/factory/ suffixed by the
490 full destination path. This is useful for populating /etc
491 with essential files, by copying them from vendor defaults
492 shipped in /usr/share/factory/etc.
494 * A new command "systemctl preset-all" has been added that
495 applies the service preset settings to all installed unit
496 files. A new switch --preset-mode= has been added that
497 controls whether only enable or only disable operations
500 * A new command "systemctl is-system-running" has been added
501 that allows checking the overall state of the system, for
502 example whether it is fully up and running.
504 * When the system boots up with an empty /etc, the equivalent
505 to "systemctl preset-all" is executed during early boot, to
506 make sure all default services are enabled after a factory
509 * systemd now contains a minimal preset file that enables the
510 most basic services systemd ships by default.
512 * Unit files' [Install] section gained a new DefaultInstance=
513 field for defining the default instance to create if a
514 template unit is enabled with no instance specified.
516 * A new passive target cryptsetup-pre.target has been added
517 that may be used by services that need to make they run and
518 finish before the first LUKS cryptographic device is set up.
520 * The /dev/loop-control and /dev/btrfs-control device nodes
521 are now owned by the "disk" group by default, opening up
522 access to this group.
524 * systemd-coredump will now automatically generate a
525 stack trace of all core dumps taking place on the system,
526 based on elfutils' libdw library. This stack trace is logged
529 * systemd-coredump may now optionally store coredumps directly
530 on disk (in /var/lib/systemd/coredump, possibly compressed),
531 instead of storing them unconditionally in the journal. This
532 mode is the new default. A new configuration file
533 /etc/systemd/coredump.conf has been added to configure this
534 and other parameters of systemd-coredump.
536 * coredumpctl gained a new "info" verb to show details about a
537 specific coredump. A new switch "-1" has also been added
538 that makes sure to only show information about the most
539 recent entry instead of all entries. Also, as the tool is
540 generally useful now the "systemd-" prefix of the binary
541 name has been removed. Distributions that want to maintain
542 compatibility with the old name should add a symlink from
543 the old name to the new name.
545 * journald's SplitMode= now defaults to "uid". This makes sure
546 that unprivileged users can access their own coredumps with
547 coredumpctl without restrictions.
549 * New kernel command line options "systemd.wants=" (for
550 pulling an additional unit during boot), "systemd.mask="
551 (for masking a specific unit for the boot), and
552 "systemd.debug-shell" (for enabling the debug shell on tty9)
553 have been added. This is implemented in the new generator
554 "systemd-debug-generator".
556 * systemd-nspawn will now by default filter a couple of
557 syscalls for containers, among them those required for
558 kernel module loading, direct x86 IO port access, swap
559 management, and kexec. Most importantly though
560 open_by_handle_at() is now prohibited for containers,
561 closing a hole similar to a recently discussed vulnerability
562 in docker regarding access to files on file hierarchies the
563 container should normally not have access to. Note that for
564 nspawn we generally make no security claims anyway (and
565 this is explicitly documented in the man page), so this is
566 just a fix for one of the most obvious problems.
568 * A new man page file-hierarchy(7) has been added that
569 contains a minimized, modernized version of the file system
570 layout systemd expects, similar in style to the FHS
571 specification or hier(5). A new tool systemd-path(1) has
572 been added to query many of these paths for the local
575 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
576 longer done. Since the directory now has a per-user size
577 limit, and is cleaned on logout this appears unnecessary,
578 in particular since this now brings the lifecycle of this
579 directory closer in line with how IPC objects are handled.
581 * systemd.pc now exports a number of additional directories,
582 including $libdir (which is useful to identify the library
583 path for the primary architecture of the system), and a
584 couple of drop-in directories.
586 * udev's predictable network interface names now use the dev_port
587 sysfs attribute, introduced in linux 3.15 instead of dev_id to
588 distinguish between ports of the same PCI function. dev_id should
589 only be used for ports using the same HW address, hence the need
592 * machined has been updated to export the OS version of a
593 container (read from /etc/os-release and
594 /usr/lib/os-release) on the bus. This is now shown in
595 "machinectl status" for a machine.
597 * A new service setting RestartForceExitStatus= has been
598 added. If configured to a set of exit signals or process
599 return values, the service will be restarted when the main
600 daemon process exits with any of them, regardless of the
603 * systemctl's -H switch for connecting to remote systemd
604 machines has been extended so that it may be used to
605 directly connect to a specific container on the
606 host. "systemctl -H root@foobar:waldi" will now connect as
607 user "root" to host "foobar", and then proceed directly to
608 the container named "waldi". Note that currently you have to
609 authenticate as user "root" for this to work, as entering
610 containers is a privileged operation.
612 Contributions from: Andreas Henriksson, Benjamin Steinwender,
613 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
614 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
615 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
616 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
617 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
618 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
619 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
620 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
621 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
622 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
623 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
625 -- Berlin, 2014-07-03
629 * As an experimental feature, udev now tries to lock the
630 disk device node (flock(LOCK_SH|LOCK_NB)) while it
631 executes events for the disk or any of its partitions.
632 Applications like partitioning programs can lock the
633 disk device node (flock(LOCK_EX)) and claim temporary
634 device ownership that way; udev will entirely skip all event
635 handling for this disk and its partitions. If the disk
636 was opened for writing, the close will trigger a partition
637 table rescan in udev's "watch" facility, and if needed
638 synthesize "change" events for the disk and all its partitions.
639 This is now unconditionally enabled, and if it turns out to
640 cause major problems, we might turn it on only for specific
641 devices, or might need to disable it entirely. Device Mapper
642 devices are excluded from this logic.
644 * We temporarily dropped the "-l" switch for fsck invocations,
645 since they collide with the flock() logic above. util-linux
646 upstream has been changed already to avoid this conflict,
647 and we will readd "-l" as soon as util-linux with this
648 change has been released.
650 * The dependency on libattr has been removed. Since a long
651 time, the extended attribute calls have moved to glibc, and
652 libattr is thus unnecessary.
654 * Virtualization detection works without priviliges now. This
655 means the systemd-detect-virt binary no longer requires
656 CAP_SYS_PTRACE file capabilities, and our daemons can run
657 with fewer privileges.
659 * systemd-networkd now runs under its own "systemd-network"
660 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
661 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
662 loses the ability to write to files owned by root this way.
664 * Similar, systemd-resolved now runs under its own
665 "systemd-resolve" user with no capabilities remaining.
667 * Similar, systemd-bus-proxyd now runs under its own
668 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
670 * systemd-networkd gained support for setting up "veth"
671 virtual ethernet devices for container connectivity, as well
672 as GRE and VTI tunnels.
674 * systemd-networkd will no longer automatically attempt to
675 manually load kernel modules necessary for certain tunnel
676 transports. Instead, it is assumed the kernel loads them
677 automatically when required. This only works correctly on
678 very new kernels. On older kernels, please consider adding
679 the kernel modules to /etc/modules-load.d/ as a work-around.
681 * The resolv.conf file systemd-resolved generates has been
682 moved to /run/systemd/resolve/. If you have a symlink from
683 /etc/resolv.conf, it might be necessary to correct it.
685 * Two new service settings, ProtectHome= and ProtectSystem=,
686 have been added. When enabled, they will make the user data
687 (such as /home) inaccessible or read-only and the system
688 (such as /usr) read-only, for specific services. This allows
689 very light-weight per-service sandboxing to avoid
690 modifications of user data or system files from
691 services. These two new switches have been enabled for all
692 of systemd's long-running services, where appropriate.
694 * Socket units gained new SocketUser= and SocketGroup=
695 settings to set the owner user and group of AF_UNIX sockets
696 and FIFOs in the file system.
698 * Socket units gained a new RemoveOnStop= setting. If enabled,
699 all FIFOS and sockets in the file system will be removed
700 when the specific socket unit is stopped.
702 * Socket units gained a new Symlinks= setting. It takes a list
703 of symlinks to create to file system sockets or FIFOs
704 created by the specific Unix sockets. This is useful to
705 manage symlinks to socket nodes with the same life-cycle as
708 * The /dev/log socket and /dev/initctl FIFO have been moved to
709 /run, and have been replaced by symlinks. This allows
710 connecting to these facilities even if PrivateDevices=yes is
711 used for a service (which makes /dev/log itself unavailable,
712 but /run is left). This also has the benefit of ensuring
713 that /dev only contains device nodes, directories and
714 symlinks, and nothing else.
716 * sd-daemon gained two new calls sd_pid_notify() and
717 sd_pid_notifyf(). They are similar to sd_notify() and
718 sd_notifyf(), but allow overriding of the source PID of
719 notification messages if permissions permit this. This is
720 useful to send notify messages on behalf of a different
721 process (for example, the parent process). The
722 systemd-notify tool has been updated to make use of this
723 when sending messages (so that notification messages now
724 originate from the shell script invoking systemd-notify and
725 not the systemd-notify process itself. This should minimize
726 a race where systemd fails to associate notification
727 messages to services when the originating process already
730 * A new "on-abnormal" setting for Restart= has been added. If
731 set, it will result in automatic restarts on all "abnormal"
732 reasons for a process to exit, which includes unclean
733 signals, core dumps, timeouts and watchdog timeouts, but
734 does not include clean and unclean exit codes or clean
735 signals. Restart=on-abnormal is an alternative for
736 Restart=on-failure for services that shall be able to
737 terminate and avoid restarts on certain errors, by
738 indicating so with an unclean exit code. Restart=on-failure
739 or Restart=on-abnormal is now the recommended setting for
740 all long-running services.
742 * If the InaccessibleDirectories= service setting points to a
743 mount point (or if there are any submounts contained within
744 it), it is now attempted to completely unmount it, to make
745 the file systems truly unavailable for the respective
748 * The ReadOnlyDirectories= service setting and
749 systemd-nspawn's --read-only parameter are now recursively
750 applied to all submounts, too.
752 * Mount units may now be created transiently via the bus APIs.
754 * The support for SysV and LSB init scripts has been removed
755 from the systemd daemon itself. Instead, it is now
756 implemented as a generator that creates native systemd units
757 from these scripts when needed. This enables us to remove a
758 substantial amount of legacy code from PID 1, following the
759 fact that many distributions only ship a very small number
760 of LSB/SysV init scripts nowadays.
762 * Priviliged Xen (dom0) domains are not considered
763 virtualization anymore by the virtualization detection
764 logic. After all, they generally have unrestricted access to
765 the hardware and usually are used to manage the unprivileged
768 * systemd-tmpfiles gained a new "C" line type, for copying
769 files or entire directories.
771 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
772 lines. So far, they have been non-globbing versions of the
773 latter, and have thus been redundant. In future, it is
774 recommended to only use "z". "m" has hence been removed
775 from the documentation, even though it stays supported.
777 * A tmpfiles snippet to recreate the most basic structure in
778 /var has been added. This is enough to create the /var/run →
779 /run symlink and create a couple of structural
780 directories. This allows systems to boot up with an empty or
781 volatile /var. Of course, while with this change, the core OS
782 now is capable with dealing with a volatile /var, not all
783 user services are ready for it. However, we hope that sooner
784 or later, many service daemons will be changed upstream so
785 that they are able to automatically create their necessary
786 directories in /var at boot, should they be missing. This is
787 the first step to allow state-less systems that only require
788 the vendor image for /usr to boot.
790 * systemd-nspawn has gained a new --tmpfs= switch to mount an
791 empty tmpfs instance to a specific directory. This is
792 particularly useful for making use of the automatic
793 reconstruction of /var (see above), by passing --tmpfs=/var.
795 * Access modes specified in tmpfiles snippets may now be
796 prefixed with "~", which indicates that they shall be masked
797 by whether the existing file or directory is currently
798 writable, readable or executable at all. Also, if specified,
799 the sgid/suid/sticky bits will be masked for all
802 * A new passive target unit "network-pre.target" has been
803 added which is useful for services that shall run before any
804 network is configured, for example firewall scripts.
806 * The "floppy" group that previously owned the /dev/fd*
807 devices is no longer used. The "disk" group is now used
808 instead. Distributions should probably deprecate usage of
811 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
812 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
813 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
814 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
815 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
816 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
819 -- Berlin, 2014-06-11
823 * A new "systemd-timesyncd" daemon has been added for
824 synchronizing the system clock across the network. It
825 implements an SNTP client. In contrast to NTP
826 implementations such as chrony or the NTP reference server,
827 this only implements a client side, and does not bother with
828 the full NTP complexity, focusing only on querying time from
829 one remote server and synchronizing the local clock to
830 it. Unless you intend to serve NTP to networked clients or
831 want to connect to local hardware clocks, this simple NTP
832 client should be more than appropriate for most
833 installations. The daemon runs with minimal privileges, and
834 has been hooked up with networkd to only operate when
835 network connectivity is available. The daemon saves the
836 current clock to disk every time a new NTP sync has been
837 acquired, and uses this to possibly correct the system clock
838 early at bootup, in order to accommodate for systems that
839 lack an RTC such as the Raspberry Pi and embedded devices,
840 and to make sure that time monotonically progresses on these
841 systems, even if it is not always correct. To make use of
842 this daemon, a new system user and group "systemd-timesync"
843 needs to be created on installation of systemd.
845 * The queue "seqnum" interface of libudev has been disabled, as
846 it was generally incompatible with device namespacing as
847 sequence numbers of devices go "missing" if the devices are
848 part of a different namespace.
850 * "systemctl list-timers" and "systemctl list-sockets" gained
851 a --recursive switch for showing units of these types also
852 for all local containers, similar in style to the already
853 supported --recursive switch for "systemctl list-units".
855 * A new RebootArgument= setting has been added for service
856 units, which may be used to specify a kernel reboot argument
857 to use when triggering reboots with StartLimitAction=.
859 * A new FailureAction= setting has been added for service
860 units which may be used to specify an operation to trigger
861 when a service fails. This works similarly to
862 StartLimitAction=, but unlike it, controls what is done
863 immediately rather than only after several attempts to
864 restart the service in question.
866 * hostnamed got updated to also expose the kernel name,
867 release, and version on the bus. This is useful for
868 executing commands like hostnamectl with the -H switch.
869 systemd-analyze makes use of this to properly display
870 details when running non-locally.
872 * The bootchart tool can now show cgroup information in the
875 * The CFS CPU quota cgroup attribute is now exposed for
876 services. The new CPUQuota= switch has been added for this
877 which takes a percentage value. Setting this will have the
878 result that a service may never get more CPU time than the
879 specified percentage, even if the machine is otherwise idle.
881 * systemd-networkd learned IPIP and SIT tunnel support.
883 * LSB init scripts exposing a dependency on $network will now
884 get a dependency on network-online.target rather than simply
885 network.target. This should bring LSB handling closer to
886 what it was on SysV systems.
888 * A new fsck.repair= kernel option has been added to control
889 how fsck shall deal with unclean file systems at boot.
891 * The (.ini) configuration file parser will now silently
892 ignore sections whose name begins with "X-". This may be
893 used to maintain application-specific extension sections in unit
896 * machined gained a new API to query the IP addresses of
897 registered containers. "machinectl status" has been updated
898 to show these addresses in its output.
900 * A new call sd_uid_get_display() has been added to the
901 sd-login APIs for querying the "primary" session of a
902 user. The "primary" session of the user is elected from the
903 user's sessions and generally a graphical session is
904 preferred over a text one.
906 * A minimal systemd-resolved daemon has been added. It
907 currently simply acts as a companion to systemd-networkd and
908 manages resolv.conf based on per-interface DNS
909 configuration, possibly supplied via DHCP. In the long run
910 we hope to extend this into a local DNSSEC enabled DNS and
913 * The systemd-networkd-wait-online tool is now enabled by
914 default. It will delay network-online.target until a network
915 connection has been configured. The tool primarily integrates
916 with networkd, but will also make a best effort to make sense
917 of network configuration performed in some other way.
919 * Two new service options StartupCPUShares= and
920 StartupBlockIOWeight= have been added that work similarly to
921 CPUShares= and BlockIOWeight= however only apply during
922 system startup. This is useful to prioritize certain services
923 differently during bootup than during normal runtime.
925 * hostnamed has been changed to prefer the statically
926 configured hostname in /etc/hostname (unless set to
927 'localhost' or empty) over any dynamic one supplied by
928 dhcp. With this change, the rules for picking the hostname
929 match more closely the rules of other configuration settings
930 where the local administrator's configuration in /etc always
931 overrides any other settings.
933 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
934 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
935 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
936 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
937 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
938 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
939 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
940 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
941 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
942 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
943 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
944 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
945 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
946 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
947 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
948 Lindskog, WaLyong Cho, Will Woods, Zbigniew
951 -- Beijing, 2014-05-28
955 * When restoring the screen brightness at boot, stay away from
956 the darkest setting or from the lowest 5% of the available
957 range, depending on which is the larger value of both. This
958 should effectively protect the user from rebooting into a
959 black screen, should the brightness have been set to minimum
962 * sd-login gained a new sd_machine_get_class() call to
963 determine the class ("vm" or "container") of a machine
964 registered with machined.
966 * sd-login gained new calls
967 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
968 to query the identity of the peer of a local AF_UNIX
969 connection. They operate similarly to their sd_pid_get_xyz()
972 * PID 1 will now maintain a system-wide system state engine
973 with the states "starting", "running", "degraded",
974 "maintenance", "stopping". These states are bound to system
975 startup, normal runtime, runtime with at least one failed
976 service, rescue/emergency mode and system shutdown. This
977 state is shown in the "systemctl status" output when no unit
978 name is passed. It is useful to determine system state, in
979 particularly when doing so for many systems or containers at
982 * A new command "list-machines" has been added to "systemctl"
983 that lists all local OS containers and shows their system
984 state (see above), if systemd runs inside of them.
986 * systemctl gained a new "-r" switch to recursively enumerate
987 units on all local containers, when used with the
988 "list-unit" command (which is the default one that is
989 executed when no parameters are specified).
991 * The GPT automatic partition discovery logic will now honour
992 two GPT partition flags: one may be set on a partition to
993 cause it to be mounted read-only, and the other may be set
994 on a partition to ignore it during automatic discovery.
996 * Two new GPT type UUIDs have been added for automatic root
997 partition discovery, for 32-bit and 64-bit ARM. This is not
998 particularly useful for discovering the root directory on
999 these architectures during bare-metal boots (since UEFI is
1000 not common there), but still very useful to allow booting of
1001 ARM disk images in nspawn with the -i option.
1003 * MAC addresses of interfaces created with nspawn's
1004 --network-interface= switch will now be generated from the
1005 machine name, and thus be stable between multiple invocations
1008 * logind will now automatically remove all IPC objects owned
1009 by a user if she or he fully logs out. This makes sure that
1010 users who are logged out cannot continue to consume IPC
1011 resources. This covers SysV memory, semaphores and message
1012 queues as well as POSIX shared memory and message
1013 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1014 limits. With this functionality, that is corrected. This may
1015 be turned off by using the RemoveIPC= switch of logind.conf.
1017 * The systemd-machine-id-setup and tmpfiles tools gained a
1018 --root= switch to operate on a specific root directory,
1021 * journald can now forward logged messages to the TTYs of all
1022 logged in users ("wall"). This is the default for all
1023 emergency messages now.
1025 * A new tool systemd-journal-remote has been added to stream
1026 journal log messages across the network.
1028 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1029 controller trees are mounted into it. Note that the
1030 directories mounted beneath it are not read-only. This is a
1031 security measure and is particularly useful because glibc
1032 actually includes a search logic to pick any tmpfs it can
1033 find to implement shm_open() if /dev/shm is not available
1034 (which it might very well be in namespaced setups).
1036 * machinectl gained a new "poweroff" command to cleanly power
1037 down a local OS container.
1039 * The PrivateDevices= unit file setting will now also drop the
1040 CAP_MKNOD capability from the capability bound set, and
1041 imply DevicePolicy=closed.
1043 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1044 comprehensively on all long-running systemd services where
1045 this is appropriate.
1047 * systemd-udevd will now run in a disassociated mount
1048 namespace. To mount directories from udev rules, make sure to
1049 pull in mount units via SYSTEMD_WANTS properties.
1051 * The kdbus support gained support for uploading policy into
1052 the kernel. sd-bus gained support for creating "monitoring"
1053 connections that can eavesdrop into all bus communication
1054 for debugging purposes.
1056 * Timestamps may now be specified in seconds since the UNIX
1057 epoch Jan 1st, 1970 by specifying "@" followed by the value
1060 * Native tcpwrap support in systemd has been removed. tcpwrap
1061 is old code, not really maintained anymore and has serious
1062 shortcomings, and better options such as firewalls
1063 exist. For setups that require tcpwrap usage, please
1064 consider invoking your socket-activated service via tcpd,
1065 like on traditional inetd.
1067 * A new system.conf configuration option
1068 DefaultTimerAccuracySec= has been added that controls the
1069 default AccuracySec= setting of .timer units.
1071 * Timer units gained a new WakeSystem= switch. If enabled,
1072 timers configured this way will cause the system to resume
1073 from system suspend (if the system supports that, which most
1076 * Timer units gained a new Persistent= switch. If enabled,
1077 timers configured this way will save to disk when they have
1078 been last triggered. This information is then used on next
1079 reboot to possible execute overdue timer events, that
1080 could not take place because the system was powered off.
1081 This enables simple anacron-like behaviour for timer units.
1083 * systemctl's "list-timers" will now also list the time a
1084 timer unit was last triggered in addition to the next time
1085 it will be triggered.
1087 * systemd-networkd will now assign predictable IPv4LL
1088 addresses to its local interfaces.
1090 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1091 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1092 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1093 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1094 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1095 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1096 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1097 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1100 -- Berlin, 2014-03-25
1104 * A new unit file setting RestrictAddressFamilies= has been
1105 added to restrict which socket address families unit
1106 processes gain access to. This takes address family names
1107 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1108 attack surface of services via exotic protocol stacks. This
1109 is built on seccomp system call filters.
1111 * Two new unit file settings RuntimeDirectory= and
1112 RuntimeDirectoryMode= have been added that may be used to
1113 manage a per-daemon runtime directories below /run. This is
1114 an alternative for setting up directory permissions with
1115 tmpfiles snippets, and has the advantage that the runtime
1116 directory's lifetime is bound to the daemon runtime and that
1117 the daemon starts up with an empty directory each time. This
1118 is particularly useful when writing services that drop
1119 privileges using the User= or Group= setting.
1121 * The DeviceAllow= unit setting now supports globbing for
1122 matching against device group names.
1124 * The systemd configuration file system.conf gained new
1125 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1126 DefaultMemoryAccounting= to globally turn on/off accounting
1127 for specific resources (cgroups) for all units. These
1128 settings may still be overridden individually in each unit
1131 * systemd-gpt-auto-generator is now able to discover /srv and
1132 root partitions in addition to /home and swap partitions. It
1133 also supports LUKS-encrypted partitions now. With this in
1134 place, automatic discovery of partitions to mount following
1135 the Discoverable Partitions Specification
1136 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1137 is now a lot more complete. This allows booting without
1138 /etc/fstab and without root= on the kernel command line on
1139 systems prepared appropriately.
1141 * systemd-nspawn gained a new --image= switch which allows
1142 booting up disk images and Linux installations on any block
1143 device that follow the Discoverable Partitions Specification
1144 (see above). This means that installations made with
1145 appropriately updated installers may now be started and
1146 deployed using container managers, completely
1147 unmodified. (We hope that libvirt-lxc will add support for
1148 this feature soon, too.)
1150 * systemd-nspawn gained a new --network-macvlan= setting to
1151 set up a private macvlan interface for the
1152 container. Similarly, systemd-networkd gained a new
1153 Kind=macvlan setting in .netdev files.
1155 * systemd-networkd now supports configuring local addresses
1158 * A new tool systemd-network-wait-online has been added to
1159 synchronously wait for network connectivity using
1162 * The sd-bus.h bus API gained a new sd_bus_track object for
1163 tracking the life-cycle of bus peers. Note that sd-bus.h is
1164 still not a public API though (unless you specify
1165 --enable-kdbus on the configure command line, which however
1166 voids your warranty and you get no API stability guarantee).
1168 * The $XDG_RUNTIME_DIR runtime directories for each user are
1169 now individual tmpfs instances, which has the benefit of
1170 introducing separate pools for each user, with individual
1171 size limits, and thus making sure that unprivileged clients
1172 can no longer negatively impact the system or other users by
1173 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1174 RuntimeDirectorySize= has been introduced that allows
1175 controlling the default size limit for all users. It
1176 defaults to 10% of the available physical memory. This is no
1177 replacement for quotas on tmpfs though (which the kernel
1178 still does not support), as /dev/shm and /tmp are still
1179 shared resources used by both the system and unprivileged
1182 * logind will now automatically turn off automatic suspending
1183 on laptop lid close when more than one display is
1184 connected. This was previously expected to be implemented
1185 individually in desktop environments (such as GNOME),
1186 however has been added to logind now, in order to fix a
1187 boot-time race where a desktop environment might not have
1188 been started yet and thus not been able to take an inhibitor
1189 lock at the time where logind already suspends the system
1190 due to a closed lid.
1192 * logind will now wait at least 30s after each system
1193 suspend/resume cycle, and 3min after system boot before
1194 suspending the system due to a closed laptop lid. This
1195 should give USB docking stations and similar enough time to
1196 be probed and configured after system resume and boot in
1197 order to then act as suspend blocker.
1199 * systemd-run gained a new --property= setting which allows
1200 initialization of resource control properties (and others)
1201 for the created scope or service unit. Example: "systemd-run
1202 --property=BlockIOWeight=10 updatedb" may be used to run
1203 updatedb at a low block IO scheduling weight.
1205 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1206 now also work in --scope mode.
1208 * When systemd is compiled with kdbus support, basic support
1209 for enforced policies is now in place. (Note that enabling
1210 kdbus still voids your warranty and no API compatibility
1213 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1214 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1215 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1216 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1217 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1218 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1219 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1220 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1221 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1222 Zbigniew Jędrzejewski-Szmek
1224 -- Berlin, 2014-03-12
1228 * systemd will now relabel /dev after loading the SMACK policy
1229 according to SMACK rules.
1231 * A new unit file option AppArmorProfile= has been added to
1232 set the AppArmor profile for the processes of a unit.
1234 * A new condition check ConditionArchitecture= has been added
1235 to conditionalize units based on the system architecture, as
1236 reported by uname()'s "machine" field.
1238 * systemd-networkd now supports matching on the system
1239 virtualization, architecture, kernel command line, host name
1242 * logind is now a lot more aggressive when suspending the
1243 machine due to a closed laptop lid. Instead of acting only
1244 on the lid close action, it will continuously watch the lid
1245 status and act on it. This is useful for laptops where the
1246 power button is on the outside of the chassis so that it can
1247 be reached without opening the lid (such as the Lenovo
1248 Yoga). On those machines, logind will now immediately
1249 re-suspend the machine if the power button has been
1250 accidentally pressed while the laptop was suspended and in a
1251 backpack or similar.
1253 * logind will now watch SW_DOCK switches and inhibit reaction
1254 to the lid switch if it is pressed. This means that logind
1255 will not suspend the machine anymore if the lid is closed
1256 and the system is docked, if the laptop supports SW_DOCK
1257 notifications via the input layer. Note that ACPI docking
1258 stations do not generate this currently. Also note that this
1259 logic is usually not fully sufficient and Desktop
1260 Environments should take a lid switch inhibitor lock when an
1261 external display is connected, as systemd will not watch
1264 * nspawn will now make use of the devices cgroup controller by
1265 default, and only permit creation of and access to the usual
1266 API device nodes like /dev/null or /dev/random, as well as
1267 access to (but not creation of) the pty devices.
1269 * We will now ship a default .network file for
1270 systemd-networkd that automatically configures DHCP for
1271 network interfaces created by nspawn's --network-veth or
1272 --network-bridge= switches.
1274 * systemd will now understand the usual M, K, G, T suffixes
1275 according to SI conventions (i.e. to the base 1000) when
1276 referring to throughput and hardware metrics. It will stay
1277 with IEC conventions (i.e. to the base 1024) for software
1278 metrics, according to what is customary according to
1279 Wikipedia. We explicitly document which base applies for
1280 each configuration option.
1282 * The DeviceAllow= setting in unit files now supports a syntax
1283 to whitelist an entire group of devices node majors at once,
1284 based on the /proc/devices listing. For example, with the
1285 string "char-pts", it is now possible to whitelist all
1286 current and future pseudo-TTYs at once.
1288 * sd-event learned a new "post" event source. Event sources of
1289 this type are triggered by the dispatching of any event
1290 source of a type that is not "post". This is useful for
1291 implementing clean-up and check event sources that are
1292 triggered by other work being done in the program.
1294 * systemd-networkd is no longer statically enabled, but uses
1295 the usual [Install] sections so that it can be
1296 enabled/disabled using systemctl. It still is enabled by
1299 * When creating a veth interface pair with systemd-nspawn, the
1300 host side will now be prefixed with "vb-" if
1301 --network-bridge= is used, and with "ve-" if --network-veth
1302 is used. This way, it is easy to distinguish these cases on
1303 the host, for example to apply different configuration to
1304 them with systemd-networkd.
1306 * The compatibility libraries for libsystemd-journal.so,
1307 libsystem-id128.so, libsystemd-login.so and
1308 libsystemd-daemon.so do not make use of IFUNC
1309 anymore. Instead, we now build libsystemd.so multiple times
1310 under these alternative names. This means that the footprint
1311 is drastically increased, but given that these are
1312 transitional compatibility libraries, this should not matter
1313 much. This change has been made necessary to support the ARM
1314 platform for these compatibility libraries, as the ARM
1315 toolchain is not really at the same level as the toolchain
1316 for other architectures like x86 and does not support
1317 IFUNC. Please make sure to use --enable-compat-libs only
1318 during a transitional period!
1320 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
1321 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1322 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
1323 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
1324 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
1325 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
1326 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
1327 Zbigniew Jędrzejewski-Szmek
1329 -- Berlin, 2014-02-24
1333 * A new component "systemd-networkd" has been added that can
1334 be used to configure local network interfaces statically or
1335 via DHCP. It is capable of bringing up bridges, VLANs, and
1336 bonding. Currently, no hook-ups for interactive network
1337 configuration are provided. Use this for your initrd,
1338 container, embedded, or server setup if you need a simple,
1339 yet powerful, network configuration solution. This
1340 configuration subsystem is quite nifty, as it allows wildcard
1341 hotplug matching in interfaces. For example, with a single
1342 configuration snippet, you can configure that all Ethernet
1343 interfaces showing up are automatically added to a bridge,
1344 or similar. It supports link-sensing and more.
1346 * A new tool "systemd-socket-proxyd" has been added which can
1347 act as a bidirectional proxy for TCP sockets. This is
1348 useful for adding socket activation support to services that
1349 do not actually support socket activation, including virtual
1350 machines and the like.
1352 * Add a new tool to save/restore rfkill state on
1355 * Save/restore state of keyboard backlights in addition to
1356 display backlights on shutdown/boot.
1358 * udev learned a new SECLABEL{} construct to label device
1359 nodes with a specific security label when they appear. For
1360 now, only SECLABEL{selinux} is supported, but the syntax is
1361 prepared for additional security frameworks.
1363 * udev gained a new scheme to configure link-level attributes
1364 from files in /etc/systemd/network/*.link. These files can
1365 match against MAC address, device path, driver name and type,
1366 and will apply attributes like the naming policy, link speed,
1367 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
1368 address assignment policy (randomized, ...).
1370 * The configuration of network interface naming rules for
1371 "permanent interface names" has changed: a new NamePolicy=
1372 setting in the [Link] section of .link files determines the
1373 priority of possible naming schemes (onboard, slot, mac,
1374 path). The default value of this setting is determined by
1375 /usr/lib/net/links/99-default.link. Old
1376 80-net-name-slot.rules udev configuration file has been
1377 removed, so local configuration overriding this file should
1378 be adapated to override 99-default.link instead.
1380 * When the User= switch is used in a unit file, also
1381 initialize $SHELL= based on the user database entry.
1383 * systemd no longer depends on libdbus. All communication is
1384 now done with sd-bus, systemd's low-level bus library
1387 * kdbus support has been added to PID 1 itself. When kdbus is
1388 enabled, this causes PID 1 to set up the system bus and
1389 enable support for a new ".busname" unit type that
1390 encapsulates bus name activation on kdbus. It works a little
1391 bit like ".socket" units, except for bus names. A new
1392 generator has been added that converts classic dbus1 service
1393 activation files automatically into native systemd .busname
1396 * sd-bus: add a light-weight vtable implementation that allows
1397 defining objects on the bus with a simple static const
1398 vtable array of its methods, signals and properties.
1400 * systemd will not generate or install static dbus
1401 introspection data anymore to /usr/share/dbus-1/interfaces,
1402 as the precise format of these files is unclear, and
1403 nothing makes use of it.
1405 * A proxy daemon is now provided to proxy clients connecting
1406 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1407 compatibility with classic D-Bus.
1409 * A bus driver implementation has been added that supports the
1410 classic D-Bus bus driver calls on kdbus, also for
1411 compatibility purposes.
1413 * A new API "sd-event.h" has been added that implements a
1414 minimal event loop API built around epoll. It provides a
1415 couple of features that direct epoll usage is lacking:
1416 prioritization of events, scales to large numbers of timer
1417 events, per-event timer slack (accuracy), system-wide
1418 coalescing of timer events, exit handlers, watchdog
1419 supervision support using systemd's sd_notify() API, child
1422 * A new API "sd-rntl.h" has been added that provides an API
1423 around the route netlink interface of the kernel, similar in
1424 style to "sd-bus.h".
1426 * A new API "sd-dhcp-client.h" has been added that provides a
1427 small DHCPv4 client-side implementation. This is used by
1430 * There is a new kernel command line option
1431 "systemd.restore_state=0|1". When set to "0", none of the
1432 systemd tools will restore saved runtime state to hardware
1433 devices. More specifically, the rfkill and backlight states
1436 * The FsckPassNo= compatibility option in mount/service units
1437 has been removed. The fstab generator will now add the
1438 necessary dependencies automatically, and does not require
1439 PID1's support for that anymore.
1441 * journalctl gained a new switch, --list-boots, that lists
1442 recent boots with their times and boot IDs.
1444 * The various tools like systemctl, loginctl, timedatectl,
1445 busctl, systemd-run, ... have gained a new switch "-M" to
1446 connect to a specific, local OS container (as direct
1447 connection, without requiring SSH). This works on any
1448 container that is registered with machined, such as those
1449 created by libvirt-lxc or nspawn.
1451 * systemd-run and systemd-analyze also gained support for "-H"
1452 to connect to remote hosts via SSH. This is particularly
1453 useful for systemd-run because it enables queuing of jobs
1454 onto remote systems.
1456 * machinectl gained a new command "login" to open a getty
1457 login in any local container. This works with any container
1458 that is registered with machined (such as those created by
1459 libvirt-lxc or nspawn), and which runs systemd inside.
1461 * machinectl gained a new "reboot" command that may be used to
1462 trigger a reboot on a specific container that is registered
1463 with machined. This works on any container that runs an init
1464 system of some kind.
1466 * systemctl gained a new "list-timers" command to print a nice
1467 listing of installed timer units with the times they elapse
1470 * Alternative reboot() parameters may now be specified on the
1471 "systemctl reboot" command line and are passed to the
1472 reboot() system call.
1474 * systemctl gained a new --job-mode= switch to configure the
1475 mode to queue a job with. This is a more generic version of
1476 --fail, --irreversible, and --ignore-dependencies, which are
1477 still available but not advertised anymore.
1479 * /etc/systemd/system.conf gained new settings to configure
1480 various default timeouts of units, as well as the default
1481 start limit interval and burst. These may still be overridden
1484 * PID1 will now export on the bus profile data of the security
1485 policy upload process (such as the SELinux policy upload to
1488 * journald: when forwarding logs to the console, include
1489 timestamps (following the setting in
1490 /sys/module/printk/parameters/time).
1492 * OnCalendar= in timer units now understands the special
1493 strings "yearly" and "annually". (Both are equivalent)
1495 * The accuracy of timer units is now configurable with the new
1496 AccuracySec= setting. It defaults to 1min.
1498 * A new dependency type JoinsNamespaceOf= has been added that
1499 allows running two services within the same /tmp and network
1500 namespace, if PrivateNetwork= or PrivateTmp= are used.
1502 * A new command "cat" has been added to systemctl. It outputs
1503 the original unit file of a unit, and concatenates the
1504 contents of additional "drop-in" unit file snippets, so that
1505 the full configuration is shown.
1507 * systemctl now supports globbing on the various "list-xyz"
1508 commands, like "list-units" or "list-sockets", as well as on
1509 those commands which take multiple unit names.
1511 * journalctl's --unit= switch gained support for globbing.
1513 * All systemd daemons now make use of the watchdog logic so
1514 that systemd automatically notices when they hang.
1516 * If the $container_ttys environment variable is set,
1517 getty-generator will automatically spawn a getty for each
1518 listed tty. This is useful for container managers to request
1519 login gettys to be spawned on as many ttys as needed.
1521 * %h, %s, %U specifier support is not available anymore when
1522 used in unit files for PID 1. This is because NSS calls are
1523 not safe from PID 1. They stay available for --user
1524 instances of systemd, and as special case for the root user.
1526 * loginctl gained a new "--no-legend" switch to turn off output
1529 * The "sd-login.h" API gained three new calls:
1530 sd_session_is_remote(), sd_session_get_remote_user(),
1531 sd_session_get_remote_host() to query information about
1534 * The udev hardware database now also carries vendor/product
1535 information of SDIO devices.
1537 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1538 determine whether watchdog notifications are requested by
1541 * Socket-activated per-connection services now include a
1542 short description of the connection parameters in the
1545 * tmpfiles gained a new "--boot" option. When this is not used,
1546 only lines where the command character is not suffixed with
1547 "!" are executed. When this option is specified, those
1548 options are executed too. This partitions tmpfiles
1549 directives into those that can be safely executed at any
1550 time, and those which should be run only at boot (for
1551 example, a line that creates /run/nologin).
1553 * A new API "sd-resolve.h" has been added which provides a simple
1554 asynchronous wrapper around glibc NSS host name resolution
1555 calls, such as getaddrinfo(). In contrast to glibc's
1556 getaddrinfo_a(), it does not use signals. In contrast to most
1557 other asynchronous name resolution libraries, this one does
1558 not reimplement DNS, but reuses NSS, so that alternate
1559 host name resolution systems continue to work, such as mDNS,
1560 LDAP, etc. This API is based on libasyncns, but it has been
1561 cleaned up for inclusion in systemd.
1563 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1564 "sd-daemon.h" are no longer found in individual libraries
1565 libsystemd-journal.so, libsystemd-login.so,
1566 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1567 merged them into a single library, libsystemd.so, which
1568 provides all symbols. The reason for this is cyclic
1569 dependencies, as these libraries tend to use each other's
1570 symbols. So far, we have managed to workaround that by linking
1571 a copy of a good part of our code into each of these
1572 libraries again and again, which, however, makes certain
1573 things hard to do, like sharing static variables. Also, it
1574 substantially increases footprint. With this change, there
1575 is only one library for the basic APIs systemd
1576 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1577 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1578 library as well, however are subject to the --enable-kdbus
1579 switch (see below). Note that "sd-dhcp-client.h" is not part
1580 of this library (this is because it only consumes, never
1581 provides, services of/to other APIs). To make the transition
1582 easy from the separate libraries to the unified one, we
1583 provide the --enable-compat-libs compile-time switch which
1584 will generate stub libraries that are compatible with the
1585 old ones but redirect all calls to the new one.
1587 * All of the kdbus logic and the new APIs "sd-bus.h",
1588 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1589 and "sd-utf8.h" are compile-time optional via the
1590 "--enable-kdbus" switch, and they are not compiled in by
1591 default. To make use of kdbus, you have to explicitly enable
1592 the switch. Note however, that neither the kernel nor the
1593 userspace API for all of this is considered stable yet. We
1594 want to maintain the freedom to still change the APIs for
1595 now. By specifying this build-time switch, you acknowledge
1596 that you are aware of the instability of the current
1599 * Also, note that while kdbus is pretty much complete,
1600 it lacks one thing: proper policy support. This means you
1601 can build a fully working system with all features; however,
1602 it will be highly insecure. Policy support will be added in
1603 one of the next releases, at the same time that we will
1604 declare the APIs stable.
1606 * When the kernel command-line argument "kdbus" is specified,
1607 systemd will automatically load the kdbus.ko kernel module. At
1608 this stage of development, it is only useful for testing kdbus
1609 and should not be used in production. Note: if "--enable-kdbus"
1610 is specified, and the kdbus.ko kernel module is available, and
1611 "kdbus" is added to the kernel command line, the entire system
1612 runs with kdbus instead of dbus-daemon, with the above mentioned
1613 problem of missing the system policy enforcement. Also a future
1614 version of kdbus.ko or a newer systemd will not be compatible with
1615 each other, and will unlikely be able to boot the machine if only
1616 one of them is updated.
1618 * systemctl gained a new "import-environment" command which
1619 uploads the caller's environment (or parts thereof) into the
1620 service manager so that it is inherited by services started
1621 by the manager. This is useful to upload variables like
1622 $DISPLAY into the user service manager.
1624 * A new PrivateDevices= switch has been added to service units
1625 which allows running a service with a namespaced /dev
1626 directory that does not contain any device nodes for
1627 physical devices. More specifically, it only includes devices
1628 such as /dev/null, /dev/urandom, and /dev/zero which are API
1631 * logind has been extended to support behaviour like VT
1632 switching on seats that do not support a VT. This makes
1633 multi-session available on seats that are not the first seat
1634 (seat0), and on systems where kernel support for VTs has
1635 been disabled at compile-time.
1637 * If a process holds a delay lock for system sleep or shutdown
1638 and fails to release it in time, we will now log its
1639 identity. This makes it easier to identify processes that
1640 cause slow suspends or power-offs.
1642 * When parsing /etc/crypttab, support for a new key-slot=
1643 option as supported by Debian is added. It allows indicating
1644 which LUKS slot to use on disk, speeding up key loading.
1646 * The sd_journald_sendv() API call has been checked and
1647 officially declared to be async-signal-safe so that it may
1648 be invoked from signal handlers for logging purposes.
1650 * Boot-time status output is now enabled automatically after a
1651 short timeout if boot does not progress, in order to give
1652 the user an indication what she or he is waiting for.
1654 * The boot-time output has been improved to show how much time
1655 remains until jobs expire.
1657 * The KillMode= switch in service units gained a new possible
1658 value "mixed". If set, and the unit is shut down, then the
1659 initial SIGTERM signal is sent only to the main daemon
1660 process, while the following SIGKILL signal is sent to
1661 all remaining processes of the service.
1663 * When a scope unit is registered, a new property "Controller"
1664 may be set. If set to a valid bus name, systemd will send a
1665 RequestStop() signal to this name when it would like to shut
1666 down the scope. This may be used to hook manager logic into
1667 the shutdown logic of scope units. Also, scope units may now
1668 be put in a special "abandoned" state, in which case the
1669 manager process which created them takes no further
1670 responsibilities for it.
1672 * When reading unit files, systemd will now verify
1673 the access mode of these files, and warn about certain
1674 suspicious combinations. This has been added to make it
1675 easier to track down packaging bugs where unit files are
1676 marked executable or world-writable.
1678 * systemd-nspawn gained a new "--setenv=" switch to set
1679 container-wide environment variables. The similar option in
1680 systemd-activate was renamed from "--environment=" to
1681 "--setenv=" for consistency.
1683 * systemd-nspawn has been updated to create a new kdbus domain
1684 for each container that is invoked, thus allowing each
1685 container to have its own set of system and user buses,
1686 independent of the host.
1688 * systemd-nspawn gained a new --drop-capability= switch to run
1689 the container with less capabilities than the default. Both
1690 --drop-capability= and --capability= now take the special
1691 string "all" for dropping or keeping all capabilities.
1693 * systemd-nspawn gained new switches for executing containers
1694 with specific SELinux labels set.
1696 * systemd-nspawn gained a new --quiet switch to not generate
1697 any additional output but the container's own console
1700 * systemd-nspawn gained a new --share-system switch to run a
1701 container without PID namespacing enabled.
1703 * systemd-nspawn gained a new --register= switch to control
1704 whether the container is registered with systemd-machined or
1705 not. This is useful for containers that do not run full
1706 OS images, but only specific apps.
1708 * systemd-nspawn gained a new --keep-unit which may be used
1709 when invoked as the only program from a service unit, and
1710 results in registration of the unit service itself in
1711 systemd-machined, instead of a newly opened scope unit.
1713 * systemd-nspawn gained a new --network-interface= switch for
1714 moving arbitrary interfaces to the container. The new
1715 --network-veth switch creates a virtual Ethernet connection
1716 between host and container. The new --network-bridge=
1717 switch then allows assigning the host side of this virtual
1718 Ethernet connection to a bridge device.
1720 * systemd-nspawn gained a new --personality= switch for
1721 setting the kernel personality for the container. This is
1722 useful when running a 32-bit container on a 64-bit host. A
1723 similar option Personality= is now also available for service
1726 * logind will now also track a "Desktop" identifier for each
1727 session which encodes the desktop environment of it. This is
1728 useful for desktop environments that want to identify
1729 multiple running sessions of itself easily.
1731 * A new SELinuxContext= setting for service units has been
1732 added that allows setting a specific SELinux execution
1733 context for a service.
1735 * Most systemd client tools will now honour $SYSTEMD_LESS for
1736 settings of the "less" pager. By default, these tools will
1737 override $LESS to allow certain operations to work, such as
1738 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1739 influence this logic.
1741 * systemd's "seccomp" hook-up has been changed to make use of
1742 the libseccomp library instead of using its own
1743 implementation. This has benefits for portability among
1746 * For usage together with SystemCallFilter=, a new
1747 SystemCallErrorNumber= setting has been introduced that
1748 allows configuration of a system error number to be returned
1749 on filtered system calls, instead of immediately killing the
1750 process. Also, SystemCallArchitectures= has been added to
1751 limit access to system calls of a particular architecture
1752 (in order to turn off support for unused secondary
1753 architectures). There is also a global
1754 SystemCallArchitectures= setting in system.conf now to turn
1755 off support for non-native system calls system-wide.
1757 * systemd requires a kernel with a working name_to_handle_at(),
1758 please see the kernel config requirements in the README file.
1760 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1761 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1762 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1763 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1764 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1765 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1766 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1767 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1768 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1769 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1770 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1771 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1772 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1773 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1774 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1775 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1776 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1777 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1778 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1779 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1780 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1781 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1782 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1783 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1785 -- Berlin, 2014-02-20
1789 * logind has gained support for facilitating privileged input
1790 and drm device access for unprivileged clients. This work is
1791 useful to allow Wayland display servers (and similar
1792 programs, such as kmscon) to run under the user's ID and
1793 access input and drm devices which are normally
1794 protected. When this is used (and the kernel is new enough)
1795 logind will "mute" IO on the file descriptors passed to
1796 Wayland as long as it is in the background and "unmute" it
1797 if it returns into the foreground. This allows secure
1798 session switching without allowing background sessions to
1799 eavesdrop on input and display data. This also introduces
1800 session switching support if VT support is turned off in the
1801 kernel, and on seats that are not seat0.
1803 * A new kernel command line option luks.options= is understood
1804 now which allows specifying LUKS options for usage for LUKS
1805 encrypted partitions specified with luks.uuid=.
1807 * tmpfiles.d(5) snippets may now use specifier expansion in
1808 path names. More specifically %m, %b, %H, %v, are now
1809 replaced by the local machine id, boot id, hostname, and
1810 kernel version number.
1812 * A new tmpfiles.d(5) command "m" has been introduced which
1813 may be used to change the owner/group/access mode of a file
1814 or directory if it exists, but do nothing if it does not.
1816 * This release removes high-level support for the
1817 MemorySoftLimit= cgroup setting. The underlying kernel
1818 cgroup attribute memory.soft_limit= is currently badly
1819 designed and likely to be removed from the kernel API in its
1820 current form, hence we should not expose it for now.
1822 * The memory.use_hierarchy cgroup attribute is now enabled for
1823 all cgroups systemd creates in the memory cgroup
1824 hierarchy. This option is likely to be come the built-in
1825 default in the kernel anyway, and the non-hierarchial mode
1826 never made much sense in the intrinsically hierarchial
1829 * A new field _SYSTEMD_SLICE= is logged along with all journal
1830 messages containing the slice a message was generated
1831 from. This is useful to allow easy per-customer filtering of
1832 logs among other things.
1834 * systemd-journald will no longer adjust the group of journal
1835 files it creates to the "systemd-journal" group. Instead we
1836 rely on the journal directory to be owned by the
1837 "systemd-journal" group, and its setgid bit set, so that the
1838 kernel file system layer will automatically enforce that
1839 journal files inherit this group assignment. The reason for
1840 this change is that we cannot allow NSS look-ups from
1841 journald which would be necessary to resolve
1842 "systemd-journal" to a numeric GID, because this might
1843 create deadlocks if NSS involves synchronous queries to
1844 other daemons (such as nscd, or sssd) which in turn are
1845 logging clients of journald and might block on it, which
1846 would then dead lock. A tmpfiles.d(5) snippet included in
1847 systemd will make sure the setgid bit and group are
1848 properly set on the journal directory if it exists on every
1849 boot. However, we recommend adjusting it manually after
1850 upgrades too (or from RPM scriptlets), so that the change is
1851 not delayed until next reboot.
1853 * Backlight and random seed files in /var/lib/ have moved into
1854 the /var/lib/systemd/ directory, in order to centralize all
1855 systemd generated files in one directory.
1857 * Boot time performance measurements (as displayed by
1858 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1859 performance information if that's available to determine how
1860 much time BIOS and boot loader initialization required. With
1861 a sufficiently new BIOS you hence no longer need to boot
1862 with Gummiboot to get access to such information.
1864 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1865 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1866 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1867 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1868 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1869 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1870 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1872 -- Berlin, 2013-10-02
1876 * The Restart= option for services now understands a new
1877 on-watchdog setting, which will restart the service
1878 automatically if the service stops sending out watchdog keep
1879 alive messages (as configured with WatchdogSec=).
1881 * The getty generator (which is responsible for bringing up a
1882 getty on configured serial consoles) will no longer only
1883 start a getty on the primary kernel console but on all
1884 others, too. This makes the order in which console= is
1885 specified on the kernel command line less important.
1887 * libsystemd-logind gained a new sd_session_get_vt() call to
1888 retrieve the VT number of a session.
1890 * If the option "tries=0" is set for an entry of /etc/crypttab
1891 its passphrase is queried indefinitely instead of any
1892 maximum number of tries.
1894 * If a service with a configure PID file terminates its PID
1895 file will now be removed automatically if it still exists
1896 afterwards. This should put an end to stale PID files.
1898 * systemd-run will now also take relative binary path names
1899 for execution and no longer insists on absolute paths.
1901 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1902 paths that are optionally prefixed with "-" to indicate that
1903 it should not be considered a failure if they do not exist.
1905 * journalctl -o (and similar commands) now understands a new
1906 output mode "short-precise", it is similar to "short" but
1907 shows timestamps with usec accuracy.
1909 * The option "discard" (as known from Debian) is now
1910 synonymous to "allow-discards" in /etc/crypttab. In fact,
1911 "discard" is preferred now (since it is easier to remember
1914 * Some licensing clean-ups were made, so that more code is now
1915 LGPL-2.1 licensed than before.
1917 * A minimal tool to save/restore the display backlight
1918 brightness across reboots has been added. It will store the
1919 backlight setting as late as possible at shutdown, and
1920 restore it as early as possible during reboot.
1922 * A logic to automatically discover and enable home and swap
1923 partitions on GPT disks has been added. With this in place
1924 /etc/fstab becomes optional for many setups as systemd can
1925 discover certain partitions located on the root disk
1926 automatically. Home partitions are recognized under their
1927 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1928 partitions are recognized under their GPT type ID
1929 0657fd6da4ab43c484e50933c84b4f4f.
1931 * systemd will no longer pass any environment from the kernel
1932 or initrd to system services. If you want to set an
1933 environment for all services, do so via the kernel command
1934 line systemd.setenv= assignment.
1936 * The systemd-sysctl tool no longer natively reads the file
1937 /etc/sysctl.conf. If desired, the file should be symlinked
1938 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1939 legacy support by a symlink rather than built-in code, it
1940 also makes the otherwise hidden order of application of the
1941 different files visible. (Note that this partly reverts to a
1942 pre-198 application order of sysctl knobs!)
1944 * The "systemctl set-log-level" and "systemctl dump" commands
1945 have been moved to systemd-analyze.
1947 * systemd-run learned the new --remain-after-exit switch,
1948 which causes the scope unit not to be cleaned up
1949 automatically after the process terminated.
1951 * tmpfiles learned a new --exclude-prefix= switch to exclude
1952 certain paths from operation.
1954 * journald will now automatically flush all messages to disk
1955 as soon as a message of the log priorities CRIT, ALERT or
1958 Contributions from: Andrew Cook, Brandon Philips, Christian
1959 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1960 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1961 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1962 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1963 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1964 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1965 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1966 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1967 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1968 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1969 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1970 William Giokas, Zbigniew Jędrzejewski-Szmek
1972 -- Berlin, 2013-09-13
1976 * The documentation has been updated to cover the various new
1977 concepts introduced with 205.
1979 * Unit files now understand the new %v specifier which
1980 resolves to the kernel version string as returned by "uname
1983 * systemctl now supports filtering the unit list output by
1984 load state, active state and sub state, using the new
1987 * "systemctl status" will now show the results of the
1988 condition checks (like ConditionPathExists= and similar) of
1989 the last start attempts of the unit. They are also logged to
1992 * "journalctl -b" may now be used to look for boot output of a
1993 specific boot. Try "journalctl -b -1" for the previous boot,
1994 but the syntax is substantially more powerful.
1996 * "journalctl --show-cursor" has been added which prints the
1997 cursor string the last shown log line. This may then be used
1998 with the new "journalctl --after-cursor=" switch to continue
1999 browsing logs from that point on.
2001 * "journalctl --force" may now be used to force regeneration
2004 * Creation of "dead" device nodes has been moved from udev
2005 into kmod and tmpfiles. Previously, udev would read the kmod
2006 databases to pre-generate dead device nodes based on meta
2007 information contained in kernel modules, so that these would
2008 be auto-loaded on access rather then at boot. As this
2009 does not really have much to do with the exposing actual
2010 kernel devices to userspace this has always been slightly
2011 alien in the udev codebase. Following the new scheme kmod
2012 will now generate a runtime snippet for tmpfiles from the
2013 module meta information and it now is tmpfiles' job to the
2014 create the nodes. This also allows overriding access and
2015 other parameters for the nodes using the usual tmpfiles
2016 facilities. As side effect this allows us to remove the
2017 CAP_SYS_MKNOD capability bit from udevd entirely.
2019 * logind's device ACLs may now be applied to these "dead"
2020 devices nodes too, thus finally allowing managed access to
2021 devices such as /dev/snd/sequencer whithout loading the
2022 backing module right-away.
2024 * A new RPM macro has been added that may be used to apply
2025 tmpfiles configuration during package installation.
2027 * systemd-detect-virt and ConditionVirtualization= now can
2028 detect User-Mode-Linux machines (UML).
2030 * journald will now implicitly log the effective capabilities
2031 set of processes in the message metadata.
2033 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2035 * The initrd interface has been simplified (more specifically,
2036 support for passing performance data via environment
2037 variables and fsck results via files in /run has been
2038 removed). These features were non-essential, and are
2039 nowadays available in a much nicer way by having systemd in
2040 the initrd serialize its state and have the hosts systemd
2041 deserialize it again.
2043 * The udev "keymap" data files and tools to apply keyboard
2044 specific mappings of scan to key codes, and force-release
2045 scan code lists have been entirely replaced by a udev
2046 "keyboard" builtin and a hwdb data file.
2048 * systemd will now honour the kernel's "quiet" command line
2049 argument also during late shutdown, resulting in a
2050 completely silent shutdown when used.
2052 * There's now an option to control the SO_REUSEPORT socket
2053 option in .socket units.
2055 * Instance units will now automatically get a per-template
2056 subslice of system.slice unless something else is explicitly
2057 configured. For example, instances of sshd@.service will now
2058 implicitly be placed in system-sshd.slice rather than
2059 system.slice as before.
2061 * Test coverage support may now be enabled at build time.
2063 Contributions from: Dave Reisner, Frederic Crozat, Harald
2064 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2065 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2066 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2067 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2068 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2069 Giokas, Zbigniew Jędrzejewski-Szmek
2071 -- Berlin, 2013-07-23
2075 * Two new unit types have been introduced:
2077 Scope units are very similar to service units, however, are
2078 created out of pre-existing processes -- instead of PID 1
2079 forking off the processes. By using scope units it is
2080 possible for system services and applications to group their
2081 own child processes (worker processes) in a powerful way
2082 which then maybe used to organize them, or kill them
2083 together, or apply resource limits on them.
2085 Slice units may be used to partition system resources in an
2086 hierarchial fashion and then assign other units to them. By
2087 default there are now three slices: system.slice (for all
2088 system services), user.slice (for all user sessions),
2089 machine.slice (for VMs and containers).
2091 Slices and scopes have been introduced primarily in
2092 context of the work to move cgroup handling to a
2093 single-writer scheme, where only PID 1
2094 creates/removes/manages cgroups.
2096 * There's a new concept of "transient" units. In contrast to
2097 normal units these units are created via an API at runtime,
2098 not from configuration from disk. More specifically this
2099 means it is now possible to run arbitrary programs as
2100 independent services, with all execution parameters passed
2101 in via bus APIs rather than read from disk. Transient units
2102 make systemd substantially more dynamic then it ever was,
2103 and useful as a general batch manager.
2105 * logind has been updated to make use of scope and slice units
2106 for managing user sessions. As a user logs in he will get
2107 his own private slice unit, to which all sessions are added
2108 as scope units. We also added support for automatically
2109 adding an instance of user@.service for the user into the
2110 slice. Effectively logind will no longer create cgroup
2111 hierarchies on its own now, it will defer entirely to PID 1
2112 for this by means of scope, service and slice units. Since
2113 user sessions this way become entities managed by PID 1
2114 the output of "systemctl" is now a lot more comprehensive.
2116 * A new mini-daemon "systemd-machined" has been added which
2117 may be used by virtualization managers to register local
2118 VMs/containers. nspawn has been updated accordingly, and
2119 libvirt will be updated shortly. machined will collect a bit
2120 of meta information about the VMs/containers, and assign
2121 them their own scope unit (see above). The collected
2122 meta-data is then made available via the "machinectl" tool,
2123 and exposed in "ps" and similar tools. machined/machinectl
2124 is compile-time optional.
2126 * As discussed earlier, the low-level cgroup configuration
2127 options ControlGroup=, ControlGroupModify=,
2128 ControlGroupPersistent=, ControlGroupAttribute= have been
2129 removed. Please use high-level attribute settings instead as
2130 well as slice units.
2132 * A new bus call SetUnitProperties() has been added to alter
2133 various runtime parameters of a unit. This is primarily
2134 useful to alter cgroup parameters dynamically in a nice way,
2135 but will be extended later on to make more properties
2136 modifiable at runtime. systemctl gained a new set-properties
2137 command that wraps this call.
2139 * A new tool "systemd-run" has been added which can be used to
2140 run arbitrary command lines as transient services or scopes,
2141 while configuring a number of settings via the command
2142 line. This tool is currently very basic, however already
2143 very useful. We plan to extend this tool to even allow
2144 queuing of execution jobs with time triggers from the
2145 command line, similar in fashion to "at".
2147 * nspawn will now inform the user explicitly that kernels with
2148 audit enabled break containers, and suggest the user to turn
2151 * Support for detecting the IMA and AppArmor security
2152 frameworks with ConditionSecurity= has been added.
2154 * journalctl gained a new "-k" switch for showing only kernel
2155 messages, mimicking dmesg output; in addition to "--user"
2156 and "--system" switches for showing only user's own logs
2159 * systemd-delta can now show information about drop-in
2160 snippets extending unit files.
2162 * libsystemd-bus has been substantially updated but is still
2163 not available as public API.
2165 * systemd will now look for the "debug" argument on the kernel
2166 command line and enable debug logging, similar to what
2167 "systemd.log_level=debug" already did before.
2169 * "systemctl set-default", "systemctl get-default" has been
2170 added to configure the default.target symlink, which
2171 controls what to boot into by default.
2173 * "systemctl set-log-level" has been added as a convenient
2174 way to raise and lower systemd logging threshold.
2176 * "systemd-analyze plot" will now show the time the various
2177 generators needed for execution, as well as information
2178 about the unit file loading.
2180 * libsystemd-journal gained a new sd_journal_open_files() call
2181 for opening specific journal files. journactl also gained a
2182 new switch to expose this new functionality. Previously we
2183 only supported opening all files from a directory, or all
2184 files from the system, as opening individual files only is
2185 racy due to journal file rotation.
2187 * systemd gained the new DefaultEnvironment= setting in
2188 /etc/systemd/system.conf to set environment variables for
2191 * If a privileged process logs a journal message with the
2192 OBJECT_PID= field set, then journald will automatically
2193 augment this with additional OBJECT_UID=, OBJECT_GID=,
2194 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2195 system services want to log events about specific client
2196 processes. journactl/systemctl has been updated to make use
2197 of this information if all log messages regarding a specific
2200 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2201 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2202 Reisner, David Coppa, David King, David Strauss, Eelco
2203 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2204 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2205 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2206 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2207 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2208 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2209 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2210 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2211 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2212 Łukasz Stelmach, 장동준
2216 * The Python bindings gained some minimal support for the APIs
2217 exposed by libsystemd-logind.
2219 * ConditionSecurity= gained support for detecting SMACK. Since
2220 this condition already supports SELinux and AppArmor we only
2221 miss IMA for this. Patches welcome!
2223 Contributions from: Karol Lewandowski, Lennart Poettering,
2224 Zbigniew Jędrzejewski-Szmek
2228 * systemd-nspawn will now create /etc/resolv.conf if
2229 necessary, before bind-mounting the host's file onto it.
2231 * systemd-nspawn will now store meta information about a
2232 container on the container's cgroup as extended attribute
2233 fields, including the root directory.
2235 * The cgroup hierarchy has been reworked in many ways. All
2236 objects any of the components systemd creates in the cgroup
2237 tree are now suffixed. More specifically, user sessions are
2238 now placed in cgroups suffixed with ".session", users in
2239 cgroups suffixed with ".user", and nspawn containers in
2240 cgroups suffixed with ".nspawn". Furthermore, all cgroup
2241 names are now escaped in a simple scheme to avoid collision
2242 of userspace object names with kernel filenames. This work
2243 is preparation for making these objects relocatable in the
2244 cgroup tree, in order to allow easy resource partitioning of
2245 these objects without causing naming conflicts.
2247 * systemctl list-dependencies gained the new switches
2248 --plain, --reverse, --after and --before.
2250 * systemd-inhibit now shows the process name of processes that
2251 have taken an inhibitor lock.
2253 * nss-myhostname will now also resolve "localhost"
2254 implicitly. This makes /etc/hosts an optional file and
2255 nicely handles that on IPv6 ::1 maps to both "localhost" and
2258 * libsystemd-logind.so gained a new call
2259 sd_get_machine_names() to enumerate running containers and
2260 VMs (currently only supported by very new libvirt and
2261 nspawn). sd_login_monitor can now be used to watch
2262 VMs/containers coming and going.
2264 * .include is not allowed recursively anymore, and only in
2265 unit files. Usually it is better to use drop-in snippets in
2266 .d/*.conf anyway, as introduced with systemd 198.
2268 * systemd-analyze gained a new "critical-chain" command that
2269 determines the slowest chain of units run during system
2270 boot-up. It is very useful for tracking down where
2271 optimizing boot time is the most beneficial.
2273 * systemd will no longer allow manipulating service paths in
2274 the name=systemd:/system cgroup tree using ControlGroup= in
2275 units. (But is still fine with it in all other dirs.)
2277 * There's a new systemd-nspawn@.service service file that may
2278 be used to easily run nspawn containers as system
2279 services. With the container's root directory in
2280 /var/lib/container/foobar it is now sufficient to run
2281 "systemctl start systemd-nspawn@foobar.service" to boot it.
2283 * systemd-cgls gained a new parameter "--machine" to list only
2284 the processes within a certain container.
2286 * ConditionSecurity= now can check for "apparmor". We still
2287 are lacking checks for SMACK and IMA for this condition
2288 check though. Patches welcome!
2290 * A new configuration file /etc/systemd/sleep.conf has been
2291 added that may be used to configure which kernel operation
2292 systemd is supposed to execute when "suspend", "hibernate"
2293 or "hybrid-sleep" is requested. This makes the new kernel
2294 "freeze" state accessible to the user.
2296 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
2297 the passed argument if applicable.
2299 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
2300 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
2301 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
2302 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
2303 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
2304 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
2305 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
2310 * The output of 'systemctl list-jobs' got some polishing. The
2311 '--type=' argument may now be passed more than once. A new
2312 command 'systemctl list-sockets' has been added which shows
2313 a list of kernel sockets systemd is listening on with the
2314 socket units they belong to, plus the units these socket
2317 * The experimental libsystemd-bus library got substantial
2318 updates to work in conjunction with the (also experimental)
2319 kdbus kernel project. It works well enough to exchange
2320 messages with some sophistication. Note that kdbus is not
2321 ready yet, and the library is mostly an elaborate test case
2322 for now, and not installable.
2324 * systemd gained a new unit 'systemd-static-nodes.service'
2325 that generates static device nodes earlier during boot, and
2326 can run in conjunction with udev.
2328 * libsystemd-login gained a new call sd_pid_get_user_unit()
2329 to retrieve the user systemd unit a process is running
2330 in. This is useful for systems where systemd is used as
2333 * systemd-nspawn now places all containers in the new /machine
2334 top-level cgroup directory in the name=systemd
2335 hierarchy. libvirt will soon do the same, so that we get a
2336 uniform separation of /system, /user and /machine for system
2337 services, user processes and containers/virtual
2338 machines. This new cgroup hierarchy is also useful to stick
2339 stable names to specific container instances, which can be
2340 recognized later this way (this name may be controlled
2341 via systemd-nspawn's new -M switch). libsystemd-login also
2342 gained a new call sd_pid_get_machine_name() to retrieve the
2343 name of the container/VM a specific process belongs to.
2345 * bootchart can now store its data in the journal.
2347 * libsystemd-journal gained a new call
2348 sd_journal_add_conjunction() for AND expressions to the
2349 matching logic. This can be used to express more complex
2350 logical expressions.
2352 * journactl can now take multiple --unit= and --user-unit=
2355 * The cryptsetup logic now understands the "luks.key=" kernel
2356 command line switch for specifying a file to read the
2357 decryption key from. Also, if a configured key file is not
2358 found the tool will now automatically fall back to prompting
2361 * Python systemd.journal module was updated to wrap recently
2362 added functions from libsystemd-journal. The interface was
2363 changed to bring the low level interface in s.j._Reader
2364 closer to the C API, and the high level interface in
2365 s.j.Reader was updated to wrap and convert all data about
2368 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
2369 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
2370 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
2371 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
2372 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
2373 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2377 * journalctl --update-catalog now understands a new --root=
2378 option to operate on catalogs found in a different root
2381 * During shutdown after systemd has terminated all running
2382 services a final killing loop kills all remaining left-over
2383 processes. We will now print the name of these processes
2384 when we send SIGKILL to them, since this usually indicates a
2387 * If /etc/crypttab refers to password files stored on
2388 configured mount points automatic dependencies will now be
2389 generated to ensure the specific mount is established first
2390 before the key file is attempted to be read.
2392 * 'systemctl status' will now show information about the
2393 network sockets a socket unit is listening on.
2395 * 'systemctl status' will also shown information about any
2396 drop-in configuration file for units. (Drop-In configuration
2397 files in this context are files such as
2398 /etc/systemd/systemd/foobar.service.d/*.conf)
2400 * systemd-cgtop now optionally shows summed up CPU times of
2401 cgroups. Press '%' while running cgtop to switch between
2402 percentage and absolute mode. This is useful to determine
2403 which cgroups use up the most CPU time over the entire
2404 runtime of the system. systemd-cgtop has also been updated
2405 to be 'pipeable' for processing with further shell tools.
2407 * 'hostnamectl set-hostname' will now allow setting of FQDN
2410 * The formatting and parsing of time span values has been
2411 changed. The parser now understands fractional expressions
2412 such as "5.5h". The formatter will now output fractional
2413 expressions for all time spans under 1min, i.e. "5.123456s"
2414 rather than "5s 123ms 456us". For time spans under 1s
2415 millisecond values are shown, for those under 1ms
2416 microsecond values are shown. This should greatly improve
2417 all time-related output of systemd.
2419 * libsystemd-login and libsystemd-journal gained new
2420 functions for querying the poll() events mask and poll()
2421 timeout value for integration into arbitrary event
2424 * localectl gained the ability to list available X11 keymaps
2425 (models, layouts, variants, options).
2427 * 'systemd-analyze dot' gained the ability to filter for
2428 specific units via shell-style globs, to create smaller,
2429 more useful graphs. I.e. it is now possible to create simple
2430 graphs of all the dependencies between only target units, or
2431 of all units that Avahi has dependencies with.
2433 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2434 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2435 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2436 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2437 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2438 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2439 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2443 * The boot-time readahead implementation for rotating media
2444 will now read the read-ahead data in multiple passes which
2445 consist of all read requests made in equidistant time
2446 intervals. This means instead of strictly reading read-ahead
2447 data in its physical order on disk we now try to find a
2448 middle ground between physical and access time order.
2450 * /etc/os-release files gained a new BUILD_ID= field for usage
2451 on operating systems that provide continuous builds of OS
2454 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2455 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2456 William Douglas, Zbigniew Jędrzejewski-Szmek
2460 * systemd-python gained an API exposing libsystemd-daemon.
2462 * The SMACK setup logic gained support for uploading CIPSO
2465 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2466 ReadOnlyDirectories= and InaccessibleDirectories= has
2467 changed. The private /tmp and /var/tmp directories are now
2468 shared by all processes of a service (which means
2469 ExecStartPre= may now leave data in /tmp that ExecStart= of
2470 the same service can still access). When a service is
2471 stopped its temporary directories are immediately deleted
2472 (normal clean-up with tmpfiles is still done in addition to
2475 * By default, systemd will now set a couple of sysctl
2476 variables in the kernel: the safe sysrq options are turned
2477 on, IP route verification is turned on, and source routing
2478 disabled. The recently added hardlink and softlink
2479 protection of the kernel is turned on. These settings should
2480 be reasonably safe, and good defaults for all new systems.
2482 * The predictable network naming logic may now be turned off
2483 with a new kernel command line switch: net.ifnames=0.
2485 * A new libsystemd-bus module has been added that implements a
2486 pretty complete D-Bus client library. For details see:
2488 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2490 * journald will now explicitly flush the journal files to disk
2491 at the latest 5min after each write. The file will then also
2492 be marked offline until the next write. This should increase
2493 reliability in case of a crash. The synchronization delay
2494 can be configured via SyncIntervalSec= in journald.conf.
2496 * There's a new remote-fs-setup.target unit that can be used
2497 to pull in specific services when at least one remote file
2498 system is to be mounted.
2500 * There are new targets timers.target and paths.target as
2501 canonical targets to pull user timer and path units in
2502 from. This complements sockets.target with a similar
2503 purpose for socket units.
2505 * libudev gained a new call udev_device_set_attribute_value()
2506 to set sysfs attributes of a device.
2508 * The udev daemon now sets the default number of worker
2509 processes executed in parallel based on the number of available
2510 CPUs instead of the amount of available RAM. This is supposed
2511 to provide a more reliable default and limit a too aggressive
2512 paralellism for setups with 1000s of devices connected.
2514 Contributions from: Auke Kok, Colin Walters, Cristian
2515 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2516 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2517 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2518 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2519 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2520 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2521 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2522 Zbigniew Jędrzejewski-Szmek
2526 * Configuration of unit files may now be extended via drop-in
2527 files without having to edit/override the unit files
2528 themselves. More specifically, if the administrator wants to
2529 change one value for a service file foobar.service he can
2530 now do so by dropping in a configuration snippet into
2531 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2532 will load all these snippets and apply them on top of the
2533 main unit configuration file, possibly extending or
2534 overriding its settings. Using these drop-in snippets is
2535 generally nicer than the two earlier options for changing
2536 unit files locally: copying the files from
2537 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2538 them there; or creating a new file in /etc/systemd/system/
2539 that incorporates the original one via ".include". Drop-in
2540 snippets into these .d/ directories can be placed in any
2541 directory systemd looks for units in, and the usual
2542 overriding semantics between /usr/lib, /etc and /run apply
2545 * Most unit file settings which take lists of items can now be
2546 reset by assigning the empty string to them. For example,
2547 normally, settings such as Environment=FOO=BAR append a new
2548 environment variable assignment to the environment block,
2549 each time they are used. By assigning Environment= the empty
2550 string the environment block can be reset to empty. This is
2551 particularly useful with the .d/*.conf drop-in snippets
2552 mentioned above, since this adds the ability to reset list
2553 settings from vendor unit files via these drop-ins.
2555 * systemctl gained a new "list-dependencies" command for
2556 listing the dependencies of a unit recursively.
2558 * Inhibitors are now honored and listed by "systemctl
2559 suspend", "systemctl poweroff" (and similar) too, not only
2560 GNOME. These commands will also list active sessions by
2563 * Resource limits (as exposed by the various control group
2564 controllers) can now be controlled dynamically at runtime
2565 for all units. More specifically, you can now use a command
2566 like "systemctl set-cgroup-attr foobar.service cpu.shares
2567 2000" to alter the CPU shares a specific service gets. These
2568 settings are stored persistently on disk, and thus allow the
2569 administrator to easily adjust the resource usage of
2570 services with a few simple commands. This dynamic resource
2571 management logic is also available to other programs via the
2572 bus. Almost any kernel cgroup attribute and controller is
2575 * systemd-vconsole-setup will now copy all font settings to
2576 all allocated VTs, where it previously applied them only to
2579 * libsystemd-login gained the new sd_session_get_tty() API
2582 * This release drops support for a few legacy or
2583 distribution-specific LSB facility names when parsing init
2584 scripts: $x-display-manager, $mail-transfer-agent,
2585 $mail-transport-agent, $mail-transfer-agent, $smtp,
2586 $null. Also, the mail-transfer-agent.target unit backing
2587 this has been removed. Distributions which want to retain
2588 compatibility with this should carry the burden for
2589 supporting this themselves and patch support for these back
2590 in, if they really need to. Also, the facilities $syslog and
2591 $local_fs are now ignored, since systemd does not support
2592 early-boot LSB init scripts anymore, and these facilities
2593 are implied anyway for normal services. syslog.target has
2596 * There are new bus calls on PID1's Manager object for
2597 cancelling jobs, and removing snapshot units. Previously,
2598 both calls were only available on the Job and Snapshot
2601 * systemd-journal-gatewayd gained SSL support.
2603 * The various "environment" files, such as /etc/locale.conf
2604 now support continuation lines with a backslash ("\") as
2605 last character in the line, similarly in style (but different)
2606 to how this is supported in shells.
2608 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2609 now implicitly appended to every log entry logged. systemctl
2610 has been updated to filter by this field when operating on a
2611 user systemd instance.
2613 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2614 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2615 the container. This makes it easier to boot unmodified
2616 Fedora systems in a container, which however still requires
2617 audit=0 to be passed on the kernel command line. Auditing in
2618 kernel and userspace is unfortunately still too broken in
2619 context of containers, hence we recommend compiling it out
2620 of the kernel or using audit=0. Hopefully this will be fixed
2621 one day for good in the kernel.
2623 * nspawn gained the new --bind= and --bind-ro= parameters to
2624 bind mount specific directories from the host into the
2627 * nspawn will now mount its own devpts file system instance
2628 into the container, in order not to leak pty devices from
2629 the host into the container.
2631 * systemd will now read the firmware boot time performance
2632 information from the EFI variables, if the used boot loader
2633 supports this, and takes it into account for boot performance
2634 analysis via "systemd-analyze". This is currently supported
2635 only in conjunction with Gummiboot, but could be supported
2636 by other boot loaders too. For details see:
2638 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2640 * A new generator has been added that automatically mounts the
2641 EFI System Partition (ESP) to /boot, if that directory
2642 exists, is empty, and no other file system has been
2643 configured to be mounted there.
2645 * logind will now send out PrepareForSleep(false) out
2646 unconditionally, after coming back from suspend. This may be
2647 used by applications as asynchronous notification for
2648 system resume events.
2650 * "systemctl unlock-sessions" has been added, that allows
2651 unlocking the screens of all user sessions at once, similar
2652 to how "systemctl lock-sessions" already locked all users
2653 sessions. This is backed by a new D-Bus call UnlockSessions().
2655 * "loginctl seat-status" will now show the master device of a
2656 seat. (i.e. the device of a seat that needs to be around for
2657 the seat to be considered available, usually the graphics
2660 * tmpfiles gained a new "X" line type, that allows
2661 configuration of files and directories (with wildcards) that
2662 shall be excluded from automatic cleanup ("aging").
2664 * udev default rules set the device node permissions now only
2665 at "add" events, and do not change them any longer with a
2666 later "change" event.
2668 * The log messages for lid events and power/sleep keypresses
2669 now carry a message ID.
2671 * We now have a substantially larger unit test suite, but this
2672 continues to be work in progress.
2674 * udevadm hwdb gained a new --root= parameter to change the
2675 root directory to operate relative to.
2677 * logind will now issue a background sync() request to the kernel
2678 early at shutdown, so that dirty buffers are flushed to disk early
2679 instead of at the last moment, in order to optimize shutdown
2682 * A new bootctl tool has been added that is an interface for
2683 certain boot loader operations. This is currently a preview
2684 and is likely to be extended into a small mechanism daemon
2685 like timedated, localed, hostnamed, and can be used by
2686 graphical UIs to enumerate available boot options, and
2687 request boot into firmware operations.
2689 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2690 the rest of the package. It also has been updated to work
2691 correctly in initrds.
2693 * Policykit previously has been runtime optional, and is now
2694 also compile time optional via a configure switch.
2696 * systemd-analyze has been reimplemented in C. Also "systemctl
2697 dot" has moved into systemd-analyze.
2699 * "systemctl status" with no further parameters will now print
2700 the status of all active or failed units.
2702 * Operations such as "systemctl start" can now be executed
2703 with a new mode "--irreversible" which may be used to queue
2704 operations that cannot accidentally be reversed by a later
2705 job queuing. This is by default used to make shutdown
2706 requests more robust.
2708 * The Python API of systemd now gained a new module for
2709 reading journal files.
2711 * A new tool kernel-install has been added that can install
2712 kernel images according to the Boot Loader Specification:
2714 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2716 * Boot time console output has been improved to provide
2717 animated boot time output for hanging jobs.
2719 * A new tool systemd-activate has been added which can be used
2720 to test socket activation with, directly from the command
2721 line. This should make it much easier to test and debug
2722 socket activation in daemons.
2724 * journalctl gained a new "--reverse" (or -r) option to show
2725 journal output in reverse order (i.e. newest line first).
2727 * journalctl gained a new "--pager-end" (or -e) option to jump
2728 to immediately jump to the end of the journal in the
2729 pager. This is only supported in conjunction with "less".
2731 * journalctl gained a new "--user-unit=" option, that works
2732 similarly to "--unit=" but filters for user units rather than
2735 * A number of unit files to ease adoption of systemd in
2736 initrds has been added. This moves some minimal logic from
2737 the various initrd implementations into systemd proper.
2739 * The journal files are now owned by a new group
2740 "systemd-journal", which exists specifically to allow access
2741 to the journal, and nothing else. Previously, we used the
2742 "adm" group for that, which however possibly covers more
2743 than just journal/log file access. This new group is now
2744 already used by systemd-journal-gatewayd to ensure this
2745 daemon gets access to the journal files and as little else
2746 as possible. Note that "make install" will also set FS ACLs
2747 up for /var/log/journal to give "adm" and "wheel" read
2748 access to it, in addition to "systemd-journal" which owns
2749 the journal files. We recommend that packaging scripts also
2750 add read access to "adm" + "wheel" to /var/log/journal, and
2751 all existing/future journal files. To normal users and
2752 administrators little changes, however packagers need to
2753 ensure to create the "systemd-journal" system group at
2754 package installation time.
2756 * The systemd-journal-gatewayd now runs as unprivileged user
2757 systemd-journal-gateway:systemd-journal-gateway. Packaging
2758 scripts need to create these system user/group at
2761 * timedated now exposes a new boolean property CanNTP that
2762 indicates whether a local NTP service is available or not.
2764 * systemd-detect-virt will now also detect xen PVs
2766 * The pstore file system is now mounted by default, if it is
2769 * In addition to the SELinux and IMA policies we will now also
2770 load SMACK policies at early boot.
2772 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2773 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2774 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2775 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2776 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2777 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2778 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2779 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2780 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2781 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2782 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2783 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2784 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2785 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2789 * Timer units now support calendar time events in addition to
2790 monotonic time events. That means you can now trigger a unit
2791 based on a calendar time specification such as "Thu,Fri
2792 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2793 or fifth day of any month of the year 2013, given that it is
2794 a thursday or friday. This brings timer event support
2795 considerably closer to cron's capabilities. For details on
2796 the supported calendar time specification language see
2799 * udev now supports a number of different naming policies for
2800 network interfaces for predictable names, and a combination
2801 of these policies is now the default. Please see this wiki
2802 document for details:
2804 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2806 * Auke Kok's bootchart implementation has been added to the
2807 systemd tree. It is an optional component that can graph the
2808 boot in quite some detail. It is one of the best bootchart
2809 implementations around and minimal in its code and
2812 * nss-myhostname has been integrated into the systemd source
2813 tree. nss-myhostname guarantees that the local hostname
2814 always stays resolvable via NSS. It has been a weak
2815 requirement of systemd-hostnamed since a long time, and
2816 since its code is actually trivial we decided to just
2817 include it in systemd's source tree. It can be turned off
2818 with a configure switch.
2820 * The read-ahead logic is now capable of properly detecting
2821 whether a btrfs file system is on SSD or rotating media, in
2822 order to optimize the read-ahead scheme. Previously, it was
2823 only capable of detecting this on traditional file systems
2826 * In udev, additional device properties are now read from the
2827 IAB in addition to the OUI database. Also, Bluetooth company
2828 identities are attached to the devices as well.
2830 * In service files %U may be used as specifier that is
2831 replaced by the configured user name of the service.
2833 * nspawn may now be invoked without a controlling TTY. This
2834 makes it suitable for invocation as its own service. This
2835 may be used to set up a simple containerized server system
2836 using only core OS tools.
2838 * systemd and nspawn can now accept socket file descriptors
2839 when they are started for socket activation. This enables
2840 implementation of socket activated nspawn
2841 containers. i.e. think about autospawning an entire OS image
2842 when the first SSH or HTTP connection is received. We expect
2843 that similar functionality will also be added to libvirt-lxc
2846 * journalctl will now suppress ANSI color codes when
2847 presenting log data.
2849 * systemctl will no longer show control group information for
2850 a unit if a the control group is empty anyway.
2852 * logind can now automatically suspend/hibernate/shutdown the
2855 * /etc/machine-info and hostnamed now also expose the chassis
2856 type of the system. This can be used to determine whether
2857 the local system is a laptop, desktop, handset or
2858 tablet. This information may either be configured by the
2859 user/vendor or is automatically determined from ACPI and DMI
2860 information if possible.
2862 * A number of PolicyKit actions are now bound together with
2863 "imply" rules. This should simplify creating UIs because
2864 many actions will now authenticate similar ones as well.
2866 * Unit files learnt a new condition ConditionACPower= which
2867 may be used to conditionalize a unit depending on whether an
2868 AC power source is connected or not, of whether the system
2869 is running on battery power.
2871 * systemctl gained a new "is-failed" verb that may be used in
2872 shell scripts and suchlike to check whether a specific unit
2873 is in the "failed" state.
2875 * The EnvironmentFile= setting in unit files now supports file
2876 globbing, and can hence be used to easily read a number of
2877 environment files at once.
2879 * systemd will no longer detect and recognize specific
2880 distributions. All distribution-specific #ifdeffery has been
2881 removed, systemd is now fully generic and
2882 distribution-agnostic. Effectively, not too much is lost as
2883 a lot of the code is still accessible via explicit configure
2884 switches. However, support for some distribution specific
2885 legacy configuration file formats has been dropped. We
2886 recommend distributions to simply adopt the configuration
2887 files everybody else uses now and convert the old
2888 configuration from packaging scripts. Most distributions
2889 already did that. If that's not possible or desirable,
2890 distributions are welcome to forward port the specific
2891 pieces of code locally from the git history.
2893 * When logging a message about a unit systemd will now always
2894 log the unit name in the message meta data.
2896 * localectl will now also discover system locale data that is
2897 not stored in locale archives, but directly unpacked.
2899 * logind will no longer unconditionally use framebuffer
2900 devices as seat masters, i.e. as devices that are required
2901 to be existing before a seat is considered preset. Instead,
2902 it will now look for all devices that are tagged as
2903 "seat-master" in udev. By default framebuffer devices will
2904 be marked as such, but depending on local systems other
2905 devices might be marked as well. This may be used to
2906 integrate graphics cards using closed source drivers (such
2907 as NVidia ones) more nicely into logind. Note however, that
2908 we recommend using the open source NVidia drivers instead,
2909 and no udev rules for the closed-source drivers will be
2910 shipped from us upstream.
2912 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2913 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2914 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2915 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2916 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2917 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2918 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2919 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2920 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2921 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2922 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2923 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2928 * udev gained support for loading additional device properties
2929 from an indexed database that is keyed by vendor/product IDs
2930 and similar device identifiers. For the beginning this
2931 "hwdb" is populated with data from the well-known PCI and
2932 USB database, but also includes PNP, ACPI and OID data. In
2933 the longer run this indexed database shall grow into
2934 becoming the one central database for non-essential
2935 userspace device metadata. Previously, data from the PCI/USB
2936 database was only attached to select devices, since the
2937 lookup was a relatively expensive operation due to O(n) time
2938 complexity (with n being the number of entries in the
2939 database). Since this is now O(1), we decided to add in this
2940 data for all devices where this is available, by
2941 default. Note that the indexed database needs to be rebuilt
2942 when new data files are installed. To achieve this you need
2943 to update your packaging scripts to invoke "udevadm hwdb
2944 --update" after installation of hwdb data files. For
2945 RPM-based distributions we introduced the new
2946 %udev_hwdb_update macro for this purpose.
2948 * The Journal gained support for the "Message Catalog", an
2949 indexed database to link up additional information with
2950 journal entries. For further details please check:
2952 http://www.freedesktop.org/wiki/Software/systemd/catalog
2954 The indexed message catalog database also needs to be
2955 rebuilt after installation of message catalog files. Use
2956 "journalctl --update-catalog" for this. For RPM-based
2957 distributions we introduced the %journal_catalog_update
2958 macro for this purpose.
2960 * The Python Journal bindings gained support for the standard
2961 Python logging framework.
2963 * The Journal API gained new functions for checking whether
2964 the underlying file system of a journal file is capable of
2965 properly reporting file change notifications, or whether
2966 applications that want to reflect journal changes "live"
2967 need to recheck journal files continuously in appropriate
2970 * It is now possible to set the "age" field for tmpfiles
2971 entries to 0, indicating that files matching this entry
2972 shall always be removed when the directories are cleaned up.
2974 * coredumpctl gained a new "gdb" verb which invokes gdb
2975 right-away on the selected coredump.
2977 * There's now support for "hybrid sleep" on kernels that
2978 support this, in addition to "suspend" and "hibernate". Use
2979 "systemctl hybrid-sleep" to make use of this.
2981 * logind's HandleSuspendKey= setting (and related settings)
2982 now gained support for a new "lock" setting to simply
2983 request the screen lock on all local sessions, instead of
2984 actually executing a suspend or hibernation.
2986 * systemd will now mount the EFI variables file system by
2989 * Socket units now gained support for configuration of the
2990 SMACK security label.
2992 * timedatectl will now output the time of the last and next
2993 daylight saving change.
2995 * We dropped support for various legacy and distro-specific
2996 concepts, such as insserv, early-boot SysV services
2997 (i.e. those for non-standard runlevels such as 'b' or 'S')
2998 or ArchLinux /etc/rc.conf support. We recommend the
2999 distributions who still need support this to either continue
3000 to maintain the necessary patches downstream, or find a
3001 different solution. (Talk to us if you have questions!)
3003 * Various systemd components will now bypass PolicyKit checks
3004 for root and otherwise handle properly if PolicyKit is not
3005 found to be around. This should fix most issues for
3006 PolicyKit-less systems. Quite frankly this should have been
3007 this way since day one. It is absolutely our intention to
3008 make systemd work fine on PolicyKit-less systems, and we
3009 consider it a bug if something does not work as it should if
3010 PolicyKit is not around.
3012 * For embedded systems it is now possible to build udev and
3013 systemd without blkid and/or kmod support.
3015 * "systemctl switch-root" is now capable of switching root
3016 more than once. I.e. in addition to transitions from the
3017 initrd to the host OS it is now possible to transition to
3018 further OS images from the host. This is useful to implement
3019 offline updating tools.
3021 * Various other additions have been made to the RPM macros
3022 shipped with systemd. Use %udev_rules_update() after
3023 installing new udev rules files. %_udevhwdbdir,
3024 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3025 %_sysctldir are now available which resolve to the right
3026 directories for packages to place various data files in.
3028 * journalctl gained the new --full switch (in addition to
3029 --all, to disable ellipsation for long messages.
3031 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3032 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3033 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3034 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3035 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3036 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3037 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3038 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3039 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3043 * journalctl gained new --since= and --until= switches to
3044 filter by time. It also now supports nice filtering for
3045 units via --unit=/-u.
3047 * Type=oneshot services may use ExecReload= and do the
3050 * The journal daemon now supports time-based rotation and
3051 vacuuming, in addition to the usual disk-space based
3054 * The journal will now index the available field values for
3055 each field name. This enables clients to show pretty drop
3056 downs of available match values when filtering. The bash
3057 completion of journalctl has been updated
3058 accordingly. journalctl gained a new switch -F to list all
3059 values a certain field takes in the journal database.
3061 * More service events are now written as structured messages
3062 to the journal, and made recognizable via message IDs.
3064 * The timedated, localed and hostnamed mini-services which
3065 previously only provided support for changing time, locale
3066 and hostname settings from graphical DEs such as GNOME now
3067 also have a minimal (but very useful) text-based client
3068 utility each. This is probably the nicest way to changing
3069 these settings from the command line now, especially since
3070 it lists available options and is fully integrated with bash
3073 * There's now a new tool "systemd-coredumpctl" to list and
3074 extract coredumps from the journal.
3076 * We now install a README each in /var/log/ and
3077 /etc/rc.d/init.d explaining where the system logs and init
3078 scripts went. This hopefully should help folks who go to
3079 that dirs and look into the otherwise now empty void and
3080 scratch their heads.
3082 * When user-services are invoked (by systemd --user) the
3083 $MANAGERPID env var is set to the PID of systemd.
3085 * SIGRTMIN+24 when sent to a --user instance will now result
3086 in immediate termination of systemd.
3088 * gatewayd received numerous feature additions such as a
3089 "follow" mode, for live syncing and filtering.
3091 * browse.html now allows filtering and showing detailed
3092 information on specific entries. Keyboard navigation and
3093 mouse screen support has been added.
3095 * gatewayd/journalctl now supports HTML5/JSON
3096 Server-Sent-Events as output.
3098 * The SysV init script compatibility logic will now
3099 heuristically determine whether a script supports the
3100 "reload" verb, and only then make this available as
3103 * "systemctl status --follow" has been removed, use "journalctl
3106 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3107 have been removed since they are hardly useful to be
3110 * And I'd like to take the opportunity to specifically mention
3111 Zbigniew for his great contributions. Zbigniew, you rock!
3113 Contributions from: Andrew Eikum, Christian Hesse, Colin
3114 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3115 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3116 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3117 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3118 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3119 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3123 * If /etc/vconsole.conf is non-existent or empty we will no
3124 longer load any console font or key map at boot by
3125 default. Instead the kernel defaults will be left
3126 intact. This is definitely the right thing to do, as no
3127 configuration should mean no configuration, and hard-coding
3128 font names that are different on all archs is probably a bad
3129 idea. Also, the kernel default key map and font should be
3130 good enough for most cases anyway, and mostly identical to
3131 the userspace fonts/key maps we previously overloaded them
3132 with. If distributions want to continue to default to a
3133 non-kernel font or key map they should ship a default
3134 /etc/vconsole.conf with the appropriate contents.
3136 Contributions from: Colin Walters, Daniel J Walsh, Dave
3137 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3138 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3142 * journalctl gained a new --cursor= switch to show entries
3143 starting from the specified location in the journal.
3145 * We now enforce a size limit on journal entry fields exported
3146 with "-o json" in journalctl. Fields larger than 4K will be
3147 assigned null. This can be turned off with --all.
3149 * An (optional) journal gateway daemon is now available as
3150 "systemd-journal-gatewayd.service". This service provides
3151 access to the journal via HTTP and JSON. This functionality
3152 will be used to implement live log synchronization in both
3153 pull and push modes, but has various other users too, such
3154 as easy log access for debugging of embedded devices. Right
3155 now it is already useful to retrieve the journal via HTTP:
3157 # systemctl start systemd-journal-gatewayd.service
3158 # wget http://localhost:19531/entries
3160 This will download the journal contents in a
3161 /var/log/messages compatible format. The same as JSON:
3163 # curl -H"Accept: application/json" http://localhost:19531/entries
3165 This service is also accessible via a web browser where a
3166 single static HTML5 app is served that uses the JSON logic
3167 to enable the user to do some basic browsing of the
3168 journal. This will be extended later on. Here's an example
3169 screenshot of this app in its current state:
3171 http://0pointer.de/public/journal-gatewayd
3173 Contributions from: Kay Sievers, Lennart Poettering, Robert
3174 Milasan, Tom Gundersen
3178 * The bash completion logic is now available for journalctl
3181 * We do not mount the "cpuset" controller anymore together with
3182 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3183 started if no parameters are assigned to it. "cpuset" hence
3184 broke code that assumed it it could create "cpu" groups and
3187 * journalctl -f will now subscribe to terminal size changes,
3188 and line break accordingly.
3190 Contributions from: Dave Reisner, Kay Sievers, Lennart
3191 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3195 * nspawn will now create a symlink /etc/localtime in the
3196 container environment, copying the host's timezone
3197 setting. Previously this has been done via a bind mount, but
3198 since symlinks cannot be bind mounted this has now been
3199 changed to create/update the appropriate symlink.
3201 * journalctl -n's line number argument is now optional, and
3202 will default to 10 if omitted.
3204 * journald will now log the maximum size the journal files may
3205 take up on disk. This is particularly useful if the default
3206 built-in logic of determining this parameter from the file
3207 system size is used. Use "systemctl status
3208 systemd-journald.service" to see this information.
3210 * The multi-seat X wrapper tool has been stripped down. As X
3211 is now capable of enumerating graphics devices via udev in a
3212 seat-aware way the wrapper is not strictly necessary
3213 anymore. A stripped down temporary stop-gap is still shipped
3214 until the upstream display managers have been updated to
3215 fully support the new X logic. Expect this wrapper to be
3216 removed entirely in one of the next releases.
3218 * HandleSleepKey= in logind.conf has been split up into
3219 HandleSuspendKey= and HandleHibernateKey=. The old setting
3220 is not available anymore. X11 and the kernel are
3221 distuingishing between these keys and we should too. This
3222 also means the inhibition lock for these keys has been split
3225 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3226 Poettering, Lukas Nykryn, Václav Pavlín
3230 * Whenever a unit changes state we will now log this to the
3231 journal and show along the unit's own log output in
3234 * ConditionPathIsMountPoint= can now properly detect bind
3235 mount points too. (Previously, a bind mount of one file
3236 system to another place in the same file system could not be
3237 detected as mount, since they shared struct stat's st_dev
3240 * We will now mount the cgroup controllers cpu, cpuacct,
3241 cpuset and the controllers net_cls, net_prio together by
3244 * nspawn containers will now have a virtualized boot
3245 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
3246 over with a randomized ID at container initialization). This
3247 has the effect of making "journalctl -b" do the right thing
3250 * The JSON output journal serialization has been updated not
3251 to generate "endless" list objects anymore, but rather one
3252 JSON object per line. This is more in line how most JSON
3253 parsers expect JSON objects. The new output mode
3254 "json-pretty" has been added to provide similar output, but
3255 neatly aligned for readability by humans.
3257 * We dropped all explicit sync() invocations in the shutdown
3258 code. The kernel does this implicitly anyway in the kernel
3259 reboot() syscall. halt(8)'s -n option is now a compatibility
3262 * We now support virtualized reboot() in containers, as
3263 supported by newer kernels. We will fall back to exit() if
3264 CAP_SYS_REBOOT is not available to the container. Also,
3265 nspawn makes use of this now and will actually reboot the
3266 container if the containerized OS asks for that.
3268 * journalctl will only show local log output by default
3269 now. Use --merge (-m) to show remote log output, too.
3271 * libsystemd-journal gained the new sd_journal_get_usage()
3272 call to determine the current disk usage of all journal
3273 files. This is exposed in the new "journalctl --disk-usage"
3276 * journald gained a new configuration setting SplitMode= in
3277 journald.conf which may be used to control how user journals
3278 are split off. See journald.conf(5) for details.
3280 * A new condition type ConditionFileNotEmpty= has been added.
3282 * tmpfiles' "w" lines now support file globbing, to write
3283 multiple files at once.
3285 * We added Python bindings for the journal submission
3286 APIs. More Python APIs for a number of selected APIs will
3287 likely follow. Note that we intend to add native bindings
3288 only for the Python language, as we consider it common
3289 enough to deserve bindings shipped within systemd. There are
3290 various projects outside of systemd that provide bindings
3291 for languages such as PHP or Lua.
3293 * Many conditions will now resolve specifiers such as %i. In
3294 addition, PathChanged= and related directives of .path units
3295 now support specifiers as well.
3297 * There's now a new RPM macro definition for the system preset
3300 * journald will now warn if it ca not forward a message to the
3301 syslog daemon because its socket is full.
3303 * timedated will no longer write or process /etc/timezone,
3304 except on Debian. As we do not support late mounted /usr
3305 anymore /etc/localtime always being a symlink is now safe,
3306 and hence the information in /etc/timezone is not necessary
3309 * logind will now always reserve one VT for a text getty (VT6
3310 by default). Previously if more than 6 X sessions where
3311 started they took up all the VTs with auto-spawned gettys,
3312 so that no text gettys were available anymore.
3314 * udev will now automatically inform the btrfs kernel logic
3315 about btrfs RAID components showing up. This should make
3316 simple hotplug based btrfs RAID assembly work.
3318 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
3319 (but not for its children which will stay at the kernel
3320 default). This should allow setups with a lot more listening
3323 * systemd will now always pass the configured timezone to the
3324 kernel at boot. timedated will do the same when the timezone
3327 * logind's inhibition logic has been updated. By default,
3328 logind will now handle the lid switch, the power and sleep
3329 keys all the time, even in graphical sessions. If DEs want
3330 to handle these events on their own they should take the new
3331 handle-power-key, handle-sleep-key and handle-lid-switch
3332 inhibitors during their runtime. A simple way to achiveve
3333 that is to invoke the DE wrapped in an invocation of:
3335 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
3337 * Access to unit operations is now checked via SELinux taking
3338 the unit file label and client process label into account.
3340 * systemd will now notify the administrator in the journal
3341 when he over-mounts a non-empty directory.
3343 * There are new specifiers that are resolved in unit files,
3344 for the host name (%H), the machine ID (%m) and the boot ID
3347 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
3348 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
3349 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
3350 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
3351 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
3352 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3353 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
3357 * Support for reading structured kernel messages from
3358 /dev/kmsg has now been added and is enabled by default.
3360 * Support for reading kernel messages from /proc/kmsg has now
3361 been removed. If you want kernel messages in the journal
3362 make sure to run a recent kernel (>= 3.5) that supports
3363 reading structured messages from /dev/kmsg (see
3364 above). /proc/kmsg is now exclusive property of classic
3365 syslog daemons again.
3367 * The libudev API gained the new
3368 udev_device_new_from_device_id() call.
3370 * The logic for file system namespace (ReadOnlyDirectory=,
3371 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
3372 require pivot_root() anymore. This means fewer temporary
3373 directories are created below /tmp for this feature.
3375 * nspawn containers will now see and receive all submounts
3376 made on the host OS below the root file system of the
3379 * Forward Secure Sealing is now supported for Journal files,
3380 which provide cryptographical sealing of journal files so
3381 that attackers cannot alter log history anymore without this
3382 being detectable. Lennart will soon post a blog story about
3383 this explaining it in more detail.
3385 * There are two new service settings RestartPreventExitStatus=
3386 and SuccessExitStatus= which allow configuration of exit
3387 status (exit code or signal) which will be excepted from the
3388 restart logic, resp. consider successful.
3390 * journalctl gained the new --verify switch that can be used
3391 to check the integrity of the structure of journal files and
3392 (if Forward Secure Sealing is enabled) the contents of
3395 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3396 and similar symlinks pre-created. This makes running shells
3397 as container init process a lot more fun.
3399 * The fstab support can now handle PARTUUID= and PARTLABEL=
3402 * A new ConditionHost= condition has been added to match
3403 against the hostname (with globs) and machine ID. This is
3404 useful for clusters where a single OS image is used to
3405 provision a large number of hosts which shall run slightly
3406 different sets of services.
3408 * Services which hit the restart limit will now be placed in a
3411 Contributions from: Bertram Poettering, Dave Reisner, Huang
3412 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3413 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3417 * When running in --user mode systemd will now become a
3418 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3419 tree a lot more organized.
3421 * A new PartOf= unit dependency type has been introduced that
3422 may be used to group services in a natural way.
3424 * "systemctl enable" may now be used to enable instances of
3427 * journalctl now prints error log levels in red, and
3428 warning/notice log levels in bright white. It also supports
3429 filtering by log level now.
3431 * cgtop gained a new -n switch (similar to top), to configure
3432 the maximum number of iterations to run for. It also gained
3433 -b, to run in batch mode (accepting no input).
3435 * The suffix ".service" may now be omitted on most systemctl
3436 command lines involving service unit names.
3438 * There's a new bus call in logind to lock all sessions, as
3439 well as a loginctl verb for it "lock-sessions".
3441 * libsystemd-logind.so gained a new call sd_journal_perror()
3442 that works similar to libc perror() but logs to the journal
3443 and encodes structured information about the error number.
3445 * /etc/crypttab entries now understand the new keyfile-size=
3448 * shutdown(8) now can send a (configurable) wall message when
3449 a shutdown is cancelled.
3451 * The mount propagation mode for the root file system will now
3452 default to "shared", which is useful to make containers work
3453 nicely out-of-the-box so that they receive new mounts from
3454 the host. This can be undone locally by running "mount
3455 --make-rprivate /" if needed.
3457 * The prefdm.service file has been removed. Distributions
3458 should maintain this unit downstream if they intend to keep
3459 it around. However, we recommend writing normal unit files
3460 for display managers instead.
3462 * Since systemd is a crucial part of the OS we will now
3463 default to a number of compiler switches that improve
3464 security (hardening) such as read-only relocations, stack
3465 protection, and suchlike.
3467 * The TimeoutSec= setting for services is now split into
3468 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3469 of individual time outs for the start and the stop phase of
3472 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3473 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3474 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3475 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3476 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3477 Gundersen, Zbigniew Jędrzejewski-Szmek
3481 * The journal and id128 C APIs are now fully documented as man
3484 * Extra safety checks have been added when transitioning from
3485 the initial RAM disk to the main system to avoid accidental
3488 * /etc/crypttab entries now understand the new keyfile-offset=
3491 * systemctl -t can now be used to filter by unit load state.
3493 * The journal C API gained the new sd_journal_wait() call to
3494 make writing synchronous journal clients easier.
3496 * journalctl gained the new -D switch to show journals from a
3499 * journalctl now displays a special marker between log
3500 messages of two different boots.
3502 * The journal is now explicitly flushed to /var via a service
3503 systemd-journal-flush.service, rather than implicitly simply
3504 by seeing /var/log/journal to be writable.
3506 * journalctl (and the journal C APIs) can now match for much
3507 more complex expressions, with alternatives and
3510 * When transitioning from the initial RAM disk to the main
3511 system we will now kill all processes in a killing spree to
3512 ensure no processes stay around by accident.
3514 * Three new specifiers may be used in unit files: %u, %h, %s
3515 resolve to the user name, user home directory resp. user
3516 shell. This is useful for running systemd user instances.
3518 * We now automatically rotate journal files if their data
3519 object hash table gets a fill level > 75%. We also size the
3520 hash table based on the configured maximum file size. This
3521 together should lower hash collisions drastically and thus
3522 speed things up a bit.
3524 * journalctl gained the new "--header" switch to introspect
3525 header data of journal files.
3527 * A new setting SystemCallFilters= has been added to services
3528 which may be used to apply blacklists or whitelists to
3529 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3531 * nspawn gained a new --link-journal= switch (and quicker: -j)
3532 to link the container journal with the host. This makes it
3533 very easy to centralize log viewing on the host for all
3534 guests while still keeping the journal files separated.
3536 * Many bugfixes and optimizations
3538 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3539 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3540 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3545 * Several tools now understand kernel command line arguments,
3546 which are only read when run in an initial RAM disk. They
3547 usually follow closely their normal counterparts, but are
3550 * There's a new tool to analyze the readahead files that are
3551 automatically generated at boot. Use:
3553 /usr/lib/systemd/systemd-readahead analyze /.readahead
3555 * We now provide an early debug shell on tty9 if this enabled. Use:
3557 systemctl enable debug-shell.service
3559 * All plymouth related units have been moved into the Plymouth
3560 package. Please make sure to upgrade your Plymouth version
3563 * systemd-tmpfiles now supports getting passed the basename of
3564 a configuration file only, in which case it will look for it
3565 in all appropriate directories automatically.
3567 * udevadm info now takes a /dev or /sys path as argument, and
3568 does the right thing. Example:
3570 udevadm info /dev/sda
3571 udevadm info /sys/class/block/sda
3573 * systemctl now prints a warning if a unit is stopped but a
3574 unit that might trigger it continues to run. Example: a
3575 service is stopped but the socket that activates it is left
3578 * "systemctl status" will now mention if the log output was
3579 shortened due to rotation since a service has been started.
3581 * The journal API now exposes functions to determine the
3582 "cutoff" times due to rotation.
3584 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3585 immediately flushing of runtime logs to /var if possible,
3586 resp. for triggering immediate rotation of the journal
3589 * It is now considered an error if a service is attempted to
3590 be stopped that is not loaded.
3592 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3594 * systemd-analyze now supports Python 3
3596 * tmpfiles now supports cleaning up directories via aging
3597 where the first level dirs are always kept around but
3598 directories beneath it automatically aged. This is enabled
3599 by prefixing the age field with '~'.
3601 * Seat objects now expose CanGraphical, CanTTY properties
3602 which is required to deal with very fast bootups where the
3603 display manager might be running before the graphics drivers
3604 completed initialization.
3606 * Seat objects now expose a State property.
3608 * We now include RPM macros for service enabling/disabling
3609 based on the preset logic. We recommend RPM based
3610 distributions to make use of these macros if possible. This
3611 makes it simpler to reuse RPM spec files across
3614 * We now make sure that the collected systemd unit name is
3615 always valid when services log to the journal via
3618 * There's a new man page kernel-command-line(7) detailing all
3619 command line options we understand.
3621 * The fstab generator may now be disabled at boot by passing
3622 fstab=0 on the kernel command line.
3624 * A new kernel command line option modules-load= is now understood
3625 to load a specific kernel module statically, early at boot.
3627 * Unit names specified on the systemctl command line are now
3628 automatically escaped as needed. Also, if file system or
3629 device paths are specified they are automatically turned
3630 into the appropriate mount or device unit names. Example:
3632 systemctl status /home
3633 systemctl status /dev/sda
3635 * The SysVConsole= configuration option has been removed from
3636 system.conf parsing.
3638 * The SysV search path is no longer exported on the D-Bus
3641 * The Names= option is been removed from unit file parsing.
3643 * There's a new man page bootup(7) detailing the boot process.
3645 * Every unit and every generator we ship with systemd now
3646 comes with full documentation. The self-explanatory boot is
3649 * A couple of services gained "systemd-" prefixes in their
3650 name if they wrap systemd code, rather than only external
3651 code. Among them fsck@.service which is now
3652 systemd-fsck@.service.
3654 * The HaveWatchdog property has been removed from the D-Bus
3657 * systemd.confirm_spawn= on the kernel command line should now
3660 * There's a new man page crypttab(5) which details all options
3661 we actually understand.
3663 * systemd-nspawn gained a new --capability= switch to pass
3664 additional capabilities to the container.
3666 * timedated will now read known NTP implementation unit names
3667 from /usr/lib/systemd/ntp-units.d/*.list,
3668 systemd-timedated-ntp.target has been removed.
3670 * journalctl gained a new switch "-b" that lists log data of
3671 the current boot only.
3673 * The notify socket is in the abstract namespace again, in
3674 order to support daemons which chroot() at start-up.
3676 * There is a new Storage= configuration option for journald
3677 which allows configuration of where log data should go. This
3678 also provides a way to disable journal logging entirely, so
3679 that data collected is only forwarded to the console, the
3680 kernel log buffer or another syslog implementation.
3682 * Many bugfixes and optimizations
3684 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3685 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3686 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3687 Shawn Landden, Tom Gundersen
3691 * "systemctl help <unit>" now shows the man page if one is
3694 * Several new man pages have been added.
3696 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3697 MaxLevelConsole= can now be specified in
3698 journald.conf. These options allow reducing the amount of
3699 data stored on disk or forwarded by the log level.
3701 * TimerSlackNSec= can now be specified in system.conf for
3702 PID1. This allows system-wide power savings.
3704 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3705 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3710 * logind is now capable of (optionally) handling power and
3711 sleep keys as well as the lid switch.
3713 * journalctl now understands the syntax "journalctl
3714 /usr/bin/avahi-daemon" to get all log output of a specific
3717 * CapabilityBoundingSet= in system.conf now also influences
3718 the capability bound set of usermode helpers of the kernel.
3720 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3721 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3722 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3723 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3727 * Note that we skipped 139 releases here in order to set the
3728 new version to something that is greater than both udev's
3729 and systemd's most recent version number.
3731 * udev: all udev sources are merged into the systemd source tree now.
3732 All future udev development will happen in the systemd tree. It
3733 is still fully supported to use the udev daemon and tools without
3734 systemd running, like in initramfs or other init systems. Building
3735 udev though, will require the *build* of the systemd tree, but
3736 udev can be properly *run* without systemd.
3738 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3739 should be used to create dead device nodes as workarounds for broken
3742 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3743 no longer supported. udev_monitor_new_from_netlink() needs to be
3744 used to subscribe to events.
3746 * udev: when udevd is started by systemd, processes which are left
3747 behind by forking them off of udev rules, are unconditionally cleaned
3748 up and killed now after the event handling has finished. Services or
3749 daemons must be started as systemd services. Services can be
3750 pulled-in by udev to get started, but they can no longer be directly
3751 forked by udev rules.
3753 * udev: the daemon binary is called systemd-udevd now and installed
3754 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3755 to adapt to that, create symlink, or rename the binary after building
3758 * libudev no longer provides these symbols:
3759 udev_monitor_from_socket()
3760 udev_queue_get_failed_list_entry()
3761 udev_get_{dev,sys,run}_path()
3762 The versions number was bumped and symbol versioning introduced.
3764 * systemd-loginctl and systemd-journalctl have been renamed
3765 to loginctl and journalctl to match systemctl.
3767 * The config files: /etc/systemd/systemd-logind.conf and
3768 /etc/systemd/systemd-journald.conf have been renamed to
3769 logind.conf and journald.conf. Package updates should rename
3770 the files to the new names on upgrade.
3772 * For almost all files the license is now LGPL2.1+, changed
3773 from the previous GPL2.0+. Exceptions are some minor stuff
3774 of udev (which will be changed to LGPL2.1 eventually, too),
3775 and the MIT licensed sd-daemon.[ch] library that is suitable
3776 to be used as drop-in files.
3778 * systemd and logind now handle system sleep states, in
3779 particular suspending and hibernating.
3781 * logind now implements a sleep/shutdown/idle inhibiting logic
3782 suitable for a variety of uses. Soonishly Lennart will blog
3783 about this in more detail.
3785 * var-run.mount and var-lock.mount are no longer provided
3786 (which prevously bind mounted these directories to their new
3787 places). Distributions which have not converted these
3788 directories to symlinks should consider stealing these files
3789 from git history and add them downstream.
3791 * We introduced the Documentation= field for units and added
3792 this to all our shipped units. This is useful to make it
3793 easier to explore the boot and the purpose of the various
3796 * All smaller setup units (such as
3797 systemd-vconsole-setup.service) now detect properly if they
3798 are run in a container and are skipped when
3799 appropriate. This guarantees an entirely noise-free boot in
3800 Linux container environments such as systemd-nspawn.
3802 * A framework for implementing offline system updates is now
3803 integrated, for details see:
3804 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3806 * A new service type Type=idle is available now which helps us
3807 avoiding ugly interleaving of getty output and boot status
3810 * There's now a system-wide CapabilityBoundingSet= option to
3811 globally reduce the set of capabilities for the
3812 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3813 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3814 even CAP_NET_ADMIN system-wide for secure systems.
3816 * There are now system-wide DefaultLimitXXX= options to
3817 globally change the defaults of the various resource limits
3818 for all units started by PID 1.
3820 * Harald Hoyer's systemd test suite has been integrated into
3821 systemd which allows easy testing of systemd builds in qemu
3822 and nspawn. (This is really awesome! Ask us for details!)
3824 * The fstab parser is now implemented as generator, not inside
3827 * systemctl will now warn you if .mount units generated from
3828 /etc/fstab are out of date due to changes in fstab that
3829 have not been read by systemd yet.
3831 * systemd is now suitable for usage in initrds. Dracut has
3832 already been updated to make use of this. With this in place
3833 initrds get a slight bit faster but primarily are much
3834 easier to introspect and debug since "systemctl status" in
3835 the host system can be used to introspect initrd services,
3836 and the journal from the initrd is kept around too.
3838 * systemd-delta has been added, a tool to explore differences
3839 between user/admin configuration and vendor defaults.
3841 * PrivateTmp= now affects both /tmp and /var/tmp.
3843 * Boot time status messages are now much prettier and feature
3844 proper english language. Booting up systemd has never been
3847 * Read-ahead pack files now include the inode number of all
3848 files to pre-cache. When the inode changes the pre-caching
3849 is not attempted. This should be nicer to deal with updated
3850 packages which might result in changes of read-ahead
3853 * We now temporaritly lower the kernel's read_ahead_kb variable
3854 when collecting read-ahead data to ensure the kernel's
3855 built-in read-ahead does not add noise to our measurements
3856 of necessary blocks to pre-cache.
3858 * There's now RequiresMountsFor= to add automatic dependencies
3859 for all mounts necessary for a specific file system path.
3861 * MountAuto= and SwapAuto= have been removed from
3862 system.conf. Mounting file systems at boot has to take place
3865 * nspawn now learned a new switch --uuid= to set the machine
3866 ID on the command line.
3868 * nspawn now learned the -b switch to automatically search
3871 * vt102 is now the default TERM for serial TTYs, upgraded from
3874 * systemd-logind now works on VT-less systems.
3876 * The build tree has been reorganized. The individual
3877 components now have directories of their own.
3879 * A new condition type ConditionPathIsReadWrite= is now available.
3881 * nspawn learned the new -C switch to create cgroups for the
3882 container in other hierarchies.
3884 * We now have support for hardware watchdogs, configurable in
3887 * The scheduled shutdown logic now has a public API.
3889 * We now mount /tmp as tmpfs by default, but this can be
3890 masked and /etc/fstab can override it.
3892 * Since udisks does not make use of /media anymore we are not
3893 mounting a tmpfs on it anymore.
3895 * journalctl gained a new --local switch to only interleave
3896 locally generated journal files.
3898 * We can now load the IMA policy at boot automatically.
3900 * The GTK tools have been split off into a systemd-ui.
3902 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3903 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3904 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3905 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3906 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3907 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3908 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3909 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3910 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3915 * This is mostly a bugfix release
3917 * Support optional initialization of the machine ID from the
3918 KVM or container configured UUID.
3920 * Support immediate reboots with "systemctl reboot -ff"
3922 * Show /etc/os-release data in systemd-analyze output
3924 * Many bugfixes for the journal, including endianness fixes and
3925 ensuring that disk space enforcement works
3927 * sd-login.h is C++ comptaible again
3929 * Extend the /etc/os-release format on request of the Debian
3932 * We now refuse non-UTF8 strings used in various configuration
3933 and unit files. This is done to ensure we do not pass invalid
3934 data over D-Bus or expose it elsewhere.
3936 * Register Mimo USB Screens as suitable for automatic seat
3939 * Read SELinux client context from journal clients in a race
3942 * Reorder configuration file lookup order. /etc now always
3943 overrides /run in order to allow the administrator to always
3944 and unconditionally override vendor supplied or
3945 automatically generated data.
3947 * The various user visible bits of the journal now have man
3948 pages. We still lack man pages for the journal API calls
3951 * We now ship all man pages in HTML format again in the
3954 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3955 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3956 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3961 * This is mostly a bugfix release
3963 * systems lacking /etc/os-release are no longer supported.
3965 * Various functionality updates to libsystemd-login.so
3967 * Track class of PAM logins to distuingish greeters from
3970 Contributions from: Kay Sievers, Lennart Poettering, Michael
3975 * This is an important bugfix release for v41.
3977 * Building man pages is now optional which should be useful
3978 for those building systemd from git but unwilling to install
3981 * Watchdog support for supervising services is now usable. In
3982 a future release support for hardware watchdogs
3983 (i.e. /dev/watchdog) will be added building on this.
3985 * Service start rate limiting is now configurable and can be
3986 turned off per service. When a start rate limit is hit a
3987 reboot can automatically be triggered.
3989 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3991 Contributions from: Benjamin Franzke, Bill Nottingham,
3992 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3993 Schmidt, Michał Górny, Piotr Drąg
3997 * The systemd binary is installed /usr/lib/systemd/systemd now;
3998 An existing /sbin/init symlink needs to be adapted with the
4001 * The code that loads kernel modules has been ported to invoke
4002 libkmod directly, instead of modprobe. This means we do not
4003 support systems with module-init-tools anymore.
4005 * Watchdog support is now already useful, but still not
4008 * A new kernel command line option systemd.setenv= is
4009 understood to set system wide environment variables
4010 dynamically at boot.
4012 * We now limit the set of capabilities of systemd-journald.
4014 * We now set SIGPIPE to ignore by default, since it only is
4015 useful in shell pipelines, and has little use in general
4016 code. This can be disabled with IgnoreSIPIPE=no in unit
4019 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4020 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4025 * This is mostly a bugfix release
4027 * We now expose the reason why a service failed in the
4028 "Result" D-Bus property.
4030 * Rudimentary service watchdog support (will be completed over
4031 the next few releases.)
4033 * When systemd forks off in order execute some service we will
4034 now immediately changes its argv[0] to reflect which process
4035 it will execute. This is useful to minimize the time window
4036 with a generic argv[0], which makes bootcharts more useful
4038 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4039 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4040 Mike Kazantsev, Ray Strode
4044 * This is mostly a test release, but incorporates many
4047 * New systemd-cgtop tool to show control groups by their
4050 * Linking against libacl for ACLs is optional again. If
4051 disabled, support tracking device access for active logins
4052 goes becomes unavailable, and so does access to the user
4053 journals by the respective users.
4055 * If a group "adm" exists, journal files are automatically
4056 owned by them, thus allow members of this group full access
4057 to the system journal as well as all user journals.
4059 * The journal now stores the SELinux context of the logging
4060 client for all entries.
4062 * Add C++ inclusion guards to all public headers
4064 * New output mode "cat" in the journal to print only text
4065 messages, without any meta data like date or time.
4067 * Include tiny X server wrapper as a temporary stop-gap to
4068 teach XOrg udev display enumeration. This is used by display
4069 managers such as gdm, and will go away as soon as XOrg
4070 learned native udev hotplugging for display devices.
4072 * Add new systemd-cat tool for executing arbitrary programs
4073 with STDERR/STDOUT connected to the journal. Can also act as
4074 BSD logger replacement, and does so by default.
4076 * Optionally store all locally generated coredumps in the
4077 journal along with meta data.
4079 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4080 writing short strings to files (for usage for /sys), and for
4081 creating symlinks, character and block device nodes.
4083 * New unit file option ControlGroupPersistent= to make cgroups
4084 persistent, following the mechanisms outlined in
4085 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4087 * Support multiple local RTCs in a sane way
4089 * No longer monopolize IO when replaying readahead data on
4090 rotating disks, since we might starve non-file-system IO to
4091 death, since fanotify() will not see accesses done by blkid,
4094 * Do not show kernel threads in systemd-cgls anymore, unless
4095 requested with new -k switch.
4097 Contributions from: Dan Horák, Kay Sievers, Lennart
4098 Poettering, Michal Schmidt
4102 * This is mostly a test release, but incorporates many
4105 * The git repository moved to:
4106 git://anongit.freedesktop.org/systemd/systemd
4107 ssh://git.freedesktop.org/git/systemd/systemd
4109 * First release with the journal
4110 http://0pointer.de/blog/projects/the-journal.html
4112 * The journal replaces both systemd-kmsg-syslogd and
4113 systemd-stdout-bridge.
4115 * New sd_pid_get_unit() API call in libsystemd-logind
4117 * Many systemadm clean-ups
4119 * Introduce remote-fs-pre.target which is ordered before all
4120 remote mounts and may be used to start services before all
4123 * Added Mageia support
4125 * Add bash completion for systemd-loginctl
4127 * Actively monitor PID file creation for daemons which exit in
4128 the parent process before having finished writing the PID
4129 file in the daemon process. Daemons which do this need to be
4130 fixed (i.e. PID file creation must have finished before the
4131 parent exits), but we now react a bit more gracefully to them.
4133 * Add colourful boot output, mimicking the well-known output
4134 of existing distributions.
4136 * New option PassCredentials= for socket units, for
4137 compatibility with a recent kernel ABI breakage.
4139 * /etc/rc.local is now hooked in via a generator binary, and
4140 thus will no longer act as synchronization point during
4143 * systemctl list-unit-files now supports --root=.
4145 * systemd-tmpfiles now understands two new commands: z, Z for
4146 relabelling files according to the SELinux database. This is
4147 useful to apply SELinux labels to specific files in /sys,
4150 * Output of SysV services is now forwarded to both the console
4151 and the journal by default, not only just the console.
4153 * New man pages for all APIs from libsystemd-login.
4155 * The build tree got reorganized and a the build system is a
4156 lot more modular allowing embedded setups to specifically
4157 select the components of systemd they are interested in.
4159 * Support for Linux systems lacking the kernel VT subsystem is
4162 * configure's --with-rootdir= got renamed to
4163 --with-rootprefix= to follow the naming used by udev and
4166 * Unless specified otherwise we will now install to /usr instead
4167 of /usr/local by default.
4169 * Processes with '@' in argv[0][0] are now excluded from the
4170 final shut-down killing spree, following the logic explained
4172 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4174 * All processes remaining in a service cgroup when we enter
4175 the START or START_PRE states are now killed with
4176 SIGKILL. That means it is no longer possible to spawn
4177 background processes from ExecStart= lines (which was never
4178 supported anyway, and bad style).
4180 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4181 reloading of units together.
4183 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4184 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4185 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4186 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4187 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek